This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Geological Society Special Publications Society Book Editors R. J. PANKHURST (CHIEF EDITOR) P. DOYLE F. J. GREGORY J. S. GRIFFITHS A. J. HARTLEY R. E. HOLDSWORTH
A. C. MORTON N. S. ROBINS M. S.STOKER J. P. TURNER Special Publication reviewing procedures The Society makes every effort to ensure that the scientific and production quality of its books matches that of its journals. Since 1997, all book proposals have been refereed by specialist reviewers as well as by the Society's Books Editorial Committee. If the referees identify weaknesses in the proposal, these must be addressed before the proposal is accepted. Once the book is accepted, the Society has a team of Book Editors (listed above) who ensure that the volume editors follow strict guidelines on refereeing and quality control. We insist that individual papers can only be accepted after satisfactory review by two independent referees. The questions on the review forms are similar to those for Journal of the Geological Society. The referees' forms and comments must be available to the Society's Book Editors on request. Although many of the books result from meetings, the editors are expected to commission papers that were not presented at the meeting to ensure that the book provides a balanced coverage of the subject. Being accepted for presentation at the meeting does not guarantee inclusion in the book. Geological Society Special Publications are included in the ISI Index of Scientific Book Contents, but they do not have an impact factor, the latter being applicable only to journals. More information about submitting a proposal and producing a Special Publication can be found on the Society's web site: www.geolsoc.org.uk.
It is recommended that reference to all or part of this book should be made in one of the following ways: VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, A.J. & MORLEY, C.K. (eds) 2003. Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216. HURST, A., CARTWRIGHT, J. & DURANTI, D. 2003. Fluidization structures produced by upward injection of sand through a sealing lithology. In: VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, A.J. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216,123-137.
GEOLOGICAL SOCIETY SPECIAL PUBLICATION NO. 216
Subsurface Sediment Mobilization
EDITED BY P.VANRENSBERGEN Ghent University, Belgium
R.R. HILLIS University of Adelaide, Australia
A.J. MALTMAN University of Wales, UK and
C.K. MORLEY University of Brunei Darussalam, Brunei Darussalam
Distributors USA AAPG Bookstore PO Box 979 Tulsa OK 74101-0979 USA Orders: Tel. + 1 918 584-2555 Fax+1918 560-2652 E-mail [email protected] India Affiliated East-West Press PVT Ltd G-l/16 Ansari Road, Daryaganj, New Delhi 110 002 India Orders: Tel.+91 11 327-9113 Fax+ 91 11326-0538 E-mail [email protected] Japan Kanda Book Trading Company Cityhouse Tama 204 Tsurumaki 1-3-10 Tama-shi, Tokyo 206-0034 Japan Orders: Tel. + 81 (0)423 57-7650 Fax+ 81 (0)423 57-7651 E-mail [email protected]
Contents
VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, AJ. & MORLEY, C.K. Subsurface sediment mobilization: introduction MALTMAN, AJ. & BOLTON, A. How sediments become mobilized Shallow subsurface sediment mobilization OWEN, G. Load structures: gravity-driven sediment mobilization in the shallow subsurface HARRISON, P. & MALTMAN, A. J. Numerical modelling of reverse-density structures in soft non-Newtonian sediments PARIZE, O. & FRIES, G. The Vocontian clastic dykes and skills: a geometric model BANKWITZ, P., BANKWITZ, E., BRAUER, K., KAMPF, H. & STORR, M. Deformation structures in Plioand Pleistocene sediments (NW Bohemia, Central Europe) LEDESERT, B., BURET, C., CHANIER, E, FERRIERE, J. & RECOURT, P. Tubular structures of northern Wairarapa (New Zealand) as possible examples of ancient fluid expulsion in an accretionary prism: evidence from field and petrographical observations DRAGANITS, E., GRASEMANN, B. & SCHMID, H.P. Fluidization pipes and spring pits in a Gondwanan barrier-island environment: Groundwater phenomenon, palaeo-seismicity or a combination of both? HURST, A., CARTWRIGHT, J. & DURANTI, D. Fluidization structures produced by upward injection of sand through a sealing lithology L0SETH, H., WENSAAS, L., ARNTSEN, B. & HOVLAND, M. Gas and fluid injection triggering shallow mud mobilization in the Hordaland Group, North Sea PRALLE, N., KULZER, M. & GUDEHUS, G. Experimental evidence on the role of gas in sediment liquefaction and mud volcanism GAY, A., LOPEZ, M., COCHONAT, P., SULTAN, N., CAUQUIL, E. & BRIGAUD, F. Sinuous pockmark belt as indicator of a shallow buried turbiditic channel on the lower slope of the Congo basin, West African margin NOUZE, H. & BALTZER, A. Shallow bottom-simulating reflectors on the Angola margin, in relation with gas and gas hydrate in the sediments VAN RENSBERGEN, P., POORT, J., KIPFER, R., DE BATIST, M., VANNESTE, M., KLERKX, J., GRANIN, N., KHLYSTOV, O. & KRINITSKY, P. Near-surf ace sediment mobilization and methane venting in relation to hydrate destabilization in Southern Lake Baikal, Siberia Polygonal faults and sediment mobilization CARTWRIGHT, J., JAMES, D. & BOLTON, A. The genesis of polygonal fault systems: a review NICOL, A., WALSH, J.J., WATTERSON, J., NELL, P.A.R. & BRETAN, P. The geometry, growth and linkage of faults within a polygonal fault system from South Australia STUEVOLD, L.M., FAERSETH, R.B., ARNSEN, L., CARTWRIGHT, J. & MOLLER, N. Polygonal faults in the Ormen Lange Field, M0re Basin, offshore Mid Norway BERNDT, C., BUNZ, S. & MIENERT, J. Polygonal fault systems on the mid-Norwegian margin: a long-term source forluidfflow HIBSCH, C., CARTWRIGHT, J. HANSEN, D.M., GAVIGLIO, P., ANDRE, G., GUSHING, M., BRACQ, P., JUIGNET, P., BENOIT, P. & ALLOUC, J. Normal faulting in chalk: tectonic stresses vs. compactionrelated polygonal faulting MERTENS, J., VANDENBERGHE, N., WOUTERS, L. & SINTUBIN, M. The origin and development of joints in the Boom Clay Formation (Rupelian) in Belgium WATTRUS, N.J., RAUSCH, D.E. & CARTWRIGHT, J. Soft-sediment deformation in Lake Superior: evidence for an immature Polygonal Fault System?
1 9 21 35 51 73
95
109 123 139 159
173 191
207 223 245 263 283
291 309 323
vi
CONTENTS
Deep subsurface sediment mobilization MORLEY, C.K. Mobile shale related deformation in large deltas developed on passive and active margins HILLIS, R.R. Pore pressure/stress coupling and its implications for rock failure TINGAY, M.R.R, HILLIS, R.R., MORLEY, C.K., SWARBRICK, R.E. & OKPERE, B.C. Pore pressure/stress coupling in Brunei Darussalam - implications for shale injection MORLEY, C.K. Outcrop examples of mudstone intrusions from the Jerudong anticline, Brunei Darussalam, and inferences for hydrocarbon reservoirs VAN RENSBERGEN, P. & MORLEY, C.K. Re-evaluation of mobile shale occurrences on seismic sections of the Champion and Baram deltas, offshore Brunei McCLAY, K, DOOLEY, T. & ZAMORA, G. Analogue models of delta systems above ductile substrates TOTTERDELL, J.M. & KRASSAY, A. A. The role of shale deformation and growth faulting in the Late Cretaceous evolution of the Bight Basin, offshore southern Australia TALUKDER, A.R., COMAS, M.C. & SOTO, J.L Pliocene to Recent mud diapirism and related mud volcanoes in the Alboran Sea (Western Mediterranean) YASSIR, N. The role of shear stress in mobilizing deep-seated mud volcanoes: geological and geomechanical evidence from Trinidad and Taiwan DEVILLE, E., BATTANI, A., GRIBOULARD, R., GUERLAIS, S., HEREIN, J.P., HOUZAY, J.P., MULLER, C. & PRINZHOFER, A. The origin and processes of mud volcanism: new insights from Trinidad DEYHLE, A., KOPF, AJ. & ALOISI, G. Boron and boron isotopes as tracers for diagenetic reactions and depth of mobilization, using muds and authigenic carbonates from eastern Mediterranean mud volcanoes PARNELL, J. & KELLY, J. Remobilization of sand from consolidated sandstones: evidence from mixed bitumen-sand intrusions Index
335 359 369 381 395 411 429 443 461 475
491 505 515
Subsurface sediment mobilization: introduction PIETER VAN RENSBERGEN1, RICHARD R. HILLIS2, ALEX J. MALTMAN3 & CHRISTOPHER K. MORLEY4 1
Renard Centre of Marine Geology, Universiteit Gent, Krijgslaan 281-S8, Gent, Belgium. (e-mail: pieter_yanrensbergen @yahoo. com) 2 National Centre of Petroleum Geology and Geophysics, University of Adelaide, Adelaide, SA 5005, Australia 3 Institute of Geography and Earth Sciences, Aberswyth University of Wales, Aberystwyth, SY23 2AK, UK. 4 Department of Petroleum Geosciences, University of Brunei Darussalam, Bandar Seri Begawan 2023, Brunei
Subsurface sediment mobilization (SSM) - which includes soft sediment deformations, sand injections, shale diapirs and mud volcanoes - is more widespread than previously thought. The ever-increasing resolution of subsurface data yielded many new observations of SSM, not only from regions obviously prone to sediment remobilization, such as an active tectonic setting or in a region with exceptionally large sediment supply, but also from tectonically quiescent areas. Until now, all the different aspects of SSM have largely been treated as separate phenomena. There is very little cross-referencing between, for example, studies of mud volcanoes and those of sand injections, although both are caused by sediment fluidization. Divisions according to sediment type, mobilization depth or triggering mechanism make little sense when trying to understand the processes of SSM. There is a gradation in mobilization processes that cause considerable overlap between categories in any classification. Hence, it is necessary to integrate our understanding of all types of SSM, regardless of scale, depth, location, grain size or triggering mechanism. In addition, polygonal faults are important in this context, as this nontectonic structural style is closely associated with sedimentary injections and may also reflect large scale mobilization. The main goal of this volume is to help develop a more integrated understanding of subsurface sediment mobilization. It contains specific case studies and a number of overview papers about the mechanisms of sediment mobilization in the subsurface (Maltman & Bolton), about polygonal faulting (Cartwright) and about shale diapirs (Morley). Other recent review papers were published about sand injections by Jolly & Lonergan (2002) but additional pertinent remarks are presented here by Hurst et al and about mud volcanism and mud diapirism by Kopf (2002) and Dimitrov (2002). Older but important reviews are found in Maltman (1994) and Jones & Preston (1987).
The contributions in this volume are divided into three sections: shallow subsurface sediment mobilization, polygonal faulting and deep-seated sediment mobilization. We have arbitrarily put the limit between the shallow and deep subsurface at 500-1000 m, which is about the maximum depth of sand injections (Jolly & Lonergan 2002) and the depth of onset of sand lithification, the lower depth limit of shallow fluid generating processes and the common transition depth to regional abnormal pore fluid pressure. Although there are significant differences between shallow processes, deep processes and polygonal faults, the boundaries between these processes are not clear and there is considerable overlap. This review paper aims to link the contributions together into this common framework.
Shallow subsurface sediment mobilization Unconsolidated sediment in the shallow subsurface is most prone to subsurface remobilization: the porosity is high, cohesion and intergranular bonds are low and fluid production rate by compaction is high. Liquefaction, plastic flow and fluidization can occur. The most important fluid source in the shallow subsurface is pore water being lost through consolidation. At greater depths, but still within what we have classified here as the shallow zone, bacterial gas generation and gas hydrate accumulation/decomposition can be important, depending on the concentration of organic matter, bacterial processes, inflow of hydrocarbon gases, temperature and pressure. This section is further divided here into three subsections: mobilization related to pore water - mostly confined to very shallow depths; mobilization related to in situ gas generation and gas hydrates; and mobilization related to injection of gas or overpressured fluids from a deeper source.
Shallow mobilization related to expelled pore water Maltman & Bolton provide a review of the mechanisms by which sediments can be mobilized, how solid sediment can change to a fluid and back to solid again. The two most important requirements are pore-fluid overpressure, to cause fluidal state by liquefaction or fluidization and the occurrence of an hydraulic gradient to induce flow. Sediment mobilization structures can be divided into load structures, characterized by folding of an interface and fluidization structures, characterized by injection of fluidized sediment into the host rock. Load structures range from simple load casts to pseudo-nodules, ball-and-pillow structures and water-escape cusps (Owen); their size varies from millimetres to several metres. They most commonly affect the interface separating sand from clay; hence the deformation has implications for the lateral and vertical permeability and connectivity of sandstones. The morphology of load casts and flame structures are mostly explained in terms of viscosity contrast and fluidal behaviour. For example, Harrison & Maltman use numerical modelling to simulate flow driven by buoyancy forces resulting from reversedensity stratification to study the different roles of initiating conditions, inertia and rheological behaviour on the morphologies and timing of formation of natural features such as load casts and flame structures. Owen, on the other hand, suggests that the varying morphologies of load casts and flame structures may be caused by different growth rates rather than differences in viscosity. The varying degrees of deformation in this case could be controlled by the magnitude and duration of the driving force. Fluidization structures result from rapid migration of fluids through unconsolidated sediment. A large variety of loading mechanisms can trigger fluid flow in the shallow subsurface, ranging from earthquakes to footsteps (Maltman & Bolton). Sediment volcanoes and sand injectites are typical results of fluidization. Rapid fluid expulsion through sands often occurs along isolated pipe-like conduits. Differences in mineralization, cementation and grain packing may preserve such fluidization pipes in the rock record and bear witness to past fluid flow events. Draganits etaL interpreted concentric cylindrical pipes (1.5m high, 0.8 m in diameter) in Lower Devonian barrier island arenites of the North Indian Gondwana coast in the Himalayas as fluidization pipes. Based on the depositional setting and the stratigraphic framework Draganits et al. inferred that increased hydraulic head in the beach sands pore fluids following a rapid sea level rise may have caused the numerous spring pits and fluidization pipes. Ledesert et al. studied tubular concretions (0.6 m in diameter) in Mio-Pliocene sediments at the
Hikurangi subduction margin, on the northern Wairarapa coast of New Zealand's North Island. The concretions are attributed to the flow of carbonaterich water through silty sediment. The extent of cementation suggests that possibly two episodes of fluid expulsion occurred. The first episode occurred relatively early, when the sediment was unconsolidated during the fluid flow. The second episode occurred after consolidation of the sediment, possibly during a compressional event. Another type of fluidized sand injection is shown in spectacular outcrops in the Vocontian Basin (SE France). Complex networks of clastic sills (up to 10 m thick and several km long) and dykes, injected into a thick marly deep-water succession, are found in the channel banks, fed laterally from sandy channels. From the outcrop it appears that sand was injected into the channel banks contemporaneously with sand deposition in the channel. Dykes branching off lateral sills were injected downwards into the deep-water mud. These injections are probably not related to upward fluid migration along a hydraulic gradient but rather to fractures formed by sudden loading by turbidites (Parize etaL).
Sediment mobilization related to gas and gas hydrate The role of gas bubbles in sediment with regard to subsurface sediment liquefaction and sediment extrusion was studied by laboratory experiments on soil samples by Pralle et al. They found that small amounts of enclosed gas bubbles render the soil compressible and consequently enhances local shearing, pore pressure build-up and structural damage. It was observed that liquidized, overpressured sediments form mud chambers, whose excess pressure is released through cracks and other discontinuities. Subsequently the liquidized sediment migrates upward, driven by its own overpressure and by the buoyancy of the enclosed gas bubbles, until extruded at the surface. This model has strong similarities to the description of the mud volcano feeder system in Barbados by Deville etaL In shallow subsurface sediment in marine or deep lacustrine basins, the occurrence of gas accumulations and migrations, together with gas hydrate formation and decomposition, causes a complex hydrodynamic setting with localized fluid and sediment extrusion features. Gay etaL document pockmark distribution that, in the absence of gas hydrate, follows closely the flanks of the meandering turbidite channel on the lower slope of the Congo fan, offshore West Africa. They also found that active gas venting occurred in a zone where the excess pore pressure exceeded the vertical confining pressure of the overlying sediment wedge and that this active zone migrates seaward in step with progra-
SUBSURFACE SEDIMENT MOBILIZATION: INTRODUCTION
dationof the sediment wedge. Sediment properties and thus fluid migration patterns change drastically when gas hydrates are present. Hydrates block the sediment pores and decrease the permeability and hydrate formation requires large amounts of water and locally causes overcompaction. Hydrate dissociation releases large volumes of gas (mostly methane) and can locally generate overpressure. A detailed geophysical study by Nouze & Baltzer of anomalous reflections also at the lower slope of the Congo fan identified three zones of abnormal pressure lying sub parallel to the sea floor. Overpressure related to hydrate accumulation and dissociation as follows: free gas accumulation at the base of the hydrate occurrence zone, free gas accumulation by hydrate dissociation at the top of the hydrate occurrence zone and a shallow layer of overpressured, undercompacted sediment, also interpreted as the result of hydrate dissociation. From geophysical studies in Lake Baikal, Siberia, Van Rensbergen etal. describe localized hydrate dissociation by injection of hydrothermal fluids at the base of the hydrate layer, triggering gas injection and short-lived mud volcanism.
Sediment mobilization related to gas and fluid injection. The ability of fluids to mobilize shallow sediment is demonstrated by L0seth et al. for vertical fluid migration and injection into the Tertiary Hordaland Group in the northern North Sea. Gas, oil and formation water from the Jurassic reservoirs in the deep Viking Graben gave rise to injection along vertical chimneys into the shallow (<1000 m) unconsolidated sediment forming the upper part of the Hordaland Group. The result was large-scale mobilization of sediments to form low-density mud diapirs, up to 100 km long and 40 km wide and circular mud volcanoes, 1-3 km in diameter. It is remarkable that the deformed unit continues laterally in a section affected by polygonal faults. Large sand injections as imaged on 3D seismic data, are also partly attributed to the injection of fluids from deep sources. Sand injections occur between <10 m and 500 m (max 800 m, Jolly & Lonergan 2002) below the sea floor. Hurst et al. suggest that small-scale sand fluidization features can occur by expulsion of formation water following sand liquefaction. However major injection structures visible on seismic data need substantial external fluid sources to provide the necessary fluid flux to initiate and sustain fluidization and sand injection on such a scale. Polygonal faults in fine-grained sediment form preferential pathways for sand injections, regardless of whether the fault system developed before, during or after the overpressure increase in the isolated sand body.
3
Injection of volatiles from mantle CO2 degassing can be a cause of sediment deformation in unconsolidated fine-grained sediment and may possibly occur in combination with earthquake deformation. This combination of factors is discussed by Bankwitz et al. with regards to a large variety of deformation structures in the unconsolidated sediments of the Neogene Cheb Basin (Czech Republic).
Polygonal faults and sediment mobilization Polygonal fault systems are intraformational faults, typically developed in fine-grained sediment, that often exhibit a polygonal pattern in plan view. They occur independently of any tectonic process, instead being entirely related to post-depositional processes within the sedimentary layer (although their mechanism of formation is not yet well understood). Polygonal faults are important in discussions about subsurface sediment mobilization because: •
•
•
they are frequently associated with fluidized sediment injections along the fault planes, which in places affects hydrocarbon reservoir architecture; they are thought to be important pathways for concentrated fluid flow through low-permeability sequences and the volume contraction associated with the genesis of polygonal faults may also be an important fluid source, and; some authors interpret polygonal faults as a result of sediment mobilization caused by largescale density inversion.
The genesis of layer-bound fault arrays with polygonal map view patterns is attributed to three main mechanisms: density inversion (Henriet etal 1992), volumetric contraction by syneresis (Cartwright & Dewhurst 1998) and gravitational loading of weak sediments with low frictional strength (Goulty 2001). Cartwright et al. provide an overview of these genetic mechanisms, discussing the evidence in favour of and against each of these mechanisms. Cartwright et al. favour the syneresis model but also indicate that so far there is no conclusive evidence for the dominance for any one of the mechanisms; they may even be complementary to some extent. Detailed case studies included in this volume advance discussions about the genesis of polyogonal faults, their capacity for generating and concentrating fluid flow and their importance for the reservoir architecture of deep-water sand deposits. Nicol et al. analyse the Lake Hope polygonal fault system in Australia and interpret it as a result of incipient gravitational overturn of a thin (c. 40 m) low-density, overpressured layer. Quantitative
4
P. VAN RENSBERGEN ETAL
systematics of fault geometries are consistent with the scale-bound nature of the system, in which the presence and the thickness of an underlying mobile layer is crucial. Fault linkage patterns are consistent with a thickening of the mobile layer in the footwall and the thinning and eventually grounding of the overlying sequence in the hanging walls. In this model, fault movement is entirely attributed to the movement of a mobile layer at the base. Mertens et al describe a field example of an orthogonal set of tensional joints in the Lower Oligocene Boom Clay, Belgium, probably caused by shrinkage of the clay due to water loss. The tensional joints formed during the Late Oligocene-Early Miocene at a maximum depth of 40-50 m below the surface. The joints are spaced between 0.5 and a few metres and formed by lateral contraction. Mertens et al. discuss the boundary conditions necessary for such joints to form. They may represent an outcrop example of contraction joints initiating a polygonal fault system. Another, quite different set of normal faults and fractures, observed in glaciolacustrine sediment in Lake Superior, is interpreted by Wattrus etal. as an immature polygonal fault system caused by dewatering of glacio-lacustrine sediment. This interpretation is based on the occurrence of circular patterns on the lake floor, which are, according to Wattrus et al. caused by fluid venting along polygonal faults in glacio-lacustrine sediment underlying the irregular Holocene sediment drape. Wattrus et al. suggest that the normal faults form by volumetric contraction during syneresis. Expulsion of fluids during contraction along the faults and fractures prevented settling of the Holocene cover. Berndt et al. suggest that polygonal fault systems may not only provide important pathways for fluid migration but also may be important sources for fluids. Berndt et al. describe fluid migration features on 3D seismic data over a Miocene polygonal fault system, from which they infer that fluid expulsion has been an ongoing process following initiation of the fault system. The absence of fluid flow indicators below the polygonal fault system seems to indicate that most of the fluid is produced within the faulted sequences, probably by continued sediment contraction. The influence of polygonal fault systems on the hydrocarbon reservoir characteristics of the Ormen Lange field, offshore Norway, is discussed by Stuevold et al. This is the first case study of a major hydrocarbon accumulation in which polygonal faults play an important role in the reservoir architecture and fluid communication. Displacements and fault distributions are highly variable and controlled by fault intersection geometry. Fault intersections are thought by Stuevold et al. to be sites for preferential fluid communication. The relationship between
polygonal fault systems and deep-water reservoir sands is also briefly discussed by Hurst etal. Most examples of polygonal faults are found in fine-grained siliciclastic sediments, but recently polygonal faulting by sediment contraction has also been interpreted to occur in chalk. This possibility is investigated by Hibsch et al. for various outcrops in France and in the United Kingdom. Hibsch et al. suggest that intraformational fault sets with greatly varying strikes observed in outcrops are difficult to fit in a complex poly-phased tectonic history, but may be the result of a single deformational phase involving volumetric contraction and polygonal faulting.
Deep subsurface sediment mobilization Sediment mobilization in the deep subsurface, i.e. generally > 1 km depth is usually associated with large pore fluid overpressure and is common in settings where large overpressures can develop, i.e. hydrocarbon provinces, basins with large sediment supply and regions of compressional tectonic stresses. Overpressure does not reinstate ductile sediment properties exhibited at shallow depth, sands are lithified and cannot liquefy, undercompacted shale may move by shearing at critical state conditions. However, high pore-fluid pressures can cause fluidization after brecciation and the breaking of grain bonds. This section considers three main types of deep-seated sediment mobilization, which are closely interrelated: mud or shale diapirism, mud volcanism and sand-bitumen mobilization. Fluidized sediment injection along hydrofractures Fracture generation by pore fluid overpressure is one of the most important processes for sediment mobilization in the deep subsurface. Seal rupture, which is generally required for SSM, occurs if the effective stresses (difference between total, applied stresses and pore pressure) are such that brittle failure is caused. As discussed by Hillis, the effect of increasing pore pressure on rock failure is more complex than commonly assumed because changes in pore pressure are coupled to (not independent of) changes in total stress. In normal fault regimes, pore pressure/stress coupling increases the propensity for tensile failure with respect to that for shear failure. Furthermore, the amount of overpressure that can be maintained by a seal is larger than would be predicted if pore pressure/stress coupling is ignored. The principle of pore pressure/stress coupling is applied in a study offshore Brunei by Tingay et al., who analyse an internal blowout event offshore
SUBSURFACE SEDIMENT MOBILIZATION: INTRODUCTION
Brunei. A deep overpressured reservoir was drilled and overpressure was transferred up the open hole to a shallower reservoir. A remarkable aspect of the internal blowout was the transfer of overpressured fluids from the shallower reservoir to the sea-bed resulting in a new vertical hydrofractured pathway. The overpressured fluids did not migrate along a preexisting fault to the seabed. Injections of mudstone through hydraulic fractures above an inferred mudstone diapir into bedded sandstone and shale deposits are described from outcrops by Morley. The outcrop examples show intrusive mudstone dykes and sills that compartmentalize the host rock reservoir. Sediment and fluid injection may inflate bedded shale layers and cause new hydrofractures propagating from the inflated shale layer. Morley suggests that diapirs may rise by hydraulic fracturing which limits the roof strength to the minimal horizontal stress plus the tensile strength of the country rock or normal faults (as opposed to the shear strength of intact rock for salt models). Hydraulic fracturing and sediment injection weakens the cap rock and may facilitate diapiric rise. Mixed sand-bitumen veins are another example of sediment injection along hydrofractures. Parnell & Kelly describe sandstone disintegration and subsequent sand fluidization, hydrofracturing and fluid flow. It is demonstrated that the passage of fluid hydrocarbons through lithified sand can isolate and spall sandstone clasts as well as individual grains from a sandstone and inject sand upwards (or occasionally downwards) into other sequences. The study demonstrates that sand injections can occur after disintegration of lithified sands and that it should not be assumed that all cases of mobilized sands are near-surf ace features.
Mud volcanoes Mud volcanoes are typically the surface expression of fluidized sediment injection and expulsion from over-pressured sediment layers in the subsurface. Sampling and analysis of mud, clasts and fluids allow an estimation of the depth and origin of overpressured fluids, the depth of sediment mobilization and the mobilization process. Deville et al. present an extensive study of mud volcanoes in Trinidad, based on surface samples and results from numerous industry oil wells. Mud volcanoes occur only in areas of tectonic compression. The fluid and gases were attributed to a deep origin, probably from depths up to 5000 m. The gas was generated by the thermogenic cracking of organic-rich horizons, probably located within the Cretaceous formations. The solid fraction of the material extruded is polygenic and originates from several
5
levels, ranging from the Cretaceous to the Pliocene. The feeding system of the mud volcanoes consists of a deep conduit in the overpressured zone, a mud chamber intruding the overburden around and above the top of the overpressured pressure zone (about 1000 m) and a superficial outlet leading to the surface vents. As such, this structure is similar to the feature described by L0seth et al. and the models generated by Pralle et al. Yassir discusses the role of shear stress in mobilizing mud in the deep subsurface, based on high pressure triaxial tests on muds from mud volcanoes in Trinidad and Taiwan. The texts show that tectonic activity plays an important role in mud volcano development and demonstrates that shear stresses are capable of creating enormous overpressures in sedimentary rocks and cause undercompacted sediments to flow at critical state conditions. The critical state model predicts that a normally consolidated sediment will contract during shear. If shear occurs in an undrained state, then the pore pressure will increase, causing a reduction in effective stress. In addition, the tests show that shearing is accompanied by only a slight increase, or even a decrease, in differential stress, or shear strength, which demonstrates the potential to flow during deformation. Deyhle et al. use boron isotopes as a tracer for the depth and origin of mud volcanic processes in the Mediterranean Ridge accretionary prism near Crete (Greece) and the Anaximander Mountains south of Turkey. The boron isotope study allows a distinction to be made between the relatively shallow origin of the mud from Mediterranean Ridge mud volcanoes (1-2 km below seafloor) and the deeper remobilization depth of rocks from the Anaximander Mountains mud domes (3-5 km below seafloor). Determining the mud origin provides better constraints on the processes involved in mud expulsion in a complex structural setting. Talukder et al present another case study of Pliocene and Quaternary mud diapirism and mud volcanism in the Alboran Sea (western Mediterrenean Sea). They, like the above studies, found that regional compression was the main driving force triggering mud diapirism and volcanism. Two major pulses of diapiric rise were distinguished within the Pliocene. Fluid expulsion from diapiric structures caused mud volcanoes and local collapse depressions.
Mobile shale in large deltas Large-scale sediment mobilization in the deep subsurface and associated deformation of the overburden occurs in large Tertiary deltas. Undercompacted prodelta clay is loaded by the progradational delta and is thought to be deformed and mobilized under the prograding differential load. In this case, the
6
P. VAN RENSBERGEN ETAL
Fig. 1. Overview of the in situ fluid sources in the subsurface, plotted against the porosity-depth curves of clay and sand (based on Karig & Hou 1992; Tissot & Welte 1984; and Claypool & Kaplan 1974). Fluid injection can occur at any depth and is not included in the figure.
undercompacted clay is considered to behave as a viscous to plastic mobile layer, like salt layers. McClay et al. studied delta evolution by analogue modelling using a polymer layer as a mobile substratum. The analogue models succeeded in producing a typical basinward stepping sequence of back-toback growth faults, observed on regional seismic sections, by delta progradation and mobile layer migration. The models show that delta top grabens are linked to delta toe contractional fold and thrust diapirs. On the other hand, it is argued by Van Rensbergen & Morley that the thickness of mobile shale on regional seismic data is often considerably overestimated. In several places the chaotic seismic facies commonly considered characteristic of mobile shale were found to be areas of poor seismic quality caused by a variety of factors. Some shale diapirs were found to be vertical fluidization pipes and mud volcanoes. On this basis they suggest that basinward fluid flow may be much more important than the flow of sediment. Morley integrates these contrasting views in a coherent overview of mobile shale-related deformation along passive and active margins. Important conclusions from this overview are that the structural style of deltas may change laterally and over time with the thickness of the mobile
shale substratum. Deposition of undercompacted mobile shale is probably not uniform but related to depocentres and sediment supply. Hence related deformation will not be uniform throughout the delta. For example, lateral parts of the delta, towards the continental slope, may exhibit gravity sliding over an inclined basin whereas in the central part, where the mobile substratum is thicker, the deformation will be more characteristic of differential loading. As a delta progrades into a basin, the thickness of the underlying mobile substratum may increase and the early tectonic style may thus be different from the tectonic style of a mature welldeveloped delta. In other words, the structural style of a delta is more variable than accounted for in most structural models. The volume closes with an example of changing structural aspects of a progradational delta in relation to the thickness of a mobile shale substratum (Totterdell & Krassay). Two delta systems underlying the Great Australian Bight, the Hammerhead and White Pointer deltas are described. A broad belt of regional growth faults and compressional deformation at the delta toe characterizes the former. The structural style is interpreted as a combination of differential loading and gravity gliding
SUBSURFACE SEDIMENT MOBILIZATION: INTRODUCTION
over a mobile substrate. The White Pointer delta developed over a steep palaeo-shelf margin and developed a narrow band of gravity deformation in absence of syndepositional shale deformation. In this case the extensional growth faults occur within a narrow band.
Conclusions The range of case studies presented in this volume illustrates that subsurface sediment mobilization is a much more widespread and important phenomenon than previously recognized. However, in several cases different authors suggest conflicting interpretations of SSM processes such that it is not yet possible to make neither definitive conclusions regarding the processes involved nor the criteria with which specific mechanisms can be recognized. Nevertheless, some general conclusions can be summarized as follows:
7
It is clear that given the many different manifestations of SSM and the many different processes involved, further advances in the understanding of SSM require a multi-disciplinary and integrated approach. This volume results from a conference 'Subsurface Sediment Mobilization' in Gent, Belgium, September 2001. This conference was organized with the financial support of the European Commission (High-Level Scientific Conferences, nr. HPCF-CT-2001-00186), the Flemish Minister of Education, the Ormen Lange Consortium and the European Association of Geoscientists and Engineers. We thank all who contributed to the succes of this event and the production of this volume. We thank the reviewers who agreed to evaluate this large number of manuscripts and we also acknowledge the input of authors of presentations at the conference but whose contributions are not represented in the volume. Figure 1 in this introduction was made with the help of J. Mertens (NIRAS, Belgium).
References •
•
•
•
•
Subsurface sediment mobilization requires sediment that is able to move as a fluid and a pressure gradient to initiate flow. Since in the shallow subsurface sediment is less consolidated and fluid production rates are high, the shallow zone is more prone to remobilization than the deep subsurface. Load structures and fluidization structures are common. As the hydraulic gradient decreases, sediment movement ceases. Hence, the distance and volume of injections is in the shallow subsurface often limited by the fluid flux through the sediment. Gas and fluid injection into the unconsolidated shallow subsurface from deeper sources can cause sediment deformation over large areas. In fine-grained sediment this may create mud chambers of liquidized mud, which can give rise to mud diapirs and mud volcanoes. The formation of polygonal fault systems is still debated. Even so, it is evident that they are important pathways for sediment injection. Where they occur close to the sediment surface they may be focussing fluid flow along the fault planes during compaction. In the deep subsurface tectonic stresses in combination with large overpressures become more important to create hydraulic fractures and to create critical state conditions at which consolidated sediment may flow. Hydraulic fractures weaken the cap rock of diapirs and are pathways for sediment injection, in some cases creating feeder systems for mud volcanoes. Plastic deformation of consolidated sediment in the deep subsurface may cause large-scale structural deformation but its significance is contested.
CARTWRIGHT, J.A. & DEWHURST, D. 1998. Layer-bound compaction faults in fine-grained Sediments. Bulletin of the Geological Society of America, 110, 1242-1257. CLAYPOOL, G.E. & KAPLAN, I.R. 1974. The origin and distribution of methane in marine sediments. In: KAPLAN, I.R. (ed.) Natural gasses in marine sediments. Marine Science. Plenum Press, New York, 99-139. DIMITROV, L.I. 2002. Mud volcanoes - the most important pathway for degassing deeply buried sediments. Earth Science Reviews, 59,49-76. GOULTY, N.R. 2001. Polygonal fault networks in finegrained sediments- an alternative to the syneresis mechanism. First Break, 19, 69-73. HENRIET, J.P., DE BATIST, M. & VERSCHUREN, M. 1991. Early fracturing of Palaeogene clays, southernmost North Sea: relevance to mechanisms of primary hydrocarbon migration. In: SPENCER, A.M. (ed.) Generation, accumulation and production of Europe's hydrocarbons. Special Publication of the European Association of Petoleum Geologists, 1, 217-227. KARIG, D.E. & Hou, G. 1992. High-stress consolodiation experiments and their geological implications. Journal of Geophysical Research, 97,289-300. KOPF, A. 2002. Significance of mud volcanism. Reviews of Geophysics, doi 10.1029/2000RG000093. JOLLY, R.J.H. & LONERGAN, L. 2002. Mechanisms and controls on the formation of sand intrusions. Journal of the Geological Society of London, 159, 605-617. JONES, M.E. 1994. Mechanical principles of sediment deformation. In: MALTMAN, A. (ed.) The Geological Deformation of Sediments. Chapman & Hall, London, 37-71. MALTMAN, A. (ed.) 1994. The Geological Deformation of Sediments. Chapman & Hall, London, pp. 362. MARTINI A.M., BUDAI J.M., WALTER L.M. & SCHOELL M. 1996. Microbial generation of economic
8
P. VAN RENSBERGEN ETAL
accumulations of methane within a shallow organicrich shale. Nature, 383,155-158. OSBORNE, MJ. & SWARBRICK, R.E. 1997 Mechanisms for generating overpressure in sedimentary basins: A
reevaluation. American Association of Petroleum Geologists Bulletin, 81,1023-1041 TISSOT B.R & WELTE D.H. 1984. Petroleum Formation and Occurrence. Springer- Verlag, New York, pp. 699.
How sediments become mobilized ALEX J. MALTMAN & ALISTAIR BOLTON Institute of Geography and Earth Sciences, University of Wales, Aberystwyth, Wales SY23 3DB, UK (e-mail: ajm @ aber. ac. uk) Abstract: Geological sediments tend to strengthen during progressive burial but the interplay of porosity and permeability, strain and effective stress gives rise to numerous circumstances in which the strength increase can be temporarily reversed. The sediment becomes capable of bulk movement - sediment mobilization. Most explanations involve overpressuring, which results from additional loading being sustained by pore-fluid that is unable to dissipate adequately, leading to frictional strength reduction. The processes are highly heterogeneous, areally and with depth. The loads can be external ('dynamic') and both monotonic (e.g. a rapidly added suprajacent mass) and cyclic (e.g. the passage of waves), internal (e.g. the result of mineral reactions) and hydraulic (e.g. injection of external fluid). The sediments may become liquidized - that is, lose strength completely and behave as a fluid - through temporary fabric collapse (sensitive sediments) because loads are borne entirely by the pore-fluid (liquefaction), or by the grains becoming buoyant (fluidization), typically due to the ingress of externally derived fluids. In response to hydraulic gradients, buoyancy forces and reversed viscosity or density gradients, the weakened sediment may undergo bulk movement, though this requires failure of the enclosing material and sustained gradients. Mobilized but non-liquidized sediments retain some residual strength but can attain large shear displacements under critical state conditions.
Sediments undergoing burial tend to progressively increase in strength until they become rock. However, this strength increase can be temporarily reversed, with the sediment becoming so weak that it is capable of mobilization in numerous circumstances. In geological usage, the term mobilization (sometimes called remobilization) involves both rendering the sediment capable of motion and the bulk movement that commonly results. This paper is chiefly concerned with the former. Actually, most sediment mobilization probably takes place at the Earth's surface - due to water currents, mass movements, etc. - but we are here concerned with subsurface mobilization. For incompletely lithified sediments, this primarily depends on some form of what can generally be termed liquidization. Following Allen (1977, 1984) a liquidized sediment is one that behaves mechanically like a fluid (irrespective of the reason), even though it previously possessed a yield strength and hence had behaved as a solid (see also Owen 1987). The present article reviews liquidization and other mechanisms by which incompletely lithified sediments - those that move chiefly by grain slippage become capable of the bulk movements associated with subsurface mobilization (see Fig. 1). Although some of the principles outlined here can be extended to weak rocks, that is, where the mobilization involves the breaking of grains and inter-grain bonds and to intrinsically weak and ductile materials such as salt, here the primary concern is with sediments. The processes of movement, ranging from in situ mixing through the intrusion of clastic bodies to
extrusion at the surface of the sediment sequence and the whole range of resulting structures, are not dealt with here.
Burial and related processes Geological sediments are essentially mixtures of relatively strong grains with an intervening fluid, usually a brine. Of course, natural sediment systems are normally much more complex, with the fluid containing gases of changing solubilities and possibly immiscible hydrocarbon phases and the particles themselves changing in volume and shape, particularly where clay aggregates are involved. Burial tends to progressively strengthen the mixture (limification), by displacing the fluid and packing the particles closer together (consolidation), hence promoting inter-grain contact (frictional strength) and chemical inter-reactions (diagenesis). The various processes of diagenesis - principally cementation, recrystallization and diffusion mass transfer, commonly called pressure solution where it is assisted by fluids - take place progressively. The ability of intact grains to slide past each other to allow bulk movement of the sediment is progressively curbed and supplanted by mechanisms that deform the grains, that is independent particulate flow is superseded by cataclastic flow, as the sediment is turned into rock (e.g. see Maltman 1994). Brown & Orange (1993) argued that this transition is likely to occur (in fine sands) at effective confining pressures between about 1 and 5 MPa, signifying,
Fig. 1. Conceptual view of the range of processes involved in sediment mobilization. The untinted area indicates the scope of the present paper.
because elevated pore-pressures are typically involved, burial depths of several kilometres. We now begin looking at situations in which this progressive evolution is temporarily halted.
Consolidation Consolidation is the chief mechanical aspect of burial. It is defined as time-dependent mechanical reduction in sediment volume, usually pore volume, in response to increased loading. In practice, consolidation normally takes place due to the weight of progressive additions of overlying sediment, that is, the increasing lithostatic load. In ideal circumstances the pore-fluid in the buried sediment only sustains the load imparted by the suprajacent fluid (pore-fluid plus any suprajacent water), but if the sediment is unable to drain adequately in response, the lithostatic load becomes proportionally added to the pore-fluid pressure, reducing the frictional contact between the grains. Where the entire load is borne by the pore-fluid, the sediment loses all its frictional strength and, assuming grain cohesion is negligible, therefore becomes liquidized. It becomes available for mobilization and remains in this state
until the pore-pressure begins to re-equilibrate. The role of consolidation in mobilization is therefore intimately involved with the ability of the sediment to dissipate its pore water (e.g. Bitzer 1996). The loss of porosity in order to maintain hydrostatic equilibrium is termed normal consolidation (e.g. path A-D on Fig. 2). Where the burial load is taken up by the pore-fluid, the sediment remains with fixed porosity despite its increasing burial depth. Such a sediment is overpressured and underconsolidated because it shows a greater porosity than expected. The situation has long been known in sedimentary basins such as the Gulf of Mexico (e.g. Stump & Flemings 2000) and is commonly recognised from positive porosity deviations from a normal depth profile. Sediments have little elasticity, with the result that reductions in effective stress acting on a partially consolidated sediment do little to restore porosity and unloading paths tend to have little gradient (e.g. path D-C on Fig. 2). Decreases in effective stress (total stress minus pore-fluid pressure) can come about either through reduction in the burial load, say by erosion of some overlying material, or by increase in pore-fluid pressure, perhaps due to injection of pressured fluid from outside. The sediment enters the
HOW SEDIMENTS BECOME MOBILIZED
11
Fig. 2. Basic principles of sediment burial and deformation, relating porosity to changes in effective stress. Where drainage is sufficient to keep pace with increasing rises in total burial load, the sediment is normally consolidated, following a line such as A-D. If drainage is prohibited, the burial load will be taken up by the pore-fluid and the sediment will remain at point A, with fixed porosity despite its increasing burial depth. Line D-C represents a sedimentunloading path, with little porosity regain. Sediment at point C is overconsolidated, having experienced higher effective stresses. Increases in deviatoric stress (not shown on this two-dimensional diagram) will prompt deformation along a path dependent on the consolidation state. Overconsolidated material (e.g. at point C) will increase in volume during deformation and tend to shear along discrete surfaces until it achieves critical state (e.g. path C-B), whereas normally consolidated sediment loses volume and tends to deform pervasively (e.g. path A-B). However, all materials will reach the critical state, where they retain strength but are capable of large shear displacements.
realm of overconsolidation, occupied by materials that once experienced greater effective stresses. Note that late overpressuring, such as might arise from the injection of externally sourced fluids, gives a different porosity value from that due to drainage being curbed from the start. The implication of this is that the absence in a depth profile of marked porosity anomalies does not signify the absence of overpressured horizons. Wherever overpressuring arises during burial, the sediment is weakened and the possibility arises of it undergoing mobilization.
Critical state deformation The overpressuring does not have to be sufficient for complete liquidization for the sediment to be involved in mobilization. Much sediment is likely to retain some residual strength during its movement, through grain cohesion or incomplete sealing of fluids and in any case liquidized sediment may dissipate pore-fluid during displacement and regain some frictional strength. In these situations, deformation at the sediment's critical state is important, as at that
12
A. MALTMAN & A. BOLTON
Fig. 3. Variation of permeability with mean effective stress for various anisotropic sediments. The permeability varies little with effective stress except below about 200 kPa, where the abrupt increases in permeability seen in fabric-parallel flow result from efficient dilation and connectivity of pores along the fabric to give a 'fracture permeability' (see Bolton et al. 2000). Apart from the sheared sand-kaolinite, which was generated in the laboratory, all the sediments were taken from Ocean Drilling Program cores, the fissile silty clay being from the Woodlark Basin.
particular combination of porosity, fluid-pressure and stress the sediment can undergo very large amounts of shear (e.g. Jones 1994), even at low deforming stresses. The sediment behaves overall in a weak, ductile manner rather than as a fluid; its actual response will depend on factors such as consolidation state (e.g. Yassk 1990). Most overconsolidated sediments have to dilate at first, and deform along discrete shear surfaces (e.g. path C-B on Fig. 2), of which the scaly fabrics commonly described from clay diapirs and (noncataclastic) deformation bands in sand dykes are manifestations. Sediments in other consolidation states undergo porosity collapse and a more pervasive style of deformation (e.g. path A-B on Fig. 2). Irrespective of the starting conditions, however, the deformation drives the sediment towards the critical state condition, represented as a line in Figure 2. The concept of a critical state only strictly applies to
homogeneous, perfectly plastic materials and in practice, of course, some fluctuation in the parameters or geometrical constraint will limit the strain achieved. However, because many sediments reach a condition of weak residual strength more easily at high porosities and low effective pressures, a situation in which relatively low deviatoric stresses drive large shear displacements, the idea is very relevant to much sediment mobilization. Permeability The extent to which pore-fluid is dissipated in response to applied loads depends on the permeability of the sediment. Permeability is the capacity of a material to transmit fluid and is a crucial influence on the extent to which overpressures develop in a sediment. In general, high permeability sediments
HOW SEDIMENTS BECOME MOBILIZED
tend to become mobilized by overpressuring due to externally-pressured fluids entering and rapidly passing through the system, whereas lower permeability materials are more likely to be weakened by overpressures in trapped fluids. In order for a sediment to undergo critical state deformation, the permeability has to be sufficient to allow the appropriate porosity changes and variations in effective stress conditions to be reached. Permeability is likely to differ from bed to bed in a sedimentary sequence, but is not fixed for any particular lithology. Active deformation affects the permeability behaviour (e.g. Stephenson et al. 1994), as do any fabrics imparted to the sediment by consolidation or by earlier deformation, especially where the effective confining pressures are low (e.g. Brown & Moore 1993; Clennell 1997). Recent laboratory work (e.g. Bolton et al 1999) has emphasized how variable permeability can be in anisotropic sediments at different effective pressures (Fig. 3). In other words, although sedimentary sequences are often assigned a single representative permeability value, in fact the drainage will vary intricately throughout - and through time, especially where burial or deformation fabrics are involved. From time to time at various levels in a sediment pile, there may well be materials with inadequate drainage, which therefore become overpressured and hence prone to mobilization.
13
Fig. 4. Fluid pressure variations with depth in the Barbados accretionary prism, estimated from laboratory consolidation tests and well-log data. From Moore et al. (1995).
Complexity of relationships For the reasons outlined above, simple relationships between burial loading, porosity and fluid pressures are unlikely in a consolidating sedimentary sequence. At their simplest, porosity gradients will be exponential and fluid pressures linear (e.g. Bahr et al 2001), but even this is unlikely (Maltman 1994). A smooth gradient of fluid pressure is only possible where the sediment is able to drain continuously to maintain equilibrium conditions. If it is not, the lithostatic gradient will also be affected, as the sediment density will be reduced to less than normal. Moreover the rate of burial-loading due to sedimentation may well vary temporally, temperature effects become increasingly relevant with depth (e.g. Graham et al 2001) and various other processes discussed below under loading - may arise. Such complexities operate differently in different lithologies, so that in a layered sequence there may well be numerous pressure anomalies corresponding to different sedimentary horizons. As an illustration, Figure 4 shows fluid pressure variations with depth offshore Barbados. The values are indirect estimates rather than actual measurements, but show well the intricacies that probably arise in nature, here in a fairly homogeneous
sequence of mudstones, though with the complicating effects of marked variations in smectite content (Brown & Ransom 1996). In this active tectonic region, deformation is adding a further complication to the consolidation history of the sediment. Figure 5 also illustrates fluid pressure heterogeneity at Barbados, but in an areal perspective. Note the marked lateral variations in fluid pressure (interpreted from differences in seismic polarity), though perhaps the degree of heterogeneity shown by this active low-angle fault may be greater than that expected in a subsiding sedimentary layer. Evidence from rocks (e.g. Vannucchi & Maltman 2000) also reveals the complex fluctuations in fluid pressure that may affect sediments early in their burial history. In summary, while the consolidation of a sediment tends to progressively strengthen it, in a sequence of bedded lithologies the fluid pressures are likely to be evolving through time in a complex way. There may well be numerous situations at different times, at different depths, in different places where a sediment is vulnerable to enhanced or reduced drainage and hence anomalously high fluid pressures. At such times, the sediment is reduced in strength, even to the extent that it behaves like a fluid
14
A. MALTMAN & A. BOLTON
Fig. 5. Lateral fluid pressure variations on the basal decollement of the Barbados accretionary prism inferred from seismic amplitude. The dark areas indicate negative seismic polarity, thought to represent areas of high pore-fluid pressure. Based on Shipley et al (1996).
and consequently becomes available for wholesale movement.
Anomalous sediment loading As outlined above, weakening of a sediment usually comes about through the addition of some load with which pore-fluid dissipation cannot keep pace. Lithostatic loading naturally arises as sediments accumulate, but a host of additional loads are possible (Fig. 6). Most mobilization comes about through combinations and mutual interactions of several effects rather than a single loading process.
Dynamic loading A mechanical load that arises from a source external to the sediment is called dynamic loading. Such a load tends to be transmitted rapidly through the sediment and, depending on the stiffness of the sediment framework, partitioned into the pore fluids. Being virtually incompressible, the fluids typically sustain most of the excess load. The most obvious example of dynamic loading is lithostatic, the suprajacent sedimentation summarised in the preceding section. However, this varies from the slow, steady deposition of fines through to the instantaneous emplacement of major allochthonous masses. The latter process can be tectonically driven rather than sedimentary and although the rates will be lower, emplacement of nappes and thrust sheets is thought to be important for triggering overpressures in active
orogenic belts (Yassir, pers. comm.). Moreover, tectonically driven shear in conditions of reduced drainage can induce over-pressuring (Yassir 1990). Other examples of dynamic loading in addition to continuing sedimentation and the effects of tectonism are the introduction of igneous material above or within the sediment sequence (e.g. Elsworth & Day 1999) and for near-surface sediments, a whole host of surficial processes ranging from footsteps (Lewis & Titheridge 1978) to meteorite impact (Read 1988). Each of these processes can be regarded as an individual event - although some are clearly repetitive - and hence such discrete loading is described as monotonic. The loads can be very large and sediments may be instantaneously strained. Poorly permeable sediments may generate pore pressures sufficiently large to prompt hydrofracture, which will then allow rapid drainage, at least temporarily. Monotonic loads are therefore unlikely to generate over-pressuring for long periods. Two external sources of additional loading that are relatively long-lived and therefore not strictly monotonic are artesian conditions (e.g. Massari et al. 2001) and the application of tectonic stresses. The first pressurizes the pore-fluid directly whereas the second, in addition to tending to deform the sediment, will be partitioned into the fluid according to the drainage conditions. McPherson & Garven (1999) argued that tectonic loading is the primary reason for overpressuring in the Sacramento Basin (although not in this case leading to mobilization) and G. Westbrook (pers. comm.) invoked tectonic loading to explain mud volcanism offshore Barbados. In both these examples, the loading is ulti-
HOW SEDIMENTS BECOME MOBILIZED
15
Fig. 6. Conceptual representation of the range of loading processes that may affect sediments in addition to progressive burial due to continued sedimentation. Note that the representation is neither comprehensive nor a rigorous classification.
mately the result of compressive stresses due to plate motion. Repeated stressing of a sediment, arising from the oscillatory passage of waves, for example, is known as cyclic loading (e.g. Grozic et al 2000; Pestana et al 2000). Here also the duration of the effects is likely to be short, geologically speaking. The catastrophically weakening effects of seismic waves, especially shear waves (Youd et al 2001), are wellknown. Subaqueous sediments buried to a few hundred metres can experience pressure pulses from waves travelling at the sea surface (Seed & Rahman 1978), in association with winds and tides (Wang & Davis 1996). The stress rise from each of these pulses may well be much less than in a monotonic event, but if each fluid pressure response is incompletely dissipated before the next pulse, the incremental accumulation of overpressure can be large. Experimental work by Sassa & Sekiguchi (2001) demonstrated that progressive wave loading is more effective in liquefying sands than loading from standing waves. A variant on this behaviour, called cyclic mobility and arising where the periodic loads oscillate in nature between compression and tension, can induce strength loss even in relatively dense sediments (Castro 1987). Wang & Davis (1996) showed how the effects of tidal cyclic loading depended on the permeability of the sediment and the strength moduli/compressibility of the grain framework and pore fluid (see also Bachrach et al. 2001), the latter being very sensitive to the presence of gas (Wang et al. 1998). Lithostatic and even super-lithostatic loads are possible if the sediment lacks stiffness completely, at least in some layers within a sedimentary sequence (e.g. Zhao et al. 1998).
Static loading Long-lived overpressures are generally more likely to be due to processes that operate within an evolving sediment - collectively referred to as static loading - rather than those imposed externally. Static loading tends to act directly on the pore-fluids, reducing the frictional strength of the sediment by an equivalent amount (e.g. Cobbold & Castro 1999). The sediment strength depends on any remaining frictional resistance and the amount of inter-grain bonding (cohesion) that diagenesis has imparted. The common processes of static loading are likely to become more important at greater depths, as lithification proceeds. The processes include aquathermal pressuring (Luo & Vasseur 1993) and hydrocarbon generation and maturation, including the expansion of rising gases such as methane (Osborne & Swarbrick 1997). Mineral dehydration, and especially the smectite-illite transformation, may also become important. The smectite content of clayey sediments can exceed 50%, so that large volumes of water, which can comprise up to 25% of the mineral, is made available when it dehydrates to illite. Although the transition is usually taken to occur when temperatures during burial reach around 60-80°C, Fitts & Brown (1999) have argued that stress can trigger the reaction, even where it is as low 1.3 MPa. Moreover, growing packets of illite progressively coalesce and restrict dewatering, further enhancing the overpressuring effect (Freed & Peacor 1989). The role of gas-hydrates is currently unclear. Their prevalence is increasingly being recognized, together with the enormity of the gas and water they
16
A. MALTMAN & A. BOLTON
store and trap, which can be released if the stability field changes, say in response to sea-level change. Kennett & Fackler-Adams (2000) argued that dissociation as the hydrates respond to even small T-P changes would prompt widespread sediment deformation, even hundreds of metres below the sea floor. Such gas-hydrate driven overpressuring was invoked by Cherkis et al. (1999) to account for the strength reduction and consequent sediment instability offshore Spitzbergen, though such effects are disputed (e.g. Bouriak etal 2000). Hydraulic loading of sediments ranges from effects that are subtle but widespread, to more localised but dramatic processes associated with vigorous fluid movements. Any excess hydraulic head will drive pore-fluid to move and the resulting osmotic and capillary forces will tend to weaken the intergrain friction. Hydrocarbons in the pore-fluid will promote buoyancy of the grains. Such effects may be common, although Osborne & Swarbrick (1997) calculated that the effects would be small. Where the permeability allows rapid movement of the pore fluids, the drag force exerted on the sediment particles, termed the seepage force, can further add to the pore-fluid pressure and even exceed the weight of the grains. In this situation, the particles are buoyant and are readily entrained by the moving fluid, which may be a mixture of liquids and gases. This is the basis of fluidization, discussed in the following section. The relative importance of the loading sources outlined above to sediment mobilization is still much debated and no doubt varies between different geological situations (e.g. Hall 1994). For example, Osborne & Swarbrick (1997) considered that stressrelated mechanisms are the most likely cause of overpressuring in many sedimentary basins whereas theoretical calculations by Wangen (2001) suggested that they chiefly result, at least in more deeply buried sandy sediments, from cementation of the pore-spaces. Current work (e.g. Lonergan et al. 2000) suggests that fluids entering the sediment system from outside are increasingly being seen as a trigger for sediment mobilization, at least in hydrocarbon provinces.
Liquidization
neously destabilised: they are thixotropic. For example, sediments with unusually high pore-water contents, perhaps through having a honeycomb or 'house-of-cards' arrangement, collapse easily. Normally, disturbance of such sediments leads to a more stable arrangement on recovery: the remoulded sediment gains in strength. However, reflocculation of the particles may generate a framework that is weaker than before. The situation can arise, for example, through slight changes in porewater chemistry and where chemical leaching has reduced grain cohesion, so that the restored framework is less rigid (Torrance 1999). Such sediments are termed 'sensitive' (Torrance 1983) and where the ratio of undisturbed to disturbed strength is large, the sediment is regarded as 'quick'. Torrance (1983) reviews schemes for quantitively defining the quick condition and the relevant mechanisms. The structures produced by sensitive sediments appear indistinguishable from those formed in sediments liquidized by other mechanisms, though with the high porosities that are normally required they are only likely to form in situations of shallow burial.
Liquefaction Where any additional load on a sediment is wholly sustained by the pore-fluid and cohesion is negligible then the sediment loses strength completely and effectively behaves as a fluid - a state known as liquefaction. Such a situation can readily come about in relatively near surface sediments, where they are susceptible to the processes of particularly rapid loading and diagenesis may not have proceeded far. The sediment will remain in this condition until the pore pressure is reduced and some inter-particle frictional strength is restored. Volumetrically, most sediment mobilization appears to be ascribed to liquefaction, although recent work is emphasising the importance of fluidization (see below). Vaid & Sivathayalan (2000) summarized the kinds of variables that affect the susceptibility of sands to liquefaction, which includes the fabric adopted by the grains following previous disturbance. Oda et al. (2001) argued that reductions in resistance to repeated liquefaction events are due to increased void connectivity, which promotes sensitivity to future stresses.
Sensitive sediments A sediment becomes mobilized because it is in a condition of insufficient strength to resist the forces driving it to move. Usually the weakening is temporary and is related to fluid pressures, as outlined above, but there are other processes (e.g. see Owen 1987). Some sediments are intrinsically vulnerable to abrupt, albeit slight, loading and become instanta-
Fluidization Where the sediment strength is lost through moving interstitial fluids buoying the particles, the state is called fluidization. The fluid-drag force balances or exceeds the particle weight, a situation normally requiring rapid ingress of external fluids and usually
HOW SEDIMENTS BECOME MOBILIZED
in subsurface sediments operating in an upward direction. Sands and other coarse, permeable sediments are probably more vulnerable to these effects of rapid pore-fluid flow than low permeability sediments such as clays. Jolly & Lonergan (in press) discuss how theory predicts that sands with wellsorted, rounded grains should fluidize most easily, although in some natural examples the coarsest sediments have preferentially mobilized. Hovland & Judd (1988) have discussed the movement of gas in sediments and Nicholls et al (1994) discussed how in fluidized near-surface sediments the form of the resulting structures depends on factors such as permeability and strength of the sediment, and the rate of the pore-fluid movement. There is a vast amount of civil engineering literature on liquefaction and its mitigation (e.g. Seed et al 2001) and on fluidization and its engineering applications (e.g. Yang 1999).
Sediment mobilization Liquefaction and fluidization relate to the state of the sediment and in themselves are not sufficient to give bulk movement of the sediment. However, such liquidized sediments are behaving, by definition, as a fluid and are, therefore, subject to fluid pressure gradients. The elevation head, as in a fluid at rest, no longer balances out the pressure head in overpressured fluid and so the fluid will attempt to move to lower values of the potential gradient. For this to occur, the hydraulic gradient must also be able to trigger movement of the weakened sediments and the gradient has to be sustained long enough to allow the movement to be accomplished. For example, rupture of a low permeability layer that enabled liquefaction of the underlying, sealed sediments may suddenly allow their escape along the pressure gradient. The sediments can then only move as long as they are sufficiently overpressured to still be liquefied. Fluidized sediments require vigorous fluid movements but sediment can only be displaced along with the moving fluid as long as the pressure gradient is sustained. In other words, mobilization processes tend to be self-terminating. The pore fluid in a fluidized sediment is similarly impelled in the direction of lower fluid potential, but in this case the sediment particles are moving by entrainment. In both the fluidization and liquefaction cases, the fluid gradient will be upwards overall (Hovland & Judd 1989), but can locally be in any direction, including downwards (e.g. Nicholls 1995). This is why, in addition to any constraints presented by the host material, mobilized sediments do not necessarily intrude upwards. Dasgupta (1998), for example, reported liquefaction structures recumbent in orientation, presumably due to locally
17
horizontal hydraulic gradients and Huang (1988) described clastic sediments intruding downwards. In recent years, the injection of externally pressurized liquids and gases into high permeability sediments, reducing the strength of the aggregate and driving rapid fluid movements that buoy the sediment particles, has been invoked for the mobilization of sands in the North Sea (e.g. Dixon et al 1995; van Balen & Skar, 2000). Such fluidization will be very largely directed upwards and may be occurring in hydrocarbon provinces on a scale not fully realized, with major implications for reservoir geometries (e.g. Lonergan et al 2000; Jolly & Lonergan in press). Sands appear to be most amenable to fluidization, especially where porosities are high. Coarser sediments normally have greater porosities and permeabilities, but these factors seem to be offset by the greater flow velocity need to buoy the heavier clasts. In addition, it would seem likely that larger apertures would be needed to allow coarser sediments to move. Jolly & Lonergan (in press) have recently reviewed the factors that determine the structures that result from sand mobilization. In general, clays are too impermeable to allow the bulk fluid movement necessary for fluidization and their greater cohesion will resist disaggregation. Both sands and clays can liquefy at shallow burial depths, where loading processes easily exceed the sediment strength, and mobilization is facilitated by the steep hydraulic gradients resulting from the nearby free surface. However, porosity loss with burial soon increases the frictional resistance of sand and progressively improves its resistance to liquefaction. Hence, at depth relatively steeper potential gradients are required to mobilize sand. This may be one reason why sand reaches surface to be extruded as volcanoes less commonly than clays. An additional driving effect can operate. Sediments that are weakened or liquidized may well be less dense and less viscous than the overlying material and the resulting reversed gradients will be highly unstable. Because gravitational potential results from the product of height above datum and mass, the overlying layer has a higher potential energy. It will tend to sink while the buoyant liquefied layer mobilizes upwards, in an attempt to produce the more stable, greater-density-withdepth, configuration. Any perturbations at the interface between the overlying, denser material and the underlying less dense sediment will, where both are behaving as fluids, act as Rayleigh-Taylor instabilities, causing the irregularities to amplify until the gravity-driven overturn can be achieved (e.g. Ronnlund 1989; Harrison 1996). Thixotropic behaviour and Rayleigh-Taylor instabilities in layered sediments are probably confined to shallow levels of burial and the resulting structures to sizes of no more than tens of metres or
18
A. MALTMAN & A. BOLTON
less. Liquefaction and fluidization structures occur at these scales also, but range up to the kilometre sizes now being documented on seismic sections. It is important to note that most sediment mobilization, certainly at the larger scale, probably involves more than one mechanism. Liquefaction and fluidization probably often work together, sediments with sensitive fabrics may be involved, while associated material may retain significant residual shear strength and be deforming at or near its critical state. Diapiric melanges provide a good example of material in different mechanical states being intimately related (e.g. Barber et al. 1986). Clay diapirs are essentially driven by excessive fluid pressures, which also act to liquidize much of the sediment. Liquid mudflows result if there is extrusion. Other parts of the intruding diapir may be shearing under critical state conditions, forming the scaly clays typical of such material (e.g. Vannucchi et al. in press). Whether the sediment moves as a fluid, through liquidization, or as a ductile solid through critical state deformation, will depend on factors such as lithology and the physical conditions, which will vary with time and position in the diapir. For example, material at the diapir margin or near to a potential conduit will drain more efficiently and retain/regain some residual strength. Orange (1990) has described how blocks in a diapiric melange remain solid during intrusion, and Brown & Orange (1993) have documented how various mechanisms operated at different parts of a diapiric melange and how they varied through time. Critical state deformation probably increases in importance with increasing depth, as progressive lithification reduces the ability of the sediment to liquidize and begins to promote cataclasis. Jones (1994) explained how the concept relates to bonded sediments such as shales and the initiation of shale diapirs.
Conclusion Irrespective of their size, mobilization structures in incompletely lithified sediments owe their origin to the same basic cause: temporary weakening of the material as it progresses towards becoming rock. The above review has summarized how the sediments - which in contrast to rocks deform primarily by independent particulate flow - undergo such weakening. Some sediments are inherently prone to sudden weakening and are referred to as being sensitive, or in extreme cases, quick. Some sediments attain the stress/porosity combination that allows them to deform at their critical state where, typically in overpressured conditions and at low deviatoric stresses, the large shear strains associated with mobilization can be achieved even though the material still has strength.
Most sediment mobilization, however, comes about through the two main processes that are capable of liquidizing the sediment, making it behave as a fluid. Both involve overpressuring. These are: (1) liquefaction, where non-equilibrium additional loads are borne by the pore-fluid, so that friction between the cohesionless grains is effectively removed and the sediment loses its frictional strength; and (2) fluidization, where pore-fluids, commonly aided by injection of further liquid and/or gas from outside, can buoy the sediment particles. It is likely that both processes often work together to liquidize a sediment. The kinds of loads involved, additional to those due to normal sedimentation, range from monotonic (e.g. suprajacent emplacement of allochthonous masses) and cyclic (e.g. oscillatory passage of storm or seismic waves) dynamic loading to the results of internal processes such as mineralogical changes in the sediment. Because of the over-pressuring, the liquidized sediment is subject to a hydraulic gradient, normally but not necessarily upwards. However, only if the overpressuring seal is ruptured, say by hydraulic fracturing, while retaining some driving hydraulic gradient, can there be bulk movement. As long as the sediment remains liquidized or deforms at critical sate conditions, it is capable of undergoing very large displacements, even on the scale of kilometres now being reported from mobilized sediments. The authors acknowledge helpful discussions with G. Owen and D. Dewhurst and constructive refereeing comments from the latter and K. Brown.
References ALLEN, J.R.L. 1977. The possible mechanics of convolute lamination in graded sand beds. Journal of the Geological Society, 134,19-31. ALLEN, J.R.L. 1984. Sedimentary Structures: their Character and Physical Basis. Developments in Sedimentology, 30. Elsevier, Amsterdam, 663 p. BACHRACH, R., NUR, A. & AGNON, A. 2001. Liquefaction and dynamic poroelasticity in soft sediments. Journal of Geophysical Research, Solid Earth, 106, 13515-13526. BAHR, D.B., HUTTON, E.W.H., SYVITSKI, J.P.M. & PRATSON, L.F. 2001. Exponential approximations to compacted sediment porosity profiles. Computers & Geosciences, 27, 691-700. BARBER, A.J., TJOKROSAPOETRO, S. & CHARLTON, T.R. 1986. Mud volcanoes, shale diapirs, wrench faults, and melanges in accretionary complexes, Eastern Indonesia. American Association of Petroleum Geologists, Bulletin, 70, 1729-1741. BITZER, K. 1996. Modeling consolidation and fluid flow in sedimentary basins. Computers & Geosciences, 22, 467-478.
BOLTON, A.J., CLENNELL, M.B., & MALTMAN, AJ. 1999.
HOW SEDIMENTS BECOME MOBILIZED Nonlinear stress dependence of permeability: a mechanism for episodic fluid flow in accretionary wedges. Geology, 27,239-242. BOURIAK, S., VANNESTE, M. & SAOUTKINE, A. 2000. Inferred gas hydrates and clay diapirs near the Storegga Slide on the southern edge of the Voring Plateau, offshore Norway. Marine Geology, 163, 125-148. BROWN, K.M. & MOORE, J.C. 1993. Comment on 'Anisotropic Permeability and Tortuosity in Deformed Wet Sediments' by J. Arch and A. Maltman. Journal of Geophysical Research, 98,17,864-17,866. BROWN, K.M. & ORANGE, D.L. 1993. Structural aspects of diapiric melange emplacement: the Duck Creek Diapir. Journal of Structural Geology, 15, 831-847. BROWN, K.M. & RANSOM, B. 1996. Porosity corrections for smectite-rich sediments: Impact on studies of compaction, fluid generation, and tectonic history. Geology, 24, 843-846. CASTRO, G. 1987. On the behavior of soils during earthquakes: liquefaction. In: CAKMAK, A.S. (ed.), Soil Dynamics of Liquefaction, Elsevier, Amsterdam, 169-204. CHERKIS, N.Z., MAX, M.D., VOGT, PR., CRANE, K., MIDTHASSEL, A. & SuNDVOR, E. 1999. Large-scale mass wasting on the north Spitsbergen continental margin, Arctic Ocean. Geo-Marine Letters, 19, 131-142. CLENNELL, M.B. 1997. Tortuosity: a guide through the maze. In: LOVELL, M.A. & HARVEY, P.K. (eds) Developments in Petrophysics. Geological Society, London Special Publications, 122,299-344. COBBOLD, PR. & CASTRO, L. 1999. Fluid pressure and effective stress in sandbox models. Tectonophysics, 301,1-19. DASGUPTA, P. 1998. Recumbent flame structures in the Lower Gondwana rocks of the Jharia Basin, India - a plausible origin. Sedimentary Geology, 119,253-261. DlXON, R.J., SCHOFIELD, K., ANDERTON, R., REYNOLDS,
A.D., ALEXANDER, R.W.S., WILLIAMS, M.C. & DAVIES, K.G. 1995. Sandstone diapirism and clastic intrusion in the Tertiary submarine fans of the BruceBeryl Embayment, Quadrant 9, UKCS. In: HARTLEY, A.J. & PROSSER, D.J. (eds) Characterization of Deep Marine Clastic Systems. Geological Society London, Special Publications, 94,77-94. ELSWORTH D. & DAY, S.J. 1999 Flank collapse triggered by intrusion: the Canarian and Cape Verde Archipelagoes. Journal of Volcanology and Geothermal Research, 94, 323-340. FITTS T.G. & BROWN K.M. 1999. Stress-induced smectite dehydration: ramifications for patterns of freshening and fluid expulsion in the N Barbados accretionary wedge. Earth and Planetary Science Letters, 172, 179-197. FREED, R.L. & PEACOR, D.R. 1989. Geopresssured shale and sealing effect of smectite to illite transition. American Association of Petroleum Geologists, Bulletin, 73,1223-1232. GRAHAM, J., TANAKA, N., CRILLY, T. & ALFARO, M. 2001. Modified Cam-Clay modelling of temperature effects in clays. Canadian Geotechnical Journal, 38, 608-621. GROZIC J.L.H., ROBERTSON, P.K. & MORGENSTERN N.R.
19
2000. Cyclic liquefaction of loose gassy sand. Canadian Geotechnical Journal, 37, 843-856. HALL, PL. 1994. Physical and chemical aspects of the development of overpressuring in sedimentary environments. Clay Minerals, 29,425-437. HARRISON, P. 1996. Rheological and numerical modelling of reverse-density structures. Unpublished Ph.D. thesis, University of Wales, Aberystwyth, 294 p. HOVLAND, M & JUDD A.G. 1988. Seabed pockmarks and seepages: Impaction on geology, biology and the marine environment. Graham & Trotman, London, 293 p. HUANG, Q. 1988. Geometry and tectonic significance of Albian sedimentary dykes in the Sisteron area, SE France. Journal of Structural Geology, 48,453-462. JOLLY, R.J.H. & LONERGAN, L. in press. Mechanisms and controls on the formation of sand intrusions. Journal of the Geological Society. JONES, M.E. 1994. Mechanical principles of sediment deformation. In: MALTMAN, A. (ed.) The Geological Deformation of Sediments. Chapman & Hall, London, 37-71. KENNETT. J.P. & FACKLER-ADAMS, B.N. 2000. Relationship of clathrate instability to sediment deformation in the upper Neogene of California. Geology, 28,215-218. LEWIS, D.W. & TITHERIDGE, D.G. 1978. Small scale sedimentary structures resulting from foot impressions in dune sands. Journal of Sedimentary Petrology, 48, 835-837. LONERGAN, L., LEE, N., JOHNSON, H.D., JOLLY, R.J.H., & CARTWRIGHT, J.A. 2000. Remobilization and injection in deepwater despositional systems: implications for reservoir architecture and prediction. In: WIEMER, P., SLATT, R.M., COLEMAN, J., ROSEN, N.C., NELSON, H., BOUMA, A.H., STYZEN, M.J., & LAWRENCE, D.T. (eds) Deep-water Reservoirs of the world. GCSSEPM Foundation 20th Annual Bob K Perkins Research Conference, 515-532. Luo, X. R. & VASSEUR, G. 1993. Contributions of compaction and aquathermal pressuring to geopressure and the influence of environmental conditions - reply to discussion. American Association of Petroleum Geologists, Bulletin, 77, 2011-2014. MALTMAN, A.J. 1994. Introduction and overview. In: MALTMAN, A. (ed.) The Geological Deformation of Sediments. Chapman & Hall, London, 1-35. MASSARI, E, GHIBAUDO G., D'ALESSANDRO A., & DAVAUD E. 2001. Water-upwelling pipes and soft-sedimentdeformation structures in lower Pleistocene calcarenites (Salento, southern Italy). Geological Society of America Bulletin, 113, 545-560. MCPHERSON, BJ.O.L. & GARVEN, G. 1999. Hydrodynamics and overpressure: Mechanisms in the Sacramento basin, California. American Journal of Science, 299,429-466. MOORE, J.C. et al 1995. Abnormal fluid pressures and fault-zone dilation in the Barbados accretionary prism: evidence from logging while drilling. Geology, 23,605-608. NICHOLLS, R.J. 1995. The liquification and remobilization of sandy sediments. In: HARTLEY, A.J. & PROSSER, D.J. (eds) Characterization of Deep Marine Clastic Systems, Geological Society, London, Special Publications, 94,63-76.
20
A. MALTMAN & A. BOLTON
NICHOLLS, R.J., SPARKS, R.SJ. & WILSON, CJ.N. 1994. Experimental studies of the fluidization of layered sediments and formation of fluid escape structures. Sedimentology, 41,233-253. ODA, M., KAWAMOTO, K., SUZUKI, K., FUJIMORI, H., & SATO, M. 2001. Microstructural interpretation on reliquefaction of saturated granular soils under cyclic loading. Journal of Geotechnical and Geoenvironmental Engineering, 127,416-423. ORANGE, D.L. 1990. Criteria helpful in recognizing shearzone and diapiric melanges: Examples from the Hoh accretionary complex, Olympic Peninsula, Washington. Geological Society of America, Bulletin, 102,935-951. OSBORNE, MJ. & SWARBRICK, R.E. 1997. Mechanisms for generating overpressure in sedimentary basins: a reevaluation. American Association of Petroleum Geologists, Bulletin, 81,1023-1041. OWEN, G. 1987. Deformation processes in unconsolidated sands. In: JONES, M.E. & PRESTON, R.M.E (eds) Deformation of sediments and sedimentary rocks. Geological Society, London, Special Publications 29, 11-24. PESTANA, J.M., BISCONTIN, G., NADIM, E, & ANDERSEN, K. 2000. Modeling cyclic behavior of lightly overconsolidated clays in simple shear. Soil Dynamics and Earthquake Engineering, 19,501-519. READ, WE 1988. Deformation in soft Ordovician sediment produced by large masses of sandstone from a neaby, but as yet undiscovered, impact crater. In: 1st Annual Meeting of the Meteoritical Society, Lunar and Planetary Institute, Houston, TX, 665, p. D-3. RONNLUND, P. 1989. Viscosity ratio estimates from natural Rayleigh-Taylor instabilities. Terra Nova, 1, 344-348. SASSA, S. & SEKIGUCHI, H. 2001. Analysis of wave-induced liquefaction of sand beds. Geotechnique, 51, 115-126. SEED, R.B., CETIN, K.O., Moss, R.E.S., KAMMERER, A.M., Wu, J., PESTANA, J.M. & RIEMER, M.E 2001. Recent Advances in Soil Liquefaction Engineering and Seismic Site Response Evaluation. National Information Service for Earthquake Engineering, University of California, Berkeley, Paper SP-2. http://nisee.bcrkeley.edu/ SEED, R.B. & RAHMAN, M.S. 1978. Wave-induced pore pressure in relation to ocean floor stability of cohesionless soils. Marine Geotechnology, 3,123-150. SHIPLEY, T.H., MOORE, G.F., TOBIN, HJ. & MOORE, J.C. 1996. Synthesis of the Barbados decollement seismic reflection response from drilling-based geophysical observations and physical properties. In: SHIPLEY, T.H., OGAWA, Y. BLUM, P. (eds) Proceedings of the Ocean Drilling Program, Scientific Results, 156, 293-302. STEPHENSON, E.L., MALTMAN, A.J., & KNIPE, R.J. 1994. Dynamic permeability in deforming accretionary
prism sediments. In: PARNELL, J. (ed.) Geofluids 93, fluid evolution, migration, and interaction in rocks. Geological Society, London, Special Publications, 78, 113-125. STUMP, B.B. & FLEMINGS, P.B. 2000. Overpressure and fluid flow in dipping structures of the offshore Gulf of Mexico (EI330 field). Journal of Geochemical Exploration, 69-70, 23-28. TORRANCE, J.K. 1983. Towards a general model of quick clay development. Sedimentology, 30,547-555. TORRANCE, J.K. 1999. Physical, chemical and mineralogical influences on the rheology of remoulded lowactivity sensitive marine clay. Applied Clay Science, 14,199-223. VAID Y.P. & SIVATHAYALAN, S. 2000. Fundamental factors affecting liquefaction susceptibility of sands. Canadian Geotechnical Journal, 37,592—606. VAN BALEN, R.T & SKAR, T. 2000. The influence of faults and intraplate stresses on the overpressure evolution of the Halten Terrace, mid-Norwegian margin. Tectonophysics, 320,331-345. VANNUCCHI, P., MALTMAN, A. & CLENNELL, B. in press. On the nature of scaly fabric and scaly clay. Journal of Structural Geology. VANNUCCHI, P. & MALTMAN, A. 2000. Insights from the Northern Apennines, Italy, into shallow-level processes of mountain building. Journal of the Geological Society, 157, 105-120. WANG, K.L. & DAVIS, E.E. 1996. Theory for the propagation of tidally induced pore pressure variations in layered subseafloor formations. Journal of Geophysical Research, 101, 11483-11495. WANG, K.L., DAVIS, E.E. & VAN DER KAMP, G. 1998. Theory for the effects of free gas in subsea formations on tidal pore pressure variations and seafloor displacements. Journal of Geophysical Research, 103, 12339-12353. WANGEN, M. 2001. A quantitative comparison of some mechanisms generating overpressure in sedimentary basins. Tectonophysics, 334,211-234. YANG, W-C (ed.) 1999. Fluidization, Solids Handling, and Processing. William Andrew, Norwich, New York. YASSER, N. 1990. The behaviour of of overpressured clays and mudrocks. In: KNIPE, R.J. & RUTTER, E.H. (eds) Deformation Mechanisms, Rheology and Tectonics Geological Society, London, Special Publications, 54, 399^04. YOUD, T.L. et al. 2001. Liquefaction resistance of soils: summary report from the 1996 NCEER and 1998 NCEER/NSF workshops on evaluation of liquefaction resistance of soils. Journal of Geotechnical and Geoenvironmental Engineering, 127, 817-833. ZHAO, C.B., HOBBS B.E. & MUHLHAUS, H.B. 1998. Analysis of pore-fluid pressure gradient and effective vertical-stress gradient distribution in layered hydrodynamic systems. Geophysical Journal International, 134,519-526.
Load structures: gravity-driven sediment mobilization in the shallow subsurface GERAINT OWEN Department of Geography, University of Wales Swansea, Singleton Park, Swansea SA2 8PP UK (e-mail: [email protected]) Abstract: Load structures are a type of soft-sediment deformation structure comprising synforms (load casts and pseudonodules) and antiforms (flame structures and diapirs) at an interface. They form in response to unstable density contrasts (density loading) or lateral variations in load (uneven loading) when sediment becomes liquidized or otherwise loses strength. They are here classified into five varieties: simple and pendulous load casts, in which the upper (denser) layer is laterally continuous; and attached pseudonodules, detached pseudonodules and ball-and-pillow structure, in which discrete masses of the upper layer are separated by matrix. Conceptual models demonstrate that there are several possible modes of formation for each type of load structure. One interpretation of the variation of load structure morphology is as a deformation series representing varying degrees of deformation, controlled by the magnitude of the driving force and/or the duration of its effective action. An interpretation of the commonly observed pattern of wide load casts and narrow flame structures is presented in terms of their differential growth. Fluidization has an important influence on the development of load structures and their relationship to other products of sediment mobilization.
This paper considers the characteristics, formation and significance of load structures, a type of softsediment deformation structure characterized by folding of an interface. Load structures are common in clastic sediments and sedimentary rocks and record sediment mobilization in the shallow subsurface, although similar processes may also operate in the deeper subsurface. The development of load structures has implications for lateral and vertical permeability and the connectivity of sand bodies. Many authors have described load structures and a thorough overview is given by Allen (1982). They most commonly affect an interface separating sand or sandstone from underlying mud, mudstone or shale, but also occur in gravels, clastic limestones, pyroclastic rocks and even igneous intrusions (Wiebe & Collins 1998). In some cases there is little contrast in lithology across the interface. The interface is deformed into regularly spaced antiforms and synforms (Figs 1, 2). Amplitudes range from the order of a millimetre to several metres, and spacing from a few millimetres to several metres, these dimensions scaling with the layer thickness (Hindmarsh & Rijsdijk 2000). The synforms, which are commonly described as load casts, are usually relatively broad and rounded, whereas the antiforms, sometimes described as flame structures or diapirs, are commonly narrower and may terminate upwards in a point (flame structures), or bifurcate upwards (Fig. 1). Stratification is usually contorted within load casts and flame structures. The degree of contortion typically diminishes away from the interface, although it may be greatest in the cores of load casts and flame structures. In some cases the upper layer
forms discrete masses, known as pseudonodules, that may resemble diagenetically-formed concretions or the products of spheroidal weathering. The symmetry axes of load casts, flame structures and pseudonodules are commonly upright, but may be regularly or variably inclined (e.g. Dasgupta 1998; Moretti et al. 2001) and where exposed in plan view the structures range from domes and basins to linear ridges and troughs (e.g. Allen 1985). Despite a good general understanding of the processes by which load structures form (see Anketell et al 1970; Allen 1982; Rijsdijk 2001), their value in understanding sediment mobilization and rheology and the conditions affecting sediments shortly after deposition remains largely unrealized and there is a need for a comprehensive framework to which diverse studies of load structures, including field studies, experiments and theoretical analysis, can be related. The aims of this paper are: (a) to propose an objective scheme for classifying and naming load structures; (b) to consider the mechanisms by which load structures can form; (c) to highlight relationships between load structures and other products of sediment mobilization; and (d) to use load structures as a case study to show how a systematic approach can enhance the understanding of subsurface sediment mobilization.
Classification of load structures A bewildering array of names has been applied to load structures (see Potter & Pettijohn 1977; Allen 1982). The classification presented here retains
widely used names and develops descriptive criteria proposed by Allen (1982) and Owen (1987), to identify five varieties of load structures. Firstly, load structures are separated into load casts, in which the upper layer is laterally continuous, and pseudonoduies, in which separate masses are surrounded by matrix (Fig. 1). Load casts are divided into two categories: simple load casts become wider upwards, with intervening flame structures becoming narrower upwards (Fig. 2a, e, h); pendulous load casts are attached to the upper layer by a narrow neck, and their intervening flame structures widen upwards, terminating in a bulbous, rounded crest and producing a morphology of teardrops and diapirs (Fig. 2b). Three categories of pseudonoduies are recognized: attached pseudonoduies comprise a single row of pseudonoduies, the upper surfaces of which are level with the top of the surrounding matrix (Fig. 2i); detached pseudonoduies comprise a single row that is also overlain by matrix, whether or not the tops of the pseudonoduies are at a uniform level (Fig. 2d); and ball-and-pillow structure comprises verticallystacked pseudonoduies, such that a cross-section normal to bedding intersects more than one pseudonodule (Fig. 2f, g). Other criteria required for a full description of load structures include: the lithologies of the upper (or pseudonodule) and lower (or matrix) layers; the size (spacing and amplitude) of load casts, flame structures and pseudonoduies; any cross-sectional asymmetry; and the geometry in plan view.
Interpretation of load structures Before considering the mode of formation of each type of load structure, it is useful to review the processes by which soft-sediment deformation struc-
tures form. In general terms, the deformation of unconsolidated sediment requires a deformation mechanism, a driving force and, in many cases, a trigger (Owen 1987). Deformation will occur if the applied stress exceeds the sediment strength, which may be achieved through an increase in applied stress (Fig. 3a) or a temporary reduction in sediment strength (Fig. 3b). The latter mechanism was described by Allen (1977) as a change of state from solid-like to liquid-like and termed liquidization. There are several possible liquidization mechanisms, some of which are initiated by a triggering agent. As an example, seismic shaking (a trigger) may induce liquefaction (a liquidization mechanism) in loosely packed, saturated sands, rendering them liable to deformation by otherwise ineffective stresses (driving forces).
Deformation mechanisms The main processes by which sediment involved in load structures may undergo deformation are outlined below. Of these, liquefaction, fluidization, thixotropy and quick clay failure are varieties of liquidization. Liquefaction. Cohesionless granular sediments experience liquefaction when the pore fluid pressure equals the normal stress across solid contacts. The grain weight can be considered as being transferred temporarily to the pore fluid, the effective normal stress between solid contacts becomes zero and the sediment behaves as a fluid (see Allen 1982; Owen 1987; Maltman & Bolton 2003). The restoration of strength begins immediately from the base upwards and the duration of the h' queried state depends on the grain size (which affects the settling velocity) and
LOAD STRUCTURES
23
Fig. 2. Load structures in the field (a-g) and experiments (h, i; see Owen 1996a). Scale bars 10 cm, except e and f. (a) Simple load casts and flame structures in Silurian turbidites, Aberystwyth Grits, central Wales, (b) Simple and pendulous load casts in Ordovician water-lain tuffs, Ramsey Island, SW Wales, (c) Flame structures and mud diapir (arrow) in Carboniferous Bude Formation, Bude, Cornwall, (d) Detached pseudonodule (1), attached pseudonodules (2) and simple load casts (3) in deltaic Upper Carboniferous Coal Measures, Amroth, South Wales. (2) and (3) have formed from the loading of ripples, (e) Simple load cast formed from a large erosional scour, Namurian, Ragwen Point, South Wales. Scale bar 50 cm. (f) Ball-and-pillow horizon in deltaic Upper Carboniferous Coal Measures, Amroth, South Wales. Scale bar 50 cm. (g) Sand-in-sand ball-and-pillow structure, Precambrian Torridonian sandstones, Skye, Scotland, (h) Attached pseudonodule formed by uneven loading. A cone-shaped sand body with angle-of-repose internal lamination foundered into a cross-bedded sand substrate. The arrowed surface was originally the planar top of the substrate, (i) Attached pseudonodules formed by density loading of medium grained sand (B) into very fine-grained sand (A). Arrows point to sand volcano deposits that formed at the sand-water interface then foundered into the substrate.
24
G.OWEN
upwards and fluidization by hot gases is an important pyroclastic process (Wilson 1980). Selective fluidization occurs when only the finer grained fraction of a polymodal sediment becomes fluidized (Nichols et al. 1994; Moretti et al. 1999). Thixotropy. Some clay-rich, cohesive sediments with a loose particle packing may undergo a drastic loss of strength upon disturbance as inter-particle bonds are broken during shear, allowing the material to flow (Boswell 1961; Gillott 1968; Maltman 1994). Over time, the inter-particle bonds are restored and the sediment regains its initial strength. This process is analogous to liquefaction in sands.
Fig. 3. Stress-time conditions for sediment mobilization showing sediment strength (solid line) and applied stress (dashed line). Shaded area represents time during which sediment is mobilized and may deform, (a) Increase in applied stress; (b) decrease in sediment strength (h'quidization).
the layer thickness (Allen 1982). Those sediments most susceptible to liquefaction are water-saturated coarse silts to fine sands with a loose depositional packing (Obermeier 1996), although coarser grained sediments may also liquefy (e.g. Postma 1983). Liquefaction can be initiated by an increase in pore pressure, which may occur as a cyclic increase induced by seismic shaking or pressure fluctuations associated with water waves, or by a shock such as breaking waves, flood impact, or rapid sediment deposition (Allen 1982). Fluidization. Cohesionless granular sediments experience fluidization when the upward component of drag imparted by a fluid flowing through the sediment equals the downward-acting grain weight (Nichols et al 1994). The effective weight of the grains becomes zero and the sediment behaves as a fluid until the drag is reduced. As well as reducing the sediment strength, fluidization can drive deformation through the upward transport of particles (elutriation). The separation of grains involved in fluidization results in bed expansion, which lowers the sediment bulk density and may further contribute to deformation. Fluidization by water can affect sediments of a range of grain sizes from coarse silt
Collapse of quick clays. Quick clays are described as being highly sensitive, where sensitivity is defined as the ratio of strength before and after disturbance. Their behaviour is similar to thixotropic materials, involving a drastic loss of strength and a tendency to flow upon disturbance, but the interparticle bonds and original strength are not restored with ageing. Such sensitivity is thought to develop when, as a result of environmental changes, a loose depositional packing becomes out of equilibrium with the ambient conditions so that inter-particle bonds, once broken, cannot be restored (Gillott 1968; Torrance 1983). Plastic flow. Clay-rich, cohesive sediments exhibit plasticity, such that over a range of pore-water contents they will flow if their yield strength is exceeded. They may be stable when deposited, but are liable to flow if the applied shear stress increases to exceed the yield strength and will remain mobile until the shear stress is reduced or the water content decreases through drainage. Brittle deformation. Under conditions of rapid loading or high strain rates or during incomplete liquidization, both cohesive and cohesionless sediments may deform by brittle failure along discrete shear surfaces (Maltman 1987; Owen 1987). Sediment may become mobilized before it is fully liquidized, since any tendency towards liquidization reduces sediment strength, rendering applied shear stresses more effective. Several liquidization processes lead to a porosity reduction and a net decrease in sediment volume and the displacement of excess pore fluid may weaken overlying sediment, or cause localized fluidization. Deformation structures formed in association with the expulsion of pore fluid are known as fluid escape structures, or more specifically as water escape structures where the sediment is water-saturated (Lowe 1975). The recognition of criteria for identifying liquidization mechanisms from the characteristics of deformed sediments would benefit from further study.
LOAD STRUCTURES
Driving forces Density loading. The formation of most load structures is related to variations in gravitational potential energy. These variations commonly result from a gravitationally unstable vertical profile of bulk density where denser sediment overlies relatively less dense; such a system is known as a reverse density gradient, or b-a system (Anketell et al 1970). Factors that influence bulk density include mineral density (e.g. organic particles or volcanic ash fragments are less dense than quartz grains) and packing variations. Freshly deposited muds are often more loosely packed than sands and finer sands more loosely packed than coarser sands and gravels, although packing is also influenced by grain shape and depositional process (Allen 1982). Upon liquidization, such a system represents a Rayleigh-Taylor instability. Wave-like perturbations develop at the interface, either growing from pre-existing irregularities or forming along an initially flat interface (Anketell et al 1970). Deformation driven in this way can be termed density loading (Selker 1993). Uneven loading. An alternative driving force is provided by lateral variations in the distribution of sediment load. Deformation occurs if the substrate becomes liquidized and can no longer support the surface load, and will proceed whether or not there is a density contrast between the load and its substrate. Vertical variations in bulk density (density loading) and lateral variations in sediment load (uneven loading) can act together to drive deformation, an example being the deformation of sandy bedforms on a muddy substrate (Dzulyriski & Kotlarczyk 1962).
Conceptual models for the formation of load structures Conceptual models can be devised that illustrate the deformation pathways leading to the five forms of load structure. The models highlight criteria that are useful in identifying the deformation pathway from the characteristics of the deformation structure.
Simple load casts and flame structures The deformation of a planar interface between denser and underlying less dense sediment forms simple load casts and flame structures as a RayleighTaylor instability when the sediment becomes liquidized (Fig 4a). In theory, deformation could proceed until the positions of the layers were reversed (Anketell et al. 1970), but because of the limited duration of the liquidized state this complete
25
overturn is rarely, if ever, achieved. In some load structures, the base of the upper layer remains virtually undeformed, except where it is pierced by narrow flame structures (Figs 2c, 4b), possibly indicating that liquidization developed in the lower layer only (see Discussion and below). If the initial interface is not planar, an element of uneven loading adds to the density loading, and controls the positions of load casts and flame structures (Figs 2e, 4c; Dzulynski & Walton 1965). A similar combination of driving forces exists if there is relief on the surface of the upper layer, such as ripples or dunes (Figs. 2d, 4d; Dzulyriski & Walton 1965). Uneven loading can be driven by the surface relief of the upper layer even if this has the same density as the lower layer (Fig. 4e), although the collapse of surface relief may be accommodated within the upper layer if that is thick compared with the surface relief (Owen 1996a). Some examples of complexly deformed bedding have been interpreted as the foundering of sandy bedforms into a sandy substrate (Horowitz 1982; Owen 19960). Finally, a rippled surface may undergo load deformation while sedimentation continues (Fig. 4f), forming syn-depositional load structures characterized by complexly rotated internal stratification (Dzulyriski & Kotlarczyk 1962). In this case, the rapid accumulation of sediment onto a weak mud layer may trigger sediment mobilization by plastic flow within the mud.
Pendulous load casts Experimental evidence indicates that pendulous load casts can develop from the same starting conditions that form simple load casts, by a continuation of the deformation process (Fig. 4g; Kuenen 1958; Anketell et al 1970; Allen 1982). Some pendulous load coasts, however, may form directly from the loading of pre-existing relief features (e.g. Fig.4f).
Attached pseudonodules Attached pseudonodules can develop from pendulous load casts if the rising diapirs reach the top of the upper layer (Figs. 2i, 5a), as in experiments by Anketell et al (1969), Owen (19960) and Moretti et al (1999). If the top of the upper layer is not the sediment-water interface, the diapirs may puncture overlying beds to form clastic dykes. Attached pseudonodules can also form by the foundering of an unevenly distributed sediment cover, such as isolated ripples or dunes (Fig. 5b). This pathway can occur through uneven loading alone, forming pseudonodules in a matrix of a similar lithology
26
G.OWEN
Fig. 5. Deformation pathways for attached pseudonodules. Shading represents denser sediment. (a) Rising sediment reaches the top of the upper layer. (b) Uneven loading of surface irregularities combined with density loading, (c) Uneven loading of surface irregularities, (d) Post-depositional erosion of load casts.
Detached pseudonodules
Fig. 4. Deformation pathways for simple and pendulous load casts. Shading represents denser sediment, (a) Rayleigh-Taylor instability at planar interface upon liquidization of both layers, (b) Liquidization of lower layer only, (c) Loading of pre-existing irregularities. (d) Loading of surface relief - combination of density loading and uneven loading, (e) Loading of surface relief — uneven loading only, (f) Loading of surface relief as sedimentation occurs, (g) Further deformation of simple load casts forming pendulous load casts.
(Figs 2h, 5c). Finally, structures defined as attached pseudonodules could be formed by the post-depositional erosion of simple or pendulous load casts (Fig. 5d).
Several deformation pathways may give rise to detached pseudonodules (Fig. 6). Firstly, detached pseudonodules can develop from pendulous load casts if the sinking of the load casts is more vigorous than the upwelling of diapirs (Fig. 6a), as in Kuenen's (1958) experiments. A source layer of similar lithology to the pseudonodules would overlie a pseudonodule horizon formed in this way. Secondly, attached pseudonodules that formed as in Fig. 5a may become covered by sediment extruded from the crests of diapirs and re-distributed at the sediment-water interface, leaving the pseudonodules detached from the top of the bed (Figs. 2i, 6b; see Owen 1996a, fig. 7). In this case there would be no source layer above the pseudonodule horizon and lamination within the sediment overlying the pseudonodules may indicate its extruded origin. Thirdly, a combination of density loading and uneven loading (for example, sandy bedforms on a muddy substrate) may allow loaded bedforms to sink completely into their substrate as detached pseudonodules (Fig. 6c; Dzufyriski & Kotlarczyk 1962). Fourthly, the loading of a denser layer sandwiched between two beds of similar sediment would form detached pseudonodules if the rising diapirs pierced the denser sediment (Fig. 6d). Fifthly, structures defined as detached pseudonodules could be seen on certain sections through pendulous load
LOAD STRUCTURES
Fig. 6. Deformation pathways for detached pseudonodules. Shading represents denser sediment. (a) Pseudonodules detach from base of denser layer. (b) Expulsion of sediment covers attached pseudonodules. While the substrate remains liquidized, sand or mud volcanoes undergo uneven loading to form a uniform layer above the pseudonodules. (c) Sinking of surface relief under combination of density loading and uneven loading, (d) Loading of a single layer overlain and underlain by similar sediment, (e) Cross-section through pendulous load casts, (f) Isolated slump folds.
casts (Fig. 6e), although a connection with the 'source' layer might then be evident on adjacent structures. Finally, some isolated slump folds may be defined as detached pseudonodules (Fig. 6f); distinctive criteria in this case might be a tendency for anticlinal geometry in the pseudonodules, an irregular distribution of pseudonodules, and a consistent sense of asymmetry.
Ball-and-pillow structure Figure 7 illustrates pathways that lead to ball-andpillow structure. Firstly, ball-and-pillow structure may form by the repeated detachment of pseudonodules from the base of a source layer (Fig. 7a). For
27
Fig. 7. Deformation pathways for ball-and-pillow structure. Shading represents denser sediment. (a) Repeated detachment of pseudonodules from base of denser layer, (b) Lateral drift of pseudonodules as they sink, (c) Simultaneous loading of several denser layers. (d) Upward stoping of fluidized through non-fluidized sediment, (e) Isolated slump folds.
this mechanism to operate, the substrate would have to remain liquidized for a long time relative to the rate of deformation, which would be possible if the density contrast across the interface was very high, thus providing a strong driving force, or if the substrate was very thick or fine-grained and therefore needed a long time to regain strength. Secondly, ball-and-pillow structure may form through the lateral drift of pseudonodules as they sink. Lateral drift might occur, for example, as a result of eddies in the wake of other pseudonodules, as in the experiments of Kuenen (1958) (Fig. 7b). In this model, there might be a gap in the row of pseudonodules for each stacked pseudonodule. Thirdly, ball-and-pillow structure could form through the simultaneous liquidization and loading of several couplets of denser and less dense sediment, provided each loaded layer developed pseudonodules (Fig. 7c). This requires liquidization to encompass several beds, which might be possible if high sedimentation rates allowed lower layers to remain uncompacted and overpressured. If pseudonodules did not develop, the product might be layers loaded in crude harmony, as in deformed heavy mineral bands (Allen 1982). Fourthly, a mechanism of 'sedimentary stoping'
28
G.OWEN
has been proposed by Allen (1982) to explain an association of ball-and-pillow structure with dish structures (e.g. Cheel & Rust 1986). Fluidized sediment penetrates upwards along an anastomosing network of extended flame structures, leaving isolated masses of non-fluidized sediment as pseudonodules (Fig. 7d). Characteristic features of structures formed in this way include little deformation within the pseudonodules, an irregular base to the overlying bed and poorly defined upper surfaces to the pseudonodules (Fig. 2g). Similar structures have been described as 'crescentic laminae' (Selley et al 1963), 'synclinal convolute lamination' (Lowe 1975) and 'droplet swarms' (Owen 1995). Finally, as with detached pseudonodules, isolated slump folds may be preserved in a form that resembles ball-and-pillow structure (Fig. 7e). The external form of some pseudonodules in balland-pillow structure indicates that the pseudonodules have been contorted (Kuenen 1948). This contortion may be attributed to sinking masses encountering a barrier to their descent, such as nonliquidized sediment, or to shear stresses acting on the margins of a sinking mass as it falls, deforming it from a spherical ball to an ovoid pillow to a contorted shape (Moretti 1997). Such contortion may therefore characterize only deformation pathways that involve considerable fall through a substrate (e.g. Figs 7a,7b).
Discussion Extent of mobilization The characteristics of load structures reflect, in part, the extent to which liquidization and mobilization develop in a two-layer system. If only the upper layer becomes liquidized, no deformation of the interface can take place. If both layers liquidize, the interface can deform as a Rayleigh-Taylor instability (Hindmarsh & Rijsdijk 2000). If only the lower layer becomes liquidized, support is removed from the upper layer. This may be strong enough to resist deformation until the strength of the lower layer is restored, or it may founder. The upward displacement of excess pore water from the liquidized layer as it re-sediments may reduce the strength of the upper layer, allowing it to develop some ductile deformation (cf. Fig. 4b). Rijsdijk (2001) considered a limiting case where the strength of the lower layer was so low that the overlying sediment sank through it as individual particles. A special case must be considered where fluidization develops in the lower layer. Since fluidization involves grain separation and bed expansion, this will enhance the driving force through both a reduc-
tion in the bulk density of the fluidized bed and the potential transport of particles (elutriation) by the upward-directed fluid shear. Loading may occur at the interface between fluidized sediment and overlying, denser, non-fluidized or partially fluidized sediment, forming structures that could be very complex since the interface may change position during deformation. A water-filled cavity may develop at or close to the interface between fluidized and nonfluidized sediment (Allen 1982; Nichols etal 1994; Owen 1996<s; Moretti et al 1999). The build-up of pore pressure may cause the burst-out of pore fluid and sediment from the cavity, forming clastic dykes and sand volcanoes; structures in Proterozoic fluvial sandstones were interpreted in this way by Owen (1996Z?). Alternatively, deformation could develop as a Rayleigh-Taylor instability between the waterfilled cavity and the overlying sediment, forming load structures overlain by a relatively undeformed bed and underlain by sediment with evidence for fluidization. Dish and pillar structures have been interpreted in this way by Moretti et al. (1999) and deformed bedding described by Johnson (1977) has been interpreted in this way by Allen (1982). The overlap between structures formed by loading and fluidization has been analysed by Moretti et al. (1999), who showed that finer grained particles elutriated by selective fluidization in the underlying unit may be unable to pass through the pore spaces in the upper unit and may deform the interface between the two units to form a halo of finer particles at the interface.
Load structures as a deformation series: implications for sediment rheology Load structures driven by density contrasts can be considered as a deformation series that increases in complexity from simple and pendulous load casts to pseudonodules and ball-and-pillow structure (Fig. 8). The complexity, or extent of deformation, is a function of the magnitude of the driving force and the duration of its effective action. Structures further along the series imply either a stronger driving force (e.g. greater density contrast, or combination of density and uneven loading) or a longer duration of liquidization (e.g. a very thick lower layer that requires a long time to regain strength). The concept of load structures as a deformation series from reversed to normal density stratification has been considered by Anketell et al. (1970) and Rijsdijk (2001), although the deformation pathways considered above show that certain starting conditions may lead directly to pseudonodule structures. Despite the gradational links implied in Figure 8, there is a major change in the degree of mobilization between load casts and pseudonodules. While load
LOAD STRUCTURES
29
Fig. 8. Links between types of load structures, demonstrating the interpretation of load structures as a deformation series. From left to right, simple load casts can develop into pendulous load casts. These can develop into detached pseudonodules, attached pseudonodules or isolated mud diapirs. Attached pseudonodules can develop into detached pseudonodules, and detached pseudonodules into ball-and-pillow structure. See text for further discussion.
casts represent the mobilization of sediment within one layer, pseudonodules imply the transfer of sediment across layers. Pendulous load casts can develop into either attached pseudonodules, if the effects of the rise of the lower layer are more prominent, or detached pseudonodules if sinking of the upper layer dominates (Fig. 8). If diapiric rise is sufficiently vigorous, the necks of rising diapirs could close behind them, forming 'pseudonodules' surrounded by a matrix of denser sediment (Woidt 1978). Such structures are uncommon, although they have been described by Cave & Rushton (1995) and Rijsdijk (2001) and from experiments by Anketell et al (1970). The effects of sinking may be expected to dominate if the lower layer simply loses strength; for buoyancy effects to dominate, some additional process must enhance the buoyancy of the lower layer, such as fluidization in sands or the release of overpressured sediment. These qualitative considerations of the relative vigour of buoyancy versus sinking have been related to sediment rheology, particularly relative kinematic viscosity, in experiments (Anketell et al. 1970; Woidt 1978), theory (Ronnlund 1989; Hindmarsh & Rijsdijk 2000) and field analysis (Rijsdijk 2001). As has already been noted, load casts are commonly broad and rounded in contrast to narrow, pointed flame structures. Anketell et al (1970) attributed this morphology to a system in which the lower layer has a much higher kinematic viscosity than the upper (Fig. 9). Since kinematic viscosity is the ratio of dynamic viscosity to density and since density
loading requires that the lower layer has a lower density, this condition is satisfied for a range of values of relative dynamic viscosity. This includes both layers having similar values of dynamic viscosity, the lower layer having a higher value of dynamic viscosity and the lower layer having a slightly lower value of dynamic viscosity. According to Anketell et al (1970), sinusoidal load casts and flame structures develop if the kinematic viscosities of the two layers are similar - that is, if the dynamic viscosity of the lower layer is significantly less than that of the upper. If the lower layer has a lower kinematic viscosity (i.e. much lower dynamic viscosity than the upper layer), Anketell et al (1970) predicted a morphology of narrow load structures and broad flame structures, which is rarely seen in natural load structures (Fig. 9). However, many of the deformation pathways that lead to the development of load structures imply that the lower layer is more mobile than the upper layer. Therefore it might be expected to have a lower dynamic viscosity (e.g. a lower layer that becomes liquidized while the upper layer retains significant strength - Fig. 4b), yet these pathways lead to the typical morphology of narrow flames and broad loads. There are other problems with the role of relative viscosity inferred by Anketell et al. (1970). Firstly, evidence from structural geology indicates that narrow synforms (loads) and broad antiforms (flames) characterize the folding of an interface underlain by relatively more viscous material (Ramsay 1967, fig. 7-42), the opposite of the relationship proposed by
30
G.OWEN
Fig. 9. Relationship between load structure morphology and relative kinematic viscosity (k = 7j/p where 77 = dynamic viscosity and p = density) for layers of contrasting density according to Anketell et al. (1970).
Anketell et al (1970). Secondly, a theoretical analysis by Ronnlund (1989, p. 346) of structures illustrated by Johnson (1977, fig. 8b), comprising broad synforms (loads) and narrow antiforms (flames), concluded that the upper layer was more viscous, again at odds with the conclusions of Anketell et al. (1970). Finally, the theoretical analysis by Hindmarsh & Rijsdijk (2000) failed to produce multiple wavelengths (i.e. narrow flames and broad loads) with variations in viscosity ratio. These contradictions suggest that the influence of relative viscosity on load structure morphology requires further investigation and may not be the dominant control on the relative width of load casts and flame structures. A simpler alternative interpretation may relate to the relative amplification rates of load casts and flame structures. Figure 10 demonstrates that if antiforms grow at a faster rate than synforms, an initially sinusoidal perturbation will develop into a waveform that appears to comprise narrow antiforms and wide synforms. Along the median line in Figure lOb, antiforms are narrower, although along the inflection line, antiforms and synforms are of equal width. At least two factors can be expected to give rise to faster growth of flame structures than load casts. The first is the influence of layer thickness on the time taken for sediment strength to be restored after liquidization. Because liquidization of the lower layer is an essential condition for the development of loading, the time needed for the restoration of strength in the lower layer is the main control on the time available for deformation. This time depends in part on the thickness of the lower layer, especially where liquefaction is the deformation mechanism (Allen 1982). Therefore, once deformation of the interface begins and the lower layer becomes thicker beneath flame structures, it will remain mobilized for longer within flame structures than beneath load casts. The second factor is the possibility that fluidization may develop in the
rising flame structures. In the previous section it was seen that fluidization enhances the mobility of the sediment by adding to the relative density contrast and to the upward-directed stresses and so if fluidization develops it will further enhance the growth rate of the flame structures. In summary, the morphology of load structures may well be influenced by sediment rheology, but the relationship is far from clear and requires further investigation. Once an interface has begun to deform, contrasts in the growth rate of antiforms and synforms may provide a simpler explanation for the commonly observed morphology of broad load casts and narrower flame structures. Such differential amplification of antiforms and synforms can be explained partly by the development of fluidization within rising flame structures or diapirs, and partly by the dependence of the duration of the liquidized state on the thickness of the liquidized layer.
Load structures that lack lithological contrast Several explanations can be offered to explain the occurrence of load casts (or pseudonodules) that differ little in lithology from their substrate (or matrix). Firstly, if uneven loading drives deformation, it will proceed even in the absence of lithological contrast (e.g. Figs 4e, 5c). Secondly, density variations that drive deformation may be destroyed by the deformation process or during burial, examples being density contrasts related to differences in packing or fluidization. Structures formed in these ways overlap in morphology with convolute stratification and Allen (1977) developed a model for the formation of convolute lamination in normally graded beds through loading at density contrasts that arose during sedimentation following liquefaction. Convolute stratification seems to be diverse in its morphology and origin, and would benefit from a
LOAD STRUCTURES
31
geometry in plan view. Experimental results and theoretical analysis by Moretti et al (2001) have clearly demonstrated the relationship between the two-dimensional asymmetry of load structures and the orientation of lateral shearing, thereby clarifying contratictory experimental results reported by Anketell & Dzulynski (1968, fig. 4) and by Anketell et al (1970, fig. 9).
Triggers for load structure formation
The preceding discussions have demonstrated overlaps between load structures and other structures formed by subsurface sediment mobilization (Fig. 11). Some of the structures in Figure 11 are load structures, such as wrinkle marks, contorted heavy mineral laminae, some varieties of convolute stratification and some periglacial involutions (Harris et al. 2000), while others, such as clastic dykes, sand and mud volcanoes and dish and pillar structures, may form in association with load structures, particularly if fluidization is developed (Alfaro et al 1997). A third group, including slump sheets, may resemble load structures, although they form by different processes.
The presence of load structures in sedimentary successions has implications for the conditions that existed during and after deposition. The rapid accumulation of sediment may be sufficient to trigger the deformation of a weak or easily liquidized substrate, enabling load structures to develop during or immediately after sedimentation (Fig. 4f; Moretti et al 2001). In many cases, however, a triggering agent of external origin may have initiated liquidization. Seismic effects have commonly been invoked, generating liquidization through direct shock or the cyclic build-up of pore pressure (Ishihara 1993) and load structures have been used as indicators of palaeoseismicity (Sims 1975; Moretti etal 1999; Rossetti 1999), although Obermeier (1996) cautioned against using them for this purpose because their formation may be triggered by various agents. Wrinkle marks, for example, are found in a range of aseismic settings and their formation may be initiated by pore water seepage associated with tidal fluctuations (Allen 1985). Other potential triggers include the dynamic or cyclic effects of waves (Dalrymple 1979; Molina et al 1998), trapped gas (Stewart 1956; Grozic et al 2000; Pralle et al 2003), or pore pressure changes generated by the thawing of frozen ground, the release of overpressured sediment, or artesian movements (Holzer & Clark 1993; Li et al 1996; Harris et al 2000). Preliminary attempts have been made to identify criteria that allow the identification of triggering agents from the characteristics and context of preserved structures, although more work is needed to establish reliable criteria (Sims 1975; Seilacher 1984; Alfaro et al 1997; Molina et al, 1998; Rossetti 1999; Jones & Omoto 2000).
Effects of lateral stresses
Conclusions
The formation of load structures is dominated by vertical displacements, but their geometry can be modified by lateral stresses related to slopes, currents or hydraulic gradients affecting the flow of pore water (Allen 1985; Dasgupta 1998; Moretti et al 2001). The effects include consistently inclined axes of symmetry in two-dimensions and linear
Load structures are characterized by folding of an interface between sediment layers. Deformation is driven by an unstable density gradient (density loading) and/or irregular loading of a substrate (uneven loading) when the system becomes mobilized through the failure of a weak layer or the loss of strength associated with liquidization. Load
Fig. 10. Development of narrow antiforms (flame structures) and wide synforms (load casts) through differential amplification rates of antiforms and synforms. (a) An initially sinusoidal perturbation (1) develops through stages 2-4 if antiforms grow faster than synforms. (b) Stage 4 from (a). Measured along a median line (Y-Y) synforms are wider than antiforms, although along the inflection line (X-X) they are equal in width.
similar analysis to that undertaken for load structures in the present paper.
Other structures indicative of sediment mobilization
32
G.OWEN
Fig. 11. Schematic relationships between load structures and other structures indicative of sediment mobilization. The thick dashed line encloses structures that form by loading processes. Shading represents denser sediment. Arrows indicate links between structures described in this paper and other structures.
structures affect sediment bulk permeability and their formation is analogous to some processes in the deeper subsurface. Load structures can be classified into load casts (simple and pendulous), in which the upper layer is continuous and pseudonodules (attached, detached and ball-and-pillow structure), in which it comprises separate masses. Conceptual models illustrate modes of formation for each type of load structure (deformation pathways) and criteria have been highlighted to differentiate the modes of origin. There is a progressive increase in the extent of deformation from simple load casts to ball-andpillow structure, although there are other interpretations for each of the structures in the series. Load structures can form when the lower layer or both layers lose strength and the development of fluidization has important consequences for their development and their association with other products of sediment mobilization. Differential growth rates of flame structures and load casts provide an alternative to viscosity contrasts as an explanation for the typical morphology of wide load casts and narrow
flame structures. Major gaps in present understanding include the implications of load structures for sediment rheology and the identification of the agents that triggered deformation. The ideas in this paper originated from the author's PhD research many years ago, supervised by J.R.L. Allen. Subsequent discussions with K. Rijsdijk, D. McCarroll and M. Moretti and the encouragement of A. Maltman are gratefully acknowledged. A. Ratcliffe and A. Cutliffe prepared Figure 2. The original manuscript has been greatly improved by the review comments of P. Harrison, W. McCaffrey and M. Moretti.
References ALFARO, P., MORETTI, M. & SORIA, J.M. 1997. Soft-sediment deformation structures induced by earthquakes (seismites) in Pliocene lacustrine deposits (GuadixBaza Basin, Central Betic Cordillera). Eclogae Geologicae Helvetiae, 90,531-540. ALLEN, J.R.L. 1977. The possible mechanics of convolute lamination in graded sand beds. Journal of the Geological Society, 134,19-31.
LOAD STRUCTURES ALLEN, J.R.L. 1982. Sedimentary Structures: Their Character and Physical Basis. Developments in Sedimentology, 30. Elsevier, Amsterdam. ALLEN, J.R.L. 1985. Wrinkle marks: an intertidal sedimentary structure due to aseismic soft-sediment loading. Sedimentary Geology, 41,75-95. ANKETELL, J.M. & DZULYNSKI, S. 1968. Transverse deformational patterns in unstable sediments. Annales de la Societe Geologique de Pologne, 38,411-416. ANKETELL, J.M., CEGLA, J. & DZULYNSKI, S. 1969. Unconformable surfaces formed in the absence of current erosion. Geologica Romana, 8,41-46. ANKETELL, J.M., CEGLA, J. & DZULYNSKI, S. 1970. On the deformational structures in systems with reversed density gradients. Annales de la Societe Geologique de Pologne, 40, 3-30. BOSWELL, RG.H. 1961. Muddy Sediments. Heffer, Cambridge. CAVE, R. & RUSHTON, A.W.A. 1995. A natural example of mud diapirs formed in wet sediments. Geological Journal, 30,183-188. CHEEL, R.J. & RUST, B.R. 1986. A sequence of soft-sediment deformation (dewatering) structures in Late Quaternary subaqueous outwash near Ottawa, Canada. Sedimentary Geology, 47,77-93. DALRYMPLE, R.W. 1979. Wave-induced liquefaction: a modern example from the Bay of Fundy. Sedimentology, 26, 835-844. DASGUPTA, P. 1998. Recumbent flame structures in the Lower Gondwana rocks of the Jharia Basin, India - a plausible origin. Sedimentary Geology, 119,253-261. DZULYNSKI, S. & KOTLARCZYK, J. 1962. On load-casted ripples. Annales de la Societe Geologique de Pologne, 32,147-159. DZULYNSKI, S. & WALTON, E.K. 1965. Sedimentary Features ofFlysch and Greywackes. Developments in Sedimentology, 7. Elsevier, Amsterdam. GILLOTT, I.E. 1968. Clay in Engineering Geology. Elsevier, Amsterdam. GROZIC, J.L.H., ROBERTSON, RK. & MORGENSTERN, N.R. 2000. Cyclic liquefaction of loose gassy sand. Canadian Geotechnical Journal, 37, 843-856. HARRIS, C., MURTON, J. & DAVIES, M.C.R. 2000. Softsediment deformation during thawing of ice-rich frozen soils: results of scaled centrifuge modelling experiments. Sedimentology, 47, 687-700. HINDMARSH, R.C.A. & RIJSDIJK, K.F. 2000. Use of a viscous model of till rheology to describe gravitational loading instabilities in glacial sediments. In: MALTMAN, A.J., HUBBARD, B. & HAMBREY, MJ. (eds) Deformation of Glacial Materials. Geological Society, London, Special Publications, 176,191-201. HOLZER, T.L. & CLARK, M.M. 1993. Sand boils without earthquakes. Geology, 21, 873-876. HOROWITZ, D.H. 1982. Geometry and origin of large-scale deformation structures in some ancient wind-blown sand deposits. Sedimentology, 29,155-180. ISHIHARA, K. 1993. Liquefaction and flow failure during earthquakes. Geotechnique, 43, 351-415. JOHNSON, H.D. 1977. Sedimentation and water escape structures in some late Precambrian shallow marine sandstones from Finnmark, North Norway. Sedimentology, 24,389-411. JONES, A.P. & OMOTO, K. 2000. Towards establishing crite-
33
ria for identifying trigger mechanisms for soft-sediment deformation: a case study of Late Pleistocene lacustrine sands and clays, Onikobe and Nakayamadaira Basins, northeastern Japan. Sedimentology, 47,1211-1226. KUENEN, PH. H. 1948. Slumping in the Carboniferous rocks of Pembrokeshire. Quarterly Journal of the Geological Society, 104, 365-385. KUENEN, PH. H. 1958. Experiments in geology. Transactions of the Geological Society of Glasgow, 23,1-28. Li, Y., CRAVEN, J., SCHWEIG, E.S. & OBERMEIER, S.F. 1996. Sand boils induced by the 1993 Mississippi River flood: Could they one day be misinterpreted as earthquake-induced liquefaction? Geology, 24,171-174. LOWE, D.R. 1975. Water escape structures in coarsegrained sediments. Sedimentology, 22,157-204. MALTMAN, AJ. 1987. Shear zones in argillaceous sediments - an experimental study. In: JONES, M.E. & PRESTON, R.F. (eds) Deformation of Sediments and Sedimentary Rocks. Geological Society, London, Special Publications, 29,77-87. MALTMAN, A.J. 1994. Introduction and overview. In: MALTMAN, A.J. (ed.) The Geological Deformation of Sediments. Chapman and Hall, London, 1-35. MALTMAN, A.J. & POOLTON, A. 2003. How sediments become mobilized. In: VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, A.J. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216,9-20. MOLINA, J.M., ALFARO, P., MORETTI, M. & SORIA, J.M. 1998. Soft-sediment deformation structures induced by cyclic stress of storm waves in tempestites (Miocene, Guadalquivir Basin, Spain). Terra Nova, 10,145-150. MORETTI, M. 1997. Le strutture sedimentarie deformative. Studio delle modalita di deformazione e dell'origine attraverso esempifossili e modellizzazione in laboratorio. PhD thesis, University of Bari. MORETTI, M., ALFARO, P., CASELLES, O. & CANAS, J.A. 1999. Modelling seismites with a digital shaking table. Tectonophysics, 304,369-383. MORETTI, M., SORIA, J.M., ALFARO, P. & WALSH, N. 2001. Asymmetrical soft-sediment deformation structures triggered by rapid sedimentation in turbiditic deposits (Late Miocene, Guadix Basin, southern Spain). Fades, 44,283-294. NICHOLS, R.J., SPARKS, R.S.J. & WILSON, C.J.N. 1994. Experimental studies of the fluidization of layered sediments and the formation of fluid escape structures. Sedimentology, 41,233-253. OBERMEIER, S.F. 1996. Use of liquefaction-induced features for paleoseismic analysis - an overview of how seismic liquefaction features can be distinguished from other features and how their regional distribution and properties of source sediment can be used to infer the location and strength of Holocene paleoearthquakes. Engineering Geology, 44,1-76. OWEN, G. 1987. Deformation processes in unconsolidated sands. In: JONES, M.E. & PRESTON, R.F. (eds) Deformation of Sediments and Sedimentary Rocks. Geological Society, London, Special Publications, 29, 11-24. OWEN, G. 1995. Soft-sediment deformation in Upper
34
G.OWEN
Proterozoic Torridonian Sandstones (Applecross Formation) at Torridon, northwest Scotland. Journal of Sedimentary Research, A65,495-504. OWEN, G. I996a. Experimental soft-sediment deformation structures formed by the liquefaction of unconsolidated sands and some ancient examples. Sedimentology, 43, 279-293. OWEN, G. 1996/7. Anatomy of a water-escape cusp in Upper Proterozoic Torridon Group sandstones, Scotland. Sedimentary Geology, 103,117-128. POSTMA, G. 1983. Water escape structures in the context of a depositional model of a mass flow dominated conglomeratic fan-delta (Abrioja Formation, Pliocene, Almeria Basin, SE Spain). Sedimentology, 30,91-103. POTTER, P.E. & PETTUOHN, FJ. 1977. Paleocurrents and Basin Analysis. Second edition. Springer-Verlag, Berlin. PRALLE, N., KULZER, M. & GUDEHUS, G. 2003. Experimental evidence on the role of gas in sediment liquefication and mud volcanism. In: VAN RENSBERGEN, P., HILLIS, R.R. MALTMAN, A.J. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216,159-189. RAMSAY, J. G. 1967. Folding and Fracturing of Rocks. New York, McGraw-Hill. RUSDIJK, K.F. 2001. Density-driven deformation structures in glacigenic consolidated diamicts: examples from Traeth y Mwnt, Cardiganshire, Wales, UK. Journal of Sedimentary Research, 71,122-135. RONNLUND, P. 1989. Viscosity ratio estimates from natural Rayleigh-Taylor instabilities. Terra Nova, 1, 344_348.
ROSSETTI, D.F. 1999. Soft-sediment deformation structures in late Albian to Cenomanian deposits, Sao Luis Basin, northern Brazil: evidence for palaeoseismicity. Sedimentology, 46, 1065-1081. SEILACHER, A. 1984. Sedimentary structures tentatively attributed to seismic events. Marine Geology, 55,1-12. SELKER, J. 1993. Expressions for the formation of load casts in soft sediment. Journal of Sedimentary Petrology, 63, 1149-1151. SELLEY, R.C., SHEARMAN, D.J., SUTTON, J. & WATSON, J. 1963. Some underwater disturbances in the Torridonian of Skye and Raasay. Geological Magazine, 100, 224-243. SIMS, J.D. 1975. Determining earthquake recurrence intervals from deformational structures in young lacustrine sediments. Tectonophysics,29, 141-152. STEWART, H.B. 1956. Contorted sediments in modern coastal lagoon explained by laboratory experiments. Bulletin of the American Association of Petroleum Geologists, 40, 153-161. TORRANCE, J.K. 1983. Towards a general model of quick clay development. Sedimentology, 30, 547-555. WIEBE, R.A. & COLLINS, W.J. 1998. Depositional features and stratigraphic sections in granitic plutons: implications for the emplacement and crystallization of granitic magmas. Journal of Structural Geology, 20, 1273-1289. WILSON, C.J.N. 1980. The role of fluidization in the emplacement of pyroclastic flows; an experimental approach. Journal of Volcanology and Geothermal Research, 8, 231-249. WOIDT, W.-D. 1978. Finite element calculations applied to salt dome analysis. Tectonophysics, 50,369-386.
Numerical modelling of reverse-density structures in soft nonNewtonian sediments PHILIP HARRISON1 & ALEX J. MALTMAN2 1
Department of Applied Mathematics, University of Wales, Aberystwyth, Wales SY23 3DB, UK 2 Institute of Geography and Earth Sciences, University of Wales, Aberystwyth, Wales SY23 3DB, UK Abstract: A numerical code has been used to simulate the flow patterns in geological soft sediments that are driven by buoyancy forces resulting from reverse-density stratification. The aim was to provide a clearer understanding of the different roles of initiating conditions, inertia and rheological behaviour on the morphologies and timing of formation of natural features such as load casts and flame structures. Particular attention was paid to the cuspate form of rising intrusions that is commonly seen in nature but that has proved elusive in most earlier experiments. The numerical results demonstrate that large localised initiating perturbations and inertial influence during flow both tend to cause a decrease in the wavelength of the resulting flow pattern and can, under certain circumstances, serve to promote a cuspate morphology. The use of a relatively low viscosity Newtonian fluid as an approximation of the coarse-grained upper layer coupled with, critically, power-law behaviour in the underlying clayey layer was also found to promote a cuspate form in the rising intrusion.
Mobilization structures in geological soft sediments have long been described but the need has grown for a more quantitative understanding of their genesis (e.g. Prentice, 1961; Brodzikowsk & Van Loon 1985; Cegla et al. 1987). This paper reports our attempts to model the gravity driven soft sedimentary features known in geology as load casts and flame structures (Keunen 1953; Pettijohn & Potter 1964). Such structures are widely seen in rocks formed from interlayered sequences of sand-rich and clay-rich sediments, which are inferred to have had different densities at the time of deformation. The features arise by a combination of the upward growth of the lower clay-rich sediment and the downward loading of the upper sand-rich sediment (for example, see Fig. 1). Although flames are sometimes loosely referred to as dewatering structures, most analyses have regarded them as natural examples of Rayleigh-Taylor instabilities (Allen 1984; Collinson & Thompson 1987; Ronnlund 1989). These arise where one homogeneous fluid of constant density is superposed above one of lesser density. The extent to which any water loss plays a part is unclear, but RayleighTaylor instability theory has proved to be a most useful basic approach to understanding these structures. For example, information is easily found on both the wavelength of the flow pattern and the rate of growth of a perturbation within a system that contains Newtonian fluids of known viscosities and densities (Ramberg 19680, b). Although some of the principles are applicable to mobilization on a larger scale, and in materials such as salt and magma, this investigation is restricted to soft sediments. Analytical approaches to the problem (e.g. Bio &
Ode 1965; Turcotte & Schubert 1982) are limited in a number of significant ways. For example, they are all restricted to the initial stages of growth of an instability, that is until the amplitude of the perturbation reaches around only 10% of the wavelength and many natural structures are likely to have progressed well beyond this stage (Ramberg 1981; Johnson & Fletcher 1994). In addition, most considerations are restricted to the linear terms that occur in the related Navier-Stokes equations. In particular, they consider strictly Newtonian behaviour - which is not normally the case in soft sedimentary materials - and they assume non-inertial flow (e.g. Maltman 1984; Coussot & Piau 1994). The exceptions to these last two shortcomings are Biot (1963a, b), who introduced elasticity and Chandrasekhar (1955), who included inertial terms from the Navier-Stokes equations. An alternative approach to understanding reversedensity structures is to simulate them experimentally in the laboratory. This can provide insights beyond the scope of theoretical analysis but also involves significant shortcomings. For example, simple experiments designed to model these structures using geological materials such as mud and sand (e.g. Anketell et al. 1970) are difficult to control because of the complex rheologies involved and the developing structures are impossible to observe in opaque sediments. Other experiments, often originally designed to model the growth of salt domes, have used simpler and more translucent fluids such as oil and syrup (Whitehead & Luther 1975; Talbot 1977) or have involved more elaborate modelling techniques such as pressure driven accelerations or centrifuges (e.g. Lewis 1950). These experiments have added
Fig. 1. Rounded load casts composed of lithified volcanic sand (upper, dark coloured layer) and cuspate flame structures (lower, pale coloured layer) in mudstone. Note the pointed, cuspate form of the flame structures, well shown at the left-hand side of the photograph. Eocene volcanogenic turbidite sequence, Boso Peninsula, Japan. Coin is approximately 2 cm in diameter.
considerably to the understanding of reverse-density structures but are limited in describing the flow behaviour of soft sediments because of their inherent differences from the geological situation. Over the last 30 years or so, advances in computer technology have led to the rise of numerical modelling as a powerful means of predicting the behaviour of physical systems. This numerical approach (Schmeling 1988; Romer & Neugebauer 1991; Poliakov et al 1993; Van Keken et al 1993) has allowed many of the simplifications and limitations of the analytical and experimental analysis of Rayleigh-Taylor instabilities to be overcome. In particular, numerical simulations have provided results from stages of diapiric development far beyond the scope of analytical theory and have included factors such as depth- and temperaturedependent viscosities, syn-depositional diapiric growth, multiple layers, erosion at the surface and non-Newtonian behaviour. We report here our attempts to use modern numerical and rheological techniques to examine the initiation and development of reverse-density structures in unlithified sediments. First, the method underlying the numerical code is summarized. All the simulations were two-dimensional and were either planar or axisymmetric. Then the use of the code to provide insights into factors that are thought to be linked with the development of naturally occurring reversedensity soft sedimentary structures, including the effects of the size and form of the initial perturbation
and the possibility of inertial flow is described. Rheological experiments allowed appropriate fluid behaviours to be incorporated in the code. Once the problem was correctly stated within the program, material parameters, boundary conditions or geometry were varied in order to observe the effects of these changes on a given system. In particular, specific attention was paid to the pointed or cuspate morphology (Walton 1956; Brodzikowski et al. 1987) of many natural flame structures (see, for example, Fig. 1). Although this cuspate form has already been explained by second order analysis of 'mullion structure', in which sediment layers are subjected to uniform shortening or extension (Johnson & Fletcher 1994), its connection with reverse-density structures has intrigued earlier workers but has proved difficult to simulate in most experimental investigations. Dzufyriski and collaborators made the most successful attempts during the 1960s (e.g. Anketell et al. 1970), although the precise reasons behind the cuspate form are impossible to identify from their experiments.
The numerical code The commercially available numerical-modelling package 'Polyflow' (Polyflow s.a., PL de L'Universite, 16, B-1348, Louvain-la-Neuve, Belgium: e-mail; [email protected]) was employed, which was designed to simulate industrial
MODELLING OF REVERSE-DENSITY STRUCTURES
non-linear fluid flow problems (e.g. Debbaut et al 1997; Yao & McKinley 1998). Polyflow is a generalpurpose program that solves partial differential equations using the finite element technique and is designed for simulating flow-processes dominated by non-linear viscous flow phenomena and viscoelastic effects. It is based on the general principles of continuum mechanics together with various phenomenological and kinetic theoretical models of the rheological behaviour of fluids. The models used in this investigation were Newtonian and power-law. The program was run on a 'Dec Alpha 3000' computer. All the simulations of this investigation incorporate an incompressible fluid condition. Using this condition we have for the Cauchy stress tensor, where p is the pressure, I is the unit tensor and r is the extra stress tensor. The form of the extra stress tensor depends on the type of fluid; in the case of a generalised Newtonian fluid the extra stress tensor is given by where 17(1 III) is the coefficient of viscosity of the fluid and for complex flows is assumed to be a function of the second invariant of the rate of deformation tensor, 2D (Macosko 1994). The coefficient of viscosity of a Newtonian fluid is constant whereas power law behaviour is given by which involves a consistency factor K with dimensions (Pa.sn) and a power law index n. Eqs 1 and 2 can be substituted into an equation of motion
37
tions of the boundary, the concepts of sub-domains and boundary sets are used. Each sub-domain corresponds to a group of finite elements and boundary sets are groupings of finite element connectors located on the domain boundary. A well-posed timedependent problem requires the specification of initial conditions throughout the domain along with the necessary boundary conditions. During this investigation certain boundary conditions have been used repeatedly. These include a no-slip condition where vn is the normal velocity and v^ is the tangential velocity, and a plane of symmetry condition where fy is the tangential force. Additionally, in certain simulations a free surface condition was imposed on the upper boundary. For this condition one requires simultaneously that the normal force f n , the tangential force fy, and the normal velocity \n be prescribed. Surface tension enters in the system as a force that has the direction of the normal to the free surface
where nis the unit vector normal to the surface, R is the Gaussian curvature of the surface and the parameter s is the surface tension coefficient. This normal force introduces tangential forces at both ends of the free surface
where s is the unit vector tangent to the free surface and directed away from the surface. Finally
which arises from the conservation of linear momentum. Here p is the density of the fluid, F is the external body force per unit mass, v is the velocity vector and the operator D/Dt is the material time derivative. When considering the flow of a Newtonian fluid the substitution produces the classical Navier-Stokes equations. A similar substitution can be made based on the power law model, i.e. by incorporating Eq. (3). Upon specifying the boundary and initial conditions of the problem the resulting equations, along with the continuity equation which results from the conservation of mass of an incompressible fluid
where x is the position of a node on the free surface. Eq. (10) is the kinematic condition for a time dependent problem and means that the free surface must follow material points in the normal direction, the tangential displacement being left arbitrary. The final condition used in the simulations is a moving interface condition, which occurs along the intersection of two sub-domains, 1^ and fl2. This condition holds when the kinematic condition, Eq. (10), is added to the conditions
are solved by the numerical calculations. In order to accommodate different sets of equations existing in different parts of a domain, for example due to changes in the material properties and different boundary conditions on different sec-
which ensures the continuity of the velocity vector and surface force over the interface. Remeshing techniques are necessary when free
38
P. HARRISON & A. MALTMAN
surface or moving interfaces are present. Throughout this investigation the Thompson transformation (Thompson 1985), which was designed to accommodate complex deformations in all directions of the Cartesian space, is implemented. The Galerkin method is used in choosing the shape functions for the calculations. The interpolant used for the velocity, pressure and co-ordinate fields are respectively quadratic, linear and quadratic. A first order Euler explicit-implicit predictor corrector method is used to calculate the evolution of the system in time.
Rheological behaviour of sediments An important step in moving towards the reality of natural geological situations is to incorporate into the simulations the density and Theological behaviour of typical sediments. It is convenient to begin by considering the higher and lower density sediments separately, i.e. the rheological behaviours of the upper and lower components of a two-layer instability. The difference in densities is usually a result of the packing efficiency of the constituent particles. Muddy sediments, consisting of very fine-grained clay particles with equivalent diameter 4 microns or less, will form the less dense sediments whereas sandy sediments tend to form the denser layers. This is because the surfaces of clay particles are subject to weak but significant electrostatic forces. These forces limit the packing efficiency of the particles (Cheng 1987). If the principles of continuum mechanics are to be used in simulating flow the system has to be treated as homogeneous, hence the particles within the fluids must be small in relation to the dimensions of the system under consideration (Anketell and Dzulyriski 1968). The numerical simulations presented in the present study involve layers measuring approximately one or two centimetres thickness and so the rheological behaviours of materials containing particles with equivalent diameters no larger than 1000 microns are considered. In order to differentiate between fine and coarse-grained particles, arbitrary size limits are used in accordance with convention. Thus, coarse-grained particles are limited to sand particles measuring between 63 and 1000 microns equivalent diameter and fine-grained particles include clay and silt with equivalent diameter less than 63 microns. These limits provide the definition of what we mean by 'fine-grained', 'coarse-grained' and 'watersaturated sand' sediments. Fine-grained sediments include a mixture of water and clay and silt size particles, coarse-grained sediments include a mix of water and clay, silt and sand size particles and watersaturated sand sediments include a mix of water and sand particles. For a given solids concentration, the content of fine particles in these sediments influences
their overall bulk density. As indicated earlier, finegrained sediments are usually the least dense, followed by coarse-grained sediments and finally water-saturated sand sediments. The rheological behaviours of each type of sediment are discussed in turn below. The multiphase nature of sediments results in solid-solid, solid-fluid and electrostatic interactions that combine to cause complicated non-Newtonian rheological behaviours such as viscoplasticity, unsteady gravitational flow, thixotropy, and liquefaction (Takahashi 1991; Iverson 1997). Thixotropy involves the agitation or shearing of an aggregate of clay particles, which disturbs the interactions of their electrostatic potentials and leads to a temporary decrease in both the apparent yield stress and viscosity of the bulk material. Sediments containing larger coarse-grained particles are subject to liquefaction, a process similar in effect to thixotropy. Many liquefaction mechanisms have been described and different accounts of the physics involved have been given (Allen 1985; Owen 1987; Maltman 1994; Nichols 1995). The consensus between all the different explanations is that the solid matrix of coarse-grained particles which in this case exists due to intergranular friction rather than electrostatic interactions - is temporarily disrupted by effects such as intense vibration caused by earthquakes or increases in pore fluid pressure above hydrostatic values. However, for the puiposes of this study it is enough to know that the action of a liquefaction mechanism results in a sudden loss of strength of the sediment. Geologists commonly view reverse-density features such as flame structures in sands as having involved some degree of liquefaction, for example through the sediment having been disturbed seismically. Experiments on geological materials seem to support this inference (e.g. Anketell et al 1970). Thus the gravity structures are taken to have developed in a stratified soft sediment that was initially at rest metastably but began to flow spontaneously, through thixotropic behaviour or liquefaction, following triggering disturbances. A liquefied state is assumed in the following numerical simulations, leading to the advantage of enabling us to ignore the apparent yield strength that would normally exist in undisturbed sediments and which would consequently tend to resist flow. Flame structures are often loosely referred to as dewatering structures. The extent to which drainage alone would generate such forms is debatable but the models reported here do not involve water loss.
Fine-grained slurries Sedimentation tests were conducted using kaolinite clay (Harrison 1996). The significance of electro-
39
MODELLING OF REVERSE-DENSITY STRUCTURES Table 1. Density and solids concentrations determined for kaolinite-clay sediments Type of suspension
Kaolinite-distilled water Kaolinite-salt water
Table 2. Parameters of Equation (13) determined from equations fitted to rheological tests on kaolinite-clay sediment
Average percentage concentration of solids by weight
Density of deposit
Type of slurry
A (Pa.s")
B
24.9
1230
Kaolinite-distilled water Kaolinite-salt water
0.129 0.111
0.101 0.085
30.3
1270
static forces in determining the degree of flocculation of clay particles means that the addition of electrolytes such as salt can produce drastic changes in the physical properties of the material. For this reason, some of the experiments involved the clay being mixed with distilled water and others with salt water at a concentration of 40 grams of salt per litre of distilled water. Sedimentation tests gave approximate solids concentrations and densities of the resulting deposits. The results displayed in Table 1 suggest that the addition of salt led to an increase in the compaction of the deposits. It should be noted that values in Table 1 are average values of the percentage concentration of solids by weight in the sediments. In reality, the concentration of solids will vary as a function of burial. However, for the purposes of this study, the average value of the solids concentration is considered sufficient to provide insights into the effects of the non-Newtonian behaviour of clay slurries in gravity driven sedimentary structures. Using the information of Table 1, fine-grained slurries of similar solids concentrations to the deposited sediments were produced for rheological analysis. The aim of these experiments was to produce a simple rheological equation that could relate viscosity to shear rate and concentration of solids. Experiments were performed using a Cammed Constant Stress Rheometer. Eq. 3, the power-law model, was found to give a good fit to the test data. Within the concentration range considered for a given shear rate, increases in concentration of solids caused an exponential increase in the slurry viscosities, in agreement with previous work (e.g. Coussot & Piau 1994). The sensitivity of this relationship depended on the type of clay under consideration. Hence, the dependence of the slurry viscosities on concentration and shear rate was fitted using an equation of the form
n
0.29 0.37
A is the consistency factor (Pa.sn), B is a dimensionless constant, C is the concentration of solids in the mixture expressed as a percentage of the total weight and n is the power-law index which determines the shear-rate exponent of the slurry. The temperature dependency for each slurry was found to be negligible between the range of 5 to 25°C. Table 2 shows the values of A, B and n determined for the kaolinite-clay slurries. These values were determined at solids concentrations ranging between 1 to 1.5 times the solids concentrations found for the deposited sediments of the sedimentation tests. This range was chosen for analysis since preliminary experiments (Harrison 1996) suggested the small-scale sedimentary structures investigated here are much more likely to flow while the sediments contain high water contents and offer little resistance to deformation. As a result of these experiments, the power law model was chosen to approximate the behaviour of fine-grained sediments in the numerical simulations.
Coarse-grained slurries and water saturated sand
Attempts to derive simple equations that relate the concentration of solids to the relative viscosity between the liquefied dispersion and the interstitial fluid have been reviewed by Allen (1984). While these theories provide estimates for the viscosity of liquefied sand of the order 0.1 to 1 Pas, they fail to include shear rate dependence of the viscosity. Furthermore, experimental observation of asynchronously varying stresses in the solid and fluid phases of debris flows suggests the complex interaction between granular temperature and pore fluid pressure necessitates the use of two-phase models in explaining the behaviour of coarse-grained slurries (Iverson 1997). Even so, having drawn attention to the difficulties in modelling the flow behaviour of large-grained slurries, we point out that in using the numerical code, we are currently restricted to choosing well-defined one-phase rheological models. Attempts to study the rheological behaviour where 17 is the viscosity and is given here as a function of the shear rate, 7, rather than the second invar- of coarse-grained slurries experimentally have iant of the rate of deformation tensor in order to prompted various techniques, such as experiments reflect the viscometric nature of the measurements. using large-scale rotational rheometers capable of
40
P. HARRISON & A. MALTMAN
accommodating large particle size, or alternatively, velocity measurements of debris flowing down inclined channels. However, the difficulties in investigating the precise behaviour of particulate flows are often reflected by the ambiguity of flow curve data (e.g. Phillips & Davies 1989; Holmes etal 1990). In general, therefore, while results presented in a previous section point clearly to the use of the power-law model in describing the behaviour of the lower layer of fine-grained sediment, the choice for the upper layer is less obvious. As indicated earlier, both the upper and lower layers of the instabilities considered here are presumed liquefied by a triggering disturbance before the onset of flow. This means that cohesive and intergranular frictional strengths are ideally reduced to zero. Consequently, terms representing sediment strength in rheological models such as the one presented by Johnson & Martosudarmo (1997) vanish. The effects of pore fluid rheology and intergranular collisions then govern the behaviour. Each of these, according to theory, can have non-linear dependencies on the shear rate. However, flow-data following yield can often be fitted equally well by linear or non-linear flow curves. In particular, linear fitting becomes increasingly plausible for sand-water mixtures of decreasing solids concentration (e.g. see data presented in Johnson & Martosudarmo 1997). Although in doing so, factors such as the 'dispersion gap' (Johnson 1984) and the strong dependence of viscosity on solids concentrations (Allen 1984) must be neglected. As an approximation then, both power-law and Newtonian behaviour might be used in modelling the behaviour of fully liquefied coarse-grained and sand-water sediments. Use of these simple rheological models in approximating the behaviour of natural sediments, although distancing the simulations from the details of the real situation, facilitate the use of numerical models as a first step in investigating the effects of non-linear rheological behaviour in the lower or possibly both sediment layers.
Investigating the form of the deforming interface using the numerical code In this section, we use the code to investigate factors that could possibly influence the form of the interface between the layers and consequently the form of resulting sedimentary structures. The factors considered are the: • • •
size and form of the initiating perturbation; effects of inertial flow; introduction of a power-law rheological model.
Each of these factors is indeed found to influence the form of the rising intrusion. In particular, these
factors are found to be conducive to the formation of a more pointed or cuspate intrusion form for the rising lower layer. Note that for all the simulations of this investigation, a very low value for the surface tension coefficient for all fluids was used (10~7 NrrT1). Thus, the results and conclusions of the work apply only to systems in which surface tension between the fluid interfaces plays a negligible role. Since all the geological materials modelled here are all water based, they are presumed to have similar surface tension coefficients, consequently surface tension should have little effect at the fluid-fluid interfaces.
Effect of initiating perturbation Theoretically, if two fluid layers of differing densities are arranged with the layer of greater density on top, they can remain in this meta-stable state for an indefinite period of time. Thus, in order to initiate flow, a perturbation must be introduced into the system. For the purposes of analytical theory, a periodic sinusoidal interface between the fluid layers is often used as the starting perturbation of the instability (e.g. Ramberg 1981; Johnson & Fletcher 1994). Other experimental, theoretical and numerical investigations have incorporated more localised perturbations in which second generation structures develop due to the disturbance provided by the growth of primary structures (e.g. Parker & McDowell 1955; Woidt 1978). Kelling & Walton (1957) postulated that reverse-density structures in natural submarine sediments are often initiated at water current marks on the seafloor, which would often tend to have a localised, discontinuous form. With this in mind, results of simulations presented in this section were initiated by small, localized perturbations of an otherwise flat interface. An investigation into the influence of the initiating perturbation, on the wavelength that eventually dominates the flow pattern of a system, has been performed previously (Schmeling 1988). The results suggested that under certain circumstances, the characteristic wavelength predicted by analytical theory is not necessarily the one that dominates the mature flow pattern. In this section, we show that the size of a localized disturbance can have profound effects on the subsequent form of a rising intrusion. Figure 2 shows the development of a fluid dome inside a cylindrical geometry. The simulations incorporated a no-slip condition on the walls of the container and an axis of rotation about which the domain was rotated. Thus, the simulation is a twodimensional mathematical problem but shows the rise of a dome in three-dimensional Cartesian space. The only difference between the two simulations of Figure 2 was the shape and size of the initial pertur-
MODELLING OF REVERSE-DENSITY STRUCTURES
41
Fig. 2. Effect of perturbation size: axisymmetric simulation with rotational symmetry. Domes forming in rising fluid from small (left-hand side) and large (right-hand side) interfacial perturbations. Cylindrical geometry measuring 20 cm diameter and 25 cm high. Both the upper and lower fluids are Newtonian and of viscosity equal to 0.1 Pas. The form of the interface is shown at progressive stages in the evolution of the domes.
bation. The influence of the perturbation is dramatic and totally alters the form of the rising intrusion. This result is explored in subsequent simulations.
Effects of inertial flow As discussed earlier, disturbed sediments may temporarily possess relatively low viscosities and thus provide scope for inertial flow during their deformation. Graphical comparisons have been drawn by Ramberg (1981) between the predictions of his own non-inertial theory and Chandrasekhar's (1955) inertial theory. These comparisons show how the predictions of the two theories diverge as inertial flow becomes increasingly important. In particular, they demonstrate how Chandrasekhar's (1955) inertial theory predicts a smaller wavelength for the flow pattern than the prediction made by Ramberg's noninertial theory, for identical systems in which inertial terms in the analytical equations are large enough to be considered significant. In this section, the effects of inertial flow on the flow patterns of instabilities initiated from localised perturbations are investigated. In order to do so, a series of four numerical simulations was conducted in which the effects of inertia in each successive simulation were increased. The Reynolds number (e.g. Turcotte & Schubert 1982) provides a ratio between inertial and viscous forces during flow
where RN is the Reynolds number, v is the velocity of flow, L is the characteristic length of the system, p is the density of the fluid and r\ is the dynamic viscosity of the fluid. Increasing values of RN indicate an increasing influence of inertia in relation to viscosity. Thus, by increasing the density of the upper fluid layer in a reverse-density system, RN is increased, not only because of the increase in p of the upper fluid, but also because of the subsequent increase in v which is a direct consequence of the greater buoyancy forces in the system. (The complex flow, involving two fluids and both spatially and temporally changing shear rates means that it is difficult to give actual values for RN. However, using the depth of the fluids for L and the rate of rise for v, we give a very rough estimate of RN in the order of IC^-IO 2 for the upper fluid in the final simulation of Figure 3. The effects of inertia are investigated by comparing simulations in which inertial effects are altered). Figure 3 and all subsequent simulations are twodimensional planar simulations. In each figure, the simulations mirror image in the plane of symmetry is shown in order to better illustrate the form of the rising intrusions. Figure 3 clearly demonstrates how increasing the importance of inertia results in a decrease in the wavelength of the flow pattern within
42
P. HARRISON & A. MALTMAN
Fig. 3. Effect of inertia: viscosities and densities are shown in the figure. A no-slip boundary condition was used on three sides of the geometry along with a moving interface condition and the vertical plane of symmetry. The diagrams show the initial form of the interface, along with its shape at later stages of development.
the system and that the form of the rising structure is seen to decrease in width. Also evident is the fact that the rate of lateral progress of the flow pattern, in relation to the vertical growth of the perturbation, decreases as inertia becomes more important. This latter observation has implications regarding the form of reverse-density sedimentary structures that are created as a result of inertial flow and are initiated from localised perturbations. As inertial flow becomes more important in their genesis, these structures will tend to be more mature in relation to neighbouring structures and will become progressively more intrusive. Thus, as with the effects of the initiating perturbation, the influence of inertial flow on the form of mature rising intrusions is significant. These two results highlight the need for caution when estimating the relative viscosities of fluid layers through observations of the form of mature rising intrusive structures (Anketell et al. 1970; Whitehead & Luther 1975; Ramberg 1981; Jackson &Talbotl986).
Incorporating the power-law model into the numerical simulations In light of the previous discussion on the rheology of natural sediments, the simulations that follow use the power law model to simulate the flow behaviour of the fine-grained lower layer and either the Newtonian or power law models to simulate the behaviour of the coarse-grained upper layer. Figure 4 shows an attempt to determine the maximum value of n, the measure of shear thinning, which could be incorporated in the simulations. A two-dimensional planar geometry with a free-surface condition imposed at the upper boundary of each system was used, with planar symmetry on the left side and no-slip conditions on other outer boundaries of the geometry. The value of AeBC of Eq. (14) was chosen as 1.6 Pas", which corresponds to the solids concentration of a kaolinite-distilled water deposit (see Tables 1 and 2). The value of n in Eq. (14) was then decreased from a value of 1 (the Newtonian model) towards 0. The
MODELLING OF REVERSE-DENSITY STRUCTURES
43
Fig. 4. A series of three simulations (a-c) using a power-law model for the lower fluid and a Newtonian model for the upper layer. The values of material parameters are shown in the figure. The depth of the layers was 7 cm.
minimum possible value of n was 0.5, hence the maximum possible degree of shear-thinning which could be incorporated in the simulations without numerical solutions calculated by the code diverging, was with a shear-thinning exponent of -0.5. The effect of this shear thinning behaviour is apparent in Figure 4, i.e. a decrease in the wavelength of flow. Three possible reasons for this decrease are: inertial effects due to a combination of decreasing vis-
cosity and increasing shear rates within the shearthinning fluid; effects attributable directly to the nonlinearity of the power-law model; a combination of these two factors. Estimates of the magnitude of the RN for the three flows are more difficult to make here due to the changing viscosity and deformation rate in the lower layer. However, using a viscosity map of the lower layer, the width of the rising intrusion and the maximum rate of rise of the intrusion, rough
44
P. HARRISON & A. MALTMAN
Fig. 5. (a) Similar simulation to Fig. 4c, but with the value ofAeBC increased from 1.6 to 10 Pas. The result was an increase in the width and wavelength of the flow pattern. The duration of flow was 32.5 s. (b) Similar simulation to (a), but with the height of the initiating perturbation increased from 0.1 to 0.3 cm. The result is a dramatic change in the form of the rising intrusion. The duration of flow was 10.6 s. Viscosity maps of the lower shear-thinning fluid are superposed on the right side of the diagrams.
estimates of between 0.1 to 10 were made for RN of the lower layer of Figure 4 (a-c). From this range of RN it is difficult to say whether or not the decreasing wavelength was indeed due to inertial effects. Thus, to determine the reason for the decrease in wavelength in Figure 4 further simulations were conducted. The value ofAeBC in the lower layer was increased from 1.6 to 10 Pas", holding n at 0.5. Thus, the viscosity of the lower power-law fluid was increased.
Any possible influence of inertia was consequently decreased without altering the non-linearity introduced by the power-law model. The result is shown in Figure 5(a). Figure 5 shows that the width of the rising intrusion increases with respect to that of Figure 4c. For non-inertial flow of Newtonian fluids, if one increases the viscosity of the lower layer relative to the upper layer, one expects to see the formation of
MODELLING OF REVERSE-DENSITY STRUCTURES
45
Fig. 6. Simulation in which both fluids are Newtonian. The material parameters are shown in the figure. The depth of each layer is 7 cm. The initiating perturbation was 0.3 cm high with a total width of 0.8 cm. The form of the interface between the fluids is plotted at various stages in the evolution of the instability; the final stage was achieved after 125 s. Even with this large perturbation the rising intrusion failed to adopt the cuspate form.
an increasingly narrow rising intrusion. This is not observed in the simulations of Figure 4c and Figure 5a; in fact, the opposite is observed. Certainly, the increased viscosity of Figure 5a relative to Figure 4c means that inertia must play a smaller role in the simulation of Figure 5 a than in Figure 4c. In addition, since increasing inertial flow was previously found to decrease the wavelength and width of rising intrusions, the results of the simulations suggest that inertial flow was indeed the mechanism behind the decreasing wavelength in Figure 4 (a-c). Furthermore, since the simulations of Figure 5 included viscosities comparable to those expected in natural sediments, we conclude that inertia is likely to be important during the genesis of reverse-density soft sedimentary structures in nature. Figure 5b shows a simulation in which all parameters were identical to that of Figure 5 a, apart from the height of the initiating perturbation that was increased from 0.1 to 0.3 cm in height, keeping the base width of the perturbation constant at 0.6 cm. As in Figure 1, the change in the perturbation produced a dramatic change in the form of the rising intrusion. The simulation initiated using a small perturbation, Figure 5a, is rounded whereas that initiated using a large perturbation, Figure 5b is of a pointed or cuspate form, very similar to the flame structure of Figure 1. Following this result, various tests were performed to identify more precisely the reason behind the cuspate morphology of the rising intrusion.
Cuspate morphology in reverse-density soft sedimentary structures As with previous experiments, many of the simulations have produced various bulbous forms for both the sagging load casts and the rising intrusions.
However, many natural flame structures are distinctly pointed and cuspate but this form has proven particularly elusive in previous simulations. Anketell et al. (1970) surmised that a large viscosity ratio is probably the critical condition for generating this morphology, but even experiments that included the non-inertial flow of Newtonian fluids with an extremely large viscosity ratio have had little success in its simulation (Whitehead & Luther 1975). In the present work, the factors considered in the attempt to determine the conditions necessary for the creation of the cuspate morphology were: • • • •
viscosity of the upper and lower fluids and the ratio between them; size of the initiating perturbation in the system; inertial flow of the fluids; shear thinning fluid behaviour.
Viscosity ratio Numerical simulations were carried out on Newtonian fluids with various viscosities and ratios as high as 10000 and in no instance did the rising structure develop a cuspate morphology (e.g. Fig. 6). Thus it is evident that a high viscosity ratio is not on its own, capable of producing a cuspate structure. The experiments were conducted with both small and large perturbations but these also failed to provide appropriate conditions for cuspate structures to form in these Newtonian fluids.
Inertial flow The only experiment we know of that has produced cuspate morphologies in Newtonian fluids is that
46
P. HARRISON & A. MALTMAN
Fig. 7. (a) Photograph from the experiment conducted by Lewis (1950). The result was unusual in that the fluids used in the experiment were Newtonian and yet the form of the rising intrusions was cuspate. The depth, length and width of the water layer was 2.8, 6.25 and 1.25 cm respectively. We infer that the cause of the cuspate form is the non-linear influence of the inertia-dominated flow, (b) Simulation of a system containing Newtonian fluids in which inertia plays an important role during the flow. The depth of each layer is 7 cm. The form of the rising intrusion is cuspate during the early stages of its evolution. Material parameters are shown in the figure. The time taken to reach the last stage of deformation was 0.6 s.
conducted by Lewis (1950) (Fig. 7a). This experiment was originally conceived to verify the linear inviscid analysis proposed by Taylor (1950). It is inferred from the speed of the experiment that it involved significant inertial flow. The experiment used fluids of low viscosities, namely air and water, initially stratified in a stable configuration. The interface was of approximately sinusoidal form, due to a standing wave on the surface of the water. The fluids were rapidly accelerated downwards so that they experienced approximately 21 g and flowed extremely quickly: the photographs recording the flow were separated by only milliseconds. These high speeds acting on fluids of low viscosities mean that inertia must have played a dominant role during
the flow (see Eq. 14). Consequently, in order to increase the influence of inertia in this study's simulations were carried out using low viscosity Newtonian fluids for both layers. The result, shown in Fig. 7b, was partially successful in that the rising intrusion assumed a cuspate form during the early stages of flow, although it tended towards a more rounded structure during the later stages. The results indicate that non-linear effects due to inertial flow do indeed play a role in producing pointed forms. We continued the investigation of the cuspate form by considering the effects of non-Newtonian behaviour. Numerical simulations similar to that shown in Figure 5b, but which incorporated the same power-law model in both the upper and lower layers,
MODELLING OF REVERSE-DENSITY STRUCTURES
47
Fig. 8. (a) Simulation of a shear-thinning fluid rising through a Newtonian fluid. Material parameters are shown in the figure. The interface between the two fluids is shown at progressive stages in the evolution of the flow pattern. The time taken to reach the last stage was 27 s. (b) Same parameters as (a) but with AeBC = 40 Pas". The time taken to reach the latest stage of the evolution was 105 s. The morphology of the rising intrusion is seen to be more cuspate than that in (a).
completely failed to produce a cuspate morphology. Two further simulations were conducted similar to Figure 5b, but with differing viscosity ratios between a Newtonian fluid in the upper layer and a power-law fluid in the lower layer. The size of the initiating perturbation and the viscosity of the upper layer was held constant in each simulation, but the viscosity of the lower, power-law, fluid was increased by changing the value of AeBC in Figure 8a to 20 Pas" and in Figure 8b to 40 Pasn. Here, pointed morphologies did result and the rising intrusion became increasingly cuspate as the viscosity of the shear-thinning fluid was increased. Although the simulations that involved low-viscosity Newtonian fluids for both layers suggested that inertial flow helped promote a cuspate morphology, the increase in viscosity of the lower layer of Figure 8b means that the importance of inertia is smaller relative to the simulation of Figure 8a. Thus, it is inferred that the influence of inertial flow is not itself the cause of the increasingly cuspate forms of Figure 8a and b. This investigation has also illustrated how extremely high viscosity ratios between
Newtonian fluid layers fail to produce a cuspate morphology. Thus, neither can the increasing viscosity ratio between the fluid-layers fully account for the increasingly cuspate forms of Figure 8. The only other possible factor is the non-linear influence of the lower layer. It is concluded, therefore, that in Figure 8 it is the non-linear power-law behaviour of the lower layer that is particularly conducive to the production of a cuspate morphology. In addition, the greater the viscosity ratio between the lower, shear thinning fluid and the upper fluid, then the more attenuated is the form of the rising intrusion. It should be noted that all the simulations that produced a pointed intrusion, required a relatively large initiating perturbation.
Discussion The simulations reported in the previous sections provide insights into the various influences on flow patterns and the resulting structures in geological sediments with reverse-density stratification.
48
P. HARRISON & A. MALTMAN
Although bulbous forms are produced under a range of conditions, we have demonstrated that under certain circumstances cuspate morphologies in rising intrusions are favoured. In particular, two possible explanations have been identified. The first is a non-linear effect caused by significant inertial flow and occurs whenever the viscosities of the fluid layers are sufficiently low or when the density contrast between the fluid layers is sufficiently high. The second is also a non-linear effect due to nonNewtonian fluid behaviour in the lower layer. In this case, it is a high viscosity ratio between the lower and upper fluid layers that is conducive to the formation of the pointed morphology. How likely are these various conditions in natural systems? Geological sediments are deposited with a range of densities and it is well known that they can exhibit non-Newtonian behaviour and undergo liquefaction processes, which allow inertial flow and induce temporary reductions in apparent yield strength. Given these reasonably common situations, acting in conjunction with the relatively large surface perturbations that readily form during the deposition of sediment layers in dynamic, subaqueous environments, all the conditions identified in our investigation are perfectly possible in nature for the production of cuspate morphologies. Of the two means thought here to be favourable for generating these forms, it is thought that the role of non-Newtonian behaviour in the lower layer is the more important in the geological setting. Although inertial flow provides a mechanism for inducing cuspate rising intrusions, it only appears to operate, with instabilities containing Newtonian fluids and small initiating perturbations, under artificially high accelerations unlikely during the flow of natural sediments that are driven purely by buoyancy forces. In the simulation that used the inertial mechanism to produce a cuspate form of the rising intrusion, Figure 5b, a relatively large initiating perturbation was a necessary additional condition. However, a recently deposited fine-grained sediment of low solids concentration, and consequently low viscosity and insignificant apparent yield strength, would seem unable to support such a large perturbation on its upper surface during burial. The following scenario is, therefore, envisaged, involving non-linear flow of the lower layer, for the production of the cuspate morphology. A finegrained sediment is deposited and left undisturbed for a period of time, during which it undergoes selfweight consolidation. The sediment's viscosity increases exponentially with the increasing solids content and eventually a small apparent yield strength develops, sufficient to allow any scours or other irregularities that develop on the sediment surface to be preserved. Ensuing deposition of a coarse-grained layer takes place; once settled its density is immedi-
ately greater than the lower fine-grained clayey layer and effectively creates a Rayleigh-Taylor instability. The two layers stay in this reverse-density stratification until the onset of some liquefying mechanism. If the self-consolidation process continues for a sufficiently long period before the deposition of a denser layer, or if lithification of the layers becomes significant before any liquefaction, then the sediments remain in this metastable state without undergoing deformation. The layers become preserved in the sedimentary record without reverse-density structures ever having been initiated. However, if some disturbance occurs while the metastable sediments are still susceptible to liquefaction or thixotropic behaviour, the apparent yield strengths of both the upper and lower sediments are drastically reduced and the interfacial irregularities preserved on the upper surface of the lower fine-grained layer behave as the relatively large initiating perturbations necessary for the production of a cuspate morphology. While the lower layer flows in a non-linear shear-thinning manner, the low packing efficiency of the recently deposited coarse-grained upper sediment layer is conducive to low-viscosity Newtonian flow. The simulations of our experiments suggest that the flow of the sediments occurs probably over a time-scale of seconds. Since initiating irregularities will vary in size in nature, the requirement of a large, localised initiating perturbation in creating a cuspate morphology may explain why reverse-density soft sedimentary structures of different morphologies can develop side by side, as in Figure 1, with only some of the structures developing a cuspate form. Finally, during liquefaction and deformation of the sediments, the packing efficiency of the upper sand and water or coarse-grained sediment increases. Thus, upon restoration of grain contacts, the sediment's apparent yield stress is significantly increased. Consequently, the likelihood of further liquefaction and deformation of the sediments is immediately reduced. Continuing sedimentation, burial and lithification progressively increase the strength of the materials, eventually precluding further flow altogether. Ultimately, lithification becomes complete and the reverse-density structures are preserved in the sedimentary record. This work would not have been possible without the help and support of Professor K. Walters. The expertise of V. Navez and E. Degand in using the Polyflow numerical code was also invaluable. C. Talbot and R.A. Hindmarsh provided helpful comments on the manuscript.
References ALLEN, J.R.L. 1984. Sedimentary Structures: Their Character and Physical Basis, volumes I and II. Elsevier, Amsterdam.
MODELLING OF REVERSE-DENSITY STRUCTURES ALLEN, J.R.L. 1985. Principles of Physical Sedimentology. George Allen and Unwin, London. ANKETELL, J.M. & DZULYNSKI, S. 1968. Patterns of density controlled convolutions involving statistically homogeneous and heterogeneous layers. Annals of the Geological Society of Poland, 38,401-409. ANKETELL J.M., CEGLA, J. & DZULYNSKI, S. 1970. On the deformational structures in systems with reversed density gradients, Annals of the Geology Society Poland, 40,3-30. BIOT, M.A. & ODE, H. 1965. Theory of gravity instability with variable overburden and compaction. Geophysics, 30,213-227. BIOT, M.A. 19630. Theory of stability of multi-layered continua in finite anisotropic elasticity. Journal of the Franklin Institute, 276,128-153. BIOT, M.A. 1963/7. Stability of multi-layered continua including the effect of gravity and viscoelasticity. Journal of the Franklin Institute, 276,231-252. BRODZIKOWSKI, K. & VAN LOON A.J. 1985. Inventory of deformational structures as a tool for unravelling the Quaternary geology of glaciated areas, Borea, 14, 175-188. BRODZIKOWSKI, K., KRYSZKOWSKI, D & VAN LOON, A.J. 1987. Endogenic processes as a cause of penecontemporaneous soft-sediment deformations in the fluviolacustrine Czyzow Series Kleszcow Graben, central Poland). In: JONES, M.E. & PRESTON, R.M.F. (eds) Deformation of sediments and sedimentary rocks, Geological Society, London Special Publications, 29, 269-278. CEGLA, J., BRODZIKOWSKI, K., KIDA, J, & MORAWSKI, S. 1987. Penecontemporaneous deformation of structures formed within some fine grained Quaternary sediments (Dalkowskie Hills S.W. Poland), Bulletin of the Polish Academy of Science, 35,221-234. CHANDRASEKHAR, S. 1955. The character of the equilibrium of an incompressible heavy viscous fluid of variable density. Proceedings of the Cambridge Philosophical Society, 51,162-178. CHENG, D.C.H. 1987. Thixotropy. International Journal of Cosmetics Science, 9,151-191. COLLINSON, J.D & THOMPSON, D.B. 1987. Sedimentary Structures, 2nd edition. Chapman and Hall, London. COUSSOT, P. & PIAU, J.M. 1994. On the behaviour of fine mud suspensions. Rheologica Acta 33,175-184. DEBBAUT, B., AVALOSSE, T., DOOLEY, J. & HUGHES, K. 1997. On the development of secondary motions in straight channels induced by the second normal stress difference: experiments and simulations. Journal of NonNewtonian Fluid Mechanics, 69,255-271. HARRISON, P. 1996. Rheological and numerical modelling of reverse-density structures. Ph.D. thesis, University of Wales, Aberystwyth, UK. HOLMES, R.R., WESTPHAL, J.A & JOBSON, H.E. 1990. Mudflow rheology in a vertically rotating flume. Hydraulics/Hydraulogy of Arid Lands, Proc. A.S.C.E. 1990 International Symposium, 212-217. IVERSON, R.M. 1997. The Physics of Debris Flows. Reviews of Geophysics, 35,245-296. JACKSON M.P.A. & TALBOT C.J. 1986. External shapes, strain rates, and dynamics of salt structures. Geological Society of America Bulletin, 97, 305-323. JOHNSON, A.M. 1984. Debris flow. In: BRUNSDEN, D. &
49
PRIOR, D.B. (eds) Slope Instability. John Wiley, New York, 257-361. JOHNSON, A.M. & FLETCHER, R.C. 1994. Folding of Viscous Layers; Mechanical Analysis and Interpretation of Structures in Deformed Rock, Columbia University Press, New York. JOHNSON, A.M. & MARTOSUDARMO, S.Y. 1997. Discrimination between inertial and macro-viscous flows of fine-grained debris with a rolling-sleeve viscometer. Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment, Proceedings of first international conference, Hyatt Regency San Francisco, San Francisco, California, 229-238. KELLING, G. & WALTON, E.K. 1957. Load-cast structures: Their relationship to upper-surf ace structures and their mode of formation. Geological Magazine, 44, 481-491. KUENEN, PH. H. 1953. Significant features of graded bedding. Bulletin of the American Association of Petroleum Geologists, 37,1044-1066. LEWIS, DJ. 1950. The instability of liquid surfaces when accelerated in a direction perpendicular to their planes, II. Proceedings of the Royal Society 202, 81-96. MACOSKO, C.W. 1994. Rheology: Principles, Measurements and Applications. VCH Publishers, New York. MALTMAN, A.J., 1984. On the term "soft-sediment deformation". Journal of Structural Geology 6,589-592. MALTMAN, A.J. (ed.) 1994. The Geological Deformation of Sediments. Chapman and Hall, London. NICHOLS, R.J. 1995. The liquefaction and remobilization of sandy sediments. In: HARTLEY, A.J. & PROSSER DJ. (eds) Characterization of Deep Marine Clastic Systems. Geological Society, London, Special Publications, 94,63-76. OWEN, H.G. 1987. Deformation processes in unconsolidated sands. In: JONES, M.E. & PRESTON, R.M.F. (eds) Deformation of sediments and sedimentary rocks. Geological Society, London, Special Publications, 29, 11-24. PARKER, T.J. & MCDOWELL, A.N. 1955. Model studies of salt dome tectonics. Bulletin of the American Association of Petroleum Geologists, 39, 2384-2471. PETTIJOHN, FJ. & POTTER, P.E. 1964. Atlas and Glossary of Primary Sedimentary Structures. Springer-Verlag, NewYork. PHILLIPS, J.P. & DAVIES, R.H. 1989. Generalised viscoplastic modelling of debris flow, A.S.C.E. Journal of Hydraulic Engineering 115,1160-1162. POLIAKOV, A., VAN BALEN, R., PODLADCHIKOV, Y, DAUDRE, B., CLOETING, S. & TALBOT, C.J. 1993. Numerical analysis of how sedimentation and redistribution of surficial sediments affects salt diapirism. Tectonophysics, 226,199-216. PRENTICE, J.E. 1961. Some sedimentary structures from a Weald Clay sandstone at Warnham brickworks, Horsham, Sussex. Proceedings of the Geologists Association, 73,171-185. RAMBERG, H. 19680. Fluid dynamics of layered systems in a field of gravity, a theoretical basis for certain global structures and isostatic adjustment. Physics of the Earth and Planetary Interiors, 1, 63-87. RAMBERG, H. 1968&. Instability of layered systems in a field of gravity. Physics of the Earth and Planetary Interiors, 1,427-474.
50
P. HARRISON & A. MALTMAN
RAMBERG, H. 1981. Gravity, Deformation and the Earth's Crust, 2nd edition. Academic Press, New York. ROMER, M. M. & NEUGEBAUER, H. J. 1991. The salt dome problem: A multilayered approach, Journal of Geophysical Research, 96, B2,2389-2396. RONNLUND, P. 1989. Viscosity ratio estimates from natural Rayleigh-Taylor instabilities. Terra Nova, 1, 344_348. SCHMELING, H. 1988. Numerical models of RayleighTaylor instabilities superimposed upon convection. Bulletin of the Geological Institutions of the University of Uppsala, 14, 95-109. TAKAHASHI, T. 1991. Debris Flow. A.A. Balkema, Brookfield,Vt Talbot, C. 1977. Inclined and asymmetric upward-moving gravity structures. Tectonophysics, 42,159-181. TAYLOR, G. 1950. The instability of liquid surfaces when accelerated in a direction perpendicular to their planes, I. Proceedings of the Royal Society, 201, 192-196 THOMPSON, J.F. 1985. Numerical Grid Generation Foundations and Applications. Elsevier, Amsterdam.
TURCOTTE, D.L. & SCHUBERT, C. 1992. Geodynamics; Applications of Continuum Physics to Geological Problems. John Wiley, New York. VAN KEKEN, P.E., SPIERS, C.J., VAN DEN BERG, A.P & MUYZERT, EJ. 1993. The effective viscosity of rocksalt: implementation of steady-state creep laws in numerical models of salt diapirism. Tectonophysics, 225,457-476. WALTON, E.K. 1956. Limitations of graded bedding and alternative criteria of upward sequence in the rocks of the Southern Uplands. Transactions of the Geological Society Edinburgh, 16, 262-271. WHITEHEAD, J.A. & LUTHER, D.S. 1975. Dynamics of laboratory diapir and plume models. Journal of Geophysical Research, 80, 705-716. WOIDT, W.D. 1978. Finite element calculations applied to salt dome analysis. Tectonophysics, 50,369-386. YAO, M. & McKiNLEY G.H. 1998. Numerical simulation of extensional deformations of viscoelastic liquid bridges in filament stretching devices. Journal of Non-Newtonian Fluid Mechanics, 74,47-88.
The Vocontian clastic dykes and sills: a geometric model O. PARIZE1 & G. FRIES2 1
Ecole Rationale Superieure des Mines de Paris, CGES - Sedimentologie, 35 rue Saint-Honore, 77300 Fontainebleau, France (e-mail: [email protected]) 2 Institut Franc, ais du Petrole, 1&4 avenue de Bois-Preau, 92500 Rueil-Malmaison, France (e-mail: [email protected]) Abstract: Distal Aptian-Albian deep water channelled massive sands of the Vocontian Basin (SE France) are often associated with sand injections. The Bevons and Rosans areas in the Vocontian domain present probably the most spectacular outcrops showing complex networks of clastic sills and dykes injected into a thick marly/limy succession. Most injections are found in the channel banks, fed laterally from sandy channels. The sills are up to 10 metres thick in the vicinity of the connection with the channel feeder; they thin out and die into marls 2 or 3 kilometres away from it. Most dykes are injected from the sills rather from the channel itself: a few small dykes can be found under the channel fill. They are most abundant within a few hundred metres of the channel. Today, injections extending downwards from sills have up to 275 metres vertical extent, whereas injections extending upwards from sills never reach the contemporaneous palaeo-sea floor. Ptygmatic folding of the dykes by mechanical compaction indicates the amount of local post-injection compaction of shale and clearly shows that sand injection occurred prior to burial. Outcrop mapping shows that channel bank fracturing is contemporaneous with channel infilling. This is evidence of early syndepositional injection of the sandy material. Vocontian clastic injections provide good geometrical analogues to deep offshore clastic injectite networks and the opportunity to better understand genetic processes.
Clastic injectites form complex reservoir sandbodies. To date it has proven difficult to draw realistic models from subsurface data even if the seismic data are calibrated by cores (Jenssen et al 1993; Me Leod et al 1999; Rigollet 2001). The two most important problems are directly related to the geometry of these complex sandbodies as follows: (i) the difference between the small-scale observation from core and the large-scale interpretation from seismic: often the small scale (centimetre- to decimetre-scale) injectites described in core cannot explain decametre- to hectometre-scale injectite features observed at seismic scale; and (ii) the under-importance given to bedded injectites or sills and the lack of realistic models of dyke and sill networks. The study of outcrop examples at the same scale as injectites observed on seismic reflection data can help improve the interpretation of seismic data and advance our understanding of the geometry of injectite network. A world-wide screening of known injection outcrops produced only a few well-exposed outcrops that met these criteria: the Tourelle Formation in North America, Alpine Flysch and the Vocontian AptianAlbian succession in SE France (Gottis 1953; Dzulinski & Radomski 1956; Rutten & Schonberger 1957; Hiscott 1979). The Bevons and Rosans areas, located in SE France (Fig. 1), provide two of the best outcrops because: (i) the injections can be described individually and at network-scale; (ii) their relations with the feeder are observed directly in the field; and (iii) they can be integrated with the
morpho-sedimentary setting of the Vocontian Basin. Since the early 1980s, Ecole des Mines de Paris have used the Vocontian injectites and associated outcrops to constrain mechanical and stochastic simulations of sandy injections and associated fracturation of marly-shaly formations. Because the Vocontian outcrops are large, continuous and well exposed, the analysis of the injectites in the Bevons and Rosans areas allows the characterisation of the complete complex sandbody from the feeder to the distal fingers of the injection network, and consideration of the timing of sand injection (Aboussouan 1963; Beaudoin & Fries 1982 1984; Beaudoin et al. 1983 1985; Fries etal 1984; Fries & Beaudoin 1985; Fries 1987; Parize et al 1987; Parize 1988). The present paper details the main results of the field analysis of these Vocontian per descensum (downwards propagating) clastic injectites.
Geological setting The Vocontian clastic injectites cannot be studied as an isolated or individual sandbody because: (i) the sills and dykes form part of a complex network; (ii) they are injected into a sedimentary host formation; and (iii) they are connected with turbiditic channelled massive sandbodies. The knowledge of the morpho-sedimentary setting of the Vocontian Basin is a necessary preliminary step to the analysis of the injection network.
Fig. 1. General setting of the field observations, (a) Palaeogeographical setting of the Vocontian domain; (b) Location of Bevons (BE), Lesche-en-Diois (LD) and Rosans outcrops.
Morpho-structural setting Bevons and Rosans are located in the Diois and Baronnies areas, in the Vocontian Basin (SE of France), in the external part of the Western Alps (Fig. la). The complete sedimentary succession, from the basal Hercynian unconformity to uppermost Miocene deposits, is 12 kilometres thick (DebrandPassard et al. 1984). During the Mesozoic era, siliciclastic sedimentation was restricted to the Trias and Aptian-Albian intervals (Debrand-Passard et al 1984; Ferry 1984; Beaudoin & Fries 1984). The structural style of the Diois and Baronnies areas is of a box fold type and is characterized by east-west fold axes, narrow anticlines and large synclines (Flandrin 1963). From the Berriasian to the early Cenomanian, this area was part of the northern margin of the Valais ocean (Dercourt et al 1993). Today it forms a rare and a well preserved part of this palaeo-passive margin with two contrasting domains (Fig. la): (i) a marly sandy to sandy marlydominated shelf domain, which crops out largely in Dauphine, Vivarais and Provence; and (ii) a marly dominated pelagic domain, the Vocontian Basin, which crops out in the Diois and Baronnies and corresponds to the upper slope domain of the margin (Fries 1987; Rubino 1989). The transition zone between the shelf and the
upper slope domain is well defined because the facies of the two domains are very different; this shelf-break built out progressively during the early Cretaceous culminating in the deposition of the 'Urgonian' Formation, during the Barremian to early Aptian. The basin morphology at this time was characterized by a well marked shelf break with a bathymetric relief of up to 250 metres (Amaud-Vanneau et al. 1979). During the Middle Aptian a maximum depth of approximately 1000 metres was reached in the Gap area (Fig. Ib; Guerin 1981; Fries 1987; Breheret 1997). From late Aptian to Middle Albian, the regional tectonic setting became transpressive and then compressive; this is the time of 'Durancian isthmus' (Fig. la; Masse & Philip 1976). As a result, during the Albian there was less contrast in morphology along the margin and it corresponded to a ramp setting (Rubino 1989). The bathymetric depth decreased to less than 500 m in the Rosans area (Guerin 1981; Fries 1987; Breheret 1997) The regularity of the pelagic sequence is considerably disturbed in the Vocontian area by the interstratification of various mass movement deposits (Figs 2 & 3). These various gravity-controlled deposits are grouped under two headings: 'sandy gravity flow deposits' and mass movement complexes, which includes slumps, debris flow deposits or debrites and pebbly mudstones. Some can be correlated up to 80
VOCONTIAN CLASTIC DYKES AND SILLS
53
Fig. 2. The Bevons area, (a) Lithostratigraphical log of the Aptian-Albian succession in the Bevons area; (b) Close-up of some upper Albian key beds (KB) at Le Puy Hill section. The clastic sill changes of stratigraphic horizon when connecting the vertical dykes, whereas the limy key beds remain unaffected. Two main directions of dykes are visible on this photograph: the first on is perpendicular to the beds (dyke 1); the second is parallel to the outcrop (dyke 2). The marly-limy succession is cut into decametre scale blocks by the horizontal and vertical fractures filled with sand (sills and dykes).
kilometres from the shelf edge (Rubino 1981, 1989; Fries 1987). The mapping of each gravity-driven body demonstrates that their maximum thickness axis and the channel axis are superposed and that they are today superposed with the synclinal axis (Rubino 1981; Fries 1987; Parize 1988; Joseph et al 1989). This confined setting explains the channelled expression of the turbiditic systems. Field observations and isopach mapping of the different mass movements show thickness diminution from the present-day synclinal axis laterally towards the anticline flanks (Fries 1987; Parize 1988; Rubino 1989). These observations point to a continuous topographic control of the sedimentation during Aptian and Albian interval (Scott & Tillman 1981). For these reasons, the Vocontian domain is a good field analogue for subsurface examples in similar geodynamic settings (Bouma 1982; Apps et al. 1994; Rigollet 2001). These erosive, transport and depositional axes define palaeo-valley systems. During the Aptian, the main active palaeo-valley system was located in the NW corner of the Vocontian Basin. The Bevons and Rosans areas are located on a palaeo-valley axis, up to 40 kilometres from the shelf-break (Figs Ib and 4)
Lithostratigraphic succession Clastic dykes were first described in the Vocontian Basin from the Serre Chaitieu outcrop near Leschesen-Diois (Rutten & Schonberger 1957; location on Fig. Ib) and the Puy hill near Bevons (Aboussouan 1963). However, clastic sills were not described until the 1980s from the Rosans (Beaudoin et al. 1983; Fries etal. 1984; Fries & Beaudoin 1985) and the Bevons area (Beaudoin & Fries 1984; Beaudoin et al. 1985). A good understanding of the lithostratigraphic setting is necessary in order to distinguish these sills from depositional beds. The host formation of Vocontian clastic injectites is the Aptian-Albian Blue Marls Formation, up to 800 metres thick (Flandrin 1963; Fries 1987; Breheret 1997). Its depositional environment is pelagic, without storm or tide influence. The Blue Marls Formation is typically composed of 50% carbonate, 30% clays and 15% quartz, with the remainder consisting of accessory minerals such as pyrite and feldspar (Aboussouan 1963). In lithologic terms, the CaCO3 content of a marl is less than 15-20% lower than that of a limy bed (Parize 1988).
54
O.PARIZE&G. FRIES
Fig. 3. The Rosans area, (a) Lithostratigraphical log of the Aptian succession in the Rosans area. Note the abundance of mass movements (slumps) and more specifically the 'Grand Slump' ('Large Slump') at the base of the Fromaget beds; (b) Close-up of upper Aptian Fromaget bundle and its key-beds (1: top of 'Grand Slump' correspond to a sandy turbidite; two, three, four, five, six and seven are decimetre thick bioturbated limestone beds; X3 to S7 correspond to metre thick slumps; dashed horizontal lines correspond to the different stratigraphic levels where clastic sills have been observed: they generally correspond to local mechanical unconformities).
The Vocontian clastic sills and dykes are injected mainly into the upper Albian interval in the Bevons area (Figs 2 and 5 ) and the uppermost Aptian interval in the Rosans area (Figs 3 and 6); the corresponding channel feeder are respectively of uppermost Albian and uppermost Aptian age. This AptianAlbian interval is organised into eight depositional sequences (Fries & Rubino 1990). The feeder system corresponds in both cases to lowstand deposits, and the sandy injection takes place laterally then downward from the palaeo-sea-bottom into the older sequence. A robust lithostratigraphic knowledge, based on lithologic markers as key-beds (KB) and various and numerous biostratigraphic data allows definition of a high-resolution geometric framework described below (Fries, 1987). In the Bevons area, the upper Albian Blue Marls Formation is composed of decimetre thick marly-limy to limy beds interbedded with marly intervals organised into 10 to 20 m thick, lime-enriched upwards sequences (Fig. 2). Each bed and each sequence can be mapped over 200 km2 in the Sisteron area (Fries 1987) and form a geometric reference for the clastic injectites in this area.
In the Rosans area, the upper Aptian 'Fromaget' 15 m thick limy bundle (Fig. 3) is composed of decimetre-thick limestone beds and marly intervals. Each key bed can be mapped over 5000 km2, i.e. the whole surface cropping out today in the Vocontian area (Fries 1987). The "Fromaget" bundle will be used to analyse the clastic injectites in the Rosans area (Figs Ib and 4). Several yellowish smectitic layers a few centimetres thick have been mapped over a large area and also used as marker beds. Examples are the Cezanne level (Dauphin 2002) in the Fromaget bundle or the Bonnard level (Fries and Parize unpubl.) in upper Albian interval (Figs 2 & 3). They are interpreted as alteration products of volcanic ash beds (Dauphin 2002).
Location of clastic injections The clastic injectites are associated with massive sand systems that head at the shelf-break (Parize 1988; Rubino & Parize 1989). These sand systems fill erosive channels 500 to 1000 metres wide and 20
VOCONTIAN CLASTIC DYKES AND SILLS
55
Fig. 4. Palaeomorphological organization of the upper slope Vocontian environment and 3D blocks of Rosans and Bevons areas, (a) Western part of the Vocontian Basin during Aptian and Albian; dashed area represents its margin (shelf break) whereas the white domain correspond to the pelagic realm; (b) and (c) The upper Aptian turbiditic system in the Rosans area. Solid line: main turbiditic channel axe; dashed line on (b): the Risou fault; Vaucluse: synsedimentary anticline, (d) and (e) The upper Albian turbiditic systems in the Sisteron and Bevons area. Solid line: main turbiditic channel axis; dashed: shelf break.
to 40 metres deep; their filling are mono- or multistoried: the thickness of the lower sandy section is up to 15 metres. The clastic injections occur in the distal parts of the turbidite depositional systems (Fries 1987; Parize 1988). The mapping of the injectite network and its feeder shows that the location of the injectites is controlled by the depositional location of the feeder. This location is related to a topographic change of the palaeo-seafloor, as proposed by Kneller (1995), such as a decrease in the slope or widening of the palaeo-valley in front of a syndepositional anticline (Fig. 4). The clastic injections appear as an architectural element, sensu Mutti & Normark (1987, 1991), of these massive sand systems. In the Vocontian area, per descensum (downwards propagating) clastic injections are always associated with thick massive sands filling deep erosive channels. This relationship is also observed in the Numidian flysch (Parize & Beaudoin 1986, 1988; El Maherssi 1992; Parize et al. 1999) and has been inferred in other places such as the Tourelle Formation (Hiscott 1979; Hiscott & Middleton 1979).
Field observations The main objective this study is to propose a realistic model of the geometry of the injectites and associated sand bodies. Although the Vocontian outcrop conditions are excellent, it is necessary to develop a field acquisition methodology that is adapted to their complex geometry and will allow comparison of the injections in different locations (Fig. 7). In the Bevons area, it is possible to walk 400 m along an 8 to 12 metre thick sill (Fig. 5) and to observe 3 metre thick dykes. The field conditions of the Rosans area permit continuous observations of the injections-feeder complex along numerous strike and dip cross-sections up to 5 kilometres long (Figs 6 and 8). A close geometric relationship exists between the sills and the dykes (Figs 2a, 5,7 and 8): they can define complex metre to tens of metrescale injected boxes near the feeder where sills are well developed. When the geometric characters of Vocontian injections are organised according to their relations to the host formation, it can be seen that dykes and sills constitute the same injected body.
Fig. 5. Le Puy hill outcrops near Bevons: close-up view of the injectites-feeder complex. Note on both the photograph and the sketch that the horizontal bedding is cut by the vertical and zig-zag trajectories of the clastic dykes (solid lines on the sketch).
Fig. 6. Saint-Andre-de-Rosans outcrops near Rosans: close-up view from the injectite network to the feeder channel. The thick sills parallel to the bedding are difficult to identify as clastic injection from a brief and too distant inspection.
58
O.PARIZE&G. FRIES
Fig. 7. 3D field acquisition: an example of clastic sills and dykes in the Rosans area. The complex geometry requires at least a plan view and two side views (location on Fig. 8).
Dykes In outcrop, clastic dykes look like large walls cutting across the host marly-limy formation (Fig. 9a and b). If their course is parallel to the outcrop trend, they can appear to have a lenticular shape and can be mistaken for channel features. Therefore, an accurate 3D drawing is essential (e.g. Fig. 7). The mean thickness of dykes is 30 centimetres, but frequently reaches 50 cm in the vicinity of the feeder and as little as a few centimetres or millimetres in their distal basal part. The observed horizontal length is up to 300 metres but in the Bevons area, mapping suggests that dyke segments reach up to 8 kilometres in length (Parize 1988): these dimensions are similar to other described examples (e.g. Diller 1889; Marschalko 1965; Peterson 1966; Smyers & Peterson 1973; Parize & Beaudoin 1986; El Maherssi 1992). Finally, the vertical continuity of the dykes is very important; in the Bevons area they cut down 275 metres into the surrounding marls from the feeder channel. The dykes form various patterns from straight to winding and are frequently branching (Figs 9 and 10). A main dyke is frequently associated with subparallel off shots or minor dykes; at a given stratigraphic level, its width can vary rapidly. The dykes are organised in dykes sets or swarms; when dykes of different sets intersect, one trend is generally shifted relative to the other (Fig. lOa). Tangential
to almost perpendicular connections have been observed (Figs lOb and c). In the Bevons area, the outcrop conditions allow analysis of the vertical evolution of dyke away from the feeder. No preferential direction can be observed near the channel feeder in the first 10 metres below the palaeo-seafloor; the density of dykes increases in the vicinity of synsedimentary faults (Fig. 11; Beaudoin & Fries 1984; Eckert 2000). In the deeper part of the section, the patterns are more closely related to local syn-sedimentary faults and the deepest dykes are parallel to escarpment faults present in the BarremianBedoulian limy bedrock (Beaudoin et al. 1986; Joseph etal. 1987). Dyke geometries vary at a variety of scales. Commonly dykes are planar and vertical, but can also exhibit oblique dips and zig-zag trajectories (Fig. 9; Gottis 1953). These changes can be related to such factors as lithological contrasts at bedding surfaces, particularly between marl-limestone contacts, bed thickness and the dip of bedding (Fig. 12). The control by lithology on dyke geometry indicates some degree of compaction and differentiation of the mechanical behaviour of different lithologies at the time of sand injection. The sandy dyke infill is massive, however subtle laminations are locally observed near its walls. These walls can be regular or contain crenulation or crumpling related to mechanical compaction (Fig. 13). The boundary between clastic dykes and
VOCONTIAN CLASTIC DYKES AND SILLS
59
Fig. 8. An example of a 600 metre long sill-dominated network in the Rosans area (1, injected sandstones; 2, mixture between injected sandstones and host formation; 3, turbiditic sands; 4, 'Grand Slump'; 5: the Fromaget bundle and the 3 and 5 key beds; 6: post depositional faults).
surrounding beds can be associated with ptygmatic folds (Figs 7, 12 and 13): the bedding seems unaffected or only slightly upturned close to the dyke; the upturning is most marked adjacent to thocker and more erected dyked, the so-called pillar- or prodeffect (Fig. 7). It is important that these small-scale features (folds, crenulations, etc.) are not mistaken with flow indicators. All these features are vertical deformation features and are related to compaction deformation after sand injection into the host formation. They can be used to calculate a compaction ratio (e.g. Andrieux 1967; Borradaile 1977; Hiscott 1979; Beaudoin & Fries 1982; Pinoteau 1986; Beaudoin et al. 1987; Truyol 1991) and to estimate a palaeoporosity curve ^(z) (Schneider & Parize 1989).
Sills Clastic sills are bedded, massive, structureless, centimetre to several metre thick sand bodies and can easily be mistaken for depositional sandstone sheets (Fig. 6). Sills are distinguished from true beds deposited on the palaeo-seafloor due to abrupt strati-
graphic jumps or steps (Fig. 14) in the surrounding marls and also sudden lateral changes in thickness (Fig. 8), often in association with off-shoots (sensu Truswell 1972) or dykes. These characteristics correspond to those previously defined by Strickland (1840), Dzulinski & Radomski (1956), Truswell (1972) and Hiscott (1979). The mean thickness of the sills is one metre but the maximum observed thickness of an individual sill is up to 12 metres. The cumulative thickness of injected sand for cogenetic sills is up to 10 metres. Sills walls are sharply defined, regular and flat (Figs 8 and 14). The heading wall may contain frondescent-type casts, but never primary sole-marks such as flute or groove-casts. The hanging wall is very regular and never shows biologic colonization. True 'dish and pillar' features (Lowe 1975) within the sill are observed in only two places in the Rosans area. Sills can be mapped up to 2.5 kilometres away from the channel feeder in the Rosans area (Fig. 15). A detailed volumetric study conducted in the SaintAndre-de-Rosans region (approximately 10 km2) indicates that the injected sand volume in the southern part of the network corresponds to 8 X106 m3 and the channel volume to 35X106 m3, so slightly less
60
O.PARIZE&G. FRIES
Fig. 9. An example of a large dyke and location of the close-up views (90 metres under the feeder), (a) and close-up of (a) and (b): relationships between the main dyke and the host formation and branching offshoot; (c) and (d) Various shapes of the dykes with straight to winding trends.
VOCONTIAN CLASTIC DYKES AND SILLS
61
Fig. 10. Examples of dyke patterns (explanations in the text), (a) vertical view; (b) and (c) plan view.
Fig. 11. Relationships between dyke density and Albian fault. 1: main synsedimentary faults; 2: dykes; 3: tabular sands (channel fills and sills); keybed 3: fossiliferous bioturbated limestone from the Fromaget bundle (upper Aptian); keybed 31: metre thick clastic sill injected above the Bonnard ash level (upper Albian).
62
O. PARIZE & G. FRIES
Fig. 12. Relationships between dyke dip and the lime bundle or beds (130 metres under the feeder). GTl vertical dyke is partly behind the tree. Ptygmatic folds by dyke GT2 absorb post depositional compaction, whereas dyke GTl exhibits more crenulation and cataclasis.
Fig. 13. Various observed deformation patterns of vertical parts of dykes, (a) Crenulation and cataclasis zones; (b) differential behaviour related to dipping variation; (c) conceptual model from outcrop data. The thinner the dyke, the more intense the ptygmatic folds; the thicker the dyke, the more intense crenulation and the cataclasis.
VOCONTIAN CLASTIC DYKES AND SILLS
63
Fig. 14. Stratigraphic jump of a sill and example of an apparently upwards propagating dyke (see location on Fig. 8; key beds 5, 6 and 7: limestones from the Fromaget bundle).
than 20% of the sand volume is injected. These dimensions are similar to those described in subsurface examples, e.g. Balder field (Jenssen et al. 1993) and Alba field (Newton & Flanagan 1993; McLeod etal 1999). Injectite network The injectite network, i.e. the continuous injected sand body, is composed of both dykes and sills (Figs 8 and 11). A typology of injectites according to their location can be proposed (Fig. 16). Within the adjacent marls, a sill-dominated network is developed from which downwards and upwards propagating dykes develop. The upwards-propagating dykes never cross the palaeosea floor (Fig. 16a) and their thickness decreases abruptly within a few metres of it, as observed in Rosans area. A good geometric characterisation of sills is necessary to correctly analyse the significance of upwards propagating dykes (Fig. 14) and to not mistake them with per ascensum, post depositional, dykes. At a small scale, the Stratigraphic arrangement of each geometric facies is well predicted: (i) the silldominated network is located above the dykedominated network and; (ii) some dykes developed upwards from the sills (but never cross the palaeosea floor), but most numerous dykes develop downwards from the sills; and (iii) only thin and small dykes are observed under the feeder (Fig. 16). The lithology of the host formation controls the injectite network geometry (Fig. 16). In the Bevons area where both dykes and sills are well represented (dykes reaching up to 275 metres from the feeder are
organised into several swarms), the host formation is a thick interbedded marl and limestone succession (Fig. 2). In the Rosans area, the injection network, which is dominated by sills, developed within a marl-lime decametre-scale bundle. This formation covers a very large, thick slump, the 'Grand Slump' (Fig. 3; Fries 1987), the top of which corresponds to a major unconformity that may have limited the vertical development of dykes. Petrographic characters of the injectites At a medium scale observation, two types of sandstone bank morphology can be described: (i) very regular, blocky and indurated and (ii) poorly indurated. The first type describes all well cemented sandstone beds or injections whose thickness ranges from few centimetres to several metres; the second type is frequently called 'safre', a vernacular term used in Diois and Baronnies to describe all 'smooth' sandstones whose thickness is up to 5 metres. It is possible to observe a hardness-evolution between the two facies: 'safre' facies, then indurated sandstones, then host-rock. Petrographic analysis reveals the evolution of the sandy injected material from time of deposition until present day and allows comparison with the feeder material. At the moment of injection or just after it, the petrographic compositions of the injection and feeder was similar. The average composition of the Aptian-Albian Vocontian sandstones, both turbiditic beds and clastic injections, is very homogeneous comprising 60-70% quartz grains (100-300 urn), 20-30% detrital glauconite grains (100-200 urn)
64
O.PARIZE&G. FRIES
Fig. 15. The upper Aptian turbiditic system: schematic map from the Rosans area showing the massive channel sand system and associated injectites. The split of the main channel is most likely caused by the Vaucluse anticline, which was a palaeo-seafloor high. Note the large extension of the sills laterally from the channel feeder. (Sar: Saint-Andre-deRosans).
Fig. 16. The relationships between the injectites and their feeder: schematic models of the present day observations in the Rosans and Bevons areas. In the Bevons model, just after injection, the depth of penetration of the vertical injections was 300 to 400 metres, reduced today after post depositional compaction to 200-250 metres.
VOCONTIAN CLASTIC DYKES AND SILLS
and minor siliceous and carbonated shells debris, some feldspar, detrital mica and rare tourmaline and zircon grains. The first diagenetic step corresponds to mechanical compaction. During this phase, the soft glauconite grains are deformed and then intruded between quartz and feldspar grains. Micas can also be deformed, squeezed out and punched by rigid grains (Diller 1889). The development of silicification corresponds to the second step of diagenetic evolution. The micas and glauconite grains are silicified, with the development of microquartz. This silicification is important in the feeder and can explain the 'safre' diagenetic facies, which corresponds to a cohesionless facies. The final observed diagenetic step is carbonatation/sparitization. This appears to be the most important diagenetic event and can explain the epigenesis of glauconite grains and siliceous cement; the microsparry transformation of primary grains such as quartz, feldspar and detritic mica is rare. This carbonatation is well marked along injectite or channel walls, while the channel itself is unaffected if its thickness is up to 3 to 4 metres. Around calcareous clasts, a spherical diagenetic envelope is developed and explains the doggers in the channel infilling or in some sills; some other examples show polyhedral limy blocks surrounded by hardened diagenetic sandstone. These diagenetic features cut across the primary structures or textures.
Geometric and genetic model Vocontian clastic sills are located in the vicinity of massive turbiditic systems (Figs 5, 6 and 16). The continuity between the injection network and the feeder can be observed directly at the outcrops (Figs 5 and 6; Beaudoin & Fries 1984; Parize et al 1987). The injection feeder corresponds to massive sands filling erosive channels. In cross-section, feeders display an almost rectangular section with the channel walls often vertical. This shape is related to the strongly erosive process that formed the channel; the vertical extent can be increased during the sand filling as large undermining seems to suggest.
Relations between channel feeder and injectite network The geometry of the channel filling is well characterized at the outcrop. In some cases, because of the relationship between dyke orientation and the present day outcrop surface, dykes can appear to be lenticular. It is easy to mistake this topographic effect with a lenticular-shape channel section and the crenulation observed on the dyke wall with primary sedimentary features (Fig. 9). A 3D field
65
approach, as detailed in this study can prevent this mistake. The channelled feeder is always an erosive body; the contact between the walls of this depositional body and its enclosing marly beds is sharp (Fig. 16). The basal and upper contacts are regular and flat. Some chondrites bioturbations or burrows are observed, but current marks such as flute or groove marks are rare. Scourings are located at the bottom of the banks; some of these are metre- to decametre-deep and these indicate turbulent depositional conditions. Rare and thin dykes can be locally observed at the base of the channel fill. The injection network is connected with the feeder from the channel walls and often near the base of these walls. These Vocontian field observations conform to rare previous published data (e.g. Bielenstein & Charlesworth 1965).
La Baume outcrop An extensive outcrop on the southern flank of Le Puy Hill (Fig. 1) allows access to horizontal and vertical sand injectites and a feeder complex (Figs 17a and b). The feeder corresponds to a large scour filled by one massive turbidite. The scouring occurred due to a hydraulic jump within the high-density turbulent flow, linked to movement on a main synsedimentary fault (Fig. 17a; Beaudoin & Fries 1984; Fries 1987; Parize et al. 1987). The feeder is 15 metre thick at its apex and its maximum width is 160 metres. Its arcuate flank is vertical near the apex. Current casts indicate a flow to NE, in accordance with the mapping of the scour fill. This feature is comparable in size and shape with modern turbidite channels (Parize et al 1989; Mutti & Normark 1987, 1991; Oilier et al. 1998). Near the apex of the scour, the vertical surface supports numerous and various flute casts, simple, corkscrew-like, or spiralling, which indicate horizontal or upward-eddying currents (Beaudoin et al. 1985; Parize 1988). Five major injections (perpendicular to low angle, dyke-like, 3 metres thick) are connected with the scour fill (Fig. 17b). Such a thickness is the maximum recognized in the Vocontian domain. These injections rapidly evolve into metre- to several metre thick sills or metre thick dykes. The banks are fractured, indicating that sand injection took place during the channel feeder infilling, that is, the sand injection is contemporaneous with the deposition of the massive turbiditic sediments (Beaudoin etal. 1985).
Vocontian injectite emplacement The clastic injection is an instantaneous event related to the 'en masse' deposition of a massive
66
O.PARIZE&G. FRIES
Fig. 17. Evidence of syndepositionnal sand injection, (a) Mapping of upper Albian turbiditic-channelled systems in the Bevons area. Large faults controlled the deposition of the successive channel systems, (b) The La Baume outcrop: the feeder exhibit a large scouring feature, linked to an hydraulic jump related to a syn-sedimentary fault. Large sills and thick dykes are injected near the apex of the scour, demonstrating that deposition and injection of the sand were directly related.
VOCONTIAN CLASTIC DYKES AND SILLS
67
turbidite sand in a erosive channel. Three successive stages can be defined: (i) erosion of the seafloor and deposition of a shaly succession immediately prior to the turbidite flow; (ii) 'en masse' deposition of the turbidite sand and formation of the injectite network and; (iii) post-depositional compaction to form the present day exposures. Stage 1: channel cutting (Fig. 18a). The submarine floor is cut by a sediment flow. The initial trigger may be attributed to sand liquefaction on the shelf during storms or tsunami (Imbert et al. 1995). This flow is associated with only thin overbank deposits and erosion of the superficial sediments. Stage 2: channel infilling (Fig. 18b). The sudden deposition of a sandy turbiditic flow induces fracturing of the bank and the injection into them. This deposition is inferred as 'en masse', characteristic of high density turbiditic flow sensu Lowe (1982) or granular flow sensu Mutti et al. (1999). This turbiditic event is associated with bank undercutting and undermining. The injection into the bank causes localised uplift, referred to as a lift effect. An equilibrium state is reached between the injected sand and the sand infilling the channel: this infilling cannot overtake the banks and the channel infilling and the injection cease at the same time. The uplift can induce a lateral warping from the edge of the bank. It can be associated with sliding or slumping similar to bank caving and breccia facies can occur locally. Stage 3: the today shape, after compaction (Fig. 18c). After burial and compaction, the network achieved its present day form. This shape is much like a channel-levee complex. On seismic data, this similarity in form could cause misinterpretation.
Conclusion The Vocontian area is a preserved upper slope domain of the northern passive margin of the Lower Cretaceous Valais Ocean and gravity-driven deposits dominate its sedimentary fill. During the Aptian and Albian time interval, deposition was siliciclasticdominated. The massive sand bodies and thin turbidite bed sets are confined within submarine valleys, 5 to 8 kilometres wide. Their deposition is related to topographic changes on the palaeoseafloor. The clastic dykes and sills are associated with massive erosive channel fill and are injected into the channel banks (Fig. 16). These per descensum injections are located in the distal part of these massive channelled turbiditic bodies (Fig. 15). The injectite body is massive, very similar to the feeder fill; no sedimentary structures are observed.
Fig. 18. Model for Vocontian injectite emplacement (explanations in the text). Stage 1: channel cutting and bypass; Stage 2: channel infilling, syndepositional injections and local bank sliding; Stage 3: post depositional compaction (computed rate of compaction: 2).
Its shape is directly controlled by the host formation lithology. The sills are characterized by particular geometric features which allow them to be distinguished from the sheet-type turbiditic beds: (i) stratigraphic jumps associated with dykes and (ii) sudden thickness variations, sometimes completely disappearing (Figs 8 and 14). The headwall of a sill can show local load-like and frondescent-like features; the hanging wall is more regular and is not affected by bioturbation. The maximum depth reached by the sills and the dykes under the base of the channel (Fig. 16) is 30 metres at Rosans and 275 metres at Bevons. In the Rosans area, injection perpendicular to the channel wall, reaches 2.5 kilometres away from the feeder. A
68
O.PARIZE&G. FRIES
Fig. 19. Conceptual model for Vocontian erosive massive sands and their associated injectite networks as architectural element, (a) Erosion and bypass by submarine currents; (b) Depositional phase starts with the arrival of massive turbiditic sandflow, clastic injection and lateral slumps are initiated; (c) Rapid 'en-masse' deposition and injection related to topographic changes; (d) Post depositional compaction and present-day geometry.
detailed volumetric study conducted in the same area indicates that the injected sand volume would be up to 30% of the total (feeder plus injectites) sand volume (Jenssen et al. 1993; Newman et al. 1993; Newton et al 1993; Dixon et al 1995; Rigollet 2001). Detailed study of La Baume outcrop (Fig. 17) demonstrates that the sand injection took place in the host formation during the channel filling. Therefore, these Vocontian clastic injectites and Vocontian-type injections, e.g. in the Numidian flysch and possibly in the Tourelle Formation, can be considered as architectural element, sensu Mutti & Normark (1987, 1991), of the massive turbiditic erosive channel fill (Fig. 19). The smectitic levels which are interbedded into the host formation are levels preferentially used by the sand injections because they are without cohesion: in Rosans areas, the Cezanne level in the main injected level and in Bevons area, the Bonnard level is the deepest and the most continuously injected level. A geometric model is proposed from field data (Fig. 16). This model takes into account the size and the shape of individual injectites and of the network, the close association between sills, dykes and
feeders and the facies of the host formation. A genetic model can be also inferred from analysis of these data and is supported by numeric simulations (Parize 1988; Eckert 2000; Eckert et al 2001; Parize et al 2001, 2002). This genetic model (Figs 18 and 19) for the Vocontian injectites begins with the formation of a channel by submarine currents. These bypass and erosive events do not deposit within the channel but result in laterally equivalent overbank deposits. The timing of this phase relative to subsequent events is not precisely established. This erosive phase is followed by the depositional phase of the massive turbiditic event. Deposition is possible because of topographic changes, such as the opening of a submarine valley or a decrease in slope. The 'en masse' deposition is rapid (Lowe 1982; Middleton & Hampton 1976; Mutti 1992; Mutti et al 1999) and contemporaneous with the continuing turbiditic flow. The sand injection into the adjacent marls is accompanied by bank uplift, which locally causes gravity destabilisations. As this turbiditc event is not a finite, short event but a complex event, the injection phase can last as long as the feeder sand moves. The final phase of this genetic model is the compaction deformation phase and the diagenetic formation of a channel-levee-type morphology.
VOCONTIAN CLASTIC DYKES AND SILLS These results have been obtained since 1983 within the framework of different research programs conducted by the Ecole des Mines de Paris with sponsorship from ANDRA, ELF, TOTAL and TOTALFINAELF. R. Bouchet, S. Eckert, Ch. Cabrol, S. Lalande, B. Paternoster, (Ecole des Mines de Paris), Y. Callec, C. El Maherssi, J. Maillart, S. Mostefai, B. Pinoteau, C. Rigollet, V. Truyol, (PhD students, Ecole des Mines de Paris), T. Bouchery (PhD student, Lille University), collaborated in the field work as part of their respective degree courses. La Baume 3D mapping was conducted with F. Schneider. J-Y Clement and M. Thiry supervised the petrographic analysis of numerous X-Ray diagrams and thin-sections. We are extremely grateful to them for their friendly availability and their scientific support. The authors give special thanks to B. Beaudoin, Ph. Crumeyrolle, F. Hadj-Hassen, P. Imbert, B. Pinoteau, J-L. Rubino, F. Schneider and M. Tijani for their advice and exchanges of opinion; special thanks are also owed to M. Berger, G. Gauthier, F. Leandri and Ph. Le Caer for their technical support. We thank all the participants of Vocontian field trips, workshops or training, in particular JM. Champanhet, P. Datillo, J-M. Fonck, Y. Grosjean, D. Laurier, Ph. Legrand, R. Martin, E. Ousset, J. Pouzet, F. Rodot, F. Temple (TOTALFINAELF), G. Oilier (CEE DG XII), N. Mavilla and R. Valentinetti (AGIP) for their discussions and constructive comments. We also express our gratitude to H. Loseth and J. Totterdell for their constructive reviews of this article and to C. Morley and P. van Rensbergen for their determinant support.
References ABOUSSOUAN, M.T. 1963. Contribution a 1'etude sedimentologique et paleoceanographique des terrains aptiens - albiens de la Provence aux Baronnies. These Doctorat Oceanographie Universite Marseille. Recueil des Travaux de la Station Marine d'Endoume (Marseille), 30, 95-164. ANDRIEUX, J. 1967. Etude de quelques filons intraformationels du fly sen albo-aptien des zones externes du Rif (Maroc). Bulletin de la Societe geologique de France, 9, 844-849. APPS, G.M., PEEL, F.J., TRAVIS, CJ. & YEILDING, C.A. 1994. Structural controls on Tertiary deep water deposition in the Northern Gulf of Mexico. In: WEIMER, P., BOUMA A.H. & PERKINS B.F. (eds) Submarine Fans and Turbidite Systems. 15th Annual Research Conference, Gulf Coast Section of Society of Economic Paleontologists and Mineralogists Foundation, 1-7. ARNAUD-VANNEAU, A., ARNAUD, H., CHAROLLAIS, J., CONRAD, M-A., COTILLON, P., FERRY, S., MASSE, J-P. & PEYBERNES, B. 1979. Paleogeographie des calcaires urgoniens du Sud de la France. Geobios Lyon, Memoire special, 3, 363-383. BEAUDOIN, B. & FRIES, G. 1982. Les filons greseux sedimentaires per descensum dans un systeme de fractures ouvertes. Le cas de FAlbien de Bevons (Alpes-de-Haute-Provence). Comptes-Rendus de VAcademie des Sciences de Paris, 295, 385-387.
69
BEAUDOIN, B. & FRIES, G. 1984. Phenomenes de resedimentation dans le Cretace inferieur subalpin. In: 5th Regional European Meeting of Sedimentology, Marseille, field-trip Guidebook, 6,51 pp. BEAUDOIN, B., FRIES, G., JOSEPH, PH., BOUCHET, R. & CABROL, C. 1986. Tectonique synsedimentaire cretacee a 1'Ouest de la Durance (S.E. France). ComptesRendus de VAcademie des Sciences de Paris, 303, 713-718. BEAUDOIN, B., FRIES, G., JOSEPH, PH. & PATERNOSTER, B. 1983. Sills greseux sedimentaires injectes dans 1'Aptien superieur de Rosans (Hautes-Alpes). Comptes-Rendus de VAcademie des Sciences de Paris, 296,387-392. BEAUDOIN, B., FRIES, G., PARIZE, O. & PINAULT, M. 1985. L'origine des injections sableuses: les sills et les dykes albiens du Ravin de la Baume, Bevons (Alpes de Haute-Provence). Comptes-Rendus de VAcademie des Sciences de Paris, 301,407-410. BEAUDOIN, B., COJAN, I., FRIES, G., MAILLART, J., PARIZE, O., PINAULT, M., PINOTEAU B. & TRUYOL, V. 1987. Mesures directes de la compaction dans les sediments. Notes et Memoires de TOTAL Compagnie Francaise des Petroles, 21, 235-247. BlELENSTEIN, H.U.
& CHARLESWORTH, H.A.K.
1965.
Precambrian sandstone sills near Jasper, Alberta. Canadian Petroleum Geology Bulletin, 13,405-408. BORRADAILE, G.J. 1977. Compaction estimate limits from sand dykes orientation. Journal of Sedimentary Petrology, 47,1598-1601. BOUMA, A.H. 1982. Intraslope basins in Northwest Gulf of Mexico: a key to ancient submarine canyons and fans. In: WATKINS, J.S. & DRAKE, C.L. (eds) Studies in continental margin geology. American Association of Petroleum Geologists, Memoir, 34,567-581. BREHERET, J.G. 1997. L'Aptien et 1'Albien de la fosse vocontienne (des bordures au bassin). Evolution de la sedimentation et renseignements sur les evenements anoxiques. These de Doctorat es Sciences de 1'Universite de Tours, 1995. Publication de la Societe geologique du Nord, 25, 614 pp. DAUPHIN, L. 2002. Litho-, bio-, chrono-stratigraphie comparees dans le bassin Vocontien a 1'Aptien. PhD Thesis, Ecole nationale superieure des Mines de Paris. Memoire des Sciences de la Terre de VEcole des Mines de Paris, 516 pp. Distributed by Editions du Bureau de Recherches geolologiques et minieres, Orleans, France. DEBRAND-PASSARD, S., COURBOULEIX, S. & LIENHARDT MJ. (eds) 1984. Synthese geologique du Sud-Est de la France: Stratigraphie et paleogeographie. Memoire du Bureau de Recherches geolologiques et minieres, 125, 614pp. DERCOURT, J., Ricou, L.E. & VRIELYNCK, B. (eds) 1993. Atlas Tethys palaeoenvironmental maps. 307 pp., 14 maps, 1 pi., Gauthier-Villars PubL, Paris. DILLER, J. S. 1889. Sandstone dikes. Geological Society of America Bulletin, 1,411^42. DlXON, R.J.,
SCHOFIELD, K., ANDERTON, R., REYNOLDS,
A.D., ALEXANDER, RWS., WILLIAMS, M.C. & DAVIES, K.G. 1995. Sandstone diapirism and clastic intrusion in the Tertiary submarine fans of Bruce-Beryl Embayment, Quadrant 9, UKCS. In: HARTLEY, A.J. & PROSSER, DJ. (eds) Characterization of Deep Marine
70
O.PARIZE&G. FRIES
Clastic Systems. Geological Society, London, Special Publications, 94,77-94. DZUTYNSKI, S. & RADOMSKI, A. 1956. Clastic dikes in the Carpathian Flysch. Annales de la Societe geologique de Pologne, 26,226-264. ECKERT, S. 2000. Modelisation geometrique et mecanique de la fracturation d'un massif argilo-carbonate. Etude des filons greseux sedimentaires albiens de Bevons (Alpes-de-Haute-Provence). Rapport d'Option des Sciences de la Terre de 1'Ecole nationale superieure des Mines de Paris, 151 pp. unpubl. ECKERT, S., PARIZE, O., HAJ-HASSEN, F., TUANI, M., BEAUDOIN, B., HOTEIT, N. 2001. Numerical simulation of the early fracturation in a heterogeneous shaly/limy massif. Abstract book Origin, Processes and Effects of Subsurface Sediment Mobilisation on Reservoir to Regional Scale, September 2001. Gent, Belgium, p. 8. EL MAHERSSI, C. 1992. Dynamique de depot du flysch numidien de Tunisie (Oligo-Miocene). PhD Thesis, Ecole nationale superieure des Mines de Paris et Universite de Lille I. Memoire des Sciences de la Terre de 1'Ecole des Mines de Paris, 15, 246 pp. Distributed by Editions du Bureau de Recherches geolologiques et minieres, Orleans, France. FERRY, S. 1984. Apports detritiques dans le Bassin vocontien. In: COTILLON, P. (ed.) Cretace inferieur. Synthese geologique du Sud-Est de la France. Memoire du Bureau de Recherches geolologiques et minieres, 125, 332-334. FLANDRIN, J. 1963. Remarques stratigraphiques, paleontologiques et structurales sur la region de Sederon. Bulletin du Service de la Carte geologique de France, 272, 815-845. FRIES, G., 1987. Dynamique du bassin subalpin meridional de 1'Aptien au Cenomanien. These Doctoral es Sciences de FUniversite de Paris VI, 1986. Memoire des Sciences de la Terre de I'Ecole des Mines de Paris, 4, 370 pp. Distributed by Editions du Bureau de Recherches geolologiques et minieres, Orleans, France. FRIES, G. & BEAUDOIN, B. 1985. L'eventail sous-marin de Ceiise (SE France) a 1'Apto-Albien. Memoire de I'Association des Sedimentologistes francais, 2, 124 pp. FRIES, G., BEAUDOIN, B., JOSEPH, PH. & PATERNOSTER, B. 1984. Les Gres de Rosans et les slumpings aptiens associes : restitution paleomorphologique. Bulletin de la Societe geologique de France, 26, 693-702. FRIES, G. & RUBINO, J-L. 1990. Testing the application of sequence stratigraphy to Aptian deposits in Southeast France. In: GINSBURG, R.N. & BEAUDOIN, B. (eds) Cretaceous Resources, Events and Rhythms. NATO ASI Sciences Series C, 304,47-62. Kluwer Academic Publ., Amsterdam. GOTTIS, C. 1953. Les filons "intraformationnels" du "Flysch" numidien tunisien. Bulletin de la Societe geologique de France, 236,2521-2532. GUERIN, S. 1981. Utlisation des foraminiferes planctiques et benthiques dans 1'etude des paleo-environnements oceaniques au Cretace moyen: application au materiel des forages D.S.D.P. de 1'Atlantique Nord et Sud. Comparaison avec la Tethys. PhD Thesis, Universite de Nice, 198 pp. unpubl.
HISCOTT, R.N. 1979. Clastic sills and dikes associated with deep-water sandstones, Tourelle Formation, Ordovician, Quebec. Journal of Sedimentary Petrology, 49, 1-10. HISCOTT, R.N. & MIDDELTON, G.V. 1979. Depositional mechanics of thick-bedded sandstones at the base of a submarine slope, Tourelle formation (lower Ordovician), Quebec, Canada. In: DOYLE, LJ. & PICKLEY, O.M. (eds) Geology of Continental Slopes. SEPM, Special Publications, 27,307-326. IMBERT, P., PARIZE, O. & RUBINO, J-L. 1995. Massive "turbidite" sandstones in the Vocontian basin (South-East France): an analogue for blocky "basin floor fan" reservoirs. AAPG Bulletin, 79,1223. JENSSEN, A.I., BERGSLIEN, D., RYE-LARSEN, M., & LINDHOLM, R.M. 1993. Origin of complex mound geometry of Paleocene submarine-fan sandstone reservoirs, Balder Field, Norway. In: PARKER, J.R. (ed.) Petroleum Geology of Northwest Europe. Proceedings of the 4th Conference. Geological Society, London, 135-143. JOSEPH, PH., BEAUDOIN, B., FRIES, G. & PARIZE, O. 1989. Les vallees sous-marines enregistrent au Cretace inferieur le fonctionnement en blocs bascules du Domaine vocontien. Comptes-Rendus de VAcademic des Sciences de Paris, 309, 1031-1038. JOSEPH, PH., CABROL, CH. & FRIES, G. 1987. Blocs bascules et passes sous-marines dans le champ de Banon (France, S.E.) a 1'Apto-Albien: une paleotopographie directement controlee par la tectonique synsedimentaire decrochante. Comptes-Rendus de I'Academie des Sciences de Paris, 304,447-452. KNELLER, B. 1995. Beyond the turbidite paradigm: Physical models for deposition of turbidites and their implications for reservoir prediction. In: HARTLEY, A.J. & PROSSER, D.J. (eds) Characterization of Deep Marine Clastic Systems. Geological Society, London, Special Publications, 94,31^9. LOWE, D.R. 1975. Water escape structures in coarsegrained sediments. Sedimentology, 22, 157-204. LOWE, D.R. 1982. Sediment gravity flows: EL Depositional models with special reference to the deposits of highdensity turbidity currents. Journal of Sedimentary Petrology, 52,280-297. MARSCHALKO, R. 1965. Clastic dykes and their relations to synsedimentary movements (Flysch of Central Carpathians). Geologicke Prdce Bratislava, 36, 139-148. MASSE, J.P. & PHILIP, J. 1976. Paleogeographie et tectonique du Cretace moyen en Provence: revision du concept d'isthme durancien. Revue de Geographic physique et de Geologie Dynamique, 23, 49—66. McLEOD, M.K., HANSON, R.A., BELL, C. R. & McHuco, S. 1999. The Alba Field ocean bottom cable seismic survey: Impact on development. In: Proceedings of the 1999 Offshore Europe Conference. Leading Edge, 18, 1306-1312. MIDDLETON G.V. & HAMPTON M.A. 1976. Subaqueous sediment transport and deposition by sediment gravity flows. In: STANLEY, D.J. & SWIFT, D.J.P (eds) Marine Sediment Transport and Environmental Management. John Wiley and Sons Publ., New York, 197-218. MUTTI, E. 1992. Turbidite sandstones. AGIP & Istituto di Geologia della Universita di Parma Publ., 275 pp.
VOCONTIAN CLASTIC DYKES AND SILLS MUTTI, E. & NORMARK, W.R. 1987. Comparing examples of modern and ancient turbidite systems: problems and concepts. In: J.K. LEGGETT, J.K. & ZUFFA G.G. (eds) Marine Clastic Sedimentology. Concepts and Case Studies. Graham and Trotman Publ, London. MUTTI, E. & NORMARK, W.R. 1991. An integrated approach to the study of turbiditic systems. In: WEIMER, P. & LINK M.H. (eds) Seismic Fades and Sedimentary process of Submarine Fans and Turbiditic Systems. Frontiers in Sedimentary Geology. Springer Verlag Publ., New York, 75-106. MUTTI, E., TINTERRI, R., REMACHA, E., MAVILLA, N., ANGELLA S. & FAVA L., 1999. An introduction to the analysis of ancient turbidite basins from an outcrop perspective. American Association of PetroleumGeologists, Continuing Education Course Note Series, 39,96pp. NEWMAN, M.ST.J., REEDER, M.L., WOODRUFF, A.H.W & HATTON, I.R. 1993. The geology of the Gryphon Oil Field. In: PARKER, J.R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference, Geological Society, London, 123-133. NEWTON, S.K. & FLANAGAN, K.P. 1993. The Alba field. Evolution of the depositional model. In: PARKER, J.R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference, Geological Society of London, 161-171. OLLIER, G., COCHONAT, P. LENAT, J.F. & LABAZUY, P. 1998. Deep-sea volcanoclastic sedimentary systems: an example from La Fournaise volcano, Reunion Island, Indian Ocean. Sedimentology, 45,293-330. PARIZE, O., 1988. Sills et dykes greseux sedimentaires: paleomorphologie, fracturation precoce, injection et compaction. PhD Thesis, Ecole nationale superieure des Mines de Paris - Universite de Lille I. Memoire des Sciences de la Terre de 1'Ecole des Mines de Paris, 7, 333 pp. Distributed by Editions du Bureau de Recherches geolologiques et minieres, Orleans, France. PARIZE, O. & BEAUDOIN, B. 1986. Les filons greseux sedimentaires du flysch numidien des regions de Tabarka (Tunisie) et de Geraci Siculo (Sicile): fracturation precoce et paleomorphologie. Memorie della Societa Geologica Italiana, 36,243-253. PARIZE, O. & BEAUDOIN, B. 1988. Clastic dykes and sills from Numidian Flysch (Sicily and Tunisia), sandy injection related to a high-density turbidity deposit. AAPGBulletin,^, IGIH. PARIZE, O., BEAUDOIN, B. & FRIES, G. 1999. Deep-water massive sands: facies, processes and channel geometry in the Numidian Flysch, Sicily - comment. Sedimentary Geology, 127, 111-118. PARIZE, O., BEAUDOIN, B. FRIES, G., PINAULT, M. & PINOTEAU, B. 1987. Les filons greseux sedimentaires. Notes et Memoires de TOTAL Compagnie frangaise duPetrole, 21, 211-233. PARIZE, O., ECKERT, S., HAJ-HASSEN, F, TIJANI, M., BEAUDOIN, B., Su, K., HOTEIT, N. & TROUILLER, A. 2001. Analyse quantitative de la fracturation precoce des massifs argileux. Predictivite sur la presence des reseaux. In: BEAUDOIN, B. & TROUILLER, A. (eds) MinAndra 2000,5-9. PARIZE, O., SAVOYE, B., SAHABI, M. & COCHONAT, P. 1989. A braided channel pattern in the head of the Var sub-
71
marine canyon floor (Nice, France). Origin and implications. Memoire de VAssociation des Sedimentologistes franc, ais, 10,227-228. PARIZE, O., SCHNEIDER, E, ECKERT, S., BEAUDOIN, B. & TROUILLER, A. 2002. La fracturation precoce des massifs argileux: quel role pour la compaction. In: BEAUDOIN, B. & TROUILLER, A. (eds) MinAndra 2001. 9-12. PETERSON, G.L. 1966. Structural interpretation of sandstone dykes, Northwest Sacramento Valley, California. Geological Society of America Bulletin, 77, 833-842. PINOTEAU, B. 1986. Nouvelles methodes d'analyse de donnees diagraphiques. Etude sedimentologique d'un champ petrolier. PhD Thesis, Ecole nationale superieure des Mines de Paris. Memoire des Sciences de la Terre de I'Ecole des Mines de Paris, 1, 306 pp. Distributed by Editions du Bureau de Recherches geolologiques et minieres, Orleans, France. RIGOLLET, C. 2001. Valorisation sedimentologique de 1'information sismique: application au comblement des bassins profonds de la Marge Atlantique NordEuropeenne du cretace a F Actuel. PhD Thesis, Ecole nationale superieure des Mines de Paris. Memoire des Sciences de la Terre de 1'Ecole des Mines de Paris, in press, 507 pp. Distributed by Editions du Bureau de Recherches geolologiques et minieres, Orleans, France. RUBINO, J-L. 1981. Les gres apto-albiens du bassin vocontien et de sa bordure occidentale. Diplome d'Etudes Approfondies de Universite de Lyon, 33 pp. unpubl. RUBINO, J-L. 1989. Introductory remarks on Upper Aptian to Albian siliciclastic/carbonate depositional sequences. In: FERRY, S. & RUBINO, J-L. (eds) Mesozoic Eustacy on Western Tethyan Margins. Post-meeting field trip in the "Vocontian Trough". Memoire de VAssociation des Sedimentologistes francais, 12, 28-45. RUBINO, J-L. & PARIZE, O. 1989. Caracteristiques des systemes turbiditiques silicoclastiques apto-albiens du bassin vocontien (SE de la France): position sequentielle - comparaison avec 1'actuel. Memoire de VAssociation des Sedimentologistes francais, 10, 253-254. RUTTEN, M.G. & SCHONBERGER H.J.M. 1957. Synsedimentary sandstone dyke in the Aptian of SerreChatieu, Southern France. Geologic en Mijnbouw, 19, 214-220. SCHNEIDER, F. & PARIZE, O. 1989. Determination d'une loi de compaction pour la tranche superficielle des sediments (0-400 m) par 1'utilisation des filons greseux sedimentaires: approche theorique. Bulletin de la Societe geologique de France, 5, 869-873. SCOTT, R.M. & TILLMAN, R.W. 1981. Steve sandstone (Miocene), San Joaquin basin, California. In: SIEMERS, C.T., TILLMAN, R.W. & WILLIAMSON, C.R. (eds) Deep-water clastic sediments: core workshop. SEMP Core Workshop, 2,116-248. SMYERS, N.B. & PETERSON, G.L. 1973. Sandstone dikes and sills in the Moreno shale, Panoche hills, California. Geological Society of America Bulletin, 82, 3201-3208. STRICKLAND, H.E. 1840. On some remarkable dikes of calcareous grit, at Ethie in Ross-shire. Philosophical
72
O.PARIZE&G. FRIES
Transactions of the Royal Society of London, 5, 599-600. TRUSWELL, J.F. 1972. Sandstone sheets and related intrusions from Coffee Bay, Transkei, South Africa. Journal of Sedimentary Petrology, 42,578-583. TRUYOL, V. 1991. La compaction: appreciation et quantifi-
cation directes et indirectes. Implications diagenetiques. PhD Thesis, Ecole nationale superieure des Mines de Paris - Universite de Lille I. Memoire des Sciences de la Terre de VEcole des Mines de Paris, Distributed by Editions du Bureau de Recherches geolologiquesetminieres, Orleans, France, 13,190pp.
Deformation structures in Plio- and Pleistocene sediments (NW Bohemia, Central Europe) P. BANKWITZ1, E. BANKWITZ1, K. BRAUER2, H. KAMPF3 & M. SlORR4 1
2
Gutenbergstr. 60, 14467 Potsdam, Germany (e-mail: [email protected]) Center for Environmental Research Leipzig-Halle, Th. Lieserstr. 4, 06120 Halle, Germany 3 GeoForschungsZentrum Potsdam, Telegrafenberg, 14473 Potsdam, Germany 4 Ostklune 11, 17406 Usedom, Germany Abstract: The small intracratonic Cheb (Eger) Basin in NW Bohemia (Central Europe) is characterized by swarm earthquakes, many mineral springs and mofettes with upper mantle CO2 degassing and by neotectonic graben and basin structures. Especially in non-lithified Upper Pliocene clay formations of the basin, a variety of deformation patterns is exposed. They include non-tectonic and tectonic activity and comprise faulting and folding from jxm- to km-scale. Previously unrecognized N-S- and ENE-striking faults are sites of mantle degassing and seismic activities. Confined-layer deformation and liquefaction structures hint to palaeoseismic events and gas escape activity. Cleavage-like arranged clay mineral plates represent the microfabric of clay within fault zones. For the first time the degassing channels of Upper Mantle fluids/gases through the Pliocene clay sediments can be documented: (Jim-scale micro-tubes were produced by the opening of Riedel shear planes induced by fault movements.
Sediment mobilization commonly involves the introduction of external fluids, including gases and the resulting structures grade into those resulting from neotectonic deformation. Evidence for recent or neotectonic deformation is most widely recognized in unconsolidated sediments of Pliocene or Quaternary age. Examples of deformational structures (Owen 1987) in Cenozoic soft sediments have been published worldwide. For example, by Collinson (1994) from recent river terrace deposits in Norway and from the modern Mississipi delta, from Quaternary deposits in NE Europe. Also, especially from glacigene deformation in Germany by Eissmann (1987) and Eissmann et al (1995); by Maltman (1994) from glacial Pleistocene sands in Poland with mineralized shear zones, from Eocene and Pliocene in Japan and Wyoming /USA and from earthquake-induced structures of Quaternary sediments in Scotland. McCalpin & Nelson (1996) published faults and scarps from several places in USA (Utah, Nevada, Idaho). In the Belgian Basin (Eocene) tectonic faults in clay were detected by Verschuren & van Rensbergen (2001). However, sedimentary basins of that age, which possess well exposed deformation structures, in association with features attributed to active mantle fluid degassing (CO2) and migration, are rare in Central Europe. One example of such a sedimentary basin is the Cheb (Eger) Basin of NW Bohemia. This Tertiary basin is located within the western part of the Ohre (Eger) Graben, to the SE of the GermanCzech national border (Fig. 1). This paper describes the various deformation structures developed within the Cheb Basin and aims to demonstrate that there is
a close relationship between earthquakes, deformation and the introduction of fluids by permanent mantle degassing. A local Moho updoming from 30 km in the surroundings up to 27 km below the Cheb Basin, with a diameter of about 40 km, was detected by P-to-S converted phases of seismic waves received from the Moho (receiver function method, Geissler et al., pers. comm. 2002). The converted phases were observed in the depth range of 30 to 60 km of the upper mantle. The area of Moho updoming and the occurrence of uppermost mantle conversions coincide with the location of CO2 escape centres, according to Weinlich et al. (1999). Geissler et al. (pers. comm. 2002) suppose that isolated active magma/fluid reservoirs in the uppermost mantle exist, which point to a presently active magmatic underplating beneath the earthquake swarm region of NW-Bohemia and the Cheb Basin. This could be one of the reasons for the various neotectonic processes that contribute to sediment mobilization and deformation in the Cheb Basin.
Geological setting of the Cheb Basin The Cheb Basin (Malkovsky 1987) is a small (c. 265 km2) intracratonic basin which formed during the Late Tertiary as a result of the reactivation of earlier Hercynian faults present within the basement. The basement comprises a multiphase deformed sequence of Proterozoic to Lower Palaeozoic monotonous slaty and quartzitic sediments, para- and orthogneisses and mica schists (Mlcoch et al. 1997)
Fig. 1. Tertiary Cheb (Eger) Basin that crosscuts the Ohfe Graben, SE of the German-Czech national border (D/CZ). Grey: post-collisional granite plutons. Ellipses: areas of earthquake swarms (Griinthal et al. 1990). Dark grey ellipse: earthquake swarm region near Novy Kostel (Fischer & Vavrycuk Internet comm. 2000). (l)-(3) new recognized fault zones in Plio- to Pleistocene clays and sands: (1) the N-S-trending Pocatky-Plesna Zone (PPZ); (2) the ENE-trending Nova Ves Fault (exposed in the open cast mine Nova Ves: circle); (3) the Luzni Fault. Grey line: faults ace. to the Wismut GmbH and the Geological Survey Freiberg. Black hatched lines: remote-sensing lineations; short hatched: evidence for current activity vom GPS measurements (Wendt 1999). Frant: Frantiskovy Lazne (Franzensbad). Tertiary volcanoes: K, Komorni hurka (Kammerbiihl); E, Zelezna hurka (Eisenbiihl).
DEFORMATION IN PLIO- AND PLEISTOCENE SEDIMENTS
intruded by the 326 to 289 Ma in age (Hecht et al 1997) Fichtelgebirge (Smrciny) and Karlovy Vary granites (Fig. 1). The basin developed at the intersection of the older (Lower Tertiary) ENE-trending Ohre (Eger) Graben and the NW-trending preNeogene Marianske Lazne Fault (MLF; Fig. 1). This fault system forms the eastern boundary of the basin and is marked by a 50 to 100 m high escarpment that can be traced, along strike, for over 100 km. At the other sides of the basin, the Tertiary sediments are transgressing on various basement series. The Cheb Basin has been actively subsiding since the Late Oligocene and comprises a sequence of Miocene lignite, clay and sand of the Miocene Cypris Formation, overlain by Upper Pliocene sands, gravels and kaolinitic clays of the Vildstejn (Wildstein) Formation (Fig. 2a). The Cypris Formation is only rarely exposed. However, the younger sediments of the Vildstejn Formation can be observed in a number of open cast mines (for example Nova Ves I &II east of the village Skalna). These two formations are separated by a diastem of 12 Ma. The sedimentary fill of the Cheb Basin also locally includes Eocene fluvial deposits. The total thickness of the Neogene sediments is less than 300 m (Spicakova et al. 2000). Pleistocene fluvial sediments present within the basin occur along the main river valleys and lacustrine Holocene diatom beds in the Soos area (Rehakova 1988). The area of the Basin was never covered by Pleistocene glaciers, but not outside the limits of permafrost. The Cheb Basin is situated in an area of remarkable neotectonic and recent crustal activity. This activity is characterized by many mineral springs, CO2 mofettes and earthquakes caused probably by active magmatic underplating (Geissler et al., pers. comm. 2002). Isotopic (813C, He3/He4,815N) and gas geochemical studies as well as gas flow mapping in the western part of the Ohf e Graben indicate that the CO2 and partly N2 which are being expelled from the mofettes are derived from the upper mantle (Weinlich et al. 1999; Brauer et al. 2003). As a result the Cheb Basin is believed to lie above an active subcontinental mantle volatile system, with recent fault movements allowing the rapid CO2 transport through the upper crust of 50 m per day and more, and release of this mantle derived CO2 (Weise et al. 2001; Brauer et al. 2003). The associated earthquakes (magnitude M^4.6) occur as swarms and at depths of 6 to 15 km, with the most recent prolonged period (August to December 2000) of seismic activity consisting of more than 10,000 events with a magnitude of <3.7 (Fischer & Vavrycuk, Internet comm. 2000; Horalek et al 2000). The epicentres of the earthquakes are distributed along N-S-trending faults, in particular the Pocatky-Plesna Fault Zone (Figs 1 and 4). The N-S faults have probably been active since Pliocene-Quaternary times and are
75
Fig. 2. (a) Stratigraphic units of the Cheb Basin, modified according to Spicakova et al. (2000) and incorporating data of Rehakova (1988) and Vtelensky et al (1990). (b) Simplified geological map with faults in the Cheb Basin (according to the Geologicka Mapa CSSR 1:200,000 sheet Karlovy Vary, 1990) and sites of recognized new faults (1 to 3 and thick lines) and areas of detailed studies.
known to have moved in Hercynian time to the Lower Palaeozoic basement in SW Saxony (Vogtland), where they are marked by 10 to 20 m wide zones of cataclasite. Within the basin the faults are hidden by Neogene non-lithified sediments covered with vegetation, and in part known from borehole interpretation (Balz 1908). The mapping (refined localization) of the hypocenters of the recent earthquakes (1985/86, 1994, 1997/98, 2000) has demonstrated that the prominent, Hercynian, Marianske Lazne Fault (Fig. 1) has remained seismically inactive during this recent phase of fault movement. The Pocatky-Plesna Zone (PPZ, azimuth 175°) crosscuts the NW-trending Marianske Lazne Fault (MLF, azimuth 150°) at Novy Kostel and will be blocked up by creep movements of the MLF due to an offset at the intersection point. This interaction
RBANKWITZCTAL.
causes stress accumulation along the fault segment of the PPZ, which cannot move by creep, and resulting earthquake activity in the area of intersection (Schneider 1993).
Statement of problem The aim of the study was to define the tectonical frame of the epicentral area Novy Kostel that was not well understood in the past. The mapped faults
(Geologicka Mapa CSSR; (Fig. 2b) are hidden by Pliocene and Quaternary soft sediments and vegetation, and have been derived mostly on the base of bedding dislocations in boreholes. None of them coincides with the distribution of the epicentres in the northern part of the Cheb Basin within a narrow N-S-trending band at Novy Kostel. The occurrence of previously unrecognized major faults in the Cheb Basin and adjacent areas, and the timing of fault activity, was to clarify the relations between the frequent earthquake swarm events and
DEFORMATION IN PLIO- AND PLEISTOCENE SEDIMENTS
the active mantle CO2 degassing in the northern part of the basin. In this paper the location of the new faults, together with their associated deformational pattern, and the spatial distribution of epicentres and mofettes are considered to determine if these features correlate. Fracturing and folding of Pliocene soft sediments, adjoining one of the new faults and the Nova Ves shear zone will be discussed; whether they have formed as a result of syndepositional or periglacial processes or seismic-induced, although evidence for seismic activity in Pliocene time is not investigated. Nonetheless, significant amounts of crustal movements are known (Peterek et al. 1996) and expected to be associated in part with seismic activity analogous to the actual behaviour of the crust. Along the Nova Ves Fault, which is of endogenic origin, the scaling of shear plane formation and its relationship to the mantle CO2 degassing through the Pliocene clay had to be investigated, providing unique observations of tiny gas escape channels of 0.005 mm in diameter in clay using SEM techniques.
Methods of investigation Three faults of the Cheb Basin were recognized by field observations for the first time during mapping of unknown, supposingly active zones indicated by morphological and drainage pattern anomalies in the northern part of the Cheb Basin, and where earthquakes and mofette activities occur. In addition, measurements of fault planes and shear indicators of the Nova Ves Fault were taken in the open cast mine Nova Ves II. Furthermore, several types of folds within this mine were recognized, in most cases only two-dimensionally at the high faces, although a few could be documented three-dimensionally. Tectonic research was not favoured because exposures within the basin are rare, although the sequence of strata of Late Pliocene age can be studied in some temporary clay and gravel pits. Outside of the open casts, neotectonic processes are only indicated by anomalies of the drainage system and relief. Faults cannot be followed outside the clay mines, because the whole basin is pasture-ground with several swamps accompanying flat depressions, or it is partly covered by agriculture and brush-wood in former sand pits. Structural measurements can only be taken in the open cast mines and only from the ground floors and the lower faces, often at a level about 40 m below the surface. The faces are always steep and consist mostly of water-bearing soft sediments. In some cases, it is risky to work at the foot of the current exploitation slope. The faces and only in part preserved benches change weekly due to the progress of mining. From the site of the ENE-trending shear zone in
77
the mine Nova Ves (exposed length: 350 m; approximately width: 20 m) 23 clay samples were investigated using a variety of techniques including mesoand microscopy, and Scanning Electron Microscopy (SEM: Jeol JXA-840A) with mineral analysis concerning the micro-tectonical behaviour of the clay (EDX: LINK Oxford EDX System with Quantum Si(Li)-Detector, Resolution 136 eV) at the University of Greifswald. For the purpose of microtectonical investigations, SEM images from joint and fault plane surfaces were made. The oriented samples were Pd-sputtered. For the mineral analysis, two common clay types (dark brown and grey) were selected and in addition, clay samples from the subfaults of the Nova Ves shear zone with different clay matrix from the slickensides domains (dark grey or black dense). Secondary minerals on the fault planes and slickensides were identified by microanalysis of C-coated samples, dried in air and high vacuum. The investigation comprised: grain size fractionation <0.63 to >63.0 (Jim, X-ray analysis of oriented specimens, X-ray powder analysis to determine the composition of the grain fraction and thermo-analytical investigations to quantify the mineral ratios. Approximately 50 mofettes are connected with the Nova Ves Fault in the open cast mine. Four gas samples were collected in glass vessels with two stopcocks. The vessels were filled with water, which was replaced by the free gas bubbling out of the water in the glass vessel. Triplet samples were taken to measure the gas composition and the 13C/12C and 3 He/4He ratios (Weise et al 2001). The CO2 content was determined volumetrically, while other components such as N2, O2, Ar, He, H2, and CH4 were measured by gas chromatography after CO2 absorption in KOH solution (Weinlich et al. 1999). For 13C analyses, CO2 and water were separated from the non-condensable gas in a vacuum at liquidnitrogen temperature. A mixture of dry ice and alcohol at -78°C removed water vapour. The isotopic analysis of carbon was carried out using a Finnigan MAT Delta-S mass spectrometer after the cryogenic separation of CO2. 613C was related to PDB standard (Pee Dee Belemnite), the reproducibility of the 513C measurement was <0.1%o.
Major faults Within the Pliocene Cheb Basin, previously unrecognized active faults have been found. Two of them, the N-S-trending Pocatky-Plesna Zone (PPZ) and the Luzni Fault, were active during Late Pleistocene and fit into the Mid-European RegensburgLeipzig(Rostock) Zone, in part 40 km wide (length: 700 km) and seismically active in its middle part (Behr et al. 1994). This zone consists, especially in the German part, of numerous N-S-trending faults,
78
RBANKWTTZZrAL.
which are composed of en echelon segments measuring a few kilometres, corresponding to the Pocatky-Plesna Zone. Similar conditions of fault segmentation in epicentral areas were detected by Tchalenko & Ambraseys (1970) at an Iranian earthquake fault zone. Figure 1 presents a section of the Regensburg-Leipzig Zone with selected faults on the German side. Similarities exist with other seismically active N-S-trending zones with sinistral shear sense in central Europe (e.g. the 9°E earthquake zone Albstadt-Stuttgart, the N-S zone within the Upper Rhine Graben, the N-S zones in the western Vosges and the Massif Central). The recognized ENE-trending Nova Ves Fault represents a dextral shear zone with its strike following the northern boundary of the Ohfe Graben (Fig. 1). It covers the Variscan margin between the Moldanubian (SE) and Saxothuringian (NW) crust and is considered to form part of the Ohfe Graben boundary system. The extended eastern prolongation could not be traced. Certainly, the Nova Ves Fault, exposed in clays of Pliocene age, has been active earlier than the Pocatky-Plesna Zone. The intersection with the Pocatky-Plesna Zone is not exposed, but dislocation at the intersection point can be expected. The time of initiation of the faults is unknown; only the time of activity can be derived from the deformed sediments. Pocatky-Plesna Zone The Pocatky-Plesna Zone (PPZ) trends N-S through the Cheb Basin intersecting mapped oblique faults (Fig. 2b), and to the north entering the Variscan basement of the mountainous Vogtland region. The active fault zone could be traced over a distance of 15 km, due to expression by a morphological anomaly in the soft rock based landscape (Fig. 3a). Fault planes of the Pocatky-Plesna Zone (PPZ) are not exposed. A prominent high-angle scarp (25-30 m high) gives evidence for the occurrence and orientation of a border fault between two separated blocks. Along the PPZ, the height difference of more than 20 m (Fig. 3b) between the higher level of the eastern block and the lower western flank and the relative steepness of the slope, point out that the movement is still active and that the fault displacement is greater than the regional erosion rate (McCalpin & Nelson 1996). The N-S-trending scarp is to the south adopted by the Plesna stream. Erosion in front of the geomorphological step deepened the fault trace and formed the often swampy valley depression. The mofettes Hartousov and Bublak occur along the PPZ (Fig. 3c). The true level difference between both flanks amounts only 5 to 15 m. Cross sections demonstrate the changing height of the scarp along strike of the PPZ (Fig. 4a).
The orientation of the Pocatky-Plesna Zone is dominantly NNW (175°). In detail the PPZ fault, rather than being a single N-S-striking structure is, in fact, composed of a number of enechelon arranged NNE- and NNW-trending scarp segments (Fig. 4b, c). The orientation and seismically documented sense of displacement on these fault segments is consistent with them having formed as Riedel shears (Riedel 1929) within a much wider N-S-trending zone of sinistral shear. All segments are of km length, but the total length of the zone itself is about 35 km. The dip of the Pocatky-Plesna Zone is not known, because a trench across the fault was not available. Several arguments for a steep dip exist, e.g. the distribution of hypocentres along the fault at depth between 6 and 15 km, dominantly between 8 and 10 km (Fig. 4d). The alignment of the epicentres is closely associated with the scaip of the PPZ (Nehybka & Skacelova 1995; Horalek et al 2000), schematically shown in Figure 3b. At present sinistral strike-slip movement components are to be expected at the PPZ according to the focal solutions and the known stress regime in the study area. About 80% of the seismic events recorded since 1985 in the Vogtland/NW-Bohemia area (Neunhofer & Guth 1990; Nehybka & Skacelova 1995; Horalek et al. 2000), occurred in the epicentral area of Novy Kostel, distributed along the Pocatky-Plesna Zone. They occur as swarms (1985/86, 1994, 1997/98, 2000) consisting of many thousands of small earthquakes up to magnitude M 2.2 and at depths below 6 km. The maximum magnitude of a main event was M<4.6, most recently with a magnitude of <3.7 (Fischer & Vavrycuk, Internet comm., 2000; Horalek et al. 2000). The N-S-trending chain-like distribution of the epicentres hints to a causative relationship with the Pocatky-Plesna Zone. Episodic cmstal movements of the epicentral area Novy Kostel with changing sense of movement, uplift and subsidence of 2 to 8 mm related to seismic events, were recorded geodetically. These temporal changes of height were measured in annual campaigns of precise levelling. After an earthquake-swarm in 1997 an episodic elevation of 4 to 6 mm (Mrlina 2000) of the area with epicentres along of the N-S zone occurred, in contrary to the swarm in 1994. They show different senses of motion. The division line of different displacements of the surface for the swarms of 1994 and 1997 was determined (Mrlina 2000), coinciding with the axis of the main seismo-active zone and with the Pocatky-Plesna Zone. From the geophysical point of view, Svancara et al. (2000) and Havif (2000) stated that the epicentre distribution to the north of Novy Kostel requires a vertical crustal discontinuity. They inferred a line in the epicentral area, which they called the 'Pocatky-Novy Kostel-Zwota line'. A visible fault was not recognized by these
DEFORMATION IN PLIO- AND PLEISTOCENE SEDIMENTS
79
Fig. 3. (a) Scarp of the Pocatky-Plesna Zone (PPZ) in Late Pleistocene soft sediments along the Plesna creek near the site of 'Briick' (section U Mostku - Horka in Fig. 4). The scarp is preserved along its strike over a length of more than 10 km. (b) Sketch of the Pocatky-Plesna Zone within the Cheb Basin and its intersection with the marginal Marianske Lazne Fault (MLF; not to scale). Dots mark schematically the distribution of epicentres in the area of Novy Kostel. The different levels of the separated blocks at both sides of the PPZ according to the elevation contour map of Svancara et al. (2000). CTH, maximum horizontal stress orientation, (c) The scarp segments of the PPZ (not to scale) and the locations of associated mofettes (circles) with documented CO2 mantle degassing (Brauer et al. 2003).
authors, either in that region extended to the north of the Cheb Basin or further to the south. Groups of events in the swarms in the period 1985 to 2000 show variations in focal mechanisms, e.g. strike-slip motion, normal or reverse faulting. Sinistral strike-slip-movements along approximately N-S planes connected with a vertical shear component dominate in the epicentral area Novy Kostel (at the PPZ) and fit into the general NW orientation of the maximum horizontal axis of compression (Wirth et al 2000). The long-term shaking
of the soft sediments that fill the basin, daily during several months (10,000 events in the swarm period August to December 2000) were considered to produce some effects in sediments at the PPZ and its further surroundings.
Luzni Fault The newly recognized Luzni Fault is located 1 km to the west of the Pocatky-Plesna Zone, connected with
80
P.BANKWITZE7AL
Fig. 4. N-S trending Pocatky-Plesna Zone (PPZ) of the Cheb Basin. 10.000 epicentres of the last earthquake swarm period, August to December 2000 (ellipse, Fischer & Vavrycuk, Internet comm. 2000) were distributed along the fault, (a) Sections across the scarp of the PPZ (vertical exaggeration 1:10). Scarp height: 25 to 30 m in non-lithified sediments. The level difference between the two flanks of the PPZ: 5 to 20 m. (b) Segments of the escarpment were indicated by bending of its strike in the middle part of the PPZ. Hatched lines outside mark the supposed boundaries of the PPZ fault. Dots indicate main degassing centres, (c) Scheme of the estimated dip of the PPZ. During the period of the earthquake swarm in 1997 the hypocenters occurred at depths between 8.7 and 10.0 kms (circles, Spicak & Horalek 2000).
the N-S-trending part of the Luzni creek (Fig. 5). The fault itself is not exposed, but indicated by a geomorphological feature. The Luzni stream drains from the western border of the Cheb basin to the SE. In the past, its water flowed into the main stream of the Plesna stream. Now, the Luzni stream turns to the south, no longer reaching the Plesna stream. The formerly-used part of the Luzni valley, prolonged downwards to the Plesna valley, is now a dry depression. A crest-like chain of hillocks that has interrupted the primary Luzni valley and forced it to drain southward marks the bending point of the drain. The ridge-like heights are observed over several kilometres. This elevation of the eastern bank of the Luzni valley marks the N-S-trending fault with observed length along strike of 3 km. Latest Pleistocene sands of the Wurm glacial period (Q4, Fig. 6) were deposited only at the western bank, thus indicating the initiation of the fault somewhat earlier.
Timing of fault activities Late Pleistocene deposits give evidence for the time of activation of the Pocatky-Plesna Zone (PPZ) and the Luzni Fault. The distribution of the Quaternary sediments along the Plesna stream reveals a specific pattern (Fig. 6): the youngest Pleistocene Q4 Formation was only deposited in front of the scarp of the PPZ, and on the western lower side of the Luzni valley, thereby dating the initiation of the scarps. The Q2 Formation is main exposed at both flanks. Q3 fluviatile sediments characterize the valleys of the Plesna creek and its tributaries. Q4 sediments occur exclusively along the western slope of the Plesna and of the Luzni stream ahead the bending section of the river slope. Obviously, the escarpments were formed before or during the sedimentation of the Q4 sands, nearly 120,000 years b. p.
DEFORMATION IN PLIO- AND PLEISTOCENE SEDIMENTS
81
Fig. 5. Part of the drainage system east of Nova Ves (Cheb Basin). The Luzni creek is cut off (arrows NE of Nova Ves) due to a N-S-trending fault. The water flow into the Plesna creek was truncated (thick arrows). The flow down from the eastern slope into the Plesna creek changed also (thick arrows) into a drainage to the south, parallel to the Pocatky-Plesna Zone, on the elevated flank. ®, uplift side of the faults. Hatched: dry valleys.
Nova Ves Fault The Nova Ves Fault, recognized for the first time, occurs within the Cheb Basin to the east of Skalna related to a basement fault in the hidden part of the Hercynian Smrciny Granite. In the open cast mine Nova Ves II (Fig. 1, circle) the ENE-trending largescale Nova Ves Fault is temporarily exposed in the Vildstejn (Wildstein) Formation, Upper Pliocene in age. The Nova Ves Fault is accessible at the lowest level of the quarry (40 m below surface) over a length of 300 m (azimuth 070-060°). It presents a dextral shear zone of c. 20 m width (Fig. 7). The timing of the fault activation within the Pliocene clays is not exactly definable; dislocated strata could not be recognized due to the mining conditions. The slope of the mine consists completely of lower clay layers and upper clay-sand intercalations of the Nova Ves Beds (Fig. 2a). The only evidence for the timing is the age of the deformed layers (not determined in detail), confining an intraPliocene period of strike-slip movements. A later generation of the discrete shear planes within the Nova Ves Fault seems improbable. Comparable deformation features are absent in the overlying sandy beds of the same Vildstejn Formation, but possibly due to the variable behaviour of the unlimited sands. The shear zone consists of small, narrow, standing discrete shear planes, mesoscopically mostly of decimetre-size (Fig. 8), occasionally up to 2 m, covered with fine or coarse horizontal slickensides.
Fig. 6. (a) Ages of the Pleistocene sedimentary succession Q3 to Q4 (modified from Skvor & Sattran 1974). (b) Schematical section across the PPZ. The Pleistocene Q4 Formation occurs only in front of the scarp, the Q3 Formation at both sides of the PPZ. The fault was activated after sedimentation of Q3 that dates the scarp initiation (0.12 Ma).
These numerous shear planes of various strike and dip (dominantly subvertical) are arranged only within the shear zone (Bankwitz et al. 2001). The Nova Ves Fault differs in its structure from shear zones in hard rock in the way that hundreds of single planes of <m2 dimension are scattered, separated from each other, but all with mostly horizontal slickensides. In addition, they are often covered with traces of intersecting cleavage-like planes (Fig. 8b, c). Because of differences in dip, they locally form an anastomosing grid. The distance between the individual shear planes or lenses is from 50 cm to 0.005 mm. The subhorizontal striae of meso- and micro-scale (cm to [Jim) give evidence of dextral movements on these planes. Dislocations could not be observed due to the monotonous clay layers of up to several metres thickness and due to the exploitation technique. The only shear sense indicators are the slickensides. The dextral movement coincides with the recent stress field orientation, which is known from seismological
82
P.BANKWITZ^rAL.
Fig. 7. Cheb Basin: open cast mine Nova Ves II in kaolinite clay (Fig. 1). Hatched lines mark the boundaries of the Nova Ves Fault, which is a dextral shear zone. Red: locations of clay samples for SEM investigations. Blue and arrows: main degassing locations. Blue with red rim: four samples of CO2 degassing vents number 1 to 4 (Tab. 1). Length of the exposed strike-slip zone in view direction: 300 m, width: c. 20 m.
(Klinge & Plenefisch 2001) and geological (Peterek etaL 1997) investigations.
Microfabric of gas jet faults within the Nova Ves shear zone The Nova Ves shear zone is a site of permanent CO2 degassing. Gas jet streams escape through hairline cracks at the surface, which are traces of small fault planes (Fig. 8). The blow out of the gas can be heard and seen within small water-filled mud holes of several centimetres to decimetres in diameter. Data of geochemical and isotopic analyses from four main degassing locations within the shear zone (1-4 in Fig. 7) confirmed the mantle source of the CO2
(Table 1). The gas jet polishes the mostly dark surfaces of such shear planes in the grey kaolinite clay of the Vildstejn Formation. Faults planes without gas escape are not so finely smoothed. No chemical interaction of the gas with the clay minerals was found. Results of the mineral analysis of such black surface domains ruled out graphitic coating or a gouge formation with another clay composition. Only films of preferred oriented small clay plates or flakes were recognized on the polished fault surface. From SEM micrographs, a complex inventory of shear planes becomes obvious. At least four sets of shear planes can be recognized. All of them dip sub-vertical and their space-relationship hints at the same shear plane geometry, as it was found in
DEFORMATION IN PLIO- AND PLEISTOCENE SEDIMENTS
83
Fig. 8. (a), (b) Vertical striae on fault planes, superimposed by fine horizontal striae, (b) Fault plane composed of two shear plane directions: ENE and NNE; vertical striae are intersecting lines of R- and supposedly P-shears. (c) Characteristic feature of fault planes within the Nova Ves shear zone: most frequent horizontal striae (the photograph is turned because of the lighting), (d) Subvertical ENE-trending shear plane with (primary) vertical traces of steps possibly due to intersecting R-shears as in the sketch (left).
macro-scale. The micro shear plane system correlates with that measured at the faces of the mine. SEM micrographs reveal the analogy of the micro shear plane structure with cleavage planes in slate or shale (Maltman 1994). The lamellae-like dense planes (Fig. 9 and lOa) correspond with R-, R'- Pand Y-shears of the Riedel model (Riedel 1929). R' -anti-Riedel shear planes are assumed to be latest formed. They were opened by a dextral movement and bending of the R-shear at a distance of 0.02 mm.
The micrograph in Figure lOa shows a section from the gas jet fault (GIF) plane at sample location (2) in Fig. 7. The fault surface is parallel to the micrograph and presents the main shear plane. The other three groups of traces are very dense shear planes, illustrated schematically in Figure 9. The distance between them is about 0.005 mm. One shear plane set reveals opened fissures of R'-shears. Of most interest are black stripes at the wall of such fissures. These are tube-like pores resulting from widening of the R'-fissure due to gas jet escape.
84
RBANKWITZ£TAL.
Fig. 9. (a) Fabric of the dextral Nova Ves shear zone resulting from micro-scale investigation. The scheme was reconstructed from SEM micrographs as in Fig. lOa. R-, R'- P- and Y- are Riedel shears. GIF is one of the Gas Jet Faults (sample NVII-3-B3) where the stream escapes at the bottom of the open cast mine (Fig. 8). The GIF surface coincides with the Y-plane of the scheme. The micro-fabric correlates with the meso- and macro-fabric of the shear zone (b) measured at the lower face and the bottom of the mine. Ellipses mark a subsurface cave along a gas stream conduit. Sand box pattern after Mandl (1988).
In addition, the micro-scale shear deformation occurs in discrete disc-like domains, not in the whole sample. Anastomosing fault lamellae develop from the coalescence of minor shear planes. The thickness of the disc-like shear plane bands is between 100 |mm to 1 cm. They are darker than the non-sheared clay and therefore easy to detect.
'Secondary porosity' and degassing channels in clay Studies of rock samples (kaolinite clay) of the surroundings of gas vent number 1 and 2 (Fig.7) gave quasi-three-dimensional insights into the interior behind small-size shear planes, by using SEM investigations. The mesoscopic shear plane surfaces as they are shown in Figure 8c and 8d are composed of
DEFORMATION IN PLIO- AND PLEISTOCENE SEDIMENTS
85
Fig. 10. (a) Part of the subvertical Gas Jet Fault plane within the Nova Ves shear zone from the site of the gas vent number 2 (SEM micrograph). The shear plane surface (parallel to the micrograph) was polished by a gas jet stream. In addition, three groups of further shear planes are traced on the plane surface. One set is visible as fissures or like holes, with dark stripes at their wall, (b), (c) Details from (a): degassing channels (arrows).
sub-planes, which can be seen in the example of the subvertical gas jet fault plane of Fig. lOa. The Rshear planes have a spacing of 0.02 mm. SEM investigations made clear that the gas/fluids escape through vertical micro-tubes in a size of 1-2 fxm in diameter (Figs 9 and 10). Such tiny channels, opened from 0.001 to 0.005 mm, are closed for flow of water, but not for gas (CO2). Through the tube-like pores, the mantle derived CO2 (Kampf et. al 1999; Weinlich et al. 1999) discharge with high pressure (and with noise) at the bottom of the open pit within the Nova Ves shear zone. At intersection sites of two micro-shear plane sets the micro-tubes occur at a high density.
Mantle degassing associated cavities In addition, meso-scale conduit structures comparable to the micro-tubes in the clay samples occur within the ENE-trending shear zone. One cave located in the lower pure clay sequence between gas vent 1 and 2 (length: 5 m, depth: 1 m, width: 0.5 m) trends parallel to the shear zone. The wall of the cave was in part mineralized (gypsum) and covered with sand, although in the mine no sand layer exists, neither directly below nor above this clay formation. This points to a cavity system within the shear zone that was filled with CO2-saturated water (sand inflow from much deeper unexposed levels) and worked as
P.BANKWITZETAL
86
Table 1. Gas composition of air-free fraction and I3CC02 isotopic composition of gases from the ENE strike-slip zone in the kaolinite clay mine (locations of the gas vents in Fig. 7) Gas vent (from ENE to WSW)
He (ppmv)
N2 (vol.%)
Ar (ppmv)
CH4 (ppmv)
CO2 (vol.%)
813C
Sample 2 Sample 1 Sample 3 Sample 4
10 11 20 34
0.11 0.19 0.34 0.57
19 36 77 125
2 3 6 9
99.89 99.80 99.65 99.41
-2.39 -2.06 -2.25 -2.18
a gas separator system (inflow from below: CO2-saturated water; in the cave: gas/water separation; outflow: mostly gas). The cavity system reacts like a cold geyser, possibly with a low liquefaction rate. During earthquake events, the process of fluid migration within the cavity could be increased and may generate liquefaction structures in the sediments.
Evidence for mantle fluid flow through the Vildstejn Clay Formation At about 50 locations, gas emanates within the ENE shear zone through the kaolinite clay beds. Along an ENE to WSW transect four gas samples were collected (Fig. 7). The gas composition and the 513C values were analysed (Table 1). From ENE to WSW the CO2 content decreases slightly, whereas the nonacid gas components N2, Ar, CH4 and He in Nova Ves increase in the same direction. The 813C values gas composition and 813C values of the four analysed gas exhalations of Nova Ves are comparable with those of the mofettes Bublak (-2.0%o), Hartousov (-2.1%c) and Devin (- 1.8%o) in the east of the Cheb Basin. Based on their gas characteristics expressed by the high gas flux, the high CO2 concentration and the high mantle-like 3He/4He ratios (Weinlich et al 1999) gave evidence for the magmatic origin of the gas from these locations. The Bublak mofettes contain the highest content of mantle gas and are considered as the main degassing centre above a supposed upper mantle plume and above the updoming Moho.
Deformation and related structures developed in unconsolidated sediments of the Cheb Basin The Cheb Basin and its surroundings are sites of recent intra-plate crustal processes, as mentioned above. Frequent seismic swarm events, permanent CO2 degassing and, in some cases, crustal uplift or subsidence of that region are being observed. But the deformations of the unconsolidated sediments, asso-
ciated with such processes in the past, have rarely been reported before. Recent deformation of soft sediments is not known, depending on the rare exposures and on the short-term instability of the slopes in non-active open cast mines. Previously unrecognized folding and faulting are exposed in several open mines without any definite relation to the most recent earthquakes because the features do not pass through into the Quaternary sediments (1 to 3 m thick) which occur locally.
Large-scale convolute bedding/garland-like folds In the open cast mine Nova Ves II (east of Skalna; Figs 1 and 2b) intrafolial folds of various types and scales are developed within the unlithified Pliocene sediments, which fill the Cheb Basin. The mine is deepened by about 40 m due to exploitation of the Nova Ves Beds (Vildstejn Formation). The lower part of the exposed profile consists of clay beds (dark grey, brownish-grey and light grey; thickness: approximately 10 m) overlain by sandy clays and argillaceous sands. The upper part of the sequence is composed of interbedded arenaceous and argillaceous sediments. Folded beds occur at the NW exploitation slope of the open mine, adjacent to the Nova Ves Fault in front of the slope (Fold Layer 2). The folds are restricted to a layered sequence of 6 to 8 m in thickness in the upper third of the slope (length of the face: about 300 m). In cases of interbedded clay layers (20 cm to > 1 m) and sandy beds lying upon porous sandy beds, garland-like folds are developed, with 5-20 m wide flat synclines and narrow steep 2-5 m wide anticlines (Fig. 11 a) with an amplitude of several metres. The axial traces of the antiforms in Figure lib trend 048° to 055°NE, the strike of the fold axial surfaces is about 050°NE and the dip varies between 35° and 65° NW, indicating a sense of vergence of some folds towards SE. The quarry face is not at 90° to the fold axes, but ongoing excavation gave evidence of a partial fold asymmetry, changing in the degree of vergence along the trend of the fold axes. The folded unit occurs above flat lying
DEFORMATION IN PLIO- AND PLEISTOCENE SEDIMENTS
sandy beds, without a flat fault in between. Overand underlying sequences are sharply limited without any disturbance of the bedding. The truncated antiforms indicate a sedimentation gap and that the deformation happens under only a thin burial (few metres) near the surface. The small reverse fault labelled with an arrow in the figure, associated with the antiform, occurs in the thinned flank of the folded brown-red laminated layer. This layer is thickening between two antiforms (Fig. lie). The thickness changes from 20 cm at the reverse fault to 50 cm ahead of the antiform within the steep flank of the fold, up to 80 cm within the synform. A rough cleavage is diverging fan-like into the synform and flat dipping below the top of the fold hinge in the steep flank. Below the white clay layer, in part the cleavage planes are developed as small shear planes and cause a kind of crenulation pattern within several domains (ellipses) of the laminated clay sands and sandy clays. At some places, the shear planes act as normal faults with small offsets (0.2 to 1 cm downthrow to NW). According to Collinson (1994), sharp anticlines and more gentle synclines are typical of convolute bedding and several tens of metres thick units can be involved in convolute bedding. The structures result from the occurrence of density inversion, so that a denser layer overlies a less dense layer. This is an unstable situation and can lead to sinking of the dense layer into the less dense layer. The lithology of the folded sequence (Fig. 11) could promote deformation by loading, but the indications are rare. Within many synforms, the layer below the foundering white clay was thickened, not thinned. Therefore, additional mechanisms have to be considered. Similar shapes and sizes of folds were reported by Eissmann et al (1995) from regions with glacial deformation. The Cheb Basin was part of the permafrost region, but evidence for the deformation of this sequence due to later glacial processes is absent. The overlying sequence (thickness c. 10 m) of interbedded clayey sands and sandy clays with intercalated clays beds (c. 20 cm thick) is undisturbed and flat lying. Evidence for a period of simple shear deformation comes from the occurrence of the cleavage and its regular orientation which correlates with the fabric of the adjacent Nova Ves shear zone in front of the face. This relationship hints at a temporarily triggered motion within the subsurface sediments around the fault zone, possibly seismically induced. Normal faulting and associated deformation structures Confined-layer faulting with injections along the fault planes and associated inclined folds occurs at a
87
deep level of the Nova Ves II quarry (Fold Layer 1; Fig. 12). The layer is part of the lower Nova Ves Beds, exposed 20 m below the mentioned sequence with garland-like folds, at the foot of the exploitation slopes to the NW and east, within the uppermost 5 m of the pure clay formation. In the quarry, the deformed sequence is subdivided into four layers of interbedded pure clay and quartz-bearing clay of various colour and variable quartz content; from the top to the bottom of the face: (a) (b) (c) (d)
layer IV, light grey clay (up to 80 cm); layer III, thin dark grey clay (2-30 cm); layer II, light grey clay with scattered single quartz grains (1-2 m); in some thin horizons (5-20 cm) quartz and feldspar enriched; layer I, thick dark grey, pure clay layer (1-2 m exposed at this face).
The intense confined-layer deformation of the sequence is developed above and adjoining the Nova Ves Fault, which trends parallel to the NW slope; the features are best visible at the east slope which is at 100° to the main fault zone. Layers I to IV are dislocated along a set of northerly dipping, listric normal faults. The offsets are varying in size in intervals related to the varying distance between the faults, superimposing a wide-span basal folding with low amplitude. This wave-like folding is the basic structure at the 160 m long eastern quarry-face, which underlines the intervals of the associated faulting. Approximately 30 to 40 m wide antiforms interchange with smaller, about 15 m wide synforms, both disturbed by the fault offsets up to a maximum 2 m. The offsets within the dark clay layer I produced the basal 'shark flipper' pattern (Fig. 12a and b). The slightly asymmetrical antiforms consist of shorter northern and longer southern limbs, indicating a sense of vergence to the SE. Small-scale inclined folds (dm- to m-size) superimpose the southerly dipping large-scale limbs increasing in size and vergence up to the top within the dark clay III. The fold axes trend approximately NE, although they are difficult to observe and measure. Overturned anticlines suggest water escape, perhaps coexisting with shear at the sediment surface. These confined-layer faults strike NE to east (azimuth 235° to 245° and 280°) dipping 50° to 75° with throw to NW and north. Sometimes they deviate from vertical in the basal layer, indicating a listric fault style. Normal faults of different scale were concentrated at the southern limbs of the basal anticlines. The most dense and regular fault pattern has been developed in the layers II and III. It comprises fault planes of different sizes: long fault plane traces can be followed sub-vertical through the face at a distance of 5 m; more than 100 smaller faults (visible length: 2-4 m) occur at distances of 2 m to
88
P.BANKWITZEr/lL.
Fig. 11. Garland-like folds at the NW slope of the open cast mine Nova Ves II developed in a layer sequence of interbedded clay, sandy clays and sands beds, adjacent to the Nova Ves Fault, (a) Wavelength of upright folds 10 to 20 m. (b) In some sections of the face the folds are slightly asymmetric: the top of an antiform was thrust (thick arrow) with sense of vergence towards SE. (c) Deformation of the synform: cleavage planes and small-scale normal faulting. At selected sites fine hatched lines hint at domains with rough cleavage patterns.
DEFORMATION IN PLIO- AND PLEISTOCENE SEDIMENTS
89
Fig. 12. Open cast mine Nova Ves II (Vildstejn Formation). Confined-layer faulting and folding within a clay sequence (c. 5 m thickness); interbedded is a brighter horizon of quartz-bearing clay, (a) Schematical sketch of the face (length: 160 m; cross-section vertically exaggerated) with the Fold Layer 1 of Fig. 7. (b) Section from the southern part of (a) (c. 50 m long). Hatched lines: suggested tensional and compressional faults (interpretation), (c) Offsets (20 to 80 cm) of the upper dark clay due to normal faulting; detail from the northern part of the face in (a). Arrows point at fault planes with injections of darker into lighter clay and vice versa.
90
RBANKWITZCTAL.
0.60 m with layer dislocations between 2 m and 80 cm. Dense small planes (length: 1 m to 1 decimetre; Fig. 12c) at distances of 2 to 20 cm occupy the sections between the larger faults. Injections of mobilized clay along the extensional faults into the light grey, quartz-bearing middle layer (IE) are frequent, sometimes it cannot be decided whether injection or a deposition of a clay cutan took place. In part, the patterns are reminiscent of the flame structures (clay smears) that were studied by Lehner & Pilaar (1997) in a Tertiary lignite mine. Evidence for clear injections of the lower dark clay (I) into the light clay (II) above were recognized (Fig. 12c), located at the lower boundary of the light clay, but not passing through the layer. The structures also occur at sites where no fault is realized and no offset exists. The injections are small (2-5 mm, rarely more than 10 mm wide), but mostly long in extension across the light layer (several decimetres), often slightly curved. They are densely developed at distances below 1 cm indicating that they follow cleavage planes. Not only the dark clay (I) was mobilized, but even the quartz-bearing light clay (II) injected the upper dark clay layer (III) sub-vertically along a fault, but in opposite direction to the throw and laterally into the layer III, starting from an area of offset (seen above the hammer in Fig. 12c). Further towards the top, dismembered beds cannot be observed, because the deformed sequence was cut by an erosion level and finally overlain by flat lying sands. It is not to exclude that above the preserved part of this section a disturbed layer has been developed, taking into consideration the occurrence of some locations with non-clarified mixing structures of various clay types within the section. The fault pattern of this confined-layer deformation resembles features in Quaternary sediments in Scotland and is very similar to the corresponding structures of Ringrose (1988, in: Maltman 1994, p. 290, fig. 9.28a) pointing out formation due to earthquake shock. The confined-layer folding of the clay sequence in Nova Ves is of the same type as published by Allen (1986, p. 150, fig. 9.1c) considered as 'rumpling' of newly deposited water-laid sediments, which has been used as an indication of palaeoseismicity.
Discussion: evidence for the relationship between soft sediment deformation, earthquake-induced water escape/liquefaction and gas escape Earthquake swarm activity and permanent mantle (CO2) degassing of the Tertiary Cheb Basin and its surroundings can be one source of the sediment mobilization and deformation, as well as periglacial
and endogenic processes. Recognition of liquefaction features and confined-layer deformation structures gives a tool for constraining the reason for soft sediment mobilization and the generation of associated structures, such as disturbance of bedding, folding, shear planes, cleavage and confined-layer faulting with injections. All individual features are exposed within several open pits of the studied region (Fig. 2b). All sites of recognized soft sediment deformation are not far from major faults associated with previous volcanic or seismic activity, or which have recently been affected by crustal movements or seismic events. Therefore we do not exclude a model of triggered initiation of the recognized faulting and folding above or adjacent to major faults, but the trend of these features is considered to be controlled by the regional stress field and this was stable during Pliocene and Quaternary times (Peterek et al 1997). Field evidence for this model comes from the confined-layer fracturing and folding at the Nova Ves quarry and from the structural coincidence of the orientations of fault planes and fold axes with the geometry of the Nova Ves Fault, which is also exposed within the quarry. Typical structures due to earthquake shock, e.g. particularly a fault-grading stratigraphy (closely spaced faults with throws increasing upwards and passing upwards into broken-up and ball-and-pillow structures) according to Ringrose (1988, in: Maltman 1994), are incompletely preserved, but suggest that seismic triggering may have supported the initiation of the exposed deformation structures. Concerning the fractured folds near the base of the Nova Ves Beds (Fig. 12), we suggest that these structures may have formed by one single process and within a relative short time span. The structures show overprinting of extension on compression without a large degree of net mass transfer in a single direction. This may have been produced by the passing of compressive and extensional surface waves in water-saturated sediments during seismic events. The vergence of these structures, away from the epicentre, is an indication of their origin. This folded sequence (Fold Layer 1) is located within and closely to the Nova Ves Fault and it is suggested that the basal folding represents shearing-induced forced folds above a deep-reaching shear zone.
Seismic triggering of soft sediment deformation The mobilization of clay within the deformed lower Nova Ves clay formation suggests that the sediments have not been deeply buried, as any consolidation of the clay would retard its mobilization. Such features in wet soft sediments in seismic active regions can be considered an indicator of past earthquakes
DEFORMATION IN PLIO- AND PLEISTOCENE SEDIMENTS
(Allen 1986; Mailman 1994). The mobilization of clay in Fig. 12 above the Nova Ves shear zone is restricted to a single clay sequence (c. 5 m) and is located in the surroundings of the epicentral area of Novy Kostel. These facts suggest seismically induced deformation with partially liquefaction, e.g. represented by flame-like injections. Liquefaction features generate preferentially in newly deposited water-laid sediments. Therefore it is important to distinguish between structures of syndepositional origin and earthquake-induced features. It is widely accepted that the occurrence of seismically generated liquefaction features depends upon the epicentral distance and the magnitude (and intensity) of the earthquake. Galli (2000) has shown that there is a relationship between magnitude and epicentral distance of the feature that is not only valid for strong (M > 6.5), but also for moderate earthquakes (M<5.9 to M>4.2), with smaller distances from the epicentre. The highest magnitudes of individual recent events of M = 4.6 in 1985/86 and M = 3.7 in 2000 (Horalek et al 2000, Klinge & Plenefisch 2001) were qualified to mobilize subsurface sediments and to produce liquefaction features in an adjoined area 5 km in diameter. A high probability exists that the region was seismic active also in Neogene time. Schneider (pers. comm. 2001) suggests that the occurrence of the N-S trending seismically active zones in Central Europe depends on the continuing convergence with the Adriatic Plate, causing the ongoing elevation of the Alps and within the foreland and further to north the compensation by strike slip movements associated with seismic activity. Evidence for that comes from the earthquake zones of Europe. This geodynamic process started before Neogene. Seismic activity within the Regensburg-Leipzig-Rostock zone, involving the epicentral area of Novy Kostel and from the area to the west of Skalna, is to suppose also during Pliocene time. The seismicity at that time is not known, neither is the intensity of past seismic events nor the location of the epicentres. Palaeoseismic activity is to be expected in this region, because neotectonic studies give evidence for geodynamic processes in Pliocene and earlier times (Peterek et al. 1996).
Gas escape triggered deformation of soft sediments The fluid transport velocity of the mantle volatiles through the upper crust was determined as 50 m/day. From similar investigations in the Mammoth Mountains, Sorey et al. (1998) reported 10^1-0 m/day. Such a rapid flow indicates that gas-water interaction during transport is limited to the upper groundwater aquifer and seems to support the
91
hypothesis of a free gas transport along channel-like conduits from the location of the hypocenter up to the aquifer. The main aquifer in the Cheb Basin is located beneath the Vildstejn Formation. Blow out of fluids (50 m high) from the borehole Hll near Frantiskovy Lazne, which penetrates the Vildstejn Clay Formation, points to a high fluid pressure which remains stable due to permanent CO2 migration from the depth. We suggest that the pressure condition within the mineral water aquifer beneath the Vildstejn Clay Formation is one reason for liquefaction processes in the soft sediments. The mentioned gas escape channels in the clay of the Nova Ves Fault (|mm size) are connected with measured gas streams and clearly result from the gas escape along earlier formed, possible earthquakeforced shear planes.
Conclusions Previously unrecognized major fault zones within the unconsolidated sediments of the Neogene Cheb Basin (Czech Republic) correlate with structures of the surrounding Variscan basement. Two N-S-trending faults complete the inventory of the seismically active MidEuropean Regensburg-Leipzig(-Rostock) Zone. The Pocatky-Plesna Zone (PPZ), associated with the main epicentral line of Novy Kostel, form a 20 m high scarp, which separates two crustal blocks and cross-cuts the Marianske Lazne Fault. Late Quaternary deposits in front of the scarp restrict the timing of the scarp formation. Evidence for the deep-seated nature of the PPZ comes from the hypocentres, which occurred at depth between 7 and 15 km and from the Bublak mofettes with upper mantle CO2 degassing. The ENE strike of another fault recognized for the first time in Pliocene clay, the Nova Ves shear zone, suggests that it forms part of the northern boundary of the Ohfe Graben. A unique structure was recognized within this shear zone using SEM techniques: the upper mantle CO2 escapes through the clay formation along this major fault, and uses tiny microchannels of micro-metre size (0.001 to 0.005 mm in diameter). Intense folding of clay layers within and adjacent to the dextral Nova Ves shear zone represents shearing-induced forced faults in part associated with cleavage. The deformation planes and fold axes trend NE, related to the maximum horizontal axis of compression. There is a possibility that seismic shaking and gas escape could have triggered soft sediment mobilization under subsurface conditions in the Nova Ves area during the sedimentation of the lower Nova Ves Beds (clay sequence) above the Nova Ves shear zone. The regular orientation of the resulting confined-layer fracturing and folding associated with
92
P.BANKWITZ£TAL.
cleavage planes may depend on the forced-fold formation and on the location of the epicentres. The occurrence of mantle degassing at both faults (Pocatky-Plesna Zone and Nova Ves Fault), together with the seismic activity of the Pocatky-Plesna Zone and its scarp formation since Pleistocene time, hints at a palaeoseismic activity in this region. Evidence from the Soos area suggests that CO2 escape has been continuing for at least the last 12,000 years (Rehakoval988). This paper is written in honour of Prof. Peter Moller, GeoForschungsZentrum Potsdam. The research of 'Neotectonic crustal activity' and 'Fluids' was supported by Deutsche Forschungsgemeinschaft (Germany) grants BA 1184/8, SCHN 251/2-1, STR 376/6, KA 902/7 and HE 2177/7. We thank J. Tesaf (Institute for Natural Health Resources, Frantiskovy Lazne) for measuring of gas composition and Kemat Enterprise Skalna for the permission to work in its open mines. The study was promoted by the discussion during the Meeting Subsurface Sediment Mobilisation and further with Dr. Plenefisch and the Czech colleagues, Dr. Horalek, Dr. Nehybka and Dr. Skacelova. We appreciate reviews by E. R. Phillips, Edinburgh, and F. M. Van der Wateren, Amsterdam and acknowledge their constructive criticism.
References ALLEN, C. R. 1986. Seismological and paleoseismological techniques of research in active tectonics. In: WALLACE, R.E. (ed): Active tectonics. National Academy Press, Washington. 148-154. BALZ, P. 1908. Das Egerlander Braunkohlenbecken. Gluckauf, 52,1830-1842. BANKWTTZ, P., KAMPF, H., STORR, M. & SCHNEIDER, G. 2001. Shear faults in Tertiary clay formation of the Eger Basin (Czech Republic) and their role for the degassing of the uppermost mantle. Berichte Institut Geologic Paldontologie, Universitdt Graz, 3,354-363. BEHR, H.-J., DURBAUM, H.-J. & BANKWTTZ, P. 1994 (eds). Crustal structure of the Saxothuringian Zone: Results of the deep seismic profile MVE-90(East). Zeitschrift fur geologische Wissenschaften, 22, 647—769. BRAUER, K., KAMPF, H., STRAUCH, G. & WEISE, S.M. 2003. Isotopic evidence (3He/4He, 13CH20) of fluid triggered intraplate seismicity. Journal of Geophysical Research, 108, 2070; cloi: 10.1029/2002JB002077. COLLINSON, J. 1994. Sedimentary deformational structures. In: MALTMANN, A. 1994 (ed.) The Geological Deformation of Sediments. Chapman & Hall, London, 95-125. EISSMANN, L. 1987. Lagerungsstorungen im Lockergebirge. Geophysikalische Veroffent-lichungen der Karl Marx-Universitat Leipzig, HI, 7-77. EISSMANN, L., Lnr, T. & WANS A, S. 1995. Elsterian and Saalian deposits in their type area in central Germany. In: EHLERS, J., KOZARSKI, S. & GIBBARD, P. (eds) Glacial Deposits in Northeast Europe. Balkema, Rotterdam. 439-464. GALLI, P. 2000. New empirical relationships between mag-
nitude and distance for liquefaction. Tectonophysics, 324,169-187. Geologicka Mapa CSSR, 1 : 200 000, 1990, sheet Karlovy Vary - Plauen. Ustredni ustav geologicky, Prague. GRUNTHAL, G., SCHENK, V., ZEMAN, A. & SCHENKOVA, Z. 1990. Seismotectonic model for the earthquake swarm of 1985/86 in the Vogtland/West Bohemian focal area. Tectonophysics, 174, 369-383. HAVIR, J. 2000. Stress analyses in the epicentral area of Novy Kostel (Western Bohemia). Studia geophysica et geodaetica, 44, 522-536. HECHT, L., VIGNERESSE, J. L. & MORTEANI, G. 1997. Constraints on the origin of zonation of the granite complexes in the Fichtelgebirge (Germany and Czech Republic): evidence from a gravity and geochemical study. Geologische Rundschau, 86, S93-S109. HORALEK, J., FISCHER, T, BOUSKOVA, A. & JEDLI£KA, P. 2000fl. The western Bohemia/Vogtland region in the light of the Webnet network. Studia geophysica et geodaetica, 44,107-125. KAMPF, H., BRAUER, K., STRAUCH, G. & WEISE, S.M. 1999. Indications for an active magma chamber in the upper mantle below the western Eger Graben, Czech Republic. 1999 Spring Mtg AGU, Suppl. to EOS April 27,378. KLINGE, K. & PLENEFISCH, T. 2001. Der Erdbebenschwarm 2000 in der Region Vogtland/NW-B6hmen. Deutsche Geophysikalische Gesellschaft, Mitteilungen, Nr. 2/2001,11-21. LEHNER, F. K. & PILAAR, W. F. 1997. The emplacement of clay smears in synsedimentary normal faults: inferences from field observations near Frechen, Germany. In: M0LLER-PEDERSEN, P. & KOESTLER, A. G. (eds) Hydrocarbon Seals. Norweg. Petrol. Soc., Special Publications, 7,39-50. MALKOVSKY\ M. 1987. The Mesozoic and Tertiary basins of the Bohemian Massif and their evolution. Tectonophysics, 137,31-42. MALTMAN, A. 1994. Deformation structures preserved in rocks. In: MALTMAN, A. (ed.) The Geological Deformation of Sediments. Chapman & Hall, London, 261-307. MANDL, G. 1988. Mechanics of tectonic faulting. Elsevier, Amsterdam, 407 pp. MCCALPIN, J. P. & NELSON, A.R. 1996. Introduction to Paleoseismology. In: McCALPiN, J. P. (ed.) Paleoseismology. Academic Press, San Diego, 1-32.
ML£OCH, B., SCHULMANN, K., SRAMEK, J., MANOVA, M., POKORNY, L., FlALA, J. & VEJNAR, Z. 1997. The
Saxothuringian zone. In: VRANA, S., STEDRA, V. (eds) Geological Model of Western Bohemia in Relation to Deep Borehole KTB in the FRG. J. Geol. Sci. Ser. Geol. 47,51-61, Czech Geological Survey, Prague. MRLENA, J. 2000. Vertical displacements in the Novy Kostel seismoactive area. Studia geophysica et geodaetica, 44,336-345. NEHYBKA, V. & SKACELOVA, Z. 1995. Seismotectonic analysis of the seismological measurements in the Kraslice network. Vestnik Ceskeho geologickeho ustavu, 70,97-100. NEUNHOFER, H. & Gtira, D. 1990. Detailed investigation of the great earthquake swarm in Western Bohemia by the local Vogtland network. In: BORMANN, P. (ed.) Monitoring and analysis of the earthquake swarm
DEFORMATION IN PLIO- AND PLEISTOCENE SEDIMENTS 1985/86 in the region Vogtland/Western Bohemia. Zentralinstitut fur Physik der Erde, Veroffentlichungen, 110,124-164. OWEN, G. 1987. Deformation processes in unconsolidated sands. In: JONES, M.E. & PRESTON, R.M.F. (eds) Deformation of Sediments and Sedimentary Rocks. Geological Society, London, Special Publications, 26, 11-24. PETEREK, A., SCHRODER, B., NOLLAU, G. 1996. Neogene Tektonik und Reliefentwicklung des nordlichen KTBUmfeldes (Fichtelgebirge und Steinwald). Geologica Bavarica,Wl,l-25. PETEREK, A, RAUCHE, H., SCHRODER, B., FRANZKE, H-J., BANKWITZ, P. & BANKWITZ, E. 1997. The late- and post-Variscan tectonic evolution of the Western Border fault zone of the Bohemian massif (WBZ). Geologische Rundschau, 86,191-202. REHAKOVA, Z. 1988. Biostratigraphy and Palaeoecology of Diatom-bearing Sediments within the Soos Basin in Western Bohemia. Czechoslovakia, 10th DIATOMSymposium, 1988,407-418. RIEDEL, W. 1929. Zur Mechanik geologischer Brucherscheinungen. Centralblatt Mineralogie, Geologie, Paldontologie, B 1929,354-368. SCHNEIDER, G. 1993. Beziehungen zwischen Erdbeben und Strukturen der Siiddeutschen GroBscholle. Neues Jahrbuch Geologie Paldontologie, Abh., 189, 275-288. SHRBENY, O. et al 1994. Tercier Ceskehu Masivu Tertiary of the Bohemian Massif. In: KLOMINSKY, J. (ed.) Geologicky atlas Ceske republiky. Statigrafie. SKVOR, V. & SATTRAN, V. (eds) 1974. Krusne Hory, zapadni cast, 1:50.000. Ustfedni ustav geologicky, Praha. SOREY, M.L., EVANS, W. C, KENNEDY, B.M., FARRAR, C. D., HAINSWORTH, LJ. & HAUSBACK, B. 1998. Carbon dioxide and helium emissions from a reservoir of magmatic gas beneath Mammoth Mountain, California. Journal of Geophysical Research, 103, 15.303-15.323. SPICAK, A. & HORALEK, J. 2000. Possible role of fluids in the process of earthquake swarm generation in the West Bohemia/Vogtland seismoactive region. Tectonophysics, 336,151-161
93
SPICAKOVA, L., ULICNY, D. & KOUDELKOVA, G.C. 2000. Tectonosedimentary evolution of the Cheb Basin (NW Bohemia, Czech Republic) between late Oligocene and Pliocene: a preliminary note. Studia geophysica et geodaetica, 44,556-580. SVANCARA, J., GNOJEK, L, HUBATKA, F. & DEDACEK, K. 2000. Geophysical field pattern in the west Bohemian geodynamic active area. Studia geophysica et geodaetica, 44,307-326. TCHALENKO, J.S., AMBRASEYS, N.N. 1970. Structural analysis of the Dasht-e Bayaz (Iran) earthquake fractures. Geological Society of America, Bulletin, 81, 41-60. VERSCHUREN, M. & VAN RENSBERGEN, P. 2001. An integrated approach to clay tectonic deformation and field guide to its Belgian Outcrops. In: PARIZE, O., VERSCHUREN, M. & VAN RENSBERGEN, P. (eds) Subsurface Sediment Mobilization Field Guidebook. Association des sedimentologues Francais, 25. VTELENSKY, J., SANTRUCEK, P. & HARTMAN, V. 1990. Jfly zapadnich Cech - oblast chebske panve (Clays of west Bohemia - the Cheb Basin area). Sbornik Geologickych Ved, Technologic, geochemie, 25,9-228. WEINLICH, F. H., BRAUER, K., KAMPF, H., STRAUCH, G., TESAR, J. & WEISE, S. M. 1999. An active subcontinental mantle volatile system in the western Eger rift, Central Europe: gas flux, isotopic (He, C and N) and compositional fingerprints. Geochimica et CosmochimicaActa, 63,3653-3671. WEISE, S. M., BRAUER, K., KAMPF, H., STRAUCH. G. & KOCH, U. 2001. Transport of mantle volatiles through the crust traced by seismically released fluids: a natural experiment in the earthquake swarm area Vogtland/NW Bohemia, Central Europe. Tectonophysics, 336,137-150. WENDT, J. 1999. Zur Geokinematik im sachsischen Vogtland. Deutsche Geodatische Kommission, C, nr. 517,1-124. WIRTH, W, PLENEFISCH, T., KLINGE, K., STAMMLER, K. & SEIDL, D. 2000. Focal mechanisms and stress field in the region Vogtland/Western Bohemia. Studia geophysica et geodaetica, 44,126-141.
This page intentionally left blank
Tubular structures of northern Wairarapa (New Zealand) as possible examples of ancient fluid expulsion in an accretionary prism: evidence from field and petrographical observations BEATRICE LEDESERT, CHRISTOPHE BURET, FRANK CHANIER, JACKY FERRIERE & PHILIPPE RECOURT UMR Processus et Bilans des Domaines Sedimentaires, FR 1818 CNRS, USTL, SN5, 59655 Villeneuve d'Ascq cedex, France (e-mail: [email protected]) Abstract: The Cape Turnagain area is located on the inboard portion of the Hikurangi subduction margin, on the northern Wairarapa coast of the North Island of New Zealand. A 4.5 km long coastal section of sea cliffs of Mio-Pliocene sediments contains numerous tubular carbonate-rich concretions. Their morphology and petrographical observations suggest they were possibly formed by fluid flows of carbonate-rich water through a silty sediment. These tubular concretions could be fossil fluid expulsion structures similar to dewatering chimneys described offshore in New Zealand. The external diameter of the concretions observed in situ reaches 60 cm and internal canal up to 4 cm. Up to four canals are encountered in a single concretion. A positive relationship is observed between the chimney size and the number of canals or cumulative diameter of canals, suggesting that the size of the concretion is a function of the fluid which flowed through the plumbing network. The increased number of tubular concretions in upper Miocene siltstones compared to overlying Pliocene strata could be linked to a compressive event that caused overpressuring and the expulsion of fluids through the sediment pile.
In active tectonic environments such as subduction margins, fluid seeps frequently occur offshore (Guilhaumou & Larroque 1995; Barbados: Mascle et al 1990; Japan: Agar 1990; Oregon: Lewis & Cochrane 1990; New Zealand: Lewis & Marshall 1996). They are sometimes associated with vent faunas, which are composed of gastropods, vestimentiferan tube worms and large bivalves (e.g. Lewis & Cochrane 1990; Lewis & Marshall 1996). Along convergent margins, great amounts of fluids are derived by compressional dewatering, mineral dewatering and chemical alteration of organic matter from water-saturated trench and oceanic sediments rapidly underthrusted and buried (Moore & Vrojlik 1992). Expelled fluids are often just above the ambient seawater temperature (Hashimoto et al 1989; Lewis & Marshall 1996) and are thus called 'cold seeps'. The eastern coast of New Zealand (NZ) North Island is part of the well-developed accretionary prism related to the Hikurangi subduction margin (Fig. 1). Numerous mud volcanoes and oil seeps are found within this coastal range (Ridd 1970; Field et al 1997). Thus, fluids appear to circulate intensely within the Hikurangi accretionary prism today. The geochemical and isotopic composition of active fluid flow has been studied from several mud volcanoes and gas seeps from this area (Kvenvolden & Pettinga 1989; Rogers et al 1999). Lewis & Marshall (1996) provide an overview of dewatering phenomena on NZ convergent margins, based on the description and interpretation of 13 sites, two of which were studied extensively by Orpin (1992,
1997). Lewis & Marshall (1996) show that seeps occur at the seabed in three types of geological environments: (1) on mid to upper slope ridges near the seaward edge of presubduction Neogene sediments; (2) near the shelf-edge and nearby canyons heads, particularly those in the vicinity of strike-slip deformation zones; and (3) where slope sediments have collapsed onto thick sedimentary basin-fill. Seeps are characterized by at least one of the following features: carbonate chimneys or crusts, plumes of low-density fluids and distinctive chemosynthetic faunas. Chimneys and crusts are formed from carbonate ions resulting from the metabolism of chemosynthetic microbic organisms (Han & Suess 1989). They are sometimes inferred to be the result of seepage of organic-rich fluids, some of them at the height of the last glacial age (Orpin 1992, 1997). At places where subduction is occurring, there may be some additional input of fluid from overpressured sediment at depth (Moore & Vrojlik 1992). There are only five published locations of chimneys offshore NZ (Fig. 1): (1)
(2) (3)
uppermost Otago slope. This is the bestdocumented example of chimneys in NZ (Orpin 1992, 1997). They are 10 to 25 cm in diameter with a 0.5-2.5 cm wide canal. Head of Pegasus Canyon, north of Banks peninsula (Orpin 1992,1997). Around Cook Strait Canyon (Lewis & Marshall 1996). They are around 16 cm in diameter with a 2-5 cm wide central canal.
B.LEDESERTCTAL. Off Hawke Bay, in the vicinity of an active plume (Lewis & Marshall 1996). The chimney is 16 cm wide with a 8 cm wide canal. A 2 m long chimney from a canyon wall on the upper Taranaki slope (347 m depth; Orpin 1997, p. 60).
None of the chimneys in these five locations were active by venting when they were trawled, but all had their central canal open and free of sediment. Tubular concretions have also been discovered in the cliff exposures and on the wave-cut platform of Cape Turnagain in uplifted Mio-Pliocene sediments (Figs 1 and 2). This study presents detailed field and petrographical data used to discuss their possible relationship to ancient fluid flow processes in a subduction zone setting. Similar structures have been described before (e.g. in the Cretaceous and in the mid-Miocene of the North Island East Coast; Neef 1984), but this study is, to the authors' knowledge, the first to investigate them in detail and interpret them as fluid escape structures (rather than burrows or diagenetic concretions).
Geological setting and description of the site The studied area is located on the eastern coastal range of the northern Wairarapa, North Island. This range corresponds to the highest ridge axis of the Hikurangi accretionary margin complex and is composed of Mesozoic to early Cenozoic basement sediments and Neogene sedimentary basins (Lewis & Pettinga 1993). The Neogene basins are contemporaneous with the subduction of the Pacific plate beneath North Island and are infilled with thick turbiditic formations and massive siltstones (Field et al 1997). The concretions described in this study were located on the northern coast of Cape Turnagain. The exposed cliff section goes from Wangaehu to Arataura (Fig. 2) and is composed of upper Miocene and lower Pliocene grey marine mudstones to siltstones. According to Moore (1981), the base of the Wangaehu Mudstone is midMiocene in age (Waiauan or possibly upper Lillburnian), while the top of the formation is early Pliocene (Opoitian). Microfossils indicate a paleo water depth of more than 200 m (Buret 1996). The sedimentary pile is poorly bedded. The bedding, measured at its base, near Wangaehu, dips at 45° to the SE. The site is a 4.5 km long section (Fig. 2) and observations can be made on cliff exposures and on the wave-cut platform (Fig. 3). Approximately a quarter of the cliff length is covered with vegetation, limiting rock exposure. The section is peculiar for the abundance of carbonate-rich concretions. Tubular concretions
Fig. 1. Location and geodynamical context of the studied area (after Chanier et al 1999). A number in a circle indicates the location of concretions cited in the literature.
do not appear in the overlying Pliocene sediments of similar lithology. Neither field evidence of vent faunas nor the presence of structural features that might influence fluid flow could be observed on the section. Three main concretion types can be distinguished: (1)
(2)
(3)
tubular concretions which have been studied in detail due to their morphological similarity with features attributed to fluid escape chimneys observed offshore (Lewis & Marshall 1996). They have been seen all along the section (Figs 2 and 3) and form the focus of this paper. There is no constraint on their whole length; cylindrical concretions without cavity which are up to 2 m in diameter. There is no constraint on their whole length either; spherical concretions which are tens of centimetres to 2 m in diameter. They show a more cemented matrix compared to the surrounding rock. They are composed of the same material (matrix and cement) as the cylindrical concretions without cavity.
Types (ii) and (iii) are similar to concretions frequently found in calcareous or detrital formations (El Albania al 2001).
TUBULAR STRUCTURES AND FLUID EXPULSION
97
Fig. 2. Location of the study area within New Zealand North Island. Concretions were observed in situ in the cliffs and locally on the wave-cut platform. The location of the three sub-sites studied in detail is indicated (sites 2a, 2g and 3a), as well as that of Fig. 3a.
Samples and analytical methods Measurements of dip and dip direction of the tubular concretions were collected as well as the measurements of the outer diameter and internal canal. Only 57 in situ concretions could be studied in detail along the 4.5 km section, 30 of which are concentrated into three small sub-sites (18 to 60 m2, Fig. 3). Seven tubular concretions that showed different morphological features and the surrounding host rock were sampled for further petrographical and mineralogical study. The sampled concretions were selected on criteria such as the size of the central canal (0.3 to 4 cm width), the smoothness of internal canal wall, the abundance of canal infill (absence or partial infill) and the presence or absence of geodic neoformed carbonates. All of the
sampled tubular concretions are approximately the same size (5 to 10 cm external diameter). Polished thin sections (30 urn thickness) were made for optical and scanning electron microscopy (SEM). The optical microscope is an Olympus BX 60 equipped with a digital video camera. The SEM is a Cambridge Stereoscan apparatus linked with a Kevex EDS Superdry device (cooled by Pelletier effect) allowing qualitative EDS microchemical analyses. The carbonate content has been measured by Bernard's calcimetry method using dilute HC1. X-Ray diffraction of the carbonate cement powders was performed with a Philips PW1710 diffractometer (Ni filtered Cu Ka radiation) from 3 to 60° 29 angles.
98
B.LEDESERT£TAL.
Fig. 3. Views of the cliff section and wave-cut platform along the studied section: (a) general view west (its location is shown on Fig. 2b). Most of the boulders seen on the shore are pieces of large hollow tubular concretions (up to 1 m wide); (b) detailed view showing several tubular concretions buried in the sea cliff indicated by arrows (site 2a).
Results (3)
ments. Several canals are sometimes observed within a single concretion; and, the geode.
Field observations Typical vent faunas were looked for in the vicinity of the tubular concretions but could not be observed. In addition, no structural features that could possibly focus fluids such as faults, folds or changes in lithology were encountered. Morphology of tubular concretions. The cemented concretions generally protrude from the surrounding uncemented rock and can be easily sampled. They are composed of three parts (Fig. 4): (1) (2)
the concretion itself which is sometimes subdivided into two concentric parts of different colour and induration state; the central canal, which is generally only partly infilled with highly cemented sedi-
These three parts will be described in detail in the petrography and mineralogy section. The tubular concretions are either linear or branched (Figs 5a and b), their shape is frequently tortuous and the number of canals varies from one to four (Figs 5c and d). The external diameter of the whole in situ concretions ranges from a few centimetres to 60 cm (Fig. 5). Larger ones (up to 100 cm) are observed as boulders on the wavecut platform. The diameter of the canal does not exceed 4 cm for the concretions observed in situ. The relationship between the external diameter of the concretion and the diameter of the central canal (or cumulative diameter of the canals in case of several canals within a single concretion) is given in Figure 6 for the 57 studied concretions. It shows a global positive linear relationship between these parame-
TUBULAR STRUCTURES AND FLUID EXPULSION
99
Attitude to bedding. The attitude in space of tubular concretions (Fig. 8) is shown by the projection of their dip and dip direction in a Schmidt diagram (lower hemisphere). They are approximately normal to bedding but show a large degree of scatter (± 30 °). Petrography and mineralogy Surrounding rock. The tubular concretions are found within poorly sorted mudstone composed of silt- to fine sand-sized quartz and feldspar grains, mica platelets and a suite of subordinate heavy minerals, surrounded by a lightly carbonate-cemented clay matrix. Sediments show considerable bioturbation and contain many microfossils (globigerinidae) and rare fragmented macrofossils (Pinna). They were deposited at outer shelf to upper bathyal depths (Field et al 1997). The carbonate content of the bulk rock is low, around 1 wt.%, but may reach 9 wt.% in the immediate vicinity (< 1 m) of a concretion. Fig. 4. Schematic transverse cross-section of the concretions, which are made of three parts: (1) the concretion itself (possibly made of two, or more, concentric parts of variable colour and cement content, separated on the figure by a dashed line); (2) the canal generally partly infilled with fine-grained micritic sediments; and (3) the geode.
ters, whatever the number of canals, indicating that the larger the plumbing network, the wider the concretion around it. The small number of multiple concretions does not allow the establishment of such a relationship for each individual group (two, three or four canals). In addition, the concretions observed at one of the three small sites (site 2g, Figs 2 and 7a) also show a positive relationship between the external diameter of the concretion and the number of canals within it (Fig. 7b). Simple statistics have been performed for the three small sites of concentrated concretions (Fig. 2), the results of which are given in Table 1. The proportion of tubular concretions, compared to that of all concretions, is variable: from 32 to 92% (Table 1). The abundance of tubular concretions varies, with densities ranging between 0.18 and 0.44 concretion/m2 across the three sites. Considerable variation is also observed for the number of canals in each tube, from 0% of multiple canals (meaning only one canal) to 54%. In addition, the mean external diameter of all concretions ranges from 3.3 to 28.5 cm, while that of the canals is between 0.7 and 1.23 cm. As shown by these results obtained on three different sites, it appears that the morphological characteristics of the tubular concretions are quite variable.
Concretions. The concretions are composed of the same detrital sediment as the surrounding rock, but the carbonate content is higher, ranging from 40 to 50 wt.%. These values and petrographical observations indicate a high detrital component within the concretions, generally higher than 50%. Carbonate mineralogy obtained by XRD on the cement shows that calcite (JCPDS 05-0586) and dolomite (JCPDS 36-0426) are the newly-formed carbonate phases, in decreasing order of abundance. Quartz, small amounts of feldspar and clay minerals are also present in the concretions, probably of detrital origin, as was observed in the surrounding rock. Infilled burrows (Fig. 9) are commonly observed within the concretions suggesting bioturbation occurred in unconsolidated sediments before the concretions were cemented. These bioturbations are similar to those found in the surrounding rock. Globigerinidae, similar to those described in the Miocene mudstone, are also common in the concretions. Canal geometry and infill The perimeter of the canals sometimes shows spiral rifling (already observed by Lewis & Marshall 1996) or small en echelon tension vugs (200 um long) filled with pure calcite or magnesian calcium-carbonate, presumably burial related. The perimeter can also be either simple and sharp, clearly crosscutting the mudstone matrix, or complex showing bioturbations frayed as dendritic shapes within newly formed carbonate phases (Fig. 9). These dendrites indicate that the sediment was not cemented when the flow responsible for the crystallization of the carbonate tube walls occurred. Alternatively, the sharp border would indicate a consolidated sediment at the time of the fluid flow.
100
B.LEDESERTCTAL.
Fig. 5. Examples of ranges of morphologies of the concretions: 5a and b show opposite views of a branched concretion around 1.5m long. It has eroded out of the sea cliff and is exposed on the beach face; 5c shows a cross-section of an in situ concretion with a single canal and 5d shows a cross-section of a concretion with two oblate canals. Note that c and d are smaller concretions, <20 cm in diameter. The canals are indicated by arrows.
The canal is frequently partially infilled with cemented sediment that is generally finer, with fewer detrital components, and a greater amount of micrite than sediment that forms the concretions. The infills are also richer in carbonate (up to 75 wt%). XRD shows only pure calcite. Preserved evidence of bioturbation is rare and was observed in only one thin section. Some globigerinidae were also observed in thin section. They have not been dated, but future research will undertake foraminiferal analysis in order to determine the respective ages of the sediment and of the canal infill. This future work might help determine the age of the infilling and distinguish if it occurred per ascensum or per descensum. This canal
infill is sometimes highly porous, mostly near the outer perimeter of the canal, generally heterogeneous, and shows concentric structures of contrasting grain size, and local accumulations of coarse particles. Geode. Four or fewer minerals are found in the geode and form zonations. The outer part of the geode is made of sparite-like Ca-Mg carbonates on which spheroids of fibrous silica have grown, locally associated with rare euhedral quartz crystals. The last generation of minerals is made of euhedral spary pure calcite (Fig. 10). Some of the concretions are free of siliceous phases. These neoformed phases probably formed during late-stage diagenesis.
TUBULAR STRUCTURES AND FLUID EXPULSION
101
Fig. 6. Positive relationship between the diameter of 57 concretions and that of their central canal (or cumulative diameter in case of multiple canals). Multiple canal concretions are rare: five with two canals, one with three canals, two with four canals.
Discussion Recent work in the Gisborne area, 300 km to the north, indicates that some tubular structures with morphological similarities to the examples from the Wairarapa may be attributed to large burrows (Mazengarb & Francis 1985; Hayward 1989). Orpin (1997) argues that a burrow origin could be questioned, but the largest concretions and canals observed at Cape Turnagain are possibly too large to be attributed to burrows. Lewis & Marshall (1996) and Orpin (1997) have documented similar carbonate structures sampled on the continental slope offshore New Zealand and in some cases have noted that their locations are close to highly sonar-reflective plumes attributed to coastal fluid expulsion. Three active plumes have been noted in a large offshore area around Cape Turnagain. Two of them have been observed off Hawke Bay at 900 m and 1138m deep, and another one is reported at a 725 m depth on the flanks of Uruti Ridge, on the continental slope off Wairarapa (Lewis & Marshall 1996). The palaeowater-depth of the sediments in which the studied concretions have formed is inferred to be greater than 200 m, according to microfossils (Buret 1996). Thus, they were probably formed below the shelf break, along the upper to middle continental slope, as summarized by Lewis & Marshall (1996) and Orpin (1997) for the chimneys observed offshore.
Some of the results exposed in this study are in agreement with an assessment of a fluid expulsion origin of the Cape Turnagain tubular concretions. According to Van Damme & Lemaire (1990), the dendritic mixing features encountered around the perimeter of some of the canals suggest the intrusion of a low viscosity fluid (presumably water) within the unconsolidated siltstone from which the concretion is made. The relationship established between the width of the concretion and canal diameter (or cumulative diameter in case of several canals) or the number of canals may be related to the volume of flow that has crosscut the sediment. Ledesert et al. (1993) suggest that a relationship exists between the width of hydrothermally altered zones (which can be compared to the width of the cemented concretions) and the ability of fractures to conduct fluids through a rock: the larger the altered zone, the greater the amount of fluid that flowed through the fracture it surrounds. Similarly, we can assume that the crosssection of the cemented area around the plumbing network depends on the amount of fluid that has flown through the central canal (or canals) or on the concentration of carbonate-forming ions in the fluid. Furthermore, the spiral rifling observed within some of the canals suggest that the concretions have formed due to a high-velocity fluid flow analogous to offshore examples summarized by Lewis & Marshall (1996). These observations and interpretations are in
102
B.LEDESERT£TAL.
Fig. 7. Detailed study of site 2g: (a) plan-view map of the site, where each rounded shape represents a tubular concretion; and (b) linear relationship between the diameter of the concretions and the number of central canals in each concretion at site 2g.
agreement with those presented by Han & Suess (1989) who show that high fluid flow rates are characterized by plumes (when active) and chimneys (like in the present study), whereas those with slow fluid rates tend to produce crusts. It is now well
known from studies on convergent margins that numerous carbonate-rich concretions can be located around each fluid seep (Lallemant et al. 1990)whatever the path followed by the fluids: either preferentially focused along the decollement thrust zone like
103
TUBULAR STRUCTURES AND FLUID EXPULSION Table 1. Simple statistics obtained from the three sites studied in detail Site
2a
Surface area of the site (m2) No of concretions No of tubular concretions Proportion of tubes/concretions (%) Density of concretions (/m2) Density of tubes (/m2) Proportion of simple canal/ total tube nb (%) Proportion of double canals / total tube nb (%) Proportion of triple canals / total tube nb (%) Proportion of quadruple canals / total tube nb (%) Proportion of multiple canals Mean diameter of concretions (cm) Mean diameter of canals (cm)
18 25 8 32
in Barbados (e.g. Moore 1989), or by diffusion within the sediments and circulation in fractures, like in the Nankai prism (e.g. Le Pichon et al 1992). The concretions studied at Cape Turnagain are clustered into small zones and zones with a low concentration of concretions. This pattern of clusters can be attributed to the focus of fluids in zones of greater permeability, even though there is no field evidence observed of structures that might focus fluid such as faults, folds or changes in lithology. We assume that the tubular concretions observed at Cape Turnagain are fossil fluid expulsion chimneys. The great amount of detrital particles (>50%) within the concretions indicates that these chimneys were formed beneath the sediment-water interface by cementation of the Miocene detrital sediment. Petrographic data suggest a possible model for the formation of the concretions. It is shown schematically in Figure 11 and consists of three steps. The first step is the diffusion of a carbonate-rich fluid from the central canal into the bioturbated, unconsolidated surrounding sediments allowing the growth and cementation of the concretion (one or two concentric parts with decreasing carbonate content from the canal to the outer part of the chimney) and the formation of dendrites. The second step is achieved by the partial infilling of the canal with fine-grained sediments, with few microfossils and rare bioturbation. This infill will be dated with future work to bring information on the origin of the sediments found within the canal. The final step is the crystallization of euhedral carbonate (Ca- and Ca-Mg carbonates) and occasionally silica crystals within the geode. The preservation of at least a part of the initial canal may be attributed to the duration (not quantified) of fluid expulsion within the sediment, even after the deposition of younger sediments on top of it. Consistently, Orpin (1997) suggests that, for the Otago chimneys, the cementation was probably maintained long-term by consistent geochemical
1.39 0.44 87.5 12.5
3a 25 12 11 92 0.47 0.43
12.5%
45 27 9 18 54%
3.3 0.7
28.5 1.06
-
60 29 11 38 0.48 0.18
100 0% 11.03 1.23
gradients within the near-surface bottom sediments. Concerning the chemistry of the fluids responsible for the cementation, few data are available today. However, the absence of associated typical vent faunas at the study site might suggest that fluids were not enriched with methane. The differences observed in the sharpness of the canal border suggest two episodes for the formation of the concretions: (i)
The dendritic features debated before and the bioturbation found within the cemented chimneys show that the sediment was not indurated when the fluid flow occurred. Thus, the formation of some of the concretions is probably
Fig. 8. Attitude to bedding of the 57 in situ concretions as deduced from the projection of their dip and dip direction in a Schmidt diagram, lower hemisphere. The bedding has been back tilted to its initial horizontal position.
104
B.LEDESERT£rAL.
Fig. 9. Features observed on the border of a canal: photomicrograph of a bioturbation crosscut by the canal, under plane-polarized light. The bioturbation (dark, round zone outlined in the centre of the photograph) is frayed and shows dendritic features made of clay (also outlined). These dendrites are mixed with calcite which developed in the canal to create the geode. The dashed line indicates the border of the canal.
(ii)
early in the history of the upper-Miocene sediments. However their age is difficult to establish with precision since no associated chemosynthetic faunas were observed on the outcrops and no isotopic dating was performed. Sharp borders, alternatively, tend to indicate that the sediment was consolidated when the flow occurred. These concretions probably formed after those described in (i).
Along the Hikurangi subduction margin, Lewis & Marshall (1996) note that there may be some additional input of fluids from overpressured sediments at depth. In the case of the fossil fluid-escape structures described in this paper, we have to consider the complex structural history of the Hikurangi margin to link fluid flows with the geodynamical setting. The Hikurangi subduction zone has been active since the earliest Miocene (van der Lingen & Pettinga 1980; Rait et al 1991; Chanier & Ferriere 1991) and has undertaken a complex structural development with a succession of tectonic shortening events and quiescence or extensional deformation (Cape et al 1990;Beanland^a/. 1998; Chanier et al. 1999). The end of Miocene is characterized by a major uplift event associated with some compressional deformation (Buret et al. 1991 \ Beanland et
al. 1998; Nicol et al submitted). The abundance of these chimneys in Miocene sediments and their scarcity in the overlying Pliocene series, even of similar lithology and thus porosity and/or permeability, suggest that they could be linked to the lowermost Pliocene compressional deformation episode. This compressive event could have generated overpressure favouring the circulation and expulsion of fluids to the surface through the consolidated sediments. Similarly, the late Quaternary is also characterized by a compressive episode and by the activity of numerous mud volcanoes and fluid seeps (Lamb & Vella 1987). Lewis & Marshall (1996) and Oipin (1997) suggested that fossil chimneys similar to those encountered offshore of New Zealand, and attributed to fluid expulsion, might occur onland as well. Orpin (1997) suggested the morphologically chimneylike carbonate structures described from the North Island East Coast, and interpreted as burrows by Mazengarb & Francis (1985) and Hayward (1989), might equally be fluid-escape structures. The Cape Turnagain area is a good example of such an occurrence. Its contribution to a better knowledge of fluid escape structures is important, since the observation of these structures is easier onshore than offshore, allowing measurements (e.g. attitude and distribu-
TUBULAR STRUCTURES AND FLUID EXPULSION
105
Fig. 10. Scanning electron microscope photomicrograph showing newly formed crystals within a geode. Note the presence of Ca-Mg carbonate (Ca-Mg), spheroids of silica (Si) and pure Ca-carbonate (Ca). The black cavity is on the right of the photograph.
tion) to be taken with accuracy. In addition, the sampling can be precisely controlled spatially and petrographical data can easily be related to the measurements obtained in situ.
Conclusion The concretions observed along the 4.5 km long Cape Turnagain section are packed into small zones (<100 m2), and larger areas with a lowconcentration of concretions. The concretions are made of three parts: (1) the external part resulting from the cementation of the detrital Miocene sedi-
ment; (2) the central canal, sometimes showing spiral rifling, either free of sediments or partially infilled with detrital particles; and (3) the geode. The concretions are characterized by a positive linear relationship between their external diameter and that of their canal (or canals) or the number of canals they contain, indicating that their size is a function of the fluid which flowed through the plumbing network. Their morphology (branched shape, decimetric size, partially unfilled central canal), together with their concentration into small zones make them very similar to structures attributed to dewatering chimneys offshore New Zealand East Coast. This similarity, along with petrographical observations
106
B.LEDESERTCTAL.
Fig. 11. Model of concretion growth, canal infilling and geode formation. Step (1) is the diffusion from the canal of a carbonate-rich fluid within the mudstone. The arrows indicate the direction of the fluid flow and slope of decreasing geochemical gradient; step (2) corresponds to the partial infilling of the canal with fine grained sediments; and step (3) is the crystallisation of minerals in the geode.
(spiral rifling of the canal, dendritic mixing features) suggest they were formed by the rapid expulsion of fluid flows through the upper-Miocene to lowerPliocene silty sediment. The exact dating of the fluid expulsion is not yet available. However, possibly two episodes of fluid expulsion occurred. The first one could have been early in the history of this sediment since it was unconsolidated during the fluid flow. The second one would have occurred later, after consolidation of the sediment, for example during a compressive event. Previous studies of similar concretions offshore were sometimes limited because of the difficulty of field sampling and observation. This study of an onshore occurrence of tubular concretions brings new field data of possible relict fluid expulsion structures. Financial support for this study (field trip and laboratory work) has been provided by the French Foreign Office (French Embassy in Wellington) and by CNRS. The authors gratefully acknowledge Drs D. Darby and A. Orpin for their careful review and improvement of the manuscript and Pr R. Hillis for his help.
References AGAR, S.M. 1990. The interaction of fluid processes and progressive deformation during shallow level accretion: examples from the Shimanto belt of SW Japan. Journal of Geophysical Research, 95-B6, 9133-9147.
BEANLAND, S., MELHUISH, A., NICOL, A. & RAVENS, J. 1998. Structure and deformational history of the inner forearc region, Hikurangi subduction margin, New Zealand. New Zealand Journal of Geology and Geophysics, 41, 325-342. BURET C. 1996. Les bassins sedimentaires d'un domaine d'avant-arc: la marge active de Nouvelle Zelande, PhD thesis of the University of Lille (France), 353 p. BURET, C., CHANIER, F, FERRIERE, J. & PROUST, J.-N. 1997. Individualisation d'un bassin d'avant-arc au cours du fonctionnement d'une marge active: la marge Hikurangi, Nouvelle-Zelande. Comptes Rendus de VAcademic des Sciences, Paris, 325, 615-621. CAPE, C.D., LAMB, S.H., VELLA, P., WELLS, P.E. & WOODWARD, DJ. 1990. Geological structure of Wairarapa Valley, New Zealand, from seismic reflection profiling. Journal of the Royal Society of New Zealand, 20, 85-105. CHANIER, F. & FERRIERE, J. 1991. From a passive to an active margin: tectonic and sedimentary processes linked to the birth of an accretionary prism (Hikurangi margin New Zealand). Bulletin de la Societe Geologique de France, 162,649-660. CHANIER, F, FERRIERE, J. & ANGELIER, J. 1999. Extensional deformation across an active margin, relations with subsidence, uplift and rotations: The Hikurangi subduction, New Zealand. Tectonics, 18, 862-876. EL ALBANI, A., CLOUTIER, R. & CANDDLIER, A.-M. 2002. Early Diagenesis of the Upper Devonian Escuminac Formation from the Gaspe Peninsula (Quebec): Sedimentological and Geochemical evidences. Sedimentary Geology, 146, 3-^, 209-223. EL ALBANI, A., VACHARD, D., KUHNT, W, & THUROW J. 200 Ib. The Role of diagenetic carbonate concretions
TUBULAR STRUCTURES AND FLUID EXPULSION in the preservation of original sedimentary record. Sedimentology, 48, 875-886. FIELD, B.D., URUSKI, C.I. and others. 1997. CretaceousCenozoic geology and petroleum systems of the East Coast Region, New Zealand. Institute of Geological and Nuclear Sciences monograph 19,301 p. GUILHAUMOU, N. & LARROQUE, C. 1995. Les circulations de fluides dans les prismes d'accretion: fluides fossiles et fluides actuels. Comptes Rendus de I'Academic des Sciences, Paris, serie II a, 321, 939-957. HAN, M.W. & SUESS, E. 1989. Subduction-induced pore fluid venting and the formation of authigenic carbonates along the Cascadia continental margin: implication for the global Ca-cycle. Palaeogeography, Palaeoclimatology andPalaeoecology, 71, 97-118. HASHIMOTO, J., OHTA, S., TANAKA, T, HOTTA, H., MATSUZAWA, S., SAKAI, H. 1989. Deep-sea communities dominated by the giant clam Calyptogena soyoae, along the slope foot of Hatsushima Island, Sagami Bay, Central japan. Palaeogeography, Paleoclimatology, Palaeoecology, 71,179-192. HAYWARD, B. 1989. Giant fossil carrots. Geological Society of New Zealand Newsletter, 86, 53. KVENVOLDEN, K.A. & PETTINGA, J.R. 1989. Hydrocarbon gas seeps on the convergent Hikurangi margin, North Island, New Zealand. Marine and Petroleum Geology, 6,2-8. LALLEMANT, S., HENRY, P., LE PICHON, X. & FOUCHER, J.P. 1990. Detailed structure and possible fluid paths at the toe of the Barbados accretionary wedge. Geology, 18, 854-857. LAMB, S.H. & VELLA, P. 1987. The last million years of deformation in part of the New Zealand plate boundary zone. Journal of Structural Geology, 9, 877-891. LE PICHON, X., KOBAYASHI, K & Kaiko-Nankai scientific crew. 1992. Fluid venting activity within the eastern Nankai trough accretionary wedge - a summary of the 1989 Kaiko-Nankai results. Earth and Planetary Science Letters, 109, 303-318. LEDESERT, B., DUBOIS, J., VELDE, B., MEUNIER, A., GENTER, A. & BADRI, A. 1993. Geometrical and fractal analysis of a three-dimensional hydrothermal vein network in a fractured granite. Journal of Volcanology and Geothermal Research, 56,267-280. LEWIS, B.T. & COCHRANE, G.C. 1990. Relationships between the location of chemosynthetic benthic communities and geologic structures on the Cascadia subduction zone. Journal of Geophysical Research, 95-B6,8783-8793. LEWIS, K.B. & MARSHALL, B.A. 1996. Seep faunas and other indicators of methane-rich dewatering on New Zealand convergent margins. New Zealand Journal of Geology and Geophysics, 39,181-200. LEWIS, K.B. AND PETTINGA, J.R. 1993. The emerging, imbricate frontal wedge of the Hikurangi Margin. In: BALLANCE, P.P. (ed.) Sedimentary Basins of the World, 2, South Pacific Sedimentary Basins, 225-250.
107
MASCLE, A., ENDIGNOUX, L. & CHENNOUF, T. 1990. Frontal accretion and piggyback basin development at the southern edge of the Barbados ridge accretionary complex. Proceedings ODP Scientific Results, 110, 17-28. MAZENGARB, C. & FRANCIS, D. 1985. The occurrence of Paramoudra concretions in the Gisborne district. Geological Society of New Zealand Newsletter, 68, 52-53. MOORE, PR. 1981. Geology of the Late Tertiary section at Cape Turnagain. Journal of the Royal Society of New Zealand, 11 (3), 223-230. MOORE, J.C. 1989. Tectonics and hydrogeology of accretionary prisms: role of the decollement zone. Journal of Structural Geology, 11,95-106. MOORE, J.C. & VROJLIK P. 1992. Fluids in accretionary prisms. Reviews in Geophysics, 30,113-135. NEEF, G. 1984. Late Cenozoic and Early Quaternary stratigraphy of the Eketahuna District (N153), New Zealand Geological Survey Bulletin, 96,101 p. NICOL, A., VAN DISSEN, R., VELLA, P., ALLOWAY, B. & MELHUISH, A. 2002. Growth of contractional structures during the last 10 Ma, Hikurangi forearc, New Zealand. New Zealand Journal of Geology and Geophysics, 45,365-385. ORPIN A.R. 1992. Authigenic carbonate chimneys as possible conduits for fluid expulsion on the outer Otago shelf. Unpublished M. Sc. Thesis, lodged in the Library, University of Otago, Dunedin, New Zealand, 153 p. ORPIN, A.R. 1997. Dolomite chimneys as a possible evidence of coastal fluid expulsion, uppermost Otago continental slope, southern New Zealand. Marine Geology, 138,51-61. RAIT, G., CHANIER, F. & WATERS, D.W. 1991. Landwardand seaward-directed thrusting accompanying the onset of subduction beneath New Zealand. Geology, 19,230-233. RIDD M.F. 1970. Mud Volcanoes in New Zealand. American Association of Petroleum Geology Bulletin, 54,601-616. ROGERS, K.M., COLLEN, V.B., JOHNSTON, J.H. & ELGAR, N.E. 1999. A geochemical appraisal of oil seeps from the East Coast basin, New Zealand. Organic Geochemistry, 30,593-605. VAN DAMME, H. & LEMAIRE, E. 1990. Viscous fingering and viscoelastic fracture in clays. In: HERMANN H.J. & Roux S. (eds) Statistical models for the fracture of disordered media. Series Random Materials and Processes. Elsevier Science Publishers B.V., North Holland (Amsterdam, Oxford, New York, Tokyo), 77-86. VAN DER LINGEN, G.J. & PETTINGA, J.R. 1980. The Makara Basin: a Miocene slope-basin along the New Zealand sector of the Australian-Pacific obliquely convergent plate boundary. In: BALLANCE, P.F, & READING, H.G. (eds) Sedimentation in oblique strike-slip mobile zones. Special publication of the International Association of Sedimentologists, 191-215.
This page intentionally left blank
Fluidization pipes and spring pits in a Gondwanan barrier-island environment: groundwater phenomenon, palaeo-seismicity or a combination of both? E. DRAGANITS1, B. GRASEMANN1 & H. P. SCHMID2 2
Institutfur Geologic, Universitdt Wien, Althanstrasse 14, A-1090 Wien, Austria (e-mail: Erich. Draganits @ univie. ac. at) 2 OMV, Exploration & Production, Gerasdorfer Strasse 151, A-1210 Wien, Austria Abstract: Cylindrical structures, cross-cutting stratification at right angles, occur in the Muth Formation, representing Lower Devonian barrier island arenites of the North Indian Gondwana coast. These structures are up to 1.5 m in height and 0.8 m in diameter, with an internal structure comprising concentric, cylindrical laminae. The pipes, which probably represent water conduits for laminar upward flow of ground water, initiate from relatively thin horizons, with upper terminations formed by spring pits. Thus, the structures in the Muth Formation represent a rarely observed combined occurrence of spring pits and their conduits below. Their formation is explained by rising ground water seepage in a coastal depositional environment that produced a relatively high hydrostatic head, resulting in the formation of springs. The rise in relative sea level might be related to tectonic subsidence caused by tectonic activity linked to the formation of conjugate deformation bands in the Muth Formation. This means, if tectonic activity was involved, it did not form the cylindrical structures by seismic liquefaction directly, but might be responsible indirectly through ground water seepage rise resulting from tectonic subsidence. Due to the little relief in this environment, the sea level rise affected a relatively large area and fluidization structures can be found widespread in distant sections.
Liquidization structures represent important postdepositional modifications of sediments, affecting many lithological parameters at various scales and bearing considerable importance to groundwater flow, oil migration paths and reservoir characteristics. At small scales, liquidization can modify grain size distributions, orientation of clasts, porosity and permeability (Lowe 1975), whilst at larger scales it may alter primary sedimentary structures, bed contacts and even the overall shapes of sedimentary bodies (Lonergan etal 2000). According to Allen (1982, pp. 293-295), 'liquidization' describes processes that modify loose, grain-supported sediments, reducing their shear strength so that they behave like a viscous liquid. Liquidization processes include 'liquefaction', which is the transformation of a loosely packed granular material from a solid state into a liquefied state as a consequence of increased hydrostatic pressure without any exchange of fluid and neglectable volume change (Youd 1973) and 'fluidization', which is liquidization by upward directed flow of externally derived fluid in a granular sediment body, where the fluid drag on the detrital grains exceeds their weight. Liquidization structures are common in clastic sediments and may originate from seismic and non-seismic processes (e.g. Lowe 1975; Allen 1982; Obermeier 19960). Cylindrical structures in granular sediments, cross-cutting stratification nearly at right angles,
have been mentioned from several depositional environments. Several explanations have been summarized by Dionne & Laverdiere (1972), Deynoux et al (1990), Hunter et al (1992), Dionne & Perez Alberti (2000) and Massari et al (2001). These include: (i) upward flow of liquidized sediment resulting from liquefaction (triggered by wave action, seismic events and impact shaking (Alvarez et al. 1998), mass movements or rapid sedimentation); (ii) fluid/gas escape resulting from fluidization (e.g. water springs, volcanic exhalation, hydrocarbon gas seeps, boiling water above basaltic sills (Rawlings 1998), gas blow-out pipes or conduits below mud-volcanoes (Kopf 2002, and references cited therein)); (iii) collapse following the removal of underlying material (e.g. evaporite and carbonate dissolution or melting of buried ice); (iv) filling of pipe-shaped cavities (e.g. evorsion holes, Fenninger 2000); (v) vertical, pipe-shaped weathering channels (van Husen 1999); (vi) concretions around an organic core; and (vii) animal burrows. In this context, it is worth mentioning that creationists generally interpret cylindrical liquidization structures as direct evidence for the global Flood of the bible (Cox 1977; Roth 1992; Walker 2000). Spring pits have been described by Quirke (1930), but hardly anything has been written about these structures since then (Draganits & Janda 2003); therefore relatively little is known about them compared to cylindrical liquidization structures.
E. DRAGAN1TS ETAL quartz arenites representing sediments deposited from shoreface to coastal dunes (Draganits 2000). In this environment, liquefaction by wave action and tectonic activity (Obermeier 1996&) or fluidization by ground water table variations controlled by relative sea-level fluctuations (Massari etal. 2001) seem the most probable mechanisms behind liquidization. The liquidization features in the Muth Formation are associated with conjugate deformation bands (Aydin 1978) and conjugate brittle faults in the uppermost part of the broadly upper Ordovician to lower Silurian Pin Formation below. Both deformation structures show identical orientations and document E-W directed shortening in a transcurrent tectonic setting; they are interpreted to belong to the same process and represent a previously unknown pre-Himalayan deformation stage in the NW Himalayas (Draganits 2000). Cashman & Cashman (2000) have shown that deformation bands may develop at near-surface conditions and that the formation of deformation bands can be related to seismic slip events on nearby faults. Due to the existence of deformation bands, palaeo-seismicity is a possible trigger mechanism for liquefaction in the Muth Formation.
Geological setting
Fig. 1. Geological map of the Pin Valley modified from Fuchs (1982) and Wiesmayr & Grasemann (2002), with location of sections A-A', E-E' and G-G'. Filled squares indicate villages and open squares signify seasonal shelters.
According to Quirke (1930) they form near the shoreline, above and below water, by ascending water which rises with sufficient force to sweep out finer sand grains. In this paper, we describe cylindrical structures developed in quartz arenites of the Lower Devonian Muth Formation (NW Himalayas). These pipes, together with other liquidization structures, like spring pits and slumped beds, have been found in three lithostratigraphical sections with a total restored distance of 31 km normal to the overall facies trend (Figs. 1,2). The cylindrical structures in the Muth Formation add important new information to those described in the literature, as both the source and termination of the pipes are exposed. The depositional environment of the Muth Formation has been interpreted as a barrier-island system, with the
The Muth Formation belongs to the Tethyan Zone of the Higher Himalaya tectonic unit, which records an almost continuous stratigraphic sequence from the Neoproterozoic up to the Eocene, deposited at the northern Indian continental margin. In the Pin Valley (Fig. 1), the Tethyan sediments were deformed during the Himalayan orogeny into large-scale SW-vergent folds, with maximum wavelengths of approximately 5 km (Fuchs 1982; Wiesmayr & Grasemann 2002). Crustal thickening related to folding resulted in metamorphic conditions ranging from diagenetic zone in the Mikkim and Muth sections to anchizone in the Baba La ('La' is the Tibetan word for pass) section (Wiesmayr & Grasemann 2002). The Muth Formation is underlain by the yellowbrown weathering Pin Formation of upper Ordovician to lower Silurian age (Bhargava & Bassi 1998; Talent pers. comm. 2001). At the type section, the Pin Formation reaches a thickness of some 290 m and consists of variable lithologies. The contact to the white quartz arenites of the Muth Formation above represents a disconformity in all three sections (Fig. 2). The Muth Formation is devoid of age-diagnostic fossils, but well constrained middle Devonian conodont faunas in the lower parts of the overlying Lipak Formation and the arthropod ichnofauna of the Muth Formation broadly indicate an early Devonian age (Draganits et al. 2001; Draganits et al
FLUIDIZATION PIPES AND SPRING PITS
2002). The formation comprises monotonous white, fine- to medium-grained, extremely pure quartz arenites with high textural, as well as compositional, maturity; the only exceptions are thin horizons of sandy and silty dolomites in higher levels of the formation (Fig. 2). The Muth Formation represents a relatively competent layer within the pile of Tethyan sediments, thus showing no second order folding as less competent formations do and as a result, sedimentary structures are well preserved (Wiesmayr & Grasemann 2002). In the Pin Valley, based on different, diagnostic sedimentary structures, the Muth Formation has been divided into four facies associations (FA 1 to FA 4, Fig. 2). The basal facies association (FA 1) is dominated by relatively thin-bedded, horizontally laminated arenites, with an increase in tabular crossbedding at higher levels. This association is followed by thick, large-scale tabular and tangential crossbedded beds with steep foreset angles (FA 2). The third association (FA 3) forms a conspicuous, sharpbased horizon comprising orange to brick red, highly oxidized very fine-grained dolomite, clay-, silt- and sandstone. The uppermost part (FA 4) of the Muth Formation consists mainly of horizontally bedded quartz arenite, which shows a gradual increase in impurity towards higher levels. In general, the depositional environment of the Muth Formation is interpreted as a barrier island system; arguments leading to this interpretation are discussed in detail by Draganits (2000). FA 1 is interpreted as transgressive sequence with beach sediments at the base and upper shoreface to upper foreshore deposits in the upper part. FA 2 is considered to represent shallowest foreshore, backshore, and coastal dune sediments. Sedimentary structures such as flat-topped ripples, adhesion ripples, tearshaped ridges (McKee 1957) and desiccated biofilms indicate at least temporally emergent conditions. The fine-grained, fine-laminated sediments of FA 3 are interpreted as lagoonal deposits. The final facies association probably comprises foreshore to lower shoreface and displays a transgressive trend with a gradational contact to the inner shelf deposits of the overlying Lipak Formation (Draganits 2000). Cylindrical structures are found in the uppermost beds of FA 2 and within FA 3; spring pits occur in the upper part of FA 2 and in the uppermost part of FA 4 (Fig. 2, 3).
Liquidization structures The lithostratigraphical sections of the Muth Formation in Figure 2 represent simplified bed-by-bed sections. Peculiar cylindrical structures and spring pits represent the most eye-catching liquidization structures in the Muth Formation of the Pin Valley.
111
However, a slumped bed occurs in the Mikkim section, some 220 m above the base of the formation (i.e. in the lower part of FA 2) and dish - and pillar structures (Lowe 1975) have been found rarely in some levels of the section. Both structures are of relative small-scale and do not seem to exceed water escape structures, which might be expected in the coastal depositional environment of the Muth Formation. They probably formed by de-watering induced by common processes in this environment such as storm waves or sedimentary overloading, which are different from the processes that formed the much larger cylindrical structures and spring pits.
Cylindrical structures Among the liquidization structures of the Muth Formation, the cylindrical structures are by far the most spectacular ones (Fig. 4). In both size and internal structure, they are very similar to the cylindrical structures described by Hawley & Hart (1934) and Dionne (1973), although their closest analogues are the cylinders described by Deynoux et al (1990). Two small cylindrical structures appear in an arenaceous bed in FA 3 (Fig. 2) in the Baba La section, c. 2 km to the north of the Baba La (Figs. 1, 2), a pass connecting the Pin and Baba Valleys (section G-G'; N31°40'30"; E78°00' 14"; 4520 m). With the exception of the two pipes in the Baba La section, all cylindrical structures have been found in the type section (Fig. 2), 1.3 km to the south of village Muth (section E-E'; N31°56'44"; E78°02'05"; 3860 m). There, relatively small and rare pipes are found 12m below the top of FA 2. In the uppermost three beds of FA 2 (aeolian dunes), pipes are large and abundant and some small pipes also occur in arenaceous beds within the finegrained dolomites in the upper part of FA 3, interpreted as lagoonal sediments (Figs 2, 3). Although both sections have a restored distance of 18 km (Wiesmayr & Grasemann 2002), these structures occur at similar levels in FA 3. The cylindrical structures in the uppermost beds of FA 2 in the type section at Muth have been investigated in detail; there these structures are abundant and best exposed (Fig. 4). Bed numbers refer to beds of the bed-by bed section in Fig. 3. Most pipes are found in the uppermost three beds of FA 2 (beds Me438-Me440 in Fig. 3); rare pipes are found in bed Me436 too. The contact of the uppermost bed of FA 2 (Me440) to the first bed of FA 3 above is a very sharp lithological break of quartz arenite to fine-grained dolomite (Fig. 3). All bar one of these beds comprise white to slightly greenish, pure quartz arenite with a high textural and compositional maturity, cemented by quartz. The
112
E. DRAGANITS ETAL.
FLUIDIZATION PIPES AND SPRING PITS
exception is bed Me437, which consists of dark green quartz arenite with a mixed chlorite/quartz matrix, showing a texture of pore-filling infiltrated clays (Price 1996). Grain size in all the beds is consistently in the medium sand range. Beds Me438 and Me440 show large-scale, high-angle, concave-up tangential cross bedding. Bed Me437 shows no internal structure and the rest of the concerned beds are horizontally laminated (Figs. 3,4e, 4f). Outside the pipes, the beds have preserved their primary sedimentary structures and bed boundaries very well; only beds Me435-Me437 show slightly wavy bedding surfaces with varying bed thickness. The pipes crosscut bedding surfaces at right angles and generally comprise straight, nearly perfect cylindrical shapes (Fig. 4e, f). A few funnelshaped structures have been found, which always taper downwards; generally, they are much smaller compared to cylindrical-shaped pipes and have been found only in the lowermost part of bed Me438. Pipes range in height from 5 to 155 cm with diameters from 2 to 80 cm; longer cylinders tend to have larger diameters than smaller ones, but no correlation is evident. The contact of the pipes to the completely undisturbed rest of the beds is a sharp boundary (Fig. 4e, f) and thus contrasts strongly from fluidization pipes surrounded by fluidization halos described by Mount (1993). Where the lamination of the host rock is truncated abruptly, the cylinder shape is regular and smooth. No clay enrichment at the contact between pipe and the rest of the bed as described by Mount (1993) has been observed. Analogous to observations by Best (1989), the sediment in the pipe-interior is slightly depleted in the finest sand fraction and especially in clay. Apart from this depletion of fines, grain size, composition and cement are nearly identical within cylindrical structures and host sediment.
Fig. 2 left, (a) Lithostratigraphical sections of the Muth Formation from the Pin Valley showing the correlation of the facies associations and the levels with cylindrical structures. The restored distances between the Mikkim and Muth sections and the Muth and Baba La sections are 13 km and 18 km, respectively. FA = facies association, (b) Overview of the Palaeozoic stratigraphy in the upper Pin Valley; view from the ravine NW of Muth village towards the SE; up-section to the left. Arrow indicates the location of the cylindrical structures at the contact between FA 2 and FA 3. S = Shian Formation, P = Pin Formation, M = Muth Formation, L = Lipak Formation. Fig. 3 right. Detail of the limostratigraphical section EE' at the level of the cylindrical structures. For lithological index and abbreviations see Fig. 2. Bed numbers represent numbers of the bed-by-bed section. FA 2 - mainly aeolian arenites; FA 3 - lagoonal dolomites and sandy dolomites; FA 4 - shallow marine arenites.
113
114
E. DRAGANITS ET AL.
Fig. 4. (a) Lower bedding surface view of several circular and spindle shaped liquidization structures at the base of bed Me438 (Lens cap diameter 53 mm). (b) Lower bedding surface view of an empty, spherical shaped liquidization structure at the base of bed Me438, connected to the beginning of concentric laminations of a pipe above. (c) Lower bedding surface view of axial sections of cylindrical structures at the base of bed Me438, partly obscured by lichens. (d) Lower bedding surface view of two, small cylindrical structures, some 12 cm apart within a larger, elliptical shaped liquidization pipe. (e) Sectional view of a large pipe crosscutting tangential cross-beds; up-section to the left. Note slightly darker bed Me438 at the base of the cylindrical structure. See geological compass for scale. (f) Slightly oblique, near-axial section of a cylindrical structure crosscutting bedding at nearly right angles; up-section to the left. Note the sharp boundary to the host sediment and the thin, well-developed concentric laminae at the rim and the thicker, concave upward laminae in the central part.
The internal structures of the pipes vary between two end members, which are similar to the two types of cylindrical structures described by Deynoux et al. (1990). Most of the structures are somewhere in between these two extremes and therefore no separate types are classified in this paper. No differences in size have been found between these two end members; both types occur next to each other. One
end member, identical to type 2 of Deynoux et al. (1990, fig. 7), comprises just a single, thin cylindrical-shaped lamina, representing the boundary of the structure, with more or less undisturbed primary sedimentary structures preserved inside; the primary sedimentary lamination commonly shows downward bending and/or normal faulting to variable extend.
FLUIDIZATION PIPES AND SPRING PITS
In the other end member, which is similar to type 1 of Deynoux et al (1990, fig. 5b), in axial sections the internal structures show several, regular, vertical, concentric (in rare cases slightly eccentric), some few mm thick laminae across the complete diameter of the cylinders (Fig. 4c). In contrast to type 1 of Deynoux et al. (1990) they rarely show a cone-incone arrangement in longitudinal sections, but this might just reflect which part of the internal structures of pipes actually are exposed, because conical laminae tend to become cylindrical shaped in higher parts. The concentric laminae in the pipes are formed by subtle grain size variations; there is no trend of grain size variations visible across the pipe diameter. Concentric laminae may occupy the complete diameter of the cylinder, or just parts of it (Fig. 4c, f). In the latter case, without exception, the lamination is always found at the rim but the core appears to be relatively structureless or the lamination is thicker, similar to type 1 of Deynoux et al (1990, fig. 12). In a typical cylindrical structure of this type, the lamina-free core comprises c. V3 of the whole diameter. Usually pipes rarely show any interaction; Fig. 4d shows an axial section of two, small cylindrical structures, some 12 cm apart within a larger, elliptical shaped pipe. With few exceptions, almost all pipes seem to originate exactly from the interface between beds Me437 and Me438 (Fig. 4e). Bed Me437 has the highest clay contents compared with other beds, some influence by liquidization is indicated by its highly variable thickness and irregular, wavy upper and lower bedding surfaces. The lower bedding surface of bed Me438 shows abundant, randomly distributed cylindrical structures in axial sections (Fig. 4c); from field evidence, these sections represent the base of most of the cylindrical structures. With only a few exceptions, the place of actual liquidization below a pipe is not visible at the base of Me438. In Figure 4b, the diameter of the circular casting cake-shaped structure is 32 cm at the base, tapering upwards and leading to a cylindrical structure. The surface of the pipe is irregular, but sharply defined and in its uppermost part, concentric laminations of the very base of the structure above are visible. The filling of the structure, which has been removed before taking the photograph, is massive. Figure 4a shows complex interaction of several liquidization centres at the base of bed Me438.
115
tually circular, randomly distributed deep pits, some more than 80 cm in diameter and several tens of centimetres deep (Fig. 5a). In a closer view, the rims of the pits are smooth and slightly raised and show concentric laminae; the uppermost filling has been partly weathered out (Fig. 5b). No pipe has been found continuing into the fine-grained, red, welloxidized dolomite bed (Me441); therefore, the upper surface of Me440 is regarded as the top of the liquidization pipes. At the upper surface of Me440, there is no sign of any extruded sand material; on the contrary, the filling has sunk, as already indicated by the previously described internal structures of downward bending of sedimentary lamination and normal faulting. In one case, a c. 3 cm long, finegrained sediment clast was found in the uppermost part of a pipe, possible indicating some backfilling from above. Thin, planar sand fissures occur (compare with Deynoux et al. 1990, fig. 5C), oriented perpendicular to the bedding, but they do not show radial orientation to the pits; in some cases they seem to cross-cut the rim of the spring pits (Fig. 5b). Other possible spring pits are found in the uppermost part of the Muth Formation at the type locality. There a prominent upper bedding surface that can be traced for several km is completely covered with abundant, randomly distributed circular pits (Fig. 5c). The bowl-shaped depressions are up to c. 80 cm in diameter and resemble spring pits (Quirke 1930), but no cylindrical structures have been found in the beds below them. Enigmatic donut-shaped structures on upper bedding surfaces of horizontal well-laminated beds have been found SE of village Mikkim. These structures have a striking circular, donut-shaped raised ring with a bowl-shaped, central depression; they might be confused with spring pits, but represent microbial gas domes (Noffke pers. comm. 2002).
Deformation bands
Another group of post-depositional structures in the Muth Formation are deformation bands (Aydin 1978). They represent conjugate shear fractures that develop cluster zones rather than discrete fault zones. They show evidence of effective porosity reduction and compaction by cataclastic processes, as well as reorganisation of quartz grains. The final appearance of the microstructure is controlled by porosity, effective pressure, fluid content and amount of displacement (Antonellini et al 1993). Spring pits They have been observed SE of Mikkim (section Several liquidization pipes have been described in A-A') throughout the entire Muth Formation. the literature, but their upper terminations have Deformation bands are concentrated at c. 120 m hardly ever been exposed or preserved. The upper and 180 m above the base of the Muth Formation, bedding surface of bed Me440 shows numerous, vir- but they are most common between 200-240 m, in
116
E. DRAGANITS ET AL.
Fig. 5. (a) Overview of the spring pits on the upper bedding surface of bed Me440; encircled hammer for scale. (b) Detail of Fig. 5b showing a spring pit with eroded central part and a slightly raised rim. Note thin sand fissures in the lower left. (c) Prominent upper bedding surface in the uppermost part of the Muth Formation at the type locality is completely covered with abundant, randomly distributed circular spring pits. Encircled H. P. Schmid for scale.
the lower part of FA 2, in coastal aeolian deposits (Fig. 2). Deformation bands in the Muth Formation are mm thin, planar, slightly undulating features that can be traced for some cm to several m. They hardly ever occur alone, but usually occur in 'zones of deformation bands' (Aydin & Johnson 1983) that constitute many closely spaced deformation bands. Although deformation bands comprise the same mineralogy as the host rock they are raised slightly above the average rock surface due to higher weathering resistance; in contrast brittle faults of fully cemented quartzite show reduced weathering resistance. The boundary to the host rock is well defined, but not as distinct as in brittle faults. The orientation of the deformation bands does not fit reasonably to any of the orientations of brittle faults related to Himalayan deformation in the Pin Valley (Wiesmayr & Grasemann 2002). Consequently, the deformation bands are not related to Himalayan deformation and therefore early Himalayan large-scale folding has been restored by back-rotation of the bedding surfaces to horizontal orientation together with the deformation bands. In the restored orientation, deformation bands show consistent orientations throughout the complete Muth Formation, forming conjugate sets of dominant WNW-ESE and less pronounced ENE-WSW trending faults, nearly perpendicular to stratification (Fig. 6).
Discussion Occurrence of the cylindrical structures Pipe-shaped structures with concentric internal lamination, cross-cutting stratification at right angles have been found in Lower Devonian quartz arenites of the Muth Formation. Several similar structures have been described in the literature (e.g. Deynoux etal 1990; Hunter et al. 1992;Massariera/. 2001); two features seem to be common with the occurrence of cylindrical structures. The first characteristic is the frequent occurrence in mature, arenaceous sediments, e.g. shallow marine reworked aeolian sand (Hawley & Hart 1934) or aeolian sand (Deynoux et al. 1990). There, high porosity and permeability seem to support the formation of pipes. The second frequent feature is the common link of cylindrical structures and/or spring pits with the transition of marine/limnic/fluviatile and terrestrial environments, where high water saturation occurs and water table variations are frequent (Quirke 1930; Hawley & Hart 1934; Gabelman 1955; Flint 1983; Deynoux et al. 1990; Guhman & Pederson 1992; Massari et al 2001; Netoff & Shroba 2001; Draganits & Janda 2003). Both characteristics are also found in the Muth Formation. From this point of view, the restriction of the cylindrical structures in the Muth Formation to certain levels might thus
FLUIDIZATION PIPES AND SPRING PITS
117
pipes initiation is also mentioned by Gabelman (1955). Compared with other beds of the Muth Formation, bed Me437 shows very high contents of chlorite in the inter-granular space, which might have originated from infiltrated clays (Price 1996). Lowe (1975) mentioned that many of his 'Type B pillars' originate at the bases of sand units that overlie mud or clay layers. Whether the increased clay content of bed Me437 is a result of high groundwater activity, or the clay content affected the movements of the fluids is a kind of chicken and egg problem. Relationship between cylindrical structures and spring pits
Fig. 6. (a) Bedding surface view of conjugate deformation bands SE of Mikkim with the acute bisectrix in E-W orientation; North is to the top, lens cap diameter 53 mm. (b) Equal area projection, lower hemisphere of 92 measurements of deformation bands in the Muth Formation SE of Mikkim, restored concerning Himalayan folding. Measurements indicate broadly WNW-ESE to ENE-WSW striking, near vertical orientation of the deformation bands.
reflect the control of the depositional environment; the effects of repeated seismic events might also explain it. With only few exceptions, the pipes in section E-E' near Muth are restricted to beds Me438Me440 (Figs 2, 3). From the observations, it seems that the initiation horizon of nearly all cylindrical structures is at the top of bed Me437, as indicated by tiny funnel-shaped pipes starting at the interface of beds Me437/Me438 and by small liquidization cells below the cylindrical structures (Fig. 4b); there are only a few pipes originating from below. A similar relatively narrow defined horizon of
The upper bedding surface of bed Me440 forms the boundary between FA 2 and FA 3 (Fig. 3); from its appearance, it even may represent a short period of non-deposition. This boundary is also the upper termination of all cylindrical structures at this level of the Muth Formation. No single continuation of pipes has been found above this surface and hence the formation of the pipes is interpreted to have happened before the sedimentation of the fine-grained dolomite above bed Me440. Consequently, there was no layer with reduced permeability (e.g. Plint 1983; Nichols et al 1994; Obermeier I996b) above the cylindrical structures during their formation, although the strength of the upper bed surface of Me440 might have been increased by microbial activity, observed in comparable levels of the Muth Formation in sections near Mikkim (Draganits 2000). The upper bedding surface of bed Me440 is covered with irregularly spaced pits (Fig. 5a, b) resembling spring pits (Quirke 1930), but also earthquake induced circular settlements in sand (Galli 2000). Thus, the liquidization structures in the Muth Formation represent a rare combined occurrence of spring pits and their conduits below. Pit diameters are surprisingly similar; bearing in mind that most of the pipes seem to originate from the same depth near bed Me437, this possibly implies that the final pipe height controls the diameter. The lack of small erosion channels leading away from pit centres indicates that they formed under water (compare with Quirke 1930). This setting additionally implies fully water-saturated pore spaces during pipe formation. The good preservation of the spring pits also excludes strong erosion of the bedding surface after their formation. The fine-grained dolomite covered the spring pits with slow sedimentation rates and protected it from erosion; today the dolomite weathers much more rapidly than the quartz arenite of bed Me440 below, therefore the pits are very well exposed (Fig. 5a, b).
118
E. DRAGANITS ETAL
sand, silty sand and rarely in gravel, but is very unusual in sediment with more than 15% clay content (Obermeier 1996a). Liquefaction takes place only where the sediment is completely satuThe upward flow of the water is controlled by two rated, strongly affecting sediments from a few to mechanisms; relief of a high pore-water pressure about ten meters depth; the susceptibility to liqueand reconsolidation of the sediment grains. The rims faction decreases nearly to nil at greater depths of the spring pits show only a thin cover of sand (Obermeier 1996a). rising only slightly above the bed surface with a If the liquidization structures of the Muth smooth contact to the bed, indicating that the upward Formation were caused by seismic activity, then the water flow was not competent enough to carry much existence of deformation structures would support sand. this model. Deformation bands, which have been Several different explanations of cylindrical struc- found in the Muth Formation, are suitable canditures in arenaceous sediments have been discussed dates for such deformation structures. Seismic extensively by Deynoux et al (1990), Hunter et al events during their formation can trigger liquefac(1992) andMassari etal (2001). tion, but deformation bands do not channel upwardFollowing their arguments, many of the possible directed fluid flow, because they show reduced formation mechanisms can be ruled out. The coastal permeability compared with the host sediment due barrier island depositional environment makes liq- to grain size reduction and compaction (Aydin uidization by slumping or overpressure by very 2000). The orientation of the acute bisectrix of these rapid sedimentation unlikely. The position of the conjugate deformation bands indicates E-W dipipes in aeolian sand dunes (FA 2) below lagoonal rected palaeo-strain direction, identical with the sediments (FA 3) additionally reduces the probabil- palaeo-strain direction of brittle faults in the underity of strong wave action. Collapse following disso- lying Pin Formation. These structures indicate lution or melting of underlying material is excluded seismic oblique contraction in a transcurrent tecby the lack of suitable sediments or former ice below tonic setting and document a previously unknown the pipes; the lack of organic material also argues pre-Himalayan deformation stage at the northern against a formation by hydrocarbon gas escapes. Hot passive margin of the Indian continent during the water (Rawlings 1998) or volcanic exhalation is dis- Early Devonian. qualified by the total lack of indication for volcanic In recent publications, it has been shown that activity. Filling of pipe-shaped cavities is improba- deformation bands typically form in porous sandble, as suitable cavities did not exist; concretions stone (Aydin & Johnson 1983; Antonellini et al around organic cores like tree trunks or plant roots 1994; Mair et al 2000). Thus their older age limit is look different and there are also no indications constrained by the depositional age of the sediment for former organic material. Animal burrows are and the younger age limit by the timing of thorough unlikely considering the size of the pipes, their cementation. Quartz grains of the Muth Formation regular shape and the lack of any other trace fossils rarely show fractures or pressure solution, indicating at this outcrop of the Muth Formation. relatively early diagenetic cementation. Cashman & Two explanations remain: liquefaction by earth- Cashman (2000) have shown from Pleistocene quake shaking and fluidization by water springs. marine terraces in California that deformation bands Both explanations concern an upwards-directed flow can develop at 'essentially surface conditions' and of a water/sand mixture, but differ in the reason that their formation can be related to seismic slip behind liquidization. However, distinguishing be- events on nearby faults. Therefore, the age of defortween seismic and non-seismic liquidization struc- mation bands probably is close to the age of seditures can be difficult (Holzer & Clark 1993; ment deposition and the seismic events responsible Obermeier 1996a, b\ Li etal 1996). for them might also have triggered liquefaction. However, there are also several arguments against a seismic explanation of the cylindrical structures. In Relationship between cylindrical structures the Muth Formation near bed Me437, although the and deformation bands bedding surfaces show some undulation, there is no indication of severe liquefaction. The vertical settleEarthquakes with magnitudes of five or more are ment caused by sediment densification is commonly thought capable of causing liquefaction, and magni- only some 3% of the liquefied sediment thickness tudes of about 5.5-6 are given in the literature for (Obermeier 1996a). Therefore, a single seismic seismic events at which liquefaction effects become event is regarded incapable to produce enough water relatively common (Obermeier 19960). Lique- to form the cylindrical structures and spring pits of faction of seismic origin, which is mainly caused by the Muth Formation, bearing in mind the multitude a cyclic shaking of the ground, is most common in of large pipes within unliquidized host sediments Common explanations of vertical cylindrical structures
FLUIDIZATION PIPES AND SPRING PITS
and the well-developed concentric lamination of several pipes that argue for considerable fluid flow through the pipes. Furthermore, this hypothesis is supported by the depth of spring pits that may give indications for a combination of volume loss due to compaction and by selective out-wash of finer grain sizes (Best 1989; Draganits & Janda 2003). More persistent and less vigourous flow of water than expected from short-lived liquefaction processes may explain the size, shape and internal structures of the liquidization structures of the Muth Formation. Additionally, the regular cylindrical pipes from the Muth Formation look quite different from liquefaction structures of unequivocal seismic origin, which usually are much more irregular (Obermeier 1996a). No evidence of fracturing or indications of rapid, short-lived, vigorous expulsion, as might be expected with seismic liquefaction (Obermeier 19960, b\ Takahama et al 2000) has been found.
Formation of the cylindrical structures in the Muth Formation The cylindrical structures of the Muth Formation are interpreted as fluidization pipes formed by channelled ascending ground water, while the surrounding sediment remained largely unfluidized; possible other explanations have been discussed in the previous section. Similar cylindrical structures in sandstone have commonly been attributed to non-seismic origin, often to up-welling of ground water (e.g. Hawley & Hart 1934; Gabelman 1955; Dionne 1973; Deynoux et al 1990; Li et al 1996; Dionne & Perez Alberti 2000; Massari etal 2001). Accepting a possible ground water spring explanation for the cylindrical structures, what are the possible reasons for up-welling water in the barrier island depositional environment of the Muth Formation? During deposition of the Muth Formation in early Devonian time, the climate is generally regarded as having been relatively arid. Additionally the high maturity, surprising consistency of the quartz arenites of the formation and the thorough oxidation of the fine-grained dolomite above the pipes reduces the probability of a major fluvial or precipitation influence. Therefore, freshwater, which can rise upwards in a marine phreatic zone due to gravity differences as suggested by Hawley & Hart (1934) and Deynoux et al (1990) is regarded unlikely to have formed the pipes in the Muth Formation, but cannot be ruled out completely. The sharp lithological break on top of bed Me440 from quartz arenite to very fine-grained dolomite indicates a major change in the depositional environment of the Muth Formation. Rapid rise in relative sea level indicated by the deposition of lagoonal sediments above barrier island dunes possibly also trig-
119
gered a rapid rise in ground water seepage that produced a relatively high hydrostatic head, resulting in the formation of springs. Subtle grain-size variations probably caused small inhomogeneities in fluid flow, which initiate self-accelerating channelling of the vertical upward movement of fluid (Couderc 1985), that resulted in randomly distributed pipes and related spring pits (Fig. 5a). The observation that only small amounts of sand have been extruded on top of the pipes in the Muth Formation probably argues for a relatively slow but persistent upward flow of water (seepage of Lowe (1975), compare with Guhman & Pederson (1992)), although a penecontemporaneous erosion of possible sand volcano deposits cannot be ruled out completely. Natural examples by Guhman & Pederson (1992) and Deynoux et al (1990) and experiments by Hawley & Hart (1934) indicate the formation of cylindrical water up-welling structures in loose sand without an aquiclude on top of the sand. Quirke (1930) observed fresh spring pits in lake shore sands which formed after heavy rainfalls just by rising ground water seepage; in his example there was also no low permeable layer involved. The rise in relative sea level might be related to tectonic subsidence caused by tectonic activity linked to the formation of deformation bands in the Muth Formation. This means, if tectonic activity was involved, it did not form the cylindrical structures by seismic liquefaction directly, but might be responsible indirectly through ground water seepage rise resulting from tectonic subsidence. Due to the little relief in this environment, the sea-level rise affected a relatively large area, thus pipes are found in widely separated sections. The existence of rare cylindrical structures some metres higher up in FA 3 (Fig. 3) possibly indicates some repetitions of tectonic activity, or that their formation was just linked with frequent sea-level fluctuations common in shallow marine environments.
Conclusions Numerous cylindrical structures cross-cutting stratification perpendicularly in Lower Devonian barrier island arenites from the NW Himalayas represent channels for upward flow of ground water. Pipes initiated from a relative thin horizon; their upper termination formed spring pits. These structures thus represent very rare examples of the preservation of fossil spring pits and their conduits below; their formation is explained by variations in ground water seepage in a coastal depositional environment. Rapid rise in relative sea level, indicated by the deposition of lagoonal sediments above barrier island dunes, possibly caused a rapid rise in ground water seepage that produced a relatively high
120
E. DRAGANITS ETAL.
hydrostatic head, resulting in the formation of springs. Due to the minor relief in this environment, the sea level rise affected a relatively large area and cylindrical structures can be found in widely separated sections. The occurrence of conjugate deformation bands in the Muth Formation with east to west directed palaeo-strain direction documents a previously unknown pre-Himalayan deformation stage at the northern passive margin of the Indian continent during the Early Devonian. Rapid relative sea level variations might have been triggered by tectonic activity related to the formation of deformation bands. This means, if tectonic activity was involved, it did not form the cylindrical structures by seismic liquefaction directly, but might be responsible indirectly through ground water seepage rise resulting from tectonic subsidence. Many thanks to U. Exner, S. Gier, R. Thiede and G. Wiesmayr with whom we shared unique experiences in these mountains. We are grateful to D. Banerjee for productive collaboration as well as K. Dorje and his family for immense hospitality during several field seasons. Many thanks to H. Rice who improved both contents and style of this paper. Our paper benefited greatly from thorough reviews by Geraint Owen and Massimo Moretti. The financial support by Fonds zur Forderung der wissenschaftlichen Forschung (P-14129-Geo) and Hochschul jubilaumsstiftung der Stadt Wien (H-32/2001) is acknowledged.
References ALLEN, J.R.L. 1982. Sedimentary Structures - Their Character and Physical Basis. Developments in Sedimentology, 30, Volume 2, Elsevier, Amsterdam. ALVAREZ, W., STALEY, E., O'CONNOR, D. & CHAN, M.A. 1998. Synsedimentary deformation in the Jurassic of southeastern Utah - a case of impact shaking? Geology, 26,579-582. ANTONELLINI, M.A., AYDIN, A. & POLLARD, D.D. 1993. Microstrucrures of deformation bands in porous sandstone at Arches National Park, Utah. Journal of Structural Geology, 16, 941-959. AYDIN, A. 1978. Small faults formed as deformation bands in sandstone. Pure and Applied Geophysics, 116, 913-930. AYDIN, A. 2000. Fractures, faults, and hydrocarbon entrapment, migration and flow. Marine and Petroleum Geology, 17,797-814. AYDIN, A. & JOHNSON, A.M. 1983. Analysis of faulting in porous sandstones. Journal of Structural Geology, 5, 19-31. BEST, J.L. 1989. Fluidization pipes in volcaniclastic mass flows, Volcan Hudson, Southern Chile. Terra Nova, 1, 203-208. BHARGAVA, O.N. & BASSI, U.K. 1998. Geology of SpitiKinnaur Himachal Himalaya. Geological Survey of India Memoirs, 124,1-210. CASHMAN, S. & CASHMAN, K. 2000. Cataclasis and defor-
mation-band formation in unconsolidated marine terrace sand, Humboldt County, California. Geology, 28,111-114. COUDERC, J.-P. 1985. Incipient fluidization and particulate systems. In: DAVIDSON, J.F., CLIFT, R. & HARRISON, D. (eds) Fluidization, 2nd edition. Academic Press, London, 1-46. Cox, D.E. 1977. Pillars, polystrate formations, and potholes. Creation Research Society Quaterly, 14,149-155. DEYNOUX, M., PROUST, J.N., DURAND, J. & MERINO, E. 1990. Water-transfer cylindrical structures in the Late Proterozoic eolian sandstones in the Taoudeni Basin, West Africa. Sedimentary Geology, 66,227-242. DIONNE, J.-C. 1973. Structures cylindriques verticales dans du quaternaire a Arthabaska, Quebec. Sedimentary Geology, 9,53-63. DIONNE, J.-C. & LAVERDIERE, C. 1972. Structure Cylindrique Verticale dans un Depot Meuble Quaternaire, au Nord de Montreal, Quebec. Canadian Journal of Earth Sciences, 9,528-534. DIONNE, J.-C. & PEREZ ALBERTI, A. 2000. Structures cylindriques verticales dans un depot meuble quaternaire, en Galice (Espagne). Geographic physique et Quaternaire, 54, 343-349. DRAGANITS, E. 2000. The Muth Formation in the Pin Valley (Spiti, N-India): Depositional Environment and Ichnofauna of a Lower Devonian Barrier Island System. PhD thesis, University of Vienna. DRAGANITS, E. & JANDA, C. 2003. Sub-aqueous artesian springs and associated spring pits in a Himalayan pond. Boreas 32,436—442. DRAGANITS, E., BRADDY, S.J. & BRIGGS, D.E.G. 2001. A Gondwanan Coastal Arthropod Ichnofauna from the Muth Formation (Lower Devonian, Northern India): Paleoenvironment and Tracemaker Behavior. Palaios, 16,126-147. DRAGANITS, E., MAWSON, R., TALENT, J.A. & KRYSTYN, L. 2002. Lithostratigraphy, conodont biostratigraphy and depositional environment of the Middle Devonian to Early Carboniferous Lipak Formation in the Pin Valley (Spiti, NW India). Rivista Italiana di Paleontologia e Stratigraphia, 108, 5-34. FENNINGER, A. 2000. Konglomeratorgeln aus der oberkarbonen Auernig-Gmppe der Kamischen Alpen (Nafifeld, Osterreich). Mitteilungen der Osterreichischen Geologischen Gesellschaft, 91, 53—61. FUCHS, G. 1982. The geology of the Pin valley in Spiti, H.P., India. Jahrbuch der Geologischen Bundesanstalt, 124, 325-359. GABELMAN, J. W. 1955. Cylindrical structures in Permian(?) siltstone, Eagle County, Colorado. Journal of Geology, 63, 214-227. GALLI, P. 2000. New empirical relationships between magnitude and distance for liquefaction. Tectonophysics, 324,169-187. GUHMAN, A.I. & PEDERSON, D.T. 1992. Boiling sand springs, Dismal River, Nebraska: Agents for formation of vertical cylindrical structures and geomorphic change. Geology, 20, 8-10. HAWLEY, J.E. & HART, R.C. 1934. Cylindrical structures in sandstone. Geological Society of America Bulletin, 45, 1017-1034. HOLZER, T.L. & CLARK, M.M. 1993. Sand boils without earthquakes. Geology, 21, 873-876.
FLUIDIZATION PIPES AND SPRING PITS HUNTER, R. E., GELFENBAUM, G. & RUBIN, D.M. 1992. Clastic Pipes of Probable Solution-Collapse Origin in Jurassic Rocks of the Souithera San Juan Basin, New Mexico, United States Geological Survey Bulletin, 1808,L1-L19. KOPF, A. J. 2002. The nature and significance of mud volcanism: A review. Reviews of Geophysics, 40, doi: 10.1029/2001RG000103. Li, Y, CRAVEN, J., SCHWEIG, E.S. & OBERMEIER, S.R 1996. Sand boils induced by the 1993 Mississippi River flood: Could they one day be misinterpreted as earthquake-induced liquefaction? Geology, 24,171-174. LONERGAN, L., LEE, N. & JOHNSON, H.D., CARTWRIGHT, J.A. & JOLLY, R.J.H. 2000. Remobilisation and injection in deepwater depositional systems - implications for reservoir architecture and prediction. In: WEIMER, P., SLATT, R.M., COLEMAN, J., ROSEN, N.C. NELLSON, H., BOUMA, A.H., STYZEN, M.J. & LAWRENCE, D.T. (eds) Deepwater Reservoirs of the World. GCS-SEPM Foundation, 20th Annual Bob F. Perkins Research Conference, December 3-6, Houston, 515-532. LOWE, D.R. 1975. Water escape structures in coarsegrained sediments. Sedimentology, 22,157-204. MAIR, K., MAIN, I. & ELPHICK, S. 2000. Sequential growth of deformation bands in the laboratory. Journal of Structural Geology, 22,25-42. MASSARI, F., GHIBAUDO, G., D'ALESSANDRO, A. & DAVAUD, E. 2001. Water-upwelling pipes and soft-sedimentdeformation structures in lower Pleistocene calcarenites (Salento, southern Italy). Geological Society of America Bulletin, 113,545-560. McKEE, E.D. 1957. Primary structures in some recent sediments. American Association of Petroleum Geologists Bulletin, 41,1707-1747. MOUNT, J.F. 1993. Formation of fluidization pipes during liquefaction: examples from the Uratanna Formation (Lower Cambrian), South Australia. Sedimentology, 40,1027-1037. NETOFF, D.I. & SHROBA, R.R. 2001. Conical sandstone landforms cored with clastic pipes in Glen Canyon National Recreation Area, southeastern Utah. Geomorphology, 39,99-110. NICHOLS, R.J., SPARKS, R.S.J. & WILSON, C.J.N. 1994. Experimental studies of the fluidization of layered
121
sediments and the formation of fluid escape structures. Sedimentology, 41,233-253. OBERMEIER, S.F. I996a. Using Liquefaction-Induced Features for Paleoseismic Analysis. In: McCALPiN, J.P. (ed.) Paleoseismology. Academic Press, San Diego, 331-396. OBERMEIER, S.F. I996b. Use of liquefaction-induced features for paleoseismic analysis - An overview of how seismic liquefaction features can be distinguished from other features and how their regional distribution and properties of source sediment can be used to infer the location and strength of Holocene paleo-earthquakes. Engineering Geology, 44,1-76. PLINT, A.G. 1983. Liquefaction, fluidization and erosional structures associated with bituminous sands of the Bracklesham Formation (Middle Eocene) of Dorset, England, Sedimentology, 30,525-535. PRICE, G.D. 1996. The significance of infiltrated clays within the Lower Permian Yellow Sands of north-east England. Geological Journal, 31,1-7. QUIRKE, T.T. 1930. Spring pits; sedimentation phenomena. Journal of Geology, 38, 88-91. ROTH, A.A. 1992. Clastic pipes and dikes in Kodachrome Basin. Origins, 19,44^8 RAWLINGS, D. 1998. Cover photo. The Australian Geologist, W9 (4), 1. TAKAHAMA, N., OTSUKA, T. & BRAHMANTYO, B. 2000. A new phenomenon in ancient liquefaction - the drawin process, its final stage. Sedimentary Geology, 135, 157-165. VAN HUSEN, D. 1999. Geologisch-baugeologische Erfahrungen beim Bau des Eisenbahntunnels Lambach, OO. Mitteilungen der Osterreichischen Geologischen Gesellschaft, 90,137-154. WALKER, T. 2000. Fluidisation pipes: Evidence of largescale watery catastrophe. Creation Ex Nihilo Technical Journal, 14, 8-9. WffiSMAYR, G. & GRASEMANN, B. 2002. The eo-Himalayan fold-thrust belt: Implications for the geodynamic evolution of the NW-Himalayas. Tectonics 21, doi: 10.1029/200 2TC001363. YOUD, T.L. 1973. Liquefaction, flow, and associated ground failure. United States Geological Survey Circular, 688,1-12.
This page intentionally left blank
Fluidization structures produced by upward injection of sand through a sealing lithology ANDREW HURST1, JOE CARTWRIGHT2 & DAVIDE DURANTI1 1
University of Aberdeen, Department of Geology and Petroleum Geology, Kings College, Aberdeen AB24 9UE, UK ^University of Wales, Department of Earth Sciences, PO Box 914, Main Building, Cardiff, South Glamorgan CF10 BYE, UK Abstract: Subsurface and outcrop data are used to describe sand injectites, a group of genetically related features that includes sandstone dykes and sills, but also structures within depositional sand bodies. Fluidization is identified as the process by which sand is injected but we draw attention to the lack of constraints regarding fluidization velocity and fluid viscosity. Injectites are shown to develop between < 10 m and 500 m below the seafloor. No relationship between depth of generation and injection geometry is found. Liquefaction of sand may produce sufficient excess pore fluid to create small sand injections during shallow burial. Large injectite bodies are identified on seismic data that may exceed 4 X 107 m3 are unlikely to be related to sand liquefaction. The general validity of hydraulic fracture as the mechanism for seal failure and propagation of injections is questioned. The association between the formation of polygonal faults and sand injection provides one of several alternative mechanisms for seal failure. Multi-phase intrusion is proposed as a likely mechanism for the formation of large sand intrusions, both because of the cyclical nature of most of the process invoked in their formation, and the author's own observations. Many of the processes of sand injection remain poorly constrained.
Introduction Sand injections (sand injectites) comprise dykes and sills, some of which are genetically related and features formed within depositional sand bodies during sand remobilization. From oil field borehole data sand dykes and sills are known to emanate tens to hundreds of metres above the deep-water depositional sand units from which they are assumed to be sourced (Dixon et al. 1995; MacLeod et al. 1999; Purvis et al 2002). Injectites that emanate from the edges of depositional sand bodies, sometimes referred to as wings, are low-angle (15^-0°) dykes that may be up to 25-30m thick, laterally extensive on a kilometre scale and in Alba Field (Eocene, central North Sea) are known to be sand-filled following drilling of hydrocarbon production wells (MacLeod et al 1999). Sand injectites are believed to form by natural hydraulic fracturing of a sealing layer, release of overpressure in the underlying sand body accompanied by fluidization of sand, and upward injection (Lonergan et al 2001). While natural hydraulic fracture is an appealing mechanism for producing small sand injections (Cosgrove 2001), we are unsure whether it is a satisfactory mechanism for producing decametre-scale injections. As many sand injectites seem to form during relatively shallow burial, commonly before substantial lithification of sand (Lonergan et al 2001) and probably at depths between a few metres and 500 metres burial, dise-
quilibrium compaction (Osborne & Swarbrick 1997) is a possible process by which overpressure may support hydraulic fracture. However, this process is speculative and other possible drivers for sand injection and fluidization warrant closer investigation. Injectite wings often form spectacular images on seismic data (MacLeod et al 1999 \ Duranti et al. 2002a), however little borehole data are available to examine the character of these features. Outcrop analogues of wings are not documented although there are some candidates (Duranti et al. 2002&). In contrast, borehole data, including cores, are available from many injectites present along the crestal areas of oil fields and from outcrops. Here, we characterize the geometry and internal features of injectites that emanate from the crestal areas of depositional sand units and use these to assess the origin of the injectites. As injectite complexes commonly record several injection episodes and have occurred at a range of burial depths within a single injectite complex (Duranti et al. 2002«; Purvis et al 2002), the evidence for the multi-cyclic character of sand injections is examined and the implications for the processes involved are considered.
Sand injectites Although descriptions of sand injectites date back to the early days of earth science (Murchison 1827)
Fig. 1. Interbedded, organic-rich siltstones and fine-grained sandstones are disrupted by sill-like features (si and s2) that contain fine-grained sand. s2 is connected downward into what may be the source depositional bed for the sand injection, si and s2 are part of the same injection, in part separated by the shale clast breccia, which has presumably formed by disraption of interbedded organic-rich silts and sands similar to those preserved to the left. Brittle and ductile deformation of the depositional lamination is present as is shale clast breciation (facies B4). Note the presence of a pygmatically compacted dyke emanating from the top of the intrusion (D). This is an example of a dyke developing more shallowly than a sill within a composite sand injectite. The upper surface of si has caused ductile deformation of the overlying fine-grained layer. Note the sharp contact margins of si and s2. Hammerhead is O.lm long. From the Upper Jurassic Helmsdale Boulder Beds, NE Scotland (photo courtesy of N.H. Trewin).
very few outcrop descriptions capture the same scale of features as identified in oil fields, however they share many textural characteristics and geometries. An example of sand injectites formed during seismic activity in the Upper Jurassic along the Helmsdale fault (northern Scotland). A large normal fault that defined a marine basin margin in northern Britain during the Kimmeridgian (Bailey & Weir 1932; MacDonald & Trewin 1993) and preserved a wide range of features that are similar to those seen in the subsurface, although usually at a smaller scale. An example from northern Scotland (Fig. 1) shows the development of sills (si, s2, s3), ptygmatic dykes (D), a shale clast breccia, low-angle shear planes (sp), and associated brittle and ductile deformation within the host, laminated organic-rich siltstones and fine-grained sandstones. Detailed examination of the outcrop reveals that undeformed beds occur approximately 2 m above the injectites and injection is inferred to have occurred at less than 5 m (decompacted thickness) below the seabed. The ptygmatic
character of the dykes is indicative of post-intrusion compaction. In this example there is no evidence of erosion in the overlying section or that any sand extruded onto the seafloor. Organic-rich shale, that we assume restricted the upward escape of pore fluid during compaction, forms the upper limit of si, although is locally intruded by small dykes. The top of the sill is discordant and appears to have 'eroded' into the mudstone. All contacts between injectites and host strata are sharp. The host strata appear to be prised apart and replaced by the injectites and evidence for incorporation of host strata in to the injectite bodies is limited to the shale clast breccia and s3, which contains sand-sized particles of organic siltstone. The shale clast breccia comprises angular and 'wispy' clasts similar to those described from subsurface examples (Duranti etal 2002a). Brecciation is interpreted to be the result of catastrophic failure of low permeability units during hydraulic fracture. Although the bases of si and s2 appear to intrude
FLUIDIZATION STRUCTURES PRODUCED BY UPWARD INJECTION OF SAND
Fig. 2. Lithostratigraphy of Paleocene-Eocene section in the North Sea with major sandstone reservoirs and known intervals affected by sand injectites.
downwards, cutting across underlying finer grained laminae, it is suggested that the 2-D view may be of an injection that is cutting the laminae obliquely, either upward or downward. It appears that the contact geometry of upper, lower and lateral surfaces of sand injections is similar. The source of the injected sand cannot be constrained, because of lack of exposure. Although the injectite in Figure 1 is small, certainly in terms of several subsurface examples that sand injectites exhibit some aspects of scale invariance. For example, fine-grained sand may inject into muds and form features with similar geometry and contact features both at centimetric (Fig. 1) or kilometric (subsurface oil field examples) scale. Hence, laboratory experiments of sand injection may not require scaling to produce analogous features to those recorded in nature (Hurst, unpublished data). Here, we focus on subsurface data from the Paleogene section of the North Sea, which is dominated by deep-water clastic deposition (Fig. 2) and contans many of the best documented subsurface sand injectite complexes.
Fluidization Fluidization is the widely accepted process by which sand is remobilized and injected in the subsurface. There is however remarkably little direct evidence to substantiate how large-scale sand remobilization by fluidization occurs. Fluidization velocity (Vf) associated with sand fluidization is poorly constrained, but increases as fluid viscosity decreases. Some experimental data (Wilson 1980; Nichols et al 1994) and observation of modern processes (sand boils, etc.) leads to the conclusion that Vf and viscosity of fluidized sand are very variable. Following Lowe (1975), the Vf may be as low as 0.01 ms"1 for fine sand, where,
125
and ps = density of solid, p f = density of fluid, d = mean diameter of particles (mm), g = gravity and \L = viscosity. The features described as the product of explosive dewatering by Lowe (1975) are tiny (O.lm high) in comparison with those described here, thus, we treat 0.001 ms"1 as a minimum Vf for fine sand (the average grain size of the injectites in this study). High viscosity sand fluidization may form subhorizontal flow laminae (Thompson et al. 1999; Duranti & Hurst, unpublished data) that are superficially similar to flow laminae in igneous intrusions. Flow laminae are not observed in any of the injectites described here. The absence of sub-horizontal flow laminae is interpreted to imply that the injected fluid had low viscosity; at present, there is no way of estimating fluid viscosity from the rock record. In the Alba Field, it seems likely that sand injection occurred within 200-300 metres of the sea floor (Duranti et al 20020). Any gas dissolved in pore fluids will exsolve in response to a reduction in pressure, for example, when a pressure seal is breached. Gas escape forms characteristic subvertical pipes (c.f. pillars of Lowe 1975) adjacent to bodies of unfluidized material (Wilson 1980) and upward merging structures (Nichols et al 1994), by a process termed aggregative fluidization (Lowe 1975).
Seismic data Injectites above and along the crestal area of oil fields are best viewed transverse to the depositional sand-rich fairways. A variety of geometries are identified (Fig. 3), many of which are penetrated by boreholes, thus allowing verification of the lithological interpretation. Seismic resolution of injected sands with sill or low-angle dyke geometry is possible providing the sand bodies are sufficiently thick to give a tuning response or discrete reflections from top and base of the body. We estimate that sand injectites of the order of 10m thicknesses are resolved in many North Sea Paleogene examples (Huuse et al in press). The limit of detection is strongly dependent on acoustic impedance contrast between sand and the adjacent fine-grained strata. Vertical to steeply dipping features, such as dykes cannot be imaged directly, unless there is some associated deformation such as a fault, and estimation of their presence, size and distribution, is therefore problematic. Although sand injectites are seen on p-wave (compressive wave) seismic data in several North Sea oil fields (Jenssen et al 1993; Lonergan & Cartwright 1999; Purvis et al 2002), more confident mapping of their geometry and lithology is possible if shear wave data are
126
A.HURSTErAL.
Fig. 3. Random cross sections of the depositional submarine slope channel that forms the long axis of the Alba oilfield (Eocene, central North Sea). Data are from the 3-D converted wave (3D-PS) seismic survey (MacLeod et al 1999) showing development of various sand injectite features along the crest and edges of the field. Where sand injections are not present the top of the sand is irregular with step-like geometry in the sills. The irregular surfaces are similar to those observed on outcrops (Fig. 1, Duranti et al. 2002a). Outline represents the inferred limits of the sand distribution. All interpretations are supported by well penetrations. A - Lacolith style sills connected to the depositional Nauchlan sand body by dykes. Well locations in the line of section are superimposed. B - Low (15-25 deg when corrected to true depth from two-way time) angle dykes emanating from the edges of the depositional body of the Nauchlan Member. In both sections the yellow picks are the top and base of the C2 sand. Courtesy of ChevronTexaco UK.
available, for example the 3-D converted wave seismic (3D-PS) data from Alba oilfield (MacLeod etal. 1999). Even where undisrupted by sand dykes, the top of the Nauchlan sandstone has a cuspate, often concave-upwards, surface (Fig. 3). This geometry of
this surface is unlikely to be of depositional origin and is interpreted to have formed during sand mobilization and injection along the crestal area (Duranti et al. 2002a). Similar features were inferred to have formed in the Harding field associated with largescale slumps within overlying and adjacent shale
FLUIDIZATION STRUCTURES PRODUCED BY UPWARD INJECTION OF SAND
127
Fig. 4. Seismic character of the fine-grained overburden into which sand injectites have intruded. Note the broken lateral reflections and the presence of subvertical features that, when calibrated with borehole data sometimes prove to be sand-rich. A - Lateral discontinuity of reflectors and irregular geometry dominates the Brioc Member (Fig. 2) above the Alba oil field (MacLeod et al. 1999, Duranti et al. 2002a). Minor faults may be inferred but are difficult to trace (courtesy ChevronTexaco UK). B - Highly irregular reflectors with little evidence of stratiform features in the Lower Forties Member (Palaeocene) between the sand-rich Lista and Upper Forties formations, Nelson oil field, central North Sea). Immediately to the right and below the well markings (22/11-N27) several sub-vertical features are present that may be sand injectites (courtesy Enterprise Oil UK).
units (Dixon et al. 1995). There is no direct evidence of slumping in shales from borehole data. The sand inectites are similar in geometry to igneous intrusions, lacolith (Fig. 3a) as well as dykes with low (Fig. 3b) and higher (Fig. 3a) angle discordance with bedding. Shales into which sand has intruded have an irregular pattern on seismic images, which we interpret to reflect the irregular thickness distribution of the sand injectites rather than folding and/or slumping of the shales (Fig. 4). Where the caprock shales are deformed by polygonal faults (Cartwright & Lonergan 1996; Lonergan & Cartwright 1999), it is likely that fault planes will act as potential failure planes for sand intrusion and much of the complexity in the seismic expression of the top seal units is probably linked to this deformational style.
Borehole data Most boreholes are located along inferred structural crests of target hydrocarbon reservoirs. Cores from these boreholes reveal a large number of subdecameter scale dykes and sills, none of which are resolved by seismic data. Hence, mapping of sand injectites from seismic data provides a minimal (and possible substantial under-) estimate of the number and volume of sand injectites present. An account of the petrophysical and geological relationships of borehole data from the Alba oilfield was presented by Duranti et al. (2002a) and provides the basis for the following summary. Distinctive bulk density and acoustic velocity characteristics reveal the presence of decameter thick sand injectite units within depositional sand
128
A.HURST£rAL. bilization. This grain textural relationship need not hold for all sand injectites with respect to their depositional precursors, but the likelihood for some degree of grain mixing and re-packing during remobilization is high. Most of the injectites that we have examined are fine or medium grained sandstone, as is expected from hydraulic considerations (Lowe 1975). Sand injectites are easily identified where they intrude finer grained strata and cross cut bedding (Fig. 6). Where dykes cut through depositional sand units, in particular where they have similar grain size, they may be distinguished by the absence of depositional sedimentary structures and different grain packing, which is revealed when the rocks are stained by alteration products from weathering or diagenesis (e.g. limonite or iron oxide) or residual oil (see examples in Thompson et al 1999). Deformation bands that run sub-parallel to the margins of dykes and sills are the only commonly occurring structure apart from occasional shale clasts that may tend to concentrate in the central parts of the injectite bodies. In contrast to previous studies, we identify a suite of facies that are inferred to have formed by sand remobilization within the depositional parent sand body. Thus, sand injectites include other sand units with a range of facies and associations in addition to those comprising dykes and sills.
Fig. 5. An example of the correlation between injectite and depositional facies identified in core and wireline data. Note that the acoustic velocity (DT) and bulk density (RHOB) increase in injectite units relative to depositional units. The natural gamma-radiation (SGR) lacks distinctive character. Differentiation between laminated and weakly laminated sands is made from core description and not from log responses; both are depositional facies. Depth scale in feet. Courtesy of ChevronTexaco UK.
units are (Fig. 5). The injectites are relatively denser and faster than the depositional sand units. Depositional units include those with primary sedimentary structures and those with secondary structures formed by liquefaction (Lowe 1975; Hurst & Cronin 2001). Injectite units include dykes and sills but more commonly are the unstratified facies of Duranti et al (2002a) that are interpreted to have formed by remobilization of sand within depositional sand units. Petrographic and core analysis data confirm that the sand injectite units have tighter grain packing and lower porosity than depositional sands units. The higher density and faster velocity associated with sand injectites in the Alba oilfield is explained by the resorting and repacking of the poorly sorted grain textures present in depositional sand units into more tightly packed grain textures following remo-
Structures formed by sand remobilization Four distinctive facies typify sand injectites (Table 1) that are termed unstratified facies (Duranti et al 20020) to distinguish them from depositional sedimentary structures and secondary sedimentary structures formed during liquefaction. The facies often form a vertical association that is interpreted to record successively more vigorous fluidization, as follows: oversteepened laminae (Fig. 7a), giant pillars (Fig. 7b), homogenized sand (Fig. 7c) and shale clast breccia (Fig. 7d). The facies association is observed to underlie dykes that penetrate low permeability overburden and their detailed characterization will be published elsewhere. The spatial relationship between the unstratified facies association and overlying dykes is compelling evidence for their common genesis. In the Alba oilfield, MacLeod et al (1999) were able to make accurate prediction of sand distribution from the 3-D PS data and delineation of the intrusive portions of injectites within the overburden by detailed correlation with borehole data (Duranti et al 20020). Oversteepened laminae (Fig. 7a) appear to have formed by distortion of depositional and secondary lamination. They steepen and become increasingly disrupted upward. Grain sorting associated with sedimentary laminae is frequently absent. Outcrop
FLUIDIZATION STRUCTURES PRODUCED BY UPWARD INJECTION OF SAND
129
Fig. 6. Sandstone dykes. A: oil-saturated sand dykes intruding grey mudstones from Alba Field (UKCS). The mudstones are only lightly consolidated. Dark graduations on scale bar are one tenths of one foot (approximately 30mm). B: sandstone dyke at outcrop from the Yellowbank Creek sand injectite complex, west of Santa Cruz, California (Thompson et al. 1999). Note the subhorizontal laminae in the dyke, the presence of which is enhanced by a residual oil stain and iron oxide weathering products.
examples demonstrate the lateral variation in steepness away from the most intense area of inferred fluid movement (Fig. 8). Over steepening is attributed to plastic deformation of laminae by upwardflowing pore fluid. With increased rates of flow, disruption of laminae occurs and features similar to pipes and pillars (Lowe 1975) may form. Homogenized sandstone (Fig. 7d) has no recognizable bedding. Laminar features present are deformation bands without evidence of cataclasis. An intense mixture of small pillars, pipes and other dewatering structures is responsible for the homogeneous character. Sub-vertical and folded fractures are common. Some vertical alignment of grain long axes is observed. This facies commonly, but not only, occurs at the upper contacts between thick sand bodies and overlying shales. Sharp, discordant contacts with overlying shales resemble those of intrusive igneous bodies with their host rocks and are typical of post-depositional emplacement accompanied by brittle deformation. Homogenized sandstones are interpreted to record pervasive fluidization (Duranti et al. 2002a) and probably aggregative fluidisation (Lowe 1975).
Shale clast breccias (Fig. 7d) are sand matrix supported, shale class in size that vary from sub-mm to >0.1 m. Clasts are angular and sometimes have a disrupted jigsaw configuration. Microfracture networks within individual shale clasts are often present. Shale clast breccias are typically located where dykes emanate from parent sand bodies and along the contacts between thick sandstone units and overlying shales. The shale-clast breccias are interpreted to have formed by catastrophic failure of a
Table 1. Unstratified facies of Duranti et al. (2002a) interpreted to have formed by the remobilization of sand during fluidization and sand injection Facies
Bl to B4 are in order of increased sand remobilization upward.
130
A.HURST£TAL.
Fig. 7. Images of cores with the unstratified facies typical of sand injectites (after Duranti et al. 20020). A: oversteepened laminae (facies Bl). B: giant pillars (facies B2). Large-scale disruption of primary and/or secondary laminae and pervasive subvertical pipes. C: homogenized sandstone (facies B3). No primary or secondary sedimentary structures are preserved and subvertical pipes superimpose each other. Occasional granulation seams are present. D: shale clast breccia (facies B4). Shale clasts are angular, sometimes in a jigsaw pattern. Individual clasts are commonly micro-fractured. The breccia is matrix-supported by sand. The dark graduations on the scale bars are one tenth of one foot (approximately 30mm).
shale layer when the fracture gradient was exceeded and clastic intrusion ensued. Some debrites may have chaotic internal organization and composition, similar to shale clast breccias. However, the pervasive sand matrix-supported texture, lack of clast orientation, presence of microfractures, and jigsaw clast configuration are considered characteristic of shale clast breccias in unstratified facies. Distinction between unstratified facies and debrites may be particularly difficult with borehole data and is of significance when reconstructing sand body geometry. For example, debrites occur in marginal or axial positions low in depositional fairways, whereas unstratified facies occur along the crests or upper margins of sand bodies, or more generally are located below fine-grained units (shales, mudstones, etc.). Petrographic evidence (Duranti & Hurst, unpublished data) reveals the structures in unstratified facies to be tightly packed sand grains. The structures are organized subvertically and share many characteristics with liquefaction structures (Lowe
1975, Hurst & Cronin 2001). Unstratified facies differ from liquefaction structures (e.g. dish structures, consolidation laminae), as they show evidence of upward movement of sand-sized particles as opposed to the downward settling of sand-sized particles recorded during liquefaction (Hurst & Cronin 2001).
Sand injectite facies association Structures that characterize unstratified facies form an association (Fig. 9) that records the breach of a low permeability seal (mud/mudstone) by a dyke, the breach point typically associated with a shale clast breccia (facies B4). The breach point is underlain by facies B3, B2 and Bl (Fig. 7a, b and c) in descending order and decreasing levels of deformation. This facies association may not have all facies present and the limited sampling volume of boreholes may sample only part of the association.
FLUIDIZATION STRUCTURES PRODUCED BY UPWARD INJECTION OF SAND
Fig. 8. Outcrop picture, and line drawing of the same, showing the cuspate upper boundary of an injected sand unit, Baldwin Creek, Near Santa Cruz, California. The shelfal marine Santa Margarita Sandstone (SMS) has undergone remobilization, thus disrupting and truncating the overlying Santa Cruz Mudstone (SCM). Note that the mudstone is undeformed whereas the sandstone has inclined laminae (oversteepened laminae, facies Bl, see Fig. 7a) that steepen to the right. Background information is found in Thompson et al. (1999) and Duranti et al (2002c).
Unstratified facies are illustrated in several earlier papers (Newman et al 1993; Jenssen et al 1993; Cosgrove & Hillier 2000) however, the structures are either interpreted as the products of liquefaction (e.g. dish structures) or their full significance in terms of fluidization is unrecognized. Contact geometry appears to be similar over a broad range of scales, centimetric (Fig. 1), metric (Fig. 5), and decametric (Fig. 3). From these relationships it may be possible to infer that the physical processes of sand injection are scale independent. We believe that the facies association is formed by catastrophic release of pressure, which built up below a low permeability (mudstone) seal. Pressure buildup is caused initially by the fine-grained (mud) low permeability unit (seal) restricting the rate of seepage of pore water being driven upward by sediment compaction. As the seal continues to compact, the permeability within the seal is reduced and the capillary force opposing fluid migration into the seal increases, thus further restricting seepage and increasing pore pressure in the underlying reservoir sand. Eventually, the fracture pressure of the seal is exceeded and fluidization and injection of sand ensues. The rapidity of the seal failure is recorded by the brecciation of shales, the absence of contorted beds
131
Fig. 9. Overview of the sand injectite facies association. Facies Bl, B2, B3 and B4 are as described in Fig. 7 and by Duranti et al (2002a). A crosscutting dyke and sill complex is depicted that is similar to that identified in seismic sections (Fig. 3) and at outcrop (Fig. 1). The upper surface of the main sand body is cuspate and has sharp discordant contacts with adjacent shales (c.f. Figs 3, 6b and 8). Dyke 1 feeds the sill and is post-dated by dyke 2, indicating multi-phase sand injection.
associated with large-scale slumps (in contrast to the descriptions of Dixon et al 1995) and the occurrence of sharp, discordant sand-shale contacts. Shales had neither time nor space to fold and contort, rather they were cut rapidly by the combination of brittle fracture and upward fluid movement. The sharp boundaries against their host strata indicate that the host strata were sufficiently cohesive to undergo brittle fracture and brecciation.
Burial depth relationships The depth at which sand injection occurs may have an important bearing on the geometry of the sand bodies formed, dykes tending to have a deeper origin while sills tend to develop at shallower depths (Lonergan et al 2001). This relationship is, however, ambiguous. Dykes that can be mapped back to their source depositional beds commonly extrude onto the (palaeo) seafloor (Boehm & Moore 2002) or land surface (Obermeier 1989, 1996), rather than forming sills within the shallow overburden (Figs 1 and 10; fig. lOa in Lonergan et al 2000; Obermeier 1989). Shallow (<10m burial) intrusions. A variety of evidence exists for sand injection during shallow burial (Parize 1988, 2001; Surlyk & Noe-Nygaard 2001). The most reliable evidence is when the full extent of a dyke, from source bed to palaeo-seafloor, is observed, which is uncommon in boreholes. Other evidence for shallow intrusion is ptygmatic folding of dykes. Although not all ptygmatic folding of sand dykes occurs during shallow burial, much does. The depth
132
A.HURST£TAL.
Fig. 10. A series of sills, including a composite uppermost sill (a), and dykes from near the base of the Moreno Shale (Cretaceous) at the head of Escapardos Canyon, Panoche Hills on the west side of the San Joaquim Valley, central California, which form an overall thickening upwards cycle of sills. The composite sill is two sills (that bifurcate to the south (right) out of view. The images are adjacent parts of a 25 m long outcrop over which the composite sill is continuous. All the sands are injectites. A: Sill al is cut by an irregularly bounded, but approximately vertical dyke, that feeds sill a2. At this locality sill b is not present and sill c (that feeds sill al) is cross cut by the dyke. At least two dykes (not in view) emanate approximately vertically from sill a2, which has a mounded top. B: Sill al, the lower sill, is slightly coarser and more poorly sorted than sill a2, and is fed by a dyke that cross cuts the laterally discontinuous, underlying sill b and is fed from sill c. Sill c is laterally discontinuous.
at which intrusion occurred can be estimated by decompacting the dyke (Hillier & Cosgrove 2002). Fine-grained sediment, especially mud, compacts more rapidly than sand during early burial. Initially mud may have a reasonably isotropic granular structure, however this is influenced by factors such as the content of organic matter and bioturbation, as well as mineralogical variations. In terms of initiating disequilibrium compaction, it may be that the onset of significant anisotropy in mud is the key to slowing fluid escape and generating transient overpressure. In general, shallow intrusions are volumetrically small features (Fig. 1 being typical). The low tensile strength of mud in shallow burial may inhibit buildup of the substantial overpressure needed to mobilize large sand volumes. We have seen many examples of small intrusions, generated during shallow burial, which emanate from thick sand beds
and believe that the volume of the parent sand body is not a critical factor governing the size of intrusions formed at < 10 m burial. Deeper (>10 to c. 500 m burial) intrusions. Large, decameter thick, sand injections occur during deeper burial. However, not all deeper sand intrusions are of decametric scale. Deeper intrusions, such as those observed in the various North Sea oil fields (Fig. 3; Jenssen et al 1993; Dixon et al 1995; Duranti et al 20020), and at outcrop (Fig. 10; Smyers & Peterson 1971; Thompson et al 1999), typically cross cut tens to hundreds of metres of finer grained strata. In order for these injections to occur, and in some cases to inject thousands of cubic metres of sand (Table 2), very substantial pore pressures would be required to open the fracture along which the sand injected and large pore pressure gradients from source bed to the
FLUIDIZATION STRUCTURES PRODUCED BY UPWARD INJECTION OF SAND
133
Table 2. Estimates of sand and fluid volumes incorporated into sand injections and fluid volumes released by the liquefaction and consolidation of sands
Small intrusion Large intrusion Consolidated unit
6.5-7.5 X 104 5.2-6.0 X107 6.5-7.2 X102
5X10 4 4X10 7 5X10 2
3.5 X 104 2.8 X10 7 3.5 X 102
1.5-2.5 X10 4 1. 2-2.0 X10 7 1. 5-2.7 X102
1.5 X10 4 1.2X107 1.5X102
VTI = gross volume of injected sand at the time of injection (using 0 = 60-80%). VT2 = volume of injected sand at the present day. Vg = grain (mineral) volume of injected sand body. Vfinj = fluid volume at the time of injection. Vfcon = fluid volume at the present day. All volumes are in m3. Data for the small and large intrusions are arbitrary values taken from the Injected Sands database (Universities of Aberdeen & Cardiff). Data for the consolidated unit is a maximum figure based on the large composite consolidation complex in fig. 2, Hurst & Cronin (2001).
upper limit of intrusion would be necessary to keep the fracture open during injection. Crosscutting relationships between different dykes and sills, from outcrop and subsurface data, can be used to demonstrate that sand injection occurs in several phases. A simple relationship between depth and intrusion geometry is not readily apparent. Careful consideration of the mechanism of clastic intrusion is required before the relationship between depth and style of intrusion can be substantiated and used in predictive models.
Pore fluid escape and pore pressure during shallow burial The shallow crust can be described as a continuous network of pore fluid (brine) separated by mineral grains. In the case of marine sediments, the pore fluid network is in continuity with the overlying ocean water. Hence, so long as a volume of pore fluid is in continuous contact with other pore fluid, overpressure in that fluid cannot be maintained (Bj0rkum et al 1998) and an approximately linear hydrostatic gradient should exist. Overpressured intervals are, however, common even during shallow burial (Maltman 1994) and form without the presence of extraneous pore fluids, aquathermal expansion or diagenetic reactions. During burial, pore fluid is continuously expelled upward as sediment porosity decreases. Formation of a low permeability horizon hinders fluid expulsion and fluid volume and pressure build-up below the low permeability horizon, forming transient overpressure, a process termed disequilibrium compaction (Osborne & Swarbrick 1997). Most subsurface sand injectites documented in the recent literature (Jenssen et al 1993; Dixon et al 1995; Duranti et al 2000; Lonergan et al 2000; Duranti et al 2002a,&), appear to have formed at less than 500 m burial, in some cases substantially shallower. In the example from northern Scotland where sills and dykes appear to have formed contemporaneously (Fig. 1), undeformed mudstones occur 2 m higher in the section, which, allowing for com-
pactional effects, probably means that the injectites formed at less than 5 m below the sea floor. It is interesting to note that even at very shallow burial, inferred high porosity in fine grained units, and with fine laminae as coarse as organic-rich silt (Fig. 1), the rate of pore fluid expulsion was sufficiently restricted to generate an upward force capable of rupturing the low permeability layers and injecting sand. Rock and fluid characteristics at the time of sand injection can only be inferred. In the context of deep-water sandstones, depositional porosity may be as high as 60% (Owen 1996), which commonly leads to localized syn-depositional and early postdepositional fluidization (Lowe 1975; Hurst & Cronin 2001). It may thus be reasonable to assume that during sand fluidization the fluid has <40% grains, that is no more grains than the most porous sediment. Using numerical simulations, vigorous fluctuations (pulses) of sediment fluidization are produced by maintaining a superficial gas velocity that is greater than Vf in sediment with an initial porosity of 45% (Kafui & Thornton 2001). In nature, excess fluid is generated from within sandstones during liquefaction at or near the earth's surface or seabed. Liquefaction consolidates the sediment by lowering porosity and releases pore fluid (Lowe 1975; Hurst & Cronin 2001), which is inferred to fluidize overlying sand.
Liquefaction: a linkage to large-scale fluidisation? Dish structures and consolidation laminae record periods of sand liquefaction (Lowe 1975). Evolution of water during liquefaction, and the concomitant decrease in porosity, is a mechanism by which adjacent sand may be fluidized and, hence, by estimating the volumetric abundance of dish structures and consolidation laminae an approximation of the volume of water evolved during liquefaction may be derived. The volume (V) of the sand prior to liquefaction is simply,
134
A.WRSTETAL.
Sand volume where m is the volume of mineral (granular) material present and 0 is the porosity (assumed to be fluid filled). During liquefaction, m remains approximately constant (minor movement of fine grained particles only) and the change in V is almost entirely a consequence of the reduction in 0 caused by consolidation of granular material (Hurst & Cronin 2001). If one assumes that prior to fluidization that sand has a minimum 0 = 60% (Owen 1996) and that is lowered to 30% by liquefaction (lower than the average porosity of sandstones containing dish structures and consolidation laminae in the Alba oilfield, Duranti et al. 20020) estimates can be made of the volume of water available to fluidize sand following liquefaction (Table 2). There is two orders of magnitude disparity between the volumes of granular material and fluid involved in liquefaction and that forming a small intrusion, making it necessary to invoke the action of several approximately simultaneous liquefaction events to support fluidization and intrusion. Although possible, one should note that the liquefied unit referred to in these estimates is large and possibly leads to over-estimation of the potential of liquefaction as a mechanism for largescale sand fluidization. If fluidization was initiated by multiple phases of liquefaction, abundant dish structures and consolidation laminae should be preserved below fluidized units. Using data from Table 2, at least 2 X 104 m3 of rock volume needs to undergo liquefaction in order to fluidize and inject a small sand intrusion and three orders of magnitude more for a large intrusion. Although dish structures and consolidation laminae are recorded from sandstones associated with sand remobilization (Newton & Flanagan 1993; Dixon et al 1995; Purvis et al 2002) they are insufficiently common to support the inference that liquefaction caused sand fluidization. The suite of unstratified facies (Fig. 7, Table 1) and their association have a clear relationship with sand injections into shaly overburden, and crosscut and otherwise disrupt secondary structures that formed during liquefaction. Disruption of liquefaction structures by sand injection is consistent with large-scale sand injection occurring at depth (down to 500m burial, Lonergan et al 2001; Duranti et al 20020) rather than a shallow origin associated with sediment consolidation (Lowe 1975, Hurst & Cronin 2001). We conclude that liquefaction that is preserved in the rock record is an unlikely source for the volumes of fluid needed to support large-scale sand injection features that are identified on seismic data (Fig. 3). Liquefaction, if it occurs during an appropriate postdepositional period, can almost certainly support small-scale fluidization (c.f. Lowe 1975 and possibly form features such as those in Fig. 1).
If one considers the volume of sand injected in a large intrusion, for example, 2.8 X 107 m3 of sand plus fluids (Vg, Table 2), it is worth considering both the dimensions of the depositional sand body that this was derived from, and the low permeability unit that overlay the sand prior to remobilization. The volume of sand in the injected sand body is assumed to be part of an original depositional sand volume within which overpressure developed below a low permeability shale-dominated unit. The shale unit may have been at the top of the sand unit or intrasand body shale. It is assumed that the area of the shale, acting as a sealing layer, dictates the volume of sand that can be remobilized.
Shale at the top of a sand unit Shale units at the tops of sand units (top seals) are generally laterally continuous on a regional scale and, in the context of hydrocarbon reservoirs, constitute seals, which are a fundamental component of any hydrocarbon play (Downey 1984). All the reservoir sands referred to in this study are overlain by, and inject into, regionally developed top seals that extend beyond the limits of sand deposition. The assumed volume of a large sand intrusion, 4 X 107 m3 (VT2, Table 2), is equivalent to a quite small depositional sand unit (10m thick and an area of 2 X 2 km2 thus, we envisage no mass-balance difficulties in sourcing several large sand intrusions from 20m or greater thick, few square kilometres or greater depositional sand units.
Intra-sand body shales (ISBS) Shales within sand units may form important barriers to fluid flow and cause localized variations in pressure within otherwise sand-rich sections (Weber 1986). In many of the sand injectites we have studied, intra-sand body shales (ISBS) have similar breaches and associated unstratified facies to those observed in top seals, direct evidence that ISBS can cause overpressure to develop locally within larger sand units. ISBS may be laterally discontinuous within the overall volume of a sand body. Using data from Table 2 and assuming sand bodies of 10,5 and 2 metre thickness, the areas of the ISBS below which they lay is estimated as 2 X 107 m2, 8 X 106 m2 and 4 X 106 m2, respectively (approximately 4.5 X 4.5 km, 2.8 X 2.8 km and 2 X 2 km). These areal dimensions are of a similar scale to the dimensions of the oil fields referred to in this paper (Newton & Flanagan 1993, Jenssen et al 1993; Dixon et al 1995; Purvis et al 2002) but
FLUIDIZATION STRUCTURES PRODUCED BY UPWARD INJECTION OF SAND
larger than shales in the shale dimension data derived from outcrops of sand-rich deep-water systems (Weber 1986; Stephen et al 2001). Assuming that the largest shale dimension (850m) from Stephen et al (2001) is the diameter of the ISBS unit (area = 7.2 X 105 m2) 2 m, 5 m and 10 m thick sand units sealed below it would have volumes in the range 1.5-7.2 X 106 m3, an order of magnitude smaller than the volume estimated for a typical large intrusion (Table 2). Despite the order of magnitude discrepancy between the volumes of large sand intrusions and ISBS dimensions, we believe that our estimates support that ISBS within sand-rich turbidites will, given appropriate seal character, cause build-up of overpressure that may lead to eventual seal rupture, fluidization and formation of large sand intrusions (this is substantiated by the occurrence of injectites within depositional sand bodies, Fig. 5). In support of this assertion, it is noted that the limitations of shale dimension data from outcrops, which probably fails to capture the dimensions of the largest ISBS because of the limitations of outcrop size. It seems unlikely that very large sand intrusions, such as the 'wings' identified at the margins of the Alba oilfield (MacLeod et al. 1999) can have been sourced below ISBS units. The potential for generating sand intrusions within sand bodies is clearly greater if several discrete ISBS units are present. Thus, if ISBS units have good seal capacity, sand injectites within sand bodies are likely to be more common than in sand bodies that lack ISBS units.
Multi-phase intrusion Multi-phase intrusion is an appealing mechanism for forming large clastic intrusions both because the physical processes for emplacing such large volumes of sand are poorly constrained and difficult to envisage and because pulsed phases of sand fluidization are known to occur during earthquaketriggered sand remobilization (Obermeier 1996). It is relevant to note that the volumes of sand extruded in association with recent seismicity are small (Obermeier 1989, 1996) compared to those identified in the subsurface (Dixon et al. 1995; Duranti et al. 2002a). Of course, larger volumes of remobilized sand associated with recent sand extrusions may occur at depth. Apart from sand volcanoes, ancient (rock record) sand extrusions are sparsely documented (e.g. Boehm & Moore 2002) although they may be commonly misidentified as primary depositional units (D.I. Macdonald 2000, pers. comm.). While the association between unstratified facies and a breached seal (Fig. 9) is compelling evidence of single dramatic seal failure, fluidized flow through the breach may be pulsed and vary in rate.
135
As noted earlier, evidence for ruptured seals is found both in shales at the tops of sand bodies and in those within sand bodies, thus creating the possibility of several seal failures, possibly at different times, and resultant sand fluidizations within a single sand body. The rock record preserves evidence of multiphase intrusion where intrusions cross cut and sills act both as feeder units for dykes and as upward terminations of dykes (Fig. 10). In this example the dykes do not appear to have been reactivated by later intrusive events, rather new dykes have formed. The last sill (a2) to be emplaced is sourced from below the sill that sourced the earlier sill (al) and an overall thickening of sills upwards is preserved. Structures inferred to have formed by fluidization of sand within depositional sand bodies (Fig. 7 and Duranti et al. 2002a) provide compelling evidence for inferring genetic relationships between these structures and sand intrusions within overlying fine grained seals (Fig. 9). It is unclear whether seal failure occurs by hydraulic fracturing (Cosgrove 2001) caused by disequilibrium compaction and build up of pressure in the sand (Osborne & Swarbrick 1997), as is likely in shallow burial intrusions, or because of other local conditions of seal failure. It is unlikely that sufficient overpressure can always be generated by disequilibrium compaction (Osborne & Swarbrick 1997) and whether an additional source of pressure is necessary, for example, gaseous hydrocarbon as in shale diapirism (Morley &Guerinl994). Differential compaction, polygonal faulting or tectonic faulting and folding may all cause or contribute to appropriate seal failure. If fractures already exist within the cap rock, as the cycle of pressure build-up commences, both the magnitude of the overpressure required to fracture the cap rock and the spatial organization of intrusions into the cap rock will be strongly influenced by the pre-existing stress field and deformational fabric. Periods of basin inversion are likely to lead to episodic breaching of cap rocks (Dore, T. pers. comm. 2000) and thus could promote sandstone intrusion from crestal positions of folded sandstone reservoirs. Wrenchrelated deformation could equally be an ideal mechanism to promote sandstone intrusion (Molyneux 2001). It is thus perhaps an oversimplification to ascribe hydraulic failure conditions as the means for predicting the occurrence and spatial organization of sandstone intrusions, as suggested by Lonergan et al (2001). Formation of polygonal faults in fine-grained strata (Cartwright 1994) is a candidate mechanism for generating fractures. Three end members of polygonal faulting are considered, where intrusions: (a) predate polygonal fault growth; (b) are synchronous with fault growth so that they are triggered by faults intersecting the base of the cap rock; or (c)
136
A.HURST£rAZ,
Fig. 11. Three possible end-member scenarios for the generation of upward sand injection. A: sand injection predates polygonal fault growth, which causes later modification of the intrusions, B: sand intrusion and polygonal fault growth are synchronous i.e. the fault growth triggers sand remobilization by causing seal failure, and C: sand intrusions postdate polygonal fault growth and exploit the mechanical anisotropy created by the faults. All three end-members can be inferred from sub-surface data and cannot be differentiated using seismic data.
post-date fault growth, exploiting the mechanical anisotropy provided by the presence of the faults (Fig. 11). End members (b) and (c) appear most appropriate as mechanisms for developing large intrusions, although small sand injections almost certainly occur in shallow burial independent of polygonal faults. These models may be used to expand upon existing models and postulate several alternative relationships between polygonal fault systems and exploitative sandstone intrusions such as those discussed from Alba Field (Lonergan & Cartwright 1999). The end members may repeat, interact or act independently, thus providing a variety of mechanisms for multi-phase clastic intrusion. Natural hydraulic fracture of cap rocks is likely to be an episodic process due to a cyclic process of pore pressure build-up within the cap rock followed by failure by hydraulic fracturing, pressure release through fluid expulsion along fractures and a repeat of the pressure build-up as the fractures close. A possible analogue for this cyclic phenomenon, which records cyclic fluid escape, is the mineral zonation in cement phases occurring along fracture networks within cap rock sequences, as observed in the pioneering study of Cathles & Smith (1983). Time scales for the mineral zonation cycles vary depending on many parameters such as lithology, burial rate and geometry, but are typically in the range of
10,000 to 500,000 years (Roberts & Nunn 1995). By analogy, it is reasonable to expect crestal intrusions of dykes and sills above sandstone reservoirs to be episodic rather than discrete events.
Conclusions Sand injectites form a genetically related group of features that includes unstratified facies and sandstone dykes and sills. All are inferred to have formed by remobilization of sand from depositional units by fluidization. The conditions of fluidization have not been constrained, but the fluidization velocity is likely to be in excess at 0.01 ms"1 and the grain content of the fluid <40%. Independent of the size and geometry of dykes and sills, they have sharp discordant boundaries and cause little deformation of adjacent fine-grained strata. The sharp boundaries, and occurrence of shale clast breccias, are inferred to reflect catastrophic failure of a seal and forceful injection of sand. Hydraulic fracturing may cause catastrophic seal failure but alternative mechanisms such as differential compaction, polygonal faulting or tectonism may be more important, or act in tandem. Downward propagation of polygonal faults is an attractive mechanism that can facilitate seal failure. Consideration of volumetrics makes liquefaction
FLUIDIZATION STRUCTURES PRODUCED BY UPWARD INJECTION OF SAND of sand an unlikely candidate for producing sufficient excess pore fluid to fluidize large volumes of sand. Hence, some small (<1 X 105 m3) sand injections developed during shallow burial may have been initiated by liquefaction, but large (>4 X 107 m3) deeper bodies, such as observed on seismic data, will not. We recognize that sealing shale horizons within sand units may be sufficiently extensive to have the potential for holding back volumes of fluidized sand of sufficient size to form large injections. The presence of several seals within, as well as above, a depositional sand body gives the possibility of several independent episodes of seal failure that may provide an explanation for the commonly observed, cross-cutting sand injections. Multi-phase intrusion may also originate because a variety of processes can elevate pore pressure and cause seal failure during burial. Processes implicated in the generation of sand injection such as disequilibrium compaction, pore fluid (including hydrocarbon gases) migration and formation of polygonal faults are cyclical, hence creating many periods in burial when sand injectites can occur. We have no evidence that the style of sand intrusion is related to the depth at which they formed; no simple sill - shallow, dyke - deep relationship is apparent. Dykes frequently emanate upward from sills, crosscut sills and reach the palaeo seafloor or earth's surface. The multiplicity of processes highlights the difficulty of constraining conditions and controls of sand injectites. Sponsors of the Injected Sands Research consortium (Chevron, Enterprise, Kerr McGee, Norsk Hydro, Shell, Statoil, TotalFinaElf) are acknowledged for their financial support and access to subsurface data. Comments from reviewers P. Cobbold and H. Johnson provided ample food for thought. Our colleagues B. Cronin, M. Huuse, R. Jonk, D. Macdonald, J. Parnell and A. Schwab are thanked for their collaboration, support and inspiration.
References BAILEY, E.B. & WEIR, J. 1932. Submarine faulting in Kimeridgian times, East Sutherland. Transactions of the Royal Society of Edinburgh, 57,429-467. BJ0RKUM, P.A., WALDERHAUG, O. & NADEAU, PH. 1998. Physical constraints on hydrocarbon leakage and trapping revisited. Petroleum Geoscience, 4, 237-239. BOEHM, A.MOORE, J.C. 2002. Fluidised sandstone intrusions as an indicator of paleostress orientation, Santa Cruz, California. Geofluids, 2,147-161. CARTWRIGHT, J.A. 1994. Episodic basin-wide fluid expulsion from geopressured shale sequences in the North Sea basin. Geology, 22,447-450. CARTWRIGHT, J.A. & LONERGAN, L. 1996. Volumetric contraction during the compaction of mudrocks: a mechanism for the development of regional-scale polygonal fault systems. Basin Research, 8,183-193.
137
CATHLES, L.M. & SMITH, A.T. 1983. Thermal constraints on Mississippi Valley Type lead zinc deposits and their implications for episodic basin dewatering and deposit genesis. Economic Geology, 78,983-1002. COSGROVE, J.W. 2001. Hydraulic fracturing during the formation and deformation of a basin: A factor in the dewatering of low-permeability sediments. American Association of Petroleum Geologists Bulletin, 85, 4, 737-748. COSGROVE, J.W. & HILLIER, R.D. 2000. Forced-fold development within Tertiary sediments of the Alba Field, UKCS: evidence of the differential compaction and post-depositional sandstone remobilisation. In: COSGROVE, J.W. & AMEEN, M.S. (eds) Forced Folds and Fractures. Geological Society, London, Special Publications, 169,61-71. DlXON R.J., SCHOFIELD K., ANDERTON R., REYNOLDS A.D.,
ALEXANDER R.W.S., WILLIAMS M.C. & DAVES K.G. 1995. Sandstone diapirism and clastic intrusion in the Tertiary sub-marine fans of the Bruce-Beryl Embayment, Quadrant 9, UKCS. In: HARTLEY AJ. & PROSSER, D.J. (eds) Characterization of Deep Marine Clastic Systems. Geological Society, London, Special Publications, 94,77-94. DOWNEY, M.W. 1984. Evaluating seals for hydrocarbon accumulations. American Association of Petroleum Geologists Bulletin, 68,1752-1763. DURANTI, D., HURST, A., HANSON, R., BELL, C. & MACLEOD, M. 2000. Reservoir characterisation of a remobilised sand-rich turbidite reservoir: the Alba Field. In: 62nd EAGE Conference & Technical Exhibition, Glasgow. Paper X5. DURANTI, D., HURST, A., BELL, C., GROVES, S. & HANSON, R. 2002a. Injected and remobilized sands from the Alba Field (Eocene, UKCS): core and wireline log characteristics. Petroleum Geoscience, 8,99-107. DURANTI, D., HUUSE, M., CARTWRIGHT, J.A. HURST, A., CRONIN, B.T., MAZZINI, A. & FLANAGAN, K. 2002& Unusual facies and geometries of the Paleogene deepwater systems in the North Sea - effects of sand remobilisation. In: 64th EAGE Conference & Technical Exhibition, Florence. Paper P057. DURANTI, D., HURST, A., HUUSE, M. & CARTWRIGHT, J.A. 2002c. Sand diapiric structures and polyphase sand remobilization in the Santa Cruz area (Central Coastal California). 64th EAGE Conference and Exhibition, Florence, Extended Abstract H-24. HILLIER, R.D. & COSGROVE, J.W. 2002. Core and seismic observations of overpressure-related deformation within Eocene sediments of the Outer Moray Firth, UKCS. Petroleum Geoscience, 8,141-149. HURST, A. & CRONIN, B.T. 2001. The origin of consolidation laminae and dish structures in some deep-water sandstones. Journal of Sedimentary Research, 71, 136-143. HUUSE, M., DURANTI, D. STEINSLAND, N., GUARGENA, C., PRAT, P., HOLM, K. CARTWRIGHT J.A. & HURST, A. in press. Seismic characteristics of large-scale remobilised and injected sand bodies in the Paleogene of the South Viking Graben (North Sea): steep-sided mounds, wings and Vs. In: DAVIES, R. etal. (eds) 3-D seismic: advances in imaging of the shallow crust. Memoir of the Geological Society of London. JENSSEN A.I., BERGSLIEN D., RYE-LARSEN M. & LINDHOLM
138
A.HURSTCTAL.
R.M. 1993. Origin of complex mound geometry of Paleocene submarine-fan sandstone reservoirs, Balder Field, Norway. In: PARKER, J. (ed.) Petroleum Geology of Northwest Europe: proceedings of the 4th Conference. Geological Society of London, 135-143. KAFUI & THORNTON 2001. Granular Dynamics Research Group website, http://civsgl.aston.ac.uk/~gdrg/ public_html/ LONERGAN, L. & CARTWRIGHT, J.A. 1999. Polygonal faults and their influence on deep-water sandstone reservoir geometries, Alba Field, United Kingdom Central North Sea. American Association of Petroleum Geologists Bulletin, 83, 3,410-432. LONERGAN, L., LEE, N., CARTWRIGHT, J.A., JOLLY, R.H. & JOHNSON, H.D. 2001. Remobilisation and injection in deep water depositional systems: Implications for reservoir architecture and prediction. In: WEIMER, P., SLATT, R.M., COLEMAN, J., ROSEN, N.C., NELSON, H., BOUMA, A.M., STYZEN, M.J. & LAWRENCE, D.T. (eds) Deep-water Reservoirs of the World; GCSSEPM Foundation 20th Annual Bob F. Perkins Research Conference, 515-532. LOWE. D.R. 1975. Water escape in coarse-grained sediments. Sedimentology, 22,157-204. MACDONALD, A.C. & TREWIN, N.H. 1993. The Upper Jurassic of the Helmsdale area. In: TREWIN, N.H. & HURST, A. (eds) Geology of East Sutherland and Caithness. Geological Society of Aberdeen, Scottish Academic Press, 75-114. MACLEOD M.K, HANSON R.A., BELL C.R. & McHuoo S. 1999. The Alba Field ocean bottom cable seismic survey: Impact on development. The Leading Edge, 18, November, 1306-1312. MALTMAN, A. 1994. Prelithification deformation. //I.HANCOCK, P.L. (ed.) Continental Deformation. Permagon Press, Oxford, 143-158. MOLYNEUX, S. 2001. Sandstone remobilisation in the Eocene to Mio-Pliocene of the Central and Northern North Sea. PhD Imperial College (unpublished). MORLEY, C.K. & GUERIN, G. 1994. Comparison of gravitydriven deformation styles and behaviour associated with mobile shales and salt. Tectonics, 15, 1154-1170. MURCHISON, R.I. 1827. Supplementary remarks on the Oolitic Series in the Counties of Sutherland and Ross, and in the Hebrides. Transactions of the Geological Society, 2, 353. NEWMAN, M.ST. J., REEDER, M.L., WOODRUFF, A.H.W. & HATTON, I.R. 1993. The geology of the Gryphon oilfield. In: PARKER, J.R. (ed.) Petroleum Geology of Northwest Europe, Proceedings of the 4th Conference, Geological Society of London, 123-133. NEWTON, S.K. & FLANAGAN, K.P. 1993. The Alba Field: evolution of the depositional model. In: PARKER, J.R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference, The Geological Society, London, 161-171. NICHOLS, R., SPARKS, S. & WILSON, C.J.N. 1994. Experimental studies of the fluidisation of layered sediments and the formation of fluid escape structures. Sedimentology, 41,233-253. OBERMEIER, S. 1989. The New Madrid earthquakes: an
engineering-geologic interpretation of relict liquefaction features. US Geological Survey Professional Paper1336-B,114p. OBERMEIER, S. 1996. Use of liquefaction-induced features for paleoseismic analysis - an overview of how seismic liquefaction features can be distinguished from other features and how their regional response distribution and properties of source sediment can be used to infer the location and strength of Holocene paleo-earthquakes. Engineering Geology, 44, 1-76. OSBORNE, M.J. & SWARBRICK, R.E. 1997. Mechanisms for generating overpressure in sedimentary basins: a reevaluation. American Association of Petroleum Geologists Bulletin, 81,1023-1041. OWEN, G, 1996. Experimental soft-sediment deformation: structures formed by the liquefaction of unconsolidated sands and some ancient examples. Sedimentology, 43,279-293. PARIZE, O. 1988. Sills et dykes greseux sedimentaires: paleomorphologie, fracturation precoce injection et compaction. These Doct. Geol. ENSMP. Universite de Lille. 333p. PARIZE, O. 2001. Sand injections in deep marine shale formations in SE France. PARIZE, O., VERSCHUREN, M. & VAN RENSBERGEN, P. (eds) Subsurface Sediment Mobilization, Field Guide Book. Association des Sedimentologues Francois. 35. PURVIS, K., KAO, J., FLANAGAN, K., HENDERSON, J. & DURANTI, D. 2002. Complex reservoir geometries in a deep-water clastic sequence, Gryphon Field UKCS: injection structures, geological modelling and reservoir simulation. Marine and Petroleum Geology, 19, 161-179. ROBERTS. SJ. & NUNN, J.A. 1995. Episodic fluid expulsion from geopressured sediments. Marine and Petroleum Geology, 12, 195-204. SMYERS & PETERSON, 1971. Sandstone Dikes and Sills in the Moreno Shale, Panoche Hills, California. Geological Society of America Bulletin, 82, 3201-3208. STEPHEN, K.D., CLARK, J.D. & GARDINER, A.R. 2001. Outcrop-based stochastic modelling of turbiditic amalgamation and its effects on hydrocarbon recovery. Petroleum Geoscience, 7, 163-172. SURLYK, F. & NOE-NYGAARD, N. 2001. Sand remobilisation and intrusion in the Upper Jurassic Hareelv Formation of East Greenland. Bulletin Geological Society of Denmark, 48, 169-188. THOMPSON, B.J., GARRISON, R.E. & MOORE, C.J. 1999. A late Cenozoic sandstone intrusion west of Santa Cruz, California: fluidised flow of water and hydrocarbonsaturated sediments. In: GARRISON, R.E., AIELLO, I.W. & MOORE, C.J. (eds) Late Cenozoic fluid seeps and tectonics along the San Gregorio fault zone in the Monterey Bay region, California. American Association of Petroleum Geologists Pacific Section, Volume and Guidebook, GB-76, 53-74. WEBER, K.J. 1986. How heterogeneity affects oil recovery. In: LAKE, L.W. & CARROLL, JR. H.B. (eds) Reservoir Characterisation. Academic Press, 487-544. WILSON C.J.N. 1980. The role of fluidization in the emplacement of pyroclastic flows: an experimental approach. Journal Volcanic and Geothermal Research, 20, 231-249.
Gas and fluid injection triggering shallow mud mobilization in the Hordaland Group, North Sea HELGE L0SETH, LARS WENSAAS, B0RGE ARNTSEN & MARTIN HOVLAND Statoil Research Centre, N-7005 Trondheim, Norway (e-mail: heloe @ statoil. com) Abstract: During a regional seismic interpretation study of leakage anomalies in the northern North Sea, mounds and zones with a highly chaotic seismic reflection pattern in the Tertiary Hordaland Group were repeatedly observed located above gas chimneys in the Cretaceous succession. The chaotic seismic reflection pattern was interpreted as mobilized sediments. These mud diapirs are large and massive, the largest being 100 km long and 40 km wide. Vertical injections of gas, oil and formation water are interpreted to have triggered the diapirs. On the eastern side of the Viking Graben, another much smaller type of mud diapir was observed. These near-circular mud diapirs are typically 1-3 km in diameter in the horizontal plane. Limited fluid injection from intra-Hordaland Group sands, through sand injection zones, into the upper Hordaland Group is interpreted to have triggered the near-circular diapirs. This observed 'external' type of mobilization was generated at shallow burial (<1000 m) and should be discriminated from the more common 'internal' type of mud diapirism that is generated in deep basins (>3000 m). The suggested model has implications for the understanding of the palaeofluid system, sand distribution, stratigraphic prediction within the chaotic zone, seismic imaging, and seismic interpretation of the hydrocarbon 'plumbing' system.
The northern North Sea sedimentary basin is one of the world's major offshore hydrocarbon provinces. During a seismic interpretation study of leakage anomalies in the northern North Sea, massive zones with a chaotic reflection character, with mounded tops, were repeatedly found located in the Tertiary Hordaland Group, vertically above gas chimneys in the Cretaceous succession. They have large lateral extent, the largest being 100 km long and 40 km wide. Another much smaller type of the chaotic reflection zone, which is 1-3 km wide, was later observed during a 3D seismic interpretation study east of the Viking Graben. These small chaotic zones were located above high amplitude V-shaped brights (V-brights), but have no underlying gas chimneys. The chaotic zones are all located in the upper part of the Eocene to Miocene Hordaland Group (Isaksen & Tonstad 1989; Jordt 1996, Jordt et al 2000). The objective of this paper is to describe and interpret the chaotic reflection patterns and associated anomalies and discuss new models and implications for the observations. The study area is approximately 50 000 km2 and is located between 58°45'N and 62°N (Fig. 1).
Geological setting During the early Tertiary, the Shetland Platform and mainland Norway became uplifted and formed important clastic source areas for the North Sea (Pegrum & Spencer 1990; Jordt et al 2000). Simultaneously, the Viking Graben and its flanks
subsided to form a sedimentary depocentre where the sediments of the Rogaland, Hordaland and Nordland groups were deposited (Isaksen & Tonstad 1989; Jordt 1996; Jordt et al 2000) (Fig. 2). The early Tertiary subsidence of the North Sea basin and uplift of the flanks most likely occurred in response to the increased extensional stress in the lithosphere that resulted in the opening of the North Atlantic Ocean during earliest Eocene (Skogseid et al 2000). Vast amounts of volcanic material were extruded and the tuffs of the Balder Formation were deposited throughout the North Atlantic. Parts of the land-laid ash-deposits and volcanic rocks were later eroded and transported into the marine North Sea basin and contributed significantly to the Eocene, Oligocene and early Miocene deposits of the Hordaland Group (Tyberg et al 2000). The large amounts of smectite in these clays are mainly sourced from the volcanically related rocks. Deep marine clastic fans were occasionally deposited throughout parts of the basin during Paleocene to Oligocene times (Pegrum & Spencer 1990; Jordt et al 2000). The top of the Hordaland Group defines a regional unconformity in the study area (Jordt 1996; Jordt et al 2000). The overlying Utsira Formation (Gregersen et al 1997) has recently been redated as early Pliocene (Piasecki et al 2002). Glaciomarine sediments were deposited above the Utsira Formation. Parts of the Hordaland Group sediments were modified syn and post deposition. Three types of post depositional alteration of the sediments are commonly observed in the Hordaland Group: (1) polygonal faulting (Cartwright 1994a, 1994£;
Fig. 1. Location map for the study area in the northern North Sea. The seismic database comprises an open 2D seismic grid between 60°N and 62°N and minor areas with 3D data coverage. The locations of the figures are indicated.
GAS AND FLUID INJECTION TRIGGERING SHALLOW MUD MOBILIZATION
141
suggest that fluid and gas injection triggered the mobilization of the upper parts of the Hordaland Group, much in the same way as has been suggested by Hoviand et al. (1998), concerning ooze diapirism on the V0ring Plateau.
Characteristics of the Hordaland Group sediments
Fig. 2. Simplified Cenozoic stratigraphy in the Norwegian North Sea.
Cartwright 1996; Cartwright & Lonergan 1996; Lonerganera/. 1998;Dewhurst^a/. 1999); (2) sand injection (Larsen 1994; Candace et al 1995; Dixon et al 1995; Nicholls 1995; Lonergan & Cartwright 1999; Cosgrove & Hillier 2000) and; (3) sediment mobilization. The first two types are well described in the literature. The focus of this paper is on the seismic features interpreted as mobilized sediments. The mobilized succession in the Hordaland Group can easily be recognized in seismic data and three different formation models have been suggested. From a study confined to Block 35/8, Jordt (1996) proposed that rapidly deposited Miocene sands caused the deformation and diapirism in the middleupper Oligocene clays. Rundberg & Nystuen (pers. comm. 1999) suggested that the entire Oligocene section slid on the top Eocene surface causing compressional folding and large-scale slumping, sliding and soft sediment deformation. In this paper we
The Hordaland Group represents a more or less continuous succession of fine-grained mudstones and interbedded sandstones deposited during the Eocene and Oligocene. It overlies the clays and turbiditic Paleocene sands without any stratigraphic breaks and is separated from the overlying Nordland Group by a regional unconformity (Isaksen & Tonstad 1989; Jordt 1996; Jordt et al 2000; Eidvin et al 2000; Piasecki et al 2002). The smectite-rich mudrocks in the Hordaland Group is characterized by small grain size, high porosity, low permeability and ductile behaviour (Aplin et al 1995; Wensaas et al 1998)(Fig. 3). The mechanical strength of montmorillonite (smectite) is about half the strength of other clays at a given stress level (Wang et al 1980). Because of its high porosity (high water content), distinct velocity and density inversions are observed across the boundary between the Nordland and the Hordaland groups (Fig. 3). The fine-grained nature and low permeability of the Hordaland Group mudrocks inhibits drainage of pore fluid during burial and consequently the mudrocks appear undercompacted. Hence, the weak behaviour of non-cemented highly smectitic Hordaland Group mudrocks is likely to be maintained during burial with a high potential for remobilization by ductile deformation following changes in the stress level. The rates of fluid expulsion in mudrocks are controlled by the rate of compaction. The compressibility of mudrocks is also highly dependent upon their lithology and composition (Aplin et al 1995). At depths below about 2 km (70°C), the rock strength will gradually increase by mineral reactions (diagenesis) (Bj0rlykke 1998).
Seismic expression and interpretation of abnormal features in the Hordaland Group The smectite-rich sediments unit in the Hordaland Group show an anomalous internal reflection pattern and also an abnormal upper surface morphology throughout large parts of the northern North Sea. The contrast between the abnormal seismic features in the Hordaland Group and neighbouring units is observed on seismic data by the character of the acoustic reflection. Outside the anomalous zone the reflection pattern laterally is often subparallel and
142
R.L0SETHETAL
Fig. 3. Lithostratigraphy, wire line logs and mudstone mineralogy obtained by X-ray diffraction analyses (XRD) in Well 34/10-2 (Gullfaks South). The Hordaland Group mudstones are characterized by high clay content (i.e. fine grain size) and high concentrations of smectite (SM) and mixed layer clay minerals (Sm/Ill). The petrophysical logs and laboratory measurements indicate that these mudstones have high porosity (low bulk density and acoustic velocity) and low permeabilities.
discontinuous on a small scale due to polygonal faulting (Fig. 4). On a larger scale, the reflection pattern is layered and quite coherent but with a wavy appearance, due to minor offsets caused by polygonal faults. The top of the Hordaland Group is characterized by a smooth, relatively high amplitude and continuous reflection. In contrast, the seismic expression of the mobilized unit is characterized by (Fig. 4): (1) (2) (3) (4) (5) (6)
chaotic internal reflection pattern; low amplitude V-shaped reflections within the chaotic reflection zone; listric normal faults at the flanks of the chaotic zone; mounds at the top of the Hordaland Group above zones with the chaotic reflection pattern; limited thickness of the zones with chaotic reflection pattern and; often high amplitude V-shaped brights (Vbrights) below the base of the chaotic zone.
The characteristics of these anomalies are described below.
Chaotic reflection zones. The chaotic internal reflection zones comprise short, disconnected internal reflections and low amplitude V-shaped reflections (Fig. 4). The internal disconnected reflections can only be traced for short distances, normally significantly less than one km. These reflection segments dip in a non-systematic way in any direction and at variable angles. The disconnected reflection segments are inteipreted as remnants of the primary bedding that has been broken and rotated to variable degree. The chaotic reflection pattern is interpreted to reflect zones of mobilized sediments. Low amplitude V-shaped anomalies. Low amplitude V-shaped anomalies commonly occur within the chaotic reflection zone. These V-shaped reflections appear in the same position on repeated seismic surveys [time lapse data (4D)] above the Gullfaks Field (Fig. 5), indicating that they reflect real underground events. The low amplitude V-shaped reflectors are interpreted as point refractions, either from the edge of vertical beds or from segments of beds that are so small that they acts as seismic refraction points.
GAS AND FLUID INJECTION TRIGGERING SHALLOW MUD MOBILIZATION
143
Fig. 4. A seismic time section from northern part of the Norwegian Block 30/9 that illustrates the chaotic reflection pattern with V-shaped amplitudes and mounds at the top of the Hordaland Group. Note the rim zone with significantly rotated beds on the left-hand side of the figure. A vertical noise zone, which is interpreted as a gas chimney, is located below the chaotic zone. A correlation map demonstrates that the gas chimney, which is a low correlation area expressed with dark colour, has a circular shape while the fault is linear in map view.
Rim zone. The lateral transition boundary from the chaotic reflection pattern often runs through a rim zone with listric normal faults that soles out within the Hordaland Group (Fig. 4). There is a gradual transition from these faults to the pure polygonal faulted succession that reflects the non-mobilized units. The beds in the transition zones can be separated from polygonal faulted parts by an increase in both fault throws and dips of bedding. Reverse faults have not been observed in the transition zone. Mounds. The top of the chaotic reflection zone is characterized by a combination of irregular mounds and bowl-shaped depressions at the top of the Hordaland Group (Fig. 5; Fig. 6). The top of the mounds can reach up to 2-300 m above the regional mean level of the top Hordaland Group. The top is interpreted as a palaeo-surf ace that was formed when the top of the chaotic zone was at the free surface. The bowl-shaped depression demonstrated that the shape of the top of the Hordaland Group was not formed by erosion processes because erosion does not develop irregular bowl-shaped depressions.
Thickness. The top of the chaotic reflection pattern coincides with the top of the Hordaland Group, while the base can vary from a relatively well-defined event, which is parallel to the underlying reflection, to an undefined blurred base (Fig. 4). The observed chaotic reflection pattern is always located in the uppermost 300-700 m of the Hordaland Group, but the thicknesses of chaotic reflection zones vary from place to place. Reprocessing has improved the imaging of the base from diffuse to quite well defined. Seismic depiction problems may wrongly result in an interpretation indicating an irregular base. V-shaped brights. High amplitude V-shaped brights (V-brights) are repeatedly observed either close to the base or below the base of the chaotic reflection pattern. One type of the V-brights (Fig. 7) is located just above the top of the tuffs of the Balder Formation (Malm et al 1984). These V-brights occur in clusters (Fig. 8) that are repeatedly observed above noise-filled seismic data that are located in the underlying Cretaceous succession (Fig. 9). Wells penetrating the noisy zones have
144
H.L0SETH£rAL.
Fig. 5.4D seismic time data from Gullfaks Field from 1985 (upper) and 1999 (lower). The V-shaped reflections appear in the same positions on repeated seismic surveys indicating that they reflect real events within the chaotic reflection zone. The V-shaped reflections are interpreted as point refractions, either from the edge of vertical beds or from segments of beds that are so small that they acts as seismic refraction points. Note the local infilling in the depression at top Hordaland Group just above the arrow head. Location indicated in Fig. 6.
increased mud gas readings and an increase in heavier hydrocarbons (only hydrocarbons up to C5 are recorded by mud gas readings) relative to wells drilled outside the noisy zones (in-house data). These noise zones are interpreted as gas chimneys. The rock type that corresponds to these top Balder Formation V-brights is not yet known. Both injected tuffs and/or cementation along fluid escape routes are possible explanations for the anomalies. It is not likely that these V-brights represent sand injections, as they are located in areas where sand is not present at these stratigraphic levels. The close connection to the underlying gas chimneys suggests that these Vbrights are related to vertical fluid movements from underlying Jurassic structures. The V-brights are interpreted to reflect areas of focused gas, oil and/or formation water flow through the Balder Formation. The distribution of the chaotic reflection pattern is mapped based on an open 2D grid located between 60° and 62°N (Fig. 10). The largest massive mud diapir zone, which is approximately 100 km long and 40 km wide, has been mapped above the North
Viking Graben and its flanks between 60°15'N to 61°30'N. The anomalies terminate to the west along a NNW-SSE striking line, that cross over the Viking Graben between 60°30'N and 60°N (Fig. 10). The anomalies are not observed above the Viking Graben just north of 60°N. Nor are the anomalies present along a NW-SE striking belt from the northern Lomre Terrace to the northern Tampen Spur. There is a marked shift in boundary style from west to east. The western boundary defines a relative straight line, while the eastern boundary of the chaotic reflection zone has an irregular shape that was difficult to map based on the open 2D grid. Minor chaotic reflection zones, which are more or less continuous, are observed along the eastern boundary of the mobilized zone (Fig. 4).
Near-circular structures One special type of the chaotic reflection zones occurs as minor near-circular structures in the hori-
GAS AND FLUID INJECTION TRIGGERING SHALLOW MUD MOBILIZATION
145
Fig. 6. Depth-structure map at top Hordaland Group above the Gullfaks Field. Irregular shaped mounds (H) and irregular bowls-shaped depression (L) are observed. The top is interpreted as a palaeo-surface that was formed by mobilized sediments.
zontal plane (Fig. 11; Fig. 12). The structures have been studied in detail on 3D seismic data in Norwegian blocks 25/6, 16/2 and 16/3. The anomalies are located in the upper part of the Hordaland Group. The zones with chaotic reflection patterns are typically between 200 and 400 m thick. The chaotic reflection zones comprise fewer and lower amplitude reflections than the laterally surrounding beds and the lateral extent and distribution can therefore be mapped with RMS-amplitude attribute maps (Fig. 12). The RMS-map demonstrates that the low reflective zones form 1-2 km wide near-circular structures in the horizontal plane in the blocks 16/2 and 16/3. In Block 25/6, the structures are slightly wider, up to 3 km wide. In Block 16/3 the circular structures terminate to the east along an irregular NNE-SSW striking boundary (Fig. 12). The succession laterally outside the chaotic reflection is perforated by polygonal faults that cut into the circular zone at various angles (Fig. 12). The lack of relation between the strike of the polygonal
faults and the circular structures is interpreted to indicate that mobilization took place in sediments where polygonal faults had already developed and that the polygonal faults had little influence on the formation of the chaotic reflection zone. Detailed automatic velocity analysis in block 25/6 demonstrates that the zones with chaotic reflection have lower velocities (2100 m/s) than the laterally surrounding succession (2250 m/s) rocks. Reprocessing with such detailed velocity data has improved the data quality both within and below the chaotic reflection zones. V-brights are almost always observed below the near-circular structures (Fig. 11). They are located in the middle part of the Hordaland Group. Vertical noise zones and minor pull-up are often observed below the V-brights. Well 24/12-1 penetrated one of these V-brights that proved to be a sandy interval that was partly carbonate cemented (Fig. 13). The carbonate cemented sand beds have a velocity of almost 4500 m/s, compared to the background sand
146
H.L0SETH£TAL.
Fig. 7. Seismic time section showing V-shaped high amplitude reflections (V-brights) that are located above the top Balder Formation on the SE flank of the Gullfaks Field. Location indicated in Fig. 8.
Fig. 8. Reflection Intensity (RI) amplitude map from a 200 ms interval above the top Balder reflector. The Rl-map demonstrates that the V-brights (red areas) are distributed in patches.
GAS AND FLUID INJECTION TRIGGERING SHALLOW MUD MOBILIZATION
147
Fig. 9. A seismic time section from the Gullfaks South Field shows that zones with chaotic reflections in the Hordaland Group are located above V-brights at top Balder level, which again are located above gas chimneys that are rooted in the underlying Jurassic rotated fault block.
and clay velocity of 2200 m/s (Fig. 14). The high velocity layers have limited lateral extent, often only a few hundred metres. The high velocity caused critical refraction to occur at relatively small offsets (Fig. 15). The mute function is normally not constructed to avoid this problem and the refracted waves are therefore included in the full-stacked data. In reality these layers act as diffraction points, setting up diffracted waves with relatively large amplitudes. During processing, diffractions are removed in the migration step, but if a non-optimal velocity function is used, not all diffractions may be properly collapsed. A residual appearing as a bright V-shaped reflector in the seismic sections may be the result. However, judging from the near trace reflection on pre-stack data, some of these carbonate cemented sands may also have a genuine Vshape(Fig. 15). The vertical noise zones that appear below Vbrights (Fig. 13) have the same seismic expression as gas chimneys. Well 25/12-1, which drilled through such a noise zone, had no indication of increased mud gas and consequently ruled out that gas caused the vertical noise. The disturbances appearing below
these V-brights are probably due to the ray bending, scattering caused by the high velocity contrasts and strong multiples. The V-brights that appear on full stack migrated data have been found to represent a combination of reflections from real V-shaped underground events and processing artefacts. The real events are interpreted as partly carbonate cemented sand injection structures. The near-circular mounded structures are observed in Norwegian blocks 25/6,16/2 and 16/3 and in the SE side of the Viking Graben within the area mapped as partly mobilized sediments (Fig. 10).
Discussion Generally, the driving force for vertical movements of mud is in most cases considered to be buoyant forces resulting from the density contrast between the mobile low density and the immobile higher density sediments. The Hordaland Group was mobilized at a relatively shallow burial depth (<1000 m) and it was the uppermost succession that was
148
H.L0SETH£rAL.
Fig. 10. Approximate location of massive mud mobilization and partly mobilized mud in the Hordaland Group superimposed on a structural map. The mapping is based on a 2D seismic grid. Where both mound shapes and internal chaotic reflection pattern are observed, the anomalies are interpreted as massive mud diapirs. In areas where only one of these characteristics is observed, where they are poorly developed or have limited lateral extension (<5 km), they are mapped as partly mobilized mud. The small near-circular mounds are incoiporated in the partly mobilized mud.
GAS AND FLUID INJECTION TRIGGERING SHALLOW MUD MOBILIZATION
149
Fig. 11. Seismic time section through a chaotic reflection zone with limited extent, mound at the top and V-bright below. The chaotic zone has a near-circular shape in the vertical plane and is from 1 to 3 km wide. It is interpreted as a mud diapir. The location of the seismic line is shown on the RMS-attribute map and in Fig. 12.
mobilized when it was close to or at the palaeo surface. Internal generation of fluids by thermal processes is not likely to occur at such low temperatures, so thermal driven processes cannot have caused the mobilization. The top of the Hordaland Group is represented by an unconformity with an associated time gap, hence rapid subsidence is not a likely model for diapir formation. Nor is it likely that any type of slumping mechanism formed the mud diapirs because such process cannot generate the observed near-circular mud diapirs. A slumping mechanism would also generate a frontal trust zone that is not observed. Injection of gas, oil and formation water is suggested to be the main triggering mechanism, as discussed below.
Fluid movements The Eocene and Oligocene deposition in the Viking Graben caused temperatures to rise in the Jurassic succession, which again caused a thermal porosity reduction reaction in the reservoirs (Bj0rkum & Nadau 1998) in the deep Viking Graben. The porosity reduction led to a pressure build-up in compartmentalized reservoirs. Basin modelling shows that
the Oligocene was a time of massive gas expulsion from the Upper Jurassic source rock in the deep Viking Graben (Sylta & Krogstad 1992; Johannesen et al. 2002). The pressure increase caused by thermal processes and hydrocarbon generation most likely resulted in hydro fracturing of the top seal and vertical leakage. Large amounts of formation water, oil and gas could thus move vertically through the hydro fractured fine-grained rocks as long as the high pressure could open the hydro-fractures (Mandl & Harkness 1987). Gas chimneys are interpreted as the seismic expressions of zones that have had, or still have focused vertically flow of gas, oil and formation water through fine-grained rock. During a general mapping of leakage anomalies between 60° and 62°N it was repeatedly observed that the zones of massive chaotic reflection pattern were located above V-brights at the top of the Balder Formation, which again were located above gas chimneys in the Cretaceous succession (Fig. 9). The gas chimneys were rooted at the crest of underlying Jurassic fault blocks, e.g. Gullfaks, Gullfaks South and Kvitebj0rn fields. This pattern did not prevail in the S W part of the Viking Graben just north of 60°N. Here the gas chimneys (e.g. the gas chimney above the Hild Field) did not have overlying V-brights and
150
H.L0SETHCTAL.
Fig. 12. RMS-amplitude map in the lower part of the chaotic reflection zone (70 ms above and 30 ms below the red reflector in Fig. 11). These chaotic zones comprise fewer and lower amplitude reflections than the lateral surrounding beds and appear as low reflective 1-2 km wide near-circular structures in the horizontal plane in blocks 16/2 and 16/3. The near-circular structures terminate along an irregular eastern boundary. Note that polygonal faults also have developed east of the boundary. The high amplitude V-brights below the chaotic zone is expressed with red colour. Vbrights are observed below almost all of the chaotic reflection zones.
chaotic reflection zone. The Hild Field observations appeared to be in conflict with a model of gas chimneys as zones with vertically focused gas, oil and formation water flow that triggered mobilization when injected to the Hordaland Group. However, the eastern limit of the Paleocene sand distribution in the North Viking Graben (Pegrum & Spencer 1990) roughly coincides with the western limit of the mobilized parts of the Hordaland Group (Fig. 10). Thus, gas and fluids that leaked vertically through the Cretaceous gas chimneys could effectively be 'gathered' and transported laterally within the Paleocene/Eocene 'thief sand. Hence, our interpretation suggests that fluids and gas did not reach the Hordaland Group in the SW part of the Viking Graben, which subsequently did not mobilize. A similar model may explain the absence of mobilization in a WNW-ESE belt, from the northern Lomre Terrace to the Tampen Spur (Fig. 10). This area may coincide with a belt of Cretaceous and Paleocene
sands that at least partly have their provenance on the Norwegian mainland (Jordt et al 2000). Thus, the absence of mobilized sediments above gas chimneys may indicate the presence of laterally continuous 'thief sands. Another type of gas and fluid injection model was required to explain the 1-3 km wide near circular mud diapirs. The absence of gas chimneys below these structures, which is the opposite of what was found for the large massive mud diapirs, is an important observation because it suggests that direct gas injection from Paleocene or deeper levels cannot be responsible for their formation. The near-circular shapes (Fig. 12) suggest that some kind of point sources must have created the structures. High amplitude V-brights are located below almost all the near-circular mud diapirs (Figs 11, 12, 13), indicating that they are genetically linked. The sand injection structures, which are the interpretation of this type of V-brights, may represent point-sourced
GAS AND FLUID INJECTION TRIGGERING SHALLOW MUD MOBILIZATION
151
Fig. 13. Near trace seismic time section at the location, e.g. Well 25/12-1. The well drilled though a mud diapir with an underlying V-bright. A vertical noise zone appears below the V-bright but a mud gas anomaly was not recorded associated with this noise zone. Thus, the noise zone cannot be interpreted as a gas chimney. injections. Thus, it appears that pressure built up in the intra-Hordaland Group sand and caused fluidization and sand injection into the overlying sequence. This pressure build-up could occur because the intra-Hordaland Group sands pinch out up-dip to the east and hence represent a closed fluid system (Fig. 16). Formation water, and possibly hydrocarbons were injected into the overlying, shallow buried upper parts of the Hordaland Group, which responded with mobilization and formation of the near-circular mud diapirs. The lack of mud diapirs in the eastern part of Block 16/3 (Fig. 12) may be related to the lack of a permeable layer to transport the fluids laterally, if the intra-Hordaland Group sands, which were sourced from the Shetland Platform to the west, wedge-out to the east. The eastern boundary of the near-circular mud diapirs may therefore correspond to the eastern pinch-out of Grid/Skade sand. Polygonal faults, which typically develop in fine-grained mud (Dewhurst et al 1999), are indeed also present east of the boundary (Fig. 12). Thus, the boundary cannot be explained by the absence of mobilizable sediments. Nor are there any significant thickness variations of the mobilized interval across the boundary that could explain the transition.
Sediment mobilization The key characteristic of the mobilized clay in the Hordaland Group is the significant amount of finegrained clays that are reflected by the high smectite concentrations (Fig. 3). The smectite enables large amounts of water to be trapped in the clay, which thus maintains low density and low shear strength. Polygonal faults typically develop in such finegrained sediments (Dewhurst et al. 1999). When the sediments were mobilized at shallow burial depths they had very high water content and consequently very low shear strength. The low permeable finegrained mud retained pore fluids. As fluids and gas intruded into these sediments, they responded by deforming and mixing to generate a low-density sediment. The low-density material moved vertically upwards in a convective fashion. This vertical movement of muddy material broke the primary layering into larger and smaller segments with preserved primary bedding, which are floating in the mixed up matrix (Fig. 4). At deeper burial and higher consolidation levels, the same clay would most likely respond to high-pressure injection of gas, oil and formation water, by hydro-fracturing. This may explain why the sediments do not continue to
152
H.L0SETHCTAL.
Fig. 14. Gamma ray, P-wave velocity logs and seismic tie to Well 25/12-1. The V-brights are tied to the top of the two carbonate cemented intervals in the sands, which has a velocity twice as great as the background sand velocity.
respond to fluid injection by mobilization at the present day. Generally, it is likely that sediments where polygonal faults develop also have the property to mobilize in response to fluid injection. Pervasive polygonal faulting was also observed in the lateral surrounding oozy sediments for the shallow mud diapirs on the present day seafloor in the V0ring Basin (Hovland et al 1998). Suggested model We propose the following model of formation for the massive mud diapirs (Fig. 16). Overpressure was generated in the deep Viking Graben due to chemical compaction and hydrocarbon generation. The hydrocarbons were trapped in Jurassic structures and the seals were hydraulically
fractured. Gas, oil and formation water leaked vertically through the Cretaceous succession in areas that now can be seismically mapped as gas chimneys. In the absence of Paleocene 'thief sands, the fluids and gas penetrated the Balder Formation tuffs in discrete point locations reflected as clusters of V-brights, continued upwards and were injected into the unconsolidated smectite-rich sediments in the upper part of the Hordaland Group. The sediments responded by deformation and mixing to form a low-density mixture that moved vertically and built up the diapirs. At the palaeo-surface, relatively large mounds and bowl-like depressions developed. The vertical mass movement in the centre of the structure resulted in lateral mass flow from the surrounding contracting areas. The resulting thinning in the upper part of the Hordaland Group is expressed as listric normal faults at the flanks of the mobilized succession. The largest mud diapirs were formed in
GAS AND FLUID INJECTION TRIGGERING SHALLOW MUD MOBILIZATION
153
Fig. 15. Pre-stack data through a V-bright. Note that the near trace data indicate that the V-bright has a genuine V-shape because the event is located gradually higher towards the flanks. The high velocity makes the reflection from the Vbright go critical at relatively small offset angles. Parts of the refracted wave are therefore included in the full stack data.
Fig. 16. Schematic diagram illustrating how the mobilized sediments reflect parts of the basin 'plumbing' system. The mobilization reflects two fluid systems, one sourced from Jurassic structures in the Viking Graben and a smaller eastern system sourced through intra Hordaland Group sands.
154
H.L0SETH#rAL.
the area of the most extensive fluid and gas flow, located above the Viking Graben and flank structures draining the deeper Viking Graben. The fluids and gas were transported laterally in areas where continuous Cretaceous or Paleocene 'thief sands overlie the gas chimneys. Here the Hordaland Group sediments were not mobilized (Fig. 16). The 1-3 km wide near-circular mud diapirs were formed by limited fluid injection from intraHordaland Group sands. These sands pinch out updip to the east and are a closed permeable system that could become overpressured. The fluids and possibly hydrocarbons were transported vertically through underlying sand injection structures, which are expressed as V-brights in seismic data. The injected sands represent point injectors that mobilized the sediments in the near-circular mud diapirs. The absence of near-circular clay diapirs in the eastern part of the Norwegian Block 16/3 can be explained by the pinch-out of lateral sand conduits (Fig. 16). Relation to other type of mud diapirism Mud diapirs are often associated with thick deltaic successions, which are activated as higher density silts and sands are deposited over the pro-delta facies in actively prograding systems (Elliot 1986; Miall 1985). Generally, at high deposition rates overpressures develop in fine-grained sediments. The clay may start to flow to form clay diapirs in response to the pressure build-up. The internal structures are well described by Morley et al (1998). Such mud diapirs (Van Rensbergen et al 1999) are generated in the deeper part of the basin, most often below 3 km and in deltaic settings. This type of mud diapirism is termed 'internal' as it is the internal overpressure processes in the sediments, which causes the pressure build-up that generate mobilized mud. The mud diapirs described herein from the North Sea represent another end-member. This type is generated at shallow burial (<1000 m) by the injection of gas and/or fluids into a potentially unstable sediment unit. It is termed 'external' because an external source of fluids and gas is responsible for the mobilization. This subdivision is a new classification of mud diapirs (Fig. 17). V-shaped reflections Three separate types of V-shaped reflections have been discussed in this paper. (1) The V-brights below the near-circular mud diapirs are interpreted as partly carbonate cemented sand injection structures. These V-brights, which appear on full stack migrated data, have been found to represent a combination of
Fig. 17. Schematic figure illustrating the difference between the external mud diapirs described herein and the more commonly described internal mud mobilization. The external type, which is observed in the upper parts of the Hordaland Group in the North Sea, developed at shallow burial (< 1000 m) by the injection of gas and/or fluids from an external source. The internal form of mud diapirism is generated in deep basins (>3000 m) by internal overpressure generation mechanisms.
reflections from real V-shaped subsurface events and processing artefacts (Figs 13, 14 and 15). (2) Clusters of V-brights above the Balder Formation tuffs are observed above gas chimneys (Figs 7, 8 and 9). These V-brights are interpreted to reflect areas of focused gas, oil and/or formation water flow through the Balder Formation. Even though the rock type, which corresponds to these top Balder Formation Vbrights, is not yet known, both injected tuffs and/or cementation along fluid escape routes are possible explanations for the anomalies. The extent to which these are real events or processing artefacts, has not been evaluated for these V-shaped events, but it is likely that the real events are partly hampered by imaging problems. (3) Low amplitude V-shaped anomalies commonly occur within the mobilized sediments (Fig. 5). They are interpreted as point refractions, either from the edge of vertical beds or
GAS AND FLUID INJECTION TRIGGERING SHALLOW MUD MOBILIZATION
from segments of beds that are so small that they acts as seismic refraction points. Unpublished in-house data show that the V-shape is mainly a geophysical processing artefact. In the North Sea, V-shaped reflections are relatively commonly observed on seismic data. Some of the V-shaped reflections are good images of the real events, e.g., erosional channels and sand injections (Lonergan & Cartwright 1999; Cosgrove & Hillier 2000). This study indicates that there exist significant image problems related to some events in the subsurface and that the V-shape at least partly can be processing artefact. A detailed geophysical study is in many cases required to work out the shape of the events that are expressed as V-shaped reflection on processed seismic data.
155
distribution of submarine sediment. Hence, the relief formed on the mounded surface above the diapirs may have guided the distribution of the immediately overlying Utsira Formation (Fig. 5). Seismic imaging. Mobilization has also modified the structure of the sediments from a layered, well-stratified succession to a complexly disrupted succession with disorganized remnants of the original bedded succession. The disruption can explain the observed low velocity zone within the mobilized succession. The identification of significant lateral velocity changes in the Hordaland Group has been shown to greatly improve seismic imaging both within and below the mobilized zone. Proper detailed velocity analyses should therefore be carried out in areas with lateral variability in mud mobilization, e.g., areas with near-circular clay diapirs.
Implications for the oil industry Palaeo-fluid system and underlying sand distribution. The most important implication of the suggested model is that the mud diapirs can be used to interpret the North Sea palaeo-fluid system during late Oligocene to Miocene time (Fig. 16). Massive mud diapirs reflect areas of significant vertical fluid flow. Absence of mud diapirs above gas chimneys reflects that the underlying lateral continuous Cretaceous or Palaeocene/Eocene 'thief sand has 'gathered' the fluids and transported them laterally. The mobilization map (Fig. 10) may therefore be a rough guide to the sand distribution. The nearcircular mud diapirs are indicative of fluidization, sand injection and limited vertical fluid transport through intra-Hordaland Group sands. The lateral extension of the intra-Hordaland Group sands is normally difficult to map using seismic data. As the overlying near-circular mud diapirs may be closely linked to the distribution of this sand, the eastward termination of the near-circular mud diapirs can be used as a guide to the pinch-out of intra-Hordaland Group sands. Stratigraphy within the chaotic zone. One of the consequences of the mud mobilization model is that the primary stratigraphic layering has been disrupted. The interbedded sand and shales in the mobilized part have been displaced vertically and with extensive mobilization the sediment flow may resemble convection cell motions. An implication of such behaviour is that the stratigraphy within the chaotic zone is unpredictable and biostratigraphic ages may appear in random order. The stratigraphy may be very different even in closely located wells. Mud mobilization also tends to generate rather rough palaeo-seafloor topography, particularly in areas of massive mud mobilization. It is well known that the shape of the deposition surface influences the
Seismic anomalies. The new aspect from this study is that it improves our ability to interpret seismic anomalies that previously were largely ignored. Vbrights, chaotic reflection zones, mounds, bowlshaped depressions, and gas chimneys can now be interpreted as features that reflect parts of the fluid motion processes in the subsurface. These features reflect palaeo-fluid motions in the North Sea, but similar processes may also be ongoing today in both the North Sea and other basins. The observations presented and new interpretations can therefore be regarded as a tool that improves our understanding of the hydrocarbon 'plumbing' system.
Conclusions Mobilized sediments in the Hordaland Group, North Sea, are fine-grained, with high water content and low shear strength. The mobilized sediments are characterized in seismic data by highly chaotic seismic reflection patterns and a top with irregular mounds and bowl-shaped depressions. Polygonal faults are normally developed in the laterally located non-mobilized succession. Two variants of mud diapirs are observed, large massive diapirs, the largest being 100 km long and 40 km wide, and minor near-circular mud diapirs that are typically 1-3 km in diameter. The massive mud diapirs are repeatedly observed above V-brights at the top of the Balder Formation, which are located above gas chimneys in the Cretaceous succession, which again are rooted at the crest of underlying Jurassic fault blocks. Fluid injection is suggested to have triggered the formation of the massive mud diapirs. Overpressure in compartmentalized Jurassic reservoirs in the deep Viking Graben hydraulically fractured the cap rock and gas, oil and formation water were transported
156
H.L0SETHCTAL.
vertically through the gas chimneys and injected into the unconsolidated smectite-rich sediments in the upper part of the Hordaland Group. The sediments formed a low-density mixture that was mobilized in a diapiric process. Relatively large mounds and bowl-like depressions developed on the palaeosurface. The Hordaland Group sediments were not mobilized where the gas and fluids were transported laterally in Cretaceous or Paleocene/Eocene 'thief sands. The 1-3 km wide near-circular mud diapirs are interpreted to have been formed by limited and focused fluid injection from intra-Hordaland Group sands. Formation water and possibly hydrocarbons were transported laterally through underlying intraHordaland Group sands and injected vertically where V-brights now are observed on seismic data. The injected sands represent point injectors that mobilized the sediments, forming near-circular mud diapirs. The absence of near-circular clay diapirs in the eastern part of the Norwegian Block 16/3 may be explained by pinch-out of lateral sand conduits (Fig. 16). This type of mud diapir is very common east of the Viking Graben. All the observed mobilization developed at shallow burial (<1000 m) by the injection of gas and/or fluids from an external source into an unstable sediment unit and is termed 'externa'. It should be discriminated from the more typical 'internal' form of mud diapirism that is generated in deep basins (>3000 m) by internal overpressure generation mechanisms. The proposed interpretation has revealed a new way of interpreting seismic anomalies and may be regarded as a tool that improves our understanding of the hydrocarbon 'plumbing' system. We are grateful to Statoil for permission to publish these results. We would also like to thanks the partners, Norsk Hydro, Esso-Mobil and Enterprise for permission to publish the data. J. Henden is thanked for sharing his knowledge of the Gullfaks and Gullfaks South fields and L.N. Jensen for preparing the structural map. We would also like to thank the referees D. Long and J. Iliffe for constructive comments that have improved the quality of the paper.
References APLIN, A.C, YUNLAI, Y. & HANSEN, S. 1995. Assessments of {3, the compression coefficient of mudstones and its relationship with detailed lithology. Marine and Petroleum Geology, 12, 955-963. BJ0RKUM, PA. & NADEAU, P.H. 1998. Temperature controlled porosity/permeability reduction, fluid migration, and petroleum exploration in sedimentary basins. Australia Petroleum Production and Exploration Association Limited Journal, 38, 453-464.
BJ0RLYKKE, K. 1998. Clay mineral diagenesis in sedimentary basins - a key to the prediction of rock properties. Examples from the North Sea Basin. Clay Minerals, 33,15-34. CANDACE, M., BROOKE, TJ. & MACKAY, TA. 1995. Mounded shallow gas sands from Quaternary of the North Sea: Analogues for the formation of sand mounds in deep water Tertiary sediments? In: HARTLEY, AJ. & PROSSER, DJ. (eds) Characterization of Deep Marine Clastic Systems. Geological Society, London, Special Publications, 94,95-101. CARTWRIGHT, JA. I994a. Episodic basin-wide fluid expulsion from geopressured shale sequences in the North Sea basin. Geology, 22,447-450. CARTWRIGHT, J.A. 1994&. Episodic basin-wide hydrofracturing of overpressured early Cenozoic mudrock sequences in the North Sea basin. Marine and Petroleum Geology, 11, 587-607. CARTWRIGHT, J.A. 1996. Polygonal fault systems: A new type of fault structure revealed by 3D seismic data from the North Sea basin. American Association of Petroleum Geologists. Studies in Geology, 42, 225-230. CARTWRIGHT, J.A. & LONERGAN, L. 1996. Volumetric contraction during the compaction of mudrocks: A mechanism for the development of regional-scale polygonal systems. Basin Research, 8,183-193. COSGROVE J.W. & HILLIER, R.D. 2000. Forced fold development within Tertiary sediments of the Alba Field, UKCS: Evidences of differential compaction and post-depositional sandstone remobilization. In: COSGROVE, J.W. & AMEEN, M.S. (eds) Forced Folds and Fractures. Geological Society, London, Special Publications, 167,61-71. DEWHURST, DAVID, N., CARTWRIGHT, J.A. & LONERGAN, L. 1999. The development of polygonal fault systems by syneresis of colloidal sediments. Marine and Petroleum Geology, 16, 793-810. DDCON, R.J., SCHOFIELD, K., ANDERTON, R., RENOLDS, A.D., ALEXANDER, R.W.S., WILLIAMS M.C. & DAVES, K.G. 1995. Sandstone diapirism and clastic intrusion in the Tertiary Sub-marine fans of the Bruce-Beryl Embayment, Quadrant 9, UKCS. In: HARTLEY, AJ. & PROSSER, D.J. (eds) Characterization of Deep Marine Clastic Systems, Geological Society, London, Special Publications, 94,77-94. EIDVIN, T, JANSEN, E., RUNDBERG, Y., BREKKE, H. & GROGAN, P. 2000. The upper Cainozoic of the Norwegian continental shelf correlated with the deep sea record of the Norwegian Sea and the North Atlantic. Marine and Petroleum Geology, 17,579-600. ELLIOT, T. 1986. Deltas. In: READING, H.G. (eds) Sedimentary Environments and Fades. Blackwell Scientific Publications. GREGERSEN, U., MICHELSEN, O. & S0RENSEN J. C. 1997. Stratigraphy and facies distribution of the Utsira Formation and the Pliocene sequences in the northern North Sea. Marine and Petroleum Geology, 14, 893-914. HOVLAND, M., NYGAARD, E. & THORBJ0RNSEN, S. 1998. Piercement shale diapirism in the deep-water Vema Dome area, V0ring Basin, offshore Norway. Marine and Petroleum Geology, 15, 191-201. ISAKSEN, D. AND TONSTAD, K. 1989. A revised Cretaceous
GAS AND FLUID INJECTION TRIGGERING SHALLOW MUD MOBILIZATION and Tertiary lithostratigraphic nomenclature for the Norwegian North Sea. The Norwegian Petroleum Directorate-Bulletin, 5. JOHANNESEN, J., HAY, S.J., MlLNE, J.K., JEBSEN, C, GUNNESDAL, S.C. & VAYSSAIRE, A. 2002. 3D Oil
migration modelling of the Jurassic petroleum system of the Statfjord area, Norwegian North Sea. Petroleum Geoscience, 8,37-50. JORDT, H. 1996. The Cenozoic geological evolution of the Central and Northern North Sea based on seismic sequence stratigraphy. Dr. Scient. thesis, Department of Geology, University of Oslo. JORDT, H., THYBERG, B. I. AND NOTTVEDT, A. 2000. Cenozoic evolution of the central and northern North Sea with focus on differential movements of the basin floor and surrounding clastic source areas. In: N0TTVEDT, A. et al. (eds) Dynamics of the Norwegian Margin. Geological Society, London, Special Publications, 167,219-243. LARSEN, M. 1994. The Balder field: Refined reservoir interpretation with the aid of high resolution seismic data and seismic attribute mapping. North sea oil and gas Reservoirs - III, Kluwer Academic Publishers, 115-124. LONERGAN, I., CARTWRIGHT, J. A. & JOLLY, R. 1998. The geometry of polygonal fault systems in Tertiary mudrocks of the North Sea. Journal of Structural Geology, 20,529-548. LONERGAN, L. & CARTWRIGHT, J. A. 1999. Polygonal faults and their influence on deep-water sandstone reservoir geometries, Alba Field, United Kingdom Central North Sea. AAPG Bulletin, 83,410-432. MALM, O.A., BRUUN CHRISTENSEN, O., FURNES, H., L0VLIE, R., AND RUESLATTEN, H. 1984. The Tertiary Balder Formation: An organogenic and tuffaceous deposit of the North Sea region. In: SPENCER, A.M. et al. (eds) Petroleum Geology of the North European Margin. Norwegian Petroleum Society (Graham & Trotman), 149-170. MANDL, G. & HARKNESS, R.M. 1987. Hydrocarbon migration by hydraulic fracturing. In: JONES, M.E. & PRESTON, R.M.F. (eds) Deformation of Sediments and Sedimentary Rocks. Geological Society, London, Special Publications, 29, 39-53. MIALL, A.D. 1985. Principles of Sedimentary Basin Analysis. Springer Verlag, 490 p.
157
MORLEY, C.K., CREVELLO, P. & AHMAD, Z.H. 1998. Shale tectonics and deformation associated with active diapirism: the Jerudong Anticline, Brunei Darussalam. Journal of the Geological Society, London,155,475-490. NICHOLLS, R.J. 1995. The liquification and remobilisation of sandy sediments. In: HARTLEY, A.J. & PROSSER, D.J. (eds) Characterization of Deep Marine Clastic Systems, Geological Society Special Publications, 94, 63-76. PEGRUM R.M. & SPENCER A.M. 1990. Hydrocarbon plays in the northern North Sea. In: BROOKS, J. (ed.) Classic Petroleum Provinces. Geological Society, London, Special Publications, 50,441-470. PIASECKI, S., GREGERSEN, U. & JOHANNESSEN, P.N. 2002. Lower Pliocenen dinoflagellate cysts from cored Utsira Formation in the Viking Graben, northern North Sea. Marine and Petroleum Geology, 19, 55-67. SKOGSEID, J., PLANKE, S., FALEIDE, J.I., PEDERSEN, T., ELDHOLM, O. & NEVERDAL, F. 2000. NE Atlantic continental rifting and volcanic margin formation. In: N0TTVEDT, A. et al. (eds) Dynamics of the Norwegian Margin. Geological Society, London, Special Publications, 167,295-326. SYLTA, 0. & KROGSTAD, W. 1992. Migration challenges in the North Viking Graben. IKU report 92.095,185p. TYBERG, B.I., JORDT, H., BJ0RLYKKE, K., & FALEIDE, J.I. 2000. Relationship between sequence stratigraphy, mineralogy and geochemistry in Cenozoic sediments of the northern North Sea. In: N0TTVEDT, A. et al (eds) Dynamics of the Norwegian Margin. Geological Society, London, Special Publications, 167,245-272. VAN RENSBERGEN, P., MORLEY, C.K., ANG, D.W., HOAN, T.Q. & LAM, N.T. 1999. Structural evolution of shale diapirs from reactive rise to mud volcanism: 3D seismic data from the Baram Delta, offshore Brunei Darussalam. Journal of the Geological Society, London, 156,633-650. WANG, C., MAO, N. AND Wu, FT. 1980. Mechanical properties of clays at high pressure. Journal of Geophysical Research, 85,1462-1468. WENSAAS, L., AAGAARD, P., BERRE, T. & ROALDSET, E. 1998. Mechanical properties of North Sea Tertiary mudrocks: investigations by triaxial testing of sidewall cores. Clay Minerals, 33,171-183.
This page intentionally left blank
Experimental evidence on the role of gas in sediment liquefaction and mud volcanism NORBERT PRALLE, MICHAEL KULZER & GERD GUDEHUS Institute of Soil and Rock Mechanics, University of Karlsruhe, P.O.Box 6980, D-76128 Karlsruhe, Germany (e-mail: kuelzer@ ibf-tiger. bau-verm. uni-karlsruhe.de) Abstract: Mud volcanoes are structures that are formed through 'cold volcanism' and indicate soil liquefaction. Their evolution depends on the structure, state and excitation of fine-grained feeding sediments. The disturbance of the framework of a loose, fine-grained, saturated sediment causes shear deformation leading to a pore fluid pressure increase. Effective stresses are thereby reduced and can vanish; the soil is then totally liquefied. Small amounts of enclosed gas bubbles render the soil compressible and enhance local shearing, pore pressure build-up and structural damage. Liquefied, overpressurized sediments form mud chambers, whose excess pressure is released through cracks and other inherent weak channels caused, for example, by density variations in the overlying strata. Experiments were conducted on small-scale soil columns and on model slopes, using quartz powder as model material. Soil column tests were aimed at determining the influence of gas bubbles with regard to the liquefaction process, whereas model slope tests were targeted at the collapsability of gentle slopes, subject to atmospheric pressure changes. Different initial porosities were achieved through ion content variation. Liquefaction of fine-grained sediments and subsequent volcano evolution could be produced.
Mud volcanoes are peculiar and intriguing geological structures. They range from a centimetre scale in recent limnic sediments to larger, seismically induced examples up to several meters in diameter (Ferentinos 2001 pers. comm), to much larger ones. For example, in Azerbaijan (Fig. 1), an area which is heavily explored for natural gas and petroleum, mud volcano cones have about 100-200 m in vertical relief and cover areas up to 5 km in diameter. The expulsion of methane gas is a constantly associated feature, regardless of the volcano's size. The composition of mud ejecta ranges from clays to clasts with an age from Cretaceous to Recent (Cooper 20010). On all scales, mud volcanoes have in common what can be described as 'cold volcanism', i.e. ejecting liquid like mud mixed with gas. Analogous to magmatic volcanism, a mud-feeding chamber must exist, which is filled with an overpressured mud suspension, producing typical volcanic features such as cones and craters upon its extrusion. Mud volcanism is often related to basin evolution with high sedimentation rates of fine-grained material with high water content, also with orogenic and tectonic activities as well as the occurrence of gaseous hydrocarbons. Some of those factors probably contribute to mud volcano evolution. From a soil mechanics point of view, several questions arise: which mechanisms and what initial soil states are favourable for the liquefaction of fine-grained soils; what role does the sedimentary environment play and which agents drive volcanic activity?
Liquefaction of granular sediments Mud volcanoes are produced mainly by temporarily and locally liquefied fine-grained sediments, which were extruded under pressures higher than hydrostatic from their loci of liquefaction in the subsurface to form a new deposit on the surface. Sediment liquefaction has long been a focus of geology and in research in geotechnical engineering, where it is significant with regard to potentially causing foundation or slope failure, mainly associated to earthquakes (e.g. Castro 1969; Dobry 1995; Ishihara 1996). The approach of the present paper, and the terminology employed, is drawn very largely from geotechnical engineering. Conventionally, saturated granular soils are considered to be liquefied when effective stresses of the soil approach zero. Shearing without drainage is isochoric (at constant volume) and increases pore fluid pressure, which can drastically reduce the skeleton pressure according to Terzaghi's principle of effective stresses. However, from a physical point of view the term liquefaction can be misleading since that implies a phase transition, viz decay of the skeleton. Liquefaction should therefore only be used for soils when becoming fluid-like, allowing unlimited deformations. In the geotechnical view, this can only occur for very porous granular soils: a granular skeleton can only decay under continued shearing if its void ratio e (ratio of void volume and solid volume - related to porosity n through n = e/(l + e)) exceeds the critical one for zero skeleton pressure,
Fig. 1. Large mud volcanoes encountered in Ayranteken, Azerbaijan (Cooper 200lb}. All volcanic features regardless of size are associated to methane gas.
ec0 (Gudehus 1996). Figure 2 illustrates the realm of permitted states according to the hypoplastic material law (Gudehus 1996), with e^ and ed being the upper and lower limit, respectively. The admixture of fines or the presence of capillary forces can render very porous soils potentially collapsible. For example, open pit mining in former East Germany left behind large collapsible deposits because of the common exploration practice of dumping fine, moist sand loosely from conveyor belts. Such soils with macropores show void ratios that are higher than limit void ratio et (Herle et al 1998; Pralle et al 2001). Dense soils also tend to undergo a pore pressure increase under cyclic isochoric shearing with small amplitudes, which is termed cyclic mobility (e.g. Castro 1969; De Albaef al 1976; Robertson & Fear 1995). Subsequent sufficient monotonic shearing increases effective stresses owing to soil's dilatant characteristic (Reynolds 1885). Certain sands are more prone to cyclic mobility than others, due to their granulometric properties such as roundness and grain size and due to their state, in particular porosity and pressure. Uniform (ratio of cumulative mass at d60 and cumulative mass at d]0) sands show a higher liquefaction potential than non-uniform ones. Completely saturated soils require shear to reduce effective stresses. However, soils containing gas bubbles are compressible and show considerable reduction of shear modulus during 'undrained' deformation because local yield zones are enabled by the gas bubbles (Duffy et al. 1994). The shear strength of dense soils is affected more strongly by the presence of gas than loose soils (Rad et al. 1994). Grozic et al. (1999) found that soils with degrees of saturation higher than 90% can experience significant strain softening. Muds consist of very fine hydrophilic mineral grains. The mechanical response of muds differs from that of sands, which can be explained by
Fig. 2. Void ratio e (density) vs. normalized grain skeleton pressure pjhs depicting the permitted soil states according to hypoplasticity between £/ and ed (Gudehus 1996; Bauer 1996). Shaded area shows range of void ratio where soils can turn into a suspension when sheared. (ed: max. density, ec\ critical density, e,: min. theoretical density for a granular body without macropores, ps: mean skeleton pressure, hs: granulate hardness).
contact mechanics only, as they contain colloids, whose texture and mechanical behaviour are dominated by water, electric forces, gas bubbles and additives. Because the evolution of their structure is very sensitive to the variability of ion concentrations and pH, highly porous honeycomb structures can evolve under certain electrostatic conditions and are higly liquefiable when disturbed. Liquefaction is sometimes used synonymously for fluidization, whereas in geology the latter is also used for the special case where effective stresses are reduced through the drag of water flow, such as through vertical flow force (e.g. Nichols etal. 1994; Owen 1996). In the following we shall refer to liquefaction as the liquid like state of the soil with zero effective stress', fluidization is used when a transfer of the liquefied granular sediment is involved.
Influence of ions on structure of fine grained soils It is known that the sedimentation of very finegrained siliciclastic particles is strongly controlled by their physico-chemical environment (Boggs 1987). Changing the pH-environment or ion concentration of a suspension will alter the structure of the sediments. Fig. 3 shows schematic and corresponding SEM photographs of quartz powder that sedimented in nearly ion free water, (a), and in salty water, (b). Sedimentation in a solution with high NaCl concentration reduces the repulsive forces
ROLE OF GAS FOR MUD VOLCANISM
161
Fig. 3. Sedimentation process in fresh (a) and salty (b) water shown as schematic (top) and corresponding SEM-photos of quartz powder (bottom). Illustrations from Reed (1995), photos from Zou (1998).
between the grains, which enhances flocculation. Grains are subjected to stronger attractive forces than before, which leads to the formation of clusters, of which smaller particles are preferably affected, since they are more sensitive to ionic variations in the free water than larger ones. Cluster formation will eventually produce a sediment with a so called honeycomb fabric (as illustrated in Fig. 3b) with a very low overall mean density that lies in the realm of the shaded area above et in Fig. 2. Such soils are naturally very sensitive to mechanical and chemical disturbances. Minute shear deformation will immediately cause a high pore pressure, which reduces effective stresses. The removal of ions, e.g. through leaching, can also destabilize the prevailing equilibrium between repulsive and attractive forces and will cause the honeycomb structure to collapse.
Criteria for laboratory tests Various laboratory experiments on granular sediments with different boundaries and initial conditions have aimed at the agents that favour liquefaction and/or the evolution of fluid escape structures, which is a more generic term for sand or mud boils or volcanism, respectively. Nichols et al (1994) investigated the evolution of sediment volcanoes in soils that consisted of a homogeneous granular layer confined at the top by a less permeable layer and subjected at its base to a vertical upward water flow. Prior to the breakthrough of water and fluidized
sediment through the upper layer, conical voids filled with water between the two layers evolved. Kokusho (1999) found similarly that a water film evolves under a silty seam that was sandwiched between two saturated sand layers after shock loading the sand. Owen (1996) conducted shaking table experiments and reproduced liquefaction induced sedimentary structures, such as sand volcanoes. Our goal was to improve the qualitative understanding of the basic mechanisms that lead to the formation of mud volcanism from a soil mechanics point of view, although without the complex conditions encountered in nature, such as compressional tectonic regimes, high pressured gas, fluid escapes and structural settings. Furthermore, the study aimed at simulating the processes that can develop within the soil after an external disturbance, i.e. exploring what processes continue to operate, especially related to gas bubbles, since these influence the effective stress evolution of the soil and constitute material heterogeneities. Several criteria had to be met: (a) drastic simplification of the boundary conditions; (b) generation of very porous homogeneous sediments; (c) presence of large gas bubbles (compared to mean grain size d50); and (d) a well defined method to produce a collapsible state. Two experimental set-ups were chosen. Firstly, a submerged quartz powder column with gas bubbles was subjected to quasistatic air pulses in order to investigate the mechanical role of gas bubbles with regard to liquefaction and its subsequent fluidization. Secondly, a submerged quartz powder body
162
N.PRALLECTAL.
Fig. 4. Void ratio e and degree of saturation Sr vs. salt concentration c for quartz powder using two different salts at neutral pH conditions.
with gas bubbles and with a gently inclined surface was also subjected to quasistatic air pulses in order to trigger liquefaction induced slope motion. This paper presents the results of laboratory tests that investigated the influence of entrapped gas bubbles on the liquefaction of mud-sized loose sediment, which was then subsequently transferred through cracks to the surface to produce sediment volcanoes. Mechanisms for their evolution are presented and discussed and related to mud volcanism encountered in situ.
dried quartz powder into salty water under stirring. The degree of saturation was then determined after sedimentation. Many gas bubbles were visible through the transparent lateral surface (resolution of unaided eye ca. 0.1 mm) being much larger than d50 and were rather randomly distributed within the model soil. Their presence and their spatial distribution within the soil were verified using the nuclear magnetic resonance (NMR) technology (Pralle 2002).
Column tests Experiments Model soil, porosity control and gas bubbles Quartz powder (Mikrosil LS300) was used as model material with the maximum particle size being about 40 fjim and an average particle size d50 of 8 jjim (no data were available for particles smaller than 6 JJUTI). The photos in Figure 3 reveal the significant angularity of the quartz grains. The ion concentration has a strong control on porosity as illustrated in Figure 4, where void ratios e differ for different concentrations of NaCl and CaQ2. The lower valency of Na+ compared to Ca2+ allows a more gradual variation of porosity. Model soils were prepared by adding sodium chloride to demineralized water, as shown in Figure 4. Although gas bubbles contributed to the obtained porosity, they were assumed to have had the same effect for both types of salt, based on the similar degrees of saturation. After trying various methods, gas bubbles were introduced through pouring oven
A perspex tube 10 cm in diameter and 40 cm height was filled with salty water. Quartz powder was then poured in, forming a submerged, but not entirely saturated column of ca. 30 cm height (Fig. 5) after its sedimentation (within two hours). Initial void ratios ranged around 1.3 and the degree of saturation was about 0.88. Undrained cohesion was estimated to be approx. 2 kPa. The tube was sealed to be airtight and was equipped with three pore pressure transducers to record pore water pressure pw, total stress p and air pressure pa. Pore water and total pressures were recorded at the bottom of the column and air pressure at the top of the tube (Fig. 5). Vertical motion of the sediment column surface was recorded using a laser displacement meter through the transparent top of the perspex tube. The water column was then subjected to quasistatic air pulses, which is equivalent to changes of the hydraulic load. Figure 6a shows the resulting total stress histories during ca. 35 load cycles of 5 s duration and ca. 14 kPa amplitude with a ca. 7 seconds pause in between.
ROLE OF GAS FOR MUD VOLCANISM
Fig. 5. Experimental set-up for column tests.
Observations and measurementsnts
Due to the pulses, water flowed into the soil column and caused a slight vertical compression during loading and expansion upon unloading. During each unloading phase air bubbles emerged from the surface and rose through the water column. In the following, reference to recorded data will be facilitated with marked key points in Figures 6 and 7. A denotes the initial state, E an initial loading with air pressure (required by the setup), C, Cl, C2 denote points during air pulse disturbance and D the final state. Note that initial total stress corresponds approximately to the column height and the water column beyond, i.e. at this point wall friction between quartz powder and perspex tube is not yet
163
effective (point A in Fig. 6a). However, the wall friction presumably accounts for the initial delay of total stress approaching air-pulse pressure amplitude, which is finally reached at all depths after six air pulses (C in Fig. 6a). Figures 6b and 7a show the evolution of pore pressure and vertical compression due to pulses, respectively. It can be seen that the quartz powder column, being only compressible due to the gas bubbles (incompressibility of water and quartz as well as undrained conditions is assumed), is apparently elastic during air pulse loading and unloading (Fig. 7a). Total pressure does not follow air pressure at the same rate during the first six cycles (up to point C), and pore pressure lags behind even more (Fig. 6b), which leads to an initial increase of effective stress (Fig. 7b). The reason why the pore water pressure increases gradually lies in the low permeability of the soil. Hydraulic loading causes water to flow into the soil, since the pore fluid is compressible due to the presence of gas bubbles. Although the bubbles can be treated as elastic, a residual settlement of the soil column can be observed, which is cumulative for the first nine loading cycles (up to point Cl). This can be explained by the friction between wall and soil column, which inhibits full expansion upon unloading because the soil is 'jammed' with the tube. However, if the gas bubbles cannot fully expand in accordance with their particular internal air pressure, they locally increase pore fluid pressure, which then dissipates according to the sediments (low) permeability. That can be well seen in Figure 7b: at point Cl effective stresses have been reduced to a local minimum. Friction between wall and soil is lowered during the next loading cycle and from then on effective stresses increase again and the cumulative net settlements are reduced (C2 in Fig. 7a). Upon unloading, water flow is reversed and flows upward, dragging tiny gas bubbles out of the sediment. After several more loading pulses, effective pressure approaches a limit value at D (grey dashed line in Fig. 7b) during unloading phases which is lower than the initial value at point A. After ending the pulses (point D), total stress (+1.5 kPa) and pore fluid pressure (+1.9 kPa) remain at higher values than before (Fig. 6), leaving effective pressure reduced by 0.4 kPa due to a remaining pore fluid excess pressure (Fig.7b). The quartz powder net settlement is ca. 0.2 mm, which indicates that despite continuous degassing during unloading and many loading cycles changes of density and saturation are negligible. Evolution of quartz powder volcanoenoess
Several minutes after the first air pulses, extrusive structures started to develop, forming a landscape
164
N.PRALLEEFAL.
Fig. 6. Histories of total pressure (a) and pore (fluid) pressure (b), recorded at the column bottom.
Fig. 7. Histories of vertical compression of quartz powder column (a). Shown in grey are the settlement trends during air pulses (bottom line) and during air pulse pauses (top line). Effective stress evolution (black line) during air pulses (b). Peaks in grey are data acquisition artefacts.
with many morphological features, such as encountered in nature (Fig. 8): pools (VP), fields of conlets small granular cones - (circled), large individual volcanoes (e.g. VI) and nearly horizontal flows (F). The uneven distribution of volcanoes is striking, with their concentration in the NW and their complete dearth in the SE, which presumably depends on the uneven initial distribution of gas bubbles in the sediment or on the evolution and geometry of the tributary conduits that feed into the main volcano conduit. Volcano cones grew simultaneously and successively, in intermittent spurts or continuously. Fluidized sediments extruded through craters, causing repeated mud flows resembling the ones in nature and building volcanoes of different sizes, almost always with a slope of ca. 35°, the angle of repose for quartz powder (Zou 1998). Flows are particularly well seen on volcanoes VI and V3 (Fig. 8 - arrows point at flows). Extrusion rates differed between volcanoes; volca-
noes VI,V2 and V3 grew rather simultaneously; V2 and V3 grew at approx. the same rate (it seems that flows were deposited alternate on each volcano, since neither dominates the other); VI had a larger extrusion rate and/or its main conduit was feed by various tributary ones. Figure 8b suggests a simultaneous evolution of volcanoes VI, V2 and V3 since their slope angles are equal. Apparently, disintegration of the volcanic cones began immediately after their creation since apparently older volcanoes (V5) show gentler slopes than young volcanic slopes (V4). The main formation phase of this volcanic landscape ended after approximately 30 minutes. Within approximately the next 12 hours the volcanic landscape had entirely disappeared. At almost zero pressure these structures are very sensitive to the slightest disturbance - continuing upward interstitial flow driven by the residual pore water excess pressure resulted once again in a completely flat surface.
ROLE OF GAS FOR MUD VOLCANISM
165
Fig. 8. Quartz powder volcanoes observed in column tests after repeated air pulses. Areal view (a), oblique view on a volcanic landscape (b).
Slope Tests This series of tests was conducted in a perspex container instrumented with a manometer, a pressure transducer, and two air taps for pressure control (Fig. 9). The tests aimed chiefly at the simulation of structural features with regard to slope motion (Pralle, 2002). As significant volcanic features also evolved during these experiments, they are presented here. To generate a slope, the container was initially
tilted by approx. 8° and filled with salty water. Dry quartz powder was then poured in and carefully stirred to obtain a well-dispersed suspension. Gas bubbles rose throughout that process, but degrees of saturation were only changed within the error margin. Through sedimentation a submerged sediment with an approximate degree of saturation of ca. 92% was formed, then the container's tilt was reversed. Utmost care was given to avoid failure initiation of the sensitive grain skeleton at this stage. The quartz powder sediment was marked with vertical ink lines along its front to better observe localization phenomena and with a thin black haze on the surface to better distinguish surface processes. The outside of the container front was scaled with a grid size of 1cm. After hermetically closing the container, further consolidation was awaited. Before pressure pulses were applied, only very slight creep movements were registered, probably due to reversal of the tilt. Quasistatic air pulses of approximately 4 kPa (approx. doubling hydrostatic pressure) were then applied to the water surface. Tests were conducted with various initial porosities.
Piping and sediment volcanism Fig. 9. Set-up of slope experiments after sedimentation and after reversing the initial tilt.
Firstly, cracks developed on the surface of the sediment (Fig. 10). After several pulses some cracks became more prominent. Gas bubbles and fluidized
N.PRALLE£rAL.
Fig. 10. Plan view of slope surface. Small extrusive conlets (labelled with c) close to cracks show tiny conduit opening at their top, whereas many warping and blisters occur randomly on the surface (labelled w). sediment expulsion could be observed primarily along cracks, as can be seen by tiny conlets in the enlarged section of Figure 10. Detachment surfaces developed along which slope motion occured (Pralle 2002). Between 5-15 minutes after ending the quasistatic pulses, several mud volcanoes evolved on the sub-horizontal surface after slope motion halted. Suddenly tiny fluid escape structures appeared on the surface spouting a fluidized mix of water and quartz powder. On and off gas bubbles were also relieved through the conduits. Blister-type warping structures appeared randomly on the surface, well distinguished from extrusive structures by the intact black haze (Fig. 10). Extrusive small structures appeared first next to cracks, where sediment material could be expelled at higher rates than through the undisturbed sediment. Analogous to real volcanoes, small cones evolved and grew successively to become larger structures as more subsurface sediment was extruded and deposited onto the surface, as can be seen clearly in Figure 11. The maximum diameter of these volcanoes was about 20 mm, the maximum height about 7 mm. Volcanoes were often accompanied by many small cones (Fig.ll). Figure 12 shows a plan view of a volcano. This circular cone with a completely filled crater evolved through repeated flows, which clearly can be identified. Figure 13 shows a profile of several volcanoes with their feeding conduits. The latter originated in chambers (ch) that consist of liquefied quartz powder and which are presumably the source of the extruded volcanic material. This assumption is based on the comparison of volume needed for building the volcano and the size of the chambers. The relative size of the chambers correlates with the relative size of the volcano. The slope angle of the volcanoes averaged about 35°, which indicates that
the material has lost its cohesion during the process of its disintegration, its transfer and resedimentation. In Figure 13 large gas bubbles (compared to d50) are visible, which are ubiquitous, i.e. at deeper and shallower levels (see arrows - gb). Also shown is an enlarged section of a deep quartz powder chamber with several gas bubbles and a larger, growing and vertically elongated gas bubble, which is about to connect to a chamber; this could be interpreted as being depleted of sediment (located right above). Note the chambers, which prevailed for many hours after the sediment was replaced with water. Stable chambers indicate that liquefaction and fluidization occurred only locally.
Summary and discussion of laboratory tests Sedimentation of silt-grade quartz powder with mean grain size of d50 = 8 jjim in salty water produced a very loose material with gas bubbles, their diameter of ca. 0.1 mm being much larger than dso. The submerged sediment was then subjected to quasistatic hydraulic load cycles. These experiments yield the following results: (1)
(2) (3)
The combination of quartz powder and ion concentration as a tool for porosity control yields a useful model material to study fluidization phenomena of nearly saturated granular sediments. Heterogeneities of the sediment fabric were due to large (compared to d50) gas inclusions. During unloading, gas bubbles were extruded from the quartz powder surface and rose up through the water column. Repeated pulsations of gas containing sediments produced residual pore water pressures.
ROLE OF GAS FOR MUD VOLCANISM
Fig. 11. A large volcano in the central part is surrounded mainly by conlets. Extrusive structures, i.e. consisting of originally subsurface material are evident by their white colour as opposed to blistering/warping (labelled w).
167
Fig. 12. Plan view of quartz powder volcano evolved after slope collapse.
low permeability (kf ca. 10 7 m/s), the hydraulic load is initially not only taken up by the pore fluid (4) Undrained compression of a sediment with gas above (Cl in Fig.6b), but the grain skeleton is also bubbles alters effective stresses. subject to loading and therefore effective stress (5) Volcano evolution can be reproduced under increases at first. The hydraulic load produces a simplified boundary condition with several seepage pressure, which is largest during the first morphological features similar to those load pulses and higher than the estimated cohesion encountered in nature, such as cones, craters cu of 2 kPa. Thus, as the water attempts to flow and sediment flows as well as mud chambers through the grain skeleton to compress the gas and conduits. bubble, the grain skeleton can be destroyed (6) Volcanoes formed in sediments after distur- (Fig.l4b). Resulting grain rearrangements and the bance ended. Volcano distribution was uneven, prevailing flow force locally increase pore pressure. i.e. some parts of the surface did not show any Kohler et al (1999) report that gas bubbles in fineexpulsion of sediment, whereas other parts grained soils greatly contribute to their fluidization showed high volcanic activity. This uneven during cyclic hydraulic loads and found the fluidizavolcano distribution can be attributed to the tion to be more pronounced the shorter the loading initial distribution of gas bubbles within the sed- time is, compared to the dissipation time, whereas iment, which controls the geometry of the the latter is controlled by the sediment's permeabilevolving conduits. ity. During unloading, this process is reversed; gas (7) Volcanoes were ephemeral and disappeared bubbles expand, water flows in the reverse direction quickly due to continuous vertical upward flow and within the vicinity of the gas bubbles the develin consequence of residual excess pore pres- oped excess pore pressure is sufficiently high to maintain a local suspension (Fig.l4c) through which sure. gas bubbles can migrate more easily due to their The mechanisms can be explained as follows: sedi- buoyant lift. Since pore pressure can only dissipate mentation of a granular soil with various grain sizes through the surface, upward flow further reduces within salty water enhances cluster formation, overall effective stresses and also drags small gas chiefly of small particles. Those clusters attach to bubbles out of the soil that are seated near the sedilarger grains forming a honeycombed soil structure. ment surface. As only excess pore pressure can Gas bubbles larger than d50 also form macropores. produce interstitial flow, local suspension zones produce a heterogeneous pore pressure distribution within the soil.
Onset of liquefaction Increase of hydraulic load compresses gas bubbles. In order to reach the bubbles, inflow of water is required (Fig.l4b). The rate of bubble compression depends at first on the permeability of the intact, i.e. continuous sediment. However, due to the limited duration of the pulses (5 seconds) and the sediment's
Onset of piping As the excess pore-pressure gradient can only point in the direction of the surface (in this experimental setup) and repeated pulses have increased the local suspension bubbles, they start to connect to each
168
N.PRALLE£7AL.
Fig. 13. Profile of quartz powder volcano chain in slope test clearly showing subsurface structures of volcanoes, ch denotes quartz powder chambers just below volcanoes and connected through conduits, gb denotes gas bubbles.
other (Fig.l4d) to form channels through which fluidized sediment migrates (a snapshot of a gas bubble joining a chamber is given in the insert of Fig. 13). Once a channel has surfaced, fluidized sediment is rapidly extruded; smaller neighbouring channels that had not surfaced yet, change their direction towards the surfaced channel, as if attracted by it. After eight pulses (Cl in Fig. 6b) at least one contiguous channel from bottom to top within the sediment column must have evolved (Fig.l4e), which is suggested by the pore pressure being in synchronicity with hydraulic loading at that point. In other words, the load that is applied at the top of the column is instantaneously recorded at the column bottom. Since the permeability of the soil has not changed (same grain size and negligible alteration of density) and based on the fact that at this point the first large extrusive structures appear on the surface it can be concluded that the sediment has locally undergone a phase transition into a suspension, which is confined to the geometry of channels. This is further evidenced by the fact that shortly after sediment extruded, pore pressure rising time increased again (C2 in Fig. 6b) and a response characteristic similar to the very first pulses could be observed (Fig. 14f). However, at this stage the sediment had lost its macroscopic homogeneity and had been structurally altered because of the formation of conduits and water-filled (apparently stable) chambers (Fig. 13). Further extrusion eventually formed volcanoes, which are the surface expressions of piping; this process could be termed 'cold volcanism'. The fact that sediment expulsion and volcano evo-
lution continued after termination of the pulses suggests an excess pore pressure driven mechanism. The combined volume of suspension bubble and conduit seems to control the volcano's size. The larger the liquefaction chamber, the larger the volcano (e.g. VI in Fig. 13). However, deeply-seated smaller suspension chambers with longer conduits produced similarly sized volcanoes with shallowseated and larger chambers, but shorter conduits (compare V2 and V3 in Fig. 13). The fact that the cones have a slope angle of approximately the quartz powder friction angle indicates that the material has lost its cohesion during fluidization. The volcanoes were also ephemeral - within hours they had disappeared again. During piping, excess pore pressure is rapidly dissipated through the channels. It is reasonable to assume that excess pore pressure had not entirely dissipated through piping, because of the residual pore pressure recorded (Fig. 6b), i.e. the remaining excess pressure was dissipated via a homogeneous seepage (upward) flow, which decreased effective stresses in the entire sediment column. Since the volcanic material has lost its cohesion due to temporary fluidization the slightest excess pressure can deform the surface structures, since its state of pressure is nearly zero. Volcano distribution was uneven, i.e. parts of the surface did not show any extrusive structures, whereas other parts showed high volcanic activity (Fig. 8). This uneven volcano distribution is attributed to the heterogeneous initial distribution of gas bubbles within the sediment, which is evidenced using NMR (Pralle 2002) and which sets the basis for the geometrical distribution of the evolving conduits.
ROLE OF GAS FOR MUD VOLCANISM
169
Fig. 14. Schematic for the evolution of local liquefaction, piping and sediment volcanism. Illustrated are macroscopic (left) and mesoscopic (right) processes. Light grey: sediment. Dark grey: areas of excess pore pressure. Thin wiggly arrows indicate interstitial flow. Straight bold arrows indicate flow of suspension.
General discussion and conclusions Even without the very complex geological boundary conditions, it is difficult to recognize the mechanisms of a three-phase sediment (granulate, water and gas) volcanism in the laboratory, mainly because of the intricate phase transition from solid to fluid. A key role for sediment liquefaction and its fluidization in these experiments can be attributed to the gas bubbles, which provide significant heterogeneities within the model sediment with regard to the soil's compressibility and thus stiffness, and its permeability. Their compressibility enables local structural alterations, i.e. shear deformations. These lead to
local pore water excess pressures, which are presumably the continuing driving force for the formation of suspension bubbles; termination of that process is strongly controlled by the permeability of the sediment. These loci with fluidized quartz powder join each other to form larger bubbles or channels and rise through the model sediment to eventually reach the surface and extrude. In contrast to experiments where fluid escape structures were produced due to a vertical flow force produced by water inflow (Nichols et al. 1994), these experiments showed that soil can be liquefied due to compression when soil is gassy and that liquefied sediment continues to migrate upward driven by
170
N.PRALLE£TAL.
its own overpressure. Wheeler (1988) showed that gas bubbles larger than void space significantly reduce the shear strength of a soil if the gas bubbles remain as such in the soil during loading, as opposed to being flooded or dissolved, which depends on the initial pore pressure and consolidation pressure. Kohler et al (1999) reports that gas bubbles in soils greatly contribute to the liquefaction of fine grained sediment when subjected to cyclic hydraulic loads. They found fluidization to be more pronounced the shorter the loading time is compared to the dissipation time, which is controlled by the sediment's permeability. This occurs often in sediments that are subjected to hydraulic changes due to tidal influence or ship traffic. Seismically induced liquefaction in shallow seabeds is enhanced by the presence of gas, as observed through rising gas bubbles during the Aigion earthquake (Ms = 6.1) in the Gulf of Corinth, Greece in 1995 (Papatheodorou & Ferentinos 1997). Roughly a fortnight after the main shock a reconnaissance survey was conducted. High resolution (3.5 kHz) seismic reflection profiles revealed the presence of sediment mounds that conformably overly the undeformed seafloor just above gassy marine sediments (Papatheodorou & Ferentinos 1997), which allow for an interpretation of mud volcanoes. Normally consolidated or underconsolidated sediments with high water content, generated in quickly subsiding basins with high sedimentation rates, require only very small deformation to be liquefied. Tectonically driven deformation (so-called internal drive) can cause regimes with overpressured sediments. That source of deformation is mostly encountered in greater basin depths and deformation rates are usually in the order of mm/year, hence the evolution of mud diapirism is more likely, rather than an suspension bubble sufficiently overpressured to be able to rise. However, if high pressure gas and fluid escapes, which are mostly fault/crack controlled (also similarly seen in Fig. 10) intrude such high porosity sediments they might cause immediate local liquefaction and provide the liquefied sediment with a buoyancy force that keeps the system selffeeding, analogously to the gas bubbles in the described experiments. Numerous investigations indicate a strong correlation between mud volcanism and high pressure fluid escapes that originate in greater depths (Graue 2000; Cooper 200la) or seismically induced liquefaction (Papatheodorou & Ferentinos 1997). If one wishes to reconcile in-situ observations with the experimental results from the laboratory, it is concluded that gas is a key agent during various stages of mud volcano evolution: for the local liquefaction of the source sediment due to intrusions of highly pressurized gas; once liquefied, gas drives the formation of suspension chambers as well as their
joining also after external input has ceased; gas provides buoyancy for the upward transport of the fluidized sediment. The volcano's size in-situ might be also be indicative of chamber size and its depth. Support was provided through EU-Project: Mechanisms of Catastrophic Landslides (LAME) Project Nr. ENV4CT97-0619, DFG Research Group 371: Equilibrium, Rearrangement and Transport Phenomena of Peloids, and Collaborative Research Programme (SFB) 461: Strong Earthquakes. Many thanks to C.V. Arnold for carefully reading the manuscript and M. Coop and G. Owen for their many valuable remarks.
References BAUER, E. 1996. Calibration of a comprehensive hypoplastic model for granular materials. Soils and Foundations, 36 (1), 13-26. BOGGS, S. 1987. Principles of Sedimentology and Stratigraphy, Merill Publishing Company, Englewood Cliffs, NY. CASTRO, G. 1969. Liquefaction of Sands. Ph.D. Thesis, Harvard University, Cambridge. COOPER, C. 200la. Mud volcanoes of the South Caspian Basin - seismic data and implications for hydrocarbon systems. AAPG Annual Meeting Abstract. Denver Colorado. COOPER, C. 2001/7. Photo provided through personal communication. DE ALBA, P., SEED, H.B. & CHAN, C.K. 1976. Sand liquefaction in large-scale simple shear tests. Journal of the Geotechnical Engineering Division, GT9, Sept. 1976, 909-927. DOBRY, R. 1995. Liquefaction and deformation of soils and foundations under seismic conditions (state of the art paper). In: Proceedings of the 3rd International Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics, St. Louis, MD, 3,1465-1490. DUFFY, S.M., WHEELER, SJ. & BENNELL, J.D. 1994. Shear modulus of kaolin containing methane bubbles. Journal of Geotechnical Engineering, 120, 5, May 1994. GRAUE K. 2000. Mud volcanoes in deepwater Nigeria. Marine and Petroleum Geology, 17 (8), 959-974. GROZIC, J.L., ROBERTSON, PK. & MORGENSTERN, N.R. 1999. The behaviour of loose gassy sand. Canadian Geotechnical Journal, 36,482-492. GUDEHUS, G. 1996. A comprehensive constitutive equation for granular materials. Soils and Foundations, 36 (1), 1-12. HERLE, I., WEHR, W. & GUDEHUS, G. 1998. Influence of macrovoids on sand behaviour. 2nd International Conference on Unsaturated Soils, Beijing 1998. Springer-Verlag, 60-65. ISHIHARA, K. 1996. Soil behaviour in earthquake geotechnics. Publications of Oxford Engineering Science Series, 46. KOHLER, H.-J., SPIES, H., BERINGER, O. & HAUBECKER, H. 1999. Fluidization and deformation of submerged soil due to fluctuating water level. In: BARENDS, F.B.J.,
ROLE OF GAS FOR MUD VOLCANISM LlNDENBERG, J., LUGER, H.J., DE QUELERIJ, L. &
VERRUIJT, A. (eds) Proceedings of XII. European Conference on Soil Mechanics and Geotechnical Engineering, Amsterdam, Netherlands, June 7th-10th, 1999. Geotechnical Engineering for Transportation Infrastructure, Balkema, Rotterdam, 1109-1115. KOKUSHO, T. 1999. Water film in liquefied sand and its effect on lateral spread. Journal of Geotechnical and Geoenvironmental Engineering, 817-826. NICHOLS, R.J., SPARKS, R.S.J. & WILSON, C.J.N. 1994. Experimental studies of the fluidization of layered sediments and the formation of fluid escape structures. Sedimentology, 41,233-253. OWEN, G. 1996. Experimental soft-sediment deformations: structures formed by the liquefaction of unconsolidated sands and some ancient examples. Sedimentology, 43,279-293. PAPATHEODOROU, G. & FERENTINOS, G. 1997. Submarine and coastal sediment failure triggered by the 1995, Ms = 6.1. Aegion earthquake, Gulf of Corinth, Greece. Marine Geology, 137,287-304. PRALLE, N., BAHNER, M.L. & BENKLER, J. 2001. Computer tomograhic analysis of loose sands. Canadian Geotechnical Journal, 38, (4), 770-781. PRALLE, N. 2002. Mechanisms in Nearly Saturated Sandy Soils Subject to Quasi-Static and Dynamic Loading.
171
Ph.D. Thesis, Publication Series of Institute of Soil and Rock Mechanics, University of Karlsruhe, Germany. No. 158. RAD, N.S., VIANNA, AJ.D. & BERRE, T. 1994. Gas in soils. II: Effect of gas on undrained static and cyclic strength of sand. Journal of Geotechnical Engineering, 120 (4), 716-736. REED, J.S. 1995. Principles of ceramics processing. Wiley and Sons, New York. REYNOLDS, 0.1885. On the dilantancy of media composed of ridgid particles in contact. With experimental illustration. Philosophical Magazine, Series 5, 20, 469-481. ROBERTSON, P.K. & FEAR, E.C. 1995. Application of CPT to evaluate liquefaction potential. In: Proceedings of the International Symposium on Cone Penetration Testing. Swedish Geotechnical Society Report, 3 (95). WHEELER, S.J. 1988. The undrained shear strength of soils containing large gas bubbles. Geotechnique, 38 (3), 399-413. Zou, Y. 1998. Der Einfluss des gebundenen Wassers auf die Leitfahigkeit und die mechanischen Eigenschaften feinkorniger Boden. Ph.D. Thesis, Publication Series of Institute of Soil and Rock Mechanics, University of Karlsruhe, Germany. No. 144.
This page intentionally left blank
Sinuous pockmark belt as indicator of a shallow buried turbiditic channel on the lower slope of the Congo basin, West African margin A. GAY1'2'*, M. LOPEZ2, P. COCHONAT3, N. SULTAN3, E. CAUQUIL4 & E BRIGAUD4 J
Universite de Lille I, Laboratoire Sedimentologie et Geodynamique, FRE 2255, Bat SN5, 59655 Villeneuve d'Ascq, France. 2 Universite de Montpellier II, Laboratoire Geophysique Tectonique et Sedimentologie, CC60, Bat 22, place E. Bataillon, 34095 Montpellier cedex, France (e-mail: gay @ elstu-univ-montp2.fr) 3 IFREMER, Institut Franc, ais de Recherche et d'Exploitation de la MER, Departement Geosciences Marines, Laboratoire Environnements Sedimentaires, BP70, 29280 Plouzane, France 4 Total-Fina-Elf, 64018 Pau Cedex, France Abstract: Pockmarks on the slope of the Lower Congo Basin are distributed along a meandering band on seafloor coincident with a shallow buried palaeochannel imaged from the 3D-seismic database. Each pockmark originates systematically at the channel-levee interface and the seafloor expression of the palaeochannel's sinuosity is mimicked by the sinuous trend of pockmarks. 3Dseismic on the slope, calibrated by biostratigraphic data from cores of the Leg ODP 175, indicate a seaward decrease of the sedimentation rate. We suggest that this condition induces a differential loading of the hemipelagic cover over the palaeochannel and propose a model for episodic dewatering of fluids trapped in the buried turbiditic channel. The consequence is a fluid flow caused by a longitudinal pressure gradient along the buried channel. A hydromechanical model proposed for the formation of shallow pockmarks indicates that the sedimentation rate cannot generate the overpressure required for pockmark formation on the seafloor. Therefore, it is suggested that hydrocarbon migration from deeper overpressured reservoirs is added to the pore fluid pressure in the shallow subsurface sediments. Horizontal drainage by the turbiditic palaeochannel and vertical migration along many vertical conduits (seismic chimneys) probably initiated at shallow subbottom depth. It is concluded that these shallow processes have important implications for fluid migration from deeply buried hydrocarbon reservoirs.
Fig. 1. Bathymetric map of the Zaire turbidite system, extending from the Zaire estuary to the deep sea fan. The shaded circle represents the study area in the Lower Congo Basin. The three sites of the Leg ODP175 in this zone are indicated.
buried anomalies such as seismic chimneys (Heggland 1998) and acoustically 'blanked' layers, which are interpreted as gas accumulations (Yun et al. 1999) or gas-charged sediments (Hovland et al. 1984; Hempel et al 1994). Because of the nature of fluids expelled, pockmarks may represent openwindows above the petroleum system and could be valuable indicators for deeper reservoir strategy. On the slope of the Lower Congo Basin, pockmarks are not randomly distributed, but always associated with fault zones, salt diapirs or gas hydrates intervals. Moreover, detailed analysis of bathymetric maps and 3D-seismic data permitted to characterize a sinuous belt of pockmarks that mimicked a shallow buried meandering channel of Pliocene age acting as a horizontal drain for interstitial fluids. This paper focuses on morphological aspects, distribution and hydromechanical model of this new kind of pockmarks. The later has to be regarded as an indicator of overpressure at shallow subsurface levels that can initiate fluidization features in unconsolidated sediments, if the seepage forces due to fluid flow are larger than its own weight.
Geological setting The West African passive margin was initiated during the opening of the South Atlantic Ocean at Early Cretaceous (130 My) (Jansen et al 1984; Marton et al 2000). Subsequent to large accumulations of evaporites (up to 1000 m) during the Aptian, the post-rift stratigraphy is characterized by two distinct seismic architectures that reflect a major change in ocean circulation and climate:
(1)
(2)
From Late Cretaceous to Eocene time an aggradational carbonate/siliciclastic ramp develops in response to low-amplitude/lowfrequency sea-level changes and stable climate (i.e. greenhouse period, Bartek et al 1991; Seranne etal 1992; Seranne 1999). From Oligocene time to Present, sedimentation was dominated by the progradation of a terrigenous wedge that reflects high-amplitude/high-frequency sea-level changes and an alternating drier and wetter climate (i.e. icehouse period; Seranne 1999).
During the icehouse period, due to the global climate cooling the increased terrigenous input to the Atlantic Ocean rejuvenated deposition of a large tubiditic fan off Congo and Angola slopes directly fed by the Zaire River (Brice et al 1982; Reyre 1984; Uchupi 1992; Droz et al 1996). The total thickness of the turbidite fan ranges from 8 to 10 kmthick and extends from the Zaire estuary down to 4000 m water depth (Fig. 1). During flooding stages, the Zaire river discharged high density bed load into the submarine canyon that originates directly at the river mouth, to feed a large sinuous channel-levee system, far onto the lower fan (Jansen et al 1984; Uenzelmann-Neben 1998; Savoye et al 2000). Only fine materials, not confined to the canyon, are delivered to the Lower Congo Basin (LCB) from riverine plumes (Cooper 1999). This suspended terrigenous material is mixed with hemipelagic sediments on the continental shelf and slope to feed the 2000-3000 m thick progradational wedge that progressively overlays the abandoned turbiditic sandy-channels at the base of the slope. During the ODP Leg 175 (1998),
SINUOUS POCKMARK BELT ON SEAFLOOR
175
Fig. 2. Seafloor dip map in the study area. The grey-scale ranges from 0° in white (horizontal surfaces) to 30° in black (sloping surfaces). The white lines indicate seismic profiles shown in the following figures. Pockmarks are not evenly distributed on the seafloor: Area 1 is characterized by small pockmarks associated with underlying gas hydrates or diapirs, and Area 2 shows a high concentration of larger pockmarks along a sinuous belt on seafloor (see text for details).
three sites were drilled at various positions from the shelfbreak (sites 1075, 1076 and 1077), which supplied new information about the nature and age of these sediments. Biostratigraphical analyses indicate an overall continuous hemipelagic settling for Mid-Pliocene sediments occurring at a rate of about 12 cm/k.y (Giraudeau etal 1998).
Data and methods This study was primarily based on a 3D-exploration seismic dataset acquired by the Total-Fina-Elf oil company and combined with a bathymetric map and a 2D seismic Pasisar profile (see below for details) acquired during the ZAIANGO project. One of the main objectives of the ZAIANGO project was to better understand the Quaternary history of the Zaire fan (Savoye et al 2000). Additional data from the Leg ODP175 on the West African Margin provided useful information on stratigraphy, sedimentation rates and mechanical properties of sediments (Wefer etal. 1998&). The bathymetric map was acquired with a Simrad EM 12 dual multibeam. Complementary data have been collected more recently with the Simrad EM300 dual multibeam and provided higher vertical and lateral resolution for acquisition in water depth less than 3500 m. The profile CD (Fig. 4) was obtained with the PASISAR system. The PASISAR is a deep seismic streamer towed behind a conventional SAR system developed by IFREMER for
high-resolution studies in water depths ranging from 200 to 6000 m (Savoye et al 1995). The 3D-dataset selected for this study covers an area of 592 square kilometres with a line spacing of 12.5 m and a CDP distance of 12.5 m. They were loaded to a station and interpreted with the SISMAGE software developed by Total-Fina-Elf. 3D seismic imagery allows extraction of continuous horizons by propagation in the 3D block and attribute calculation (Kidd 1999).
Description of pockmarks on the CongoAngola slope Pockmarks in the Lower Congo Basin seem randomly distributed (Fig. 2). We observed more than 250 pockmarks in the study area, with an average density of 0.42 pockmarks per km. They range from 100 m up to 800 m in diameter and from a few metres to 40 m in depth. Most of them have a circular shape in plan view, but the largest pockmarks are elongated in one main direction. Detailed observations show that these large pockmarks are composite features. The dip map of the seafloor of the investigated area extracted from the 3D-seismic (Fig. 2) delineates two main zones of pockmarks characterized by their distribution shape, size and density. Area 1 is characterized by patches or isolated small pockmarks that extend on a large domain of the northern slope of the Zaire canyon; as opposed to Area 2,
176
A.GAYCTAL.
where large pockmarks are distributed in a sinuous belt at the right bank of the Zaire canyon. These two domains are clearly associated with different buried structures; First, the pockmarks of Area 1 that are of more conventional origin are described, focusing on the sinuous pockmark belt of Area 2 and the link with a subsurface channel is discussed.
of 1000 m also occur within Area 1. They correspond to the imprint on the seafloor of a normal faulted network, due to the collapse above the diapir crest (Stewart 1999). Similar diapir pockmarks and BSR's offshore Nigeria have been discussed by Hovland el al (1997), Heggland (1997) and Graue (2000).
Pockmarks related to gas hydrates and salt diapirs (Area 1)
The sinuous pockmark belt (Area 2)
Area 1 is characterized by small circular pockmarks, ranging from 100 m to 300 m in diameter, and from a few metres to a maximum of 20 m in depth (Fig. 2). They are unevenly distributed and their abundance varies considerably within the area. In particular, pockmarks develop in areas covered by 1 to 3 km regularly spaced linear depressions. These furrows have a north/south orientation, perpendicular to the regional slope with about 1 km in length and have an average depth of 5 m; they are interpreted as regular deformation by creeping of the superficial slope sediments. Numerous seismic profiles through pockmarks show two superposed acoustic anomalies, vertically elongated under the pockmarks (Fig. 3). The shallowest anomaly is ovoid in shape with depressed high-amplitude reflectors interpreted as a reduction of the seismic velocities (pull-down effects) through a gas-charged column. Such acoustic anomalies are also called seismic chimneys and could be indicative of fluid flow from deeper levels (Hempel et al 1994; Tingdahl et al 2001). The deepest anomaly characterized by acoustic turbidity corresponds to an inverted cone shape, marked by a fadeout of the reflectors. On both sides of this region the bright reflectors shift upward. Profile AB (Fig. 3) shows a high-amplitude reflection parallel to the seafloor located at 250 ms TWT. This is a Bottom Simulating Reflector (BSR), which is often considered as the lower thermodynamic limit of the gas hydrates stability zone (Shipley et al 1979). BSR's are characterized by the reversed polarity compared to the seafloor reflection, indicating a downward reduction of seismic impedance and therefore of seismic velocity. This contrast in impedance is probably due to the presence of free gas entrapped below the gas hydrate stability zone and the BSR can be considered as the interface between high-velocity gas hydrates and the underlying gas-charged sediments of low acoustic velocity. On this profile, the BSR is deflected upward directly beneath the pockmark depressions, suggesting a localized positive heat flow anomaly. This dome-shaped anomaly could be due to an ascending movement of fluids through the sedimentary column. Several giant circular depressions with a diameter
Area 2 is a 3 to 4 km wide and 41 km long pockmark band, crossing the study area from the SE to the NW corner. Its southern boundary is coincident with the right bank of the Zaire canyon. Pockmarks in this area are 100 to 800 m long, with a maximum depth of over 40 m. Some pockmarks are open-ended, suggesting that they have formed from the coalescence of several smaller pockmarks. They are mainly concentrated in the central part of Area 2, along a 22 km long section. They are regularly spaced about 300 m apart, along a sinuous belt. The high-resolution PASISAR profile (Fig. 4) crosses one of these pockmarks on the right levee of the Zaire canyon. The reflections below the pockmark appear depressed and not enhanced, in contrast to the observations made on seismic profiles in Area 1. Moreover, no inverted V-shape anomaly was detected directly beneath the depressed zone. We conclude that the down bowing of these reflections is not an artifact, but the reflectors are physically depressed and represent a chimney for ascending fluids from underlying levels to the seafloor. At about 200 m below seafloor, the chimney branches on an ancient buried channel-levee system. Reflectors are depressed down to the palaeochannel and along the channel-levee interface. The base of the chimney is located on the left edge of the channel fill, where it seems to take root because of the lack of any deeper sound-speed anomaly. The close relationship between the pockmark and the buried palaeochannel is evident on the PASISAR profile CD, which has been performed by horizon mapping on 3D-seismic.
Mapping of the buried palaeochannel Automatic picking of the base of the palaeochannel is difficult because of the irregular reflection patterns. To map palaeochannels, two continuous horizons with high amplitudes (Fig. 5) were combined, one within the channel fill and one within the levee system. The combination of the two horizons provides an isochronal map of the palaeochannel. The dip map shown on Figure 6a clearly delineates the channel trend that borders the present day Zaire canyon from the SE to the NW. The sinuosity
SINUOUS POCKMARK BELT ON SEAFLOOR
177
Fig. 3. Seismic profile AB in Area 1, extracted from 3D data, and interpreted line drawing. This profile crosses two pockmarks, illustrating the close relationship between seafloor features, associated chimneys and the BSR. High amplitude reflectors in chimneys are interpreted as gas-charged intervals.
of the palaeochannel is characterized by regular, smooth curves with a constant channel width of 800 m. The central axis of the palaeochannel seems to coincide with the meandering trend of pockmarks identified on the seafloor (Fig. 6b). Moreover, pockmarks within Area 2 are located above the edges of the buried channel-levee system.
Spatial characteristics of the sedimentary cover above the palaeochannel The sedimentary cover above the palaeochannel as constrainted by using a time-to-depth empirical function by seismic-well calibration provided by Total-
Fina-Elf. With this rule, the synthetic isochronal map of the palaeochannel has been converted into an isodepth map and the sediment thickness above the channel-levee system has been represented as an isopach map (Fig. 7). This map shows the general decrease of the sediment thickness, from 360 metres in the SE to 20 metres in the NW, in agreement with the sedimentation rates calculated from three cores drilled in this area during the Leg ODP 175. The sedimentation rate decreases from the eastern shallower site (site 1076) to the western deeper site (site 1075) from 15 cm/k to 10 cm/k.y (Giraudeau et al 1998). This overall seaward decrease in sedimentation rate indicates the progressive disappearance of terrigenous and hemipelagic input.
178
A.GAY£rAL
Fig. 4. PASISAR profile CD showing the close relationship between pockmarks at the seafloor in Area 2 and a buried palaeochannel. A vertically elongated zone of depressed reflectors up to 180 ms TWT below the seafloor is interpreted as a seismic chimney. No acoustic anomalies have been identified at deeper levels, suggesting that the migration of fluids started at the channel-levee interface.
Relationship between pockmarks, palaeochannel and the sedimentary cover As previously outlined on the dip seafloor map, pockmarks are concentrated along 22 km in the central part of Area 2 (Fig. 8). Arrows indicate the real positions of pockmarks along the palaeochannel axis. The general trend of the curve presents two knick-points, which determine three individual zones highlighted by three grey backgrounds: (1)
The distal zone in the NW part of Area 2 is the zone of thinnest sediment cover. The thickness varies from 15 m to 130 m. No pockmarks have been identified on the seafloor in this zone. (2) The central zone of Area 2 includes the majority of pockmarks. Although the thickness of overlying sediments varies from a minimum of 110 m to a maximum of 245 m, with an average of 175 m. Pockmarks only appear where the thickness of the sediment cover ranges from 130 m to 240 m. (3) The landward zone, in the SE part of Area 2, is
the zone of thickest sediments. This zone is characterized by an average thickness of the sedimentary cover ranging from 165 m to 260 m. Only four pockmarks occur in this zone, all four in places where the sediment cover is less than 240 m.
Morphological evidences for fluid seepages on seismic profiles and dip seafloor maps Different morphological features on seismic profiles and dip seafloor maps have been observed on the three previously identified sub-zones of Area 2. The seismic profile GH (Fig. 9) in the distal zone shows that some depressions are not located directly over the channel axis (600 m or more), but systematically located at higher bathymetric levels. This suggests that they are markers of ancient abandoned meander loops, such as in the present Zaire canyon (Babonneau et al 2002). These perched meander loops show crescent-shaped depressions of about 600 m length that are progressively buried due to the
SINUOUS POCKMARK BELT ON SEAFLOOR
179
Fig. 5. Seismic profile EF in Area 2, extracted from 3D data, crossing the buried palaeochannel. This profile illustrates the difficulty to map the buried channel by automatic horizon picking. Manual picking of the base of the channel would be time-consuming. Two continuous high-amplitude horizons located near the top of the channel fill and near the top of the levees were traced to map the palaeochannel. This way, the general morphology of the palaeochannel is preserved.
upward increase of sedimentation rate. All parallel horizons above both levees of the palaeochannel are truncated near flanks of the depression and display top-lap structures, suggesting erosional or non-depositional features. Due to the presence of a constructive dome, it is suggested that both depressions are due to fluid escape, which locally prevents a normal sedimentation rate. In the central zone, the entire buried channel-levee system is pointed out by the meandering track of highly concentrated pockmarks on the seafloor (Fig. 10). It appears that their location seems to be dependent on sediment thickness above the palaeochannel. Pockmarks are mostly concentrated in the distal part of this zone, where long depressions result from the coalescence of two (or more) smaller pockmarks, suggesting a peanutshaped geometry (i.e., composite pockmarks). In this zone, no isolated seafloor pockmarks, nor buried chimneys or fossil pockmarks have been identified on seismic sections. These morphological features indicate that fluid escape is currently active or has been recently active. In the landward zone, only four small pockmarks have been identified. However, different to the central zone, numerous buried structures such as chimneys and buried pockmarks (Long 1992) are visible on seismic sections crossing the
palaeochannel (Fig. 11). They are identified on seismic profiles by depressed reflectors, horizontally sealed by slope sediments at the present day. Buried pockmarks are systematically located above channel flanks and indicate the past activity for fluid seepage above the channel-levee system. The few pockmarks visible on the modern seafloor could indicate secondary fluid migration through ancient reactivated chimneys.
Mechanical model for overpressure in a buried silty/sandy channel The channel fill is generally characterized by a predominantly sandy/silty sediment. Stratified hernipelagic sediments cover the channel-levee system, leading to its progressive burial and compaction. Overpressure development in sedimentary basins is directly related to the types of sediment facies deposited (controlling lithology), sedimentation rate, thermal expansion of fluids, transformation of clay minerals and to hydrocarbon generation (Yu & Lerche 1996) or bacterial methanogenesis. Among these factors, sediment facies and sedimentation rates are the main factors controlling fluid pressure
180
A.GAYCTAL.
Fig. 6. A Dip map of the two mapped horizons at the level of the palaeochannel (see Fig. 5). B Spatial correlation between seafloor pockmarks and the channel extent. All pockmarks in Area 2 (black circles) occur at the channel flanks. The meandering track of pockmarks at the seafloor represents the sinuous trend of the buried palaeochannel.
development in a basin. In this case, hemipelagic mud with low permeability entraps the palaeochannel, characterized by higher permeabilities and prevents efficient dewatering of pore fluids. A rapid increase of the overburden pressure can lead to the generation of excess pore fluid pressures (Bolton & Maltman 1998), and fluids can escape from the sand/silt body through the muddy cover, creating pockmarks on seafloor (Cole et al. 2000). In this case, pockmarks form in fine sediments and not in sandy sediments as reported by Hovland in the North Sea (Hovland & Judd 1988). In a sedimentary column, an elementary volume AV is subjected to three forces:
(1)
its own weight, Fg, due to gravity:
(2)
where psat is the specific gravity (in kN.m 3); forces of buoyancy, Fb, due to immersion in water:
(3)
where pf is the specific gravity of fluid (generally 10 kN.m-3); seepage forces, Fs, due to fluid flow:
SINUOUS POCKMARK BELT ON SEAFLOOR
181
Fig. 7. Isopach map of the sediment cover of the palaeochannel. The thickness decreases from 360 m in the landward zone to 20 m in the distal zone, illustrating the progressive decrease of terrigenous sediment input.
where pf is the specific gravity of fluid and i is the hydraulic gradient with I = -Qra(^ h, where h represents the head pressure. Without any specific pathways where fluid may circulate and/or accumulate, pore fluids can escape up to the seafloor if sediments are fluidized: grains become suspended in fluid, which can migrate upward. Therefore, the balance between ascending forces (Fs and Fb) and descending forces (Fg) must be equal and the hydraulic gradient, i, must reach the critical gradient, ic. For a vertical seepage, ic is given by the following equation:
where p' corresponds to the submerged gravity. The equation (1) becomes: For fluid migration up to the seafloor, a vertically critical gradient must be taken into account from the top of the palaeochannel to the seafloor: where Ah is the variation of head pressures between the top of palaeochannel and seafloor and L represents the thickness between these two points. Einsele (1977) and Bonham (1980) showed that flow velocities during sediment compaction and the range of compaction-driven fluid flow primarily depend on thickness of the compacting sedimentary column. During the Leg ODP175, bulk densities were measured and compiled on shipboard along each core, every 4 to 50 cm. Lithostratigraphical and mag-
netostratigraphical analyses conducted in the Lower Congo Basin show intercalations of hemipelagic and terrigenous deposits that can be easily correlated from site to site and sequences are regionally crosscorrelated. Cores from site 1077, the nearest site from Area 2, should provide a good flashover of mechanical properties in our study area. Although sediment compressibility is reduced by several orders of magnitude with increasing effective stress during compaction (Neuzil 1980), real profiles of sediment compressibility and bulk density have not been integrated in any equation. For shallow processes of compaction, an average bulk density for the first 240 m is considered as a good approximation. With a sediment thickness less than 240 m, fluids can escape along discontinuities, such as the channellevee interface. Above 240 m thickness, the sedimentation can seal the system, requiring an excess pore pressure for pockmark generation. Equation (4) yields an average value of 2.944 kN.nT3 for the submerged density. Equation (6) gives a variation of head pressures, Ah, equal to 70.7 m, which is equivalent to an excess pore pressure at the top of palaeochannel of 0.707 MPa (considering g equal to 10 m.s~2). This overpressure corresponds to the minimum excess pore pressure needed for pockmark formation with a sediment cover of 240 m.
Discussion An excess pore pressure can be generated in a buried layer by two processes: (1) a thick deposit of finegrained sediment with low permeabilities can create overpressure in an underlying level. In this case the sedimentation rate and associated lithologies are key
182
A.GAYZ7AZ,
Fig. 8. Sediment thickness distribution along the palaeochannel axis. Arrows indicate the positions of pockmarks. Based on sediment thickness, three zones are identified: a distal zone (sediment cover thickness: 15-130 m), a central zone (sediment cover thickness: 130-245 m), and a landward zone (sediment cover thickness: 165-260 m). Pockmarks only occur where the thickness of the sediment cover is between 130 and 240 m.
Fig. 9. Correlation between seafloor morphology and a cross-section in the distal zone of Area 2. Left: dip seafloor map. The white dashed line represents the axis of the buried palaeochannel. Right: seismic section GH across the palaeochannel (see Fig. 2 for location). Note that several curved depressions are visible on seafloor corresponding to the hemipelagic fill of abandoned meanders. They may represent areas of non-deposition due to fluid seepage.
parameters. A model described by Dugan (Dugan & Flemings 2000) predicts that significant overpressures will originate where loading is rapid. Due to the low permeabilities of the hemipelagic cover, fluids are preferentially entrapped in the sandy/silty body of the buried channel, which can lead to an excess of pore pressure and later to the up-dip migration along bedding planes, i.e. up the flanks of levees and basin fill.
Effect of the sedimentation rate for generating overpressure The vertical stress due to an additional load is: where psat is the bulk density (in kN.rn"3) and d is the thickness of the new deposit (in m). The average bulk density for 240 m in the core 1077 is 12.944
SINUOUS POCKMARK BELT ON SEAFLOOR
183
Fig. 10. Correlation between seafloor morphology and a cross-section in the central zone of Area 2. Left: seafloor dip map. The white dashed line corresponds to the palaeochannel axis. Right: seismic section EF (see Fig. 2 for location). Channel flanks are highlighted by a sinuous trend of pockmarks at the seafloor. Pockmarks and underlying chimneys seem to take root at the channel-levee interface.
Fig. 11. Correlation between seafloor morphology and a cross-section in the landward zone of Area 2. Left: seafloor dip map. The white dashed line represent the axis of the buried palaeochannel. Right: seismic section IJ (see Fig. 2 for location). Only four pockmarks have been identified in this zone. All other pockmarks seem to be sealed. This zone may represent an area of episodic seepage due to the high sealing capacity of the hemipelagic cover.
kN.m 3 and the vertical stress corresponding to the overpressure is 707 kN.m~2. Equation (7) gives a value of d equal or superior to about 56 m. This thickness represents the last 56 m of the sediment cover (240 m), which has been deposited very quickly over the palaeochannel, leading to the overpressure. The dissipation time of overpressured fluids, t%, depends on the hydraulic diffusivity Dz (1.10~8 rn^s"1 in the study area, calculated from ODP data, ODP175 initial report, in Wefer et al 1998b), on the maximum vertical distance of dissipation z (the dissipation can be performed upward or
downward, so z = 240 / 2= 120 m) and on the time factor Tv(in%):
Tv is related to the consolidation rate U (in %). Values of Tv are available from pre-calibrated curves or can be expressed from a Fourier series (see Appendix for details). For a consolidation rate of 50%, Tv50% = 0.197 and Equation (8) gives a time dissipation of 9000 years. For a consolidation rate of
184
A.GAYETAL
99%, Tv99% = 2, the time dissipation is 91,000 years. The average sedimentation rate at the site 1077 is about 10 cm/k.y and reaches 20 cm/k.y for the last 200000 years (Wefer et al 1998a). Based on a maximum sedimentation rate of 20 cm/k.y, 56 m would have been deposited in 275,000 years. The overpressure would have already dissipated when the sediment thickness reached 56 m. Thus, the sedimentation rate appears insufficient to explain an overloading effect capable of producing the excess of pressure needed to expel fluids up to the seafloor. Only repeated overflow deposits from the Zaire canyon could build up a thick sediment cover rapidly reaching 56m. Biostratigraphical analyses of all sites in the Lower Congo Basin indicate an overall continuous hemipelagic sedimentation, characterized by the absence of (or only minor) post-depositional sediment transport. Only one thin and isolated Bouma D/E turbidite sequence has been identified (Pufahl etal 1998, and site 1075 descriptions).
belt and associated shallow buried channel act as a by-pass zone for free gases. Due to the sandy/silty nature of the channel fill, the overpressure is uniformly distributed along the channel body. Dugan & Flemings (2000) developed a two-dimensional model to generate overpressure, in which the geometry of the reservoir and rate of loading control lateral fluid transfer. This model predicts that significant overpressures will originate where loading is rapid. Longitudinal flow occurs in the palaeochannel because a pressure gradient exists, due to the differential loading above it. The gradient is assumed to be highly dependent on the loading geometry, the bulk compressibility and the hydraulic conductivity. The pockmark distribution on seafloor is not dependent on the location of the fluid sources under the palaeochannel, because this lateral gradient exists.
Fluids migration from deeper levels as a key parameter
Area 2 is characterized by a seaward decrease of the sedimentary cover above a shallow buried channel axis, where three sub-zones have been identified as a function of morphological features on the seafloor and on seismic sections:
Pockmark structures are commonly attributed to fluid venting from overpressured biogenic/thermogenic methane, oil or other pore fluids. In the slope of the Lower Congo basin, the hemipelagic sediments play the role of an impermeable seal over the turbiditic palaeochannel. If fluids migrate from underlying levels, they are preferentially entrapped in higher permeability layers, represented here by the sandy/silty linear body (Hovland & Judd 1988; Tinkle et al 1988; Mann & Mackenzie 1990; Premchitt et al. 1992). In these conditions, the channel acts as a drainage pipe, and the supply of ascending fluid from deeper overpressured reservoir can exceed the pore pressure limit of 0.707 MPa necessary for fluidization, upward migration and pockmark formation on the seafloor. This hypothesis implies that fluids migrating up to the palaeochannel, partly originate from deep thermogenic processes (Brooks et al 1999). The expected nature of fluids escaping from seafloor pockmarks should be a mixture of interstitial water, shallow biogenic gases (produced by bacterial degradation of organic matter) and thermogenic gases or oil from deep buried reservoirs. A discontinuous BSR is evident on 3D-seismic data in Area 1, directly superposed beneath deep-buried palaeochannel bodies (Figs 5 and 10). This is in contrast to Area 2, where no distinct BSR is evident beneath the shallow buried channel, indicating the lack of free gas trapped under gas hydrates or the lack of gas hydrates themselves. It is suggested that the chimneys directly expel all fluids from the channel body reservoir. Pockmark
Conceptual model for fluid seepages above a shallow buried palaeochannel
(1)
the distal zone displays direct fluid escape because the sedimentary cover is very thin above buried meanders; (2) the central zone has a medium-thick cover that allows active fluid seepage creating new pockmarks or re-using ancient chimney, that can lead to vertically stacked pockmarks; (3) the landward zone is characterized by a thick sedimentary cover; pockmarks are sealed because the rate of sedimentation exceeds the seepage forces, but fluids can periodically escape. This pattern demonstrates that channel-related pockmarks are active over a window of sediment cover of thickness between 130 m and 240 m. This active fluid window moves basinward during time by slope progradation and pockmarks progressively mimic the palaeochannel axis (Fig. 12). Conversely, in the landward zone, numerous fossil pockmarks and associated chimneys indicate a past activity for fluid seepage (Hovland et al 1984). The thickness between the channel-levee interface and fossil pockmarks is of the same order as the present day thickness of the cover observed in the central zone of Area 2 (Fig. 10). This last observation suggests that leakage processes may stop where there is a high seal capacity of the sedimentary cover. If the mechanical model for expelling fluids
SINUOUS POCKMARK BELT ON SEAFLOOR
185
Fig. 12. Conceptual model for fluid expulsion from a sandy/silty buried channel. Fluid escape in the landward zone is assumed to be sporadic, because of a thick sediment cover and characterized by a low pockmark density on the seafloor. Central and distal zones may be active for fluid seepage at the present day. This 'active window' moves basinward in pace with slope progradation.
from shallow buried channel is clearly evidenced by 3D seismic profiles, the spatial link with deep over pressured reservoirs remains more hypothetical. It seems that at a greater depth, a rapid porosity collapse may generate overpressure and lead to the development of tensile fractures, veins and sand injections contributing to fluid migration (Fisher et al 1999). Some examples of similar features have been reported in layered strata with strong permeability contrasts from the Gulf of Mexico. In this area, the overpressured fluids from the sand layers migrate upwards in the overlying mud by fracture permeability, following the minimum principal stress (Bishop Stump & Flemings 2000; Cole et al 2000). In the North Sea, buried craters have been observed at a Pliocene horizon, which may have
been formed during an earlier period of sustained gas seepage. Hydrocarbon migration pathways are largely controlled by the distribution of high permeability conduits, such as faults (Yu & Lerche 1996) and sand-rich carrier sequences and their structural dip or geometry (Burley et al 2000). In particular, Cartwright (1994) and Lonergan et al (1998) described thick mudstone-dominated sequences disrupted by complex arrays of small extensional faults distributed at intervals of 100-500 m. The network of faults is assumed to be produced by volumetric contraction of fine-grained sediments; it may easily drive fluids from underlying stratigraphic units up to the seafloor.
186
A.GAYCTAL.
Conclusion This integrated study combining conventional seismic profiles with 3D seismic blocks in the Lower Congo Basin showed a close relationship between a type of seafloor pockmarks and a buried palaeochannel: these pockmarks seem to take root at the channel-levee interface. They are systematically distributed on both sides of the buried channel body and mimic the meandering track of the palaeochannel on the seafloor. Six key results have been established:
where u is the pore water pressure, Dz is the hydraulic diffusivity, t is time and z denotes the position where u is determined. The Terzaghi's consolidation equation can be solved using analytical or numerical techniques. The solution obtained depends on the boundary conditions. For our case, with a soil layer of height, 2H, the boundary conditions are:
(1)
(b)
The buried turbiditic channel determines a horizontal drain for lateral fluid flow. (2) The seaward decrease of the sedimentary cover activates a differential overloading responsible of down channel fluid migration and pockmarks development, when the thickness ranges between 130 m and 240 m. (3) A value of 707 kPa was calculated as the minimum excess of pore pressure needed for fluid bursting and pockmarks formation, for a channel buried at about 240 m below seafloor. (4) As the sedimentary wedges build up the slope, this open-window moves downward along the palaeochannel axis. (5) Considering the sedimentation rate in the study area, the excess of pore pressure in the palaeochannel is supposed to be created by an additional fluid supply that migrates upward from deeper levels. (6) Discontinuities in the sedimentary column, such as faults, erosional surfaces, or buried chimneys may channelize deep fluids during migration. Conversely, all sedimentary structures, such as channel bodies or sandy/silty layers, concentrate fluids before redistribution as intermediate reservoirs. Finally, a new type of pockmark closely linked to a buried palaeochannel has been described in the Lower Congo Basin and a hydromechanical model has been proposed, which implies the mixing of shallow and deep buried reservoirs. These seismic chimneys are the spatial link between source rock, reservoir trap and shallow-gas anomalies; their detection may be indicative of both potential zones of geohazards and deeper prospective reservoirs (Heggland 1998; Aminzadeh et al 2001; Tingdahl et al 2001).
Appendix The process of consolidation is directly linked to the rate of excess pore pressure dissipation. The onedimensional consolidation theory is governed by the following differential equation (Terzaghi 1943):
(a)
complete drainage at top and bottom of the layer; u = 0 at z = 0 and z = 2H; the initial excess pore water pressure Au = iij is equal to the applied stress increment ACT.
The solution is obtained as a Fourier series, which can be expressed in the following form:
where Uz is the degree of consolidation at time, t, at depth z, and Tv is a non-dimensional time factor. Uz and T are given by:
where Hdr is the length of the longest drainage path. Based on the numerical solution of equation (10), and in order to define the time factor TV as a function of the degree of consolidation Uz, Casagrande (1936) and Taylor (1948) determine a 'pre-calibrated' curve concerning the Time factor, Tv, which is given by the following equations:
From equations (11), (13) and (14) and for a given hydraulic diffusivity Dz and for a given drainage path Hdr, it is possible to evaluate the time t needed to obtain a specified degree of consolidation Uz. This work was largely improved by the data of the ZAIANGO project, co-sponsored by IFREMER and TotalFina-Elf. Authors are very grateful to A. Morash head of the Deep Offshore Project at TFE; and Bruno Savoye head of the ZAIANGO project at IFREMER, for their financial support and data supplies. M. Hovland, R Vogt and an anonymous reviewer are gratefully acknowledged for their critical reviews and their suggestions. We warmly thank R Van Rensbergen for his careful English edit and N. Babonneau for useful comments concerning the Zaire canyon.
SINUOUS POCKMARK BELT ON SEAFLOOR
References ABRAMS, M.A. 1992. Geophysical and geochemical evidence for subsurface hydrocarbon leakage in the Bering Sea, Alaska. Marine and Petroleum Geology, 9,208-221. ABRAMS, M.A. 1996. Distribution of subsurface hydrocarbon seepage in near-surface marine sediments. In: SCHUMACHER, D. & ABRAMS, M.A. (eds) Hydrocarbon Migration and its near-surface expression. American Association of Petroleum Geologists Memoir, 66,1-14. AMINZADEH, R, DE GROOT, RE, BERGE, T. & VALENTI, G. 2001. Using gas chimneys as an exploration tool. World Oil, May 2001,50-56. BABONNEAU, N., SAVOYE, B., CREMER, B. & KLEIN, B. 2002. Morphology and architecture of the present canyon and channel system of the Zaire deep-sea fan. Marine and Petroleum Geology, 19,445-467. BARAZA, J., ERCILLA, G. & NELSON, C.H. 1999. Potential geologic hazards on the eastern Gulf of Cadiz slope (SW Spain). Marine Geology, 155,191-215. BARTER, L.R., VAIL, PR., ANDERSON, J.B., EMMET, PA. & Wu, S. 1991. Effect of Cenozoic ice sheet fluctuations in Antartica on the stratigraphic signature of the Neogene. Journal of Geophysical Research, 96, 6753-6778. BISHOP STUMP, B. & FLEMINGS, P.B. 2000. Overpressure and fluid flow in dipping structures of the offshore Gulf of Mexico (E.I. 330 field). Journal of Geochemical Exploration, 69-70,23-28. BOE, R., RISE, L. & OTTESEN, D. 1998. Elongate depressions on the southern slope of the Norwegian trench (Skagerrak): morphology and evolution. Marine Geology, 146,191-203. BOLTON, A. & MALTMAN, A. 1998. Fluid-flow pathways in actively deforming sediments: the role of pore fluid pressures and volume change. Marine and Petroleum Geology, 15, (4), 281-297. BONHAM, L.C. 1980. Migration of hydrocarbons in compacting basins. American Association of Petroleum Geologists Bulletin, 64,549-567. BRICE, S.E., COCHRAN, M.D., PARDO, G. & EDWARDS, A.D. 1982. Tectonics and Sedimentation of the South Atlantic Rift Sequence: Cabinda, Angola. Studies. In: DRAKE, W.A. (ed.) Continental Margin Geology. American Association of Petroleum Geologists Memoir, 34,5-18. BROOKS, J.M., BRYANT, W.R., BERNARD, B.B. & CAMERON, N.R. 1999. The nature of gas hydrates on the Nigerian Continental Slope. Abstract book of the Third international conference of gas hydrates (18—22 July), Park City, Utah. BROWN, A. 2000. Evaluation of possible gas microseepage mechanisms. American Association of Petroleum Geologists Bulletin, 84 (11), 1775-1789. BURLEY, S.D., CLARKE, S., DODDS, A., FRIELINGSDORF, J., HUGGINS, P., RICHARDS, A., WARBURTON, I.C. & WILLIAMS, G. 2000. New insights on petroleum migration from the application of 4D basin modelling in oil and gas exploration. Journal of Geochemical Exploration, 69-70,465-470. CARTWRIGHT, J.A. 1994. Episodic basin-wide hydrofracturing of overpressured Early Cenozoic mudrock
187
sequences in the North Sea Basin. Marine and Petroleum Geology, 11 (5), 587-607. CASAGRANDE, A. 1936. Characteristics of cohesionless soils affecting the stability of slopes and earth fills. Journal of Boston Society of Civil Engineers, 23, 13-32. COLE, D., STEWART, S.A. & CARTWRIGHT, J.A. 2000. Giant irregular pockmark craters in the Palaeogene of the Outer Moray Firth Basin, UK North Sea. Marine and Petroleum Geology, 17,563-577. COOPER, C.K. 1999. Ocean currents offshore Northern Angola. Proceedings Offshore Technology Conference Paper 10749, Houston, Texas, May 3-6. DROZ, L., RIGAUT, E, COCHONAT, P. & TOFANI, R. 1996. Morphology and recent evolution of the Zaire turbidite system (Gulf of Guinea). Geological Society of America Bulletin, 108 (3), 253-269. DUGAN, B. & FLEMINGS, P.B. 2000. The New Jersey margin: compaction and fluid flow. Journal of Geochemical Exploration, 69-70,477^81. EICHUBL, P., GREENE, H.G., NAEHR, T. & MAKER, N. 2000. Structural control of fluid flow: offshore fluid seepage in the Santa Barbara Basin, California. Journal of Geochemical Exploration, 69-70,545-549. EINSELE, G. 1977. Range, velocity and material flux of compaction flow in growing sedimentary sequences. Sedimentology, 24,639-655. FISHER, Q.J., CASEY, M., BEN CLENNELL, M. & KNEPE, R.J. 1999. Mechanical compaction of deeply buried sandstones of the North Sea. Marine and Petroleum Geology, 16,605-618. GIRAUDEAU, J., CHRISTENSEN, B.A., HERMELIN, O., LANGE, C.B., MOTOYAMA, I. & Shipboard Scientific Party. 1998. Biostratigraphic age models and sedimentation rates along the southwest African Margin. In: WEFER, G., BERGER, W.H., RICHTER, C. et al (eds) Proceedings of the Ocean Drilling Program. Initial Reports, 175,543-546. GRAUE, K. 2000. Mud volcanoes in deepwater Nigeria. Marine and Petroleum Geology, 17, No. 8,959-974. HEGGLAND, R. 1997. Detection of gas migration from a deep source by the use of exploration 3D seismic data. Marine Geology, 13,41-47. HEGGLAND, R. 1998. Gas seepage as an indicator of deeper prospective reservoirs. A study based on exploration 3D seismic data. Marine and Petroleum Geology, 15, 1-9. HEMPEL, P., SPIESS, V. & SCHREIBER, R. 1994. Expulsion of shallow gas in the Skagerrak- Evidence from subbottom profiling, seismic, hydroacoustical and geochemical data. Estuarine, Coastal and Shelf Science, 38, 583-601. HOVLAND, M. 1981. Characteristics of pockmarks in the Norwegian trench. Marine Geology, 39,103-117. HOVLAND, M., JUDD, A.G. & KING, L.H. 1984. Characteristic features of pockmarks on the north sea floor and scotian shelf. Sedimentology, 31,471-480. HOVLAND, M. & JUDD, A. 1988. Seabed pockmarks and seepages. Graham and Trotman, London. HOVLAND, M., GALLAGHER, J.W., CLENNELL, M.D. & LEKVAM, K. 1997. Gas hydrate and free gas volumes in marine sediments: Example from the Niger Delta front. Marine and Petroleum Geology, 14 (3) 245-255.
188
A.GAYETAL
JANSEN, J.H., VAN WEERING, T.C.E., GIELES, R. & VAN IPRESEN, J. 1984. Middle and Late Quaternary oceanography and climatology of the Zaire-Congo fan and the adjacent eastern Angola Basin. Netherlands Journal of Sea Research, 17, 201-249. JOSENHANS, H.W., KING, L.H. & FADER, G.B. 1978. A sidescan sonar mosaic of pockmarks on the Scotian shelf. Canadian Journal Earth Science, 15, 831-840. KELLEY, J.T., DICKSON, S.M., BELKNAP, D.F., EARNHARDT, W.A. & HENDERSON, M. 1994. Giant sea-bed pockmarks: Evidence for gas escape from Belfast Bay, Maine. Geology, 22,59-62. KIDD, G.D. 1999. Fundamentals of 3D seismic volume visualization. Proceedings Offshore Technology Conference Paper 11054, Houston, Texas, May 3-6, 823-836. KING, L.H. & MACLEAN, B. 1970. Pockmarks on the Scotian shelf. Geological Society of America Bulletin, 81,3141-3148. LONERGAN, L., CARTWRIGHT, J. & JOLLY, R. 1998. The geometry of polygonal fault systems in Tertiary mudrocks of the North Sea. Journal of Structural Geology, 20, (5), 529-548. LONG, D. 1992. Devensian Late-glacial gas escape in the central North sea. Continental Shelf Research, 12 (10), 1097-1110. MANN, D.M. & MACKENZIE, A.S. 1990. Prediction of pore fluid pressures in sedimentary basins. Marine and Petroleum Geology, 7,55-65. MARION, L.G., TARI, G.C. & LEHMANN, C.T. 2000. Evolution of the Angola passive Margin, West Africa, with Emphasis on Post-Salt Structural Styles. In: MORIAK, W. & TALWANI, M. (eds) Atlantic Rifts and Continental Margins. Washington, DC, American Geophysical Union, 129-149. Neuzil, C.E. 1980. Groundwater flow in low permeability environments. Water Resources Research, 22, 1163-1195. ORANGE, D.L., GREENE, H.G., REED, D., MARTIN, J.B., MCHUGH, C.M., RYAN, W.B.F., MAKER, N., STAKES, D. & BARRY, J. 1999. Widespread fluid expulsion on a translational continental margin: Mud volcanoes, fault zones, headless canyons, and organic-rich substrate in Monterey Bay, California. Geological Society of America Bulletin, III, 992-1009. PREMCHTTT, J., VON RAD, N.S., To, P., SHAW, R. & JAMES, J.W.C. 1992. A study of gas in marine sediments in Hong Kong. Continental Shelf Research, 12, 1251-1264. PUFAHL, P. K., MASLIN, M.A., ANDERSON, L., BRUCHERT, V, JANSEN, F., LIN, H., PEREZ, M., VIDAL, L. & Shipboard Scientific Party. 1998. Lithostratigraphic summary for LEG 175: Angola-Benguela upwelling system. In: WEFER, G., BERGER, W.H., RICHTER, C. et al (eds). Proceedings of the Ocean Drilling Program. Initial reports, 175,533-542. REYRE, D. 1984. Petroleum Characteristics and geological evolution of a passive margin. Example of the Lower Congo-Gabon Basin. Bulletin du Centre de Recherche Exploration Production ElfAquitaine, 8 (2), 303-332. SAVOYE, B., LEON, P., DEROECK, Y.H., MARSSET, B., LOPES, L. & HERVEOU, J. 1995. PASISAR: a new tool for near-bottom very-high resolution profiling in deep water. First Break, 13 (6), 253-258.
SAVOYE, B., COCHONAT, P., et al 2000. Structure et evolution recente de 1'eventail turbiditique du Zaire: premier resultats scientifiques des missions d'exploration ZAIANGO 1 et 2 (Marge Congo-Angola). Comptes-Rendus de I'Academic des Sciences de la Terre et des Planetes, 331, 211-220. SERANNE, M., SEGURET, M. & FAUCHIER, M. 1992. Seismic super-units and post-rift evolution of the continental passive margin of southern Gabon. Bulletin de la Societe Geologique de France, 163,135—146. SERANNE, M. 1999. Early Oligocene stratigraphic turnover on the west Africa continental margin: a signature of the Tertiary greenhouse-to-icehouse transition ?, Terra Nova, 11 (4), 135-140. SHAW, J., COURTNEY, R.C. & CURRIE, J.R. 1997. Marine geology of St. George's Bay, Newfoundland, as interpreted from multibeam bathymetry and back-scatter data. Geo-Marine Letters, 17,188-194. SHIPLEY, T.H., HOUSTON, M., BUFFLER, H.R.T., SHAUB, F.J., MCMILLEN, K.J., LADD, J.W., WORZEL, J.L. 1979. Seismic evidence for widespread possible occurrence of gas-hydrate horizons or continental slopes and rises. American Association of Petroleum Geologists Bulletin, 63,2204-2213. SOLHEM, A. & ELVERHOI, A. 1993. Gas-related sea floor craters in the Barents Sea. Geo-Marine Letters, 13, 235-243. SOTER, S. 1999. Macroscopic seismic anomalies and submarine pockmarks in the Corinth-Patras rift, Greece. Tectonophysics, 308,275-290. STEWART, S.A. 1999. Seismic interpretation of circular geological structures. Petroleum Geoscience, 5, 273-285. TAYLOR, D.W 1948. Fundamentals of soil mechanics. John Wiley, New York. TAYLOR, M.H., DILLON, W.P. & PECKER, LA. 2000. Trapping and migration of methane associated with the gas hydrate stability zone at the Blake Ridge Diapir: new insights from seismic data. Marine Geology, 164,79-89. TERZAGHI, K. 1943. Theoretical soil mechanics, John Wiley, New York. TINGDAHL, K.M., BRBL, A.H. & DE GROOT, P.P. 2001. Improving seismic chimney detection using directional attributes. Journal of Petroleum Science and Engineering, 29,205-211. TINKLE, A.R., WENER, K.R., MEEDER, C.A., HUFF, J.F., JOHNS, D.M. & MAY, J.A. 1988. Seismic no-data zone, offshore Mississippi delta: Part I - acoustic characterization. Proceedings Offshore Technology Conference, Houston, Texas, May 2-5,5753,65-74. UCHUPI, E. 1992. Angola Basin: Geohistory and Construction of the Continental Rise. In: DE GRACIANSKY, PA. Geologic Evolution of Atlantic Continental Rifts. Nostrand Reinhold, New York, 77-99. UENZELMANN-NEBEN, G. 1998. Neogene sedimentation history of the Congo Fan. Marine and Petroleum Geology, 15,635-650. VOGT, PR., GERDNER, J., CRANE, K., SUNDVOR, E., BOWLES, F. & CHERKASHEV, G. 1999. GroundTruthing 11- to 12-kHz side-scan sonar imagery in the Norwegian + Greenland Sea: Part II: Probable diapirs on the Bear Island fan slide valley margins and the Voting Plateau. Geo-Marine Letters, 19 (1-2), 111-130.
SINUOUS POCKMARK BELT ON SEAFLOOR WEFER, G., BERGER, W.H., RICHTER, C. & Shipboard Scientific Party 19980. Facies patterns and authigenic minerals of upwelling deposits off southwest Africa. In: WEFER, G., BERGER, W. H., RICHTER, C. et al Proceedings of the Ocean Drilling Program. Initial Reports, 175,487-504. WEFER, G., BERGER, W.H., RICHTER, C. et al 1998/7. Leg ODP 175, Initial Reports. WERNER, F. 1978. Depressions in mud sediments (Eckernfoerde Bay, Baltic Sea) related to sub-bottom and currents. Meyniana, 30,99-104.
189
WHITICAR, M.J. & WERNER, F. 1981. Pockmarks: Submarine vents of natural gas or freshwater seeps? Geo-Marine Letters, 1,193-199. Yu, Z. & LERCHE, I. 1996. Modelling abnormal pressure development in sandstone/shale basins. Marine and Petroleum Geology, 13 (2), 179-193. YUN, J.W., ORANGE, D.L. & FIELD, M.E. 1999. Subsurface gas offshore of northern California and its link to submarine geomorphology. Marine Geology, 154, 357-368.
This page intentionally left blank
Shallow bottom-simulating reflectors on the Angola margin, in relation with gas and gas hydrate in the sediments HERVE NOUZE1 & AGNES BALTZER2 l
Departement Geosciences Marines, Centre IFREMER de Brest, BP 70, 29280 Plouzane, France (e-mail: herve. nouze @ ifremer.fr) 2 Univ. Caen, Centre de Geomorphologie du CNRS. LAbo M2C 24 rue des tilleuls, 14 032 Caen, France Abstract: Acoustic facies interpretation, high-resolution velocity analysis and amplitude versus offset modelling have been performed on high resolution seismic data acquired on the West African margin offshore Angola, in water depths of about 2000 m. The area has a complex structural, thermal and fluid-flow setting, in which sediments are affected by salt diapirism and faulting associated with sediment compaction. A discontinuous bottom-simulating reflector (BSR) at a depth of about 200 m below sea floor could mark the base of the gas hydrate occurrence zone, which does not always coincide with the top of the free gas zone. Within the gas hydrate stability zone, a shallow bottom-simulating reflector is observed at a depth of about 75 m below seafloor. This shallow bottom simulating reflector, that is termed 'sheep back reflector' (SR), correspond to a small amount of gas being trapped in the sediments. It could mark the top of the gas hydrate occurrence zone, where gas hydrate dissociation may occur. A reversed polarity reflector (Rl) is also observed about 25 m below the sea floor. This reflector could correspond to a limit between normally compacted and underconsolidated sediments, possibly related to a permeability change in the sediments. Thus, the occurrence of excess pore pressure generated during gas hydrate dissociation could explain some subsurface sediment mobilization processes.
Bottom-simulating reflector (BSR) events typically occur at a depth of between 200 and 600 m beneath the seafloor based on seismic reflection data. These unusual reflections cross cut sedimentary layer reflectors and are parallel to the seabed. Many authors have interpreted the BSR to be generated at the transition between sediments containing a variable amount of solid gas hydrate above and sediments containing a small volumetric fraction, typically a few percent, of free gas, below (Andreassen et al 1997; Paull et al 1996; Tinivella & Accaino 2000). Gas hydrate-related BSR should consequently lie at depths that closely correspond to the base of the gas hydrate stability zone (BGHS) (Hyndman & Spence 1992), although discrepancies have been reported (Bangs et al. 1993; Ruppel 1997). The BGHS is controlled both by temperature and pressure, as well as by chemical conditions (Zatsepina & Buffett 1998). A BSR can exist without any evidence for gas hydrates (Hoviand et al. 1999) and alternatively, gas hydrates were recovered where no BSR is present (Paull et al. 1996). These results, among others, based upon ground truth, have lead some authors to reconsider the interpretation of the BSR. Many alternative explanations have been proposed to account for the BSR, including artefacts of the recording equipment, sedimentary changes or unconformities
(Makogon 1981), velocity changes due to opalization (Laberg et al. 1998) or multiple arrival of a velocity contrast in the water layer (Pecher et al. 1996). However, when the depth of the BSR reflection approximates the predicted limit of the BGHS, most authors still consider that the BSR is related to gas hydrate formation. During further exploration into the analysis of the complexity of 'BSR occurrence areas', some authors have identified superimposed BSR. For example, Posewang & Mienert (1999) have reported the occurrence of double BSR on the margin west of Norway. These authors suggest that 'the upper double BSR may mark the top of gas hydrates and the lower double BSR may represent a relict of former changes of the gas hydrate stability field from glacial to interglacial times or the base of gas hydrates with a gas composition including heavier hydrocarbons' Furthermore, Foucher et al. (2002) demonstrate evidence and discuss the origin of a double BSR on the Nankai margin, off Japan. They interpret the uppermost BSR as an active methane hydrate related reflector and the deepest BSR as a residual hydrate-related reflector. This could record a recent migration of the base of the methane hydrate stability zone from the deepest BSR to the upper one. Sea-bottom warming and tectonic uplift could be possible explanations for this migration.
Fig. 1. Multibeam bathymetric map of the studied area, with profile Z2-30, profile Zl-01, ODP hole Leg 175 - site 1076, and salt diapir apex locations.
In this paper, the authors show seismic data over the Angola margin, where a discontinuous BSR is present at about 200 m bsf. A shallow BSR (SR) occurs at about 75 m bsf, within the gas hydrate stability zone. The SR reflector crosscuts the stratigraphic units, shows a reversed polarity with respect to the seafloor polarity and is in general parallel to the seafloor. A very shallow reflector (Rl) occurs at about 25 m bsf. It shows as well a reversed polarity with respect to the seafloor polarity. In the following, the authors study the uppermost hundreds of metres of sediments with respect to fluid flow, sediment compaction, gas and gas hydrate occurrence, with special emphasis on the SR and Rl reflectors. They also discuss possible interpretations to account for the origin and the nature of these reflectors.
General setting The Zaire Fan system is located on the mature passive continental margin that results from the opening of the South Atlantic Ocean (130 Ma) during the Early Cretaceous (Jansen et aL 1984). Along the margins of Angola, Zaire, Congo and Gabon the opening was followed during the Middle Aptian period, by a large accumulation of salt which generated diapiric structures (Emery et al. 1975; Jansen et al. 1984) under sediment loading. The investigated area is situated on the Gabon continental slope, between 1000 m to 2500 m of water depth, north of the Zaire canyon (Fig. 1). This part of the slope has an average dip of 3°. Evaporites deposited during the Aptian migrate and generate salt diapirs which, in some places, rise up to the sea-
SHALLOW BOTTOM-SIMULATING REFLECTORS ON THE ANGOLA MARGIN
193
Fig. 2. Stacked seismic section Z2-30.
floor. The Z2-30 seismic and echosounder 3.5 kHz profiles (Figs 2 and 3) presented in this paper show images of one of these sub-cropping diapirs.
Data acquisition and processing Acquisition and conventional processing of the seismic data Z2-30 seismic profile (Fig. 2) was shot during the Zaiango II cruise on board N/O FAtalante in November 1998. The seismic acquisition system was especially set up in order to acquire highresolution data in deep waters. The technical characteristics of the system that was used are the following: a 2.5 km long streamer composed of 96 channels, with a 25 m group interval, towed at 3 m immersion. The data were sampled at 1 ms and recorded using a SN358 lab. The source was composed of an array of four Generator-Injector (GI) guns and two mini GI guns, with a total volume of 217 inch3 for the generators. It was operated at a nominal pressure of 140 bars and immersion was set to 1.5 m. The shooting interval was 25 m, which leads to a maximum fold of 48. The conventional processing sequence that has been applied to the data includes SEGD format conversion, source delay correction (27 ms), band-pass filtering (25-300 Hz) and CDP gather formation. No amplitude corrections have been applied to the data in
order to preserve the relative amplitude changes on the profiles. To generate the stack sections (Fig. 2), the authors performed conventional semblance velocity analysis, normal move out corrections and stack. No post-stack migration was applied, in order not to alter the amplitudes of the data.
Acquisition and processing of the 3.5 kHz echosounder data The 3.5 kHz echosounder data was recorded using a hull mounted transducer, operated in chirp mode. The chirp signal is 20 ms long, with frequencies varying between 2.5 and 4.5 kHz. The data is bandpass filtered (1.5kHz-5.5kHz) and correlated with the source signal before being digitally recorded with a 24000 Hz sampling frequency. For display purposes (Fig. 3) the envelope of the signal is computed and low-pass (2000 Hz) filtered. No automatic gain control is applied to the data, which avoids blanking artefacts close to the seafloor reflection. This is high quality data and provides up to 120 m penetration into the sediment with a metric resolution.
Data description The multichannel seismic section is displayed in Figure 2, with zooms between CDP 400-800 and CDP 2250-2850 in Figures 4 and 5 respectively. The
194
H. NOUZE & A. BALTZER
Fig. 3. 3.5 kHz profile Z2-30 (top) and interpretation (bottom).
main structural feature is a salt diapir. It rises up to about 50 ms below the seafloor. The sedimentary layers are deformed near the diapir and are affected by closely-spaced faults, especially visible in the upper layers between 30 and 300 ms bsf. On the multichannel seismic data, the deeper reflectors (more than 700ms bsf) are rather continuous, but above 500 ms bsf the data show at least three unusual reflectors: (1)
On the multichannel data, between 200 and 300 ms bsf, a reflector parallels the sea floor, and shows a polarity reversed with respect to the sea floor polarity (Fig. 6). It cross cuts the sedimentary layers (Fig. 5) and can be traced through out the section, but is sometimes interrupted (at CDP 2500 for instance). Because of these characteristics, in the following this reflector will be termed the 'BSR'. The seismic facies above the BSR is characterized by horizons with lower amplitudes than when there is no BSR (compare amplitudes between CDP 1700-2400 and between CDP 2600-2800 for instance). The seismic facies below it is composed of high amplitude, discontinuous horizons with frequent diffractions. Assuming the interval velocities shown
in Figures 4 and 5 in the sediments when converted to depth the BSR, lies between 180 m bsf at 1450m water depth and 210 m bsf at 2000 m water depth. It is interesting to note that between CDP 1750-2500, the BSR coincides with the upper limit of high amplitude anomalies, whereas between CDP 300-600 (Figs 2 and 4), there is a gap about 30 ms wide between the BSR and the high amplitude anomalies. (2) At about 110 ms below the sea bottom, on the multichannel data, a faint but continuous and low apparent frequency reflector parallels the sea floor and crosses a well-layered unit deformed by normal faults. Its depth is about 80-90 m bsf, assuming an interval velocity of 1500 m/s in the upper 100 m of sediments. This reflector has a reversed polarity with regards to that of the sea floor (Fig. 6). Up-slope (west of the diapir), it can be defined by a double white/black full-wave, whereas down-slope (east of the diapir), it is weaker and presents just a simple white half-wave. Near the diapir it rises up close to the sea bottom. It can be correlated to the deepest series of reflectors on the 3.5 kHz data (Fig. 3). On these data, it marks the transition between a transparent or almost
SHALLOW BOTTOM-SIMULATING REFLECTORS ON THE ANGOLA MARGIN
195
Fig. 4. A detail of seismic profile Z2-30. From left to right: (1) stacked section with interval velocity superposed on the seismic data; (2) corresponding near common offset (285 m) section; (3) corresponding far common offset (2310 m) section. Due to the lower resolution of the far offset section, Rl is not visible anymore.
Fig. 5. A detail of seismic profile Z2-30. From left to right: (1) stacked section with interval velocity superposed on the seismic data; (2) corresponding near common offset (285 m) section; (3) corresponding far common offset (2310 m) section. Due to the lower resolution of the far offset section, Rl is not visible anymore. The common offset sections display preserved amplitudes. On the near offset section, note the contrast between blanked amplitudes (gas hydrate effect?) below the sea floor where BSR is present and stronger amplitudes both where BSR is absent and below the BSR (gas effect?). Note as well that SR clearly cross cuts sedimentary layer (top right).
(3)
transparent sequence above and acoustic masking below. It progressively disappears close to CDP 1800. On the 3.5 kHz data, this reflector has a shingled aspect. On account of this particular mounded aspect, we call this reflector the 'Sheep back reflector' (SR). At about 20 to 30 ms bsf (about 25 m bsf when converted to depth) on the multichannel
seismic data, a black continuous reflector (Rl) with good lateral continuity parallels the sea floor, except near the diapir where it approaches the sea bottom reflector. It shows a reversed polarity with regards to that of the sea floor (Fig. 6). East of the diapir, on the 3.5 kHz data, it can be correlated to a strong reflector that marks the base of a weak continuous
196
H. NOUZE & A. BALTZER
Fig. 6. Near offset traces for CDP 1300. Note the reversed amplitude for reflectors Rl, SR and BSR, and the relatively low absolute amplitude of these reflections. Note the rather week reflections above the BSR ('blanking effect?') and the strong anomalous reflections below the BSR (gas occurrence).
sequence of reflectors and the top of the transparent sequence mentioned above. West of the diapir, it can be correlated to the transition between the weak continuous sequence of reflectors and the top of the transparent sequence, but no strong reflector occurs in the 3.5 kHz data. The sea floor gently dips seaward (3°) over the whole length of the section displayed. On the 3.5 kHz data, as well as on the multichannel data, it is characterized by a well-marked continuous reflector affected by a series of small depressions. These depressions are more numerous on the part of the section east of the diapir. Directly overlying the diapir zone that rises through the sedimentary section from 4000 ms to 2490 ms, (Fig. 3, CDP 1390-1520) the sediments show a larger depression.
ODP site 1076 ODP site 1076, Hole 1076A, at 5° 4.1316' S - 11° 6.0917'E, near the SE end of profile Zl-01 (Fig. 1), penetrated 204 m below the sea floor and recovered 217.5 m of sediments. It has complete records of physical and geotechnical measurements throughout the length of the core (Fig. 7). Hole 1076 A is situated near the area of occurrences of the SR reflector, but fails to pass through it. The Zl-01 seismic profile has an aspect similar to that of the Z2-30
profile downslope from the diapir. A correlation of the physical properties of the sediment in the Hole 1076A and the acoustic facies of the 3.5 kHz line Zl-01 is shown in Figure 7. A synthetic description of the sediments recovered in the Hole 1076A is given below (Wefferef al 1998). The sediments in Hole 1076A had high gas content: natural gas analyses determined that much of this gas was CO2. Methane first appears in samples at 28.3 bsf and its concentration rapidly increases below 35m bsf. Core segments affected by voids and core disturbances are marked by a thick black line, on the right side of the log (Fig. 7). The curves for the physical properties can be divided into three sections: from the top to about 25 m; from 25 m to 90-100 m and from 100 m to the bottom of the core. In the upper part, a velocity value close to 1530 m/s is found whilst a slow velocity of 1500 m/s is measured around 13 and 22 m bsf. Density and porosity curves show a negative correlation, whereas the moisture content curve is parallel to the porosity curve: as the density increases from 1200 kg/m3 to 1400 kg/m3, the porosity and the moisture decrease from 85% to 75% and 70-57% respectively. All the curves show a significant change at about 25 m. Two velocity peaks (1550 m/s) occur at 75 m and 90 m. From 100 m to the bottom the thermal conductivity and the shear strength measurements are clearly perturbed by the voids, while the density and porosity curves do not show significant changes.
SHALLOW BOTTOM-SIMULATING REFLECTORS ON THE ANGOLA MARGIN
197
Fig. 7. ODP Hole 1076A physical properties logs and 3.5 kHz section of the Zl-01 profile. 3.5 kHz section is situated to the East of Hole 1076A, where the SR reflector exists although it is not well developed.
Methods of data analysis
All the studied CDPs in Figure 8 show the same evidence of a phase inversion, except for CDP 1200 and 1028 where it is less clear. A closer look at the sea bottom signature for these two localities near the Seismic polarization analysis top of the diapir reveals that the sea bottom signal is Figure 6 shows near offset traces on the multichan- reversed when compared to the sea bottom signal at nel data from CDP 400, upslope from the diapir, the other places. with Rl, SR and the BSR marked by arrows. The main point to note here is that these three reflectors show a phase inversion. A reversed polarity on the Migration velocity analysis near normal incidence seismic data traduces a decrease of the seismic impedance. Thus, it marks a In order to obtain the stacked seismic section (Fig. 2), a velocity model was built by standard semblance decrease in P-wave velocity or/and in density. Thanks to the digital recording of the 3.5 kHz data, velocity analysis using the Promax software. This the authors were able to conduct a polarization analy- analysis provides root mean square velocities in sis of the recorded 3.5 kHz signal at several locations time. To refine the model and access interval velocity in (Fig. 8). It consisted of superimposing the sea floor signal (Fig. 8: top frame) on the signal at a given depth, depth migration velocity analysis was underreflector (Fig. 8: middle frame). In order to illustrate taken on selected CDPs. It was expected to be able to phase inversions, the signal of the studied reflector interpret: was reversed and plotted together with the sea floor signal (Fig. 8: bottom frame - sea floor in grey, (1) The presence of free gas in the sediments from low velocity values because a small amount of reflector in black). A good correlation between these free gas is responsible of a large decrease of the two signals indicates that there is a phase inversion, P-wave velocity (Domenico 1976) which in turn traduces the occurrence of a sediment (2) the occurrence of gas hydrates from high layer characterized by a lower acoustic impedance. velocity anomalies. It should be noted that the source signal could be slightly different for each shot. However, for a single shot, it is considered that the comparison of the An iterative method was used that can be divided into three phases for each iteration: (a) migration phase: it phase between two different reflectors is valid.
198
H. NOUZE & A. BALTZER
Fig. 8. Phase analysis of the 3.5 kHz signal. The bottom signal is shown in the top frames, the SR signal in the middle frames, and the superposition of the bottom signal and the reversed SR signal in the bottom frames.
consists of building depth migrated common reflection point (CRP) gathers with a set of perturbed depth interval velocity models around a central velocity model; (b) picking phase: the gathers are stacked to build a semblance panel and the maximum semblance is picked (Fig. 9); (c) velocity model update: the pick in the semblance panel is converted into velocity corrections. The velocity model is updated and used to start the next iteration. As for conventional RMS velocity analysis, the theoretical resolution of this velocity analysis method depends on a large set of parameters (Yilmaz 1987). On real data, the velocity resolution can be empirically estimated as the half width of the peaks on the semblance panel. For the analysis displayed in Figure 9, it is estimated to be better than 10 m/s at 2000 m bsf. The main parameter that controls the vertical resolution is for this data, the vertical distance between reflectors that yield good focusing on the semblance panel. Due to the highly faulted sequence above the BSR and the low signal-to-noise ratio below it, it was possible to select an event every 30-40 m on average. Thus, the velocity analysis method does not solve for thin layers. Migration velocity analysis was conducted of several CDP localities and the results are superimposed on the seismic data on Figures 4 and 5. A result common to all the analyses is that interval velocities in the sediments are almost constant and not significantly different from the water velocity, within the resolution of the velocity analysis, down
to the BSR. Such low velocities in the sediments are quite surprising, but are in good agreement with the values obtained at the Hole 1076A drilling site. Sediments at that locality consist of under compacted clays with very high porosity (75-85%) and subsequently low P-wave velocity (1520-1560 m/s). In Figure 4, the results of analyses performed in the up-slope part of the profile with regard to the diapir are presented, on both sides of CDP 600, where the BSR vanishes. Two possible weak (40 m/s) velocity inversions are seen, corresponding to reflectors Rl and SR. A weak inversion is detected where the BSR is observed, but is undetected where there is no BSR. Below the BSR, velocities start to increase. In Figure 5, the analysis performed on the down slope part of the profile with regard to the diapir, on both sides of the BSR disappearance (CDP 2500) are shown. An almost constant velocity below the sea floor is shown with faint inversions at the Rl and BSR. No inversion is detected with the inversion procedure at the BSR level.
AVO analysis Amplitude Variation with Offsets (AVO) has been extensively used to study BSR (Hyndman & Spence 1992; Andreassen et al. 1997; Ecker et al 1998; Tinivella & Accaino 2000). In this paper, reflection coefficients are not inverted but their variation with offset are extracted
SHALLOW BOTTOM-SIMULATING REFLECTORS ON THE ANGOLA MARGIN
199
Fig. 9. Migration velocity analysis. From left to right: (1) semblance panel; (2) corrected CRP gather with the lowest velocities; (3) corrected CRP gather with the baseline velocities; (4) corrected CRP gather with the highest velocities
from the data to yield and quantify physical parameters (P-wave velocity, S-wave velocity, density). Instead these coefficients as well as their variations with offset are used to support or invalidate an hypothesis. To estimate the reflection coefficients for the studied reflectors, a sequence that was described in Ecker (1999) was followed. (1) The amplitude of both the sea bottom and the reflectors on the prestack CDP data are selected, with no amplitude correction. (2) The sea floor amplitudes are corrected for spherical divergence attenuation are corrected. (3) The AVO sea floor response from the velocity model, the zero offset sea floor reflection coefficient and the density curve in the ODP logs is modelled. (4) The source and receiver amplitude corrections are derived by dividing the calculated sea floor response by the sea floor amplitudes. (5) For a given reflector and for each offset, a ray tracing algorithm is used to obtain the values of the incidence angles for the reflections at the studied reflector, the transmission losses at the interfaces of the input velocity model, as well as the attenuation due to energy absorption in the sediments. (6) The spherical divergence corrections, source and receiver amplitude corrections, attenuation corrections and transmission corrections are applied to the amplitudes of the studied reflectors. The sea floor zero offset reflection coefficient was calculated by dividing the divergence corrected amplitude of the sea floor by the one of the sea floor first multiple. This calculation (R = 0.12) was com-
pared to and shows good agreement with the result obtained using the ODP density values and the velocity from the velocity analyses. P-wave attenuation Qp was set to a value of 80, to account for high attenuation in near surface sediments (Ayres & Theilen 2001). S-wave velocities for the sediments lacking gas or gas hydrates were estimated according to the formulas given by Hamilton (Hamilton 1976) for water-saturated silty clays and turbidites (Table 1). The P-wave velocities of these sediments were extracted from the ODP logs, assuming that these data represent sediments without gas or gas hydrate (Table 1). The porosity was extracted from the ODP logs. Then, P- and S-wave velocities were compared for partially gas-saturated sediments using the equations of Biot (1956) as quoted by Domenico(1977): Table 1. P-wave, S~wave velocities at Rl and SR levels used for the AVO calculations
Depth bsf (m) P-wave vel.(m/s) S-wave vel (m/s)
Rl
SR
Data from
25 1515 210
100 1550 380
Z2-30 seismic section ODP Log Hamilton formulas (*)
(*) Hamilton formulas Vs= 116 + 4.65 z Vs = 302+1.28.(z-40) Vs-404 + 0.58 (z-120)
(0
200 Table 2. Symbol
H. NOUZE & A. B ALTZER Parameters used to evaluate velocities for porous and partially gas saturated sediments Parameter
Cs Cw Gg CpO m M-
Compressibility of solid grains Compressibility of water Compressibility of methan Pore compressibility at zero differential pressure Pressure gradient of pore compressibility Shear modulus of rock matrix
h z
Water depth Depth of horizon below seafbed
ps pwO pw
Sediment grain density Density of seawater at sea surface Density of seawater at horizon level
Pg pav
Density of methane at horizon level Mean density of rock above horizon
k
Coupling factor
g Cp Cb Cf pb Pf Sw Sg Gt TO
Acceleration due to gravity Pore compressibility Compressibility of rock matrix Compressibility of pore fluid Bulk density of rock Pore fluid density Water saturation Gas saturation Temperature gradient Bottom water temperature Porosity
R Pw MCH4
Pd 0 T
Constant Pore pressure Methane molar weight Differential pressure Cs/Cb Temperature at horizon depth
In these equations, the shear modulus of the rock matrix (jx) and the coupling factor (k) were adjusted to fit the P- and S-wave velocity values for the 0% gas saturation. The parameters for these calculations are shown in Table 2. The velocity variations with gas saturation, as well as the velocity variations with porosity, are displayed in Fig. 10(a) and (b) respectively. To estimate the velocities for partially-hydrated sediments the approach of Minshull et al. (1994) was used: the time average equation was applied to determine Vh, velocity for fully hydrated sediments
Value
Units 11
3.0. 104.19. 10~10 4.24 . 10-8 5.10~9 -7.5.10-17 SR: 1.9.108 Rl:6.0.107 1450.0 SR:90 Rl:25 2500 1030.0 pwO * exp(10.347063.10~14* Pw- 4.3577. 10~ 22 *Pw 2 ) MCH4*Pw / R * T SR: 1380 Rl: 1350 SR: 15 Rl: 15 9.81 CpO + m * P d (l-4>)*Cs + c/>*Cp S w * C w + Sg*Cg c£*pf+(l-(/>)*ps Sw * pw + Sg * pg
Pa-1 Pa"1 Pa-1 Pa"1 Pa"2
Pa m m kg/m3 kg/m 3 kg/m3 kg/m 3 kg/m3
m/s2 Pa"1 Pa"1 Pa'1 kg/m3 kg/m3
% % K/m K %
0.064 277.1 SR: 0.75 Rl:0.76 8.32 p w O * g * ( h + z) 0.016 (pav - pw) * g* z
Pa kg Pa
T = TO + z * G t
K
Pa . m3 / K . mol
and then interpolated between Vw velocity for water-filled sediments with: 1/V = Sw/Vw + Sh/Vh (Sw and Sh are the relative proportion of water and hydrate in the pore space). Vs, Vp and p for gas hydrates were set to 1700 m/s, 3500 m/s and 913 kg/m3 respectively. The AVO responses have been computed using the full Zoeppritz equations (Zoeppritz 1919). Analysis was undertaken on both sides of the diapir (CDP 500 and CDP 2300) for the Rl and SR reflectors. The main results for the analysis on CDP 500 are presented in Figure 11. Reflection coefficients at zero offset for Rl and SR are —0.06 and —0.08 respectively. For both Rl and SR, the AVO responses were tested for: (1) normal sediments above under consolidated sediments, i.e. a porosity increase (76-82%
SHALLOW BOTTOM-SIMULATING REFLECTORS ON THE ANGOLA MARGIN
201
Fig. 10. Variation of computed seismic velocities with gas saturation (10-1) and porosity (10-2) using the Biot model.
Fig. 11. Computation of SR and Rl reflection coefficients with angle of incidence using the Zoeppritz equations at CDP 500. Dashed lines: sediments with 20% gas hydates over normal sediments; solid lines: normal sediments over under compacted sediments; dash-dot lines: normal sediments over sediments with 0.5% gas.
for Rl, in agreement with the ODP porosity curve, 75-80% for SR) and the associated velocity changes (Fig. 10(b)); (2) sediments with 20% gas hydrates in the pore space above normal sediments; (3) normal sediments above sediments containing 0.5% of free gas and the associated velocity changes (Fig. 10(a)). For the SR reflector (Fig. ll(b)), the tests show: (1) that a 5% porosity decrease below SR cannot account for the observed reflection coefficients; (2) that a large amount (50% of the pore space) of gas hydrate would be necessary to account for the observed reflection coefficients; (3) that a small amount of gas in the sediments (0.5% of the pore space) is sufficient to explain the observed reflection coefficients. For the large incidence angles and due to the decreasing signal to noise ratio, the reflection coefficients extracted from the data might be underestimated.
For the Rl reflector (Fig. 11 (a)), the tests show: (1) that the 6% porosity decrease observed on the ODP log below Rl can account for the Rl reflection coefficients; (2) that a large amount (50% of the pore space) of gas hydrate would be necessary to account for these reflection coefficients; (3) that even a small amount of gas in the sediments (0.5% of the pore space) would generate reflection coefficients larger than the observed reflection coefficients.
Discussion In the following section, facies and amplitude anomalies detected on the section in relation to sediment compaction, gas and gas hydrate occurrence are discussed and possible interpretations that can account for Rl and SR characteristics are considered.
202
H. NOUZE & A. BALTZER
Sea floor and faults On the sea floor imagery obtained from the EM 12 multibeam echosounder data, the sea floor depressions on the 3.5 kHz and multichannel data correspond to circular high reflectivity anomalies and are identified as pockmarks (Cochonat, pers. comm.). These pockmarks are interpreted as evidence of fluid flows of deep origin crossing the HSZ and reaching the seabed. These fluids most probably contain gas, although, with the resolution of the analysis presented here, there is no evidence (low velocity or amplitude anomalies) for substantial gas accumulation between the BSR and the sea floor. The normal faults which affect the sediments up to about 700 ms bsf, when mapped in 3-dimensions on industry 3D seismic data are arranged in polygonal patterns in plan view and could be attributed to volumetric contraction during compactional dewatering, as is observed by Dewhurst et al (1999) in the North sea basins (Sultan, pers. comm.).
High amplitudes as a hint for free gas in the sediments When gas is present in the sediment, it can be trapped preferentially in higher permeability sediment layers. Such a phenomenon can account for the anomalously high amplitude reflections observed immediately underneath the BSR (Figs 2, 4 and 5). Diffraction hyperbolas, strong acoustic attenuation and strong amplitude anomalies below the BSR (Fig. 4) confirm that there is most probably an accumulation of free gas below the BSR. Close to the sea floor depression on top of the diapir, high amplitudes and diffraction hyperbolas may indicate gas above the BSR, in a limited area.
Amplitudes blanking as a hint for gas hydrate loaded sediments The amplitude blanking that is observed on the multichannel seismic section (Fig. 5) could be related to a scattering of energy associated with the many shallow faults in the sedimentary layers above the BSR. However, on the multichannel data, there is a strong relationship between the occurrence of significant blanking and the presence of the BSR (Fig. 2 between CDP 2250 and CDP 2750), with no visible change in the degree of faulting of the sedimentary layers. On the one hand, the results from the ODP Drilling Leg 164, on the Blake Ridge (Holbrook et al. 1996; Paull et al. 2000) were interpreted as the blanking above the BSR that had nothing to do with gas hydrates. However, on the other hand, Lee & Dillon (2001) have developed models to explain
amplitude blanking in relation with pore filling of gas hydrates in the sediments. According to their models, significant blanking can occur within low velocity sediments with a relatively low gas hydrate concentration. Thus, it is proposed that the occurrence of blanking on the seismic profile strongly correlated with the presence of the BSR, together with the absence of any significantly high velocity anomaly. This could correspond to a large (more than 10% of the pore space) amount of gas hydrate in the sediments above the BSR and could be in favour of a limited and dispersed gas hydrate concentration above the BSR in the parts of the profile where blanking is observed.
BSR Temperature and heat flow measurements conducted during the Zai'ango II cruise gave a water bottom temperature of 4.2°C at 1400 m water depth, and an average geothermal gradient of 64 mK.m"1. With these values, the base of the gas hydrate stability zone (BGHZ), computed from the stability curve for methane hydrates in pure water, using the formula ln(P) = 29.112 - 7694.3 / T, where P is pressure in Mpa and T temperature in Kelvin (Istomin & Yakushev 1992) is about 300 m. Laberg etal (1998) give a P-T diagram for the stability of gas hydrate in pure- and seawater and for different gas composition. On this diagram (Fig. 12), the values for our conventional BSR (CDP 500 and 2250) fall slightly below the curves for pure methane in seawater and far below higher hydrocarbon models. However, brine composition in the sediments, methane concentration, heavy gas composition (geothermal gas coming from deep sources in the diapir area is highly possible) as well as thermal gradient anomalies (as a result of higher thermal conductivity in the diapir and fluid circulations) could lower the value for the BGHZ. Xu & Ruppel (1999) have developed a model to predict the distribution and evolution of gas hydrates in porous marine sediments. In this model, together with pressure and temperature conditions, the authors take into account the mass fraction of methane in the seawater, the solubility of methane and the methane flux. As a result of the model, a distinction needs to be made between: the sea floor/top of the gas hydrate stability zone (TGHS) where gas hydrate is stable but not present; the top of the gas hydrate occurrence zone (TGHZ) where gas hydrate is stable and actually present; the bottom of the gas hydrate occurrence zone (BGHZ); the bottom of the gas hydrate stability zone (BGHS); and the top of the free gas zone. Depending mainly on the variations of the gas solubility and concentration with depth, the
SHALLOW BOTTOM-SIMULATING REFLECTORS ON THE ANGOLA MARGIN
203
Fig. 12. Stability of gas hydrate in pure and sea water and for different gas composition. Modified from Laberg et al. (fig 5, Laberg et al. 1998). The positions of the conventional BSR for CDPs 500 and 2250 are reported as stars on the diagram and could indicate that the gas involved in the gas hydrate formation is mainly methane.
TGHS and the TGHZ, as well as the BGHZ, the BGHS and the free gas zone may not coincide. On the section presented here, the BSR lies between about 180 and 210m bsf and it can be interpreted as the top of the free gas zone, or as the BGHZ. Between CDP 400 and 600, the BSR does not coincide with the occurrence of high amplitude anomalies (that are interpreted as an evidence for free gas in the sediments) and could then be attributed to the BGHZ. Whereas between CDP 1750-2500 it coincides with the high amplitude anomalies and would then mark the top of free gas zone (and possibly the BGHZ which could coincide at this locality).
The cause of the SR reflector From the polarization analysis, it is clear that the SR reflector traduces a decrease in acoustic impedance. The AVO analysis shows that a small amount of gas in the sediments (0.5% of the pore space) is sufficient to explain the observed reflection coefficients. Taylor et al. (2000) observed that when gas-charged strata dip relative to the BGHS, BSR may display a shingled appearance on high resolution seismic profiles. This aspect is caused by slight variation of gas concentration across stratification boundaries that terminate against the BGHS. In Hole 1076A free gas occurrence is supported by the fact that severe core disturbances and large voids from gas release were
204
H. NOUZE & A. B ALTZER
observed from 80 m to the bottom core in the Hole 1076A. From the AVO results, the shingled aspect of SR, as well as gas occurrence in Hole 1076A, it is proposed that a small quantity of free gas exists at the SR level within the gas hydrate stability zone. The SR reflector crosses the stratigraphy, which means it cannot be attributed to a sedimentary change, but more probably to a physical/thermodynamic boundary. As previously mentioned, the sea floor/top of the gas hydrate stability zone (TGHS) and the top of the gas hydrate occurrence zone (TGHZ) may not coincide. When accounting for double BSRs, Posewang & Mienert (1999) and Foucher et al (2001) propose that a second BSR is generated at the BGHZ because of temperature and/or pressure condition changes at the BGHZ. In a similar way, the authors propose that the shallow BSR (SR) results from temperature or pressure conditions changes at the TGHZ. It appears that the TGHZ does not coincide with the sea bottom, that gas hydrate dissociation occurs at the TGHZ and that the free gas released in the sediments during that dissociation generates SR. Such a scenario is hypothetical. More chemical and physical analysis and modelling need to be conducted to validate or invalidate these hypotheses. The cause for the Rl reflector From the polarization analysis, it is clear that the Rl reflector traduces a decrease in acoustic impedance. On the ODP 1076A density log (Fig. 7), there is an increase in density and decrease in porosity coherent with sediment compaction, from the sea floor up to about 20-25 m bsf, where both logs show a change towards lower density and larger porosity. Then the density increases and the porosity decreases with depth again less rapidly than in the upper 25 m. The AVO tests show that this porosity change at about 20-25 m bsf accounts for the Rl reflection coefficients. Thus, the AVO tests support the following hypothesis: the upper 25 metres of sediment are normally compacted and overlie under-compacted sediments. Furthermore, core disturbance occurs at 23.8 m bsf until 45 m bsf in the Hole 1076A, which is in agreement with soupy (highly water-saturated) sediments in this interval. Uenzelmann-Neben (1998) identifies a reflector in a nearby area (reflector CF-A-int) and correlates this reflector with a minor unconformity related to a maximum of upwelling conditions in Late Pliocene times. The CF-A-int reflector shows the same characteristics as Rl. It is proposed that the change in the sedimentation processes mentioned above, is accompanied by a change in the sediments perme-
ability. Rl would then either be the result of a different behaviour of the sediments above and below Rl with regard to compaction, or a permeability barrier which would accumulate overpressure generated below Rl. This is attributed to sediment compaction (and creation of the polygonal fault system), fluid flow below Rl, or gas hydrate dissociation at the SR level, as discussed above.
Conclusions In this paper, careful analysis of high resolution seismic data was undertaken. Polarization, amplitude velocity and AVO analysis were used to determine the distribution of gas and gas hydrate in the sediments, in an area with complex structural, thermal and fluid flow environment. These analyses allow an interpretation of the features observed. At some localities, a discontinuous bottom-simulating reflector (BSR) could mark the base of the gas hydrate occurrence zone, and/or the top of the free gas zone at other positions. There is evidence for limited gas accumulations both below and within the gas hydrate stability zone. A shallow BSR (SR), crosses the stratigraphic units at about 70 m below the sea floor and is underlain by a small amount of free gas. It could mark the top of the gas hydrate occurrence zone, where gas hydrate dissociation may occur. At 25 m below the sea floor, a reflector (Rl) could mark a limit between normally compacted and under compacted sediments: possibly overpressure generated during gas hydrate dissociation. Publications concerning double BSR occurrence are still uncommon, but recent work reports such observations, for example that of Posewang & Mienert (1999) and Foucher et al (2001). New data, with increasing resolution, as well as careful interpretation of high-resolution data, would probably lead to new discoveries. As far as gas hydrate occurrence in the sediments studies have progressed, it becomes increasingly clear that a model with sediments homogeneously filled with gas hydrates overlying gas-loaded sediments, where gas is trapped by the hydrates sealing the sediments, proves to be far too simple. Ever more complex models for gas hydrate formation and dissociation, taking into account methane conservation, heat transfer, pore distribution and pore water chemistry in the sediments, predict that gas and gas hydrate distribution in the sediments, is a dynamic and discontinuous phenomenon. High resolution seismic data can help in our understanding of the distribution of gas and gas hydrates in the sediments, in in-situ conditions, thus validating or invalidating the proposed models.
SHALLOW BOTTOM-SIMULATING REFLECTORS ON THE ANGOLA MARGIN The authors are grateful to the officers and crew from the N/O FAtalante for their co-operation during data acquisition, to the scientific crew of the Zai'ango II cruise, and to P. Cochonat and N. Sultan (Ifremer) for their helpful comments on our study. Many thanks to U. Tinivella, L. De Santis and K. Andreassen for their helpful review of the manuscript. The data presented in this paper were collected in the framework of the Zai'ango (Ifremer - TotalFinaElf) project. 3.5 kHz and seismic displays as well as the migration velocity analyses were made using the Sithere software (Ifremer).
References ANDREASSEN, K., HART, RE. & MACKAY, M. 1997. Amplitude versus offset modelling of the bottom simulating reflection associated with submarine gas hydrates. Marine Geology, 137,25-40. AYRES, A. & THEILEN, F. 2001. Preliminary laboratory investigations into the attenuation of compressional and shear waves on near-surf ace marine sediments. Geophysical Prospecting 49 (1), 120-127. BANGS, N. L., SAWYER, D. S. & GOLOVCHENKO, X. 1993. Free gas at the base of the gas hydrate zone in the vicinity of the Chile Triple Junction. Geology, 905-908. BIOT, M.A. 1956. Theory of propagation of elastic waves in a fluid-saturated porous solid. Journal of Acoustics Society America, 28,168-191. DEWHURST, D.N., CARTWRIGHT, J.A. & LONEGARN, L. 1999. The developement of polygonal fault systems by syneresis of colloidal sediments. Marine and Petroleum Geology, 16,793-810. DOMENICO, S.N. 1977. Elastic properties of unconsolidated porous sand reservoirs. Geophysics, 42 (7), 1339-1368. DOMENICO, S.N. 1976. Effect of brine-gas mixture on velocity in an unconsolidated sand reservoir. Geophysics, 41, 882-894. ECKER, C. DVORKIN, J. & NUR, A. 1998. Sediments with gas hydrates: Internal structure from seismic AVO. Geophysics,^, 1659-1669. ECKER, C. 1999. Standford Exploration Project sep96. EMERY, K.O., UCHUPI, E., PHILLIPS, J., BOWIN, C., MASCLE, J. 1975. Continental Margin off Western Africa: Angola to Sierra Leone. American Association of Petroleum Geologists Bulletin, 59,2209-2265. FOUCHER, J.P., NOUZE H. & HENRY, P. 2002. Observation and tentative interpretation of a double BSR on the Nankai slope. Marine Geology, 187,161-175. HAMILTON, E. L. 1976. Shear-wave velocity versus depth in marine sediments: a review. Geophysics, 41, 985-996. HOLBROOK, W.S., HOSKINS, H., WOOD, W.T., STEPHEN, R.A., LIZZARALDE D. & LEG 164 Science Party. 1996. Methane hydrate and free gas on the Blake Ridge from vertical seismic profiling. Science, 273, 1840-1843. HOVLAND, M., FRANCIS, T.J.G., CLAYPOOL, G.E. & BALL, M.M. 1999. Strategy for scientific drilling of marine gas hydrates, JOIDES journal, 25 (1), 20-24. HYNDMAN, R. D. & SPENCE, G.D. 1992. A seismic study of methane hydrate marine bottom simulating reflectors. Journal of Geophysical Research, 97, 6683-6698.
205
ISTOMIN, V.A. & YAKUSHEV, V.S. 1992. Gas hydrate at natural conditions. Nedra Publishers Moscow, 236p. JANSEN, J.H.F., VAN WEERTNG, T.C.E., GIELES, R., VAN IPEREN, J. 1984. Late Quaternary oceanography and climatology of the Zaire-Congo fan and adjacent eastern Angola Basin. Netherland Journal of Sea Research, 17,201-249. LABERG, J.S., ANDREASSEN, K. & KNUTSEN, S.-M. 1998. Inferred gas hydrate on the Barents Sea shelf - a model for its formation and a volume estimate. GeoMarine Letters, 18,26-33. LEE, W.M. & DILLON, W.P. 2001. Amplitude blanking related to the pore filling of gas hydrate in sediments. Marine Geophysical researches, 22 (2), 101-109. MAKOGON, Y.E 1981. Hydrate of natural gas, translated from Russian by CIESLEWICZ, W.J. PennWell, Tulsa, OK.,237p. MINSHULL, T.A., SINGH, S.C., & WESTBROOK, G.K. 1994. Seismic velocity structure at a gas hydrate reflector, offshore western Colombia, from full waveform inversion. Journal of Geophysical Research, 99, 4715-4734. PAULL, C.K., MATSUMOTO, R., WALLACE, P.J. & DILLON, W.P. (eds) 2000. Proceedings. ODP, Scientific. Results, 164: College Station, TX (Ocean Drilling Program). PAULL, C.K., MATSUMOTO, R., WALLACE, PJ. et al 1996. Proceedings ODP Initial Reports, 164: College Station, TX (Ocean Drilling Program). PECKER, LA., MINSHULL, T.A., SINGH, S.C. & VON HUENE, R. 1996. Velocity structure of a bottom simulating reflector offshore Peru. Results from full waveform inversion. Earth and Planetary Sciences Letters, 139, 459-469. POSEWANG J. & MIENERT J. 1999. The enigma of double BSRs: indicators for changes in the hydrate stability field? Geo-marine letters, 19,157-163. RUPPEL, C. 1997. Anomalously cold temperatures observed at the base of the gas hydrate stability zone on the US Atlantic passive margin. Geology, 25,699-702. TAYLOR, M.H., DILLON, W.P & PECKER, LA. 2000. Trapping and migration of methane associated with the gas hydrate stability zone at the Blake Ridge Diapir: new insights from seismic data. Marine Geology, 164,79-89. TINIVELLA, U., LODOLO, E., CAMERLENGHI, A. & BOEHM, G. 1998. Seismic tomography study of a bottom simulating reflector off the South Shetland Islands (Antarctica). In: HENRIET, J.-P. & MIENERT, J. (eds), Gas hydrates: relevance to world margin stability and climate change. Geological Society, London, Special Publications. 141-151. TINTVELLA, U. & ACCAINO, F. 2000. Compressional velocity structure and Poisson's ratio in marine sediments with gas hydrates and free gas by inversion or reflected and refracted seismic data (South Shetland, Antarctica). Marine Geology, 165,13-27. UENZELMANN-NEBEN, G. 1998. Neogene sedimentation history of the Congo Fan. Marine and Petroleum Geology, 15, 635-650. WEFFER, G., BERGER W.H., RICHTER, C. et al 1998. Proceedings ODP, Initial Reports, 175, 87-115. Xu, W. & RUPPEL, C. 1999. Predicting the occurrence, distribution and evolution of methane gas hydrates in
206
H. NOUZE & A. BALTZER
porous marine sediment. Journal of Geophysical Research, 104,5081-5096. YILMAZ, O. 1987. Seismic data processing. Investigation in geophysics, vol. 2. Society of Exploration Geophysicists. 512 p. ZATSEPINA, O. YE. & BUFFET, B.A. 1998. Thermodynamic
conditions for the stability of gas hydrate in the seafloor.Journal of Geophysical Research, 103, 24,127-24,139. ZOEPPRTTZ, K. 1919. Erdbebenwellen VfflB, On the reflection and penetration of seismic waves through unstable layers: Goettinger Nachr., 66-84.
Near-surface sediment mobilization and methane venting in relation to hydrate destabilization in Southern Lake Baikal, Siberia PIETER VAN RENSBERGEN1'6 JEFFREY POORT1, ROLF KIPFER2, MARC DE BATIST1, MAARTEN VANNESTE1'6, JEAN KLERKX3, NICK GRANIN4, OLEG KHLYSTOV4 & PETR KRINITSKY5 l
Renard Centre of Marine Geology, University of Gent, Gent, Belgium (e-mail: pieter_yanrensbergen @yahoo. com) 2 EAWAG, Dubendorf, Switzerland ^Department Geology and Mineralogy, Royal Museum of Central Africa, Tervuren, Belgium 4 Limnological Institute, Siberian Branch of the Russian Academy of Sciences, Irkutsk, Russia 5 VNHOkeangeologia, St. Petersburg, Russia ^Present address: Department of Geology, University ofTroms0, Norway Abstract: Four seeps and mud extrusion features at the lake floor were discovered in August 1999 in the gas hydrate area in Lake Baikal's South Basin. This paper describes these features in detail using side-scan sonar, detailed bathymetry, measurements of near-bottom water properties, selected seismic profiles and heat flow data calculated from the depth of the hydrate layer as well as obtained from in situ thermoprobe measurements. The interpretation of these data is integrated with published geochemical data from shallow cores. The seeps are identified as methane seeps and appear as mud cones (maximum 24 m high, 800 m in diameter) or low-relief craters (maximum 8 m deep, 500 m in diameter) at the lake floor. Mud cones (estimated to be approximately 50-100 ka old) appear to be older than the craters and have a different structural setting. Mud cones occur at the crest of rollover structures, in the footwall of a secondary normal fault, while the craters occur at fault splays. The seeps are found in an area of high heat flow where the base of the gas hydrate layer shallows rapidly towards the vent sites from about 400 m to about 160 m below the lake floor. At the site of the seep, a vertical seismic chimney disrupts the sedimentary stratification from the base of the hydrate layer to the lake floor. Integration of these results leads to the interpretation that focused destabilization of gas hydrate caused massive methane release and forced mud extrusion at the lake floor and that the gas seeps and mud diapirs in Lake Baikal do not have a deep origin. This is the first time that methane seeps and/or mud volcanoes associated with gas hydrate decomposition have been observed in a sub-lacustrine setting. The finding suggests that gas hydrate destabilization can create large pore fluid overpressures in the shallow subsurface (<500 m subsurface) and cause mud extrusion at the sediment surface.
Lake Baikal is the only fresh-water basin with demonstrated presence of gas hydrate in the subsurface (Kuzmin etal. 1998). Presence of gas hydrates in the subsurface was also inferred from detailed seismic experiments (e.g. Golmshtoketal. 2000; Vanneste et al. 2001). In the summer of 1999, several gas seeps and mud extrusion features were discovered in Lake Baikal's South Basin. In this paper, we use side-scan sonar mosaics from a detailed study area, in combination with echo-sounding data, CTD casts and a number of seismic profiles to investigate the nature of the expelled fluids and gases, the morphology of the vent sites and their structural setting in relation to the thickness distribution of the hydrate-bearing section and lateral heat flow variations. The study suggests that the seeps in the southern part of Lake Baikal are gas seeps that most probably originate from methane release by hydrate decomposition at the base of the hydrate stability zone and infer that
gas release in the shallow subsurface can cause extrusion of lacustrine mud at the lake floor. In the Southern Baikal Basin (SBB) and Central Baikal Basin (CBB) extensive gas hydrate accumulations occur in the subsurface (Fig. la). The presence of gas hydrate in Lake Baikal was first inferred by Golmshtok et al. (1997) on the basis of the observation of a 'bottom-simulating reflection' (BSR) on multi-channel seismic profiles. The BSR was interpreted to mark the base of the hydrate stability zone (BHSZ), which was estimated from the seismic data to be between 35 m and 450 m thick (Golmshtok et al. 1997). The BSR could be observed in the area around the Selenga River delta (Fig. la) at water depths exceeding 580 m, i.e. on the deeper delta slope and the generally flat basin floors of the adjacent sub-basins. The first samples of gas hydrate in Lake Baikal sediments were brought to the surface in a 'Baikal
Fig. 1. A. Structure of Lake Baikal (inset) and the Selenga delta region, distribution of hydrate occurrence in the Baikal South Basin and Central Basin mapped on basis of a regional seismic grid (Golmshtok et al. 1997). B. Heat flow distribution in Lake Baikal based on thermoprobe measurements along regional transects (after Golubev 2000).
Drilling Project' (BDP) core in 1997 (Kuzmin et al 1998). The 224 m long core was taken in 1427 m water depth and the hydrates were found in coarse sandy turbidites at 121 m and 161 m sub-bottom depth. The hydrates in the collected samples occupied approximately 10% of the pore space (Golubev pers. comm. in Vanneste et al. 2001). The total volume of hydrate in Lake Baikal was calculated based on the above observations to be about 7.6 X109 m3, which represents about 1.24 X 1012 m3 of methane at STP conditions (Vanneste et al. 2001). Chromatographic analysis of the gases emitted by the BDP samples showed that methane was the only hydrocarbon in the sample representing >99% of the total gas volume (Kuzmin et al. 1998; Kuzmin et al. 2000). Carbon isotopic composition (S13C) ranges between 58 %o and —68 %o (Kuzmin et al. 1998; Kuzmin et al. 2000), indicating a bacterial origin via methanogenesis (Kvenvolden 1998; Sloan 1998). Hydrate distribution in Lake Baikal is limited to the area around the Selenga delta and the gas of the gas hydrate is most probably derived from organic matter supplied by the Selenga River, the main source of terrigenous organic matter to the lake. Sedimentation rates at the basin floor in Lake Baikal are in the order of magnitude of 0.3 mm/yr; the organic carbon in the lake floor sediments ranges from 2-4% for interglacial sediments to about 0.2% for glacial-age sediments (Colman et al. 1996).
Under such conditions methane generation from organic diagenesis can form hydrate in situ (Sloan 1998), but most likely the methane concentration in the hydrate stability zone was enriched over geological time with methane from hydrate decomposition at the base of the hydrated sediment layer due to continued sedimentation ('methane recycling'; Paull et al. 1994). Baikal water is pure and characterized by high oxygen concentrations throughout the water column (Kipfer et al. 1996) and very low concentrations of dissolved solids (Faulkner et al. 1991) and it is under-saturated for carbonates. As a result there are no authigenic or biogenic carbonates formed or preserved in Lake Baikal (Colman et al. 1996). More than 50 hot springs have been mapped in the Baikal Rift, mainly along the eastern margin of the Northern Baikal Basin (NBB) and the CBB (Pinneker & Lomonosov 1973). Hydrothermal activity is associated with zones of high geothermal heat flux related to basement fault zones. The water temperature in onshore hot springs ranges from 40 °C to 80 °C. At an offshore hydrothermal vent site in Frolikha Bay in the NBB, temperatures reach 16 °C at 0.5 m in the subsurface (Shanks & Callender 1992). This sub-lacustrine vent is located in a water depth of 440 m and the discharge of saline water can be traced deep into the NBB (Kipfer et al. 1996). Shanks & Callender (1992) interpreted stable-isotope data of the pore water in the proximity of the hydrothermal discharge to be
MUD DIAPIRS AND GAS HYDRATES IN LAKE BAIKAL
209
Fig. 2. Location map of the high-resolution seismic lines, heat flow stations and the side-scan sonar mosaic (shown in Fig. 3).
the result of intrusion of meteoric water from onshore mountain ranges along active rift faults into the bottom sediments of Frolikha Bay. Leaching of Cambrian evaporitic rocks near Lake Baikal is probably the cause of the increased salinity of the hydrothermal discharge. The sublacustrine hot spring at Frolikha Bay is related to an advective heat flow anomaly that reaches 8600 mW/m2 (Fig. Ib; Crane et al 1991; Golubev et al 1993); background heat flow in the Baikal Rift is about 40-70 mW/m2 (Golubev 2000). Other narrow positive heat flow anomalies in the NBB (Golubev & Poort 1995) occur at near-shore faults extending along bottom foothills. In the CBB, heat flow anomalies are also associated with faults in the subsurface. In the SBB a prominent positive anomaly occurs over an area of 30 km wide (Fig. Ib). Golubev (2000) suggests that all heat flow anomalies in Lake Baikal can be attributed to groundwater penetrating beneath the rift shoulders and rising as thermal waters along faults in the lake basin. Such regional heat and fluid circulation by
groundwater has been suggested to be feasible by numerical modelling (Poort & Polyansky 2002). The study area is located in a zone where on regional multi-channel seismic profiles the base of the hydrate stability zone (BHSZ) is irregular and not at all mimicking the lake floor as a 'bottomsimulating reflection' (Golmshtok et al. 2000) and where several gas seeps and mud extrusion features have recently been discovered (Van Rensbergen et al 2002). The study area is located over the hanging wall of a major rift fault, the Posolsky Fault, in a water depth of 1320 m to 1440 m (Fig. 2). The Posolsky Fault, just north of our study area, is an active fault that forms the SE flank of the Posolsky bank and continues basinwards in a SW direction for about 30 km. It has a maximum throw of over 1.5 s (about 3 km) below the SW margin of the Selenga delta (Scholz & Hutchinson 2000). However, just north of the observed seeps the Posolsky Fault offsets the lake floor by about 200 m. A number of small secondary faults with throws of less than 20 m occur within the study area.
210
P. VAN RENSBERGEN ETAL.
Fig. 3. Side-scan sonar mosaic over the discovered methane seeps.
Data and methods The data discussed here are located in a study area of about 15 km X 16 km in the SBB and were acquired in August 1999 (Fig. 2). In this study area, a sidescan sonar mosaic covers an area of about 90 km2. The 30 kHz SONIC sonar was towed at about 80-250 m above the lake floor; 0.8 km was imaged at each side of the track-line and the across-track footprint of the acoustic beam ranges from 0.75 m to 3.8 m. In addition, nine single-channel airgun seismic reflection profiles with a total length of about 180 km were obtained in the study area. A 3 litre IMPULSE-1 airgun (frequency range of 45-330 Hz) was used yielding a penetration up to 600 ms two way travel time (about 480 m) and a vertical resolution of about 3 m. One of the seeps was studied in detail using a 12 kHz echo sounder and 11 CTD casts (measurements of temperature, conductivity/salinity, light transmission, and oxygen concentration) were taken during two different surveys two weeks apart. Additional data were provided by 23 thermo probe measurements taken along two seismic profiles (Fig. 2) to calculate the lateral variation in heat flow. The 2 m long GEOS-T thermo probe applies the continuous heating method and allows the obtaining of temperature gradient measurements and thermal-conductivity values over four 0.5 m intervals in the upper sediments. Attempts to core in the seep area during the August 1999 expedition failed, but during the March 2000 coring expedition, from the ice-covered lake, samples of hydrate in diatom-rich silts and silty clays were retrieved. The sampled hydrates are
shallow accumulations (about 20 cm below the lake floor) restricted to the seep area. The cores did not penetrate the more extensive deeper hydrate layer, the base of which is imaged on the seismic profile. The geochemistry of hydrate samples and sediment pore waters were analysed by Matveeva et aL (2000). In the discussion section, the geochemical results will be integrated with geophysical data from this study.
Description and interpretation Surface and shallow subsurface expression of the methane seeps Four seeps were found in the study area (Fig. 3). They were named 'Bolshoy' (large), 'Stari' (old), 'Malyutka' (very small) and 'Malenki' (small) (Van Rensbergen et aL 2002). The seeps occur over a rollover structure in the footwall of a small fault antithetic to the Posolsky Fault (downthrown side to the north). The fault offsets the lake floor by about 20 m and the seismic profiles show the throw to increase with depth to a maximum of 100 m (about 80 m). This antithetic fault runs across the side-scan sonar mosaic as an undulating trace with high backscatter strength. The fault turns from a N30E direction in the western part of the study area to a N70E direction in the centre (parallel to the Posolsky Fault) where it splits into two perpendicular splays (N20E and Nl 10E respectively) with a throw of about 10 m each. Malenki and Malyutka are low-relief craters; Bolshoy and Stari are mud cones on the lake floor (Fig. 3).
MUD DIAPIRS AND GAS HYDRATES IN LAKE BAIKAL
211
Fig. 4. Detail of the side-scan sonar image of the Malenki crater (A), with overprint of bathymetry from echo-sounding surveys (B), and interpretation (C).
Low-relief craters. Malenki and Malyutka are located next to the southern fault splay. At Malyutka the fault splay orientation is parallel to the main fault trace (070°N); at Malenki the fault splay has turned to the more southerly direction (110°N). Short parallel fault escarpments, again with a 070°N orientation, are observed at both craters. At Malenki, one larger fault escarpment runs parallel to the northern fault splay direction (020°N). Malenki (Fig. 4) is the larger of the two craters, it has a maximum depth of 8 m and a diameter of about 500 m. Malyutka is only 200 m wide. At both sites, seep activity reveals itself on echo-sounding data as a 10-15 m high reflective plume in the lake water. The base of the hydrate stability zone (BHSZ) is generally marked on seismic records by a BSR, a continuous high-amplitude reflection sub-parallel to the lake floor and not affected by infra-basin fault traces or dipping stratigraphic reflections (Kvenvolden 1998). On high-resolution seismic data, such as the data used in this study, a BSR does not appear to consist of a continuous high-amplitude reflection but of a series of bright spots attributed to
free gas pockets accumulated below the BHSZ (Vanneste et al 2001). In Lake Baikal, the depth of the BSR corresponds well to the theoretical depth of the BHSZ (Vanneste et al. 2001), hence it is considered here to be a good approximation for the depth of the BHSZ. It has to be mentioned that in some marine hydrate provinces the correlation between BSR and BHSZ is ambiguous. Reflections similar to a BSR may occur both within the hydrate stability zone (e.g. Mienert & Posewang 1999; Nouze & Baltzer 2003) as well as below the BHSZ (Xu & Ruppel 1999; Wood & Ruppel 2000). In the study area, bright gas-enhanced reflections reveal an undulating BHSZ, strongly disrupted in the footwall of the antithetic fault (Fig. 5). They appear to be displaced along faults although their apparent vertical displacement is in places much larger than the actual fault displacement. Their distribution indicates that in places the BHSZ shallows up to a sub-bottom depth of about 150 m in the footwall block of the fault (as opposed to c. 400 m in the surrounding areas). The fault segment along which this occurs is about 10 km long; adjacent segments
212
P. VAN RENSBERGEN ETAL.
Fig. 5. Airgun seismic profile GAHY030 through the Malenki seep shows enhanced reflections at the base of the hydrate stability zone shallowing towards the Malenki vent site. Heat flow values calculated from the depth of the base of the hydrate layer and heat flow values derived from in situ thermoprobe measurements along the profile are plotted above the seismic section.
of the same fault show no anomalies in the BHSZ morphology. On a seismic profile through the Malenki seep area (GAHY030), a narrow vertical zone of chaotic reflections - a 'seismic chimney' (Fig. 6) - extends up to 200 m high, from the crest of the dome-shaped BHSZ to the seep at the lake floor. Such a 'seismic chimney' is most commonly interpreted as a vertical fluid conduit caused by hydraulic fracturing of the overburden by overpressured, often gas-charged, fluids (Van Rensbergen et al. 1999). The apparent width of the 'seismic chimney' is due to noise caused by acoustic velocity effects at shallow depth and does not represent the true width of the fluid conduit. Except for a few high-amplitude reflections immediately below the lake floor, no amplitude enhancement (indicators of free gas accumulations) or blanking (a possible indication of dense hydrate accumulations) occur along the 'seismic chimney'. There are no traces of sediment outflows on the side-scan sonar data (Fig. 4) or on the highresolution seismic data, nor is there any indication of past sediment deformation in the seep area (Fig. 6). This characteristic surface and subsurface expression, lacking extrusive mudflows or well-developed craters, suggests that the Malenki and Malyutka seeps are young features. The striking parallelism of escarpments within both seep areas with fault orientations indicates a close connection between the formation of the Malenki and Malyutka seeps and the present stress regime.
Mud cones at the lake floor. Bolshoy is the largest of the four vents. It appears as an irregular cone 24 m high and 800 m in diameter (Fig. 7), about 500 m SE of the antithetic fault trace. On side-scan sonar images, the Bolshoy cone appears to be composed of several smaller cones giving a rough appearance to the slopes. Lake-floor sediments accumulate against the NW flank of the cone, locally smoothing the irregular topography and causing a difference in level of 10 m between the opposing sides of the mud cone. Erosional moats occur at the southern and northern flanks of the cone. Again, the side-scan sonar data show no traces of mudflows at the lake floor. An area of anomalously low backscatter occurs at the apex of the cone on side-scan sonar images and is attributed to gas venting from the Bolshoy cone. On echo-sounding data parts of the Bolshoy area are obscured below a 15 m high reflective plume in the lake water. Stari lies about 2 km south from the fault trace. It is an ellipsoidal mound about 500 m long with an irregular surface and a maximum height of 10 m. Some visible lineaments may indicate fault escarpments but the parallelism characteristic for the Malenki and Malyutka craters is lacking. Seismic profiles about 1 km away from Stari do not show any sediment deformation features related to the Stari seep. There is no seismic line through the larger Bolshoy seep but the closest seismic profile (GAHY023) shows gas trapped at the margin of the dome-shaped BHSZ (Fig. 8). The BHSZ has an
MUD DIAPIRS AND GAS HYDRATES IN LAKE BAIKAL
213
Fig. 6. Detail of the high-resolution seismic section at the Malenki crater shows a vertical feeder pipe from the base of the hydrate layer to the lake floor. Methane bubbles escaping at the Malenki crater probably cause acoustic reflections in the water columns.
irregular mushroom shape within the crest of the rollover structure in the footwall of the antithetic fault. Extraction of the average energy (square of the reflection strength) in the seismic profile GAHY030 emphasizes the amplitude differences (Fig. 8b). Bright reflections occur below the BHSZ, most likely generated by small pockets of free gas. Bright reflections also occur below small (200 m deep, 500 m wide, up to 20 m thick) reflection-free lenses. These are interpreted as gas pockets below shallow mudflow lenses or below shallow gas hydrate accumulations. In a 100 ms (about 80 m) thick zone directly above the BHSZ, amplitude blanking can be observed across different seismic facies. Although amplitude blanking is not a direct indicator of gas hydrate and vice versa, it is probable that its occurrence is due to the presence of gas hydrate in the pore spaces of the sediments, effectively reducing impedance contrasts within the sediment (Rowe & Gettrust 1993). On the seismic line GAHY023, shallow sedimentary disturbances that may be related to mud extrusion at the Bolshoy seep are covered by a drape of continuous, undisturbed sediment about 20 m thick. It confirms the interpretation from the side-scan sonar data that no mud extrusion occurred recently and that the mud volcano is being buried by lake-floor sediments. Applying an average sedimentation rate of 0.3 mrn/yr (Colman et al.
1996), the timing of mud extrusion can be roughly estimated at 50-100 ka. On this basis, the Bolshoy and Stari cones are interpreted to be older than the Malenki and Malyutka seeps. The Bolshoy and Stari cones are different from the low-relief craters. They are characterized by mud extrusion at the lake floor during some phase of their evolution but not in recent times. Their location and morphology does not show as strong a relationship with active faults as the two low-relief seeps.
Heat flow anomalies and their implications for the source of methane Detailed heat flow data were obtained from thermal gradient and thermal conductivity measurements in the upper 2 m of the lake sediment at 23 stations in the study area. These in situ measurements are compared with calculated heat flow values obtained from the depth of the BHSZ on seismic data, following the procedure of Golmshtok et al. (2000). Along profile GAHY030 through the Malenki seep, the correlation between measured heat flow values and values inferred from the depth of the BHSZ is very good, the difference is only about 5% (within the range of the methodological error). In exception, a high peak value was measured in the Malenki seep where the
214
P. VAN RENSBERGEN ETAL.
Fig. 7. Detail of the side-scan sonar image of the Bolshoy mud cone (A), with overprint of bathymetry from echosounding surveys (B), and interpretation (C).
corresponding BHSZ cannot be detected on seismic data (Fig. 5). Near the Malenki seep, in-situ measured heat flow varies between 55 mW/m2 and 110 mW/m2, slightly higher than the average heat flow values for the SBB (50-70 mW/m2). In the Malenki seep crater a peak value of 165 mW/m2 was measured, which is much higher than any other measurement in this part of the SBB. The shape of this local heat flow anomaly is typical for the upward flow of warm fluids and quite similar to the local heat flow anomalies observed at thermal springs in the NBB (Golubev 2000) A contour map of inferred heat flow (Fig. 9), derived from the thickness of the BHSZ, shows the dome-shaped zone of the base of the hydrate-bearing sediment layer in the footwall of the antithetic fault (Vanneste 2000). The map is not accurate near the Bolshoy seep because of the outline of the seismic grid. Two areas of maximum heat flow are inferred, one in the vicinity of the Bolshoy seep and one associated with the Stari/Malyutka seeps. The Malenki seep, where the highest measured heat flow values were made, is located laterally from the inferred heat flow maximum. This again suggests that the base of the hydrated-sediment layer is not in equilibrium with temperatures at the BHSZ.
Near bottom water properties in the Malenki seep area Nine vertical CTD casts are located at and around the Malenki crater. Figure 10 shows a series of four CTD casts measured within a few hours of each other (see station locations on Fig. 6). The data show a small positive temperature anomaly and a negative anomaly in oxygen concentration, but no significant variations in light transmissivity or conductivity. The temperature and oxygen anomalies occur in an interval of 60 m above the NW part of the crater. To the NW of the crater, this anomaly occurs in an interval between 70 m and 125 m above the lake floor. These temperature and oxygen anomalies are transient phenomena: CTD casts at almost the same locations measured a week earlier do not show any significant anomalies. At station four, the average temperature increase over the anomalous temperature interval is 1.5 ± 0.7 mK; the average negative oxygen anomaly over the interval is about 0.087 ± 0.024 mgO2/kg. The temperature and oxygen anomalies can be attributed to the oxidization of methane. The measured average oxygen deficiency corresponds to an oxidization of 1.36 X 10~6 mole/1 CH4 free gas, which results
MUD DIAPIRS AND GAS HYDRATES IN LAKE BAIKAL
215
Fig. 8. A. Airgun seismic profile GAHY023 near to the Bolshoy seep shows again a shallowing base of the hydrate layer towards the seep. B. A detail of the high-resolution seismic profile shows a free-gas pocket trapped below the dome-shaped base of the hydrate layer. C. The relative energy attribute (square of the reflection strength) highlights bright spots caused by free gas in the sediment, as well as amplitude blanking, a possible indicator of hydrate-bearing sediments. D. An interpreted line drawing combines observations from seismic attribute analysis.
in an estimated temperature increase of about 0.3 ±0.1 mK. The estimated temperature increase is in agreement within a factor of two of the measured average temperature increase within the anomaly interval. The anomaly can therefore be explained by oxidization of rising CH4 and is related to the reflective plume observed on echo-sounding data. At Station seven, located at the Malenki seep and within the area covered by the acoustic plume, the temperature anomaly starts abruptly at 13 m above the lake floor, at the top of the plume. To the north, outside the area covered by the acoustic plume, the temperature anomaly shifts to higher levels. The acoustic plume in the lake water is therefore attributed to rising methane bubbles at the seep dissolving in the water column. Once dissolved, methane will oxidize to CO2. Deep water currents and the initial jet of the upward moving methane bubbles will cause the buoyant dissolved methane cloud to drift away from the active degassing site of Malenki. As a result, we assume that the temperature and oxygen anomalies related to methane oxidation are found at higher levels in Station four, located about 600 m NW of the Malenki crater.
These observations indicate that it is probable that only methane is escaping at the Malenki crater. If other fluids are expelled at Malenki they cannot be much different from the lake water in terms of temperature and salinity.
Discussion Geochemical characteristics of shallow
hydrate at the Malenki seep area At the Malenki site, shallow gas hydrate accumulations are found at sub-bottom depths of 16 to 42 cm. This shallow gas hydrate is restricted to the Malenki seep area and occurs independently from the deeper, laterally extensive hydrate layer. Gas hydrate samples were analysed for the gas composition by Matveeva et al (2000), sediment pore water was sampled and analysed by Matveeva et at. (2000) and Granina^a/. (2001). In the shallow gas hydrate layer, hydrate occurs as
216
P. VAN RENSBERGEN ET AL.
Fig. 9. Heat flow map based on heat flow values calculated from the thickness of the hydrate layer along the seismic lines. The SW part of the map (hatched lines) is unrealistic due to poor data coverage. The positions of the mud cones and craters in relation to heat flow and structure are indicated.
massive lumps of up to 7 cm in diameter, and as cmsized inclusions throughout the muddy host sediment. At spontaneous degassing, the clathrated gas consists (by volume) of 99.0% methane with minor to trace amounts of ethane (0.11%), propane (2.10~4), butane (3.10~4), nitrogen (0.4%) and carbon dioxide (0.5%) (Matveeva et al 2000). The C1/C2 ratio is 900, which is according to Sloan (1998) indeterminate to the thermogenic (C1/C2< 100) or biogenic (C1/C2> 1000) origin of the gas in the sample. The relative absence of propane suggests a biogenic origin (Sloan 1998) but it may also be a mixture of biogenic with some hydrocarbons from a deeper, thermogenic, source. The pore waters in the Malenki seep area show a typical increase in chloride and sulphate concentrations compared to average pore water composition in Lake Baikal. Granina et al. (2001) found that the mean concentration of chloride in sediment pore waters (11.8 mg/1) is 15 times higher compared to average pore water concentration in SBB (0.8 mg/1). At the same location, Matveeva et al. (2000) also found the ion content increasing from 'near-bottom lake "water' over 'sediment pore water' to 'water from gas hydrate-bearing sediments'. A similar evo-
lution exists for oxygen and deuterium isotopic compositions that become heavier from lake water to water in hydrate-bearing sediment, indicating isotopic fractionation during hydrate formation (Matveeva et al 2000). Matveeva et al (2000) reported also that 'very dense' sediments with low water content occurred above the hydrate-bearing interval. Shallow hydrate formation from a flow of free gas, in this case originating from the BHSZ, will draw water from the surroundings (Ginsberg & Soloviev 1997) and can create haloes of saline water due to incomplete diffusion of excluded salt together with water depletion in the vicinity of concentrated hydrate lenses (Clennell etal 1999). Both Matveeva et al (2000) and Granina et al (2001) interpret the data as evidence of injection of saline thermal water into the shallow subsurface sediments. In our opinion, the effect of ion exclusion by hydrate formation can cause the observed geochemical and isotopic signatures. The measurements are considered normal for shallow hydratebearing sediment in a gas seep area and do not - in our opinion - provide evidence for injection of thermal water into shallow sediment. This interpretation also fits with the CTD measurements at the
MUD DIAPIRS AND GAS HYDRATES IN LAKE BAIKAL
217
Fig. 10. Temperature curves and oxygen concentration curves from 4 CTD stations projected onto the seismic section show the good correlation of a positive temperature anomaly with a negative oxygen anomaly attributed to oxidization of methane in the near-bottom Baikal water.
Malenki seep where no expulsion of saline or warm fluids was detected. Integration of results The observed methane seeps and mud cones in Lake Baikal occur in an area of subsurface gas hydrate accumulation and of varying heat flow within a larger region of elevated heat flow. The variations in heat flow are interpreted to result from migration of thermal water along active fault segments and dipping permeable beds. Heat redistribution by infiltrated meteoric water is inferred to be the cause of heat flow variations in all of the Baikal sub-basins (Golubev 2000). The narrow, high-amplitude heat flow anomaly at the Malenki seep is interpreted to be a short-term event, not a steady-state situation. Consequently, it is interpreted, on the basis of the arguments below, that a pre-existing, laterally continuous hydrate layer was locally affected by a recent elevation of heat flow at the BHSZ, about 400 m subsurface (Fig. 11). The arguments for the seeps being the result of a recent perturbation of the regional geothermal field and local decomposition of a pre-existing hydrate layer are as follows: (1) (2)
Active seeps are limited to the area above the anomaly, there are no seeps observed in the area with regular BSR. The active seeps and mud diapirs are young features (estimated <100 ka), there is no indication of any older activity on the highresolution seismic data.
(3)
The young age of the seeps is in agreement with the heat flow data that indicate a nonequilibrium condition exist at the site of the Malenki seep. (4) The Malenki seep appears to be a gas seep; no warm fluids are expelled. This indicates that there is no direct fluid-flow pathway from the deep subsurface to the sediment surface. (5) A gas chimney occurs in the shallow subsurface at the site of the Malenki seep. Because of point four, the gas chimney probably roots in the zone of elevated BSR and extends from the top of the updoming BSR to the lake floor. This indicates that overpressures below the domeshaped BHSZ increased to values greater or equal to the fracture gradient. Since the rate of overpressure increase was much higher than the rate of pore fluid pressure dissipation, it indicates a relatively rapid overpressure increase. (6) Shallow hydrate accumulated just below the sediment surface in the Malenki seep crater. Despite numerous cores, no shallow hydrates were found outside of the Malenki crater. The gas in the shallow hydrates is most likely of bacterial origin, possibly a mixture with gas of thermogenic origin, and does not have the geochemical signature of gas migrating directly from deeper thermogenic sources. Figure 11 shows the interpreted suite of events following a heat pulse at the base of the hydrated sediment layer. Heat transfer to the BHSZ probably caused the rise of the BHSZ to the present depth of about 150 m in the seep areas. The gas release from
218
P. VAN RENSBERGEN ETAL.
A. Continuous gas hydrate layer, possibly with free gas pockets at the base of the hydrate stability zone(BHSZ)
B. A thermal pulse shifts the BHSZ upwards, methane is released from hydrate decomposition, and pore fluid pressure below BHSZ rises.
C. Gas blow-out occurs when pore fluid pressure reaches the fracture gradient. Shallow hydrates form out of the gas flow below the sediment surface and draw water from the surroundings. Mud may be extruded as a stiff plug into a mud cone. Fig. 11. Conclusion cartoon showing the inferred decomposition of a laterally continuous hydrate layer (A) by a pulse of increased thermal flux at the base of the hydrate layer (B, C).
MUD DIAPIRS AND GAS HYDRATES IN LAKE BAIKAL
hydrate decomposition is accompanied by an increase in pore fluid pressure, as one volume of methane hydrate may hold 150-180 volumes of free gas (Kvenvolden 1988). Pore fluid pressure at the BHSZ probably rose up to the fracture gradient and caused formation of vertical fractures and gas expulsion. The occurrence of young active seeps and mud diapirs illustrates the forcible nature of the gas expulsion. In the seep areas, local shallow hydrate formed out of the flow of free gas and accumulated almost immediately below the sediment surface. The shallow hydrates have a halo of saline pore water and isotopically heavier oxygen. Initially, oceanic gas hydrates were also interpreted to be the source of mud volcanoes but over the years, more detailed measurements established the deep origin of methane expelled at the sea floor and of methane in hydrate. We do not wish to exclude this possibility but except for the disagreement with oceanic studies, there is no indication of a direct transfer of methane from a deep source. The bulk of the hydrated gas, sampled in a deep core of the Baikal Drilling Project, is of bacterial origin (Kuzmin et al. 1998) and formed via methanogenesis of organic matter supplied by the Selenga River. There are no alternative sources of biogenic or thermogenic methane known in Lake Baikal. Based on the observations above it is inferred that the largest available source of gas in Lake Baikal is probably the extensive hydrated sediment layer that occurs to a depth of about 300-400 m. The Malenki seep is a gas seep; there is no indication of seepage of mineralised or thermal waters in the Malenki area. The entire expulsion mechanism seems to be driven by free gas; methane hydrate dissociation releases large amounts of free gas; free gas is expelled at the Malenki seep and only goes into solution after it enters the lake water. Shallow gas hydrates at the Malenki seep also from out of free gas and pore water. Additionally, shallow hydrate formation will extract fluids form the sediment, further decreasing the likelihood of fluid expulsion and fluidization of the sediment. The crater morphology is therefore not shaped by mudflows. The Bolshoy and Stariy cones may be extruded as a stiff plug of sediment without fluidization. The driving force of the mud extrusion may be the buoyancy of the gas-charged sediment, possibly in combination with a pore fluid pressure gradient. On the basis of morphology, localization and probably also age, the four seeps can be grouped into mud cones (Bolshoy and Stari) and low-relief craters (Malenki and Malyutka). Malenki and Malyutka seem to be recent features located along active fault segments. Thermal waters responsible for hydrate dissociation at the BHSZ most likely migrated upwards along these active fault segments. Bolshoy and Stari especially are older features (estimated to
219
be about 50-100 ka old). They are located away from the fault at the crest of the rollover structure in the footwall of the antithetic fault. Thermal fluids may have migrated upwards along tilted, permeable beds. The difference in morphology may be caused by the age of the seeps (older seeps having extruded a plug of mobilized mud), but may also be related to the localization of the seeps. Fluid flow along the antithetic fault segment in the study area may be recent but also ephemeral, causing short-lived pulses of gas escape, whereas thermal fluids focused in structural traps may form a more constant source of heat, releasing larger quantities of methane. Methane seeps attributed to hydrate decomposition are quite common in oceanic hydrate provinces. In these contexts, the methane is considered to originate from the regional, gradual (bottom-up) destabilization of hydrates at the BHSZ and the BHSZ is usually not characterized by distinct anomalies. Methane is usually released at the sea floor by dispersed seepage over large areas, e.g. in pockmark or crater fields (Ginsberg 1998; Solheim & Elverhoi 1993) or by more focused escape along faults (Ginsberg et al. 1993; Mienert & Posewang 1999; Suess et al 1999; Wood & Ruppel 2000). Such methane seeps are usually not accompanied by sediment extrusion at the sediment surface. The Lake Baikal example seems to indicate that localised destabilization of hydrates may create large overpressures in the shallow subsurface (<500 m) and that the consequent gas blow-out is capable of mobilizing and extruding subsurface sediments.
Conclusions In this paper, we investigated in detail the nature, morphology and structure of lake floor seeps that occur in Lake Baikal. These are the first gas seeps and mud volcanoes associated with gas hydrate discovered outside the marine environment. The seeps were encountered in the South Baikal Basin in an area with peculiar anomalies in thickness of the hydrate-bearing sediment layer. Measurements of near-bottom water properties, surface data (sidescan sonar and bathymetry) and subsurface data (high-resolution seismic profiles, heat flow data, and geochemical studies) are combined into an integrated study of the seeps and associated mud mobilization features. The seeps are gas seeps without much, if any, expulsion of associated fluids. The gas seeps occur as low-relief craters or mud cones. Mud cones appear to be older than craters and have a different structural setting. Mud cones occur at the crest of rollover structures, in the footwall of a secondary normal fault. The craters occur at fault splays and are dissected by numerous escarpments parallel to the
220
P. VAN RENSBERGEN ETAL.
main fault direction. The gas seeps are located in areas of elevated heat flow where the base of the hydrate stability zone (BHSZ) moves upwards from about 350 m to about 150 m below the lake floor. At the Malenki site, the surface seep is directly connected with the base of the hydrate stability zone through a vertical 'seismic chimney', interpreted as a feeder pipe. A shallow accumulation of gas hydrate was found in short cores in the Malenki seep area. Active gas seeps in a hydrate provinces are most often the result of a gas flow from deep subsurface hydrocarbon systems and hydrates are often only a side product of this gas flow. In this case however, it is interpreted that the seeps are caused by an ongoing process of focused hydrate decomposition and associated massive methane escape. This rare example of focused gas flow from hydrate decomposition, at this moment in geological time, may be due to the combination of thick hydrate accumulations in an active rift characterized by numerous heat flow anomalies. A heat flow anomaly at one of the active seeps is interpreted to be a young feature, probably due to a recent influx of thermal fluids in the subsurface and probably caused focused destabilization at the base of the hydrate accumulation zone. The expulsion of large quantities of methane creates craters at the lake floor but also seems capable of extruding mud, most likely without much sediment fluidization. Our interpretation of the Baikal gas seeps suggests that gas hydrate destabilization can create large pore fluid overpressures in the shallow subsurface (<500 m subsurface) and cause mud extrusion at the sediment surface. The research project was funded by the Belgian Federal Office for Scientific, Technical and Cultural Affairs (OSTC) and the INTAS project 1915. PVR and JP are postdoctoral fellows with the Fund for Scientific ResearchFlanders (FWO). The seismic data were processed and viewed using SMT's Kingdom Suite interpretation software (Educational User License). The RCMG team helpfully supported data processing. Special thanks to E. Chapron and R. Hus for their help with the interpretation. We warm-heartedly thank the captain and crew of the R/V Vereshchagin and the Director of the Limnological Institute (Irkutsk, Russia) for the excellent logistic, technical, and scientific support. Najwa Yassir and Roger Sassen kindly reviewed the manuscript and suggested useful changes.
References CRANE, K., HECKER, B. & GOLUBEV, V. 1991. Hydrothermal vents in Lake Baikal. Nature, 350, 281. COLMAN, S.M., JONES, G.A., MEYER, R., KING, J.W., PECK, J. A. & OREM, W.H. 1996. AMS Radiocarbon analysis from lake Baikal, Siberia: Challenges of dating sediments from a large oligotrophic lake. Quaternary Science Reviews, 15, 669-684.
CLENNELL, M.B., HOVLAND, M., BOOTH, J.S., HENRY, P. & WINTERS, WJ. 1999. Formation of natural gas hydrates in marine sediments 1. Conceptual model of gas hydrate growth conditioned by host sediment properties. Journal of Geophysical Research, 104, 610,22985-23003. FAULKNER, K.K., MEASURES, C.I., HERBELIN, S.E. & EDMOND, J.M. 1991. The major and minor element geochemistry of Lake Baikal. Limnology and Oceanography, 36,413-423. GINSBERG, G.D., SOLOVIEV, V.A., CRANSTON, R.E., LORENSON, T. D. & KVENVOLDEN, K.A. 1993. Gas hydrates from the continental slope, offshore Sakhalin Island, Okhotsk Sea. Geo-Marine Letters, 13,41^-8. GINSBERG, G.D. 1998. Gas hydrate accumulation in deepwater marine sediments. In: HENRIET, J.P. & MIENERT, J. (eds) Gas hydrates. Relevance to world margin stability and climatic change. Geological Society, London, Special Publications, 137,51-62. GINSBERG, G.D. & SOLOVIEV, V.A. 1997. Methane migration within the submarine gas hydrate stability zone under deep water conditions. Marine Geology, 137, 311-323. GOLMSHTOK, A.Y., DUCHKOV, A.D., HUTCHINSON, D.R. & KHANUKAYEV, S.B. 1997. Estimation of the heat flow in Lake Baikal based on seismic data of gas hydrates layer lower boundary. Russian Geology & Geophysics, 10, 1714-1727. GOLMSHTOK, A.Y., DUCHKOV, A.D., HUTCHINSON, D.R. & KHANUKAEV, S.B. 2000. Heat flow and gas hydrate of the Baikal Rift Zone. International Journal of Earth Sciences, 89,2,193-211. GOLUBEV, V.A. 2000. Conductive and convective heat flow in the bottom of Baikal and in the surrounding mountains. Bulletin du Centre de Recherches Elf Exploration Production, 22,323-340. GOLUBEV, V.A., KLERKX, J. & KIPFER, R. 1993. Heat flow, hydrothermal vents and static stability of discharging thermal water in Lake Baikal (south-eastern Siberia). Bulletin du Centre de Recherches Elf Exploration Production, 17,54-65. GOLUBEV, V.A. & POORT, J. 1995. Local heat flow anomalies along the western shore of north Baikal basin (Zavorotny area). Russian Geology and Geophysics, 36, 174-185. GRANINA, L.Z., CALLENDER, E., LOMONOSOV, I.S., MATS, V.D. & GOLOBOKOVA, L.P. 2001. Anomalies in the composition of Baikal pore waters. Geologija i Geofizika (Russian Geology and Geophysics), 42, 362-373 (in Russian). KIPFER, R., AESCHBACH-HERTIG, W., HOFER, M., HOHMANN, R., IMBODEN, D.M., BAUR, H., GOLUBEV, V & KLERKX, J. 1996. Bottomwater formation due to hydrothermal activity in Frolikha Bay, Lake Baikal, Eastern Siberia. Geochimica et Cosmochimica Acta, 6,961-971. KUZMIN, M.I., KALMYCHKOV, G.V, GELETIJ, V.F., GNILUSHA, V.A., GOREGLYAD, A.V., KHAKHAEV, B.N., PEVZNER, L.A., KAVAI, T, IOSHIDA, N., DUCHKOV, A.D.,
PONOMARCHUK, V.A.,
KONTOROVICH,
A.E.,
BAZHIN, N.M., MAHOV, G.A., DYADIN, Yu. A., KUZNETSOV, FA, LARIONOV, E.G., MANAKOV, A. Yu., SMOLYAKOV, B.S., MANDELBAUM, M.M. & ZHELEZNYAKOV, N.K. 1998. First find of gas hydrate in
MUD DIAPIRS AND GAS HYDRATES IN LAKE BAIKAL sediments of Lake Baikal. Doklady Akademii Nauk, 362,541-543 (in Russian). KUZMIN, M.I., GELETIJ, V.F., KALMYCHKOV, G.V., KUZNETSOV, F.A, LARIONOV, E.G., MANAKOV, A. Yu., MIRONOV, Yu. I., SMOLYAKOV, B.S., DYADIN, Yu. A., DUCHKOV, A.D., BAZIN, N.M. & MAHOV, G.A. 2000. The first discovery of gas hydratesin the sediments of the Lake Baikal. In: HOLDER, G.D. & BISHNOI, P.R. (eds) Gas hydrates. Challenges for the future. Annales of the New York Academy of Sciences, 912, 112-115. KVENVOLDEN, K. A. 1988. Methane hydrate - a major reservoir of carbon in the shallow geosphere? Chemical Geology, 71,41-51. MATVEEVA, T.V., KAULIO, V.V., MAZURENKO, L.L., KLERKX, J., SOLOVIEV, V. A., KHLYSTOV, O.M. & KALMYCHKOV, G.V. 2000. Geological and geochemical characteristic of near-bottom gas hydrate occurrence in the southern basin of the Lake Baikal, Eastern Siberia. Abstract Book of the VI International conference on gas in marine sediments, VNIIOkeangeologia, St. Petersburg (Russia), 91-93. MIENERT, J. & POSEWANG, J. 1999. Evidence of shallowand deep-water gas hydrates destabilizations in North Atlantic polar continental margin sediments. GeoMarine Letters, 19,143-149. NOUZE, H. & BALTZER, A. 2003. Shallow bottom simulating reflections on the Angola margin, in relation with gas and gas hydrate in the sediments. In: VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, A.J. & MORLEV, C.K. (eds) Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216,191-206. PAULL, C.K., USSLER, W. & BOROWSKI, W.S. 1994. Proposed model of hydrate formation by upward migration of free gas. In: SLOAN, E.D., HAPPEL, J. & HNATOW, M.A. (eds) Proceedings of the First International Conference on Natural Gas Hydrates, Annales of New York Academy of Sciences, 715, 392. POORT, J. & POLYANSKY, 0.2002. Heat transport by groundwater flow during the Baikal rift evolution. Tectonosphysics, 351,75-89. PINNEKER, E.V. & LOMONOSOV, I.S. 1973 On genesis of thermal waters in the Sayan-Baikal highland. In: Proceedings of Symposium on Hydrogeochemistry and Biogeochemistry, Volume I Hydrogeochemistry, The Clarke Company, Washington, 246-253. ROWE, M.M. & GETTRUST, J.F. 1993. Fine structure of hydrate bearing sediments on the Blake Outer Ridge as determined from deep-tow multichannel seismic data. Journal of Geophysical Research, 98, Bl, 463-473.
221
SCHOLZ, C.A. & HUTCHINSON, D.R. 2000. Stratigraphic and structural evolution of the Selenga Delta Accommodation Zone, Lake Baikal Rift, Siberia. International Journal of Earth Sciences, 89,212-228. SHANKS, W.C. & CALLENDER, E. 1992. Thermal springs in Lake Baikal. Geology, 20,495-497. SLOAN, E.D. 1998. Clathrate hydrates of natural gasses. 2nd edition revised and expanded. Chemical Industries Volume 73, Marcel Dekker Inc. New York, 705 pp. SOLHEIM, A. & ELVERHOI, A. 1993. Gas-related sea floor craters in the Barents Sea. Geo-Marine Letters, 13, 235-243. SUESS, E., TORRES, M.E., BOHRMANN, G., COLLIER, R.W., GREINERT, J., LINKE, P., REHDER, G., TREHU, A., WALLMANN, K., WINCKLER, G. & ZULEGER, E. 1999. Gas hydrate destabilisation: enhanced dewatering, benthic material turnover and large methane plumes at the Cascadia convergent margin. Earth and Planetary Science Letters, 170,1-15. VANNESTE, M. 2000. Gas hydrate stability and destabilisation processes in lacustrine and marine environments - Results from theoretical analyses and multi-frequency seismic investigations, unpublished Ph.D. thesis, Renard Centre of Marine Geology, Department of Geology and Soil Sciences, Ghent University, pp. 255. VANNESTE, M., DE BATIST, M., GOLMSHTOK, A., KREMLEV, A. & VERSTEEG, W. 2001. Multi-frequency seismic study of gas hydrate-bearing sediments in Lake Baikal, Siberia. Marine Geology, 172,1-21. VAN RENSBERGEN, P., DE BATIST, M., KLERKX, J, Hus, R., POORT, J., VANNESTE, M., GRANIN, N., KHLYSTOV, O. & KRINITSKY, P. 2002. Sub-lacustrine mud volcanoes and methane seeps caused by dissociation of gas hydrates in Lake Baikal. Geology, 30,7,631-634. VAN RENSBERGEN, P., MORLEY, C.K., ANG, D.W., HOAN, T.Q. & LAM, N.T. 1999. Structural evolution of shale diapirs from reactive rise to mud volcanism: 3D seismic data from the Baram Delta, offshore Brunei Darussalam. Journal of the Geological Society, London, 156, 633-650. WOOD, W.T. & RUPPEL, C. 2000. Seismic and thermal investigations of the Blake Ridge gas hydrate area: a synthesis. In: PAULL, C.K., MATSUMO, R., WALLACE, P.J. & DILLON, W.P. (eds) Proceedings of the Ocean Drilling Program, Scientific Results, 164,253-264. Xu, W. & RUPPEL, C.1999. Predicting the occurrence, distribution, and evolution of methane gas hydrate in porous marine sediments. Journal of Geophysical Research, 104, 5081-5095.
This page intentionally left blank
The genesis of polygonal fault systems: a review JOE CARTWRIGHT, DAVID JAMES & AL BOLTON Department of Earth Sciences, Cardiff University, Cardiff CF10 3 YE, UK Abstract: Polygonal fault systems are widely developed in fine-grained sedimentary successions and have been recognized in over 50 basins worldwide. They are normal faults with modest throw values (typically 10-100 m), organized with a characteristic plan form pattern that is crudely polygonal, but with considerable variation in specific planform patterns. They have been attributed to four genetic mechanisms: gravity collapse, density inversion, syneresis and compactional loading. Their strain characteristics allow them to be distinguished from tectonic normal faults. The strengths and weaknesses of the four genetic mechanisms are considered in the light of these strain characteristics. It is argued that syneresis offers the likeliest mode of genesis and best explains the local and global features of these extraordinary structures. The detailed physical mechanism driving syneresis remains poorly understood.
The aim of this paper is to summarize the current understanding of polygonal fault systems and to explore the various mechanisms offered to explain their genesis. These intriguing and enigmatic structures have considerable implications for the flow of pore water and hydrocarbons in sedimentary basins and for a more general soil mechanical view of clay consolidation (Cartwright 1996,1997). For much of the last century, the consolidation of clay soils or clay sediments has been regarded as a one-dimensional process driven by gravitational loading (e.g. Terzhagi & Peck 1948). In this process, original depositional grain fabrics are locally re-aligned as porosity decreases as a function of increasing vertical effective stress. The strain involved is three-dimensional at the grain scale, since re-packing necessitates movements in three dimensions, but the engineering approach to this process views this is as one-dimensional on a gross scale, since the net loss of volume in the horizontal plane is observably zero. This assumption of no lateral strain (e.g. Lambe & Whitman 1979) does not take into account the possibility of widespread and large displacement brittle failure of the consolidating sediments. The development of polygonal fault systems is intrinsically linked to the early consolidation of finegrained sediments, as reviewed below. Polygonal fault genesis clearly requires a critical examination of whether classical soil mechanics can explain their development and if so, how this genesis can be reconciled with previously held views of clay consolidation. The paper begins with a review of previous research on polygonal faults, showing how they are recognized and differentiated from other types of soft sedimentary structure and how models for their development have evolved over the past decade. The various competing or complimentary mechanisms are then discussed, along with an assessment of what
further data are required in order to test these genetic models. The paper then concludes with a discussion of the implications of polygonal faults for different disciplines engaged in basin analysis.
Definition In this paper, polygonal fault systems are defined as 'an array of layer-bound extensional faults within a mainly fine-grained stratigraphic interval that exhibit a diverse range of fault strikes which partially or fully intersect to form a polygonal pattern in map view'. This definition is deliberately loose when it comes to the discussion of strike because the local stress regime operative at the time when polygonal faults grow can exert a significant influence on strike and on the organisation of the fault array as a whole. The definition is also deliberately specific about the limological context, because thus far this type of fault has only been recognized in packages that are predominately composed of fine-grained sediments. Other types of layer-bound extensional fault system occur, such as classical detached growth faults and slump faults, differential compaction faults and faults formed in association with forced folding. For this discussion and for clarity in general, it is useful to separate polygonal faults from these other layer-bound faults, although there may be cases where they are not distinguishable on such simple criteria as embodied in this definition.
Review of previous research Over the past two decades, numerous descriptions of seismic data have shown small intra-formational extensional faults, which have been attributed to compaction or dewatering (see Cartwright & Dewhurst 1998 for references). However, polygonal
Fig. 1. A pioneering model for the genesis of polygonal fault systems as conceived by Henriet et al. (1989). This model is simplified here into a four-stage evolution, from deposition of clays, (I), to self-sealing and overpressure build-up, CH), to density inversion folding, (in), and finally to faulting and pore pressure collapse, (IV). The model shows how density inversion ultimately leads to folding through the development of a Rayleigh-Taylor instability. The regular pattern of folding then leads to a pattern of complementary fractures and faults. faults were first recognized as a new class of softsediment deformational feature in the late 1980s. In pioneering work based on outcrops of Eocene and Oligocene clays in quarries in Belgium and from high resolution seismic data in the Belgian offshore, Henriet et al. (1989, 1991) recognized a suite of 'clay tectonic' structures that they attributed to over pressuring of clay-rich sediments, density inversion and collapse (Fig.l). They recognized the widespread development of small extensional faults as part of their clay tectonic spectrum and compiled directional data for these faults. Their mapping, however, was limited by the problems of resolving a complex fault pattern with the sampling frequency inherent in their 2D seismic grid. Convincing polygonal geometries were only observed a few years later with the benefit of 3D seismic data. Higgs & McClay (1993) mapped upslope-facing intra-formational faults in the Oligo-Miocene on the western margin of the North Sea Basin and by comparing their geometry from 2D seismic interpreta-
tion with faults in analogue experiments proposed a model based on gravitational collapse on a slope. Cartwright (1994a, b) used regional 2D seismic data to map the extent of a system of layer-bound extensional faults developed pervasively within the Lower Tertiary of the North Sea Basin. This fault system was developed in tiers that were defined as discrete and correlatable stratigraphic packages containing genetically linked fault arrays with polygonal plan form geometry (Fig. 2). Growth sequences at the upper boundaries of the tiers suggested that the faults acted as small, syn-sedimentary faults (Fig. 3). In map view, the seismically resolvable faults were only found to be developed in the basin axis and on its slopes, but not in the marginal deltaic systems. The lateral limits to the fault system were coincident with lateral changes in facies and/or the thickness of the tier. Cartwright (19940) thus inferred that the faults were delimited primarily by the fine-grained nature of the depositional units in which they were developed, but also suggested that the updip limit of the system
THE GENESIS OF POLYGONAL FAULT SYSTEMS: A REVIEW
225
Fig. 2. Schematic geometry of polygonal fault systems showing their organization into tiers. Plan form patterns within each tier differ from those in adjacent tiers, although some faults often link from one tier to another. Note the tendency for the faults to dip in one direction along one cross-sectional face, but not on the orthogonal section. The rose plot shows some real data from the central North Sea, revealing the bias in strike direction of faults related to the palaeoslope direction (from Cartwright 19940).
might be influenced by the gross isopach of individual tiers (Fig. 3). Cartwright (1994a) characterized this North Sea polygonal fault system as consisting of small extensional faults with throws of 10-50m, with throws and lengths being scaled according to the tier thickness, with faults mainly restricted to individual tiers, but larger faults transecting tiers. Cartwright (1994a, b) extended the Henriet et al (1991) model to the basin scale and applied it to the Tertiary of the North Sea Basin. From a 3D seismic survey of UK block 30/19, he recognized the polygonal geometry of a system of minor extensional faults contained entirely within Lower Tertiary mudrich sequences and invoked episodic dewatering of overpressured cells in order to explain the tiered nature of the fault system. Cartwright & Lonergan (1996) measured apparent extensional strains of approximately 10% on line balanced depth sections through the North Sea polygonal fault system, with areal strain that was almost radially uniform over much of the basin. Importantly, strain was found to vary within tiers and between tiers, but remained almost uniform radially. This extensional strain was attributed to volumetric contraction (shrinkage) because: (1) absence of visible compressional structures to balance this extension; and (2) the underlying basement had remained essentially undeformed through the Cenozoic (no lithospheric extension). They drew an
analogy with mudcracks (vertical extensional cracks due to shrinkage of mud by dessication), but emphasized that the growth of polygonal faults involved shear failure conditions, not pure vertical tensile failure and also that there was no evidence for dessication. At that time, a physical explanation for shrinkage of the fully saturated sediments in a wholly sub aqueous environment was lacking. Cartwright & Dewhurst (1998) subsequently invoked the poorly understood process of syneresis for polygonal fault genesis. This process is well known in chemical engineering since it affects many colloidal materials (fine grained particles with a critically high surface area to mass ratio). Syneresis is defined as the 'spontaneous contraction of a gel without the evaporation of the solvent' (Brinker & Scherer 1990). Clay-rich sediments have the potential to form ideal gels on deposition because of their micron-scale particle size and the high mass to surface area ratio of the particles and thus could be expected to synerese as a normal response to their gel-like state. Cartwright & Dewhurst (1998) in a global review of polygonal faults based on 2D and 3D seismic from over 200 basins showed: (1) that they are very widely developed, sometimes over areas of >l,000,000km2; (2) they were known exclusively from passive margin or cratonic basins (e.g. North Atlantic, South Atlantic, Australian margins) and; (3) were only developed in sedimentary sequences
226
J. CARTWRIGHT ETAL
Fig. 3, A seismic profile from the UK sector of the central North Sea, showing the characteristic geometry of a polygonal fault system within Lower Tertiary mudstones. Note the lateral reduction in strain as the tier thickness decreases towards the edge of the fault system. Note also the evidence for synsedimentary fault activity in the form of divergent wedges adjacent to the upper tips of many of the faults. Only a few of the many faults in this tier are shown with interpreted fault planes.
dominantly composed of fine-grained facies such as clays or biogenic sediments. Cartwright & Dewhurst (1998) used this distribution as their prime argument to develop the case for syneresis as a driving mechanism for the shrinkage model of Cartwright & Lonergan (1996). A map showing the distribution of previously reported and unpublished occurrences of polygonal fault systems is shown in Figure 4. Although polygonal faults deform primarily fine-grained sediments, Lonergan & Cartwright (1999) and Stuevold et al (2003) have shown that they can also occur when much coarser facies are interbedded with the fine-grained intervals, but they do not occur solely and specifically within coarsegrained intervals. Polygonal faults often exhibit a 'mutually exclusive' distribution pattern where there are thick coarse-grained bodies developed within a generally fine-grained interval, the classic example being the polygonally faulted margin of the Alba Field (Fig. 5). Most recently, Watterson et al (2000) have challenged the syneresis model, and instead re-advanced the argument that density inversion is the main driving mechanism for the faulting, with contributions from gravitationally-driven extension (slope
failure). Goulty (2001) has also challenged the syneresis model, advocating that displacement on the polygonal faults is possible under normal (burial) loading conditions and that repeated slip occurs because of the low frictional strength of the compacting sediments.
Structural characteristics Geometry Polygonal faults systems are commonly developed in tiers (Cartwright 1994a). Faults in one tier may partially interconnect with those in adjacent tiers by cross-propagation of a sub-set of the total fault population (Fig. 2), but the majority of the faults in the separate tiers are contained wholly within individual tiers. Tier boundaries are often widely correlatable stratigraphic units and can be either coarser intervals, or intervals that act as a detachment or mobile layer. No attempt has been made on a sufficiently large database to correlate tier boundaries with stratigraphic properties, although an initial attempt by
THE GENESIS OF POLYGONAL FAULT SYSTEMS: A REVIEW
227
Fig. 4. The global distribution of polygonal fault systems as recognized from seismic data or, more rarely, field evidence. The basins with polygonal fault systems are mainly in passive margin or intra-cratonic settings and are indicated in outline position only (black dots). The data is drawn mainly from Cartwright & Dewhurst (1998), with additional unpublished data.
Dewhurst et al. (I999a) over part of the North Sea Basin did show that fault style in tiers (geometry, strain) correlated reasonably well with bulk grain size distributions. Polygonal faults have many of the same geometrical characteristics as tectonic normal faults. On 3D seismic data, they range in fault trace length from 100 m to several kilometres and extend vertically across discrete layers from a few tens of metres to over one kilometre in thickness. Smaller scale faults of the polygonal type have been observed in Belgium (Henriet et al 1991; Verschuren 1992) and in the Thames Estuary (J. Dix, pers. comm. 1997), suggesting that this range is limited at the lower end by seismic resolution. On good quality seismic data, the faults appear as discrete, single 'planes' (e.g. Fig. 6), but field examples show that greater complexity can occur (Verschuren 1992). The fault planes observed at outcrop (Verschuren, 1992) are notable for having thin fault zones and the displacement is concentrated at the fault zone, rather than distributed in a broader ductile shear zone, as suggested by Wattersonefa/. (2000). Polygonal faults can be planar or listric (Fig. 3). Extreme listricity is rare and is seen, for example, when a mobile unit occurs at the base of the deformed layer (e.g. Stuevold et al. 2003). More commonly, polygonal faults flatten gently with depth, changing dip by about 10° over vertical dis-
tances of a kilometre or so. True dips of polygonal fault planes measured orthogonal to strike on different 3D surveys exhibit a large range from c. 25° to c. 65°. In areas where different tiers of polygonal faults span a considerable depth range, it is apparent that fault plane dip decreases with increasing burial depth and decreasing porosity of the deformed units. This implies that the dip can be modified by compactional flattening either during or after the faults were active (Stuevold et al. 2003). Precise quantification of this effect of compaction in case studies is hampered by the fact that different tiers of faults rarely form in two or more stratigraphic intervals with identical initial physical properties. It is also possible that some rotation of fault planes occurs during extensional deformation. The cross-sectional geometry through a tier of polygonal faults is greatly influenced by the dip direction of the individual faults and if and how these dip directions are grouped or alternate. Where there is a strong influence of slope, it is common to see a strong bias to either an upslope or down slope dip direction, with only a minority of faults adopting the opposing sense of dip (Higgs & McClay 1993; Clausen etal 1999). Polygonal fault systems are characterized by a large range of fault strikes. Where strikes are almost randomly oriented, a classical polygonal plan form geometry results (Fig. 6a). However, it must be stressed that a large variety of plan form patterns
228
J. CARTWRIGHT ETAL
Fig. 5. Flattened time slice of the reservoir interval in the Alba Field, central North Sea, showing the outline of the reservoir and the polygonal fault pattern in the adjacent mudstone interval (from Lonergan & Cartwright 1999). This image shows the localization of polygonal faults to within the clay rich facies and not the sand-rich facies of the main reservoir body (sinuous feature in the centre of the map).
have been mapped, from curved (Fig. 6b) to rectangular (Fig. 6c). These variations in the basic polygonal plan form can arise from external influences such as regional slope, tectonic context, or basement topography, or from intrinsic variation in the physical properties or the thickness of the deforming interval. It is important when judging the 'polygonality' of a fault array not to make this judgement on too small an area of the entire deformed region, since it is quite common to find almost unidirectional strikes in small sub-areas of a polygonal fault system under the influence, for example, of a steep local slope. The range of fault strikes in a polygonal fault system means that as these faults grow and lengthen, a high degree of intersection occurs within the array between faults where the intersection angles may range from gently acute to orthogonal (Lonergan et al. 1998). Intersection geometries and branch line topologies are, therefore, much more complex than for many tectonic normal fault arrays. For tiers exhibiting low bulk areal strains, the number of intersections is less than for higher bulk strains. Several common types of fault intersection have
been recognised to date (Lonergan et al 1998; Stuevold et al. 2003) but the analysis of intersection geometries is in general poorly understood and in particular, we have no general model for the kinematic evolution of complex intersections. Reconstructing the growth of a set of polygonal faults requires a full understanding of the propagation mechanics. It is not clear, at present, whether the simple radial propagation models used to analyse growth of tectonic normal faults apply to polygonal faults. One general observation that has implications for the mechanics is that many intersections are orthogonal or nearly orthogonal even when the strike of the fault segments is not. Often, one segment changes its trajectory forming a curved trace so that this condition of orthogonal intersection is satisfied (Fig. 6b). This tendency to intersect orthogonally was recognized for joint systems and attributed to stress field modifications near the joint (Lachenbruch 1962) and a similar mechanical explanation might apply for polygonal fault intersections (Lonergan etal 1998). The three-dimensional geometry of polygonal fault systems is invariably complex and difficult to appreciate from simple 2D cross-sections. On any 2D profile, most of the fault planes will be intersected at an oblique angle to the maximum dip and strike and oblique fault cuts will be the norm. Where the dip direction reverses, conjugate intersections can result (between faults with similar strike) and these can occur either within the deforming interval, or in some cases at or close to the base of the tier (e.g.Watterson et al. 2000). Footwalls and hanging walls exhibit deformational features that are characteristic of the strains associated with fault propagation as described, for example, by Barnett et al. (1987). This is most commonly expressed as minor folding (normal and reverse drag) immediately adjacent to fault planes (Fig. 7a). The fold wavelength is typically much less than the vertical extent of the fault plane. Upper tips often terminate in Mode II monoclinal tipline folds. Basal tips either terminate in a detachment, with a tight rollover fold close to the zone of maximum fault plane curvature, or simply tip out at the base of a tier without becoming strongly listric (Fig. 7b). In this latter case, it is common to observe significant thickness changes (>30%) across the fault in the region adjacent to the tip. Where faults are closely spaced and particularly if they are grouped into sets with similar strikes and dip directions, these fault-related folds can combine to form complex geometries, with synclinal structures often alternating with anticlinal structures. These folds have been attributed by several authors to the density inversion mechanism. This raises the important question as to whether the folds are responsible for the development of the faults, or are a consequence of the strain associated with fault propagation (see 'Genetic mechanisms').
THE GENESIS OF POLYGONAL FAULT SYSTEMS: A REVIEW
Fig. 6. Some typical plan form patterns of polygonal fault systems varying from highly polygonal, (a), to curved, (b), rectangular, (c). Length scale units are 1 km (b, c). These maps show the large variation in the expression of the polygonal fault style i.e. in spacing, orientation, intersection relationships and linearity of fault segments.
Strain The strain associated with polygonal fault systems is difficult to quantify with any high precision from 3D seismic surveys because the observable strain is limited by the vertical and lateral resolution of the seismic data. Small faults and fractures are undetect-
229
able beneath some critical limit of throw and these could make a significant contribution to the total strain and strain distribution. The seismically resolved strain is almost certainly an underestimate of the true strain and might misrepresent the directional properties of the strain field if there are sufficient numbers of sub-seismic faults in a given orientation. This important caveat should be borne in mind throughout this discussion and more generally when assessing polygonal fault systems. The strain in polygonally faulted sediments is truly three-dimensional and includes compactional strains, so it is questionable whether 1-D or 2-D strain measurements are representative of the bulk strain. Measuring the observable strain due to faulting from seismic cross-sections only yields the linear strain if the slip vectors on all the faults are known (Fig. 8). Since the slip vectors are unknown, the linear strain cannot be measured. If it is assumed that slip is orthogonal to strike (i.e. all faults are pure dip-slip structures), an approximation to the linear strain in that direction can be obtained by taking linear traverses across the fault system and measuring the change in line length along the survey line (Cartwright & Lonergan 1996). However, any linear traverse will cross many faults obliquely and this will always produce overestimates of the true linear strain along that traverse (Fig. 8). Summing the heaves of all the faults in a given area of a map generated from 3D seismic interpretation and comparing this to the mapped area can find an approximation to the areal strain. However, there are many problems with accurate definition of hanging wall and footwall cut-offs,
230
J. CARTWRIGHT ETAL
Fig. 7. Seismic expression of features typically associated with propagation of normal faults as seen in polygonal fault systems. Fault propagation folds expressed as normal and reverse drag are seen (a), along with upper and basal tip folds, (b).
because of the sampling interval, the lateral resolution and the pixel representation used in the digital mapping (Lister 2002). Areal strain values are also of limited value in that they do not give any indication of the directional biases in the fault strike distribution that might be present in a given fault pattern. Understanding the vertical distribution of strain on polygonal faults provides critical insight into the growth history. Most polygonal faults have displacement distributions similar to those idealized for blind normal faults by Barnett et al (1987). They have a maximum value located near the middle of the fault plane, with radially decreasing displace-
ment towards the fault tips. In some cases, the maximum displacement is skewed slightly towards the basal tip (e.g. Watterson et al. 2000). Where one tier supersedes an earlier formed one, a single fault may display two displacement maxima, one located near the centre of each tier (Stuevold et al 2003). In many cases it is possible to argue that the faults nucleated near the centre of tiers if it is assumed that this corresponds to the position of maximum displacement on the fault (the radial propagation model; see Barnett et al 1987). In other cases, it has been argued from conjugate intersections sited at the base of a tier that nucleation and growth was from
THE GENESIS OF POLYGONAL FAULT SYSTEMS: A REVIEW
Fig. 8. Schematic map of a group of polygonal faults and two approaches to measuring linear strain. If it is assumed that slip vectors, (S), are orthogonal to strike, profiles drawn for line balancing should include the heave portions (HI,2, etc.), which will not lie in the line of section. Line balancing in a straight line (line segments Jl,2,3; heave segments 11,2) will result in overestimation of the extensional strain.
the base upwards (Watterson et al. 2000). These simple observations do not apply once faults form abutting intersections with neighbouring fault segments in the array or with a mechanically confining layer at depth, and the fault can no longer propagate radially in all directions. The simplest interpretation of the observed patterns of displacement variation on typical polygonal faults is that the locus of maximum displacement represents the site of initial fault nucleation and that the fault subsequently grew by radial propagation as a blind fault as the displacement accrued (Fig. 9). In any given polygonal fault system, some blind faults evidently grow to intersect the sediment/water interface, such that they become small growth faults, with thickness differences between hanging wall and footwall (e.g. Fig. 7). Other faults remained blind until they ceased to be active.
Genetic mechanisms Gravity sliding/collapse In this type of model, polygonal faults are regarded as the result of sliding down a slope, with a basal detachment at the boundary separating the intraformational faults from undeformed, deeper sedimen-
231
Fig. 9. Schematic model for upward propagation of polygonal fault systems showing (A) initial nucleation at some moderate depth beneath the seabed (c. 200 m or less), followed by (B) radial propagation until the fault tips out against the basal boundary at depth and intersects the seabed to exert an influence on sediment thickness. Some of the faults tip out without ever intersecting the seabed and thus remain as blind faults through their growth history.
tary units (e.g. Higgs & McClay 1993; Clausen et al. 1999). The downslope gravitational stress provides the necessary conditions for failure, but this would be expected to produce a strong alignment of fault strikes parallel with the slope contours and the extension should also be balanced by down slope contractional strain of a similar amount. Since this bias of strikes is obviously not the case for most polygonal fault systems, gravity sliding is not considered a viable mechanism, although down slope gravitational stresses can provide a modifying influence on the type of polygonal pattern that evolves wherever a strong slope is present. Equally, there is no evidence for the down slope contraction that this model would predict, although this could potentially be concealed in some form of pervasive ductile contractional strain/porosity loss without the development of any seismically resolvable feature. Whilst many polygonal fault systems are known from passive continental margin settings where such a slope would have been a factor (Cartwright & Dewhurst 1998), they are also widely developed on basin floors, with no discernible slope on any part of the deformed sequence, and this slope-induced mechanism cannot be viable as a primary controlling process.
J.CARTWRIGHTCTAL.
232
Density inversion In this model (Henriet et al. 1989), conditions for a reversed density gradient are established by sealing of clay units during early burial, and this density inversion then leads to folding of the sediments, with consequent fracturing of the folded regions, and dewatering of the sediments eventually leading to a restored equilibrium density gradient (Fig. 1). In detail, the model outlines how initial deposition of clays is followed by the development of local seals which prevent the uniform dewatering under increasing burial loads (Fig. 1, stages I, II). Henriet et al (1989) described how preferential drainage of pore fluids from compacting clays at the top and base of a clay-rich interval could lead to self-sealing of the centre of the body, and how this could lead to density inversion. They also suggested that overpressure would develop through under-compaction and that this would reduce frictional resistance to shear of the clay particles, thus enhancing their ability to flow into wavy deformation structures (domal folds) that grew in response to the density inversion. A deformational system consisting of regularly arranged domal folds had previously been observed in experimental studies of density inversion phenomena by Anketell et al. (1970). Henriet et al (1989) invoked the fluid dynamical concept of the Rayleigh-Taylor instability in the deformation of an interface between two viscous fluids to explain the wavy deformational pattern. In this concept, the interface between a high-density upper layer and an underlying layer with lower density (and possibly viscosity) develops a sinusoidal instability that may evolve into a pattern of regularly spaced upwellings of lower density fluid into the denser layer. The wavelength of these instabilities is dependent on the density contrast and the kinematic viscosities of the two layers (Turcotte & Schubert 1982), but Henriet et al (1989) were not able to quantify the viscosities involved at the time of fault development and hence relate this to the wavelength of the structures. This problem of lack of constraint of the physical properties of the deforming units at the time of deformation is a general one and applies to consideration of the other mechanisms. Henriet et al (1989) proposed several possible ways that normal faults could form in association with the density inversion folds: (1) (2) (3) (4)
as typical Andersonian faults due to deviatoric stress under increasing burial load; as a result of viscous drag under the overlying flowing mobile clay layer; as the result of sliding off the rising clay domes; as collapse structures resulting from mass withdrawal (of mobile clay).
Watterson et al (2000) took the main elements of the density inversion model and modified it to explain the specific features of the Lake Hope 3D survey in the Eromanga Basin, in Australia. In their case study area, Watterson et al (2000) found that a large proportion of the polygonal faults appeared to form conjugate intersections near the base of a tier and from this argued that the faults must have propagated upwards from a basal mobile layer, which originally formed the lower, less dense layer of a reverse density stratification. They also found that the three-dimensional geometry of the faults had a pattern that was composed of anticlinal ridges that were graben-like in cross-section and separated by synclinal cells. They stated that this pattern was 'so similar to those produced by Rayleigh-Taylor instabilities that we find it difficult to avoid the conclusion that density inversion was the dominant factor in itsformation'. They proposed that the faults developed during the growth of a spoke pattern of ridge-like folds. These folds are described as having a 'significant component of concentric geometry', and strain differences between anticlinal and synclinal domains of the ' concentric fold stacks' require discontinuities between the two types of domain to maintain overall strain compatibility. In this view of the faults as 'geometrically necessary discontinuities', they are regarded as intrinsic to the folding rather than the product of a separate event or stage of development. The crux of this model then, is that the polygonal faults serve to accommodate the strain 'misfit' between anticlines and synclines. This misfit arises because of different radii of curvature and hence different amounts of layer thinning due to curvature. This effect can be quantified (Watterson et al 2000, Appendix) and should match the observed vertical displacement variations on the polygonal faults. However, Watterson et al (2000) did not produce any measurements of curvature strain in support of their model for density inversion, so it was not possible for them to test the model from these data (see additional discussion in James 2000; Walsh et al 2000). Weaknesses of the density inversion model The density inversion model offers a unifying explanation for the some structural styles associated with polygonal faults. In particular, it explains the polygonal geometry of the faults by linking fault nucleation to the plan form geometry of density inversion folds, and therein lies its principal weakness. Some polygonal fault systems do indeed have a wavy or folded deformational structure upon which the fault pattern can be thought of as being superimposed. Excellent examples have been presented by Watterson et al (2000). However, for the model to be valid, this relationship between folds and faults should be universal. Many polygonal fault systems
THE GENESIS OF POLYGONAL FAULT SYSTEMS: A REVIEW
are not, however, characterized by a regular arrangement of anticlinal and synclinal density inversion folds that this model requires (e.g. Lonergan & Cartwright 1999) and it is difficult to invoke density inversion where there is no evidence for the precursor folding required by the model. Density inversion folding is commonly observed at the scale of individual beds (cm to m), but has only recently been recognized at the kilometre length scale typical of many polygonal faults (Davies et al. 1999)(Fig. 10). The density inversion folds mapped by Davies et al. (1999) have a polygonal planform with a wavelength of a kilometre (Fig. 10), amplitudes of approximately 50 m near the base of the folded sequence, and maximum flank dips of 7°. They are not associated with any seismically resolvable faults. They display an upward reduction in fold amplitude in contrast to the modelled geometry of density inversion folds depicted by Watterson et al (2000). The geometry of the density inversion folds shown in Figure 10 provides a useful template to discuss the extensional strain that could be expected to result from outer arc bending. The largest folds in this array of density inversion structures have maximum linear strains due to outer arc extension of <2% and most folds have an outer arc strain of < 1 % (Fig. 11). The linear strains observed in many polygonal fault systems are of the order of 5-10% (Cartwright & Lonergan 1996), which is significantly larger than the outer arc extensional strains measured on the density inversion folds in the Faeroe-Shetland Basin (Fig. 11). Folds with flank dips of >20° would be required to produce outer arc extension of 5% or more and such steep dips are rarely, if ever, seen in polygonally faulted areas. In summary, density inversion folding does occur in nature at the length scales exhibited by polygonal faults and may well be a contributory factor in shaping some sub-sets of polygonal fault systems. It is difficult to view it as a universal mechanism because of the mismatch of strain and the absence of a universal correlation between folds and faults. It is also difficult to test this model because the physical properties leading to density inversion are by definition transient and cannot be reconstructed with any certainty for the time of the deformation. Syneresis Syneresis is by definition a spontaneous contraction (shrinkage) without evaporation, but is a process that is specifically restricted to gels (Brinker & Scherer 1990). To invoke syneresis for the development of polygonal faults thus requires that the deforming interval be in a gel state at the time of deformation. Gels are a framework of colloidal particles, and the primary condition for gel formation is the very fine size range of the constituent particles (clay size range). The global distribution of polygonal fault
233
systems is restricted to mainly ultra-fine grained sediments with very small particle sizes. These sediments fall into the range of colloidal materials and this means they had the potential for gel formation on deposition, although it is not possible to be sure they were actually gels when deposited. Syneresis is, in general, a poorly understood physical process, although its effects are widely appreciated in other scientific disciplines such as chemical engineering and food science (Scherer 1989; Brinker & Scherer 1990). Cartwright & Dewhurst (1998) and Dewhurst et al (1999a) offered some possible physical mechanisms for syneresis that might have occurred in the specific lithological and geochemical contexts of polygonal fault development. For example, where there is a high concentration of smectite in the deforming units, changes in ionic concentrations or salinity can induce syneresis through osmotic effects or through suppression of the double layer repulsive forces (van Olphen 1977). For further explanation of the physical and chemical aspects of the syneresis mechanism as it applies to the genesis of polygonal faults the reader is referred to Dewhurst^al (1999a). The syneresis model for polygonal fault development builds directly from the kinematic argument advanced by Cartwright & Lonergan (1996), that in a pinned system (i.e. no lateral strain, Lambe & Whitman 1979), extensional strains expressed through normal faulting are the result of volumetric contraction (Fig. 12). Assuming that the errors in measuring bed length on the seismic data are small and provided the seismic imaging of reflection cutoffs against fault planes is representative of the true geometry (i.e. there are no ductile shear zones along fault planes) then the observed reduction in bed length across the deformed interval cannot be extensional in a true sense, because the sidewalls are pinned. Unless the extensional strain is balanced locally by equivalent contractional strains (e.g. thrusting, folding) and in the absence of mass transfer out of the system (e.g. by diapirism), the only remaining conclusion that can be drawn is that there must be pervasive layer-parallel shortening of an amount equivalent to the extensional strain. This means that bed length and porosity would not be conserved during the deformation. Syneresis, by definition, is volumetric contraction and thus offers a possible physical explanation for the volumetric strains of polygonally faulted intervals measured from reduction in bed lengths on seismic data. For a clay layer undergoing syneresis it is envisaged that the particles in the porous gel framework move closer together under the influence of a net increase in inter-particle attractive forces. Resistance to this motion by frictional coupling of the layer boundaries will result in a body force that is tensile, acting to oppose the contraction, but resolved in the
234
J. CARTWRIGHT ETAL
Fig. 10. (A) Seismic profile across density inversion folds from the Faeroe-Shetland Basin showing the typical geometry. Note the upward decrease in fold amplitude and the absence of any incipient faulting across the folds. (B) Dip attribute image of the density inversion folds shown in (A). From Davies et al. (1999).
direction of the frictional coupling. In the syneresis model, it is envisaged that certain conditions are established during sedimentation or early burial that trigger the spontaneous contraction of the material under fully saturated conditions. If this contraction is resisted, for example, by basal boundary layer coupling, then horizontal tension will act in all direc-
tions, superimposed on any other stresses resulting from tectonics, loading or local slopes. The tensile stresses resulting from syneresis could lead to the propagation of mode I fractures (at shallow burial depths and low vertical effective stress) or to the propagation of shear fractures at greater depths where the vertical effective stress is
THE GENESIS OF POLYGONAL FAULT SYSTEMS: A REVIEW
235
Fig. 11. (A) Comparison of linear strains measured orthogonally across a group of polygonal faults (extension = 7.5%) and across a simple density inversion fold (extension = 1.7%). (B) True scale section across a fold arc where extension = 5%. Note the much larger flank dips required for this increase in extension when solely due to folding compared to either of the examples in (A).
above a threshold value for shear failure (Dewhurst et al 1999a, b). As the material continues to contract and as new layers are added on at the surface, the shear fractures would radially propagate as a function of accumulating displacement (Fig. 9). In the syneresis model, it is proposed that the displacement is added as a direct response to horizontal contraction of the layering. This contraction is in turn a function of pore fluid expulsion: syneresis and
contraction cannot proceed without a means for the pore fluid to be expelled (Scherer 1989). Since pore fluid expulsion is likely to be heterogeneous, the nucleation sites where syneresis begins would probably be randomly distributed. There would possibly be regions that are losing pore fluid, and others that are receiving pore fluid (Fig. 13a). Hence, there would be fluid pressure gradients within the body due to syneresis, and in all probability, regions of
236
J. CARTWRIGHT ETAL
Fig. 12. Contrasting strain paths in vertical section through a polygonal fault system (from Cartwright & Lonergan 1996). For compaction with extension, the system is not pinned laterally and the extension is a typical pure shear deformation where line length is conserved. For compaction with contraction, the system is pinned (condition of no lateral strain), so line length is not conserved and the lateral contraction is expressed in extensional faulting even though there is no net extension. L0 is the original line length and the individual segments of the deformed bedding are denoted by Lj. Z0 and Z{ refer to initial and final thicknesses of the deformed interval, respectively.
inflationary overpressure. The localization of sites of anomalous fluid pressure might well play a role in triggering failure and initial fracture propagation. In this way, randomly oriented fractures might well nucleate around local regions of active syneresis (Fig. 13b). Once the body is dissected by actively propagating shear fractures, these might act as conduits for expelled pore fluid, particularly during discrete slip events. This would tend to add permeability heterogeneity to the already heterogeneous distribution of syneresing and non-syneresing regions, with possible feedbacks set up between the processes of syneresis, pore fluid expulsion and fault propagation. This could result in the localization of pore fluid expulsion via the fracture network, and inhibit syneresis in areas of net inflow of pore fluid (Fig. 13c). This in turn, would produce lateral heterogeneity in porosity and bed thickness, which could eventually be systematically arranged relative to the major fault pathways. In this context, it is important to note that Verchuren (1992) argued strongly for fluid flow along polygonal faults from observation of displaced microfossils in fault gouge.
In a sedimentary succession undergoing syneresis, therefore, it is possible to consider the distribution of contractional strain as being partly a function of the permeability structure and partly a result of fault propagation effects. Barnett et al (1987) and Walsh & Watterson (1987) elegantly described the strains in the region surrounding propagating, blind normal faults. Displacement gradients, near tip strains and wall rock folding should all be expected to develop during the growth of blind tectonic faults, and these same fault propagation strains should be equally expected for propagating polygonal faults (e.g. Fig. 7). The polygonal faults will continue to grow by addition of displacement and if possible, fault surface area, as long as the material is actively contracting, simply because there is a need for balancing the contraction with complementary extension. In the syneresis model, therefore, faults can be viewed as representing discontinuities that bound regions of differential syncretic contraction. They are not closed boundaries until the faults have propagated to form completely circumferential structures
THE GENESIS OF POLYGONAL FAULT SYSTEMS: A REVIEW
237
Fig. 13. A schematic model for polygonal fault nucleation and propagation based on syneresis showing localization of regions undergoing syneresis, (A), followed by initial stages of fault propagation, (B), leading to feedback between fluid escape and the enhanced permeability associated with the faults, (C) (see text for additional discussion).
with fully closed intersections. This view of the faulting as necessary discontinuities to permit differential strain accumulation is similar in some respects to that expressed by Watterson et al (2000), but with an entirely contrasting physical origin. The variation of displacement over fault surfaces is accompanied by folding and abrupt thickness variations close to the
basal, upper and lateral tips are all to be expected from the propagation of blind normal faults, as noted above. This variation in displacement over fault surfaces can often result in accompanying 'roll-over' folds, which, when developed on adjacent faults of opposite fault plane dip, will tend to produce anticlinal and synclinal composite roll-over folds
238
J. CARTWRIGHT ETAL
are 'driven' to propagate from the contraction of the central layer. These hypothetical cases represent just a few of the possibilities for fault growth in the context of a contracting medium, where the accumulation of strain is clearly governed by several contributory elements. To these, the component of strain due to simple one-dimensional consolidation under gravity should be added.
Fig. 14. A hypothetical propagation sequence for the development of polygonal faults in alternating depositional units that have properties that favour (SI, S2) or preclude (Nsl, Ns2) syneresis during shallow burial. Propagation would be driven by the deformation of the syneresing layers and this would lead to complex distribution of displacement on the faults and various Linkage geometries between alternating layers.
(Cartwright 19940). These can mimic density inversion folds and could well be an explanation of structures that have been interpreted as resulting from density inversion. In addition to the variation of displacement arising from propagation, if there is any lateral or vertical variation in the physical properties of the material that affect its ability to synerese, we could expect lateral and vertical variations in the amount of contractional strain and hence in the extensional strain required to balance it. Layers of a material prone to syneresis might be interbedded with some that are not so disposed and this could well lead to complex displacement distributions on fault surfaces (Fig. 14). A syneresing layer could be the nucleus for faults that subsequently propagate into nonsyneresing, but weak layers above and below and the propagation into these layers might be associated with contrasting displacement gradients, as the tips
Weaknesses of the syneresis model The syneresis model can potentially explain every geometrical and kinematic characteristic of polygonal fault systems, but it might be perceived as a weak explanation of these structures for two reasons: (1) the complex physical and chemical processes of syneresis have not been described in detail for the variety of finegrained sediment compositions in which polygonal faults are observed; and (2) known examples of syneresis are at a comparatively small scale compared with basin-wide polygonal fault systems. The problems of scale and timing apply to the syneresis model as they do equally to the density inversion model. The vast scale over which they are developed is perhaps the most difficult aspect of polygonal fault systems to appreciate (Cartwright & Lonergan 1997). Any externally forced mechanism invoked as a trigger for syneresis, such as osmosis or cation diffusion (Dewhurst et al 19990), would have to be operative approximately synchronously over many tens of thousands of square kilometres. Hydrogeological and hydrogeochemical conditions are likely to vary laterally over distances of >100 km, simply because of variable subcrop to any polygonally faulted interval, and the contrasting juxtapositions of underlying lithologies. It seems much more likely that any viable trigger for syneresis on the appropriate length scale must be related to the initial physical properties of the deforming units. Given the observation that polygonal faults are recognized in sediments with highly contrasting mineralogy and pore fluid chemistry, it would thus be necessary to invoke different physical and chemical conditions for syneresis for each major lithofacies (Cartwright & Dewhurst 1998). A final weakness of the syneresis model is the lack of detailed constraints on the timing of the deformation. Experimental work on syneresis shows that it is a rapid process under laboratory conditions (Jungst 1934; White 1961; Brinker & Scherer 1990), but the high strain rates achieved under experimental conditions are not likely to occur under real burial conditions because syneresis is buffered by the permeability of the medium and the flux of pore fluid out of the system (Scherer 1989; Dewhurst et al. I999a). It is difficult to obtain a reliable prediction of the strain rate likely with syneresis from a theoretical analysis and hence test these predictions against observations. Equally, there are no in situ physical
THE GENESIS OF POLYGONAL FAULT SYSTEMS: A REVIEW
properties data from a shallow polygonal fault system that would allow cross matching of permeability, porosity and fluid pressure from the natural example with those predicted from theory.
239
any given fault increases the horizontal component of compression in the direction containing the slip vector and this in turn forces expulsion of pore fluid so that the extra horizontal space required by the addition of fault heave is matched by a component of horizontal compaction.
Gravitational Loading The fundamental tenet of this model (Goulty 2001) is that normal gravitational loading of certain types of sediments can lead to failure without the need to invoke additional stresses (e.g. tension) or overpressure (reduction of effective stresses). In this model, it is proposed that the property of the sediments that allows them to fail under confined vertical loading is their exceptionally low coefficient of friction. This model is in many respects as radical as the syneresis model because it is counter to many decades of theoretical and experimental soil mechanics. It advances the concept that some sediments are so weak that confined loading induces failure rather than simply inducing consolidation of the material (classical ID consolidation). To advance this model, Goulty (2001) examined the conditions for failure using a Mohr-Coulomb construction. He argued that 'for many real sediments, the Mohr-Coulomb failure envelope corresponds to the residual shear strength1, i.e., the strength on an existing failure plane. The residual shear strength is significantly lower than the value of peak shear strength, which governs the conditions of initial failure. Goulty (2001) noted that average fault plane dips quoted by Cartwright & Lonergan (1996) were 45°. He then used a Mohr-Coulomb construction to show that shear failure would occur on a fault plane dipping at 49.7°, for sediments with a low coefficient of friction (u, with a value of 0.166) if the ratio of horizontal to vertical effective stress (the K0 value) was 0.72. He argued that these values are reasonable to explain continued failure on an existing fault plane if the residual shear strength is used instead of peak shear strength. In developing his model, Goulty (2001) accepted the volumetric contraction condition of Cartwright & Lonergan (1996) and linked the compaction in horizontal directions to the repeated slip of faults by stating that 'the horizontal component of slip . . . must be accommodated by compressional strains in the bedding in the same horizontal orientation'. Thus, the slip on the faults necessitates complimentary shortening in the layer-parallel direction. This is exactly the reverse of the syneresis model, where layer-parallel shortening requires accommodation by extensional faulting. Goulty (2001) elected to explain the horizontal compressional strains as resulting from a faulting-induced increase of the least principal effective compressive stress. A fundamental tenet of this model therefore, is that slip on
Weaknesses of the gravitational loading model. This model depends on the existence of critical values of two main parameters, u and KQ, for the sediments undergoing deformation. The value of u for representative types of clay-rich sediments can be estimated from laboratory measurements, but Goulty (2001) is correct in challenging the validity of these under the natural conditions represented by the development of polygonal fault systems. In situ data from a site of actively deforming polygonal faults are, of course, what is needed, but these data do not exist at present. Knowledge of K0, for example, is very poor, even by extrapolation from experimental studies (Maltman 1994) and in situ calibration of K0 is beyond current technological capabilities. These uncertainties mean that the Goulty model is virtually untestable under in situ conditions and would be extremely difficult to evaluate from units in which the polygonal faults had ceased to be active because of the assumptions required to reconstruct the values of the key parameters that applied during the initiation of failure. The problems here partly relate to the difficulty in extrapolating from laboratory test conditions to more realistic loading conditions appropriate to the basinal contexts in which polygonal faults are developed. It should be recalled that although we have no direct measurements of strain rates for polygonal faults, sedimentation rates (and overburden loading) for the fine grained sequences in which the faults grow are extremely small, typically 10-50m/Ma (Cartwright & Dewhurst 1998), so any mechanism based on overburden loading needs to address this key observation and balance the rate of load addition with the likely flux of pore fluid through highly porous media. In addition to the lack of constraints of the key parameters governing failure, the emphasis placed on the conditions for re-shear in the model, rather than initial failure is also questionable. The Goulty model carries no explicit explanation for fault nucleation and subsequent propagation and does not explain what perturbs K0 from its value during normal compaction to a value promoting failure. Prior to the establishment of normal faults with dimensions of 100s of metres, shear failure of considerable volumes of intact sediment must have occurred and it is difficult to see how the conditions for re-shear can possibly apply for this early phase of fault nucleation and growth. Some viable mechanism is required to get to the stage of a fully linked and continuous fault plane before the lower shear
240
J. CARTWRIGHTCTAL.
Fig. 15. A seismic profile showing shallow buried and possibly still active polygonal faults in the V0ring Basin, Norway. The profile illustrates the two-tier geometry of this polygonal fault system. Note that the upper tips of many of the faults in the upper tier are located within 100 m of the present sea bed. 1ms is approximately 0.9 m.
resistance of the established fault plane (residual shearing resistance) comes fully into play, as opposed to the peak shear strength. It seems unlikely that virgin failure involved in fault propagation could be approximated with the residual shear strength, unless the problem is simplified to the grain scale, when slip at all grain contacts could be regarded as overcoming friction on microfractures (grain boundaries). If that was the case, then why should these microfractures at grain boundaries aggregate to give discrete normal faults at the macro scale, rather than as an alternative, simply allow normal consolidation to proceed without macroscale failure? All sediments fail repeatedly during consolidation when viewed at the grain scale because of the need for slippage across grain boundaries. In the fully (laterally) confined conditions of oedometer (uniaxial) tests, for example, clay-rich sediments undergo consolidation without macroscopic shear failure. Why then should weak sediments in a passively subsiding basin be any different and compact with normal faulting as opposed to simple consolidation by grain scale micro-slippage, if loading by the overburden is the only mechanism driving consolidation? Shear failure of consolidating clay sediments is routinely observed in the laboratory, but only under triaxial loading conditions, where there is space for lateral extension outwith the original confines of the test material.
As noted above, one of the most critical elements of the Goulty model is the linkage between slip and accommodation of horizontal strains. Goulty (2001) argued that horizontal stress in the direction of the least principal compressive stress must increase when slip occurs and that this is accompanied by local reduction in overpressure. These effects are envisaged to combine to produce the necessary horizontal compaction to accommodate the addition of heave. The fundamental issue here is that normal faulting is linked by Goulty (2001) to an increase in horizontal compressive stress. This is counter to all classical explanations of normal faulting in which reduction in horizontal compressive stress is a fundamental component. It is difficult to envisage the exact failure conditions that allow simultaneous decrease in the least principal compressive stress and horizontal compaction without the intervention of some perturbing force (e.g. as in syneresis). Finally, in the classical Mohr-Coulomb analysis of normal faulting, fault plane dip is related to the strength of the faulted medium and to the value of u. One possible route, therefore, to estimating u in a bulk sense for a tier of polygonal faults is by measuring the dip of polygonal faults as close to their depth of origin as possible. It is important to appreciate that during burial, normal gravitational consolidation of the host sediments results in considerable vertical flattening of fault planes, so it is essential to
THE GENESIS OF POLYGONAL FAULT SYSTEMS: A REVIEW
241
polygonal faulting in a shallow buried interval. A much more comprehensive study of fault plane dips in an Oligo-Miocene smectitic claystone sequence in the North Sea by Clausen et al. (1999) also gave steep fault plane dips (range of 48° to 85° with a mean of 73°, n = 1694). In these two cases, it is difficult to reconcile the much steeper fault plane dips and high mean value of u with the shallower fault dips and low value of u required for the failure mechanism invoked by Goulty (2001). Fig. 16. A graph of fault plane dip against strike for faults from the V0ring Basin, showing no directional control on fault dip and a typical fault plane dip of 60° +/—5°. This is a typical value for normal faults and does not imply any abnormally low value of the angle of internal friction.
find polygonal faults as close to surface as possible to minimize this effect. This important point was ignored by Goulty (2001) when arguing for a low value of u. The average value of polygonal fault plane dip of 45° quoted by Goulty (2001) in support of his use of a low value of u came from data in Cartwright & Lonergan (1996), for polygonal faults that had been buried some 1-2 km after they had ceased to be active. These fault planes would have been considerably steeper when first propagating in near-surface conditions (see also Stuevold et al this volume), so it is misleading to assume this 45° dip applied to the onset of the deformation. One of the best seismically imaged polygonal fault systems for measuring fault dips at shallow burial depths is developed in the V0ring Basin, Norway (Fig. 15). This fault system deforms an Oligocene to Pliocene interval approximately 1000 m thick, consisting of claystones and biosiliceous oozes. Many of the faults approach to within a few tens of metres of the present day seabed and these faults could well be active today. Porosity values for this interval extrapolated from the nearest Ocean Drilling Project borehole (ODP site 644) are in the range 50-75%, so there has been relatively modest vertical compaction of this interval. Measurements of fault plane dip made from true dip, depth converted seismic profiles across 34 faults in this area exhibit a range from 51.8° to 64.4° with a mean value of 59.7° (Fig. 16). These values were measured at an approximate depth below seabed of 200 m. Fault plane geometry is planar in the upper 300-400 m of the cross-sectional profile. These dips are suggestive of a mean value of u of 0.565 (range of values from 0.21 to 0.80). It would be unwise to draw general conclusions about all polygonal fault systems from these data because fault plane dips are likely to vary according to gross lithology of the faulted interval. However, they do give an example of the probable value of |m, in at least one clear case of
Conclusions Polygonal faults are remarkable structures. They share many common features with regular tectonic normal faults, but their organization into a threedimensional network of great complexity sets them apart from typical tectonic fault arrays. Three main genetic models have been advanced to explain their occurrence. These three models share the general theme of early development and a role in compaction and pore fluid expulsion. From the strengths and weaknesses of these three models outlined above, it is argued here that syneresis offers the best current explanation of polygonal fault genesis because it explicitly accounts for the volumetric strain expressed in seismically observed fault offsets of bedding and it matches the observation that polygonal faults are restricted in occurrence to dominantly fine-grained depositional systems. Density inversion is not accepted here as a general explanation because the values of extensional strain required by outer arc bending of density inversion folds do not equate with typical strain values for polygonal fault systems, notwithstanding the difficulties of accurate measurement of strain. In addition, many polygonal fault systems do not show evidence of an earlier stage of regularly spaced domal folds required by this mechanism. Gravitational loading (Goulty 2001) is questioned here as a general mechanistic explanation because measured fault plane dips in polygonal fault systems that have not experienced significant post-faulting burial suggest quite normal values of internal friction, and not the anomalously low values required by this model. At this early stage of research, none of these models should be unequivocably rejected and some of these contrasting mechanisms may even be complimentary to some extent. Density inversion, for example, is likely to occur in a system of multilayers in which one or more of those layers were undergoing syneresis. Thus far, the data gathered on polygonal faults have largely come from 2D and 3D seismic interpretation of polygonally faulted sequences. Much of our current description of their geometry and kinematic evolution is thus inherently limited by seismic
242
J. CARTWRIGHT ETAL
resolution. To discriminate between these three models will require much additional data, particularly on in situ physical properties, but also from improved seismic imaging of shallow-buried faults. Little is known about the permeability behaviour of polygonal faults, both during or after active deformation. Limited evidence from outcrops suggests they could play a significant role in vertical fluid migration through otherwise highly impermeable sequences. The same reasoning can be applied to polygonal faults developed within source rock intervals. Polygonal faults evidently play a role in deposition of reservoir sands in some continental slope settings (Stuevold et al this volume) and in postdepositional reservoir remobilization, but the details of this interaction are not yet clear and thus limit our predictive capabilities. Schlumberger GeoQuest are gratefully acknowledged for donation of interpretation software used to interpret the 3D seismic data. This research was conducted under NERC grant GR3/9906. We are also grateful to Norsk Hydro (N. Moller and T. Vejum) for allowing access to 3D seismic data. J. Turner and C. Morley are thanked for their constructive and helpful reviews of the manuscript.
References ANKETELL, J.M., CEGLA, J., & DZULYNSKI, S. 1970. On the deformational structures in systems with reversed density gradients. Annales of the Geological Society of Poland, 15,1-29. BARNETT, J.A.M., MORTIMER, J., RIPPON, J.H. & WALSH, J..J. & WATTERSON, J. 1987. Displacement geometry in the volume containing a single normal fault. Bulletin of the American Association of Petroleum Geologists, 71,925-937. BRINKER, C J. & SCHERER, G.W. 1990. Sol-gel science: the physics and chemistry of gel processing. Academic press, San Diego, 908pp. CARTWRIGHT, J.A.I9940 Episodic basin-wide hydrofracturing of overpressured Early Cenozoic mudrock sequences in the North Sea Basin. Marine and Petroleum Geology, 11, 587-607. CARTWRIGHT, J.A. I994b. Episodic collapse of geopressured shale sequences in the North Sea Basin. Geology, 22,447-450. CARTWRIGHT, J.A. 1996. Polygonal Fault Systems: a new type of geological structure revealed by 3D seismic data. In: WEIMER, P. (ed.) Application of 3-D seismic Data to Exploration and Production, American Association of Petroleum Geologists, Studies in Geology,42,225-23l. CARTWRIGHT, J.A. 1997. Polygonal Fault Systems in thick shale sequences. In: YARDLEY, B, & JAMTVEIT, B. (eds) Fluid Flow Processes in the Continental Crust. Chapman & Hall, London, 81-104. CARTWRIGHT, J.A. & DEWHURST, D. 1998. Layer-bound compaction faults in fine-grained Sediments. Bulletin of the Geological Society of America, 110, 1242-1257.
CARTWRIGHT, J.A. & LONERGAN, L. 1996. Volumetric contraction during the compaction of mudrocks: a mechanism for the development of regional-scale polygonal fault systems. Basin Research, 8,183- 193. CARTWRIGHT, J. & LONERGAN, L. 1997. Polygonal fault Systems in the Eromanga and North Sea Basins: a comparison. Exploration Geophysics, 28, 323-331. CLAUSEN, J.A. GABRIELSEN, R.H., REKSNES, PA. & NYSETHER, E. 1999. Development of intraformational faults in the northern North Sea: influence of remote stresses and doming of Fennoscandia. Journal of Structural Geology, 21,1457-1475. DAVIES, R. CARTWRIGHT, J.A. & RANA, J. 1999. Polygonal density inversion structures from the Faer0e-Shetland Trough. Geology, 27,798-802. DEWHURST, D., CARTWRIGHT, J.A. & LONERGAN, L. 19990. The development of polygonal fault systems by the syneresis of fine-grained sediments. Marine and Petroleum Geology, 16,793-810. DEWHURST, D., CARTWRIGHT, J.A. & LONERGAN, L. 1999&. Three-dimensional consolidation of clay-rich sediments. Canadian GeotechnicalJournal, 36,355-362. GOULTY, N.R. 2001. Polygonal fault networks in finegrained sediments - an alternative to the syneresis mechanism. First Break, 19,69-73. HENRIET, J.P., DE BATIST, M., VAN VAERENBERGH, W. & VERSCHUREN, M. 1989. Seismic facies and clay tectonic features in the southern North Sea. Bulletin of the Belgian Geological Society, 97,457-472. HENRIET, J.P., DE BATIST, M. & VERSCHUREN, M. 1991. Early fracturing of Palaeogene clays, southernmost North Sea: relevance to mechanisms of primary hydrocarbon migration. In: SPENCER, A.M. (ed.) Generation, accumulation and production of Europe's hydrocarbons. Special Publication of the European Association of Petoleum Geologists, 1,217-227. HIGGS, W.G. & McCLAY, K.R. 1993. Analogue sandbox modelling of Miocene extensional faulting in the Outer Moray Firth. In: WILLIAMS, G.D. & DOBB, A. (eds) Tectonics and Sequence Stratigraphy. Geological Society, London, Special Publications, 71, 141-162. JAMES, D.M.D. 2000. Discussion on geometry and origin of a polygonal fault system. Journal of the Geological Society of London, 157, 1261-1265. JUNGST, H. 1934. Geological significance of syneresis. Geologische Rundschau, 25, 312-325. LACHENBRUCH, A.H. 1962. Mechanics of thermal contraction cracks and ice-wedge polygons in permafrost. Geological Society of America, Special Paper, 70. LAMBE, T.W. & WHITMAN, R.V. 1979. Soil Mechanics. John Wiley & Sons, New York. LISTER, D. 2002. Computer modelling and characterisation of intersecting fault networks. Ph.D Thesis, University of London. LONERGAN, L., CARTWRIGHT, J.A. & JOLLY, R. 1998. 3-D Geometry of Polygonal Fault Systems. Journal of Structural Geology, 20, 529-548. LONERGAN, L. & CARTWRIGHT, J.A. 1999. The role of polygonal fault systems in the reservoir geology of the Alba Field, North Sea. Bulletin of the American Association of Petroleum Geologists, 83,410-432. MALTMAN, A. 1994. The geological deformation of sediments. Chapman & Hall, London, pp.362.
THE GENESIS OF POLYGONAL FAULT SYSTEMS: A REVIEW SCHERER, G.W. 1989. Mechanics of syneresis. I. Theory. Journal of Non-Crystalline Solids. 108,28-36. STUEVOLD, L.M., FAERSETH, R.B., ARNESEN, N., MOLLER, N. & CARTWRIGHT, J.A. 2003. Polygonal faults in the Ormen Lange Field, Offshore Mid Norway. In: VAN RENSBERGEN, P., HILLIER, R.R., MALTMAN, AJ. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216,263-281. TERZHAGI, K. & PECK, R.B. 1948. Soil Mechanics in Engineering Practice. Chapman & Hall, London. TURCOTTE, D. & SCHUBERT. 1982. Geodynamics. John Wiley & Sons, New York, pp 450. VAN OLPHEN, H. 1977. An introduction to clay colloid chemistry. John Wiley & Sons, New York. VERSCHUREN, M. 1992. An integrated 3D approach to clay tectonic deformation. Ph.D Thesis, Universiteit Ghent.
243
WALSH, JJ. & WATTERSON, J. 1987. Distribution of cumulative displacement and seismic slip ona single normal fault surface. Journal of Structural Geology, 9, 1039-1046. WALSH, J.J., WATTERSON, J., NICOL, A., NELL, PR. & BRETAN, P.G. 2000. Discussion on geometry and origin of a polygonal fault system: reply. Journal of the Geological Society of London, 157,1261-1265. WATTERSON, J., WALSH, J.J., NICOL, A., NELL, PR. & BRETAN, P.G. 2000. Geometry and origin of a polygonal fault system. Journal of the Geological Society of London, 157,151-162. WHITE, W.A. 1961. Colloidal phenomena in sedimentation of argillaceous rocks. Journal of Sedimentary Petrology, 31, 560-570.
This page intentionally left blank
The geometry, growth and linkage of faults within a polygonal fault system from South Australia A. NICOL1'2, JJ. WALSH1, J. WATTERSON3, P.A.R. NELL4 & P. BRETAN4 1
Fault Analysis Group, Department of Geology, University College Dublin L69 3GP, UK (e-mail: fault@fag. ucd. ie) 2 Present address: Institute of Geological and Nuclear Sciences, Lower Hutt, New Zealand 3 Fault Analysis Group, Liverpool University Marine Laboratory, Port Erin, Isle of Man IM9 6JA, British Isles 4 Badley Earth Sciences Ltd, North Beck Lane, Hundleby, Spilsby, Lincolnshire PE23 5NB, UK Abstract: Quantitative analysis of faults within a South Australian polygonal fault system, interpreted from a 3-D onshore seismic survey, provides a basis for establishing their growth and linkage histories. The geometric characteristics of faults are consistent with an origin arising from the gravitational instability of an underlying low-density, overpressured, mobile layer. Fault size populations have scale-bound, non-power-law properties reflecting the thicknesses of the faulted and mobile layers and the strongly connected nature of the system. The spatial distributions of faults reflect the localization of conjugate faults at the top of the mobile layer and the scale of fault-bounded polygons. Displacement variations on faults show marked decreases at or adjacent to the top of the mobile layer and attest to its active role in faulting. The wide range of fault strike directions provides numerous fault intersections with high intersection angles (> 60°) forming triple-junctions at which fault linkage and capture occurred. Fault linkage and capture is attributed to a simple model in which continued movement on faults which share a mutual footwall is favoured and hanging wall faults are deactivated. The model involves thickening of the mobile layer within the footwalls of faults and thinning and eventual grounding of the overlying sequence, within their hanging walls.
This paper describes the geometric and kinematic features of faults that form part of a polygonal fault system in the Lake Hope region of South Australia. The Lake Hope polygonal fault system is typical of a distinctive and widespread type of intra-formational fault system that characteristically occurs in postrift mudstone-dominated marine sequences (e.g. Cartwright 1994; Cartwright & Lonergan 1996, 1997; Cartwright & Dewhurst 1997; Lonergan et al I998a,b; Clausen et al 1999; Lonergan & Cartwright 1999; Dewhurst et al 1999; Watterson et al 2000; Walsh et al 2000). Such fault systems are believed to have a non-tectonic origin in which gravity plays a major role, possibly arising from a density instability (Henriet et al 1991; Verschuren 1992; Watterson et al 2000; Walsh et al 2000), but lateral contraction due to volume reduction has also been proposed (Cartwright & Lonergan 1996). The Lake Hope faults are thought to have formed as a result of incipient gravitational overturn of a thin (ca 40 m) low-density, overpressured layer (Watterson etal. 2000; Walsh ef0/. 2000). In this paper, we describe the quantitative systematics of the geometry of faults within the Lake Hope fault system and suggest that they are strongly influenced by the mode of origin of these faults, i.e. deformation of a low density mobile layer. Fault population data and fault displacement analysis, together with the analysis of fault linkage are shown
to be consistent with the scale-bound nature of the system in which the presence and thickness of an underlying mobile layer is crucial.
General characteristics of the fault system The normal fault system at Lake Hope is predominantly within a Lower Cretaceous-Tertiary sequence and has been mapped from good quality 3-D seismic reflection data over a region of 138 km2 (Oldham & Gibbins 1995; Cartwright & Lonergan 1997). A map of the most intensely faulted horizon, the Top Coorikiana horizon (Fig. la), suggests an even distribution of strike directions that is confirmed by the histogram of cumulative trace length vs. strike azimuth (Fig. 2a). Maps are also available for the Top Mackunda Horizon, ca 300 m above the Coorikiana, and the Cadna-owie horizon, ca 200 m below and display a comparable distribution of fault strikes (Watterson et al 20000). The faults have throws <80m and displace all three mapped horizons in most cases where throws are >30 m. Fault dips typically range from 50 to 5 5 ° with no preferred dip direction, although cross-sections reveal that larger faults often form conjugate pairs which intersect close to the Cadna-owie horizon (Watterson et al 2000; Walsh et al 2000) on which the fault traces outline more or less regular polygons with cell diameters of mainly 0.8-2 km. The faults
Fig. 1. TWT structure contour map for Coorikiana horizon, Lake Hope 3D seismic survey (see (b) for location). Reds and yellows represent higher elevations (total elevation range is 932-1276 MS TWT). Maximum slopes on the domal structure are <1°. Boxes show the locations of fault maps in Figs 3, 6, 8, 9 and 10. Solid lines show the locations of seismic sections in (c) and (d). (b) Location map for Lake Hope polygonal fault system, (c) and (d) seismic sections with interpreted faults and horizons (Cadna-owie, Ca; Coorikiana, Co; Mackunda, M). Vertical exaggeration is c. 2.4.
THE GEOMETRY, GROWTH AND LINKAGE OF FAULTS
and fault pattern appear to have originated on the lowermost mapped horizon, the Cadna-owie and propagated upwards and laterally. Faulting has previously been attributed to the mobility of a ca 40 m thick lowdensity overpressured layer that immediately overlies the Cadna-Owie horizon (Lavering 1991; Watterson et al 2000; Walsh et al 2000). The up ward growth of the spatially ordered, but more or less randomly striking, non-vertical faults on the Cadna-owie inevitably leads to a reduced spatial ordering on younger horizons in addition to the development of numerous high-angle fault intersections up sequence.
Quantitative characteristics of fault system Fault populations Fault size populations can display a broad range of scaling properties, from the power-law distributions that are characteristic of scale-invariant (selfsimilar) systems through to the non-power distributions of scale-bound systems (e.g. Marrett & Allmendinger 1991; Gillespie et al 1993; Watterson et al 1996; Yielding et al 1996; Pickering et al 1997). Many, though not all, tectonic fault systems show scale-invariant properties, whilst scale-bound systems are perhaps best typified by strata-bound fault arrays. Here, we describe the fault population systematics of the Lake Hope fault system in an attempt to establish whether their nature reflects either a scale-invariant or scale-bound system. 1-D and 2-D fault populations are measured and analysed using conventional methods (Watterson et al 1996; Yielding 6tf al 1996). The 1-D fault throw population (Fig. 2b), comprising all throws sampled at intersections between fault traces and seismic lines, and the 2-D fault populations of both maximum throw (Fig. 2c) and trace length (Fig. 2d) were measured on the Coorikiana horizon map. Preliminary analysis of the Mackunda horizon produced comparable results. For arrays of intersecting fault segments, the 2-D data, comprising trace length and maximum throw, are strongly dependent on the criteria used for fault correlation. Where two or more faults intersect at one or more branch-points there is a degree of freedom, or subjectivity, in the correlation of the seismic fault picks that determines which fault segments are taken as constituting a through-going fault and which segment or segments are classified as splays. The data presented here are therefore comparable only with data for similar fault systems that have been correlated using the same criteria. Fault segments have been correlated mainly based on comparability of size, i.e. throw values, rather than on similarity of strike direction and the forms and
247
lengths of the resulting fault traces are a direct consequence of this choice. The rationale for this basis of correlation will be apparent in later sections but it is emphasized that there can be no 'correct' correlation criterion for faults with a history of linkage. Plots of cumulative frequency for either fault length or maximum throw do not, in some cases, allow reliable distinction between power-law and non power-law distributions (Watterson et al 1996) and frequency curves provide more reliable characterizations of distributions (Fig. 2b, c & d). A 1-D fault throw distribution (Fig. 2b) is independent of the correlation strategy but a power-law distribution, if present, is detectable only if the range of throws on the straight segment of the population curve is at least one order of magnitude (Childs et al 1990). As the valid range of recorded throws is much less than an order of magnitude (13 m-80 m), the data (Fig. 2b) cannot provide an unambiguous characterization of the throw distribution. While maximum throws range up to 80 m, a significant proportion (97% of 880) are <50 m (Fig. 2c) and the maximum throw frequency is approximately log-normal with a median value of ca 19 m (Fig. 2c). As the minimum throw resolution is between 6 m and 13 m, the decrease in frequency of throws from 30 m to 13 m is real (Fig. 2c). This decrease in frequency of small faults (in terms of their maximum throws), which demonstrably is not due to throw-resolution limitations, does not typically occur in throw distributions of tectonic fault systems with power-law size distributions. The strongly right-skewed curve for the maximum throw frequency distribution is also unusual and reflects the scale-bound nature of the faults, a consequence of the stratigraphically confined nature of this and similar fault systems (Cartwright & Dewhurst 1997; Watterson et al 2000; Walsh et al 2000). All fault systems in a layered crust are potentially scale bound (Ouillon et al 1996), but the extremely narrow stratigraphic interval within which the faults are confined, ca 1100 m in the case of the Lake Hope faults, means that the maximum up-dip dimension is attained by faults with throws of only 80 m. Because the maximum throws of faults in Figure 1 are restricted to values of no greater than twice the thickness of the mobile layer (Watterson et al 2000; Walsh et al 2000) and as fault dimensions scale with displacement, the degree to which polygonal faults are scalebound is strongly dependent on the thickness of the mobile layer. Fault systems forming in response to deformation of a relatively thin mobile layer may extend over a lesser vertical range than those faults formed in association with a thicker mobile layer. Fault-trace lengths (n = 880) on the Coorikiana horizon range from 15 m-3025 m with most (ca 91%) <1200 m and a median value of ca 350 m (Fig. 2d) and are best described by a negative-exponential
248
A.NICOLETAL.
Fig. 2. Fault populations on the Coorikiana horizon map. (a) Histogram of fault-trace azimuths binned in 5° intervals and weighted by strike length, (b) Cumulative frequency (+) and frequency (o) of fault throws (N = 1193) sampled on a 500 m spaced grid of seismic lines, (c) Cumulative frequency (+) and frequency (o) of fault-trace maximum throws (N = 880). (d) Cumulative frequency (+) and frequency (o) of fault-trace lengths (N = 880). (e) Logarithmic plot of maximum throw vs. fault-trace length (N = 880), for which best-fit line has a slope of 0.6. The inset box denotes the area of good quality/resolvable data, (f) Fault-trace spacing frequency as measured by fault-normal distance to nearest fault trace from intersection of fault traces with three cross-lines and three in-lines.
THE GEOMETRY, GROWTH AND LINKAGE OF FAULTS
distribution, i.e. random. However, trace-length data are not only dependent on the correlation criteria but also subject to the effects of throw resolution (e.g. Heffer&Bevan 1990; Watterson ef al 1996; Yielding et al 1996; Pickering et al 1997) and of the spacing of interpreted seismic lines. The marked decrease in frequency of fault trace lengths below 90 m directly reflects the 87.5 m spacing of interpreted sections. The few recorded trace-length values <87.5 m represent short traces constructed from fault picks on adjacent cross-line and in-lines and these will be only a small proportion of all traces <87.5 m. The sharp decrease in the frequency of trace lengths >lkm reflects the ca 0.8-2 km diameter typical of the faultbounded polygons (Watterson et al 2000a). Size population data for the Lake Hope fault system do not show the power-law properties that are characteristic of many tectonic fault systems. Log-normal and negative exponential populations are typical of systems that are not scale invariant and for which a characteristic scale applies. The Lake Hope fault system, as with other polygonal fault systems, are typical of scale bound geometries because they are formed and contained within well defined stratigraphic packages. In such circumstances, the dimensions and displacements of faults are linked to the thicknesses, as well as properties, of both the mobile layer and the sequence containing the faults (see Watterson et al 2000).
Spatial distributions of faults The main spatial characteristic of the Lake Hope fault system is the polygonal arrangement of faults on the Coorikiana horizon, aspects of which have previously been described by Watterson et al (2000a). Here we attempt to quantitatively characterize the spatial distribution of the faults by measuring the fault-normal spacing, which was sampled at each of the 300 intersections of fault traces with three seismic in-lines and three cross-lines spaced at intervals of approximately one quarter of the map dimension. The trace-normal distance (or spacing) to the nearest fault trace was measured at each sample point. There is no significant difference between the in-line and cross-line spacing populations; so all data were combined to provide a single spacing frequency curve (Fig. 2f). Fault spacings range from 9-1726 m with most (ca 91%) between 100-1000 m (mean 395 m, standard deviation 245 m). The spacing population distribution is approximately log-normal (Fig. 2f), with a coefficient of variation, Cv (standard deviation/mean, see Christensen & Olami 1992), of 0.62, suggesting that the fault-trace spacing has a degree of regularity (Gillespie et al 1999). This conclusion is not sensitive to the under-sampling of spacings below the
249
lateral resolution of the seismic data (ca 60 m). As many neighbouring faults dip in opposing directions and converge downwards to intersect near to the Cadna-owie horizon, fault trace spacing on the Coorikiana horizon map will not be representative of spacing on other horizons. The range of spacings and the average spacing on the Coorikiana horizon are attributed to the scale bound nature of the fault system. On this horizon, polygon boundaries are defined by closely spaced parallel faults defining graben which have widths on the scale of a few hundred metres. This spacing reflects the geometry of the graben bounding faults, which generally intersect at the top of the mobile layer. Given the ranges of fault dips for this system (44° to 60°; see Watterson et al 2000), the widths of graben are consistent with the thickness of sequence between this horizon and the mobile layer (ca 200 m). Spacings larger than ca 500 m reflect the dominant scale of the polygonal cell pattern (Watterson et al 2000), which on this horizon are ca 1 km across.
Displacement vs. length scaling of faults A plot of maximum throw (D) against fault trace length (L) shows a broad band of data with a best-fit line corresponding to D = 0.09L0 6 for all fault traces with both maximum throw > 6 m and length > 90 m. The majority of the data range in maximum displacement and trace length by 1.5 orders of magnitude (Fig. 2e). Despite the biases arising from throw resolution and the sensitivity of the data to the correlation criteria, together with the scatter of the data, the slope of the best-fit line is very likely <1.0, which is unusual. As with tectonic fault datasets (e.g. Walsh & Watterson 1988; Cowie & Scholz 1992; Gillespie et al. 1992; Schlische et al 1996) these data suggest that an increase in maximum throw is accompanied by an increase in fault dimensions. We attribute the low slope of the Lake Hope fault data set to two main factors: (i) The high degree of fault linkage which would be expected to increase fault dimensions without a commensurate increase in maximum throw (e.g. Cartwright et al 1995). (ii) The upper limit placed on maximum throws (i.e. < 80 m) by the thickness of the mobile layer defines an effective ceiling on displacement while trace lengths continue to increase; thickening and thinning of the mobile layer across faults provides a maximum throw of twice the ca 40 m thickness of the layer.
Fault connectivity An attribute that provides a measure of the connectedness of a fault system and is not dependent on fault
250
A.NICOL£TAL.
correlation criteria, is the ratio of the number of branch-points to free tip-points, which for the Coorikiana horizon map is 0.7. This value is characteristic of a relatively well-connected 2-D system and is higher than the ratios, ca 0.1-0.4, for typical tectonic fault systems mapped to the same throw resolution (e.g. Manzocchi 1997). When the map resolution is effectively increased by extrapolating fault tip-points, on the basis of measured throw gradients, the connectivity is enhanced further to branchpoint/tip-point ratios of between 1 and 10 for the Coorikiana horizon, values that are typical of 2-D maps of highly connected multi-set joint systems (e.g.Gfflespieeffl/. 1993). Summary The statistical data demonstrate some of the profound differences between this polygonal fault system and typical tectonic fault systems (see also Lonergan et al. 1998a,£). The statistical differences reflect not only the obvious differences that are apparent on 2-D fault maps (i.e. a wide range of fault strike directions and high numbers of fault intersections for polygonal faults as compared to tectonic faults), but also the less obvious differences due to the vertically scale-bound nature of the fault system. These differences are consistent with polygonal faults forming in response to gravitational overturn of a mobile layer, as proposed by Watterson et al (2000). The high degree of connectivity, as shown by the high branch-point/tip-point ratios, is a feature that makes the Lake Hope fault system ideally suited to a study of fault linkage and capture.
Fault-trace linkage patterns In map view, there is a high density of fault traces (ca 3-10 faults/km2) and neither a preferred strike (Figs 1, 2a & 3) nor a preferred dip direction (e.g. Fig.3). Approximately 80% of fault traces >1.2 km in length (n = 63) have one or more branch-points which frequently occur at high angle (>60°) triplejunctions where segments of the same fault have significantly different strike directions (Figs 1 and 4). The branch-points are interpreted as having originated at the intersection of two faults, the intersected (or abutted) fault comprising segments one and two and the intersecting (or abutting) fault, equal to segment three, which subsequently merged to form a linked fault and a splay (Fig. 4). In 85-90% of cases, the intersecting fault is in the footwall of the intersected fault, a geometry comparable to the class B fault intersection type of Lonergan et al. (1998&). The predominance of footwall abutments may be, in part, a reflection of the higher proportion of the map
area which represents footwall regions, a particular feature of this level in this individual fault system and therefore, not of general significance. As is shown below, however, there is a significant difference between hanging wall and footwall intersections insofar as hanging wall intersections only rarely resulted in fault capture. This asymmetry is believed to be a key to understanding the origin and kinematics of faults within the Lake Hope system. The angle of intersection between the two initial faults, where at least one trace length was > 1.2 km, varies from 6°-90° with about half (58%) of the hanging wall and most (82%) footwall intersections at ^ 60°. Few examples have been found of intersection angles much greater than 90°, i.e. where the intersecting and intersected faults have opposed dips. Approximately 50% of faults >1.2 km in length on the Coorikiana horizon (N = 79) have curved traces of which all but a few (14%) are convex to the hanging wall (Figs 1, 3, 5 and 6). This preponderance of convex to hanging wall fault traces reflects both the curved nature of individual fault traces (e.g. Figs 3, 5 and 9) and the high proportion of footwall intersections (e.g. Figs 3, 6, 8 and 10). In three-dimensions the curved faults are cylindrical (Fig. 5) or weakly conical with steeply plunging axes and downward apices, i.e. faults are generally less curved on the Mackunda horizon than the lower horizons (Fig. 5): there are approximately twice as many cylindrical fault surfaces as conical surfaces. As the maximum dimensions of individual faults occur mostly at about the level of the Coorikiana horizon, it is likely that intersections between faults mostly occurred first at or about this level.
Fault displacement patterns and related deformation Horizon separation diagrams and horizon contours Throws and horizon geometries were measured on and, adjacent to, many of the larger faults offsetting the Coorikiana horizon. Horizon separation diagrams, also referred to as Allan diagrams, for typical linked faults are shown in Figure 7, as viewed from the hanging wall side of each fault. To avoid distortions due to the curved fault traces, each diagram has been constructed on a vertical projection surface, which everywhere parallels the curved or angular fault trace. As would be expected, irregularities on horizon cutoff lines coincide with branch-points which, in turn, often coincide with abrupt changes in strike directions of linked faults (Fig. 6). Abrupt changes in throw at branch-points is most often achieved by changes in the altitude of the footwall
THE GEOMETRY, GROWTH AND LINKAGE OF FAULTS
251
Fig. 3. Fault map of part of the Coorikiana horizon. The downthrow direction of each fault is shown. See Figure la for location. The location of the fault analysed in Fig. 11 is shown.
horizon cutoff with hanging wall horizon cutoffs on both segments of the intersected fault typically forming a continuous line. These relationships reflect the predominance of footwall intersections and produce steep throw gradients adjacent to branch-points (Figs 6b, c and d). Flat-topped displacement profiles are typical of larger faults without splays, which are unlikely to represent merged faults. The geometry of horizon contours adjacent to a fault is related to the form of the horizon separation diagram for that fault but is also dependent on other factors, specifically the form of the fault map trace and the nature of associated horizon deformations within the volume surrounding a fault. Horizon def-
ormations associated with faults are referred to as reverse drag and typically take the form of hanging wall rollover and footwall uplift (Gibson et al 1989). The angular or curved traces typical of either 'single' or linked faults are associated with distinctive horizon geometries that are modifications of the geometries associated with more or less straight fault traces. Horizon depth contours in the footwalls of faults which are concave to their footwalls, the usual case, are typically parallel to a line joining the fault tip-points, consistent with hingeing of the footwall sequence along the line connecting the tippoints at each stage of tip-point propagation (Figs 6a, 8 and 9). The footwall horizon contours are
252
A.NICOLETAL
Fig. 4. Schematic diagram showing an initial fault with two segments (1 and 2), intersected at 90° by a second fault (segment 3) in the footwall of the first fault, with linkage and growth of segments 1 and 3. (a) Map view; (b) throw profiles and; (c) oblique view of separations for each segment on a single horizon. Locations of branchpoint throw parameters presented in Figures 14 and 15 are shown by open circles in (a) and (b).
therefore typically oblique to each individual segment of the fault trace. Horizons in the footwalls of these 'trap-door' faults dip away from either the sharp bend in an angular fault trace or the mid-point of a curved trace, each of which is usually close to the highest point on its respective structure (Figs 6a, 8 and 9). Horizons in the mutual footwall of three or more faults, e.g. within a polygonal cell, are commonly concave upwards or 'dish' shaped (Fig. 6a). The horizon geometries indicate uplift of footwalls and subsidence of hanging walls, consistent with the forms of the horizon separation diagrams (Fig. 7), most of which show upwardly convex footwall cutoff traces and downwardly convex hanging wall cutoff traces. Horizon dips immediately adjacent to faults range between ca 1.5° and 8.5° in both footwalls and hanging walls. Zones of high horizon dips adjacent to faults with throws of ca 20-50 m are typically 50-300 m wide, i.e. about an order of magnitude less than reverse drag radii on tectonic faults of similar
Fig. 5. Fault polygons on three horizons (labelled) and upper and lower tip-lines (broken line) mapped onto a common horizontal plane for fault surfaces which are (a) cylindrical and (b) weakly conical.
throw (Gibson et al 1989). The fault-related bed rotations are therefore more restricted laterally and the structures correspondingly tighter than those associated with typical tectonic normal faults and point to contrasting mechanisms of fault formation.
Displacement contour diagrams Throws on individual faults, although quite small, can be mapped on multiple seismic lines for up to nine horizons, providing detailed throw patterns over entire fault surfaces. Displacement contour diagrams for 20 of the largest faults (e.g. Figs 10 and 11), offsetting both the Coorikiana and Mackunda horizons, generally have maximum throws just below the Coorikiana Horizon. Individual faults may have multiple throw maxima, located at different stratigraphic levels and typically between, rather than adjacent to, branch-lines (Figs 10 and 11). Faults with a single throw maximum, which is not always positioned mid-way between the lateral tiplines, generally have branch-lines with only small splay faults, i.e. <20 m throw. On displacement contour diagrams, throws typically decrease rapidly downwards from points of maximum throw to sub-horizontal lower tip-lines,
THE GEOMETRY, GROWTH AND LINKAGE OF FAULTS
253
Fig. 7. Horizon separation diagrams for six faults, three of which are shown on the Coorikiana horizon map in Figure 6. Filled circles indicate hanging wall and footwall cutoff data. All faults are shown at the same scale and branchpoints are arrowed. 50 ms vertical scale bar = c. 64 m. Fig. 6. (a) Coorikiana horizon map of faults bounding a single polygonal cell showing basinal form of the cell. Note the relatively steep horizon dips adjacent to faults within the cell, (b), (c) and (d) show throw profiles along faults D, E and F with branch-points arrowed. Abrupt changes in throw are associated with footwall intersections only. Horizon separation diagrams for faults D, E & F are shown in Figure 7. Horizon 'depth' contours in ms TWT; 1 ms TWT is c. 1.3 m. See Figure 1 for location.
over vertical distances of ca 60 m-380 m (Figs 10 and 11). Dip-parallel throw gradients between the lower tip-lines and points of maximum throw, where throw contours are generally sub-horizontal, are ca 0.08-0.9 for faults with maximum throws >25 m. Interval velocities in the mudstone sequence are relatively constant and so do not contribute significantly to the calculated steep gradients which are accommodated by sequence thickness differences between the hanging wall and footwall sides of faults of 10%-70%. The lower tip lines of each of the 20 contoured faults coincide with the top of the Cadna-owie horizon and the base of the low-density mobile layer (Watterson et al 2000; Walsh et al 2000). This relationship suggests a strong stratigraphic control on the vertical distribution of faulting and is consistent with gravitational overturn of
the mobile layer providing the principal driving mechanism for faulting and associated folding. Rapid downward decreases in displacement are accommodated by ductile deformation and flowage of the faulted sequence and the mobile layer in particular. Displacement variations on faults cannot, however, be related directly to fault propagation directions. For example, simple models for tectonic faults, in which approximately radial displacement patterns are taken to reflect radial fault propagation (e.g. Gibson et al. 1989), cannot be applied to faulted sequences characterized by marked rheological changes; in such circumstances the final fault displacement variations are not simply related to initial fault propagation directions. The conjugate nature of the Lake Hope faults, where intersections of oppositely dipping graben-bounding faults coincide with the mobile layer, favours a model in which the faults initiated at and propagated upwards from that level (Walsh et al 2000). The coeval and complementary folding and faulting associated with a gravitational overturn model (Watterson et al. 2000), however, does not demand a dominant propagation direction (both upward and downward fault propagation can occur). Dip-parallel throw gradients above the throw maxima (range 0.05-0.15) are, as expected, generally lower than those below
254
A.NICOL£TAL.
Fig. 9. Convex-to-hanging wall fault on the Coorikiana horizon. Horizon 'depth' contours in ms TWT; 1 ms TWT is c. 1.3 m. Contours illustrate horizon deformation adjacent to fault with footwall horizon dips up to 8.5°. See Figure 1 for location.
Fig. 8. Polygons of intersecting faults on the Coorikiana horizon illustrating the high degree of connectedness of the faults. Horizon 'depth' contours in ms TWT; 1ms TWT is c. 1.3m.. See Figure 1 for location.
the maxima. Upper tip-lines are often irregular (Figs 10 and 11) and may be associated with slipperpendicular relay zones (i.e. fault relays seen on cross-sections; see Fig. 3) and high associated throw gradients. In 3-D the fault and splay shown in Figure lla provide indications of the fault growth and the linkage process. The intersecting fault (A', equivalent to Segment 3 in Fig. 4c) intersected the footwall of the original fault (A and B, equivalent to Segments 1 and 2 in Fig. 4c) and merged with fault A, leaving an inactive splay (fault B). The throw contours on the linked fault have two maxima, one on each segment. The maximum throw on fault A is just over 40 ms (1ms - c. 1.3m), about 20 ms greater than on de-activated fault B. Approximately 20 ms throw therefore accumulated on segments A and A'
after intersection and their tip-lines at the time of intersection corresponded approximately with the present 20 ms throw contour (fault A) and 10 ms throw contour (fault A'). The vertical extent of fault A at the time of intersection should correspond with the vertical extent of the branch-line, which approximately, it does. Throw subsequent to merging of faults A and A' will have downthrown fault B relative to the linked fault, along the branch-line. The present vertical dimension of fault B should be the same as at the time of intersection. The post-intersection slip vector or vectors are of particular interest. The junction between faults A and A' formed a branch-line that was represented, at least initially, by an angular corner on the linked fault surface. Kinematically, the easiest direction of slip on the linked fault would be parallel to this corner, which is not parallel to the original slip direction on either fault A or fault A'. Whatever the subsequent slip direction or directions, slip on a surface with an angular corner is likely to require significant additional accommodation strains within the adjacent rock.
Fault linkage model Given the available data, intersections are interpreted as having originated by lateral and vertical
THE GEOMETRY, GROWTH AND LINKAGE OF FAULTS
255
Fig. 10. (a) Fault map on Coorikiana horizon (See Fig.l for location); (b) depth-converted cross-section; and (c) throw contoured strike-projection (maximum throw 48 ms, ca 61 m) with multiple throw maxima due to linking of initially separate faults. Vertical and horizontal scales are approximately equal and the same on each figure and throw contours are in ms TWT (1ms = 1.3 m). Horizontal scale for the strike-projection is only approximate due to strike irregularities. Note that the westernmost footwall fault branch-line pitches to the west at depth, even though the intersecting fault dips east, because of the clockwise strike change as it approaches the main fault.
propagation of the intersecting fault when both it and the intersected fault were still active (Fig. 12). A linked fault is then formed by merging of the intersecting fault with one of the segments of the intersected fault. This capture process gives rise to a curved or angular fault trace on which the strike may vary by up to 90°, providing a characteristic faultmap pattern (Fig. 3). The accuracy with which the form of the curved traces is represented on the fault maps is limited by the spacing of interpreted seismic lines, i.e. 87.5 m, but in most cases there appears to have been a smoothing of what must initially have been angular intersections. The data are consistent with the sequence of events illustrated in Figure 12. Initially: (i) two separate active faults exist; (ii) one fault propagates and intersects the other; (iii) the intersecting fault links with one segment of the intersected fault; (iv) the linked fault continues to grow as a single fault, while the un-merged segment of the intersected fault becomes an inactive, or relatively less active, increasingly vestigial splay. The process is similar to that which produces vestigial inactive splays at breached relays (e.g. Childs et al. 1995; Cartwright et al 1996), which may be regarded as a special case of the general rule proposed here. The sequence of events is most easily identified where high-angle fault intersections (> 60°) are associated with an
abrupt change in strike direction and sometimes throw, of a fault coinciding with a branch-point to a splay (Figs 3 & 13). This sequence is evident only where the intersecting fault lies in the footwall of the intersected fault. While there is a possibility that the intersecting fault post-dated the intersected fault and reactivated one segment of it, there is no difference between the geometric effect of re-activating one segment of an inactive intersected fault and that of de-activating a segment of an active intersected fault. As the faults are not syn-sedimentary, there is no prospect of distinguishing between the two kinematic possibilities (Childs et al 2003). The linkage model described above is supported by displacements observed on each of the fault segments at a branch-point. Where three fault trace segments meet at a branch-point, two have substantially higher displacements adjacent to the branch-point than the third segment and these two higher-displacement segments often have significantly different strike directions from one another (Figs 12 and 13a-f). The third, lowest-displacement, segment usually has a strike direction sub-parallel to that of one of the higher displacement segments. It is this combination of features that requires a decision on what criteria to use in correlating two of the three segments as a single fault and relegating the third segment to the status of a splay.
256
A.NICOLE7VLL
Fig. 11. Example of a high angle (ca > 60°) fault intersection, (a) Fault map on Coorikiana horizon (See Fig. 3 for location); (b) oblique view of faults in 3-D; (c) strike-projection of fault B; and (d) strike-projection of fault A (viewed from east). Vertical and horizontal scales are the same on each figure and are approximately equal and throw contours are in ms TWT (1ms =1.3 m). Horizontal scale for the throw contour diagram is only approximate due to strike irregularities.
At a footwall intersection, a simple rule determines which segment of the intersected fault merges with the intersecting fault and remains active and which segment becomes 'inactive'. The rule is that the segment of the intersected fault, which merges and remains active is that which shares a footwall with the intersecting fault, i.e. Segment 1 in Figure 4a. The linked, or subsequent, fault is convex to its hanging wall. Note that the hanging wall cutoff of segment two is continuous with that of segment one both before intersection and after post-intersection movement on the linked fault consisting of segments one and three. For compatibility to be maintained the aggregate throw on segments two and three adjacent to the branch-line must equal the throw on segment one. The relative size of the intersecting fault, segment three in Figure 4a, appears not to be a factor
in the capture process as it may, at the time of intersection, be smaller than the segment of the intersected fault that it causes to become 'inactive'. Also, the segment of the intersected fault that remains active can, at the time of intersection, be smaller (i.e. shorter) than the segment that becomes 'inactive'. How much smaller the segment that remains active can be, cannot be measured directly because of postintersection growth. Figure 14a shows the relative lengths of the active (segment one) and 'inactive' (segment two) segments of intersected faults; even after post-intersection growth ca 35% of the 'inactive' segments are as long or longer than their active counterparts. The simple capture process accounts for many of the characteristics of merged faults and their splays resulting from footwall intersections. Figure 14b
THE GEOMETRY, GROWTH AND LINKAGE OF FAULTS
257
Fig. 12. Schematic diagram illustrating stages of fault linkage and capture, as exhibited by faults in Fig.l3a-f. Shaded polygon indicates the two segments correlated as a single structure that is referred to as the 'main fault' with throws greater than on the splay.
shows the ratios of throws on the two segments of intersected faults, immediately adjacent to the branch-points. Before intersection, the two values would be identical and the differences are due to post-intersection growth. Low ratios are expected to be associated with intersections that formed relatively early and therefore pre-dated most of the fault growth. High ratios, i.e. those close to 1.0, will be associated with branch-points formed late in the faulting history. A ratio of 0.6, the mean value, would represent an intersection that occurred after 60% of the displacement had occurred if that movement on the splay ceased at the time of intersection. The existence of very low ratios (e.g. 0.3) suggests that some intersections formed relatively early during fault growth and that little or no post-intersection displacement was accommodated on the 'inactive' segments. Figure 15a shows relationships between throw values, adjacent to branch-points, on the three fault segments. At the time of intersection, the throw adjacent to the new branch-point on the intersecting fault (Segment 3, Fig. 4a) is zero. Therefore, adjacent to the branch-point, the sum of the throws on the intersecting fault (Segment 3) and the 'inactive' segment of the intersected fault (Segment 2, Fig. 4a) should equal the throw on the active segment of the intersected fault (Segment 1, Fig. 4a) both at the time of intersection and at all subsequent times. The diagrammatic three-way horizon separation diagram in Figure 4c illustrates these relations for the simple case of vertical faults and their consequent vertical branch-line. The data (Fig. 15a) indicate that while this condition is approximately met in many cases, there is a tendency for the aggregate throw on Segments 2 and 3 to exceed that on Segment 1. This discrepancy between the simple model (Fig. 4c) and some branch-points was investigated further by plotting throw data adjacent to branch-points, with each branch-point individually represented on a ternary diagram (Fig. 15b). At the time of intersection each branch-point would be represented by a point midway along the left side of the triangle (point A, Fig. 15b), i.e. equal throws on Segments 2
Fig. 13. Fault polygons on the Coorikiana horizon with triple junctions indicating fault capture. Fault segments have been correlated on the basis of throw coherency and correlated faults are shown shaded. Unshaded segments became 'dead' splays when the two original faults intersected and linked. More than 50% of displacement of the correlated faults occurred subsequent to linkage. All faults shown at same scale with downthrown (i.e. hanging wall) sides marked by filled rectangles.
and 3 and zero throw on Segment 3. Following intersection, the locus of each branch-point would be a straight line between point A and a point representing the increments of post-intersection relative throws. For a simple model in which post-intersection increments of throw on Segments 1 and three are equal and throw on Segment 2 is zero, the righthand end of the line would be point B (Fig. 15b). The distribution of data suggests that post-intersection throws are better represented by a segment three/segment one ratio of 1.5:1. As displacement at fault intersections must be conserved, throw at the branch-point on Segment 1, should be equal to the sum of throws on Segments 2 and three (e.g. Fig. 4c) and thus the ratio of throws on Segment 3 to Segment 1 is expected to be ^1. The discrepancy between observed and predicted ratios of Segment 3 to Segment 1 is believed to reflect inadequacies of the data due to the sample-line spacing (87.5 m). This spacing does not permit throws always to be measured sufficiently close to branch-points which, where displacement gradients are high (e.g. > 0.15), can result in significant errors. High displacement
258
A.NICOLETAL
Fig. 14. Frequency histograms for throws and trace-lengths on intersecting faults where they offset the Coorikiana horizon. Terms used to describe the faults are summarized in Figs 4 and 12. Data from intersections where segment 3 is in the footwall and hanging wall of the first two segments are differentiated by the terms 'footwalF and 'hanging wall' (NB segment 1 is always that in the footwall of segment 3, i.e. regardless of whether the intersection is a footwall or a hanging wall intersection), (a) trace length ratios segment 2/segment 1, footwall intersections; (b) branch-point throw ratios segment 2/segment 1, footwall intersections; (c) branch-point throw ratios segment 1/segment 2, hanging wall intersections; (d) segment 3 ratios branch-point throw/max throw, footwall intersections; and (e) segment 3 ratios branch-point throw/max throw, hanging wall intersections.
gradients near branch-lines are expected because of the high heterogeneous strains needed to accommodate slip on two linked non-vertical fault segments with divergent slip vectors and a complex branchline geometry. Rapid decreases in throw close to branch-points indicate high volumetric strains that, for these seismic data, are manifest as high horizon dips (Watterson etal 2000a). The post-intersection growth of those segments of linked faults which originated as intersecting faults, is shown in Figures 14d and e by the ratios, for each of these segments, between throw adjacent to the branch-point and the maximum throw. At the time of intersection of the tip-line of the intersecting fault with the surface of the intersected fault, throw on the former adjacent to the branch-point will be negligible relative to the maximum throw on that fault and the ratio will be correspondingly small. After linking with a segment of the intersected fault (Segment 1), the point of maximum incremental slip on the intersecting fault (Segment 3) will move nearer to the branch-point, i.e. towards the centre of the linked fault. At 'mature' intersections, i.e. with relatively
large amounts of post-intersection growth, the ratio of branch-point throw to maximum throw will increase towards a value of 1.0 on both segments of the linked fault. Figures 14d and e show a full range of ratios from values indicative of immature intersections to the more common values indicative of mature intersections. Many faults are the result of two or more intersections (Figs 3 and 8) and in extreme cases a single linked fault forms a closed loop surrounding a footwall block (Fig. 6a). The correlation of fault segments based on size rather than strike direction usually identifies as a single fault the two or more segments that moved as a unit in the later stages of faulting, rather than those that were a single unit previously. In the Lake Hope fault system this correlation criterion is unambiguous where a significant proportion of the displacement is post-intersection, but objective correlation based on the size criterion alone is more difficult where an intersection post-dated most of the displacement. Although hanging wall intersections are relatively uncommon in the Lake Hope fault system, there are
THE GEOMETRY, GROWTH AND LINKAGE OF FAULTS
259
active and the one which shares its footwall with the hanging wall of another segment is destined to become relatively inactive. The difference between the post-intersection evolution of hanging wall and footwall intersections is illustrated by the ratios of throws, adjacent to branch-points, on the two segments of faults intersected in their hanging walls (Fig. 14c). In contrast to footwall intersections (Fig. 14b), there is no indication of one segment of the intersected fault having a consistently higher throw than the other and, therefore, no indication of one segment having remained active longer than the other. Fault linkage in the Lake Hope system is dominated by footwall intersections, a situtation which is not entirely due to the higher proportion of map area that represents footwall regions. This preponderance of footwall abutments, combined with the frequent characteristic of hanging wall faults showing a decrease in displacement, or even terminating, towards another fault, reflect a marked asymetry in the topology of this fault system. It is suggested that this asymmetry reflects the mechanism of formation of the system and specifically, the mobility of the underlying low-density layer (see below).
Discussion
Fig. 15. Plots showing relationships between throws on each fault segment adjacent to branch-points, (a) Throw on segment 1 v.s sum of throws on segments 2 and 3; (b) Ternary diagram showing relative branch-point throws on each of the three segments. For significance of points A and B, see text.
many instances on the fault maps where a potentially intersecting fault in a hanging wall terminates close to the other fault. Given the limited throw resolution of the maps it is likely that such cases do represent intersections, either on the mapped horizon or at some level above or below. The predominance of these immature hanging wall intersections, or near intersections, is consistent with the intersecting faults becoming inactive soon after intersection. In the case of hanging wall intersections therefore, it appears that it is the intersecting fault that becomes the 'inactive' splay, with both segments of the intersected fault remaining active. The empirical rules describing footwall intersections, therefore, also apply to hanging wall intersections, i.e. of the three segments, the two which share a footwall remain
The faulting mechanism giving rise to polygonal, or intra-formational, fault systems is clearly different from those giving rise to either tectonic fault systems or gravitational fault systems. We favour a mechanism where fault displacements are driven by gravitational instability of a low-density, overpressured, layer (Watterson et al 2000; Walsh et al 2000). This mechanism is consistent with the polygonal fault trace patterns, abrupt downward termination of the fault array at the low-density layer and folding of horizons adjacent to faults. The linkage and capture process described is also consistent with the gravitational instability model. Specifically, as the faulting is ultimately driven by lateral and vertical migration of the mobile material in an overpressured layer (Watterson et al 2000; Walsh et al 2000), the mechanism produces motion of low-density material towards fault footwalls and away from hanging walls. In the limit, the hanging walls of large faults may be grounded when the mobile layer beneath them has been completely withdrawn which will result in cessation of movement on faults in the hanging walls. By contrast, increases in the volume of the mobile layer in the footwall of larger faults will enhance the likelihood of movement in these areas. Under these circumstances post-linkage growth of footwall intersections, where fault segments bound a common footwall, would be favoured over hanging wall intersections.
260
A.NICOLErAL.
Conclusions (i)
(ii)
(iii)
(iv) (v)
Faults within the Lake Hope polygonal fault system have geometric characteristics that are consistent with an origin arising from the gravitational instability of a low-density, overpressured, mobile layer. Fault size populations have scale-bound, nonpower-law, properties that reflect the thickness of both the faulted layer and the underlying mobile layer and the strongly connected nature of the system. The spatial distributions of faults reflect the localization of conjugate faults at the top of the mobile layer, on the one hand, and the scale of polygons, on the other. Vertical displacement variations on faults show marked decreases in displacement at or adjacent to the top of the mobile layer. Fault linkage and capture subscribes to a simple model favouring the continued movement on faults which share a mutual footwall and promoting the deactivation of hanging wall faults.
The model is attributed to thickening of the mobile layer within the footwalls of faults and thinning and eventual grounding of the overlying sequence, within their hanging walls. We thank Santos Ltd (Australia) for providing the 3-D seismic data on which this work was based. This work was partly funded by an Enterprise Ireland Basic Research Grant (Contract No. SC/2001/141). Thanks are due to our colleagues in the Fault Analysis Group for assistance at all stages of the work.
References CARTWRIGHT, J.A. 1994. Episodic basin-wide hydrofracturing of overpressured Early Cenozoic mudrock sequences in the North Sea Basin. Marine and Petroleum Geology, 11,587-607. CARTWRIGHT, J.A. & DEWHURST, D.N. 1997. Layer-bound compaction faults in fine-grained sediments. Geological Society of America Bulletin, 110, 1242-1257. CARTWRIGHT, J.A., TRUDGILL, B. & MANSFIELD, C.S. 1995. Fault growth by segment linkage: an explanation for scatter in maximum displacement and trace length data from the Canyonlands Grabens of S.E. Utah. Journal of Structural Geology, 17,1319-1326. CARTWRIGHT, J.A. & LONERGAN, L. 1996. Volumetric contraction during the compaction of mudrocks: a mechanism for the development of regional scale polygonal fault systems. Basin Research, 8,183-193. CARTWRIGHT, J.A. & LONERGAN, L. 1997. Seismic expression of layer-bound fault systems of the Eromanga and North Sea Basins. Exploration Geophysics, 28, 323-331.
CARTWRIGHT, J.A., MANSFIELD, C. & TRUDGILL, B. 1996. The growth of segmented normal faults: evidence from the canyonlands Grabens of S.E. Utah. In: BUCHANAN, P.G. & NIEUWLAND, D.A. (eds) Modern developments in structural interpretation, validation and modelling, Geological Society, London, Special Publications, 99,163-177. CHILDS, C., WALSH, J.J. & WATTERSON, J. 1990. A method for estimation of the density of fault displacements below the limits of seismic resolution in reservoir formations. In: BULLER, A.T., BERG, E., HJELMELAND, O., KLEPPE, J., TORS/ETER, O. & AASEN, J.O. (eds) North Sea Oil and Gas Reservoirs - II. Graham and Trotman, London, 309-318. CHILDS, C., WATTERSON, J. & WALSH, J.J. 1995. Fault overlap zones within developing normal fault systems. Journal of the Geological Society, London, 152,535-549. CHILDS, C., NICOL, A., WALSH, J. & WATTERSON, J. 2003. The growth and propagation of synsedimentary faults. Journal of Structural Geology, 25, 633-648. CHRISTENSEN, K. & OLAMI, Z. 1992. Variation of the Gutenberg-Richter b values and non-trival temporal correlations in a spring-block model for earthquakes. Journal of Geophysical Research, 97, 8729-8735. CLAUSEN, J.A., GABRIELSEN, R.H., REKNES, P.A. & NYSAETHER, E. 1999. Development of intraformational (Oligocene-Miocene) faults in the northern North Sea: influence of remote stresses and doming of Fennoscandia. Journal of Structural Geology, 21, 1457-1475. COWIE, P.A., & SCHOLZ, C.H. 1992. Displacement-length scaling relationship for faults: data synthesis and discussion. Journal of Structural Geology, 14, 1149-1156. DEWHURST, D., CARTWRIGHT, J.A. & LONERGAN, L. 1999. The development of polygonal fault systems by the syneresis of colloidal sediments. Marine and Petroleum Geology, 16, 1-18. GIBSON, J.R., WALSH, J.J. & WATTERSON, J. 1989. Modelling of bed contours and cross-sections adjacent to planar normal faults. Journal of Structural Geology, 11,317-328. GILLESPIE, P.A., WALSH, J.J. & WATTERSON, J. 1992. Limitations of dimension and displacement data from single faults and the consequences for data analysis and interpretation. Journal of Structural Geology, 14, 1157-1172. GILLESPIE, P. A., HOWARD, C.B., WALSH, J.J. & WATTERSON, J. 1993. Measurement and characterization of spatial distributions of fractures. Tectonophysics226,113-141. GILLESPIE, P.A. JOHNSTON, J.D., LORIGA, M.A., MCCAFFREY, K.J.W, WALSH, J.J. & WATTERSON, J. 1999. Influence of layering on vein systematics in line samples. In: MCCAFFREY, K.J.W., LONERGAN, L. & WILKINSON, J.J. (eds) Fractures, Fluid Flow and Mineralization, Geological Society, London, Special Publications, 155,35-56. HEFFER, K.J. & BEVAN T.G. 1990. Scaling relationships in natural fractures - Data, theory and applications. Society of Petroleum Engineers, Reprint no. 20981, 1-12. HENRIET, J.-P, DE BATIST, M. & VERSCHUREN, M. 1991. Early fracturing of Palaeogene clays, southernmost
THE GEOMETRY, GROWTH AND LINKAGE OF FAULTS North Sea: relevance to mechanisms of primary hydrocarbon migration. In: SPENCER, A.M. (ed.) Generation, accumulation and production of Europe's hydrocarbons. European Association of Petroleum Geoscientists 1,217-227. LAYERING, I.H. 1991. Observations on the geological origin of the 'C' horizon seismic reflection, Eromanga Basin. BMR Journal of Australian Geology & Geophysics, 12,5-12. LONERGAN, L., CARTWRIGHT, J., LAYER, R. & STAFFURTH. J. 1998a. Polygonal faulting in the Tertiary of the Central North Sea - implications for reservoir geology. In: COWARD, M.P., JOHNSON, H. & DALTABAN, T.S. (eds) Structural Geology in Reservoir Characterization and Field Development. Geological Society, London, Special Publications, 127,191-207. LONERGAN, L., CARTWRIGHT, J. & JOLLY, R., 1998/7. The geometry of polygonal fault systems in Tertiary mudrocks of the North Sea. Journal of Structural Geology 20,529-548. LONERGAN, L. & CARTWRIGHT, J. 1999. Polygonal faults and their influence on deepwater sandstone reservoir geometries, Alba Field, UK Central North Sea. AAPG Bulletin, 83,410-432. MANZOCCHI, T.H.P. 1997. Quantification of flow impairment in faulted sandstone reservoirs. PhD thesis, Department of Petroleum Engineering, Heriot-Watt University, Edinburgh. MARRETT, R. & ALLMENDINGER, R.W. 1991. Estimates of strain due to brittle faulting: sampling of fault populations. Journal of Structural Geology, 13,735-738. OLDHAM, A.C. & GIBBINS, N.M. 1995. Lake Hope 3D -A case study. Exploration Geophysics, 26, 383-394.
261
OUILLON, G., CASTAING, C. & SORNETTE, D. 1996. Hierarchical geometry of faulting. Journal of Geophysical Research, 101B, 5477-5487. PICKERING, G., PEACOCK, D.C.P., SANDERSON, D.J. & BULL, J.M. 1997. Modelling tip zones to predict the throw and length characteristics of faults. AAPG Bulletin, 81, 82-99. SCHLISCHE, R.W., YOUNG, S.S. & ACKERMAN, R.V. 1996. Geometry and scaling relations of a population of very small rift-related normal faults. Geology, 24, 683-686. VERSCHUREN, M. 1992. An integrated 3-D approach to clay tectonic deformation. Doctoral thesis, University of Gent, 359 p. WALSH, J.J., & WATTERSON, J. 1988. Analysis of the relationship between displacements and dimensions of faults. Journal of Structural Geology, 10, 239-247. WALSH, J.J., WATTERSON, J., NICOL, A. & NELL, P.A.R. 2000. Discussion of the geometry and origin of a polygonal fault system: reply. Journal of the Geological Society London, 157,1261-1264. WATTERSON, J., WALSH, J.J., GILLESPIE, PA. & EASTON, S. 1996. Scaling systematics of fault sizes on a large scale range fault map. Journal of Structural Geology, 18,199-214. WATTERSON, J., WALSH, J.J., NICOL, A., NELL, P.A.R. & BRETAN, P. 2000. Geometry and origin of a polygonal fault system. Journal of the Geological Society London, 157,151-162. YIELDING, G., NEEDHAM, T. & JONES, H. 1996. Sampling of fault populations using sub-seismic data: a review. Journal of Structural Geology, 18,135-146.
This page intentionally left blank
Polygonal faults in the Ormen Lange Field, M0re Basin, offshore Mid Norway LIV M. STUEVOLD1, ROALD B. FAERSETH2, LINN ARNESEN2, JOE CARTWRIGHT3 & NICOLA MOLLER1 l
Norsk Hydro ASA, N-0246 Oslo, Norway Norsk Hydro Research Centre, Sandslivn. 90, N-5020 Bergen, Norway ^Department of Earth Sciences, Cardiff University, PO Box 914, Cardiff CF10 3YE, UK. (e-mail: [email protected]) 2
Abstract: 3D seismic and well data from the Ormen Lange Field, Mid Norway have been used to analyse the development of a system of polygonal faults affecting the Late Cretaceous-early Paleocene reservoir. These faults have the typical properties of polygonal fault systems recognized elsewhere in mainly fine-grained successions. They grew by upward propagation from the thick, shale-prone interval of the Late Cretaceous in the M0re Basin and were reactivated during the deposition of the Balder Formation. They have throws ranging from a few metres to 80 m, are typically 1-3 km in length and have highly irregular throw distributions along strike, mainly as a result of complex fault intersection geometries. The Ormen Lange Field is the first described example of polygonal faults that completely transect a major sandstone reservoir interval. The presence of these faults has important implications for the likely production behaviour of the field. Fault seal analysis shows that they are unlikely to form juxtaposition seals, except locally, but that they may have a significant risk for clay smear seals, particularly in the lower reservoir unit.
Polygonal fault systems have been recognized in many passive continental margin and intracratonic basin settings worldwide in the past decade (Henriet et al 1991; Cartwright & Dewhurst 1998). They have mainly been described from sedimentary successions that are dominantly fine-grained and are typically found in smectite-rich claystones or biogenie mudstones. The interaction of polygonal faults with clastic reservoirs is poorly documented, although their occurrence in deepwater depositional systems means that they could be expected to overlap to some degree with sand-prone depositional units. Lonergan & Cartwright (1999) described the interaction between polygonal faults and a confined deepwater sandstone reservoir of Eocene age in the Alba Field, central North Sea. In this example, the faults defined the edge of the reservoir and exploited the abrupt facies change from slope claystones to the massive sandstone reservoir. This paper describes the geometrical and kinematic characteristics of polygonal fault systems developed in the Ormen Lange Field. This field is the first yet described where a polygonal fault system could be demonstrably shown to have played a significant role in the deformation of the whole of the reservoir interval. This relationship between the polygonal faults and a major hydrocarbon reservoir offers a unique opportunity to examine the ways that polygonal faults might affect the reservoir geology and influence the likely production behaviour. The overall goal of this study was to classify the different types of faults present in the field and sur-
rounding area, and analyse their impact on fluid migration over geological and production time scales. The specific objectives were: (1) (2)
(3) (4)
To define the fault geometry (including throw characteristics) and classify on that basis; to define the relationships between fault systems and reservoir stratigraphy, including units beneath and above the main reservoir interval; to reconstruct the timing of activity of the fault systems; to determine the impact of these fault systems on fluid migration, both across faults (sealing potential) and along faults (transmission behaviour).
Geological setting of the Ormen Lange Field The Ormen Lange Field is located about 100 km off the coast of Norway in the M0re Basin (Fig. 1). This major gas accumulation was discovered by Norsk Hydro in 1997 and subsequently proven by appraisal drilling. The field occupies the southern part of a large, north to south trending domal structure (Fig. 2), one of a series of Cenozoic domes in the Norwegian Sea (Blystad et al 1995; Dore & Lundin 1996). The wavelength of the dome is c. 10 km and the structural relief at the reservoir level is of the order of 450 m. Maximum dips of the flanks of the
ment provided by a structurally-controlled depression or sub-basin. Beneath the reservoir interval is a thick and only partially drilled succession, of probably mainly finegrained sediments. Where calibrated by drilling, the pre-reservoir part of the uppermost Late Cretaceous (Springar Formation) is a smectitic claystone. Above the reservoir is another mainly fine-grained succession, passing upwards from mudstones into biogenic (siliceous) mudstones and ooze, with only minor coarser-grained units developed. The PlioPleistocene interval, is however, notably coarser, reflecting the change to cooler climatic influence on sediment supply. The present day sea bed is marked by highly irregular topography which is mainly the result of the Storrega Slide (Bugge 1983).
3D seismic interpretation Fig. 1. Location map of the Ormen Lange Dome (OL) in the M0re Basin, offshore Norway, showing major basinal structural elements. H-H = Helland Hansen Dome.
dome are 3-4°. The dome is underlain by extensional faults of Late Jurassic age and it is likely that the growth of the dome during the Eocene to Miocene was related to subtle compressional reactivation of these underlying structures. A large thickness of mainly fine-grained Cretaceous sediments separates the extensional foundations from the shallower domal structure and this layer probably played a significant role in the adoption of the particular domal structural style. Compression in the Cenozoic has been attributed to a combination of ridge-push and erogenic (Alpine) stress regimes (Dore & Lundin 1996; Vagnes et al. 1998). However, recent modelling associates the domal structures in the M0re and Voring Basins with differential loading (Kjeldstad et al. in press) The reservoir interval for the Ormen Lange Field comprises turbiditic sandstones of the Springar Formation (Maastrichtian) and the Egga Member of the Vaale Formation (Danian) (Fig. 3), delivered from a single SE source. The period of reservoir deposition thus straddles the Cretaceous/Tertiary (K/T) boundary. The boundary is represented by an abrupt change from grey-green, strongly bioturbated mudstones to black mudstones with almost no bioturbation (the Vaale Tight unit). High net to gross ratios for the reservoir interval have been proven in three wells within the field limits, with the best quality reservoir identified in the Egga Member (Fig. 3). The gross reservoir package varies in thickness from 0 m to 150 m. The reservoir is interpreted as having been deposited in a basin floor fan, with some confine-
Database The seismic database over the Ormen Lange structure consists of both 2D and 3D surveys acquired in the past decade. This study was based mainly on the latest 3D seismic data, a reprocessed and merged volume OL00ML These data were calibrated using synthetic seismograms to wells 6305/5-1 (drilled in 1997), 6305/7-1 (1998), 6305/1-1 (1998) and 6305/8-1 (2000).
Data quality There are several challenges to the interpretation of the seismic data. The data quality at the reservoir level and immediately above varies considerably over the structure. The combination of the irregular seabed with steep seabed scarps and shallow gas anomalies in the Miocene and Palaeocene over the field is a significant problem for geophysical imaging and leads to degradation of the signal to noise ratio and reduction in the frequency content over much of the area of the field. Added to these overburden imaging problems are the geophysical effects related to the gas accumulation itself. Vertical seismic resolution is limited to 20-25 m at the reservoir level, leading to interference problems along the margins of the field. However, faults with throws of c. 10 ms TWT (two way time) were resolved from systematic offsets of stratal reflections. The resolution limit for detecting fault offsets is smaller than the gross vertical stratigraphic resolution because the recognition of offsets is dependent on the sampling frequency and the quality of the migration and not on the dominant wavelength (stratigraphic resolution).
POLYGONAL FAULTS IN THE ORMEN LANGE FIELD, OFFSHORE MID NORWAY
265
Fig. 2. Perspective view from the SW of the Ormen Lange Dome illustrated by the time structure map at the Top Vaale horizon (reservoir level). The field of view is approximately 25 X 65 km and vertical exaggeration is c. 10. Purple tones are deep and red are shallow. The maximum vertical relief in this image is c. 400 m. The faulted nature of this surface is just visible at this scale. The outline of the field is shown in red and the five wells drilled on the structure are shown as black dots.
Interpretation methodology
Regional structural geometry
Previous studies of polygonal faults have demonstrated the utility of closely spaced horizon mapping for accurate definition of fault geometry and kinematics (Cartwright & Lonergan 1996). In addition, extensive use of coherence attribute displays allows the rapid and accurate construction of fault plane images, which are best viewed using 3D visualization tools (Lister 2001). Two approaches to structural mapping were employed in this study; regional mapping of key surfaces and 'local' case studies in areas of excellent data quality. Given the generally poor data quality over the northern part of the field area, the case study areas were selected to maximize the understanding of the polygonal fault systems in a wide geographical spread within and around the field area. This allowed estimation of the lateral variability of the fault systems and enabled extrapolations to be made from areas of good data quality into the areas of the field with poorer data quality.
The regional distribution and expression of polygonal faults are illustrated with two horizon maps, the Top Vaale and the Opal CT marker. The stratigraphic context of the two regionally mapped surfaces is seen on typical inline and crossline profiles across the domal structure (Fig. 4). The Top Vaale is a distinctive seismic event displayed as a minimum trough that gives a good approximation to the top of the reservoir interval (Fig. 4) and offsets on faults at this event are likely to be very close approximations to the true offsets at the top of the reservoir. The fault pattern at Top Vaale is best displayed using the dip attribute (Fig. 5). A clearly defined array of polygonal faults is well developed along the western flank of the field and appears to extend into the field area, particularly in the northern and central parts of the field. Over the northern part of the field area, the data are extremely noisy and the dip attribute map is a poor representation of the fault structure. Some faults are visible
266
L.M. STUEVOLD ETAL
Fig. 3. Stratigraphic chart for the Ormen Lange area within the M0re Basin, showing the reservoir interval straddling the Cretaceous-Tertiary boundary. The mapped horizons are shown as black triangles.
through the noise, but are difficult to trace over their full lateral extent with confidence. The visible polygonal fault pattern on Figure 5 is characterized by sets of linear fault segments with a large range of fault strikes. Some of the longest faults have a strike of east to west to WS W-ENE (labelled with an arrow in Fig. 5) and these are particularly evident in the north and the west-central flank regions. This observation is not an artefact of artificial illumination on the attribute display. Fault intersections are abundant and have mainly orthogonal to highly oblique intersection angles. The areal density of both intersections and fault segments decreases from north to south along the western flank area, with few faults visible in the extreme S W portion of the survey area. It is possible that a 'sub-seismic' fault population complicates this simple qualitative description of fault distribution because the effective limit for fault recognition is a throw value of c. 10 ms TWT. The fault pattern at the Opal CT horizon (Fig. 6) contrasts with that seen at Top Vaale. The pattern is clearly polygonal, but has a higher proportion of curved fault segments than in the Top Vaale map (Fig. 5). Qualitatively, there is no obvious preferred orientation bias evident on Figure 6. Most of the faults fall within a limited length range of 1-2 km. The polygonal fault system is well resolved over most of the field, and the only regions of poor quality
in the dip attribute are on the eastern flank, beneath the major scarps developed at the seabed and over parts of the crest of the structure. The Stratigraphic relationships between the faults developed at these two horizons are apparent on vertical seismic profiles through the survey area (Fig. 4). These profiles show that the two mapped horizons sample the fault geometry of two separate 'tiers' of polygonal faults. Tiers in this context are stratigraphically bound units characterized by a particular assemblage of polygonal faults (Cartwright 1994). Multi-tiered polygonal fault systems have been recognized in the North Sea Basin, where they have been related to the lithostratigraphy of the tiers and where tier boundaries generally correspond to regionally important Stratigraphic boundaries (Cartwright 1994; Dewhurst et al 1999). The Top Vaale horizon is transected by faults developed within the Late Cretaceous and Paleocene succession (Tier 1), whereas the Opal CT horizon is transected by faults developed within the Eocene to Miocene interval (Tier 2). This upper tier of faults can be clearly seen die out downwards at or around the level of the Top Balder marker, corresponding approximately to the base of the Eocene (Fig. 4b). The faults of the upper tier are gently to strongly listric in cross-sectional shape, whereas the faults of the deeper tier are mostly planar or gently listric on the time section over the depth range at which they are resolved. Their lower limit is not visible so it is possible that they change shape at depth. The following description focuses on the lower tier of faults, and the upper tier is not considered further in any detail.
3D Geometry of tier 1 In the context of the Ormen Lange Field, the most important group of polygonal faults is developed within tier 1, since these faults deform the reservoir units. The regional map at the Top Vaale level (Fig. 5) shows the geometrical characteristics of the polygonal faults on the scale of the field. Additional insights were obtained by detailed fault mapping of numerous horizons bracketing the reservoir interval in several sub-areas within the 3D survey area. The main observations from this detailed mapping are summarized below.
Fault patterns Many of the faults within tier 1 extend vertically for c. 1000 m from within the Late Cretaceous upwards to the level of the Top Balder (Fig. 4). The expression of the faults within the Late Cretaceous is poorly constrained in general because of the sparsity
POLYGONAL FAULTS IN THE ORMEN LANGE FIELD, OFFSHORE MID NORWAY
Fig. 4. Representative crossline (A) and inline (B) from the OLOOM1 survey. Three horizons are indicated for reference, Opal CT horizon (within tier 2), Top Balder (approximate boundary between tiers 1 and 2) and Top Vaale (reservoir level). Fault offsets of 20-40 milliseconds are clearly visible at the Top Vaale level, within tier 1. Note the degradation of the seismic response over the crest of the dome on the cross-line. Representative faults are shown on both profiles, but note the high vertical exaggeration (200 ms is approximately 200 m over this field of view). Their deeper continuation is unknown because of lack of stratal reflections deeper within the Late Cretaceous of tier 1, although it is believed that the tier 1 faults propagated upwards from below the reservoir level. Located on Fig. 6.
267
268
L.M.STUEVOLDCTAL.
Fig. 5. Dip attribute display of the Top Vaale over the entire field area. The outline of the field is shown, along with the positions of the five wells testing the Ormen Lange Dome. Note the variable response and quality of imaging of polygonal faults from the western flank onto the crest of the structure. Examples of some of the longest faults with east to west strikes are shown with black arrows. This image shows a general decrease in the fault density from north to south along the western flank. This may be related to grain size variation within the non-reservoir components of the Late Cretaceous succession.
POLYGONAL FAULTS IN THE ORMEN LANGE FIELD, OFFSHORE MID NORWAY
269
Fig. 6. Dip attribute display of the Opal CT horizon, showing polygonal faulting over much of the survey area. Map quality is poor over the crest due to poor signal to noise ratio of the seismic data. Inset map is a location map for Figs 4, 7,8, 11 and 12.
270
L.M. STUEVOLD ETAL
Fig. 7. Local time structure maps of pre-reservoir (Late Cretaceous) reflections from the western flank of the Ormen Lange Field, showing polygonal fault patterns to be well developed at this deepest mapped level, but with a north to south decrease in average throw values (e.g. 40-50 ms in (A) compared to 10-20 ms in (C). The majority of faults throw down to the north and west quadrants. Contour ranges in milliseconds are 3100-3260 (A), 3540-3750 (B), and 3360-3500 (C). Located on Figure 6.
of correlatable reflections within this opaque or low amplitude seismic interval. The lower portions of the faults therefore cannot be interpreted in this low amplitude seismic facies and their basal relationship with the stratigraphy cannot be defined. Mappable reflections are present locally and allow the basic fault geometry to be defined. These show that the faults are arranged in a rectilinear polygonal pattern
along the entire western flank region of the Ormen Lange Field. Examples of local maps are shown in Figure 7. These maps demonstrate that the plan form geometry remains fairly uniform over a distance of 50-60 km, but that there is a reduction in maximum throw values on individual faults in the array from values of 40-80 m in the north to 10-30 m in the south. By analogy with other polygonal fault systems where bulk strain is observed to correlate with grain size (Dewhurst et al 1999), it is suggested that this southerly reduction in throw and in strain generally, is the result of a change in bulk grain size of the pre-reservoir component of the Late Cretaceous succession. This proximal to distal transition in a northward direction is constrained on a basin scale from a regional well database and palaeogeographic reconstructions of the Late Cretaceous. There is some uncertainty in this interpretation because of the poor constraint on the stratigraphic position of the base of tier 1. The fault patterns are generally well defined at or close to the reservoir interval. Detailed mapping on the western flank and in a small area of the eastern flank shows the faults at reservoir level to have a classical polygonal pattern (Fig. 5). In areas of deeper
POLYGONAL FAULTS IN THE ORMEN LANGE FIELD, OFFSHORE MID NORWAY
271
Fig. 8. Local time structure maps for Top Balder horizon on the western, (A), and eastern, (B), flanks of the Ormen Lange Field. Contour ranges in milliseconds are 2675- 3325, (A), and 2700-3000, (B). Located on Figure 6. Note that the pattern on the eastern flank is much more polygonal than that on the western flank. The strong bias in orientation in faults on the western flank to east to west trends is considered most likely the result of early doming on the north to south axis. Polarization of polygonal fault trends is well known elsewhere to be due to local tectonic stresses.
map control, the faults mapped at reservoir level are observed to be upward extensions of faults mapped at the deeper, intra-Cretaceous level (Fig. 4b). There is some lateral variation in the details of the fault pattern at reservoir level on the scale of the 3D survey area (Fig. 5). In general, the fault pattern is characterized by a highly interconnected network, with a high density of fault intersections that are orthogonal to highly oblique (Fig. 5). As noted above, the most striking feature is the bias towards the longest faults having WSW-ENE or west to east strikes. These longer faults, however, are highly segmented by intersections and relay structures. The fault patterns mapped at shallower levels within tier 1 contrast markedly with those mapped at the reservoir level. Some examples of local maps at or close to the Top Balder horizon are shown in Figure 8. The maps from the western flank exhibit patterns with an almost unidirectional strike, with a dominant east to west trend. In contrast, the Top Balder map from the eastern flank shows a more typical polygonal pattern, with orthogonal or highly oblique intersections and several well developed strike directions. This eastern flank pattern is typical of the pattern found regionally at this level outwith
the limits of the Ormen Lange Dome. The strong bias towards the east to west trend seen on the crest and western flank of Ormen Lange dome is difficult to explain. One possibility is that it is a response to local stress conditions during the early growth of the anticline, possibly combined with the gravitational effects of the regional slope. Another possibility is that smaller faults with other trends are also present but are below the seismic resolution. Similar local modifications of an otherwise random polygonal pattern are well known from polygonal fault systems in the North Sea where steep local slopes, inversion folds or salt domes are all observed to distort the regional polygonal pattern (Cartwright 1994). The density of faults decreases progressively upwards through the Palaeocene and is generally a minimum at the Top Balder level. Similar reductions in fault density and total strain have been observed towards the upper limits of polygonal fault tiers in the North Sea Basin (Dewhurst et al 1999). This reduction in density upwards is matched on some faults by reduction in throw towards the upper tips, but an important group of faults exhibits an increase of throw towards the upper limit of the tier (see below).
272
L.M. STUEVOLD ETAL.
are lower, in the range 10-20 m. On the eastern flank, where measured, the maximum throw values Faults within tier 1 are planar to gently listric, in con- at reservoir level are in the range 20-40 m, but these trast to those within tier 2, which are often strongly cannot be related to any deeper mapped values. The listric near the base of the tier (Fig. 4b). On the strongest variation is thus observed in the northwestern flank and crestal areas, most faults in tier 1 south direction i.e. along the fold axis. Because of dip preferentially to the northern or western quad- the correlation with the lateral variation in the throw rants i.e. in a downslope direction. Tier 2 faults in the observed at the deepest mapped level beneath the same area dip preferentially to the eastern and south- reservoir, it is thought probable that the variation in ern quadrants, in an upslope direction (Fig. 4). Fault the maximum throw values at reservoir level is due plane dips in tier 1 range from 25-50°. The generally to lithological variations in the pre-reservoir Late shallow dips of fault planes observed in the study Cretaceous interval. Some influence on throw patarea are unlikely to reflect the original fault plane terns could be related to the thickness of the reserdips during initial propagation. Polygonal faults in voir interval itself. However, since the full extent and the Gjallar Ridge, farther north in the Voring Basin, thickness of the reservoir facies are not known and are imaged on 3D seismic data within 500 m of the there are only limited calibrations of net-to-gross seabed, where they have measured dips in the range within the reservoir interval, this possibility cannot 50-57° (Cartwright et al 2003). Although these be rigorously evaluated at present. Analysis of the vertical throw distribution allows faults are not directly analogous to those at reservoir level in the Ormen Lange dome because of differ- reconstruction of the growth history of the tier 1 ences in lithology of the deformed intervals, they polygonal faults. Two distinct types of vertical throw provide an approximate guide to likely dips of fault distribution are recognized (Fig. 9). The first type planes during the initial growth stage of polygonal (Fig. 9a) is characterized by a maximum throw at the fault development. It is therefore considered most deepest mapped horizon (intra-Late Cretaceous), likely that the observed fault plane dips have been with an approximately linear decrease upwards flattened by compaction after the faults ceased to be along the fault surface to the upper tip close to the active, which is certainly consistent with the loss of Top Balder horizon (e.g. fault X on Fig. 4b). This porosity indicated by density logs. Similar post-def- throw distribution is consistent with a growth model ormational fault plane flattening by compaction has involving propagation upwards from an unspecified been invoked to explain polygonal fault geometries location at depth within the pre-reservoir section in clay quarries in Belgium (Verschuren 1992). This (beneath the deepest mapped horizon). The second type of throw distribution (Fig. 9b) is important rotational deformation has implications characterized by two separate maxima (at near Top for fault sealing potential (see below). Balder level and at the deepest mapped horizon) and a minimum throw midway between the Top Balder and Top Vaale coinciding with the Top Lista horizon Fault throw (e.g. fault Y on Fig. 4b). This type of throw distribuFault throws were measured for two main reasons: to tion has not previously been documented in other better define the growth history of the faults and for polygonal fault systems. The preferred interpretafault seal prediction. The three-dimensional distri- tion is that there were two discrete episodes of fault bution of throw within polygonal fault systems is generation, propagation and growth during the dehighly complex (e.g. Lonergan et al 1998). For the velopment of the tier 1. This was followed by partial purpose of this study, the main focus was on predict- reactivation of the early formed faults during the ing the likely throw magnitudes on faults in the later stage of growth (during deposition of the crestal area that were too poorly imaged to allow for Balder Formation), with displacement transfer propdirect measurement and to consider the throw distri- agating downwards from the locus of maximum butions at reservoir level that might impact fault seal throw close to the upper boundary of the tier (Fig. potential (Fisher & Knipe 1998). 10). This model is based on the assumption that Maximum throw values measured at the top reser- downwards propagation would have been associated voir level for individual faults exhibit considerable with a finite displacement gradient appropriate to the lateral variation throughout the survey area. The lithologies of the deforming units (Walsh & largest values of 70-80 m are recorded in the NW Watterson 1988). and there is a progressive reduction southwards For both types of fault, initial propagation through along the western flank that mimics the reduction in the reservoir interval can be shown to have occurred throw observed at deeper, pre-reservoir levels (Fig. whilst the reservoir was being deposited. Isopach 7). In the region of the western flank adjacent to the maps of the reservoir interval show thickness 6305/5-1 well, maximum throw values are in the increases in hanging walls along many fault traces. range 30-50 m, but in the extreme south, the values These are particularly well seen for the larger, east to Fault plane geometry
POLYGONAL FAULTS IN THE ORMEN LANGE FIELD, OFFSHORE MID NORWAY
Fig. 9. Schematic representation of the two main types of vertical throw distribution (T= throw in metres, D= depth in metres below Top Balder datum) on faults affecting the reservoir interval along the western flank of the field. In (A), the throw values, (T), decrease from the reservoir interval upwards to a tip close to the Top Balder. In (B), there are two throw maxima, separated by a minimum positioned between the top reservoir and the Top Balder. The upper tips of many of the faults appear to be truncated by erosion or by detachment of tier 2 faults from above the Top Balder.
west trending faults. An example of a detailed isopach map from the western flank is presented in Figure 11 (a), along with a seismic example of divergent thickening into a fault plane indicative of syn-reservoir growth (Fig. 1 Ib). The interpretation that some of the polygonal faults were active during reservoir deposition is important for two reasons: (1) it shows that the faulting occurred whilst the sediments were relatively unlithified; and (2) it suggests that reservoir thickness and possibly quality might be impacted by the activity of the polygonal fault system. The lateral variation of throw along individual faults in the array was examined with specific emphasis on the likely impact for fault seal potential. Detailed mapping at the reservoir level both on the flanks and locally over the crest of the field shows that few if any closed polygons occur i.e. fault blocks that are entirely surrounded by completely linked faults. Instead, it is more generally the case that three or four faults interconnect via relay structures or oblique to orthogonal abutting intersections. Lateral throw distribution on such a complex interconnecting array of faults would be expected to exhibit anomalous throw patterns and departures from the idealised throw distribution normally encountered with isolated faults (Nicol et al. 1996; Lonergan
273
Fig. 10. Schematic model of fault propagation involving: (1) nucleation and early propagation with a first displacement maximum (Di); (2) subsequent nucleation in a newly deposited layer with its position influenced by the underlying fault tip, perhaps through localized tip folding; (3) propagation of the upper fault as new layers are added to develop a separate displacement maximum (Dmax2) and more complete linkage with the underlying fault.
observed adjacent to relay structures and abutting intersections, whereas away from these linkage structures, the throw varies more gradually. For tectonic normal faults, throw maxima are commonly observed close to midpoints on fault segments, suggesting that they nucleated close to those positions and propagated radially from this nucleus (Dawers & Anders 1995; Cartwright et al 1996). In contrast, in this system of polygonal faults, the abutting intersections mapped throughout the survey area are invariably associated with throw minima on the main fault close to the position where the secondary fault intersects the main fault. This would be expected in the case of sequential growth of the 'main' fault followed by the intersecting fault because the footwall of part of the main fault corresponds to the hanging wall of the intersecting fault and this changes the footwall elevation on the main fault (E. Tervoort pers. comm. 1999). An example of the complex variation of throw at fault intersections is presented in Figure 12, for a large WSW-ENE trending fault from the western flank region. This fault is over 4 km long and consists of a series of segments with mid-segment throw
274
L.M. STUEVOLD ETAL.
Fig. 11. Time isopach between the Top Vaale and Near Top Cretaceous (this approximately spans the reservoir interval) from the western flank of the field (A). Contour interval is milliseconds two way travel time vertical time difference (layer subtraction between the two mapped surfaces). Growth is observed on several of the faults from stratigraphic expansion into hanging walls, as seen for example in the seismic profile across an east to west trending fault (B). Reservoir interval is located between two arrows. Located on Figure 6.
maxima (peak values similar for three segments at c. 60 m) and minima at intersections with N-S faults. The throw values at the minima (10-15 m) drop to < 50% of the peak values observed in the middle of fault segments. Throw distributions both within segments and close to intersections are asymmetric and lateral gradients of throw on one side of each intersection are relatively high. High lateral throw or displacement gradients are typical of relay structures (Peacock & Sanderson 1991) and lateral fault tips (Cartwright & Mansfield 1998) and usually involve a number of compensatory strain mechanisms including intense fracturing. It is considered probable that the abutting fault intersections would be characterized by intense localized deformation around the branch line and that this would have a significant impact on fault seal behaviour. The complex patterns of throw variation on the example in Figure 12 emphasize the need for the highest spatial resolution of throw measurements when attempting fault seal analysis.
Fault seal analysis Fault seals can arise from juxtaposition of a reservoir with a sealing lithology (juxtaposition seal) or by
development of a fault rock with a high entry pressure for hydrocarbons (membrane seal) (Watts 1987). Membrane seals can be subdivided into six types: clay smears, phyllosilicate framework fault rocks, cataclastic fault rocks, disaggregation zones, cementation and microcrystalline quartz framework rocks (Fisher & Knipe 1998). Detailed study of cores through the reservoir interval and analysis of formation micro imaging logs (FMI) were undertaken to complement the seismic structural analysis to provide constraints on the risk of fault seal throughout the field. Two small normal faults were intersected by cores in the reservoir interval in well
Fig. 12 right. Displacement (squares) and throw values (diamonds) for a fault from the western flank region of the Ormen Lange structure. These values were measured from depth sections constructed orthogonal to strike. Circles on the dip attribute map show data positions and these correspond to data points on the plot as a function of strike length. The plot shows a highly irregular pattern of throw and displacement, with significant minima located at abutting intersections between the main fault (A) and smaller, north to south trending intersecting faults. This type of throw distribution is typical for all fault intersections in the area.
POLYGONAL FAULTS IN THE ORMEN LANGE FIELD, OFFSHORE MID NORWAY
275
276
L.M. STUEVOLD ETAL
6305/5-1 and subjected to microstructural analysis. Thin zones of clay smear were observed in both cases and together with the absence of any signs of grain crushing, suggests that slip occurred whilst the host reservoir sediments and interbedded clays were unlithified. This interpretation of the core data thus supports the seismic arguments that the polygonal faults were expressed at the seabed during reservoir deposition, and continued to be active into the Paleocene, whilst the reservoir was buried to perhaps 3-500 m. The interpretation of an early deformational history for the polygonal faults in the Ormen Lange area makes it probable that membrane mechanisms such as cataclasis, cementation and microcrystalline quartz framework development are of much less importance than clay smear and the development of phyllosilicate framework rocks. Cementation seals cannot be discounted, but are difficult to predict and core data do not indicate a significant occurrence of cemented fracture fills in the reservoir interval. In addition, the reservoir lithologies encountered in the hydrocarbon-bearing wells are mainly relatively clean sandstones and provided these facies are developed more widely outside the area of well control, this lessens the risk even of phyllosilicate framework rocks to occur in significant quantities. The fault seal analysis and prediction for Ormen Lange Field has therefore been focused on juxtaposition and clay smear sealing potential. To assess the risk of both phyllosilicate framework and clay smear fault rocks at reservoir level, the shale gouge ratio, SGR (Yielding et al. 1997) was calculated for the three key wells. This required accurate measurement of the clay content throughout the reservoir section. This was achieved by well log interpretation (Gamma Ray, Density/Neutron logs) calibrated with mud logs, sidewall cores and plugs sampled from long cores. An example of one of these plots of SGR is shown in Figure 13. This shows the clay fraction with depth for the reservoir interval and the SGR as a function of varying throw. This plot shows that SGR values in excess of 20% (assumed to be a threshold for static sealing) are restricted to the lower reservoir unit (Cretaceous Sandstone Reservoir Unit) and for small throw values (< 10 m). The main Egga reservoir unit is characterized by very low values of the SGR (< 10%) and faults with throws up to the juxtaposition threshold are unlikely to seal, or even significantly impair communication. Similar conclusions can be drawn for the other two wells, although the throw value for the threshold of juxtaposition varies from 35-75 m, as the thickness and facies of the upper reservoir unit vary laterally. Clay smear seal potentials can also be evaluated using the Shale Smear Factor (SSF), which is the ratio of fault throw to clay layer thickness (Lindsay et al. 1993). Continuous clay smear has been predicted
to occur for SSF values of less than seven (Lindsay et al. 1993), although other unpublished studies consider that five is a better threshold value. These estimates of the threshold value for continuity of the smear are based on thickness measured at in situ burial depths, which are not necessarily the depths when smear took place. Decompacted clay layer thicknesses would perhaps be more representative of the deformational process. Using in situ clay layer thicknesses of 1-4 m observed in the three key wells of the most likely candidate for clay smearing, the Vaale Tight unit (a claystone developed at the base of the main Egga reservoir unit) fault throws of 5-20 m could produce continuous clay smear to seal the Cretaceous/lowermost Tertiary reservoir beneath the Vaale Tight unit if the threshold value of five is used, or throws of 7-28 m if the value of seven is used. If decompacted thicknesses were used, these throw ranges would increase to approximately 10-50 m assuming that in situ thicknesses are approximately half their original thickness. Note that calculations of SGR and SSF were based on constant thickness of correlative units across faults. Seismic evidence suggests that some faults were growing during deposition of the reservoir interval (Fig. 11), but in the absence of direct well control of stratal expansion from hanging wall to footwall, the simplest approach of constant thickness was adopted. It is apparent from the SGR and SSF approaches that considerable uncertainty exists in the throw ranges for which effective seal might be predicted on the polygonal faults, but that even faults with throws of perhaps as much as 40-50 m could represent barriers to communication for the Cretaceous reservoir unit. Most of the faults mapped on the western and eastern flank areas closest to the area of well control have throw values within this range and could thus be predicted to have a high probability of continuous clay smear. Several additional factors should be noted, however, when considering the potential for continuous clay smear in the Ormen Lange Field. Later compactional deformation. It was inferred that the polygonal faults experienced an important phase of post-deformational rotational flattening by vertical compaction. Fault planes could have rotated by perhaps 20° or more and this might have led to disruption of original clay smears either by additional shearing of the fault planes during rotation or by folding of the fault planes (a requirement of compactional flattening). Excellent examples of folded fault planes in potential field analogues are described by Verschuren(1992). Deformation at intersections. The precise geometry of the branch lines for all the many fault intersections is poorly understood. Limits to the spatial resolution
POLYGONAL FAULTS IN THE ORMEN LANGE FIELD, OFFSHORE MID NORWAY
277
Fig. 13. A plot of Shale Gouge Ratio for well 6305/5-1, showing depth values in the reservoir interval plotted against values of fault throw. The main reservoir subdivisions are shown on the vertical axis. SGR values are much higher in the lower reservoir units, indicating an increased risk of seal. Lithological parameters were computed from petrophysical data calibrated with physical measurements of cores. This plot is based on in situ values and no account is made of decompaction or of possible variations of net-to-gross across faults with growth in the reservoir interval.
of the seismic mean that such complex structures cannot be easily imaged. Despite this uncertainty and by analogy with relay structures, the recognition of extreme lateral throw gradients at intersections encourages the prediction that branch lines will be zones of intense localized fracturing and wall rock folding. These complex zones of deformation might thus be favoured sites for clay smear disruption and hence cross-fault communication. It should also be noted that the deformation around branch lines might be exacerbated by the complex displacement transfer involved in the second phase of growth of those east to west trending faults that have been shown to have a separate throw maximum at or close to the level of the Top Balder. This second phase of motion would have occurred when the reservoir was buried to >400 m and any original clay smears would thus be more highly lithified than when first sheared into their original configuration. The opportunity could have existed, therefore, for a significant disruption of the clay smear fabric during this reactivation, possibly in a more brittle mode. Sandstone injection along fault planes. Small sandstone sills a few centimetres in aperture have been
identified in cores from the three appraisal wells in the field closely below the Vaale Tight interval. This important intra-reservoir claystone is a very distinct black unit in contrast to the generally green claystones above and below within the reservoir interval. High-resolution biostratigraphy shows that this black mudstone accumulated much more rapidly than the green mudstones above and below. The presence of sandstone dykes in all three cores close to the same marker horizon implies that the injection was linked to the physical properties of the black claystones as an intra-formational seal. Since the Vaale Tight unit is one of the main candidate units for clay smear, the association of sandstone dykes at this level raises the probability that early-formed sedimentary intrusions exploited faults and fractures at this level and that sandstones could have been injected along fault planes. Even small aperture dykes (<10 cm) can have a major impact on vertical connectivity in heterogeneous reservoirs (Dixon et al 1995; Molyneux 2001), so it is possible that if such dykes occur along fault planes, they could negate the effects of even well developed clay smears. In contrast to the prediction of clay smears, the analysis of the risk of juxtaposition seal is much
278
L.M. STUEVOLD ETAL
more straightforward and builds directly from the geometry and throw distributions mappable from the seismic data. To have a juxtaposition seal for the upper reservoir unit (Egga) in the Ormen Lange Field, it would be necessary to have throws greater than the reservoir thickness over the full width of the field. No such structure-transecting faults have been mapped, and the risk of significant juxtaposition sealing for the Egga is considered very small. Faults with throws of >60 m are uncommon and mainly found in the extreme north of the study area and even in these cases the strike length over which such high values of throw occurs is only 1-2 km at maximum. The widespread occurrence of intersections and their associated throw minima means that large throws are not laterally persistent. In summary, the polygonal faults that are so widely developed through the reservoir interval are likely to have little to no impact on the fluid communication within the main (Egga) reservoir unit, but the prediction of clay smear in particular is much less certain for the lower reservoir units. The rarity of completely isolated fault blocks completely surrounded by fully-linked polygonal faults means that continuous flow paths exist throughout the field at any given intra-reservoir reservoir level, but these may in some areas be highly tortuous.
Discussion The polygonal faults developed in the Ormen Lange area have been shown to have considerable influence on the reservoir geology. Previous accounts of the interactions between polygonal faults and reservoirs have been limited to relatively superficial effects mainly concerning the top seal to reservoir relationship (e.g. Lonergan & Cartwright 1999). In the case of the Ormen Lange Field, polygonal faults that nucleated in the underlying mudstone-dominated sequences of the Late Cretaceous grew upwards and completely transected the reservoir interval over a large proportion of the area of the hydrocarbon accumulation. These faults were active during reservoir deposition and this early deformation has a significant influence on the development of fault rocks and fault seal potential. Later reactivation of these faults and post-faulting compaction also has a bearing on fault seal potential. One of the most significant aspects of this study is the recognition of the profound impact of highly oblique to orthogonal intersection geometries on the lateral distribution of throw along polygonal faults. This particular characteristic of polygonal faults has only received limited previous attention (E. Tervoort pers. comm. 1999). The highly irregular throw patterns are reminiscent of those found on highly segmented tectonic normal faults (e.g. Dawers &
Anders et al 1995; Cartwright et al 1996). Unlike simple co-linear relay systems, however, the unique geometry of polygonal fault systems results in far more complex branch line topologies and it is likely that this complexity is reflected in the sealing potential of the intersection zones. Much further work is required on these complex intersections before we can adequately predict their impact on fluid transmissibility. This study has not addressed the origin of the polygonal fault system since it was primarily focused on the impact of the faults on the reservoir geology. However, it is worth noting that the polygonal fault systems in the Ormen Lange area provide excellent examples of the organization of these systems into tiers, defined by the stratigraphy and thus strengthen the general conceptual link between polygonal faults and the physical properties of the layers in which they are bounded (Cartwright 1994; Dewhurst era/. 1999). This study area also exemplifies the different controls on the lateral expression of polygonal faults within any single tier. The dominant control on tier 1 is thought to be the grain size distribution related to the basin margin and slope physiography during the Late Cretaceous, based on analogy with other polygonal fault systems (Dewhurst et al 1999). The role of reservoir thickness as a factor is uncertain and requires further investigation beyond the scope of the present study. Based on the limited observations presented in this paper, it could nonetheless be suggested, for example, that the tier 1 faults have smaller throws to the south of the field because the reservoir thickness increases in this direction. This could be taken further to suggest a direct link between the loading effect of the reservoir and the process of upward propagation of the polygonal faults. There is, however, no evidence from studies of growth faults in general to suggest that upwards propagation should be inhibited by the change from mud-dominated to sand-dominated intervals: many growth faults transect heterolithic facies successions with multi-layers of sandstone and mudstone (e.g. Edwards 1995). Whilst growth faults might not be precise mechanical analogues for polygonal faults, they at least share the common property of being extensional faults active at the contemporaneous sediment-water interface. An alternative explanation for the southwards decrease in throw magnitude derives from previous studies of polygonal faults where it is argued that the throw distribution and bulk strain reflects the bulk physical properties of the fine-grained components of the tier interval (Dewhurst et al. 1999). In this alternative view, the reservoir section (50-70 m thick) would be considered to have been passively deformed as part of the general upward propagation of faults whose vertical extents could exceed 1000 m in total. Any variations
POLYGONAL FAULTS IN THE ORMEN LANGE FIELD, OFFSHORE MID NORWAY
at reservoir level would then be considered to be directly related to the gross variation in strain over the whole of this much thicker interval (c. 1000 m) and not just the thickness of the reservoir interval. This view is consistent with the observation that the southernmost faults on the western flank tip out upwards well below the base of the reservoir. Since the polygonal faults in tier 1 propagated to the seabed in many cases to produce growth sequences in the hanging walls (Fig. 1 Ib), this implies that this southern group of faults ceased to be active well before the reservoir was deposited and their behaviour and throw characteristics could not have been influenced by the loading effects of the reservoir units. One of the most enigmatic aspects of the tier 1 faults is the bimodal throw versus depth distribution of the subset of the fault population that was active during deposition of the Balder Formation (Fig. 9b). This subset of faults has a dominant east to west strike orientation where they are most clearly mapped on the western flank of the structure (Fig. 8a), but this highly skewed pattern is not observed in the overlying tier 2 faults, which grew during the most active stages of folding (Fig. 6). It is possible that early growth of the Ormen Lange dome, in combination with the prevailing northward dipping basin slope polarized the local horizontal stresses such that only faults with an approximately east to west orientation were selected for reactivation. Are these faults truly part of the polygonal fault system? Were we to strictly answer this question by considering a small part of the survey area, it is unlikely that the strongly unidirectional pattern expressed at the Top Balder horizon (e.g. Fig. 8) would be equated with the polygonal fault mechanism. However, regional mapping shows that a well-defined polygonal pattern is found surrounding Ormen Lange and is visible on the eastern flank of the dome (Fig. 8b). On balance, therefore, it is perhaps better to consider the unidirectional pattern as a severely skewed, local distortion of a polygonal fault system. Similar distortions of polygonal fault networks have been recognized elsewhere, in conjunction with oversteepened slopes (Cartwright 1994), growth of salt and mud diapirs (Rundberg 1989; Lister 2001), growth faults (D. Hansen pers. comm. 2000) and differential compaction anticlines or monoclines (Clausen & Korstgaard 1993) and even impact craters (Gorter et al 1989) or pockmark fields (Cole et al 2000). In all these cases, a local tectonic stress field has modified the stress field governing polygonal fault growth to modify the resultant fault orientations. To the authors' knowledge, this is the first case study of a major hydrocarbon accumulation in which polygonal faults play a significant role in the fundamentals of the reservoir geology. It is likely that many more hydrocarbon reservoirs are deformed in
279
a similar way and that as exploration continues in frontier deepwater provinces for turbiditic sandstone plays, additional examples of polygonally-faulted reservoirs will be documented and continue to shed light on these extraordinary and widely developed structures.
Conclusions (1)
(2) (3) (4) (5) (6)
(7)
(8)
(9)
The Ormen Lange Field is deformed by polygonal faults that propagated upwards into the reservoir from the Late Cretaceous shale-dominated sequence. These faults were locally reactivated during deposition of the Balder Formation. Fault throws vary irregularly along all fault planes ranging from a few metres to over 75 m. The largest faults are in the north/NW of the area and trend approximately east to west. Fault planes dip at low angles, ranging from 25-40°. Substantial post-faulting rotational flattening has probably occurred. Some tier 1 faults (the larger faults, mainly east to west trending) were active during reservoir deposition as small growth faults. The patterns of polygonal faulting vary throughout the area on a scale of tens of kilometres. This is possibly due to lithological variation in the shale sequences underlying the reservoir. There are significant variations both in a north-south and east-west direction across the main structure. Displacement and throw distributions on all significant faults are highly segmented and irregular, with rapid lateral variations. The most important control on these variations is the intersection geometry in the polygonal network. Intersections are likely to be sites for preferential fluid communication and the intense damage zones in and around the branch line particularly should disrupt clay smears. The polygonal network is not fully connected: there are few completely closed polygons. Communication routes, although sometimes tortuous, exist throughout the main production area. Fully sealed, isolated fault blocks are not thought to be present.
The larger east to west trending faults have the main risk of juxtaposition seals, but have not been mapped continuously across the structure The licence group of licences PL208/209/250 is acknowledged for support and permission to publish the data. J. Tyssekvam and E. Berg have provided many helpful reviews and discussions along the way. We are also grateful for
280
L.M. STUEVOLD ETAL
DORE, A.G. & LUNDIN, E.R. 1996. Cenozoic compressional structures on the NE Atlantic margin: nature, origin and potential significance for hydrocarbon exploration. Petroleum Geoscience, 2, 299-311. EDWARDS, M. 1995. Differential subsidence and preservation potential of shallow water Tertiary sequences. In: PLINT, A.G. (ed.) Sedimentary Fades Analysis International Association of Sedimentologists, References Special Publications 22,265-281. BLYSTAD, P. BREKKE, H., RERSETH, R.B., LARSEN, B.T., FISHER, Q. & KNEPE, R.J. 1998. Fault sealing processes in SKOGSEID, J. & T0RUDBAKKEN, B. 1995. Structural siliciclastic sediments. In: JONES, G., FISHER, Q. & KNIPE, R.J. (eds) Fault sealing and fluid flow in hydroelements of the Norwegian continental shelf. Part II: carbon reservoirs. Geological Society, London, The Norwegian Sea Region. Bulletin of the Special Publications, 147,117-134. Norwegian Petroleum Society, 8. BUGGE, T. 1983. Submarine slides on the Norwegian conti- GORTER, J.D., GOSTIN, V & PLUMMER, P. 1989. The enigmatic Tookoonooka Complex in southwest nental margin, with special emphasis on the Storegga Queensland. In: O'NEILL, B.J. (ed.) The Cooper and area. IKUPublication, 110. Eromanga basins. Proceedings of the Petroleum CARTWRIGHT, J.A. 1994. Episodic basin-wide hydrofracExploration Society of Australia, 441-456. turing of overpressured Early Cenozoic mudrock HENRIET, J-P., DE BATIST, M. & VERSCHUREN, M. 1991. sequences in the North Sea Basin. Marine and Early fracturing of Palaeogene clays, southernmost Petroleum Geology, 11,587-607. North Sea. In: SPENCER, A.M. (ed.) Generation, accuCARTWRIGHT, J.A. & DEWHUHST, D. 1998. Layer-bound mulation and production of Europe's hydrocarbons. compaction faults in fine-grained Sediments. Bulletin European Association of Exploration Geologists, of the Geological Society of America, 110, Special Publication, 1,217-227. 1242-1257. CARTWRIGHT, J.A., MANSFIELD, C. & TRUDGILL, B.D. 1996. KJELDSTAD, A., SKOGSEID, J., LANGTANGEN, H.P., BJ0RLYKKE, K. & HOEG, K. in press. Differential The growth of faults by segment linkage: evidence loading by prograding sedimentary wedges and contifrom the Canyonlands Grabens of S.E. Utah. In: nental margins: all arch forming metamorphism. NIEUWLAND, D. (ed.) Structural validation of crosssection interpretation. Geological Society, London, Submitted to Journal of Geophysical Research. LINDSAY, N.G., MURPHY, F.C., WALSH, J.J. & WATTERSON, J. Special Publications, 99,192-214. 1993. Outcrop studies of shale smear on fault surCARTWRIGHT, J.A & LONERGAN, L. 1996. Volumetric contraction during the compaction of mudrocks: a faces. International Association of Sedimentologists Special Publication, 15,113-123. mechanism for the development of regional-scale LISTER, D. 2001. Three dimensional modelling of fault surpolygonal fault systems. Basin Research, 8,183-193. faces. Ph.D thesis, University of London. CARTWRIGHT, J.A. & MANSFIELD, C. 1998. Lateral tip LONERGAN, L., CARTWRIGHT, J.A. & JOLLY, R. 1998. 3-D geometry and lateral displacement variation on Geometry of Polygonal Fault Systems. Journal of normal faults in the Canyonlands National Park, Utah. Structural Geology, 20, 529-548. Journal of Structural Geology, 20,1-20. CARTWRIGHT, J.A., JAMES, D.M.D. & BOLTON, A. 2003. The LONERGAN, L. & CARTWRIGHT, J.A. 1999. Polygonal faults genesis of polygonal fault systems: a review. In: VAN and their influence on deep-water sandstone reservoir RENSBERGEN, P., HILLIS, R.R., MALTMAN, A. & geometries, Alba Field, UK Central North Sea. Bulletin of the American Association of Petroleum MORLEY, C. (eds) Subsurface Sediment Mobilization. Geologists. 83,410-432. Geological Society, London, Special Publications, MOLYNEUX, S. 2001. Post-depositional remobilisation of 216,223-243. CLAUSEN, O.R. & KORSTGAARD, J. 1993. Small scale faultsandstone reservoirs. Ph.D thesis, University of ing as an indicator of deformation mechanism. London. Journal of Structural Geology, 15,1343-1358. NICOL, A. WATTERSON, J. WALSH, J.J. & CHILDS, C. 1996. COLE, D., STEWART, S. & CARTWRIGHT, J.A. 2000. Giant The shapes, major axis orientations and displacement irregular pockmarks from the Eocene of the North patterns of fault surfaces. Journal of Structural Sea. Marine and Petroleum Geology, 17, 673-689. Geology, 18, 235-248. DAWERS, N.H. & ANDERS, M.H. 1995. DisplacementPEACOCK, D.P. & SANDERSON, DJ. 1991. Displacements, length scaling and fault linkage. Journal of Structural segment linkages, and relay ramps in normal fault Geology, 17, 607-614. zones. Journal of Structural Geology, 13,721-733. DEWHURST, D., CARTWRIGHT, J.A. & LONERGAN, L. 1999. RUNDBERG, Y. 1989. Tertiary sedimentary history and basin The development of polygonal fault systems by the evolution of the Norwegian North Sea. Dr Ing. Thesis, syneresis of fine-grained sediments. Marine and University of Trondheim. Petroleum Geology, 16,793-810. VAGNES, E., GABRIELSEN, R.H., & HAREMO, P. 1998. Late DDCON, R.J., SCHOFIELD, K., ANDERTON, R. & DAVIES, K.G. Cretaceous-Cenozoic intraplate deformation at the 1995. Sandstone diapirism and clastic intrusion in the Norwegian continental shelf. Tectonophysics, 300, Bruce-Beryl embayment. In: HARTLEY, A.J. & 29-46. PROSSER, DJ. (eds) Characterization of deep marine VERSCHUREN, M. 1992. An integrated approach to clay clastic systems. Geological Society, London, Special tectonic deformation. Ph.D thesis, University of Publications, 94,77-94. Ghent. excellent log interpretations provided by J. Bang as a basis for SGR estimates. Furthermore we thank P. Kjaernes, P. Gillespie, M. Shahly, M. Schneider and D. James for helpful comments to earlier presentations. D. Hansen and A. Robinson are thanked for assistance with data compilation.
POLYGONAL FAULTS IN THE ORMEN LANGE FIELD, OFFSHORE MID NORWAY WALSH, JJ. & WATTERSON, J. 1988. Analysis of the relationship between displacements and dimensions of faults. Journal of Structural Geology, 10,239-247. WATTS, N.L. 1987. Theoretical aspects of cap rock and fault seals for single and two phase hydrocarbon columns.
281
Marine and Petroleum Geology, 4,274-307. YIELDING, G. FREEMAN, B. & NEEDHAM, D.T. 1997. Quantitative fault seal prediction. Bulletin of the American Association of Petroleum Geologists, 81, 897-917.
This page intentionally left blank
Polygonal fault systems on the mid-Norwegian margin: a long-term source for fluid flow CHRISTIAN BERNDT, STEFAN BUNZ & JURGEN MIENERT Challenger Division for Seafloor Processes, Southampton Oceanography Centre, European Way, Southampton SO14 3ZH (e-mail: [email protected]) Abstract: 2D and 3D seismic data from the mid-Norwegian margin show that polygonal fault systems are widespread within the fine-grained, Miocene sediments of the Kai Formation that overlie the Mesozoic/Early Cenozoic rift basins. Outcropping polygonal faults show that de-watering and development of polygonal faults commenced shortly after burial. On the other hand, the polygonal fault system's stratigraphic setting, upward decreasing fault throw and the association with fluid flow features that are attributed to de-watering of the polygonal fault systems shows that polygonal faulting and fluid expulsion is an ongoing process since Miocene times.
The advent of 3D seismic data acquisition and interpretation led to the discovery of a new, non-tectonic class of faults called polygonal fault systems (Cartwright 1994). Such fault systems occur frequently in the fine-grained fill of sedimentary basins (Cartwright & Dewhurst 1998). Cartwright & Lonergan (1996) have suggested that the formation of polygonal fault systems is related to sediment contraction and fluid expulsion, as they are layerbound and not related to adjacent basement. The processes leading to contraction and water expulsion are still debated (cf. Cartwright et al 2003). Possible processes involved in their development include syneresis of colloidal sediments (Dewhurst et al. 1999) and Rayleigh-Taylor instabilities due to density inversions (Watterson etal. 2000). Understanding of such fault systems is important as they interact with adjacent reservoirs (Cartwright 1994; Lonergan & Cartwright 1999) and because they might control fluid flow on a regional scale (Henriet et al. 1991). Growth-related sedimentary successions at the top of the polygonal fault systems revealed that their development commences during early burial of the host sediments (Cartwright & Lonergan 1996; Lonergan et al. 1998). However, until now there was little information about how long the involved processes remain active. Here, we present evidence from 3D and 2D seismic data for long-term fluid flow from the polygonal fault systems of the mid-Norwegian V0ring Basin. It suggests that fluid expulsion related to polygonal fault development in this area is an ongoing processes since the Early Miocene.
Polygonal fault systems on the midNorwegian margin The mid-Norwegian continental margin has formed as a result of several rifting episodes leading to Late Paleocene/Early Eocene continental break-up and
development of the Norwegian-Greenland Sea (Skogseid et al. 2000). Subsequently, the margin experienced an episode of moderate compression during the Oligocene and Miocene, which led to the development of dome structures (Brekke & Riis 1987; Skogseid & Eldholm 1989; Dore & Lundin 1996; Vagnes et al. 1998). From the Pliocene to the Pleistocene the margin was glaciated yielding a thick wedge of clastic sediments on the shelf (Vorren et al. 1998). Here, the area north of the Storegga Slide (Bugge et al. 1987) is focused on, a large submarine slope failure that occurred at 8200 yrs b.p. (Haflidason et al. 2001). The sedimentary successions in this area include the Brygge Formation of the Eocene/Oligocene Hordaland Group, the Miocene/earliest Pliocene Kai Formation, which is generally characterized by fine-grained hemipelagic oozes, and the Plio-/Pleistocene glacially derived hemipelagic contourites and debris flows of the Naust Formation (Blystad et al. 1995; Rokoengen et al 1995). Hjelstuen et al. (1997) mapped the extent of small-offset faults on the mid-Norwegian margin (Fig. 1) showing that they are widespread in the upper Oligocene and Miocene successions of the V0ring Margin. 3D seismic data show the polygonal shape of these faults in plan view (Fig. 2). This shape implies a lack of dominant strike directions. The faults commonly occur in at least two tiers in the uppermost Brygge and in the Kai Formations. Their vertical extent is variable, i.e. the upper and lower terminations of adjacent faults do not necessarily occur at the same stratigraphic level (Fig. 3a). The fault frequency is much higher in the lower part than in the upper part of the tiers. Major faults cut through the whole tier, whereas many smaller faults exist in the lower part and terminate either at deeper stratigraphic levels or at major faults (e.g. Fig. 2). The fault frequency is lower where the host formation thins out towards the Tertiary dome structures.
Fig. 1. Polygonal faults on the mid-Norwegian margin are primarily located within the Cretaceous/Early Tertiary rift basin between the break-up related volcanic rocks to the west (light gray, after Berndt et al. 2001) and the Tr0ndelag platform in the East (light gray, after Blystad et al 1995). Black lines and circles indicate seismic examples shown in Figs 2-5. JMFZ, Jan May en Fracture Zone. On 2D seismic lines the average spacing between individual faults is smaller than the maximum polygon diameter determined from the 3D seismic data measured at the top of the tiers (Fig. 3b). This supports the interpretation that most of the faults in
the 2D seismic data are indeed part of polygonal fault systems. Generally, the fault throw is highest close to the lower termination (Fig. 3c), with an average between 20-40 ms, but may occasionally reach much higher
LONG-TERM DEVELOPMENT OF POLYGONAL FAULT SYSTEMS
285
Fig. 2. Regional seismic section showing polygonal fault systems (cf. Fig. 1 for location). The inset shows the dip attribute map of a horizon slice from the Ormen Lange 3D block located to the south of the seismic section. The inset is projected onto the seismic line with the dashed line indicating approximate depth.
Fig. 3. (a) High-resolution seismic profile JMF97 (courtesy Fugro Geoteam AS) showing pipes originating at the upper termination of polygonal faults (arrows). It also shows an example for a possibly still active pipe with down-bending reflectors (circle). Bottom simulating reflector (BSR) indicates the base of the gas hydrate stability zone; (b) distance between faults in 2D seismic lines and width of polygons in 3D seismic data; Norm, count, number of polygons with a given diameter divided by number of polygons evaluated in total; (c) fault throw for 30 to 50 faults on individual 2D seismic lines (names indicated). The throw along polygonal faults generally increases downward. In some areas, however, the throw near the lower fault termination decreases, e.g. line NH9753-402.
The sediments of the Naust Formation are genervalues. The height of the faults varies depending on the thickness of the hosting Kai Formation, i.e. ally not faulted. However, they frequently show where the formation is thicker the faults have a gentle deformation where polygonal faults exist in the underlying formations. greater vertical extent and vice versa. In the NW part of the study area where the Kai Formation, is not overlain by the Naust Formation, the faults reach close to the sea floor (Fig. 4a). In the Fluid flow eastern part of the study area, i.e. above the Ormen Lange Dome, they are buried by up to 1000 m of sed- The study area is characterized by a number of fluid flow indicators in the geological and geophysical iments.
286
C.BERNDT£7AZ,
Fig. 4. Seismic evidence for prolonged polygonal fault system development, (a) Polygonal faults offset the sea floor; (c-d) fluid expulsion related pipes that terminate at different stratigraphic levels, Naust A-E, Pleistocene (0.015 my.-18 m.y.); Naust F-H, Pliocene (1.8-3.6 m.y.); Kai A-C, Miocene (23.8 m.y.). Seismic panels located on Fig. 1.
data. The most prominent evidence stems from laterally narrow, i.e. 20-200 m wide, circular zones of up-bending, low-amplitude reflectors (Fig. 5). In the following, these zones are called pipes, because they are similar to those reported from the Niger Delta and onshore Rhodes by L0seth et al (2001). L0seth et al. (2001) interpreted them to be the result of episodic fluid expulsion. The pipes on the midNorwegian margin originate at different depth, most often at the base of an inferred gas hydrate layer (Mienert et al. 1998; Posewang & Mienert 1999) and in some instances at the upper termination of polygonal faults in the top of the Kai Formation, which coincides with a layer of high reflectivity (Fig. 3a). They never occur within or below the polygonal fault systems. The upper termination of the pipes is at different stratigraphic levels within the Naust Formation or at the seabed (Fig. 4). The pipes are up to 600 ms two-way-travel time or approximately 550 m high (Fig. 3a). There are some pipes with down-bending reflectors, which are interpreted as a result of velocity pull-down possibly indicating active fluid expulsion (circle in Fig. 3a). The upper terminations of pipes with downbending reflectors are at different stratigraphic levels, but none of these pipes reach the seabed. Pronounced step-wise seismic amplitude changes and step-wise changes for reflector pull-up in the pipes (e.g. Fig. 4c at 1.3 s) are consistent with episodic rather than continuous activity. A second line of evidence for substantial fluid flow can be derived from side-scan sonar data showing pockmarks above some of the pipes (Fig. 5). Pockmarks are frequently caused by fluid flow (Hovland&Juddl988). Finally, pronounced seismic amplitude anomalies
at the base of the gas hydrate stability zone also indicate a dynamic fluid flow system (Fig. 3a). These amplitude anomalies are interpreted to be the result of free gas that is trapped underneath hydrate sediments (Mienert et al. 1998; Posewang & Mienert 1999). Mienert & Posewang (1999) showed seismic evidence for fluid expulsion from the gas hydrate system. The seismic fluid flow indicators are most common near the northern sidewall of the Storegga Slide and relatively sparse in the rest of the study area.
Discussion Fluid origin The processes leading to polygonal fault system development are not well understood, but all genetic models for polygonal fault systems involve fluid expulsion from the host rock. The volume of expelled fluids might be as much as 60% (Verschuren 1992). It is therefore reasonable to anticipate that the polygonal fault systems provide a major source for the fluid flow observed in the study area. We also recognize that some fluids might ascend from greater depth. However, from the absence of fluid flow indicators underneath or within the polygonal fault systems it is inferred that the volumes of such fluids must be small compared to the fluids expelled from polygonal fault systems. Sampling of the ascending fluids in one of the still active pockmarks and their gee-chemical analysis may constrain the fluid origin in more detail, but this has not been attempted so far.
LONG-TERM DEVELOPMENT OF POLYGONAL FAULT SYSTEMS
287
Fig. 5. Composite plot of side-scan sonar sea floor backscattering image (top) and high-resolution single channel seismic profile (bottom) showing that some of the pipes terminate in circular pockmarks. Dark colours in the side-scan sonar data indicate high backscatter.
Fluid channelling Polygonal fault systems are widespread in the study area (Figs 1 and 2). However, fluid flow indicators are not similarly abundant. Most often they occur in areas with gas hydrates or glacial debris flows in the
overburden. We interpret this observation as an indication for diffuse flow of fluids out of the polygonal fault system that is not observable in the seismic data, unless the fluids get trapped either by hydrated sediments or by less permeable debris flows. In these instances pockets with high pore pressure will
288
C.BERNDTErAL.
develop that will expel fluids episodically when the pore pressure exceeds the strength of the trap and causes pipes and amplitude anomalies observed in the seismic data. In the area with observed fluid-escape features the polygonal faults terminate into a layer, which shows increased reflectivity in seismic sections (Fig. 3a), probably indicating the accumulation of gas-enriched fluids within the sediments. Although the main flow from here might be diffusive, the fact that some pipes start at the upper termination of polygonal faults (Fig. 3a) suggests that the polygonal faults or their vicinity are preferential pathways for fluid migration.
Timing Outcropping polygonal faults at the sea floor in the western part of the study area (Figs 1 and 4) show that polygonal faults start to develop immediately after burial in accordance to the observations of growth structures observed by Cartwright (1994) and Cartwright & Lonergan (1996). Assuming that dewatering of the polygonal fault system causes the observed fluid flow the seismic fluid flow indicators can be taken as a proxy for the timing of the process that leads to the development of polygonal fault systems. Figure 4 shows pipes that have their upper termination at different stratigraphic levels. These stratigraphic levels denote the earliest possible dates for the pipes and possibly time of activity assuming that they have not been re-used. The fact that some pipes continue all the way to the sea floor causing pockmarks implies that fluid flow and therefore dewatering of the polygonal fault system must have been active until recently. The monotonous downward increase of fault throw is a second line of evidence pointing to an extended dewatering history. If the sediments would expel the entire amount of pore water immediately after burial, the throw along individual faults should be constant with depth. Lithological variations, e.g. variations in water content, should give random variations in throw. Only a process that is active over the entire burial history of the polygonal fault system would result in a monotonous downward increase of fault throw. The frequently observed reduced throw in the lowermost part of the faults is not well understood. It possibly indicates that initiation of the process depends on a threshold thickness of the previously deposited sediments, before it picks up. We do not observe clustering of the upper terminations of the pipes at any particular stratigraphic level. This does not allow deducing temporal variations of fluid flow. However, the fact that the fault throw is decreasing upwards indicates that most of the faulting happens during burial. It is inferred that
fluid flow due to de-watering was most vigorous at an early stage of polygonal faulting and that it decreased subsequently. The moderate deformation of the overlying Naust Formation, i.e. no clear faults but rather sacking structures, supports the interpretation that most faulting already occurred before deposition of this unit. The area in which polygonal fault systems occur on the mid-Norwegian margin is extensive (Fig. 1) and pipes related to fluid flow are found in many places within this region. This leads to the conclusion that long-term fluid flow related to polygonal faults is a general pattern. Nevertheless, there are some areas such as the northern sidewall of the Storegga Slide, in which pipes are more abundant than elsewhere. This suggests that de-watering and polygonal faulting is variable in time and space most likely as a result of external forces such as the Storegga Slide or thickness variations of the faulted sediment units. In fact, the Kai Formation is thickest in the northern part of the Storegga Slide area (Britsurvey 1999). Finally, the downward increase in fault frequency and the ensuing downward decrease in fault block size point towards long-term and still ongoing sediment contraction and fluid expulsion. Small-size fault blocks in the lowermost parts of the polygonal fault system combine into larger fault blocks higher up in the sections. This is interpreted as a change of faulting scale. Faults apparently start to grow in the lower part and advance upwards until they terminate either at the upper boundary or into another fault and thus they focus strain along fewer faults. This is consistent with a long-term activity of the faults.
Conclusions 3D seismic data from the Ormen Lange Dome demonstrate the existence of polygonal fault systems on the mid-Norwegian margin. Extrapolating the results of 3D-seismic imaging by using 2-D seismic data and the mapping results of Hjelstuen et al. (1997) it is found that substantial parts of the outer V0ring and M0re margins are underlain by polygonal fault systems. Stratigraphically, the fault systems are located in the fine-grained, hemi-pelagic sediments of the Kai Formation, which were deposited during the Miocene. Under the assumptions that seismic fluid flow indicators can be taken as a proxy for active fluid flow the duration of polygonal fault system development can be quantified. The variable stratigraphic position of the top terminations of the fluid flow indicators implies that the processes that cause fluid expulsion and perhaps development of the polygonal fault systems have been active on the midNorwegian margin ever since Miocene times.
LONG-TERM DEVELOPMENT OF POLYGONAL FAULT SYSTEMS
Fluid flow from the polygonal fault systems and its spatial relationship to the occurrence of natural gas hydrates poses the question whether both are related and whether the polygonal fault systems might be a source of fluids that are involved in the development of gas hydrates. Coincidence of some polygonal faults and the northern Storegga Slide side-wall might indicate that the polygonal fault systems play a role for continental slope stability. These questions have to be investigated in more detail in the future. The authors would like to thank D.A. Dewhurst and L. Lonergan for careful reviews. Furthermore we thank Norsk Hydro ASA and Fugro Geoteam AS in Oslo for permission to show their seismic data. The work was funded by a Norsk Hydro grant NHT-B44-VK0768-00, NFR grant 128164/432, and EC grant EVK3-CT-1999-00006 (COSTA project).
References BERNDT, C, PLANKE, S., ALVESTAD, E., TSIKALAS, F. & RASMUSSEN, T. 2001. Seismic volcanostratigraphy of the Norwegian Margin: constraints on tectono-magmatic breakup processes. Journal of the Geological Society, 158,413-426. BLYSTAD, P, BREKKE, H., F^ERSETH, R.B., LARSEN, B.T., SKOGSEID, J. & T0RUDBAKKEN, B. 1995. Structural elements of the Norwegian continental shelf; II: The Norwegian Sea Region, Norwegian Petroleum Directorate, Stavanger. BREKKE, H. & Rns, F. 1987. Tectonics and basin evolution of the Norwegian shelf between 62° N and 72° N, Norsk Geologisk Tidsskrift, 67,295-322. BRITSURVEY. 1999. Seabed Project - Geological and geophysical interpretation in M0re/V0ring Area, Final report, Britsurvey, Norfolk, UK. BUGGE, T, BEFRING, S., BELDERSON, R. H., EIDVIN, T, JANSEN, E., KENYON, N.H., HOLTEDAHL, H. & SEJRUP, H.P. 1987. A giant three-stage submarine slide off Norway. Geo-MarineLetters,!, 191-198. CARTWRIGHT, J.A. 1994. Episodic basin-wide hydrofracturing of overpressured Early Cenozoic mudrock sequences in the North Sea Basin, Marine and Petroleum Geology, 11. No.5, 587-607. CARTWRIGHT, J.A. & DEWHURST, D.N. 1998. Layer-bound compaction faults in fine-grained sediments. Bulletin of the Geological Society of America, 110, 1242-1257. CARTWRIGHT, J.A. & LONERGAN, L. 1996. Volumetric contraction during the compaction of mudrocks: a mechanism for the development of regional-scale polygonal fault systems. Basin Research, 8,183-193. CARTWRIGHT, J.A. JAMES, D. & BOLTON, A. 2003. The genesis of polygonal fault systems: A review. In: VAN RENSBERGEN, P., HILLIS, R.R. MALTMAN, A.J. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216,223-243. DEWHURST, D.N., CARTWRIGHT, J.A. & LONERGAN, L.
289
1999. The development of polygonal fault systems by syneresis of colloidal sediments. Marine and Petroleum Geology, 16,793-810. DORE, A.G. & LUNDIN, E.R. 1996. Cenozoic compressional structures on the NE Atlantic margin: nature origin and potential significance for hydrocarbon exploration. Petroleum Geosciences, 2, 299-311. HAFLIDASON, H., SEJRUP, H.-P, BRYN, P. & LIEN, R. 2001. The Storegga Slide; Chronology and flow mechanism. Paper presented at the XI European Union of Geoscientists Meeting. 8-12 April, Strassbourg, France. HENRIET, J.P., DE BATIST, M. & VERSCHUREN, M. 1991. Early fracturing of Paleogene clays, southernmost North Sea: relevance to mechanisms of primary hydrocarbon migration. In: SPENCER, A.M. (ed.) Generation, Accumulation and Production of Europe's Hydrocarbons, Oxford University Press, Oxford 217-227. HJELSTUEN, B.O., ELDHOLM, O. & SKOGSEID, J. 1997. V0ring Plateau diapir fields and their structural and depositional settings. Marine Geology, 144 (1-3), 33-57. HOVLAND, M. & JUDD, A.G. 1988. Seabed pockmarks and seepages. Impact on biology, geology, and environment. Graham and Trotman, London. LONERGAN, L., CARTWRIGHT, J.A. & JOLLY, R.J.H. 1998. The geometry of polygonal fault systems in Tertiary mudrocks of the North Sea. Journal of Structural Geology, 20,528-548. LONERGAN, L. & CARTWRIGHT, J.A. 1999. Polygonal faults and their influence on deep-water sandstone reservoir geometries, Alba Field, United Kingdom Central North Sea. American Association of Petroleum Geologists Bulletin, 83,410-432. L0SETH, H., WENSAAS, L. ARNTSEN, B., HANKEN, N.M., BASIRE, C. & GRAUE, K. 2001.1000 m long blow-out pipes. Paper presented at the EAGE 63rd conference and technical exhibition, Amsterdam, The Netherlands, 11-15 June 2001. MlENERT, J., POSEWANG, J. & BAUMANN, M. 1998. Gas
hydrates along the northeastern Atlantic margin: possible hydrate-bound margin instabilities and possible release of methane. In: HENRIET, J.-P. & MIENERT, J. (eds) Gas Hydrates: Relevance to World Margin Stability and Climate Change. Geological Society, London, Special Publications, 137,275-291. MIENERT, J. & POSEWANG, J. 1999. Evidence of shallowand deep-water gas hydrate destabilization in North Atlantic polar continental margin sediments. GeoMarine Letters, 19,143-149. POSEWANG, J. & MIENERT, J. 1999. High-resolution seismic studies of gas hydrate. Geo-Marine Letters, 19, 150-156. ROKOENGEN, K., RlSE, L., BRYN, P., FRENGSTAD, B.,
GUSTAVSEN, B., NYGAARD, E. & S^TTEM, J. 1995. Upper Cenozoic stratigraphy on the mid-Norwegian continental shelf. Norsk Geologisk Tidsskrift, 75, 88-104. SKOGSEID, J. & ELDHOLM, O. 1989. V0ring Plateau continental margin: Seismic interpretation, stratigraphy, and vertical movements. In: ELDHOLM, O., THIEDE, J. & TAYLOR, E. (eds) Scientific Results of the Ocean Drilling Project. Leg 104,993-1030. College Station, Texas.
290
C.BERNDTCTAZ,
SKOGSEID, J., PLANKE, S., FALEIDE, J.L, PEDERSEN, T., ELDHOLM, O. & NEVERDAL, F. 2000. NE Atlantic continental rifting and volcanic basin formation. In: N0TTVEDT, A. (ed.) Dynamics of the Norwegian Margin. Geological Society, London, Special Publications, 167,295-326. VERSCHUREN, M. 1992. An integrated approach to clay tectonic deformation, and development of a new 3D surface modeling method, Phd thesis, University of Ghent, 359 pp. VORREN, T.O., LABERG, J.S., BLAUME, F. DOWDESWELL, J.A., KENYON, N.H. MIENERT, J., RUMOHR, J. & WERNER, F. 1998. The Norwegian-Greenland Sea
continental margins: Morphology and late Quaternary sedimentary processes and environment Quaternary Science Reviews, 17,273-302. VAGNES, E. GABRffiLSEN, R.H., & HAREMO, P. 1998. Late
Cretaceous-Cenozoic intraplate contractional deformation at the Norwegian Sea continental shelf: timing, magnitude and regional implications. Tectonophysics, 300, 29-46. WATTERSON, J, WALSH, J., NICOL, A., NELL, P.A.R. & BRETAN, P.G. 2000. Geometry and origin of a polygonal fault system. Journal of the Geological Society, 157,151-162.
Normal faulting in chalk: tectonic stresses vs. compaction-related polygonal faulting C. HIBSCH1, J. CARTWRIGHT2, D.M. HANSEN2, P. GAVIGLIO3, G. ANDRE1, M. GUSHING4, P. BRACQ5, P. JUIGNET6, P. BENOIT7 & J. ALLOUC8 1
UMR G2R, Universite Henri-Poincare; BP 239, 54506 Vandceuvre-les-Nancy cedex, France (e-mail: Christian, hibsch@g2r. uhp-nancy.fr) 2 Dept of Earth Sciences, Cardiff University; PO Box 914, Park Place, Cardiff CF10 BYE, Wales UK 3 EA 2642 Geosciences, Universite de Franche-Comte, 16 route de Gray 25030 Besangon cedex France 4 IRSN / DPRE / SERGD / BERSSIN; BP 17, 92262 Fontenay-aux-roses, France 5 Universite du Littoral Cote d'Opale, L.I.S.E, 32 avenue Foch, 62 930 Wimereux., France 6 Lab. Geologic de Normandie occidentale, Universite de Caen; 14032 Caen cedex, France 1 Lab. hydrologie; 10170 Mery-sur-Seine, France *ENSG - INPL, BP 40; 54501 Vandceuvre-les-Nancy cedex, France Abstract: This paper documents normal fault sets observed in chalks exposed in widely separated localities in the UK and France. These faults are characterized by having a wide range of strikes at any one locality, are developed entirely within the chalk succession and do not seem to interconnect to deeper or shallower structures. These structures may result from two different mechanisms: (1) complex polyphase deformational histories involving contrasting stress states; or (2) a single deformational phase in which the faults develop to accommodate compactional strains. Evidence is presented from microstructural and petrographic data to support the latter interpretation. In particular, the association of calcite and marcasite mineralizations with fracture surfaces and fault zones and textural observations relating flint occurrence to early fault formation point towards fault propagation at a very early stage of burial and compaction of the chalky sediments. An analogy is drawn between these outcrop-scale structures and polygonal fault systems at a larger scale recognised from seismic observations of chalk sequences deposited at passive continental margins. The origin of these structures may be related to syneresis at an early stage of deformation followed by pressure solution phenomena that may reactivate this early-inherited polygonal fault pattern until the present day.
The analysed chalk deposits range from Albian to at least Campanian in age and will be indifferently referred as chalk in the text. Previous palaeostress analysis conducted in the UK defined post-Variscan rifting and inversion stages (Hibsch etal 1993). The abundance and ubiquity of normal faults encountered in these Cretaceous chalks was one of the most startling aspects, given that geodynamic data suggest that rift tectonics were no longer active in these regions. In the southern North Sea basins, the onset of chalk sedimentation followed the end of the opening of the Bay of Biscay. The main rifting stages were over before Late Cretaceous times and the major synsedimentary normal faults were buried in the basins surrounding UK and NW France (Kent 1980; Ziegler 1990). Compaction-related driving factors may explain such widespread omni-directional normal faulting. This interpretation conflicts, however, with results derived from southern England and northern France (Vandycke & Bergerat
1992; Pairis et al. 1997) or from northern England (Peacock & Sanderson 1994; Starmer 1995) where Late Cretaceous to Cenozoic extensional tectonic stages have been defined by structural analysis of sets of normal faults measured in sediments spanning this age range. The aim of this paper is to evaluate the structural data and to propose an alternative compaction-related polygonal faulting explanation in which the chalk is considered to have been affected by radial extension. A general overview of the problem is presented, with data and references to studies from Central and northern England and from the Paris basin. The approach adopted here was: (1) to check the consistency between the published extensional regimes and the geodynamic data; (2) in cases of possible conflict between interpretations, evaluate whether a single non-tectonic omni-directional extension could be an alternative solution; (3) acquire additional field measurements at key localities to test
alternative interpretations; (4) evaluate data suggestive of a link between radial extension and diagenesis. Fieldwork was undertaken in Normandy and in the Boulonnais to analyse possible Late Cretaceous extension. The Cenozoic opening of the Lower and Upper Rhine Grabens was considered to have been related to tectonic normal faulting in the Late Cretaceous chalk and thus explain a focus on microtectonic sites from the eastern Paris Basin.
Methodology for microtectonic analysis If polygonal faulting is considered as a possible deformation in chalks, microtectonic analysis of normal faults should be considered carefully. Two endmember approaches can be applied to the treatment of normal fault populations that reflect uniaxial finite shortening in chalk. The first one is to consider that uniaxial shortening accommodates vertical heterogeneous compaction of the sediment, in which case faulting is not related to tectonics. The second one is to consider that this three dimensional strain is the result of the superimposition of distinct two dimensional deformation events each related to different tectonic events. Sorting of normal fault populations appears to be a crucial point. It can be based on relative chronologies of striations or crosscutting relationships of faults, but for most of the faults the sorting is based on apparent compatibility of fault kinematics and such compatibility depends on initial postulates about the expected tectonic regime. Published micro tectonic results from chalk outcrops in France and England were discussed to see whether such sorting did not hide a single radial extension phenomenon, especially when syndiagenetic normal faulting evidence was observed on the site. Computation of palaeostress tensors based on fault measurements was obtained with an inverse method (Carey 1979). It allows the definition of the dip-direction of three stress axes and a ratio, (R), comparing the three vector values: R = o2-crl/cr3-crl. The mathematical solution considers normal and tangential stresses for each fault plane defining a theoretical slip-vector, (r), parallel to the tangential stress resolved for each plane. The minimization process defines R and the three stress axes for which the sum of the angle values between r and the measured striae (s) for each fault is minimum. Faults characterized by r-s angles largely over 20° do not fit perfectly to the proposed mathematical solution and should be considered carefully. Consequently, compaction-related radial extension is defined with R tending towards 1 (al close to
Tectonic settings and fault analysis England and surrounding basins Three post-Variscan rifting stages were characterized with the analysis of about 400 natural outcrops and quarries (Hibsch et al 1993). The youngest of these commenced during the Late Jurassic and ended prior to the Albian (Badley et al. 1989; Chapman 1989; Ziegler 1990). A NNE-SSW direction of minimum horizontal stress (o3) was defined and possibly coeval with WNW-ESE trending transcurrent regime (transtension) (Benard et al. 1985). However, this transcurrent regime seems to have been also expressed in the chalk at least during Late Cretaceous times (Glennie & Boegner 1981; Starmer 1995; Pairis et al. 1997) and may reveal a regional palaeostress evolution between Early and Late Cretaceous times by permutation between al and crl. Sites related to the Late Jurassic Early Cretaceous rifting event are mainly located close to regional fault systems (Fig. 1). Kimmeridgian and Early Cretaceous synsedimentary movements in the Pickering Graben, between the Market Weighton Block (MWB) and the Cleveland Basin, were over prior to the deposition of the chalk (Kirby & Swallow 1987). In the East Anglia domain, away from regional faults, numerous outcrops of Jurassic beds did not show any evidence of normal faulting (Fig. 1) while in the same area, normal faults were abundant in the Late Cretaceous chalks sites (dots on Fig. 1). Most of these normal fault planes display dip-slip movements and no relative chronologies of striations related to extensional regimes were found. The dip-directions of faults are scattered in all directions and thus it was possible to compute any given direction of o3 if a sorting was applied to these sets of normal faults. If such palaeostress regimes were defined with the normal faults measured in the chalk, the three palaeostress directions characterizing postVariscan rifting events derived from localities studied in Permian to Early Cretaceous sediments could have been interpreted as a result of Late Cretaceous or younger tectonic deformation. Such a conclusion appears to be in conflict with geodynamic data. In northern England, the end of the rifting is marked by the Late Cretaceous unconformity above the MWB and adjacent Cleveland Basin (Kent 1980). The onset of chalk deposition started with the Albian Red Chalk and occurred mainly during the thermal subsidence stage without clear tectonic subsidence related to normal faulting. Late Cretaceous extensional deformations are not well documented and may be related to local transtensional regimes sometimes in association to synsedimentary folds (Mortimore & Pomerol 1991).
CHALK, TECTONICS VS. POLYGONAL FAULTING
Normal faulting encountered in the English chalk was first considered to be due to non-tectonic radial extension (Hibsch et al. 1993) and all the normal faults were computed together (Fig. 2a). In comparison, the Oxfordian to Aptian extensional regime (sites from Fig. 1) is of clear tectonic origin. A clear distinction appears when more than six groups of 10°-interval of dip-directions are involved in the computing. For chalks, the R ratio evolves over 0.8 and defines radial extension (Fig. 2a) when R evolves below 0.8 for the Oxfordian to Aptian extension (Fig. 2b). The increase of dip-directions in any computation should allow the grouping to be made with faults displaying more oblique-slip movements. One consequence for the Late Jurassic to Early Cretaceous regime is the evolution of R towards 0 since it includes faults displaying transcurrent kinematics. This means an evolution towards a transtensional regime (crl close to cr2) compatible with the Cretaceous WNW-ESE transcurrent regime. On the contrary, such associated obliqueslip kinematics were lacking in the chalk and R evolved towards 1. A very detailed study in the Flamborough Head region (Peacock & Sanderson 1994) corresponds to our sites 30 and 34 (Fig. 3, above the inverted Pickering Graben). Normal faults sets in the chalk were analysed in terms of strain. One very interesting result came from the analysis of the intersections between the normal faults. No coherent systematic chronology was found and thus, did not allow the differentiation of palaeoextensional regimes. Measuring fault displacements, they even concluded that there was a homogeneous strain in all horizontal directions. Since this pure flattening finite strain was not clearly coherent with a classical tectonic regime, the result was suggested to be due to the localization between two major inherited normal faults of the Pickering Graben. In contrast, however, our regional study suggests that this radial extension is a general phenomenon in the chalk, which does not need to be explained by the presence of any specific underlying major faults. Considering the microtectonic response of the chalk during inversion stages gives other clues and allows precise timing of the radial extension in the tectonic evolution of UK. The omni-directional extension is also recognized by Starmer (1995) in the Flamborough Head area but considered to have occurred after a first ENE-WSW to E-W trending compression related to the Late Cretaceous Laramide stage. Such chronology suggests a quite recent radial extension, long after first diagenetic processes. On the contrary, the Late Cretaceous to Palaeocene Laramide inversion stage may be interpreted with a NW-SE direction of shortening (Hibsch et al. 1993) coeval with a rifting pulse around northern UK (Ziegler 1987). Magmatic dikes
293
related to this event are observed in northern England (England 1988) (Fig. 3) and are consistent with a NE-SW direction of
294
C.fflBSCHZrAL.
Fig. 1. Palaeostress map related to the Oxfordian to Aptian rifting stage and distribution of all our sites of microtectonic analysis (dots); i.e. the abundance of sites in the Cretaceous chalk. Larger site number refer to Fig. 2.
Para Basin We focused on three zones (Fig. 4) with possible extensional faulting of tectonic origin: (1) in Normandy (Etretat), where Aptian to Coniacian synsedimentary deformations may be related to Cretaceous tectonic normal faulting; (2) in the Boulonnais (Cap Blanc-Nez), where at least four extensional regimes are documented with one of them possibly related to the influence of the Lower Rhine Graben and; (3) in the Champagne region (Chepy/Omey fault zone), where hydraulic fracturing of the chalk is related to the opening of the Upper Rhine Graben during the Oligocene. Etretat (Normandy). The chalk outcropping along the cliffs of Etretat ranges in age from Cenomanian to Coniacian (Kennedy & Juignet 1974). Flint beds
are very frequent in these exposures and outline impressive synsedimentary deformation features (underwater truncations, slumped beds, mass sliding). The regional fault system trends NW-SE from the Bray anticline to the Merlerault axis and is considered to have influenced Cretaceous sedimentation (Juignet 1971). The tectono-sedimentary activity controlling basin and swell morphologies from Aptian to Cenomanian times (Kennedy & Juignet 1974) may have affected sediment distribution until at least the Coniacian (Juignet & Breton 1992). Small buried normal faults have been described, but were related to gravitational sliding deformation (Kennedy & Juignet 1974). Outcrops along the 'le Tilleul' beach were analysed, which lies 20 km SW of the Fecamp Lillebonne Fault (Fig. 4) and is regarded as a border fault of a Late Cretaceous tilted block (Mortimore &
CHALK, TECTONICS VS. POLYGONAL FAULTING
295
Fig. 2. Evolution of the R ratio with the increase of groups of 10° of dip directions of faults in the computing. Comparison between measurements in chalk (graph [a] and associated stereogram, site [86] on Fig. 3) and palaeostress tensors related to the Oxfordian to Aptian rifting stage(graph [b] and associated stereogram, site 64] on Fig. 1). Wulff lower hemisphere projection, Nm: magnetic north, Ng: geographical north. Arrows indicate real slip motions, R = o2-crl/o-3-o-l, the histogram diagrams and the bold trace on cyclographic projections present the angle value between the measured striation (s) and the theoretical position of the slip motion (r) according to the palaeostress tensor result. 1 box = 1 fault, numbers in the boxes refer to the microtectonic database.
296
CfflBSCHCTAL.
Fig. 3. Palaeostress map related to the post-Palaeocene 'Pyrenean' inversion stage; notice the abundance of sites (dots) in the Late Cretaceous chalk from East Anglia without evidence of strike-slip faulting. Larger site numbers refer to the text or to Fig. 2.
Pomerol 1991). Slickenside striae on the analysed fault surfaces display strike-slip, dip-slip movements, but also oblique-slip faulting along certain structures such as for 'la Valleuse du Tilleul' Fault, which shows about 4 metres of apparent vertical throw. In the hanging-wall block, close to the fault plane, chalk and flints are brecciated and the pebbles are mixed in a chalk matrix showing creep features reminiscent of mylonite fabrics (Fig. 5a). This kind of ductile deformation is indicative of hydroplastic faulting. These faults affect the whole cliff section and thus post-date the Coniacian (Kennedy & Juignet 1974). A north to south trending strike-slip regime (Fig. 6a) and an east to west trending extension (Fig. 6b) can be defined. Both show an average R ratio. The direction of o3 is almost the same for
the two regimes. All these faults can be computed together to define an extensional regime (Fig. 6c) suggesting transtensional deformation. The north to south direction of al (Fig. 6b) is coherent with rightlateral inversion along NW-SE structures such as the Bray or the Fecamp-Lillebonne faults (Fig. 4) (Pomerol 1977; Mortimore & Pomerol 1991). The microtectonic analysis (Fig. 6c) may reflect Late Cretaceous faulting coherent with block tilting between the Fecamp and the Rouen-St. Valery Fault (Mortimore & Pomerol 1991) or transtensional faulting during Cenozoic north to south trending 'Pyrenean' inversion stage. One important fact is the lack of scattered distribution of normal faults. More probably of tectonic origin, a coherent single extensional regime has been computed and gives moder-
CHALK, TECTONICS VS. POLYGONAL FAULTING
297
Fig. 4. Map of location of selected microtectonic sites (dots) in chalks from the Paris Basin.
ate R ratio (Fig. 6). One main difference to other chalk outcrops is the abundance of hard-grounds and flint levels, which may have strengthened the chalk during compaction processes. Similar observations were made in Flamborough Head, northern UK (Peacock & Sanderson 1994), where the number of normal faults seems to strongly decrease in conjunction with the decrease of clay content and increase of flint beds. Cap Blanc-Nez (Boulonnais, Nord Pas-de-Calais). Four extensional regimes (N-S, NW-SE, E-W and NE-SW) were described in this area (Vandycke & Bergerat 1992). A second main point of interest is the presence of syntectonic calcite along the fault planes which might have been interpreted as a tectonic signal related to pressure-solution. To obtain their palaeostress results, the authors have done a sorting of fault populations. Faults are believed to initiate as conjugate fault sets displaying dip-slip movements. Fault planes displaying oblique-slip kinematics are interpreted as evidence of a relative chronology, since such movement is considered to have occurred after an initial dip-slip movement in the conjugate fault system (Vandycke & Bergerat 1992). Relative chronologies of striations on such fault planes seem to have confirmed this hypothesis and led the authors to the following conclusions: (1) east to west extension of Late Cenomanian age (upwards dying out of some north to south striking normal faults below Turonian chalks, but some of them also affect the Turonian; (2) occurrence of a north to south extension; (3) NW-SE extensional regime; (4) relative chronology on one fault belonging to the NW-SE regime would indicate that a NE-SW extension postdated the other phases. This last regime is related to
the Miocene to Quaternary palaeostress field developed around the Lower Rhine Graben. Sand fillings in NW-SE striking faults and the general NW-SE strike of the major drainage pattern is believed to be influenced by the neotectonic fracturing. For the three previous regimes, no convincing correlation with Late Cretaceous to Cenozoic rifting stage is available. New microtectonic analysis was performed taking also into account relative chronologies on fault planes. A similar sorting was applied and the same four directions of cr3 were obtained (Fig. 6d, e, f & g). Faults related to N-S and NW-SE cr3 did not show any relative chronology of striations between them and look like conjugate fault systems displaying dip-slip movements (Fig. 6d, e). The two other regimes, trending NE-SW and E-W, are also characterized by conjugate fault sets but seem also compatible with faults displaying oblique-slip kinematics. For the NE-SW trending extension, the faults a, f3 & X (Fig. 6f) may respectively have been initiated with a N-S, NW-SE and E-W direction of o3 before an hypothetical reactivation during a NE-SW extension. This has not been confirmed by relative chronologies of striations. On the contrary, two such relative chronologies (Fig. 6g, faults 5, s) were observed on fault planes related to the east to west trending extension suggesting its occurrence after the NE-SW regime. An oblique-slip movement on fault <$> (Fig. 6g) would also indicate the reactivation of a structure initiated in a conjugate system related to the north to south extension. Taking into account our chronologies and published ones (Vandycke & Bergerat 1992), we should admit the east to west trending direction of cr3 to have occurred twice, before and after the NE-SW
298
C.HIBSCH£rAL.
Fig. 5. Examples of deformations in chalk, (a) hydroplastic breccia along an oblique-slip normal fault (Etretat), (pencil length 13 cm); (b) unaffected contact between Kimmeridge clays and overlying Cretaceous units in the South Ferriby Quarry; (c) horizontal view of a listric fault on the Cap Blanc-Nez tidal flat (hammer length 13 cm); (d) vertical view of a listric fault on the Cap Blanc-Nez cliffs displaying orthogonal continuous fault plane with dip-slip striations; (e) radial emplacement of marcasite-filled veinlets on the tidal flat; (f) listric normal fault outlined by flint slice in the Droup St Basle Quarry; (g) detail of the flint slice with a releasing bend affected by small brittle rupture (contraction) after lithification; (h) ductile shear of a flint level due to normal faulting; (i) stereogram of poles of flint veins (slices) and radial stress tensor of associated syndiagenetic normal faults.
CHALK, TECTONICS VS. POLYGONAL FAULTING
299
Fig. 6. Stereograms for microtectonic analysis in the Paris Basin (sites located on Fig. 4). Contoured stereograms show the distribution of poles of veins. Legend for stereograms, refer to Fig. 2. Greek letters identify faults with relative chronologies referred in the text.
300
C.HIBSCHCTAL.
direction of extension. These apparent contradictions in the chronologies, the dispersion of the cr3 directions and the poor correlation with geodynamic data may also suggest that most of the faults result from a single radial extension. Faults may have been reactivated in various directions during the same deformation stage. The peculiar geometry of the faults and associated mineralizations are also critical in this respect and will enforce this interpretation as discussed below. Omey and Vittel fault zones (Champagne). This zone would have been influenced by the opening of the Upper Rhine Graben since the end of Eocene times (Coulon 1992). Hydraulic fracturing is commonly associated to the normal faulting with a significant hardening of the chalk and with concentrations of iron oxides (Richard et al. 1997). Field mapping of such bands of hardened chalk and hydraulic breccia shows widespread major NNE-SSW and subordinate NNW-SSE trends (Allouc et al 2000). Quarries around Chepy (Fig. 4) allowed the authors to define the following palaeostress succession (Coulon & Frizon de Lamotte 1988): (1) Eocene NNW-SSE to NNE-SSW Tyrenean' strike-slip tectonics prior to (2) Oligocene omni-directional extension associated with hydraulic fracturing followed by (3) 'Alpine' NW-SE trending strike-slip tectonics since Miocene times. Two directions of cr3 (ENE-WSW and NW-SE) have been deduced from the distribution of poles of calcite veins. The ENE-WSW direction is described as the minor one but would post-date the first NW-SE trending direction of extension (Coulon 1992). Despite this chronology, the authors considered both directions to belong to a single Oligocene omni-directional regime trending N77°E with a high R ratio (0.95). Such interpretation appears plausible since similar hydraulic fracturing can be encountered either along NNE-SSW and NNW-SSE striking fault zones (Allouc et al. 2000). The radial extension was obtained by computing all normal faults together, even including faults with (r-s) angles reaching 60° (see chapter methodology). The question now considered is whether if radial extension can be considered as a real tectonic regime or not. Most of these quarries are now closed, sloped or used for waste disposal. Only data coming from the Chepy Quarry allowed comparison with conclusions from Coulon & Frizon de Lamotte (1988): (1) normal faults associated with hardened chalk gave a coherent palaeostress tensor with moderate R. The calculated ENE-WSW trending cr3 is coherent with one family of calcite veins (Fig. 6i); (2) a second set of normal faults is compatible with the NW-SE direction of cr3 deduced from the distribution of calcite veins (Fig. 6j) and with also an average R; (3) a third palaeostress tensor was obtained with remaining faults and defined a NE-SW trending
extension with low R ratio (Fig. 6k). This last regime tends towards transtension (crl getting closer to cr2) since it also involves strike-slip faults compatible with a NW-SE trending contractional stage. Relative chronologies between three of these faults and calcite veinlets either striking NNW-SSE and NE-SW (compatibles with stress tensors i or j) show the post-dating of this 'Alpine' transcurrent regime. Finally, computing the normal faults all together define the palaeostress tensor (1) (Fig. 6) with R close to 1 and with a NE-SW direction of cr3 quite similar to previous results (Coulon & Frizon de Lamotte 1988). Such a radial stress tensor may also result from a grouping in the computing between a few normal faults resulting from compaction-related radial extension and others due to real tectonic stresses. The improbability of a radial extension of tectonic origin is also supported by the lack of radial distribution of poles of calcite veins. The deformation and associated hydraulic fracturing may result from real tectonic stresses with two distinct directions of cr3.
Evidence for polygonal faulting The recognition of truly omni-directional faulting within the chalk is at first sight enigmatic. However, omni-directional fault systems are now widely observed in fine-grained successions from seismic observations, where they are referred to as polygonal fault systems. The common factor for these fault systems is that they are developed in ultra finegrained sediments with high porosity and extremely low permeability. Most examples are provided by mud-dominated deposits, but recently polygonal fault systems have been documented from almost pure chalk successions in the Bay of Biscay, the NW Australian basins and the Scotian margin basins of Canada (Cartwright & Dewhurst 1998; Hansen etal 2000). Vertical offsets less than approximately 5 m are still not being resolved, leaving many faults and fractures in the sub-seismic domain. This is the case for most of the normal faults observed in the chalk during our survey. Such polygonal faulting is documented at a different scale and the abundance of such fractures with all consequences in terms of reservoir permeability is highlighted. At seismic-profile scales, one main argument for polygonal faulting is the position of the deformation in a restricted lithological unit. It consists of thousands of multi-directional, closely-spaced normal faults with throws of less than 100 m and lengths of 500-1000 m, organized into polygonal networks in plan view. Maximum fault displacement is commonly found in the middle of the tier, decreasing upward and downward towards the tier boundaries (Lonergan et al. 1998). At outcrop scale, a spectacu-
CHALK, TECTONICS VS. POLYGONAL FAULTING
lar example comes from the South Ferriby Quarry (Fig. 3, site 54 & Fig. 5b) where the chalk is affected by numerous normal faults (not visible at the scale of the picture) without displacing the contact between the Kimmeridge Clays and the few metres of overlying Early Cretaceous sands. The normal faults frequently die out downward within the chalk and rarely displace the sands and do not detach within the clays as might be expected. Basal termination is commonly achieved via listric-shaped geometry. This is also illustrated in the Champagne region with an example coming from the Droup St Basle Quarry (Fig. 4), where the listric fault trace is outlined by a flint slice (f on Fig. 5). Other 3D exposures of normal faults in the chalk are developed along the coast of the 'Cap Blanc-Nez' (Fig. 4) on the tidal flat or along the cliffs. The faults display hydroplastic slickensides associated with syntectonic crystallization of calcite and marcasite. On horizontal exposures, they can display strong curvatures suggesting listric faulting (Fig. 5c). On vertical exposures, one example (Fig. 5d) shows two orthogonal normal faults displaying dip-slip striations and smoothly joining together without any crosscutting. The lack of tilting suggests that these features correspond to sub-circular structures with vertical conic shapes rather than classical listric tectonic faults. The analogy with small conic structures surrounded by dip-slip hydroplastic striations and ranging in size from cm to a few decimetres of length and diameter is noticeable (Hibsch et al. 1993). The upward development of such larger conic features can easily account for outcrop-scale polygonal fault patterns.
Timing of radial extension during diagenetic processes Mineralization and early diagenetic processes One major issue to be resolved with the analysis of the fracture and normal fault sets in chalks is the timing of the deformation. Insights into the timing have come from new work on the mineralization and diagenetic processes associated with the brittle structures. This section describes this work based on several key localities. At the 'Cap Blanc-Nez' (Fig. 4), curved faults are outlined with syntectonic calcite, marcasite filling, or both together. Marcasite is commonly associated with epigenetic alteration of fossils or bioturbation but can also fill vertical veinlets. In the blue chalk, they are clearly unoxidized away from fracture pattern and are related to an early diagenetic history (suggesting a very poor matrix permeability not allowing pervasive oxidizing meteoric water diffusion). When associated with calcite, marcasite is fre-
301
quently oxidized but this may be due to recent meteoric water percolation along the faults. The marcasite sometimes appears to be moulding calcite crystal heads and thus would indicate an early calcite growth phase. The radial extension suspected by the microtectonic analysis is impressively marked out by radial marcasite veinlets such as on Figure 5e. The radial distribution of marcasite veinlets in the whole site is summarized on Figure 6h. It is also noteworthy that syntectonic calcite is only present along normal fault planes and was never encountered as independent vertical tension gashes. From other studies, calcite veins in chalk are more frequently associated with clear examples of tectonic faulting such as in the Champagne region (Coulon & Frizon de Lamotte 1988; Richard et al 1997) or in inverted UK basins like the Wessex Basin (Mimran 1977) or the Pickering Graben (Hibsch et al 1993). Calcite veining is supposed to occur during pressure-solution stages after the main consolidation processes due to the burial of the chalk (Jones et al 1984). The calcite related to the normal faults at the 'Cap Blanc-Nez' site seems to be intimately associated to the development of marcasite and thus, on the contrary, suggest an occurrence during early diagenetic processes. Another extremely important record of the diagenetic history is the flints. The development of flints along normal fault planes has been observed in several places in England (Mortimore 1983; Hibsch et al 1993), in Normandy (Kennedy & Juignet 1974) and in the Champagne region (Coulon 1992) (e.g. Droup St Basle Quarry - Fig. 4 & Fig. 5, f, g, h, i). In this quarry, flint veins (slices) follow listric normal fault planes (Fig. 5f) or vertical fractures on releasing bends related to these normal faults (Fig. 5g). On this last photograph, the flint has suffered post-lithification small brittle contractional horizontal displacement, whereas on Figure 5h, the normal faulting affecting the horizontal flint bed had a more ductile behaviour and occurred prior to the complete lithification of the flint (as shown by the stretching of the flint bed along the fault). The stereogram (i) (Fig. 5) plots the poles of the flint veins associated to a palaeostress tensor computed with the normal faults. The regime tends towards radial extension (high R value, and permutation between cr2 and cr3), but the stereogram shows a concentration of poles to the NW indicating a local main extensional strain towards the SE. The timing of siliceous precipitation and lithification of the flints is controversial. Overviews of these processes in Normandy are available (Kennedy & Juignet 1974; Juignet & Breton 1997). Initial siliceous concentration was favoured by higher porosity and permeability in chalk filling of burrows. Siliceous migration occurred quite early in the diagenetic processes but the lithification of the flints must
302
C.fflBSCHCTAL.
have been achieved several metres below the water/sediment interface since no brittle fracturing of the flints was found associated with the slump and debris flow features. In chalks from Etretat, horizontal shearing and brecciation of the flints followed this first development of flint levels (Fig. 7, a & b). Similar features were described in England (Mortimore 1983). It raises the question if it may correspond to a widespread phenomenon in chalks only clearly noticeable when affecting flint beds. In Normandy, this horizontal shearing is associated with chalk dikes and sills (Fig. 7a) which may be developed vertically over several metres. This points out the important role played by fluid overpressures during this shearing. When flint breccia is lacking, the horizontal shearing can be outlined by bands of anastomosed chalk sills. Both injected chalks and shear bands are harder than normal chalk, as confirmed by a clear loss of porosity (Fig. 7c). The amount of horizontal displacement, when discernible, seems quite low, in the maximum order of the metre scale. Along the tidal flat, good examples of parallel-inclined paramoudra flints outline the phenomena. This shearing and chalk injection has been either related to near surface seismic shaking or to heterogeneities of compaction caused by the underlying carbonate bank morphology (Juignet & Breton 1997). Since this chalk was buried under several tens of metres below the watersediment interface when chalk injection occurred (because of the delay in the lithification of the flints), the palaeoseismic interpretation appears doubtful. A secondary flint development occurred, frequently associated with fractures such has normal faults or en echelon tension gashes. It also developed horizontally as flint slices, which locally incorporate breccia of the previous flints. The en echelon system reveals horizontal movements and may branch downward either on flint beds or on flint slices. This points out the persistence of silica-saturated fluids long after first concretions in the burrows. The siliceous precipitation and deformation occurred repetitively during the burial of the chalk. This is summarized in Figure 7d where the flint slices may suffer contractional or extensional deformation after at least a first stage of lithification, but also with a third example where a first shear-related flint slice has been disrupted and overturned in a shear band outlined by two other bracketing flint slices. Even if the dating of such phenomena depend on the uncertainties of the velocity of flint precipitation and lithification, this must have be achieved during Late Cretaceous times before emersion and introduction of meteoric water conditions into the system. Contrasting with this early syndiagenetic polygonal faulting, a last stage of deformation was described in a previous section. It appears clearly younger since the fault breccia includes flint pieces and also debris of dolomitized hard-grounds but are never asso-
ciated with the growth of new flint. The associated extensional regime does not seem to be radial and was considered to have a tectonic origin. Normal faulting in chalk is frequently characterized by hydroplastic deformation features. Creep features are quite common in this material (Fig. 5a) and various features show that the chalk has behaved with a certain ductility (Jones et al. 1984). This is illustrated in Figure 7e, which shows twisted cylindrical features surrounded by hydro plastic striations. The ductile chalk can be envisaged as being squeezed along the fault plane.
Late diagenetic processes and geomorphological evidence Different from polygonal fault systems developed in clay, polygonal faults in chalk are subsequently strongly susceptible to post-consolidation pressuresolution. Dip-slip slickolites may be also encountered on faults related to the radial extension, indicating a deformation during pressure-solution stage, after the first consolidation stage due to the burial of the chalk. The following examples attempt to document and date these phenomena. In the Paris basin, vertical and lateral heterogeneities in seismic velocities, described as rapid and slow chalks, have been evidenced, just below the Cenozoic unconformity (Hanot & Thiry 1999). A rapid chalk is recognized at the base with a hardening possibly due to dolomitizations and slow chalks on top, with very bad seismic signals corresponding to altered chalk. Irregularities of the top chalk surface were correlated with synsedimentary thickness variations during the Cenozoic sedimentation, which could not be explained by tectonic fault control since the basal chalk contact is not displaced. They concluded that local dissolution processes affected the chalk below Cenozoic fluvial valleys. This phenomenon created a subsidence allowing continental deposits thickening. Migration of meteoric water in the chalk boosted the pressure-solution process and seems to have allowed the precipitation of dolomites in the deeper chalk. Thermohalyne convective cells may have promoted the fluid migration (Hanot & Thiry 1999). Such phenomena described for Cenozoic continental environments may be still active nowadays as suggested by Quaternary geomorphologic evidence. Numerous scarps called the 'rideaux' (lynchets) in northern France (few hundreds of metres length for a few metres high) were first related to pure anthropogenic agricultural features or dissolution of the chalk (Gosselet 1906). More recently, relations with active faults were proposed (Bracq & Delay 1997). Supporting this hypothesis, 'rideaux' can sometimes develop above underlying tectonic faults displacing
CHALK, TECTONICS VS. POLYGONAL FAULTING
303
Fig. 7. Examples of deformations in chalk, (a) Position of horizontal shear zones in the Etretat cliff section and associated multidirectional chalk injections; (b) example of horizontally brecciated flints (no clear sense of shearing identified); (c) porosity loss in the shear zone; (d) examples of deformation associated to flint development (contractional, extensional or within two horizontal shear zones); (e) twisted cylindrical creep features along a normal fault plane (*) showing thick marcasite and calcite mineralization ('Cap Blanc-Nez'); (f) shallow development of listric faults (line draw) due to gravitational forces (displacement towards the valley axis, 'Cap Blanc-Nez'); (g) quaternary age of the movement revealed by open releasing bend and fast meteoric water flow (springs and bleaching of the chalk along the fault trace).
304
C.fflBSCHCTAL.
geological units below the chalk. The example close to 'Surques' in the Boulonnais (Fig. 4), displays a rather equal vertical offset of about 10 metres on top of the basement and in the topographic profile (Brunin 2000). Given the lower resistivities in the chalk measured in the hanging-wall block, compaction/dissolution phenomena cannot be excluded as shown by the syncline geometry enhanced in the chalk at the footstep of the 'rideau'. These lynchets only appear on top of the outcropping chalk units but not on top of Jurassic limestones. The scattered spatial distribution and strikes of these scarps may argue in favour of a surficial faulting process instead of tectonic-controlled faulting. To account for such important geomorphic scarps, reaching sometimes several tens of metres and considering the susceptibility of the chalk to erosion, the neotectonic solution would imply high fault slip-rates not compatible with the low seismicity of the Paris basin. Such Quaternary faulting has been also evidenced in a north to south striking trench made across the Longueville anomaly (Fig. 4). First considered as evidence for neotectonic activity (Benoit & Grisoni 1995), reverse faults affecting Quaternary terrace deposits have been investigated in order to discriminate palaeoseismic and other processes (Baize et al 2002). The alluvial deposits, mainly composed of Jurassic limestone pebbles, are preserved in pockets surrounded by centripetal reverse fault zones. Evidence for geliturbation near the top of the trench indicates periglacial impact in the area. The occurrence of mechanisms such as solifluction of the underlying altered chalk may have be an alternative explanation for the faulting as suggested by contortions and metre-scale pocket-shape features at the top of the altered chalk. During ice age periods, deformation of the chalk can be exaggerated with ice load and can develop glacier ice-thrust features (Banham & Ranson 1965), but glaciers are not supposed to have reached this area. The fault geometry does not suggest slope-driven lateral spreading but rather an underlying loss of matter (withdrawing). Undergoing dissolution and creep in the chalk seem to have developed in association with the alluvial environment and underlying water flows, as explained by Hanot & Thiry (1999) for older series. Other evidence for recent superficial normal faulting can be observed along the coast. Recent low angle normal faults are clearly developed with centripetal displacement on both sides of small valleys reaching the sea. This can be observed in Normandy (Juignet & Breton 1997) where these movements allowed an important increase in permeability in the chalk aquifer or at the 'Cap Blanc-Nez' with an intense development of low angle normal faults (Fig. If) showing displacements towards the valley axis. Open fractures developed in releasing bends of the low-angle normal faults allowed rapid water flow
(Fig. 7g). Such recent movements related to geomorphologic structures may also explain sand fillings in open fractures parallel to the main NW-SE striking drainage pattern (Vandycke & Bergerat 1992) without invoking neotectonic stresses.
Evidence for deformation mechanism from textures and porosity Faulting and folding affecting the chalk were related to important variations of geomechanic properties (Mimran 1977; Clayton & Matthews 1987). These variations can affect its porosity and also its strength. Modifications can affect the overall chalk unit, but frequently display lateral and vertical variations. Such perturbations can be localized around the fault planes (Gaviglio et al 1999). In order to complete the overview of the physical modifications induced by fault movements in chalks, we now focus on the horizontal shear bands noticed in Normandy (Fig. 7a). The 'deformed' layer (4 cm thick) was interbedded with two layers of undeformed rock (upper bed: 8.5 cm thick, and lower bed: 12.5 cm thick). Porosity was determined from weight measurements (before and after saturation with water) on plugs about 8 cm3 in volume. Five vertical profiles (i.e. roughly normal to the bedding) are shown in Figure 7c. Plugs containing flint pieces were not considered in the measurements. The porosity in the upper and the lower beds ranges between 30-35%; in the deformed layer it drops down to 22%. The variation in porosity is very sharp and coincides with boundary layers. These characteristics are very similar to those already observed along fault planes in the Campanian chalk of the Mons basin (Gaviglio et al 1997). In most cases no microscopic evidence (e.g. SEM observations) of transformation of the material can be found beyond 50 mm from the fault plane. However, physical measurements (elastic waves velocity or capillary rises), in addition to porosity measurements, display evidence of modifications of the arrangement of the material up to 150 mm (Gaviglio et al 1999). The basic mechanisms for this transformation can be summarized as follows: (1) shearing provides an opportunity for fluid circulation; (2) packing of the grains takes place through frictional sliding and pore-collapse; (3) as a consequence of fluid escape, pressure solution can start because the effective stresses at the grain contacts increase. Fluids are essential because it is the only factor able to produce such a change in texture leading to a homogeneous arrangement. For the sample coming from Etretat, the sharp contact between the intact and the deformed materials is consistent with the formation of a shearing zone. The development of horizontal flint beds along such zone and the branching of flint-filled en echelon
CHALK, TECTONICS VS. POLYGONAL FAULTING
normal faults indicate horizontal shearing during early diagenetic stages.
(2)
305
compactional flattening, in which the same fault pattern results from a three-dimensional contraction strain due to the intrinsically weak state of chalky sediments (c.f. Goulty 2001).
Discussion Except for few sites where extension of tectonic origin seems to be documented, evidence has been presented in the previous sections to suggest that instead of interpreting small, multi-directional normal fault systems developed within the chalk in terms of several superposed phases of tectonic deformation, it is possible to re-interpret these structures as essentially the product of a single phase of deformation. This deformation is layer-bound, since the structures are confined to within chalk sequences. In most cases, there is no systematic order of fault development as revealed by the relative chronology analysis. A limited study of the textures and porosity of chalks in fracture zones shows that the faulting is associated with a strong decrease in porosity in a zone fringing the striated planes of the normal faults, as a consequence of shearing associated with pressure solution and cementation. From the important observations concerning the faults and their relationships to the early diagenetic calcite, marcasite and flint formation, it can also be argued that the normal faulting occurred even earlier during the compaction and burial process. Although regional tectonic setting might have been important in determining specific slip senses and biases in orientations, it is evident that omni-directional fault systems are found in a wide variety of tectonic settings. This implies that the driving mechanism does not depend on regional tectonic stresses, but can be locally perturbed by them. The strong evidence for polygonal faulting restricted to chalk layers raises the important question as to the likely deformation mechanism. The polygonal fault systems evidenced at seismic scale share many of the characteristic features of the outcrop-scale normal faults in chalk described in this paper. They only differ in being considerably larger, but that is mainly a function of seismic resolution and perhaps too, the limitation of the outcrop dimensions. Using polygonal fault systems as a mechanistic analogue thus seems entirely justified. On this basis, it could be argued that two mechanisms could explain the origin of the polygonal faults as described in this contribution: (1)
syneresis, in which the high porosity chalk would have contracted whilst in a gel-like condition, and the strain of layer-bound contraction would have been accommodated by normal faulting with an almost homogenous bulk strain value (c.f. Cartwright & Dewhurst 1998);
From the evidence presented in this paper, we are not able to argue for either one of these two competing mechanisms. The measured fault plane dips cannot be decompacted with any certainty to their original inclinations at the time when deformation occurred and hence the sediment strengths at the time for deformation cannot be constrained. This critical aspect of the Goulty model cannot be evaluated without this constraint. By the same token, syneresis is difficult to evaluate in chalks, where there is not the same degree of electrochemical activity as in other fine-grained sediments (e.g. smectitic clays). It is interesting to note that chalks with polygonal fault systems contain minor amounts of smectite, and experimental work on syneresis (White 1961) produced syneresis fractures in chalk-like sediments when seeded with minor traces (< 10%) of smectite. Much further work is required to rigorously evaluate these two possible mechanisms. Other arguments presented in this paper suggest that in addition to the primary development of normal faults during mechanical compaction stage, a later reactivation of these structures occurred during uplift and exposure to meteoric influences and near-surf ace weathering processes. After the early stage of compaction, the chalk possibly underwent a renewal of the polygonal faulting during the subsequently pressure-solution chemical compaction stage. On seismic sections, polygonal fault systems in the chalk are restricted between two undeformed stratigraphical boundaries. When the polygonal fault system remained under marine conditions during the subsequent geological history it has not been reactivated. On the contrary, relative vertical displacements along normal faults may occur again in continental environments, as evidenced on seismic profiles in the Paris Basin where non-tectonic synsedimentary deformation due to underlying dissolution of the chalk occurred under fluvial conditions during early Cenozoic times (Hanot & Thiry 1999). Reactivation of normal faults due to pervasive (not necessary karstic) dissolution of the chalk in the catchment area around alluvial plains may explain present-day widespread distribution of lynchets ('rideaux') on top of the chalk. In between these two extremes in age (early faulting during first stages of consolidation and very recent faulting due to diffused ground water dissolution), the introduction of meteoric (unsaturated) water irr chalk aquifers appears as a strong boost factor for the increase of pressure-solution phenomena and could promote reactivation of the normal faulting at depth. The mechanical consolidation
306
C.HIBSCHErAZ,
stages reduces the porosity and promotes the pressure solution at grain to grain scale. Associated early cementation may reduce the permeability of the chalk and impede the overpressure dissipation, which in turn can restrain the mechanical consolidation development (Jones et al 1984). This can generate undercompacted chalks. Faults during the compaction are fundamental since permeability is very low in the chalk and thus, fluids overpressures could only dissipate along the fractures. The inherited early syndiagenetic polygonal fault system is a path promoting meteoric water invasion. The meteoric origin of unsaturated fluids is revealed by oxygen isotopic ratio and Sr/Ca contents in cements and syntectonic calcite (Mimran 1977; Richard et al 1997). The travel of the water through other sedimentary horizons may be outlined by iron oxides concentrations close to the faults (Richard et al 1997; Gaviglio etal 1999). The increase of fluid pressure promotes pressure-solution at the grainfluid interface and thus saturates the fluid in carbonates. On the contrary, the lowering of the fluid pressure along the fault system promotes crystallization and thus reduces the permeability which in turn causes the fluid pressure to increase again (Mimran 1977; Richard et al 1997). A second step of deformation may produce important volume loss due to both porosity reduction and carbonate migration. This reduction can reach about 40% in chalks from Dorset according to Mimran (1977), but this loss can locally reach 95% with high carbonate migrations during a third step of deformation. In Dorset, these important modifications in geomechanical properties were related to Cenozoic compressional stages and not to an increase of normal faulting. Considering that polygonal faulting results in a pure flattening finite strain which cannot be explained by regional stretching, the amount of horizontal extension deduced from the normal fault geometry and displacements, should be compensated by horizontal contraction. Volume loss of rock material due to pressure solution may be involved in this compensation. In a closed hydrogeological system, the estimates of horizontal volume loss around normal faults of the Mons Basin (Gaviglio et al 1999) are about ten times lower than the horizontal contraction deduced from the fault movements. On the contrary, when considering an open system affected by higher dissolution processes and carbonate migrations, as deduced from the observation of Belemnita paralleled to the fault planes, coherent values of horizontal contraction are obtained (Gaviglio et al 1999). This confirms the carbonate migration along the fault planes. This horizontal contraction, compensating extension, is established in the Mons basin, a site where extension has been considered to have a tectonic origin (Gaviglio et al 1999). Pressure threshold related to the collapse of
samples of very soft chalks from England were compared to the expected natural maximum vertical pressure sustained by these chalks during their burial history (Clayton & Matthews 1987). Although considering the values to be in the same scale order, the natural maximum overburden pressure appears systematically higher than the pressure measured under laboratory conditions and raises the question why these soft chalks have not collapsed under natural conditions? It can perhaps be envisaged that isolated pockets of undercompacted soft chalks are protected within an intricate network of polygonal faults and that the faults form zones with a higher mechanical strength conditioned by the types of physical and chemical modifications around fault planes described by Gaviglio etal (1999). The evolution of compaction and associated diagenetic events could be significantly different depending on the evolution of the vertical load. Transition from the purely mechanical consolidation stage to the pressure solution stage seems quite progressive and both may occur during early syndiagenetic polygonal faulting, but at a limited extent with carbonate- (and sometimes silica-) saturated marine waters. Calcite veins in chalk point out the pressuresolution stage and are mainly encountered in areas affected by strong tectonics at later diagenetic stages (Mimran 1977; Coulon & Frizon de Lamotte 1988; Hibsch et al 1993; Richard et al 1997) except perhaps for the calcite associated to marcasite in the 'Cap Blanc-Nez' polygonal fault system suggesting earlier pressure-solution phenomena associated to syntectonic crystallization.
Conclusions Normal fault sets observed in chalks, exposed in widely separated localities in the UK and France, are characterized by a wide range of strikes at any one locality. They are developed entirely within chalk successions and do not seem to interconnect to deeper or shallower structures as shown at outcrop or seismic-scale. These structures can be interpreted in two different ways invoking: (1) complex polyphase deformational histories involving contrasting stress states; or (2) a single deformational environment in which the faults develop to accommodate compactional strains. Our general overview points out a widespread radial extensional regime expressed in the chalk. This contrasts with previous documentation of the radial regime by other authors in both countries where this pattern was interpreted as the result of peculiar local tectonic stress fields. Strong analogy can be drawn between polygonal fault systems developed in chalk successions, as revealed on seismic profiles and the smaller scale deformation described in this paper. Petrographic
CHALK, TECTONICS VS. POLYGONAL FAULTING data support the compaction-related interpretation. In particular, the association of calcite and marcasite mineralization with fracture surfaces and fault zones, and textural observations relating flint occurrence to early fault formation point towards fault propagation at a very early stage of burial and compaction of chalky sediments. The origin of polygonal fault systems is explained by two currently debated hypotheses: syneresis favoured by smectite content in the chalk and threedimensional contraction strain due to the intrinsically weak state of the chalk. In the Paris Basin, evidence from seismic profiles, geomorphologic scarps and Quaternary fault movements suggests extensional reactivation of part of this inherited fault pattern from Cenozoic times to present-day. During the uplift of the basins, considering the very low permeability of the chalk, the meteoric water invaded the chalk preferentially along fault pathways. Pressure solution phenomena were much more intense around the fault planes and promoted this reactivation. Horizontal contraction due to this pure flattening finite strain was compensated by volume loss due to significant carbonate dissolution and migration. Finally, all these data emphasize the difficulty in analysing tectonic normal faults in the chalk due to interference with non-tectonic polygonal faulting. Considering this new hypothesis, more detailed studies should be carried out to compare the specific behaviour of the different chalk units. This first overview suggests polygonal faulting to increase with the amount of smectite and to be blocked at an early diagenetic stage when lithification of large connected network of flints occurred and strengthened the chalk block.
References ALLOUC, J., LE Roux, J., BATKOWSKI, D., BOURDELLON, C., CATILLON, J., CAUSERO, L., MENILLET, R, MORFAUX, P. & RAVAUX, J.-P 2000. Notice explicative, feuille Suippes (159). Orleans, BRGM, Carte geologique de la France, (1/50 000). BADLEY, M.E., PRICE, J.D. & BACKSHALL, L.C. 1989. Inversion, reactivated faults and related structures: seismic examples from the southern North Sea. In: COOPER, M.A. & WILLIAMS, G.D. (eds) Inversion Tectonics. Geological Society, London, Special Publications, 44,201-219. BAIZE, S., GUSHING, E. M., LEMEILLE, R, GRANIER, T., GRELLET, B., CARBON, D., COMBES, P. & HIBSCH, C. 2002. Sismotectonique de la France metropolitaine volume 3: Inventaire des indices de rupture affectant le Quaternaire en relation avec les grandes structures connues en France metropolitaine et dans les regions limitrophes. Memoires de la Societe geologique de France, Hous serie, 175,142pp. BANHAM, PH. & RANSON, C.E. 1965. Structural study of
307
the contorted drift and disturbed chalk at Weybourne, North Norfolk. Geological Magazine, 102,165-174. BENARD, R, DE CHARPAL, O., MASCLE, A. & TREMOLIERES, P. 1985. Mise en evidence d'une phase de serrage EW au Cretace inferieur en Europe de 1'ouest. CompteRendus de I'Academie des sciences, Paris, 300, 765-768. BENO!T, P. & GRISONI, J.-M. 1995. Tectonique rissienne et fini-wiirmienne / holocene dans la basse terrasse alluviale (5-10 M du Riss) de la riviere Aube (Longueville-sur-Aube), dans le sud-est du B as sin parisien - France. Bulletin d'Information des Geologues du Bassin de Paris, 32,7—11. BRACQ, P. & DELAY, R 1997. Transmissivity and morphological features in a chalk aquifer: a geostatistical approach of their relation. Journal of Hydrology, 191, 139-160. BRUNIN, A.-S. 2000. Comportement hydrodynamique de 1'aquifere crayeux sur le pourtour de la boutonniere du Boulonnais. [PhD thesis], Universite du Littoral, Calais. CAREY, E. 1979. Recherche de directions principales de contraintes associees au jeu d'une population de failles. Revue de Geologie Dynamique et de Geographic Physique, 21,57-66. CARTWRIGHT, J.A. & DEWHURST, D.N. 1998. Layer-bound compaction faults in fine-grained sediments. Geological Society of America Bulletin, 110, 1242-1257. CHADWICK, R.A., LIVERMORE, R.A. & PENN, LE. 1989. Continental extension in southern Britain and surrounding areas and its relationship to the opening of the North Atlantic ocean. Memoire of the American Association of Petroleum Geologists, 46,411-424. CHAPMAN, T.J. 1989. The Permian to Cretaceous structural evolution of the Western Approaches Basin (Melville sub-basin), U.K. In: A, C.M. & WILLIAMS, G.D. (eds) Inversion Tectonics, Geological Society, London, Special Publications, 44,177-200. CLAYTON, C.R.I. & MATTHEWS, M.C. 1987. Deformation, diagenesis and the mechanical behaviour of chalk. In: JONES, M.E..& PRESTON,. M.R (eds) Deformation of Sediments and Sedimentary Rocks, Geological Society, London, Special Publications, 29,55-62. COULON, M. 1992. La distension oligocene dans le nord-est du bassin de Paris (perturbation des directions d'extension et distribution des stylolites). Bulletin de la Societe Geologique de France, 5,531-540. COULON, M. & FRIZON DE LAMOTTE, D. 1988. Les craies eclatees du secteur d'Omey (Marne, France): le resultat d'une brechification par fracturation hydraulique en contexte extensif. Bulletin de la Societe Geologique de France, 8,177-185. ENGLAND, R.W. 1988. The early Tertiary stress regime in NW Britain: evidence form the patterns of volcanic activity . In: MORTON, A.C. & PARSON, L.M. (eds) Early Tertiary Volcanism and the Opening of the NE Atlantic, Geological Society, London, Special Publications, 39, 381-389. GAVIGLIO, P., ADLER, P., THOVERT, J.-F, VANDYCKE, S., BERGERAT, R, BEKRI, S. & LESTIDEAU, R. 1997. Grainscale microstructures and physical properties of faulted chalk. Bulletin de la Societe Geologique de France, 168,727-739.
308
C.HIBSCHErAL.
GAVIGLIO, P., VANDYCKE, S., SCHROEDER, C., COULON, M., BERGERAT, F, DUBOIS, C. & POINTEAU, 1.1999. Matrix strains along normal fault planes in the Campanian White Chalk of Belgium: structural consequences. Tectonophysics, 309,41-56. GLENNIE, K.W. & BOEGNER, P.L.E. 1981. Sole Pit inversion tectonics. Petroleum Geology of the Continental Shelf of North-West Europe. Institute of Petroleum, London, 110-120. GOSSELET, MJ. 1906. Observations sur les creuses de 1'Artois et de la Picardie et reflexions sur 1'importance de la dissolution souterraine de la craie. Annales de la Societe geologique duNord, 35, 237. GOULTY, N.R. 2001. Mechanics of polygonal faulting in fine-grained sediments. First Break, 19,69-73. HANOT, F. & THIRY, M. 1999. Anomalies sismiques dans la craie et deformations superposees dans les formations tertiaires du sud-est du bassin de Paris. Bulletin de la Societe Geologique de France, 170, 915-926. HANSEN, D.M., LYKKE-ANDERSEN, H., SHIMELD, J.W. & WILLIAMSON, M A. 2000. Layer-bound faulting on the Scotian Shelf, offshore Nova Scotia, Canada, 24. Nordiske Geologiske Vinterm0de, Trondhjem, Norge. Geonytt, 1, 74. HIBSCH, C., CUSHING, E.M., CABRERA, J., MERCDBR, J.L., PRASIL, P. & JARRIGE, J.J. 1993. Paleostress evolution in Great Britain from Permian to Cenozoic: a microtectonic approach to the geodynamic evolution of the southern UK basins. Bulletin Centres de Recherches Exploration-Production Elf-Aquitaine, 17, 303-330. JONES, M.E., BEDFORD, J. & CLAYTON, C. 1984. On natural deformation mechanisms in the Chalk. Journal of the Geological Society, London, 141, 675-683. JUIGNET, P. 1971. Modalites du controle de la sedimentation sur la marge armoricaine du bassin de Paris a I'Aptien-Albien-Cenomanien. Bulletin du B.R.G.M., 1,113-126. JUIGNET, P. & BRETON, G. 1992. Mid-Cretaceous sequence stratigraphy and sedimentary cyclicity in the western Paris basin. Palaeogeography, Palaeoclimatology, Palaeoecology, 91,197-218. JUIGNET, P. & BRETON, G. 1997. Breches de silex, diagenese et deformation dans les craies de HauteNormandie (France). Annales de la Societe geologique du Nord, 5,227-240. KENNEDY, WJ. & JUIGNET, P. 1974. Carbonate banks and slump beds in the Upper Cretaceous (Upper TuronianSantonian) of Haute Normandie, France. Sedimentology, 21,1-42. KENT, P.E. 1980. Subsidence and uplift in East Yorkshire and Lincolshire: a double inversion. Proceedings of the Yorkshire Geological Society, 42, 505-524. KIRBY, G.A. & SWALLOW, P. W. 1987. Tectonism and sedimentation in the Flamborough Head region of North-
East England. Proceedings of the Yorkshire Geological Society, 46, 301-309. LONERGAN, L., CARTWRIGHT, J. & JOLLY, R. 1998. The geometry of polygonal fault systems in Tertiary mudrocks of the North Sea. Journal of Structural Geology, 20,529-548. MIMRAN, Y. 1977. Chalk deformation and large-scale migration of calcium carbonate. Sedimentology, 24, 333-360. MORTIMORE, R.N. 1983. The stratigraphy and the sedimentation of the Turonian-Campanian in the Southern province of England. Zitteliana, 10, 27-41. MORTIMORE, R.N. & POMEROL, B. 1991. Upper Cretaceous tectonic disruptions in a placid chalk sequence in the Anglo-Paris Basin. Journal of the Geological Society, London, 148, 391-404. PAIRIS, J.-L., MILLIEN, A. & SIRIEYS, P. 1997. Fracturation et deformation de la craie. Retour d'experience de la partie frangaise du Trans-Manche. Annales de la Societe geologique du Nord, 5, 257-266. PEACOCK, D.C.P. & SANDERSON, D.J. 1994. Strain and scaling of faults in the chalk at Flamborough Head, U.K. Journal of Structural Geology, 16,97-107. POMEROL, C. 1977. Dynamique comparee de trois bassins epicontinentaux: mer du Nord, Manche et bassin de Paris. Bulletin Centres de Recherches ExplorationProduction Elf-Aquitaine, 1,233-256. RICHARD, J., COULON, M., GAVIGLIO, P. & RAMSEYER, K. 1997. L'hydrofracturation : une deformation tectonique a haut potentiel diagenetique. Exemple des craies hydrofracturees de la region d'Omey (Bassin de Paris, France). Compte-Rendus de I'Academic des sciences, Paris, 325, 359-366. STARMER, I.C. 1995. Deformation of the Upper Cretaceous chalk at Selwicks Bay, Flamborough Head, Yorkshire: its significance in the structural evolution of northeast England and the North Sea Basin. Proceedings of the Yorkshire Geological Society, 50, 213-228. VAN HOORN, B. 1987. Structural evolution, timing and tectonic style of the Sole Pit inversion. Tectonophysics, 137. VANDYCKE, S. & BERGERAT, F. 1992. Tectonique de failles et paleo-contraintes dans les formations cretacees du Boulonnais (France). Implications geodynamiques. Bulletin de la Societe Geologique de France, 163, 553-560. WHITE, W.A. 1961. Colloidal phenomena in the sedimentation of argillaceous rocks. Journal of Sedimentary Petrology, 31,560-570. ZIEGLER, P.A. 1987. Late Cretaceous and Cenozoic infraplate compressional deformations in the Alpine foreland - a geodynamic model. Tectonophysics, 137, 389^20. ZIEGLER, P. A. 1990. Geological Atlas of Western and Central Europe. Geological Society, London, 239 p.
The origin and development of joints in the Boom Clay Formation (Rupelian) in Belgium J. MERTENS1'2, N. VANDENBERGHE2, L. WOUTERS1 & M. SINTUBIN2 1
ONDRAF/NIRAS, Belgian Agency for Radioactive Waste and Enriched Fissile Materials, Kunstlaan 14, BE-1210 Brussels, Belgium (e-mail: [email protected]) Structural Geology & Tectonics Group, K.U.Leuven, Redingenstraat 16, BE3000Leuven Belgium Abstract: A system of natural, vertical and mutual perpendicular joints is described in the clay pits of the Rupelian Boom Clay in the Antwerp area of North Belgium. Joints are the dominant discontinuity surfaces in the clay outcrops. Neither the stress evolution of the clay during burial and uplift nor the regional tectonic history can explain the tensional jointing. It is suggested that the negative horizontal stresses required for the joint formation in clays are caused by shrinkage of the clay when the formation was located near the surface. The origin of the loss of pore fluid in a several tens of metres thick clay layers remains unknown. The joints in the Boom Clay are a scarce field example of the possibility of lateral contraction of a clay layer, without involving tectonics or burial/uplift as an origin. When such a jointed clay layer is buried again, the presence of the vertical joints might offer pathways for fluid migration through a relatively impermeable layer. A general relationship between a maximal depth beneath which no tensional joints can occur and the cohesion of the clay has been derived. In the case of the Boom Clay this limiting depth is around 40-50 m. The time of the joint formation in the outcrop area is most probably late Oligocene/early Miocene. The burial history of the clay at a particular location can be used as a predictive tool for the presence or absence of tensional joints.
The Boom Clay formation in north Belgium Geology and geotechnical applications. The Boom Clay is a marine Oligocene deposit of several tens of metres thickness, which is of considerable stratigraphic and geotechnical significance in North Belgium (Fig. 1). It has been studied on many occasions for different geotechnical and applied geological purposes and projects. Perhaps the most visible application has been the exploitation of the clay along the Rupel and Scheldt rivers since the 13th century for brick making and at times for roof tile fabrication. In the beginning of the 20th century, ships were loading bricks along the Rupel for the English and the USA markets. The properties of the Boom clay for the coarse ceramic industry have been summarized by Decleer etal.(1983) and for the expanded clay industry by Decleer & Viaene (1993). The clay has also been used for the underground storage of gas and an experimental site was tested for the shallow underground storage of liquefied natural gas (De Sloovere 1983). Extensive geotechnical testing of the clay has been done in preparation for tunnelling for road and subway construction under the Rupel in Boom, the Scheldt in Antwerp and the Westerscheldt in the Netherlands (De Beer 1967,1971) and for the now abandoned project for the construction of a storm surge on the Scheldt north of Antwerp (Schittekat et al 1983). The main world-class geo-
technical project in the Boom clay is, however, the construction of an underground laboratory with testing facilities at more than 200 m depth in Mol in North Belgium (location in Fig. 1), with a view to a possible high-level radioactive waste disposal in the Boom clay of North Belgium (Neerdael 1996). The significance of the clay for the present volume is the nature of the jointing, which we propose is due to near-surface shrinkage and which has some similarities with strata-bound polygonal faulting. The Boom Clay Formation is the unit stratotype of the Lower Oligocene Rupelian stage and a well-published example of cyclostratigraphy (Vandenberghe 1978; Van Echelpoel & Vandenberghe 1987; Van Echelpoel & Weedon 1990; Vandenberghe et al 1997; Vandenberghe et al 1998). The clay was deposited during the Rupelian in the southern part of the North Sea basin. It is known from outcrops and from the shallow subsurface as a continuous thick clay layer in Belgium, the Netherlands and Germany. It was deposited at a water depth between 50 and 100 m. Periodically, however, the depositional characteristics changed slightly as is documented by a typical and laterally very continuous sub-horizontal layering, each bed being several tens of cm thick. The bedding shows up by changes in the relative proportion of the different lithological components of the clay: the silt-sized particles, mainly quartz and less feldspar and also vegetal organic matter grains, and the fine-grained clay mineral and carbonate particles. The beds that
Fig. 1. Location of the Boom Clay Formation and the studied clay pits.
appear in the field are alternations of silty and very fine clay with additional black organic rich layers and pale grey carbonate rich horizons. The clay mineralogy of the Boom clay is a mixture dominated by illite and irregular illite-smectite interlayers with minor quantities of kaolinite, smectite and degraded chlorite. Early diagenetic pyrite is abundant throughout the deposit, generally in the form of lithified bioturbation traces. Early diagenesis also transformed the originally marly horizons in septaria beds, which are so characteristic that in Germany the clay is known as the Septarienton. The clay is described as stiff, meaning that it is hardly mouldable by the fingers and has an undrained shear strength between 75 and 160 kPa and a wet density of about 2 ton/m3 (Attewell & Farmer 1976; Institution of Civil Engineers 1976). These values fit the observed values in the Boom clay. Table 1 shows some average characteristic values of the Boom Clay. The burial and uplift history of the Boom clay makes it an overconsolidated clay where it occurs near the surface. Deep in the subsurface of North Belgium it was overconsolidated in the past but at present it is buried deeper than ever before. Because of the many geotechnical applications in the clay, the occurrence of discontinuities in the regular bedding structure of the clay is of importance. In particular, understanding the genesis of these discontinuities is crucial in predicting their importance for new projects.
Table 1. Some characteristic properties of the Boom Clay Natural water content w Unit weight of dry soil yd Unit weight of soil y Unit weight of solid particles -ys Plasticity index IP % particles < 20 (x % particles < 2 JJL Poisson ratio v Youngs modulus E
28 14.9 19.3 26.5 40 80 54 0.4 300
% kN/m3 kN/m3 kN/m3 % % MPa
Overview of the deformational structures, fissures and jointing in the Boom clay The Rupelian Boom clay in the studied area is part of the Upper Cretaceous and Cenozoic sedimentary cover overlying the northern flank of the stable Palaeozoic London-Brabant Massif and which is merging into the North Sea sedimentary series to the north. Since the Oligocene, the area has undergone only moderate vertical movements and has been outside the direct influence of the Lower Rhine Graben activity to the NE. A review of neotectonic data for NW Europe suggests that, except for al/o-2 permutations in the NW-SE direction, with the cr3 always in the NE-SW direction, the stress field has remained constant from the Neogene to the present (Bergerat & Vandycke 1994). It is, however, known that even outside the Lower Rhine graben influence,
JOINTS IN THE BOOM CLAY FORMATION
311
Fig. 2. Faults in the clay pit near Kruibeke.
slight vertical movements have reactivitated some of the deeper basement faults in Cenozoic times affecting the Cenozoic cover (see Sintubin et al 2000). In that stable area, only one zone with faults possibly corresponding to such reworking in the Boom Clay has been observed. It occurs in the NW corner of the Kruibeke clay pit (location in Fig. 1) as a zone with at least four normal faults, striking between 45° and 62° west with northwards dips between 42° and 57° and with small offsets in the order of half a metre (Figs 2,3). Another very localized deformation in the Boom clay is the occurrence of diapiric structures underlying the present Scheldt river and only there, detected by high resolution seismic surveys in the river (Wartel 1980; Henriet 1992; Henriet et al 1986). Laga (1966) has already described the diapric deformation as a curiosum on the Scheldt Kennedy site in Antwerp. The exclusive relation of the diapirs to the river valley strongly suggests that the river valley erosion diminished the vertical stress in the Boom clay to the point that the horizontal stresses in the clay inherited from former deeper burial could squeeze the clay upwards. The uplifted clay reaching the riverbed is systematically eroded by the river current action. More generally occurring in the Boom clay, however, are discontinuity surfaces of different size and nature. The most apparent discontinuity in the clay is a large scale jointing system that can be observed in the clay pits of the outcrop area (Fig. 4). It has already been featured in a photograph by De Beer (1971). Related to this type of joint are more irregular, smaller, dipping joints with more curved and rough surfaces (Fig. 5), commonly with a plumose structure and occasionally slickensided. Both types of joints undoubtedly correspond to the fissures described by many authors in similar clay deposits (a.o. Fookes 1965; Skempton et al 1969, Burland et al 1977; Price & Cosgrove 1990) and generally related to the overconsolidation of the clays (see discussions in Attewell & Farmer 1976
Fig. 3. Fault in a corner of the clay pit near Kruibeke. Flat appearance is due to the orientation of the excavation wall.
and Bell 1981). Another common type of discontinuity are the smaller scale, smooth, shining and striated surfaces (Fig. 6). They measure some cm2 to dm2 and may or may not be related to the larger jointing system. They are also common in many other Cenozoic clays of the same area in outcrops and in boreholes. This paper will deal exclusively with the study of the jointing in the clay as these structures are an illustration of the possibility of lateral contraction of a clay layer, not related to tectonics or burial/uplift. The joints might play an important role in fluid migration through this relatively impermeable clay when it is buried again. It is also known that joints have a profound influence on the geotechnical behaviour of clays in many excavation works.
Observation of the jointing in the clay pits Field description The joint surfaces are near vertical and have crosssections with the excavation walls ranging from two to several metres (often surfaces of more than 50 m2 were observed). The spacing between successive parallel joints varies between 0.5 m and several
312
J.MERTENSCTAL.
Fig. 5. Smaller, more irregular joints. Clay pit near Terhagen.
Fig. 4. Joints in the Boom Clay. Clay pit near Rumst.
metres. The joint surface is rather smooth and occasional precipitations of limonite and dispersed gypsum crystals (pyrite oxidation) occur. Neither displacement along the joints nor slickensides were observed. Two almost perpendicular main directions of jointing can be observed. The density of jointing does not change with the depth of excavation, in general about 30 m. Slight curving of the joint surfaces can occur and sometimes small steps are observed where lithologically different layers are crossed by the joints. Related to this type of joint is a secondary denser network of smaller, more irregular dipping joints with curved and rougher surfaces (Fig. 3), often with a plumose structure and occasionally slickensided. The spacing between these joints is less than 0.5 m. There also exist two dominant, almost perpendicular directions amongst these less regular joints, broadly similar to the orientations of the large joints but with a broader spread in orientation. Orientation of the joints in the outcrop area Figure 7 depicts the measured orientations of the large and smaller joints in the different clay pits in the outcrop area. On each of the orientation rose diagrams, the orientation of the clay excavation face is
Fig. 6. Small surfaces with striations. Clay pit near Steendorp.
also given. It can be concluded that the joint orientations are independent of the excavation practice. On a regional scale (Fig. 8), the joint orientations seem consistent, suggesting that the regional stress state might be involved. Fracturation and jointing in septaria In a few cases, it was observed that the large vertical joints have influenced the outermost part of septaria nodules (Fig. 9), implying that this part of the concretion was still soft at that time. It was, however, also observed that the joints crossing the central part of the nodule follow the septae, implying that the inside was already lithified when the joints formed. Septarian fragments with a fully lithified outer shell were found to be reworked at the base of the overlying Miocene Edegem sands.
JOINTS IN THE BOOM CLAY FORMATION
313
Fig. 7. Orientation of large (>2 m cross-section with excavation wall) and smaller joints (<2 m cross-section with the excavation wall) in the clay pits. Orientation of the excavation wall is drawn in grey, (a) Large joints in Terhagen. (b) Smaller joints in Terhagen. (c) Smaller joints in Niel-Schelle (only few large joints were observed), (d) Large joints in Rumst. (e) Smaller joints in Rumst. (f) Smaller joints in Sint-Niklaas (only few large joints were observed), (g) Large joints in Steendorp. (h) Smaller joints in Steendorp. (i) Large joints in Kruibeke. (j) Smaller joints in Kruibeke.
Large slickensided surfaces (larger than 10 cm2) are also commonly observed on large fragments of septarian nodules loosely occurring on the floor of the clay pits (Fig. 10). It appears that these slickensides only occur on one surface of the septaria and under
angles lower than 20°, dipping towards the outside of the concretion. It is suggested that the pressure of the overlying compacting clay over the already rigid interior of the concretions caused slickensides and striations on the outer part of the septaria. The shrink-
314
J.MERTENSETAL.
Fig. 8. Overview of preferred orientations in the clay pits. ing of the lithifying inside of the nodule may have triggered the process. As the mechanical shear resistance of the concretions is much larger than the shear stresses caused by the overburden, the surfaces must have formed in an earlier stage when the carbonate in the outer parts of the concretion had not yet fully hardened. The surfaces are mineralized with pyrite similar to the one that is precipitated on the sepataria inside the concretions.
Discussion Although similar large fissures have often been described in literature, a soil mechanical explanation is rarely discussed. The general reference to overconsolidation of clays is not sufficient as similar jointing is also observed in young clay deposits, which have not been overconsolidated (Terzaghi 1951). Attempts for a more quantitative explanation of the joint formation were made by Price & Cosgrove (1990). They suggested that the phenomena are in fact vertical shear joints. These joints would develop due to an augmentation of the shear stress caused by a slight inclination of the layers. However, in the case of the Boom Clay, which shows a clear orthogonal set, this theory is not applicable, as only one dip direction occurred during the whole burial history. In addition, the lack of striations on the joint surfaces, in a soft clay material which easily shows impact marks and the lack of any displacement along the joints also point to a tensional origin of these structures. It was also observed that none of the preferred directions of the joint sets could be designated as primary direction, being the first direction in which joints developed. This suggests that both directions of the set developed at the same time. From the orientation of the jointing in the clay with respect to the orientation of the clay faces in the excavation pits and from the observed clay-septaria interactions, it can be deduced that jointing is not related to the excavation of the clay. Its cause is natural. As the joint orientations seem to be region-
Fig. 9. Interaction between septaria and clay. Two near vertical joints are crossing a septaria. The left joint cuts through the outer part of the septaria. The right joint cuts through the outer part of the septaria and follows the septae. The other and smaller fissures visible are caused by exposure of the clay to the air.
ally consistent, jointing might be related to a regional geological process. From the relationship with the septarian nodules it can be concluded that the large jointing occurred during the hardening but before the full hardening of the septaria. As septaria with fully lithified outer shells are reworked at the base of the overlying Miocene sands, the jointing did occur before that time. As a general conclusion, it can be stated that tensional forces in the Boom Clay must have created the observed joints at some moment during the history of the Boom Clay. We shall therefore first examine the conditions required for tensional failure in the Boom Clay.
Limiting conditions for tensional joint development using the Mohr-Coulomb criterion: a theoretical approach As the morphology of the joints clearly points towards a brittle failure and as only general condi-
JOINTS IN THE BOOM CLAY FORMATION
315
Fig. 11. Illustration on a Mohr-Coulomb diagram of the theory of maximum possible depth of tension joint formation.
Fig. 10. Large slickensided plane on a septaria from a clay pit near Rumst. Also visible (in dark) are the vertical cracks caused by shrinkage during the full hardening of the septaria, and crosscutting the sample.
tions are evaluated, a brittle Mohr-Coulomb fracture criterion is appropriate in this theoretical approach. From the geometry in a Mohr circle representation of the failure criterion in the tensional domain, it can be stated that when the maximal normal stress exceeds a particular value, the failure in the negative normal stress field becomes impossible and shear failure will occur in the positive stress field, instead of tensional failure (Fig. 1 1). In the case of the Boom Clay, the vertical stress will be a limiting factor for the possibility of tensional failure, as overpressures did not occur during the shallow burial history of the clay. Above a certain vertical stress amount, tensional failure will become impossible and only shear failure will occur. As the vertical stress is directly related to the amount of overburden, this implies that the clay must have been located above a certain maximum depth at the time of joint formation. This limit value for the maximum depth in the case of the Boom Clay can be deduced from geometrical calculations (see Appendix) and is given in the formula below:
where dmax(m) is the maximum depth where tension joint formation can occur, c the cohesion and <J> the internal angle of friction. The bulk density is 2 t/m3 and the water pressure is hydrostatic. To summarize, in clay which was never overpressured during its burial history, the maximum depth of joint formation can be estimated when the cohesion and internal friction angle are known. As can be seen in the formula, the maximum depth of joint formation depends on the cohesion and angle of internal friction. Tensional failure will
Fig. 12. Sensitivity of maximum depth dmax to changes in cohesion and internal angle of friction.
be possible at greater depths if the cohesion or the internal angle of friction of the clay is increased. The sensitivity of both parameters is illustrated in Figure 12. First, the angle of internal friction was kept constant at a representative value of 18°. The cohesion was changed between 0.1 MPa and 0.4 MPa. The maximum depth of tension joint formation changed from less than 20 m to more than 70 m. In opposition, when taking the cohesion constant at a representative value of 0.2 MPa, the changing of Jmax when the angle of internal friction is changed between 10° and 25° is rather small: from 33 m to 38 m. So only changes in cohesion provoke significant changes of dmax. Considering these results, it is acceptable to continue the theoretical approach with a constant angle of friction of around 18°, which is an average value for the Boom Clay near the surface. However, small changes in cohesion are strongly affected by maximum depth, suggesting that, even within the clay layer itself which is around 40-50 m
316
J. MERTENS ETAL.
Fig. 13. Experimentally derived relation between cohesion and depth for the Boom Clay.
thick in the Antwerp area, small differences in cohesion need to be considered. Therefore, the clay in the clay formation has a changing cohesion from the top to its base. Results from recent experimental work on the Boom Clay (Wildenborg & Bogdan 2000), show a linear relationship between cohesion c and the burial depth of the clay (Fig. 13). This makes it possible to estimate the cohesion of the clay in every depth position in the formation. The deepest burial depth from the top of the Boom Clay in the Antwerp area was around 90 m, as derived from geophysical, geomechanical and geological reasoning (Schittekat et al 1983). Assuming that the clay retains its maximum burial cohesion, the cohesions of the clay in all positions in the formation and at all times during its burial history are known. That this assumption is plausible is illustrated by the example of Terhagen, where a cohesion of 0.15 MPa near the top of the Boom Clay, derived from the experimental linear relationship, is in agreement with measurements in Van Impe (1993). In Figure 14, for the Antwerp area, the experimentally deduced relation between cohesion and depth of the clay at its highest uplift position in its history, is plotted together with the relation between cohesion and dmax. The highest uplift position of the clay formation is taken, as it is the most favourable position for joint formation. The two curves are plotted as bands rather than as lines, to incorporate some uncertainty in the values. The intersection between both lines determines the field where tensional jointing can occur in an upper part of the clay and where it is excluded and replaced by shear failure in the lower parts of the Boom Clay. This is because below the intersection point, the cohesion of the clay is too small to allow tension failure to occur. The depth value limiting the two failure states is about 40-50 m in the Antwerp area. From calculations similar to the one demonstrated for Kruibeke (location in Fig. 1) and applied to all the clay pits, the general conclusion can be drawn that the formation of tension joints in the Boom Clay is limited to the near surface (<50 m).
Fig. 14. Relationship 'cohesion — depth' and 'cohesion' maximum depth tension joint formation'. In this situation, the clay is located near the surface after the uplift. This is the most favourable condition for tension joint formation. Above the cross-section of the two curves the vertical stress is low enough to create the joints. The cohesion is large enough in this case. Beneath the cross section, no tension joints can be formed due to a too large vertical stress and the relative low cohesion of the clay.
Discussion of the origin of joints The stress history of the Boom Clay: in search of stress states with tensional forces In order to use the above mentioned mechanical state description of the Boom Clay for evaluating when the tensional jointing occurred, maximal and minimal normal stresses need to be specified throughout the history of the Boom Clay deposit. For a subhorizontal overconsolidated clay like the Boom Clay, the normal stresses are the vertical burial stress and the horizontal stress, the latter assumed to be equal in all directions. Maltman (1994) derived relationships between the effective horizontal stress (cr'h) and the effective vertical stress (cr'v) during loading and unloading. They can be expressed for the Boom Clay in the following formulae (^is the Poisson ratio):
by loading.
by unloading and reloading till previous reached maximum. The evolution of both values with burial and subsequent uplift can be derived from the relationships by Maltman and a Poisson ratio of 0.4 (Wouters & Vandenberghe 1994). This is represented in Figure
JOINTS IN THE BOOM CLAY FORMATION
317
Furthermore, as the tensional joints are vertical, the minimal stress should be horizontal and not vertical as derived from the theoretical stress history curves. Tectonism in the area
Fig. 15. Stress evolution of the clay undergoing burial and uplift over 100 m.
15. In this figure, the stress evolution of a layer in the Boom Clay is shown, buried to a depth of 100 m and uplifted afterwards towards the surface. In overconsolidated clays, there exists a point in the uplift history where the maximal stress is no longer the vertical, but the inherited horizontal stress. In this case, the point lies at about 50 m depth. The evolution of the Mohr circles during uplift is shown in Figure 16. Using the burial history (e.g. Fig. 17a, b) and the above-mentioned formulas, the horizontal and vertical stress evolution curves in time can be constructed. An example of this being the evolution of a clay layer in the Boom Clay formation at Kruibeke, shown in Figure 18. It is clear that the evolution of the stresses under normal loading and unloading conditions cannot explain the observed jointing. Indeed, as the horizontal stress can never become negative when the clay is buried, it is impossible to create the tensional forces needed to create the observed joints.
The nearest and most important tectonic event since the deposition of the Boom Clay was the Roer and Rhine Graben activity in SE Holland and NE Belgium. This graben structure had developed and was also inverted long before the deposition of the Boom Clay. However, there was an important increase in rift activity just after the deposition of the Boom Clay during the late Oligocene and during the Plio-Pleistocene. The stress pattern in the area indicating minimum horizontal stress in NE-SW direction has not changed direction since the Oligocene (Bergerat & Vandycke 1994). Stress magnitudes, however, might have changed. It seems, therefore, unlikely that positive extension in the NW-SE direction, needed to explain certain directions of discontinuities, ever occurred due to tectonism during the history of the Boom Clay. Furthermore, in some cases it can be seen that a NW-SE discontinuity was formed before a NE-SW discontinuity and in other cases the opposite was found, not to mention the sometimes more random orientation that occurred e.g. in Steendorp (Fig. 7). Tectonism as an origin itself does not seem a valid hypothesis. Origin of extension by shrinkage The origin of a horizontal extension might lie in the phenomenon of clay shrinkage. In Figure 19, the effect of shrinkage is illustrated in a Mohr-Coulomb
Fig. 16. Evolution of the Mohr circles during an uplift of 100 m.
318
J. MERTENS ETAL
Fig. 17. (a) Burial history of Kruibeke (see Fig. 1). An important uplift of around 100 m occurred near the end of the Oligocene period (24.5 Ma ago). After this uplift, the clay was always relatively shallow buried (base <100 m depth) (b) Burial history of Mol (see Fig. 1). The same uplift took place in Mol, but less pronounced (<50 m). After the uplift, the clay was buried again, till its present and deepest burial state.
JOINTS IN THE BOOM CLAY FORMATION
319
Fig. 18. Stress evolution during the burial history of Kruibeke. Fig. 20. 3D drawing from the formation of an orthogonal set of joints. Arrows indicate the direction of least horizontal stress in the initial stress state.
Fig. 19. Illustration on a Mohr-Coulomb diagram of the effect of shrinkage on the stress state of the clay.
diagram. The initial stress state is that of a point somewhere in the clay near the surface. Because the clay is allowed to shrink in the vertical direction, no extra tension will be created in this direction. However, in the horizontal direction, shrinkage will result in extra-extensional tension, as the clay is not allowed to displace in this direction. When the clay has shrunken far enough, the horizontal stress will reach the point of tensional strength, and tension joints will be created. Although freezing and thawing of the clay, in particular during the ice age, may have induced volume changes, this cause of shrinkage is very improbable as irregularities that are more plastic would be expected and the influence would be more superficial. A loss of pore fluid, due to various mechanisms, can cause considerable shrinkage, as a clay always tries to maintain full saturation. Very simple calculations — based on the elastic stress-strain relationships and an E-modulus of about 300 MPa (E-modulus in Wouters & Vandenberghe 1994) — give a very rough estimation of the pore fluid decrease needed to create tension failure in the situation of the Boom Clay. A loss of around 1 or 2% could be sufficient. This effect is due to the very weak cohesion of clays. The origin of an orthogonal set of joints can be explained by the anisotropy of the horizontal stresses on a regional scale. As already mentioned, the direction of minimal horizontal stress was
roughly NE-SW. Shrinkage (taking place in all directions) will create an equal decrease in all directions of the horizontal stress. However, as the initial horizontal stress in one direction is minimal, it will be in this direction that the value of tensional failure will be reached first. Joints will start to form perpendicular to this direction. The direction perpendicular to the developed discontinuities will have a horizontal stress close to zero, as relaxation takes place owing to the failure. With continuing shrinkage, the previous maximum horizontal stress becomes the least one and further failure will occur perpendicular to the previously developed discontinuities. With further shrinkage, this process continues and an orthogonal set is created. This is shown in Figure 20, where a 3D illustration from the development of a joint set is shown. Further shrinkage might have created the smaller and more irregular joints observed. The conditions for tension failure development (shallow burial state) and the timing of joint formation (somewhere between 20 and 24 Ma ago) also fit the shrinkage conditions.
Conclusions The dominant discontinuities that can be observed in the Boom Clay outcrops are vertical and mutually perpendicular joints, spaced between 0.5 and a few metres. The absence of striations on the surfaces and the lack of throw along the joints demonstrate that these joints are of tensional origin. The relations between the orientation of the excavation faces and the joint orientation, together with the interaction with the outer core of the septaria, show that the joints are natural and not induced by excavation. It can be demonstrated that a general relationship exists between the cohesion of the clay and a maximal depth beneath which no tensional joints can develop. For the Antwerp area this maximal depth is situated between 40-50 m. An additional mechanism has to be invoked to explain the existence of the
320
J. MERTENS ETAL
negative horizontal stresses (tensional forces) required to develop jointing. Neither burial/uplift nor regional tectonic history can explain the tensional faults. Shrinkage is a plausible mechanism to create tensional horizontal stresses leading to the jointing. The shrinkage might have been caused by a small loss of pore fluid. The origin of this loss of pore fluid in such a relatively thick clay layer (some tens of metres) remains unknown. When such a jointed clay layer is buried again, the presence of the vertical joints might offer pathways for fluid migration trough a relatively impermeable layer. The regional stress field at the moment of failure influenced the directions of the joints. Considering the presence of reworked septarian nodules with fully lithified outer shells at the base of the Edegem sands and the burial history of the clay, the time of origin of the jointing in the Boom Clay is estimated to be late Oligocene/early Miocene. The knowledge of the burial history of the clay is an important tool to predict the presence of jointing in buried clay masses. Comparing, for example, the burial history curve of the Boom Clay in the Mol area (Fig. 17b) (location on the map of Fig. 1) with the burial curve in the Antwerp area on Figure 17a, it can be seen that the Boom Clay top did not reach the surface during the Late Oligocene uplift and never did so in later times. About 20-30 m of the fine glauconitic and clayey Late Rupelian Eigenbilzen Sands remained on top of the clay. The uplift was also far less pronounced than that of the Antwerp area. Calculations based on this history of the Mol area show that only in the few upper metres of the clay, could joints have been developed. The present experience during the large tunnel excavations at the underground research facility in Mol, located 225 m below surface and 35 m beneath the top of the Boom Clay formation confirms this, as no joints Like those described in the clay pits and discussed in this paper are observed in the underground works. Appendix In Figure 21 the Mohr-Coulomb criterion is given for a continuum with a cohesion c and an angle of friction 4>- The tensile strength of the Boom Clay is not well known; the rule of thumb (being c/2) is supposed to be a real minimum in this case. Estimations of the tensile strength of the Boom Clay are higher, meaning that the derived d^^ is an utmost maximum. The maximum possible depth of tension joint formation will be reached when labl equals (c/2 +x). From geometric calculations, we obtained for labl:
Fig. 21. Simplified geometrical calculations on the MohrCoulomb criterion.
This gives us for x:
The maximum vertical effective stress is then given by:
So, the maximum burial depth can be calculated from this using a bulk density of 2 t/m3:
References ATTEWELL, P.B. & FARMER, I.W. 1976. Principles of engineering geology. Chapman and Hall, London, pp. 1045. BELL, EG. 1981. Engineering Properties of Soils and Rocks. Butterworths & Co. Publications, London, pp. 149 BERGERAT, F. & VANDYCKE, S. 1994. Palaeostress analysis and geodynamical implications of CreatceousTertiary faulting in Kent and the Boulonnais. Journal of the Geological Society, London. 151,439-448 BURLAND, J.B., LONGWORTH, T.I. & MOORE, J.F.A. 1977. A study of ground movement and progressive failure caused by deep excavations in Oxford Clay. Geotechnique 19(4), 453-477 DECLEER, J. & VIAENE, W. 1993. Rupelian Boom clay as raw material for expanded clay manufacturing. Applied Clay Science, 8, 111-128 DECLEER J., VIAENE, W. & VANDENBERGHE, N. 1983. Relationships between chemical, physical and mineralogical characteristics of the Rupelian Boom Clay, Belgium. Clay Minerals, 18, 1-10
JOINTS IN THE BOOM CLAY FORMATION DE BEER, E.E. 1967. Shear strength characteristics of the Boom Clay, Proceedings of the Geotechnical Conference, Oslo, 1, 83-88. DE BEER, E. 1971. Problemes poses par la construction du tunnel sous 1'Escaut a Anvers. Sols et fondations 83. Annales de Vinstitut technique du bdtiment et des travaux publics. DE SLOOVERE, P. 1983. Extensometer measurements in a cavity cooled down to — 196°C. Symposium International In situ testing, Paris, 2,481—491. FOOKES, P.O. 1965. Orientation of fissures in stiff over consolidated clay of the Siwalik system. Geotechnique, 15,195-206. HENRIET, J.P., MONJOIE, A. & SCHROEDER, C. 1986. Shallow seismic investigations in engineering practice in Belgium. First Break, 4 (5), 29-37. HENRIET, J.P., VERSCHUREN, M. & VERSTEEG, W. 1992. Very high resolution 3D seismic reflection imaging a small scale structural deformation. First Break, 10 (3), 81-88. Institution of Civil Engineers. 1976. Manual of applied geology for engineers. London, pp. 378. LAGA, P. 1966. Kleidiapier in de uitgraving voor de spoorwegtunnel van de E3-weg op de rechteroever te Antwerpen. Het ingenieursblad, 35 (17), 552-553 MALTMAN, A. 1994. The Geological Deformation of Sediments. Chapman and Hall, London. NEERDAEL, B. 1996. Geological radwaste disposal in Belgium. In: Geological problems in radioactive waste isolation - second worldwide review. Berkely Laboratory, US. PRICE, NJ. & COSGROVE, J.W. 1990. Analysis of geological structures. Cambridge University Press, Cambridge. SCHITTEKAT, J., HENRIET, J.P & VANDENBERGHE, N. 1983.
Geology and geotechnique of the Scheldt Surge Barier. Characteristics of an over consolidated Clay. In: 8th International Harbour Congress 2, Antwerp, 121-135.
321
SINTUBIN, M., LAGA, P., VANDENBERGHE, N., KENIS, I. & DUSAR, M. 2000. Deformation features in Paleogene sands in the Hoegaarden area (Belgium). Geologica belgica, 3/3-4,257-269. SKEMPTON, A.W., SCHUSTER, R.L., PETLEY, D.J. 1969. Joints and fissures in the London Clay at Wraysbury andEdgware. Geotechnique, 19 (2), 205-217. TERZAGHI, K. 1951. The influence of modern soil studies on the design and construction of foundations. Building Research Congress. Div, 1,139-145. VAN IMPE, W. 1993 Grondmechanica. Academia Press, Gent. VAN ECHELPOEL, E., VANDENBERGHE, N. 1987. Field guide to the Rupelian stratotype. Bulletin de la Societe beige de Geologic. Brussels. 96. Part 4, 325-337. VAN ECHELPOEL, E. WEEDON, G.P. 1990. Milankovitch cyclicity and the Boom Clay Formation: an Oligocene siliciclastic shelf sequence in Belgium. Geological Magazine, 127 (6), 599-604. VANDENBERGHE, N. 1978. Sedimentology of the Boom Clay (Rupelian)in Belgium. Verh. Kon. Acad. Belgie, Klasse der Aardwetenschappen. XL, pp 147. VANDENBERGHE, N., LAENEN, B., VAN ECHELPOEL, E. & LAGROU D. 1997. Cyclostratigraphy and climatic eustasy. Example of the Rupelian stratotype. Earth and Planetary Sciences, 325, 305-315. VANDENBERGHE, N., LAGA, P., STEURBAUT, E., HARDENBOL, J. & VAIL PR. 1998. Tertiary sequence stratigraphy at the southern border of the North Sea basin in Belgium. SEPM Special publication no. 60,119-154. WARTEL, S. 1980. The Tertiary and Quaternary sub bottom of the Schelde estuary near Antwerpen (Belgium). Geologic en Mijnbouw, 59 (3), 233-240 WILDENBORG T. & BoGDAN O. 2000. Transport of radionuclides through a clay barrier. TNO-NITG 8-11. WOUTERS, L. & VANDENBERGHE, N. 1994. Geologic van de Kempen. Nirond, Brussels, pp 208.
This page intentionally left blank
Soft-sediment deformation in Lake Superior: evidence for an immature Polygonal Fault System? N. J. WATTRUS1, D. E. RAUSCH1 & J. CARTWRIGHT2 l
Large Lakes Observatory, University of Minnesota, Duluth, MN 55812 USA (e-mail: nwattrus @ d. umn. edu) Department of Earth Sciences, University of Cardiff, PO Box 914, Cardiff, CF10 BYE Abstract: New multi-beam sonar and seismic data collected in Lake Superior document the widespread development of lake-floor rings in fine-grained lake-floor sediments. The multi-beam images reveal that the rings develop as connected clusters and that individual rings have an irregular polygonal appearance. High-resolution seismic data collected with a 28 kHz echo sounder reveal extensive fracturing and faulting in the glacio-lacustrine sediments below the lake-floor. Displacement on the faults is typically normal with throws of less than 50 cm. Three styles of faulting are recognized: (a) monoclinal flexure; (b) graben-like; and (c) conjugate. Zones of acoustic blanking below the faults may be associated with de-watering and mobilization of the sediments. Lateral thickness variation in some horizons suggests that fault and fracture development is linked with lateral movement of sediment. Piston cores collected near lake-floor rings show well-developed fractures and microfaults, suggesting that fracturing and faulting occurs on a wide range of scales. The seismic and lithological characteristics of the glacio-lacustrine section are similar to those of sediments in which Polygonal Fault Systems (PFS) have been described. This suggests that the rings in Lake Superior may be the surface expression of PFS in the near-surface sediments.
Side-scan sonar surveying in the North American Great Lakes has revealed the complex character of their lake-floors. Rather than being static, featureless basins, they are dynamic surfaces that are shaped by a variety of processes. Some of these no longer occur while others continue today. Much of this work has been done in western Lake Superior. Among the many features observed are: scour marks made by icebergs dragging in the sediments on the lake-floor (Berkson & Clay 1973a); erosional troughs produced by contour currents (Johnson 1980; Flood & Johnson 1984; Halfman & Johnson 1984; Johnson et aL 1984); parallel furrows and grooves produced by bottom currents (Flood & Johnson 1984; Johnson et al 1984; Flood 1989); and slumping (Johnson 1980). Perhaps the most interesting features seen in sidescan sonar data from Lake Superior are rings and sub-circular depressions that occur in the finegrained sediments found in the deeper parts of the basin (Fig. 1). They are typically 100-250 m across, 25-50 m wide and 1-5 m deep. The lake-floor within the rings has the same depth as the surrounding lakefloor. High-resolution seismic profiles collected across these features show that the unconsolidated glacio-lacustrine clay beneath them is fractured and faulted (Fig. 3). The lake-floor rings were first observed in sidescan records collected by Berkson & Clay (1973&) who interpreted them to be the polygonal lake-floor expression of cracks or fractures produced by syneresis in the underlying sediments. Other workers (Flood & Johnson 1984; Flood 1989) interpreted them to be segments of sub-circular ring depressions
produced by compaction-driven dewatering of the underlying sediments. Similar oval depressions have been described in Lake Michigan (Berkson et al. 1975; Colman etal 1992), Lake Huron (Moore Jr. et al. 1994) and in the Placentia Bay, Newfoundland (PsiTrottetaL 1997). In this paper, we describe the results of experiments conducted in Lake Superior to study the origin of the lake-floor rings. These experiments have produced the best images yet of the deformation in Superior's lake-floor sediments. They suggest that the rings on the floor of Lake Superior may represent the lake-floor expression of an immature Polygonal Fault System (PFS) in the fine-grained lake-floor sediments. PFS are a class of soft sediment structures that has only recently been recognized. This class was initially described in the central North Sea Basin, where PFS are widely developed in Palaeogene mudrock-dominated sequences. They are believed to result from volumetric contraction triggered by syneresis during early compaction of ultra-fine sediment (Cartwright & Dewhurst 1998). The timing and duration of the processes that lead to the development of PFS are poorly understood, largely because there are no known active PFS. Studying the rings in Lake Superior may provide the key to understanding these processes.
Geological setting Lake Superior lies at the southern margin of the Canadian Precambrian Shield. The basin is
Fig. 1. Side scan sonar image of lake-floor rings collected off Grand Marais, Minnesota near the study area.
structurally controlled and has probably been in existence since the beginning of the MesoProterozoic. Deep drilling within the basin (Zumberge & Gast 1961) failed to recover any preQuaternary rocks younger than Precambrian. Farrand & Drexler (1985) summarize the Quaternary history of the lake basin. Quaternary deposits in the western part of Lake Superior lie unconformably on the Precambrian bedrock. They represent the lateWisconsinan glaciation of the region when the Superior lobe of the Laurentide ice sheet advanced into and out of the basin several times. The ice began its final retreat out of the basin about 10,250 +/-100 yr. BP (Lowell et al 1999). The soft sediments described in this paper post-date this retreat. The post-glacial sequence in western Lake Superior consists, from oldest to youngest, of homogeneous red clay, red varved clay, grey varved clay and gray homogeneous clay. This sequence rests on a red, clayey till, which represents either the Marquette advance or the earlier Split Rock/ Nickerson advance. The thickness of the post-glacial sequence ranges from 0-45 m and averages about 25 m (Dell 1971). The appearance of grey homogeneous clay signifies the establishment of an 'ice-free' basin around 8000 yr. BP (Mothersill 1988).
Methods Since 1996, the Large Lakes Observatory (LLO) has collected over 5000 km of high-resolution seismic reflection data and side scan sonar data in western Lake Superior (Fig. 2). The seismic data were collected using an ORE Geopulse boomer generating peak energy at about 1 kHz, firing at 0.5 s intervals. In 1998, a Knudsen 320B/R 28kHz echo sounder
was added to the suite of systems used. It is possible to obtain very high-resolution subsurface images with this system (Fig. 3). In areas with fine-grained unconsolidated substrate, this system exceeded 20 m of sub-bottom penetration, thus covering a significant proportion of the soft sediments in western Lake Superior. Side scan sonar data were simultaneously acquired using an ORE 100 kHz side scan sonar system. Post-survey processing of the seismic data was carried out using the SU seismic processing package developed at the Colorado School of Mines (Stockwell Jr. 1997). Processing included editing of bad traces, muting, filtering to remove coherent noise, deconvolution to reduce multiples and to shorten the wavelet and amplification to boost lowamplitude reflections. Interpretation of the echo sounder profiles is complicated by the presence of imaging artifacts. These are a function of the operating frequency of the echosounder, its beam width, the water depth and the curvature of the lake-floor and subsurface reflectors (Flood 1980). The artefacts can complicate the appearance of subsurface features and produce apparent truncations in reflections (Fig. 4). The convex bulge in the reflectors that may develop next to a fault, for example, can create these artefacts. Since the curvature of this feature increases with depth, the magnitude of the artefact grows with depth (Figs 5 and 6). Attempts to suppress these artefacts with seismic migration techniques (Stockwell Jr. 1997) have proven unsuccessful due to the nature of the echo sounder signal and the relatively low signal-to-noise ratio of the data. In 1999 a multi-beam survey was conducted in western Lake Superior off Grand Marais, Minnesota (Fig. 2) in an area known to have lake-floor rings overlying faulted glacio-lacustrine sediments. The survey covered an approximately 5 km 2 area with an average water depth of 175 m. The multi-beam survey was acquired with a Reson Seabat 8101 multi-beam system mounted on the RAf Blue Heron. This system operates at 240 kHz and uses 101 beams of 1.5° width. Vessel motion and position information was obtained with a TSS POS/MV320. This system can achieve +7-0.05° accuracy in roll, pitch and heading, and provide submetre horizontal resolution using a fully integrated system of inertial sensors and survey-grade DGPS receivers. The data were recorded using aTritonElics ISIS system. The survey was managed using the Coastal Oceanographies Hypack. Sound velocity profiles required for refraction corrections were collected periodically during the course of the survey using an AML SVPlus sound velocimeter. Postacquisition processing of the multi beam data was performed using CARIS HIPS. After editing bad data and applying corrections for vessel motion and
SOFT-SEDIMENT DEFORMATION IN LAKE SUPERIOR
325
Fig. 2. Map of Lake Superior showing the study areas with locations of regional echo sounder survey lines, cores sites and the multibeam survey. Shaded regions indicate areas exhibiting ring development on the lakefloor. Lake Superior's position on the North American continent is indicated by the box on the inset map.
refraction effects, bathymetric grids with a 10 m sample interval were prepared. Swath width, the width of the strip of lake-floor illuminated by the multi-beam, decreases with increasing water depth. In shallow water (z<50 m) the swath width is approximately 7.2 X water depth. In deep water, the effective swath width is approximately 2 X water depth. The multi beam has a slant range resolution of 0.0125 m. Its horizontal resolution is determined by the water depth, beam width and beam angle. In water 200 m, deep the horizontal resolution of the nadir beam is 5.2 m. This decreases with increasing beam angle growing to 7.8 m at the outer beams. The actual working resolution of the system is somewhat better, since the data are collected with overlapping swaths and the outer beams are typically discarded during processing. The multi beam data were collected on track lines orientated parallel to the regional bathymetric contour. The track line spacing was selected to ensure 20% overlap between adjacent swaths and 100%
coverage of the survey area. Average vessel speed during the surveys was 8 knots. High-resolution seismic profiles were also collected in the study area using both the Knudsen echo sounder and the Geopulse. Although it poorly imaged the internal structure of the glacio-lacustrine sediments, the Geopulse's deeper penetration allowed mapping of the entire soft sediment package down to the bedrock. This data was used to investigate the influence of deeper structures on the glacio-lacustrine section. In order to trace deformation patterns in the glacio-lacustrine section across the area, track lines for this part of the experiment were spaced at 15 m intervals. Several gravity and Kullenberg piston cores, 7.8 cm in diameter, were recovered from the survey areas (Fig. 2, Table 1). The longest core recovered was 715 cm long. Coring locations were selected using the echo sounder records. Their positions were obtained with the ship's navigational system. The recovered cores were cut into 1.5m sections, capped and stored for later analysis at the LLO.
326
NJ.WATTRUSErAL.
Fig. 3. Typical echo sounder record collected in area with lake floor rings. This example was collected off Grand Marais, MN as part of a regional seismic survey of western Lake Superior. The interpreted section in the lower panel illustrates fracture systems developed in two separate tiers.
Results
storm-induced currents no longer sweep the lakefloor (Johnson 1980). Rings and subsurface deforInterpretation of the regional data confirms the wide- mation are sometimes not found in the deep-water spread development of the ring structures through- areas, indicating that the presence of fine-grained out much of the western basin of the lake (Fig. 2). sediments does not guarantee the occurrence of rings They are only found in areas wherefine-grainedsedor fractures in the glacio-lacustrine section. iments occur below the lake-floor. In the western The rings often appear to be partially filled in by half of the lake, this typically corresponds to areas sediments that were transported across the lake-floor where water depths exceed 100 m. At these depths, by bottom currents (Flood & Johnson 1984; Flood
SOFT-SEDIMENT DEFORMATION IN LAKE SUPERIOR
327
Fig. 4. Examples of the three structural styles recognized in the glacio-lacustrine sediments. Note the flexure that typically develops adjacent to the fault. This deformation increases with depth along the fault and terminates in a basal layer, which exhibits evidence of lateral sediment flow(indicated by the arrows). The sediments below the faults are commonly underlain by zones of acoustic blanking, (a) Monoclinal flexure; (b) Graben-type; (c) Conjugate type.
1989). The eroded interface between the post-glacial and glacio-lacustrine sections suggests that the near surface is susceptible to erosion from bottom currents (Fig. 3). Therefore, the present day rings are either: (a) relict features that are protected from bottom currents; or (b) forming at a rate fast enough for their growth to exceed the flux of sediments
across the lake-floor. The former hypothesis is supported by submersible observations (Flood 1989), which described glacio-lacustrine sediments outcropping in the walls of the rings, suggesting that they are not filling in. The echo sounder data show that the surface of the glacio-lacustrine sequence is commonly eroded and
328
NJ.WATTRUSErAL.
Fig. 5. Example of the imaging artefacts typically encountered in the echo sounder profiles. This example was collected off Isle Royale, Michigan (Fig. 2) in an area with lake floor rings.
that the underlying sediments exhibit fractures and small-scale faulting (Fig. 3). Some of these continue to the surface where they coincide with a ring or partial ring structure. Many, however, are truncated at the post-glacial grey clay interface. Apparent displacement on the faults is typically normal and throws on the faults are usually less than 50 cm. Three structural styles dominate the faults and fractures: monoclinal flexures, graben-like, and conjugate (Fig. 4). These faults, especially the grabenlike and the conjugate type, are often associated with acoustic blanking of the sediments immediately below. In some areas, the echo sounder records show two layers or tiers with continuous, sub-parallel internal reflections. In places, these layers are separated by a poorly resolved facies that lacks internal reflections (Fig. 3). Elsewhere, the boundary between the two layers is marked by an unconformity, which in some cases is angular. More typically, the boundary is simply disconformable with very little evidence to suggest any hiatus in the depositional history. All the layers have well-developed fracture systems. Most of the fractures do not extend beyond the layer in which they appear. In general, these systems are confined to distinct intervals in the subsurface and appear to be genetically unique. Based upon their acoustic expression and the available core record, it appears that these fractures only occur in fine-grained sediments.
Fig. 6. Schematic representation of the origin of imaging artefacts seen in the echosounder profiles. Note how the location of the fault is obscured by diffractions from the abruptly terminated reflectors.
The sediments adjacent to the faults are frequently deformed into low-relief bulges. Like their associated fractures and faults, these deformations are layer-bound. The amplitude of the deformation appears to increase with depth along the fault, reaching a maximum in the lowermost portion of the tier. This portion lacks significant internal reflections, and shows marked lateral variation in thickness. No high-resolution 3D seismic data are available for mapping the spatial distribution of fractures in the glacio-lacustrine sediment. In addition, poor weather conditions during the survey are the cause of the relatively low signal-to-noise ratio in the echo sounder data. Therefore, producing an accurate map of fracture distribution from echo sounder data is not possible. Nevertheless, mean fracture densities were
SOFT-SEDIMENT DEFORMATION IN LAKE SUPERIOR
329
Table 1. Gravity and piston cores collected during the multi-beam survey Core Name
47-35.130 N 46-57.325 N 46-54.574 N 46-53.221 N 46-50.059 N 47-39.284 N 47-29.246 N 47-33.967 N 47-30.031
90-11. 694 W 86-36.774 W 86-36.498 W 86-36. 144 W 86-35.724 W 89-44.537 W 89-44.627 W 89-16.424 W 89-43.3 15 W
715cm 186cm 152cm 185cm 14cm 305cm 473cm 484cm 449 cm
gc,gvc gc gc gc bs gvc, 1-10 mm faults - not split gc, gvc, 1-10 mm faults gv, 1-10 mm faults
Cores with names ending in P are piston cores, those with names ending in G are gravity cores, gc = grey clay, gvc = grey varved clay, bs = brown sand. The gravity cores were all collected in the Blue Heron Trough in the eastern multibeam study area (Fig. 2). Table 2. Average fracture density on echo-sounder lines recorded off Grand Marais, MN in an area with rings on the lake floor Line Name
Fracture densities vary from greater than 4/200m to none. Lines WLS99-08 and WLS99-09 are closer to the shore and do not intersect the other lines. calculated using regional echo sounder data collected in an area exhibiting lake-floor rings off Grand Marais on the Minnesota North Shore. These show little variation with azimuth of the track-line (Table 2). The multi beam data show well-developed rings on the lake-floor. The rings appear both as isolated features and as connected networks. They appear to become more angular and polygonal in appearance in the SW third of the surveyed area (Fig. 7). An isopach map depicting variations in the thickness of the post-glacial grey clay was constructed from the echo sounder profiles (Fig. 8). It indicates that the postglacial sediment thins southward and is completely absent in much of the SW half of the study area. This observation suggests that the change in appearance of the lake-floor rings is linked to the thinning and pinch-out of the postglacial sediment. The Geopulse seismic data were used to create depth maps and isopach maps of the glaciolacustrine section and the underlying till (Fig. 8). In this, albeit small area, no correlation between these maps and the location, size, or density of the lakefloor rings is apparent.
Fig. 7. (a) Grey-shaded contoured representation of the lake floor surveyed by the multi-beam. The surveyed area is centered at 47-29-39 N, 89-43-52 W and covers approximately 5 km2; (b) shaded relief image of the multi-beam bathymetry. Surface illuminated by a southern light source located 60 ° above the horizon. Vertical exaggeration = 10 X. The varved glacio-lacustrme clays recovered in piston cores from the study area exhibit extensive small-scale fracturing and faulting (Fig. 9). The faults typically exhibit normal displacement with
330
N.J.WATTRUSErAL.
Fig. 8. Contoured lake bathymetry (note ring shapes) superimposed on various subsurface maps in order to show correlation between the lake-floor and subsurface morphologies. The maps are: (a) Isopach map of the post-glacial Holocene gray clay; (b) isopach map of the varved glacio-lacustrine clay; (c) depth map of till surface; (d) isopach of the till horizon. A constant velocity of 1500 m/s was used to calculate the isopach and depth maps.
SOFT-SEDIMENT DEFORMATION IN LAKE SUPERIOR
331
332
N.J.WATTRUSCTAZ, Table 3 Seismically identifiable criteria for recognizing PFS (1) Polygonal fault patterns in map view. (2) Layer-bound fault system delineated by regionally correlatable stratigraphic surfaces. (3) Fault system distributed over a large part of the basin. (4) Normal style faulting, small throws. (5) Closely spaced faulting. (6) Tiered development ie. stacked, unique sequences of faults. (7) Fault polarity: Switching between apparent 'synthetic' to 'antithetic' faulting. Listed in order of diagnostic value. Taken from Cartwright &Dewhurst(1998).
Fig. 9. Example of the mierofaults observed in the varved glacio-lacustrine clays recovered from piston core LS99-1 IP located in the SW corner of the multi-beam survey (Fig. 2). Its location (47-29-14.75 N, 89-44-37.64 W) represents the edge of a network of rings, (a) Split core photographed under natural daylight; (b) Interpreted split core. Fault throws are typically less than 10 mm.
throws that vary between 0-1 cm. The varved nature of the sediments permits their reconstruction to their pre-faulted condition, demonstrating that coring probably did not induce the faults.
Discussion The most significant features in the data are the widespread layer-bound fractures and small faults in the glacio-lacustrine sediments. Similar layer-bound faulting is visible in seismic reflection data from the North Sea (Cartwright & Lonergan 1996). In 2D profiles, the faults appear to be part of a small-scale extensional system. They occur over large areas of the sedimentary basin. Early models proposed that their origin and occurrence were associated with overpressure, density inversion and hydraulic fracturing (Henriet et al 1991; Cartwright 1994a, 19941?). Only when mapped with 3D seismic data was it recognized that the deformation associated with these faults resulted in radially uniform extensional strains. In map view these faults have a polygonal appearance. This unusual state of strain means that a tectonic origin for these structures is unlikely.
Cartwright & Lonergan (1996) proposed that these Polygonal Fault Systems (PFS) are the result of volumetric contraction. Cartwright & Dewhurst (1998) used 3D seismic data from the North Sea to describe seismically identifiable criteria (Table 3) that can be used to identify PFS development in other sedimentary basins. They used these criteria in a global survey to analyse the distribution of PFS and to identify common characteristics among the different basins where they occur. Their study identified regionally extensive PFS in 28 sedimentary basins around the world. The only factors common to all 28 examples were that the PFS occurred in marine depositional settings containing either ultra-fine (<2 nm) smectitic clay stones or carbonate chalk. High porosity and extremely low permeability also characterized the systems. There did not appear to be any systematic correlation with other factors such as sedimentation rate, age, organic carbon content or depth of burial. Reconstruction of the North Sea PFS to their prefaulted configuration shows that they have undergone 3D contraction rather than the 2D contraction resulting from gravity-driven compaction. Cartwright & Dewhurst (1998) proposed that the colloidal properties of the sediments during the early stages of burial are important criteria for PFS formation. They suggested that syneresis was the agent responsible for the volumetric contraction described in the model proposed by Cartwright & Lonergan (1996). In this scenario, pore fluids during early diagenesis are expelled under chemical rather than hydraulic gradients (Dewhurst etal. 1999). Using Cartwright & Dewhurst's (1998) criteria for recognizing PFS shows that the interpreted data from Lake Superior meet nearly all of the requirements for PFS development. This suggests that the fracturing and faulting in the glacio-lacustrine sediments of Lake Superior may be a PFS. Given the very young age of these sediments, this PFS would be extremely immature, perhaps still active.
SOFT-SEDIMENT DEFORMATION IN LAKE SUPERIOR
The sediment in the offshore basins of Lake Superior is extremely fine grained. Published grain size analyses of sediment samples collected from the lake show that 84.5% of the grey clay and 92.2% of the grey varved clay is smaller than 1 (Jim (Farrand 1969). Berkson & Clay (19736) reported high water content (>50%) in the 6 m cores collected by researchers from the Carnegie Institute. The water content of the piston cores collected as part of this study was very high, ranging from 80-90% just below the surface to 70-75% at the bottom of the core. Fractures are well developed over large areas of the western basin, apparently where the lake-floor is composed of very fine-grained sediments. If the rings are the surface expression of a PFS developed in the near-surface sediments, the following model may explain their formation. High sedimentation rates associated with the retreat of glacial ice out of the basin led to the deposition of ultra-fine clays with high water content. Some time after deposition syneresis produced a volumetric contraction of the sediments and created a PFS in the glaciolacustrine clays. Pore fluid migrating upward from the underlying sediments may trigger syneresis, although no correlation between the morphology of the rings and that of deeper reflectors is apparent. The timing and duration of this event is poorly understood. Since many of the fractures and faults truncate against the Holocene lower boundary, it appears that most of the volumetric contraction ceased before the deposition of the post-glacial gray clay began. Alternatively, it is possible that reworking of the recent lake-floor sediments removed the upper portions of the fractures and faults. Some fractures reach the lake-floor and may have acted as conduits for the expelled water. Water venting from the lake-floor removed or prevented the deposition of sediment near the fracture forming the ring. There is no evidence to indicate the duration or rate of this venting. It seems more likely that it was a prolonged, slow venting which caused the surface sediment to be fluidized and susceptible to erosion by bottom currents. The presence of multiple tiers of layer-bound fractures suggests that the conditions necessary to form them occurred several times during the deposition of the glacio-lacustrine sediments in Lake Superior. The nature of the interval separating the uppermost two tiers is not clearly understood. In some places, the intervening unit is absent and tiers of fractures are developed directly one above the other and separated by an unconformity. Presumably, these represent periods when lower lake levels led to widespread erosion of the palaeolakefloor. Similar conditions may be responsible for the scoured, commonly unconformable boundary between the glacio-lacustrine section and the overlying Holocene grey clay.
333
There are several differences between the examples described by Cartwright & Dewhurst (1998) and features in Lake Superior. The most striking difference is the scale of the latter. They appear to be approximately an order of magnitude smaller than those described by Cartwright & Dewhurst. This may simply be because the North Sea seismic data, which were acquired with airguns, cannot resolve features having the size of those in Lake Superior. This difference could also be due to the glaciolacustrine sediments in Lake Superior being much thinner than the Paleogene sediments in the North Sea, where Cartwright & Lonergan (1996) first described PFS. A more extensive survey across nearsurface sediments that vary in thickness could test this hypothesis. An alternative explanation may link changes in ring size and density to variations in one or more of the material properties of the sediment. Although too small to resolve with high-resolution seismic data, the micro-faults observed in the cores raises the possibility that they could be linked to the formation of the rings. This would suggest that fractures develop at all scales and possibly have a fractal character. Similar fractal relationships have been demonstrated in mudcracks (Velde 1999). If the rings in Lake Superior are the surface expression of a PFS in the underlying glaciolacustrine sediments, they represent the first example of PFS development in a glacio-lacustrine sediments and a freshwater environment. Dewhurst et al. (1999) suggest that the presence of saline pore water may play a part in triggering syneresis in marine sediments. The freshwater environment of Lake Superior argues that a different agent is responsible for triggering syneresis in this case. The presence of a thick glacio-lacustrine sequence does not guarantee the development of fractures or rings. In some parts of the lake, the glacio-lacustrine section exceeds 20 m but is free of fractures and shows no significant deformation. There are no rings developed above these sediments. Clearly, if the rings are the surface expression of a PFS developed in the near-surface sediments, the volumetric contraction necessary to produce a PFS has not occurred in those regions. Perhaps the mechanism responsible for triggering syneresis did not occur. Even if it did, other factors, such as the presence of coarser-grained sediments may have limited contraction.
Conclusions Acoustic data collected in western Lake Superior document the widespread development of ring-shaped features on the floor of the lake. The rings are underlain by systems of layer-bound fractures and faults within glacio-lacustrine sediments. Piston cores collected near lake-floor rings exhibit micro-faulting,
334
N. J. WATTRUS ETAL.
suggesting that ring formation may be accompanied by fracturing and faulting at many scales. According to the model, high sedimentation rates associated with the retreat out of the basin of the glacial ice, led to the deposition, of ultra-fine clays with high water content. Sometime after deposition, syneresis produced a volumetric contraction of the sediments, creating a PFS. Several questions remain and require further study. At present, the timing of the deformation is poorly understood. If these systems developed over extended periods of time, what processes could have sustained them? If syneresis was the mechanism responsible for their formation, what triggered it? If these features are associated with an immature PFS developed in the young glacio-lacustrine section, studying them further may shed new light on the formation of PFS elsewhere. This work was funded in part through grants from the National Science Foundation (OCE-9724432), Sea Grant, the Natural Environment Research Council (GR3/R9906), and the Weinert Foundation. We thank the Captain and crew of the R/VBlue Heron.
References BERKSON, J.M. & CLAY, C.S. 19730. Microphysiography and possible iceberg grooves on the floor of western Lake Superior. Geological Society of America Bulletin, 84,1315-1328. BERKSON, J.M. & CLAY, C.S. 19736. Possible syneresis origin of valleys on the floor of Lake Superior. Nature, 245,89-91. BERKSON, J.M., LINEBACK, J.A. & GROSS, D.L. 1975. A side-scan sonar investigation of small-scale features on the floor of southern Lake Michigan. Illinois State Geological Survey, Report 74. CARTWRIGHT, J.A. 19940. Episodic basin-wide fluid expulsion from geopressured shale sequence in the North Sea basin. Geology, 22,447-450. CARTWRIGHT, J.A. 19946. Episodic basin-wide hydrofracturing of overpressured Early Cenozoic mudrock sequences in the North Sea Basin. Marine Petroleum Geology, 11, 587-607. CARTWRIGHT, J.A. & DEWHURST, D.N. 1998. Layer-bound compaction faults in fine grained sediments. Geological Society of America Bulletin, 110,1242-1257. CARTWRIGHT, J.A. & LONERGAN, L. 1996. Volumetric contraction during the compaction of mudrocks: A mechanism for the development of regional-scale polygonal fault systems. Basin Research, 8,183-193. COLMAN, S.M., FOSTER, D.S. & HARRISON, D.W. 1992. Depressions and other lake-floor morphologic features in deep water, southern Lake Michigan. Journal of Great Lakes Research, 18, 267-279. DELL, C.I. 1971. Late Quaternary sedimentation in Lake Superior. PhD Thesis, University of Michigan, Ann Arbor, MI. DEWHURST, D.N., CARTWRIGHT, J.A. & LONERGAN, L. 1999. The development of polygonal fault systems by
syneresis of colloidal sediments. Marine and Petroleum Geology, 16,793-810. FARRAND, W.R. 1969. The Quaternary record of Lake Superior. In: 12th Conference on Great Lakes Research. International Association of Great Lakes Research. FARRAND, W.R. & DREXLER, C.W. 1985. Late Wisconsinan and Holocene history of the Lake Superior basin. In: KARROW, RF. & CALKIN, RE. (eds) Quaternary evolution of the Great Lakes, Geological Association of Canada, 30,17-32. FLOOD, R.D. 1980. Deep-sea sedimentary morphology: modelling and interpretation of echo-sounding profiles. Marine Geology, 38,77-92. FLOOD, R.D. 1989. Submersible studies of current -modified bottom topography in Lake Superior. Journal of Great Lakes Research, 15, 3-14. FLOOD, R.D. & JOHNSON, T.C. 1984. Side-scan targets in Lake Superior - evidence for bedforms and sediment transport. Sedimentology, 31, 311-333. HALFMAN, J.D. & JOHNSON, T.C. 1984. The sediment texture of contourites in Lake Superior. In: STOWDORRIK, A.V & PIPER, DJ.W. (eds) Fine-grained Sediments; deep-water Processes and Fades. Geological Society, London, Special Publications, 15,293-307. HENRIET, J.P., DE BATIST, M. & VERSCHUREN, M. 1991. Early fracturing of Paleogene clays, southernmost North Sea. In: SPENCER, A.M. (ed.) Generation, accumulation and production of Europe's hydrocarbons, European Association of Petroleum Geologists Special Publication, 217-227. JOHNSON, T.C. 1980. Late-glacial and postglacial sedimentation in Lake Superior based on seismic-reflection profiles. Quaternary Research, 13, 380-391. JOHNSON, T.C., HALFMAN, J.D., BUSCH, W.H. & FLOOD, R.D. 1984. Effects of bottom currents and fish on sedimentation in a deep-water, lacustrine environment. Geological Society of America Bulletin, 95, 1425-1436. LOWELL, TV., LARSON, G.J., HUGHES, J.D. & DENTON, G.H. 1999. Age verification of the Lake Gribben forest bed and the Younger Dryas advance of the Laurentide Ice Sheet. Canadian Journal of Earth Sciences, 36, 383-393. MOORE JR., T.C., REA, D.K., MAYER, L.A., LEWIS, C.F.M. & DOBSON, D.M. 1994. Seismic stratigraphy of Lake Huron - Georgian Bay and postglacial lake level history. Canadian Journal of Earth Sciences, 31, 1606-1617. MOTHERSILL, J.S. 1988. Paleomagnetic dating of late glacial and post-glacial sediments in Lake Superior. Canadian Journal of Earth Sciences, 25, 1791-1799. PARROTT, R.D., SHAW, J. & HUGHES-CLARKE, J. 1997. Use of multibeam bathymetry in a regional geophysical survey in Placentia Bay, Newfoundland. In: 66th Annual International Meeting of the Society of Exploration Geophysicists. Dallas, TX, 833-834. STOCKWELL JR., J.W. 1997. Free software in education; a case study of CWP/SU; Seismic Un*x. The Leading Edge, 16, 1045-1049. VELDE, B. 1999. Structure of surface cracks in soil and muds. Geoderma, 93 (1-2), 101-124. ZUMBERGE, J.H. AND CAST, P. 1961. Geological investigations in Lake Superior. GeoTimes, 6,10-13.
Mobile shale related deformation in large deltas developed on passive and active margins C. K. MORLEY Department of Petroleum Geoscience, University of Brunei Darussalam, Bandar Seri Begawan, 2028, Brunei Darussalam Abstract: Understanding the nature of mobile shale deformation in deltas requires appreciation of several critical factors: (1) the tectonic setting (deposition on an active or passive margin); (2) the range of potential structural styles and how to identify them on seismic reflection data; (3) how structural styles evolve with time; (4) how to apply and modify what is understood about salt deformation to mobile shale deformation; and (5) how delta structures influence sedimentation. Differences between active and passive margin settings include: (a) active margins tend to display shorter, higher gradient drainage systems associated with high rates of erosion per unit area when compared with passive margins; (b) active uplift of the hinterland will force delta progradation (not seen on passive margin); (c) uplift causes erosion and re-cycling of older deltaic deposits which can cause comparatively fine grained sandstone reservoirs; and (d) the development of large compression-related folds and thrusts in the shelfal region of deltas on active margins can create areas of active uplift. Consequently, structurally controlled embayments may develop that cause distinct changes in the dominant depositional process (wave, tidal, fluvial) along the margin. On both active and passive margins, growing gravity-driven structures locally create sea floor highs and depressions and create changes in the dip of slopes which influence depocentre location and sediment pathways from the shelf to deep water. A wide range of mud diapir geometry can be found in deltas. Commonly mud diapirs exhibit reactive, active, passive and collapse features. Some diapirs develop as compressional features seaward of fault bounded depocentres. Also frequently associated with delta provinces are fluid pipes, shale intrusions, fluid fronts and gas clouds. All the overpressure phenomena can produce dimming, or disruption of reflections on seismic reflection data, however the characteristics of the different phenomena are sufficiently diverse that they can frequently be differentiated on good quality data.
Introduction Deformation caused by loading of overpressured, under-compacted (mobile) shales associated with deltas produces spectacular, rapidly evolving structures, thick, rapidly forming depocentres and massive fluid fluxes in basins. Commonly in the literature the large-scale deformation associated with gravity tectonics on deltas is referred to as mobile shale, or shale tectonics. Although lithologically mud or clay is a more appropriate term than shale, this paper will not try to resist the common usage of shale, Deltas are commonly significant hydrocarbon provinces, e.g. the Caspian Sea, Baram Deltaic Province, Columbus Basin-Trinidad (Orinoco Delta), Bengal Basin, Gulf Coast of the USA, Niger, Mahakam and Nile Deltas (Fig. 1). Some under-explored deltaic provinces also remain, particularly in Asia, for example the Mekong Delta stands out. Remarkably, the literature on mobile shale deformation remains comparatively sparse and understanding of shale tectonics is commonly linked with the much larger salt tectonics literature. The barriers to better understanding of mobile shale deformation include sparsity of outcrop examples of shale deformation and the need for (proprietary) subsurface data to understand what is a very large-scale phenomenon.
Early work describing the basic principals of deformation in deltas came from the Gulf Coast of the US and the onshore region of the Niger Delta. Key papers include Morgan et al (1968), Merki (1972), Bruce (1973), Dailly (1976), Bishop (1978), Evamy (1978) and Doust & Omastola (1990). Many valid and some redundant ideas about mobile shale deformation developed in these papers remain entrenched in our general perceptions of delta tectonics, However subsequent to those papers seismic reflection data quality and quantity has improved considerably; data is available from a larger number of deltas and seismic data coverage is more comprehensive, particularly in the deeper water areas. Additionally, numerous analogue models have enhanced our understanding of the dynamics and kinematics of gravity structures (e.g. Vendeville & Cobbold 1987; Vendeville & Jackson \992a,b\ Jackson & Vendeville 1994; Ge et al 1991 \ McClay et al. 1998, 2000). In a review of deformation in the Niger Delta Morley & Guerin (1996) noted that considerable similarities and differences existed between shale and salt deformation. Two of their key conclusions were: (1) on seismic data deep, once reflective, stratified sequences were somehow being lost within regions of chaotic reflections (thought to represent mobile shales); and (2) the overpressured
Fig. 1. Location map of the main mobile shale provinces associated with deltas world-wide.
nature of mobile shales resulted in significant differences in structural style and diapir evolution compared with salt tectonics, including migration of overpressure 'fronts'. In this paper a review and revision of the basic understanding of delta tectonics is undertaken because considerably more data has recently become available. In particular, outcrop and seismic reflection data has enabled the nature of the dimmed or chaotic seismic reflection data to be better understood (e.g. Morley et al 1998; Van Rensbergen et al 1999; Van Rensbergen & Morley 2000). Also, the characteristics of deltas on active margins, particularly around Borneo have become better described (e.g. Sandal 1996; McClay et al 2000; Wood 2000; Madon & Abu Hassan 1999; Lambiase in press). Some of the key points addressed are: (1) problems with seismic interpretation of mobile shale deformation, in particular how to distinguish the various types of diapirism and mobile shale intrusions; (2) the interaction between structure and sedimentation (accommodation space creation, bathymetry, impact on sediment transport pathways), and in particular how structure may affect reservoir sandstone distribution; and (3) how the tectonic setting of deltas affects structural style and sedimentation.
Tectonic setting of deltas Sedimentologists recognize that many different factors influence sedimentation within deltas. These include climate, fluvial processes, energy of the receiving basin and characteristics of the receiving basin (such as shelf width, platform stability, basin
geometry, location of sediment input points) (e.g. Davies 1980). The characteristics of the receiving basin are discussed in this section. Most of the largest deltas in the world are formed on passive margins (e.g. Nile, Niger, Mississippi, Mekong deltas) and comparatively few lie on active margins (e.g. Ganges-Brahmaputra rivers-Bengal Basin), (e.g. Coleman & Prior 1980; Davis 1980). Hydrocarbonbearing deltaic provinces are found extensively on both kinds of margin; passive margin delta hydrocarbon-bearing provinces include the Gulf of Mexico, Nile and Niger Deltas (e.g. Evamy et al. 1978; Doust & Omastola 1990; Morton & Galloway 1991; Wever 2000). Active margin deltas include the Volga River (Caspian Sea) (Katz et al 2000), GangesBrahmaputra rivers-Bengal Basin (Shamsuddin & Abdullah 1997), Columbus Basin-Orinoco Delta (Gibson 1994; de Gamero 1996; Wood 2000); Baram Deltaic Province, NW Borneo (Sandal 1996), and Kutai Basin-Mahakam Delta, SE Borneo, (van de Weerd & Armin, 1992; McClay etal 2000), (Fig. 1). A single, great river feeds the classic large deltaic provinces on passive margins. Typically the drainage basin of the river is large and of low gradient. For example the river Niger is over 4030 km long, with headwaters lying below 600 m elevation. Approximately 1.3-1.5X106 km3 of sedimentary rocks have been deposited in the Niger Delta, which has a drainage area of 2.1 X 106km2. This equals approximately 0.73 km3 of sediment in the basin for every square kilometre of drainage area. However, not all passive margin settings are similar. For example, the Amazon River has not developed a major delta, instead a massive canyon system feeds most of the sediment directly from the fluvial system
MOBILE SHALE RELATED DEFORMATION IN LARGE DELTAS
into an enormous submarine fan system (e.g. Flood et al 1995). Perhaps the narrowness of the passive margin due to highly oblique extension during continental break-up was an important contributing factor to this situation. The Gulf of Mexico is a rather unusual setting of an abandoned, propagating oceanic spreading centre and passive margin system. It is fed by numerous rivers, including the Mississippi, that contains delta-related deposits from multiple fluvial sources and mobile shale deformation is sub-ordinate to the effects of salt. Active margins exhibit a variety of geometries. One common type is where multiple deltaic basins are developed on a rapidly uplifting, narrow, linear region. One example is the Baram Deltaic Province, NW Borneo, which formed on the margin of an accretionary prism complex (e.g. Sandal 1996). Approximately 2.3 X 105km3 of (Middle MioceneRecent) sedimentary rocks lie in the basin, derived from a drainage area of about 35,000 km2. The amount of sedimentary rock derived per square kilometre of drainage area is 6.6 km3, a considerably higher yield than the Niger Delta. The Baram River (170 km long, with headwaters at about 1500 m elevation) is part of a linear front of drainage where several large rivers are concentrated to form a coalesced system of deltas. The presence of different fluvial systems feeding into a basin may lead to different timing of deformation and loading in different parts of the basin as the different fluvial systems evolve. Active margins are not necessarily linear; in extreme cases they may be U-shaped or completely enclosed and form a strongly subsiding thrust belt hinterland surrounded by mountains. For example, the Alboran Sea (Rif-Betic Cordillera) and the Caspian Sea (Perez-Belzuz et al 1997; Katz et al 2000; Fig. 1). Deposition from large and small fluvial systems originating from the uplifting mountain belts surrounding the marine basins resulted in rapid loading of the basin by fine grained sediments and the development of mud diapirs. Mobile shales may be widespread throughout the basin, but can be concentrated on active tectonic trends, particularly along strike-slip faults. Phenomenal amounts of sediment can rapidly accumulate in these basins, the South Caspian Basin for example contains up to 25 km of Tertiary sediment deposited on thinned continental or oceanic crust. Mechanics of mobile shales Early studies of deltas identified the single most important factor in delta tectonics as the rapid progradation of delta sediments across thick, water-rich pro-delta shales (Bruce 1973; Evamy et al 1979; Knox & Omatsola 1989). Deposition focused on one
337
part of the margin causes a broad flexural isostatic responses to accommodate the load. Even in deepwater areas of the Niger deltas 3-6 km of MioceneRecent have accumulated, while on the shelf 12 km of delta sedimentary rock is commonly imaged on seismic reflection data (Morley & Guerin, 1996; Fig. 2). Deep, rapid burial of the pro-delta shales preserves porosity and pore fluids in the shales, hence the shales remained undercompacted and pore fluids became overpressured. Sediment loading triggers ductile flow of the mobile shales and results in the development of growth faults, mud diapirs, mud volcanoes and toe thrusts in the overlaying deltaic succession (e.g. Morgan et al 1968; Evamy et al 1979; Knox & Omatsola 1989). At any one time sedimentation tends to be concentrated in a narrow band of depo-belts, some 30-60 km wide, bounded by large displacement (up to 6-7 km throw) growth faults. Step wise progradation of deltas via successive depo-belts has been proposed (e.g. Evamy et al 1978). As subsidence in a depo-belt ceases the zone of active subsidence jumps oceanwards and delta top and alluvial sands tend to be deposited over the largely inactive growth faults. This style of progradation was termed the 'Escalator Regression Model' by Knox & Omatsola (1989). The driving mechanism for gravity tectonics include buoyancy forces, gravity sliding and differential loading (see reviews by Jackson & Talbot 1986; Kehle 1989; Vendeville & Jackson 1992a). On passive margins where sediments are comparatively thin, the post-rift thermal subsidence section tends to have a gentle regional offshore dip, which promotes structures forming by gravity sliding (Fig. 2b). The offshore Angolan passive margin exhibits spectacular examples of this kind of gravity sliding where salt forms the main detachment (e.g. Duval et al 1992). Gravity sliding associated with overpressured shale detachments can produce a series of seaward dipping (regional) faults which sole out into one or more detachment horizons and pass down dip into a zone of toe thrusting and folding. There is usually little significant mud diapirism. Some relatively small deltas such as the West Luconia Delta, W Borneo, may be dominated by gravity sliding structures (e.g. Madon & Hassan 1999). Gravity sliding is also commonly invoked as the mechanism of deformation in the main parts of major deltas (e.g. Bruce 1973; Evamy et al 1978; Cohen & McClay 1996). However major deltas are considerably different from smaller deltas, the margins of large deltas, and passive margins. In major deltas, mobile shales form a very thick succession and delta sediments increase in thickness significantly passing from the toe thrust belt (sediments 4-5 km thickness, covered by up to 3 km of water) towards the delta (sediment thickness 12-14 km,
338
C. K. MORLEY
Fig. 2. The structural setting of the Niger Delta, illustrated by a regional map of the deltaic tectonic provinces, and sketched sections based on seismic reflection data, (modified from Morley & Guerin 1996). (b) gravity sliding on low sedimentation rate part of passive margin, (c) and (d) regional sections through the delta; (e) examples of tall shale diapirs with downbuilt synformal depocentres, height of diapirs about 8-9 km; and (f) examples of offshore prograding counter regional fault depocentres.
MOBILE SHALE RELATED DEFORMATION IN LARGE DELTAS
water depth less than 200 m). Hence flexural loading of the crust by the delta pile results in the base of the delta wedge dipping gently back towards the continent (Fig. 2c, d), consequently gravity sliding would not seem to be the driving mechanism for the major structural features of large deltas. Instead differential loading of the mobile shales by uneven deltaic sedimentation would seem to be the main driving mechanism (e.g. Edwards 1976; Morley & Guerin 1996; Ge etal 1997; McClay etal 2003). It is reasonable to argue that at the start of major delta deposition the oceanward slope was preserved and so gravity sliding was important in the early stages of delta development. Morley & Guerin (1996) noted that on parts of the Niger Delta seismic data an early series of comparatively small, closely spaced normal faults was present over printed by the main delta structures. The author has noted similar features on seismic data across the Baram Delta (Fig. 3b). These early, closely spaced faults could represent the gravity sliding stage. Unfortunately, these features have been little studied or documented. One significant characteristic of mobile shales is that mobility is not confined to a specific stratigraphic unit. Mobility can change with time depending upon burial rate, the amount of overburden, lateral changes in shale thickness, regional and local stresses, local drainage conditions (faults, carrier beds, fluidized pipes) and the internal overpressuring conditions (e.g. Osborne & Swarbrick 1998) that operate at a particular time. To maintain and regenerate overpressured conditions when fluids are lost during deformation requires mechanisms other than burial-undercompaction. As reviewed by Osborne & Swarbrick (1998), most diagenetic mechanisms (such as the Illite-Montmorillonite transition) for the production of water can only produce volume increases of a few percent. The main diagenetic mechanism for 'topping up' established overpressure conditions is the conversion of oil to gas (Osborne & Swarbrick 1998), which can cause a volume increase of 50-80 times (Barker 1990). The top of the main overpressured shale mass commonly lies at depths of 4-5 km, which is coincident with the depth of maximum oil generation, or the start of the gas window (e.g. Khalivov & Kerimov 1983; Sandal 1996; Morley & Guerin 1996; Paterson 1997). Commonly at outcrops of shale pipes the plastic mud matrix has a petroliferous odour. Methane and other hydrocarbons have been widely verging on ubiquitously reported from active mud diapirs, thereby supporting the notion that hydrocarbons are important to mud diapirism (Hedberg 1974). Gas migration from lower to higher depths may also provide a charge to overpressured shaly units higher in the section (Graue2000). Changing stress conditions can also enhance
339
overpressure. Hydrofracturing and consequent loss of overpressure is related to the value of the minimum principal stress. Evolution from a stress regime favouring extensional faulting to one favouring thrust faulting will cause the minimum principal stress to change from a sub-lithostatic pressure value, to near lithostatic. Hence the pore fluids can support a higher pore fluid pressure ratio before hydrofracturing occurs (e.g. Osborne & Swarbrick 1998). Such a mechanism might be significant for deltas on on active margins, but not on passive margins. Vendeville & Jackson (19920) recognized that the old buoyancy mechanism of diapir rise as a less dense viscous fluid rising through a denser viscous fluid did not match the mechanical behaviour of rocks in the upper 10 km of crust. They proposed that salt diaprism equated with a viscous fluid trying to rise through a brittle material. The resistance of the overburden to (salt) diapir rise was assumed to be controlled by the brittle shear strength of the roof rocks and subsequent slip was controlled by the frictional behaviour of the block bounding faults (Vendeville & Jackson 1992a). Consequently considerable resistance to the rise of a diapir was predicted. Vendeville & Jackson (1992a) suggested that a diapir would not pierce a roof thicker than about one third of the adjacent total overburden thickness. The situation for diapir rise by shale differs from salt because overpressures are involved. In the one well documented outcrop example of mobile shale deformation associated with a delta, networks of dykes, sills and other intrusive bodies are found in the roof of a mud diapir (Morley et al 1998; Morley 2003). The presence of intrusive complexes may explain dimmed reflections bordering diapirs on seismic reflection data (Van Rensbergen et al 1999; Van Rensbergen & Morley 2000). If mud diapirs do rise by hydraulic fracturing and stoping of the country rock, then the roof rocks are weaker still than proposed by Vendeville & Jackson (1992a) and are controlled by the stress conditions required to initiate hydraulic fracturing. When the pore fluid pressure is large enough to overcome the tensile rock strength and the minimum horizontal stress in an adjacent mass natural hydraulic fracturing will occur (e.g. Sibson 1996). The diapir may creep upward due to the effects of buoyancy and differential loading, but caprock strength is reduced to the conditions necessary for hydraulic fracturing.
Types of mud intrusion and their seismic signature Classic mud diapirs exhibit different stages of diapir development, similar to salt diapirism, including the reactive, active and passive stages
340
C. K. MORLEY
Fig. 3. Map and sections of the deltaic province of NW Borneo (modified from Morley et al. in press), (a) Regional map of the NW Borneo margin showing mixed Middle Miocene age growth faults and compressional/transpressional folds; (b) Regional cross section through onshore and offshore Brunei and Sarawak; (c) Cross-section through the western offshore area of Brunei illustrating predominantly growth fault features, with a later series (late MioceneRecent) of widely spaced mostly regional faults prograding over an earlier, more closely spaced sequence of regional and counter regional faults.
MOBILE SHALE RELATED DEFORMATION IN LARGE DELTAS
defined by Vendeville & Jackson 1992a, b\ Cohen & McClay 1996; Morley & Guerin 1996; PerezBelzuz et al 1997 (Fig. 4). In addition, overpressured fluid/intrusive fronts, pipes and gas clouds have been identified, which on poorer quality seismic reflection data would probably be interpreted as mud diapirs (e.g. Van Rensbergen et al. 1999). A distinction is made here between mud diapirs and mud pipes, however in the literature commonly no distinction is made and both features are referred to as mud (or shale) diapirs. Sometimes the pipes are referred to as shale diapirs, which emanate from mud pillows or rollers. An analogy is sometimes drawn between shale pipes and igneous pipes and the igneous term diatreme is applied to shale pipes. The reason for making a distinction between pipes and diapirs is that large masses of mobile muds which form diapirs, pillows and rollers behave in somewhat similar ways to salt diapirs, whilst shale pipes do not. For example both mud and salt diapirs display reactive, active and passive phases, welds and downbuilding depositional synclines (e.g. Morley & Guerin 1996; Warren 1999). Mud pipes are long, narrow forcefully intrusive bodies or networks of intrusions filled with overpressured mud and fluids, with gas either in solution or free (depending upon depth). Often the pipes break through to the surface and create mud volcanoes. Mud pipes develop by hydraulic fracturing associated with overpressured fluids, hence there seems to be no equivalent feature in salt tectonics. Pipes may emanate from shale diapirs, or overpressured non-diapiric mud or shale masses (Graue 2000). Once the overpressured fluids have been driven off, a plastic or scaly clay matrix containing angular blocks of lithified rock fragments is left filling the pipe (e.g. Barber et al. 1986). It is important to discriminate between emplaced masses of mobile muds and injection and intrusion features that diminish seismic data quality and which may be mistaken for shale diapirs. Summarizing data from Morley & Guerin (1996), Van Rensbergen et al. (1999); Haskell et al (1999), Stewart (1999), Graue (2000), and Van Rensbergen & Morley (2000), some key distinguishing characteristics of chaotic masses seen on seismic reflection data are given in the following lists.
Criteria for a mud diapir (Figs. 2e and 4) (1) Chaotic seismic reflection data in a relatively broad, steep-sided mass. (2) Commonly circular to elongate, can be irregular and may form walls. Large diapirs can be 10 km or greater at their base, but tend to taper upwards and may be only a few kilometres in diameter at their top. The maximum height of
(3) (4) (5)
(6) (7)
(8)
(9) (10)
341
large diapirs (measured from the touchdown at the base to the crest) is commonly around 6-8 km. Broad, large velocity pushdown beneath the diapir. This is the opposite of salt diapirs, which are associated with velocity pull-ups. Touchdown geometry on flanks of diapir where the mobile shale has been almost entirely evacuated. Evidence for reactive, active and passive phases. The reactive phase can sometimes be identified as section expanding towards the diapir on one side (into a growth fault) and thinning onto the other side. The passive and active phases are associated with (6) and (7) below. Downbuilding - as shown by deep, typically synformal basins adjacent to diapirs. Passing towards the diapir flank strata are folded upwards, syn-kinematic strata display decreasing rotation up section and onlaps onto more highly rotated strata (Fig. 4). This may be an essential distinguishing feature from feeder pipes (below) when the shale diapirs are tall and narrow. Sag basin, collapse faults (volume loss in shale mass due to dewatering) overlying diapir. Major fields of conjugate faults occur if the diapir collapse phase is present. If the diapir has not dewatered and compacted, it will be associated with a negative gravity anomaly. Broad field of multiple mud volcanoes if emergent of sea floor.
Chaotic seismic reflection data in narrow zone, with subvertical sides, usually much taller than it is wide. Circular to oval cross-section a few hundred metres up to about 3 km in diameter. Modern or palaeo-evidence for mud volcano emerging to the sea floor. May emanate from a deeper overpressured shale mass, or a shale diapir. Usually velocity pull-down effects are small. May display several pulses of mud volcano emergence at the sea floor. Not associated with reactive, active and passive phases, or downbuilding; no associated touchdown geometries. Relatively small collapse-related faults and short wavelength (few 100 metres) synformal down-bending of reflections from strata into the pipe are common.
342
C.K.MORLEY
Fig. 4. Seismic lines illustrating some characteristics of shale diapirs. (a) Typical large shale diapir with associated responses of adjacent sedimentary section to diapir growth, inferred to indicate reactive, passive/active growth, (b) Mass of fingers of chaotic seismic facies. Not all chaotic pipe or diapir-like bodies are easy to interpret, in seismic line b despite the tall, narrow, pipe-like nature of the chaotic seismic facies is difficult to interpret as one style. The adjacent sediments on the left half of the Fig. display synformal geometries indicative of active or passive diapirism. The right half of the figure does not show synformal reflections from the country rock suggesting the chaotic facies represents shale pipes and/or gas chimneys.
MOBILE SHALE RELATED DEFORMATION IN LARGE DELTAS
Criteria for broad fluid/intruded front (Fig. 5) (1)
(2)
(3) (4)
Generally chaotic seismic reflection data, but contains some weak reflections that display similar dips, and some continuity with the adjacent high reflective 'stratified' succession. Commonly associated with dipping strata, forms an inclined front moving up dip, can resemble a faulted edge, except in detail the edge tends to be ragged (Van Rensbergen & Morley2000). Easily mistaken as the edge of a shale diapir and may pass deeper or laterally into a shale diapir. Inclined, diffraction-like reflections which crosscut reflections from strata are commonly present. The anomalous reflections are probably from the fluid/intrusion front and form pseudo-fault plane reflections.
Criteria for gas clouds and gas chimneys (Fig. 6) (1)
(2) (3) (4) (5) (6)
Generally chaotic seismic reflection data, but contains some weaker reflections that display similar dips to, and continuity with, adjacent high reflective 'stratified' succession. Formed from a few percent gas saturation in permeable layers and fractures (Stewart 1999). May overlie a shale diapir. Is not associated with basin formation such as down-building synclines adjacent to shale diapirs. May cause numerous, small velocity push downs, and break up reflections. Cloud-like zone of poor data quality. Usually kilometre-scale in diameter. Not associated with negative gravity anomaly. Gas chimneys tend to rise off deeper hydrocarbon accumulations. They form a narrow, vertical chaotic seismic data zone and may extend vertically hundred of metres to several kilometres. More diffuse gas features also exist and can give rise to broad patches of poor quality data that in outline have a tree or bush-like shape.
Overpressured fluids and shale intrusions render shale diapirs sufficiently different from salt diapirs that an alternative approach needs to be taken when describing and classifying them. Shale diapir growth can follow three main paths (Fig. 7): (1) reactive to passive development of diapirs, similar to salt diapirs as described by Vendeville & Jackson (1992a). A large shale mass is emplaced as a diapir which initially grows passively in the footwall of normal faults (reactive rise), followed by active and passive phases, with associated down-building on
343
the flanks (e.g. Morley & Guerin 1996; Fig. 7a,b). (2) A diapir forms by reactive rise, but then the emplacement of shale as a large mass ceases. Instead, mud-filled hydraulic fractures intrude the country rock, propagating both laterally from the sides and upwards from the top of the mobile shale mass (Fig. 7c). Overpressured fluids are expelled, but further rise of an entire chaotic shale mass is not achieved. This phase may occur during diapir emplacement to millions of years after diapir emplacement ceased (Van Rensbergen et al 1999). At the top of the diapir one or more long, narrow pipes are sometimes developed by which mud and fluids are expelled to the sea floor (e.g. Van Rensbergen et al 1999; Van Rensbergen & Morley 2000; Figs. 2e, 4b, 5 and 7c). (3) Toe diapirs associated with local compression (Fig. 7al). Mobile shale (and fluids) are driven oceanward due to loading, commonly by fault-bounded depocentres (e.g. Cohen & McClay, 1996). Typically the overburden is thin, and the diapir may just cause folding (forming a shale pillow), or the oceanward margin of the diapir is bounded by thrusts (Morgan et al 1968; Morley & Guerin, 1996). In this paper such diapirs are referred to as toe diapirs, to emphasize their link with toe thrusts and folds. These three categories are not mutually exclusive and some diapirs may display elements of all three as they have evolved through time. Collapse of shale diapirs Movement and evacuation of large volumes of mobile sediment is commonly indicated by the presence of touchdowns and welds between rock units once separated from each other by the mobile sediment (e.g. Vendeville & Jackson 1992b). However, with mobile shales, a layer of compacted shale is typically left after dewatering. Consequently complete touchdowns are rare, although on seismic data features that look very much like touchdowns typical of salt tectonics can be seen (e.g Figs 2f, 4a and 8). The activity of shale diapirs, pipes and intrusions is related to the quantity of fluids they contain. Water and gas loss will cause the shale volume to shrink. For shale diapirs this produces a crestal sag in the overburden (Fig. 7b). Ring faults or even well developed arrays of conjugate minor faults may also form. These features are developed whilst the diapir is buried, and not emergent at the sea floor. The history of water loss associated with diapirs means that newly emplaced shale bodies will tend to have low densities (typically about 2.20 g/cm3 at 4 km depth), whereas older shale bodies tend to be denser (may approach 2.65 g/cm3 at 4 km) than the surrounding rocks. Hence, younger shale diapirs tend to be associated with negative gravity anomalies and considerably slower velocities than the surrounding
344
C. K. MORLEY
Fig. 5. Seismic line illustrating a typical mud-diapir feeder pipe and a lateral emplacement of a fluid/intrusive front, SW Ampa, offshore Brunei (from Van Rensbergen et al. 1999).
Fig. 6. Seismic line illustrating a gas cloud in detaic sediments (North Sea, data provided by BP).
MOBILE SHALE RELATED DEFORMATION IN LARGE DELTAS
Fig. 7. Schematic cross-sections illustrating some of the characteristics of shale diapirs and pipes, (al) and (a2) variations on shale bulges developing in response to sedimentary loading, by deposition in the hangingwall of a reactive diapir (al) and in a withdrawal syncline (a2); (b) development of shale diapirs similar to classic salt diapirs with reactive, active and passive phases; (c) shale diapirs commonly do not develop like (b) but may superficially resemble them. Instead a complex of gasrich fluids and shale intrusions may intrude laterally and vertically (pipes) into the country rock from mobile shale masses at depth. The well-developed synformal depocentres in (b) are not seen in (c).
345
about 82°C (Hedberg 1974). Hence near surface the quantity of dissolved methane is low. Brown (1990) has shown that the expansion of methane in intrusive muds is a significant mechanism for incorporating material into the diapir, and must aid fluid escape as the fluidized overpressured shales approaches the surface. Hence, it is unlikely that most shale diapirs can sustain prolonged activity once they lie near or are exposed at the surface. Estimated volumes of material extruded over the lifetime of individual mud volcano-feeder pipe systems is in the order of 1-11 km3 (e.g Guliev 1992; Graue 2000). The following rough calculation shows how important mud pipe volcanoes may be in draining mobile shales. First it is assumed that the average life of a mud volcano is 100,000 years, so that if delta sedimentation has occurred over 10 million years, there have been 100 phases of mud volcano activity. The volume of mobile shales available to be drained for a large delta such as the Niger, is estimated as 4 km thickness over a region 300 km X 400 km = 4.8 X 105km3. Within the mobile shales, the initial volume of water=70%, whereas if drained the volume is likely to be about 20%, hence the volume of fluid available to be drained = 2.4 X 105km3. For each phase of the 100 'episodes' 2.4 X 103 km3 of fluid is lost over a area of 120,000 km2. If each mud volcano pumped 1 km3 of fluid over its 100,000 year life then one mud volcano every 50 km2 would drain the entire shale mass. This calculation is very simplistic, for example it does not take into account fluid lost through fault systems and large shale masses arriving at the sea floor, the changing areas of active diapirism with time and whether a single mud volcano system could drain an area of 50 km2. However, it does demonstrate the potential for a system of apparently widely spaced mud volcanoes to permit enormous volume changes in the mobile shale system to occur.
Interaction of sedimentation and structural development Growth faults
rock. For example a shale diapir in the Niger Delta, 4 seconds high, is underlain by strong reflections that display a velocity push-down 1.1 seconds deeper than the same reflection below a stratified sequence. This indicates a considerable mass of overpressured fluids that have reduced the average interval velocity by 25%. The solubility of methane at 20°C (atmospheric pressure) is only about 34.7 ml/1 in pure water and 28 ml/1 in seawater. The solubility increases rapidly with pressure but decreases with temperature up to
Large growth fault depocentres are remarkable for the amount of sediment they can accumulate in a short time period. Although difficult to date, individual depocentres in general show active life spans in the order of 1-5 million years. In the Niger Delta, the largest counter-regional fault depocentres have accumulated syn-kinematic sediments some 6-7 km thick in areas 20-25 km wide, that extend 40-50 km along strike (Fig. 2). The strike-length is based on the length of individual curved fault segments; several growth faults commonly propagate together
346
C. K. MORLEY
Fig. 8. Sketch from seismic data of a shale touchdown associated with a regional fault depocentre, Niger Delta. Commonly as a result of differential loading below the depocentre a shale diapir, and/or toe thrust and fold develop on the seaward side of the depocentre (modified from Morley & Guerin 1996).
to produce a very long fault (hundreds of kilometres) composed of linked, scallop-shaped segments in map view. Models of growth fault development by Bruce (1973) are based on fault development occurring on a shallow shelf, landward of the delta foresets. Rider (1978) made field observations that suggested that growth faulting occurred preferentially in (paralic) sand units and was absent during (pro-delta) shale deposition. However, large active growth faults depocentres actually appear to preferentially develop on the outer shelf-upper slope in the Niger Delta and Baram Delta Province (e.g. Sandal 1996; Hiscott 2001; Figs 2 and 3), a region where sedimentation is dominantly fine grained. In particular, it is common to find that active counter-regional faults cause delta foreset clinoforms to stack in their hangingwall, while active regional faults are also commonly located near or oceanward of the delta topset-foreset transition (e.g. Sandal 1996; Hiscott 2001). The stacking of the delta top-foreset transition does not always occur near the counter regional growth fault, and can occur between one third and half way towards the hinge, or regional fault margin of the depocentre (Figs 2f and 9). This indicates that in terms of section expansion it is not always the topset, shallow shelf section that is thickest and exhibits the fastest depositional rates. Instead it is the deeper shelf-slope pro-delta section (dominated by shelfal mudstones, slumps and turbidite deposits). Morton & Galloway (1991) provide detailed examples of how sequence development on growth fault margins
is a complex mix of local structural effects, eustasy, sediment supply and tectonics. The distribution of growth fault systems is linked to relative rates of deposition versus subsidence (e.g. Bruce 1973; Evamy et al 1978; Fig. 10). Sediment loading triggers subsidence in the hanging wall of a fault and evacuation of mobile shales below the hangingwall. Subsidence tends to be concentrated on a single large fault as long as deposition does not exceed the maximum strain rate of ductile flow in the underlying mobile shale. At the start of deposition, mobile shales have the potential to accommodate subsidence faster than deposition can cause subsidence. Fault activity will slow, or cease to move when strain rates in the mobile shales slow due to dewatering and loss of overpressure, or when they are mostly expelled beneath the fault and a touchdown occurs. Once motion on the fault slows and the deposition rate begins to exceed the subsidence rate, barring compensating eustatic sea-level changes, delta progradation occurs. A new fault develops where mobile shales, oceanwards of the previous fault can more effectively accommodate displacement, and the process described above is repeated. Over a period of hundreds of thousands of years, activity on growth faults may be episodic. Cartwright et al (1998) found for 17 Gulf Coast growth faults their late Pleistocene-Recent history showed at least three cycles of activity separated by periods of inactivity. A partial explanation may lie in variable sediment loading during transgressive-regressive depositional cycles. However, in addition the
MOBILE SHALE RELATED DEFORMATION IN LARGE DELTAS
347
Fig. 9. Schematic section illustrating the control on sand distribution by counter regional faults. Commonly the counterregional fault depocentre will cause aggradation of the shelf edge and force aggradational stacking of sand-prone shallow marine facies. If the shelf edge is location some distance landward of the counter-regional fault then the counter-regional fault pond turbidites adjacent to the fault.
timing of growth differed on neighbouring faults, suggesting that pore-fluid pressure distribution, friction and detachment geometry also played a role (Cartwright et al. 1998). The detailed link between fault activity and sedimentation is clearly an area requiring new studies. The location of the major fault depocentres lies between sediment-starved accommodation space (the continental slope towards deep water) and the delta topset where sediment supply balances or exceeds accommodation space creation. Whilst depocentres are narrow regions of focused subsidence, the flexural isostatic reponse to sediment loading creates additional widespread subsidence. The importance of isostacy is seen where dams have cut off sediment supply to deltas and consequently coastal areas have subsided and begun to flood. Two notable examples are the Aswan Dam on the Nile, and the Mississippi-Missouri system in North America. As a result of the Aswan dam causing negligible sediment influx to the delta, isostatic subsidence is no longer compensated by sediment supply and the coastal region is sinking at a rate of 3-5 mm/yr (Stanley & Warne 1988).
Regional and counter-regional faults The onshore Niger delta area is dominated by growth faults that most commonly dip towards the offshore, due to their importance these faults have been termed regional faults (e.g. Evamy et al 1978). The less abundant faults that dip towards the continental interior are called counter-regional faults. Both fault types are capable of forming large depocentres. With increased offshore exploration, a more complete view of the delta is now available. Counter-regional faults apparently more important than regional faults during the early stages of growth
fault formation (Morley & Guerin 1996; Fig. 2). This is based on the observation that a belt of dominantly counter-regional faults forms the transition between the diapir zone and the onshore regional fault zone in many deltas including the Baram and Niger deltas (e.g. Morley & Guerin 1996; Sandal 1996). An example of a counter-regional fault zone (Fig. 9) shows three main depocentres; each counterregional fault (except the last, most seaward fault) terminates against a younger, relatively small displacement regional fault, which forms the up-dip termination of the more seaward and younger depocentre. Each depocentre represents a relatively short period in the life of the delta, yet there are no older deltaic sediments visible below the depocentres. Their absence led Morley & Guerin (1996) to propose that older, stratified deltaic sediments tend to get buried by younger depocentres and converted into mobile shales. Hence, much of the older deltaic history is lost. The dominance of regional faults in the onshore area is a late stage overprint. Analogue modelling by McClay et al (1998), demonstrated that counter-regional fault systems in single and double load models post-dated regional fault formation and was linked with pronounced thinning beneath the regional fault of the ductile layer flowing outboard (i.e. offshore). The very thick counter regional fault depocentres described above do not fit the idea of a greatly thinned shale layer very well. So while the analogue model provides one set of conditions under which counterregional faults may form, those conditions are probably not the ones in operation in the Niger Delta. However, the notion that deformation precedes large counter-regional fault formation is important. There are hints from seismic data (Morley & Guerin, 1996; Van Rensbergen & Morley 2000) and analogue modelling (McClay et al 2003) that at least in places the early, pre-major
348
C. K. MORLEY
Fig. 10. Growth on fault systems related to depocentre evolution schematically illustrating some of the potential differences between depocentre shifts caused by eustacy and by cessation of growth fault activity.
depocentre history of deltaic deformation is characterized by numerous, closely spaced growth faults and associated diapirs (Fig. 3b). These structures would influence the development of later structures as discussed by Van Rensbergen & Morley (2000) and McClay et al (2003). An incorrect view, still widely held, is that counter-regional faults are anomalous, difficult to explain features. This view was loosely based on certain assumptions such as normal faults originated by gravity sliding, not differential loading and the observation that regional faults were dominant in the onshore region of the Niger Delta (e.g. Evamy et al. 1978). For example, James (1984) tried to explain the occurrence of counter-regional faults in the Baram Delta of NW Borneo as following pre-existing faults, whereas Evamy et al (1978) suggested that counterregional faults in the Niger delta formed along facies boundaries where sandstones abruptly passed into shales. Differential loading due to differences in the amount of compaction between the two lithologies initiated the faulting. In order to locate the counter regional faults at a marked sand-shale facies boundary the counter regional faults were thought to be initiated landwards of the delta slope (Evamy et al 1978). This is contrary to the frequent observation from seismic sections that the delta front tends to stack kilometres away from the counter regional fault, in its hanging wall (Fig. 9). However this model for counter-regional faults still remains popular within the oil industry (e.g. Wood 2000). Actually a
counter regional fault more efficiently accommodates a prograding load than a regional fault and may originate from a shale bulge that develops in front of the prograding load, as shown in physical models (Ge etal 1997). In older models, diapirs were regarded as developing first, with faults forming later on the sites of diapirs (e.g. Bruce 1973; Dailly 1976). Today the picture is more varied. With improved seismic data quality, it is quite common to see that some faults actually controlled the initial diapir activity. For example, in Brunei uplift has exposed the early stages of the prograding delta sequence onshore. Outcrops show the early deltaic deposits prograded over a non-mobile substratum; these most proximal units show no evidence for growth faulting (Back et al 2001). Near the coast the overpressured marine shales become sufficiently thick to permit growth faults to develop. These early growth faults are mostly counter-regional and did not nucleate on earlier diapirs, instead they created reactive diapirs (Figs 3, 5 and 9). Commonly, younger regional faults seem to nucleate around the unfaulted margin of reactive diapirs formed in the footwalls of counter regional faults (Fig. 2f). A section through a relatively simple growth faulted delta is illustrated in Figure 3b, which crosses the SW part of the Baram Delta, NW Borneo. It shows a series of closely spaced, mixed regional and counter regional faults affect the lower part of the section (unit A) and directly overlie a unit interpreted
MOBILE SHALE RELATED DEFORMATION IN LARGE DELTAS
349
Fig. 11. Line drawing of a seismic line from the Niger Delta illustrating how different styles of diapir interact with growth faults through time.
to be overpressured, undercompacted Setap Formation. A second unit (B), full of prograding clinform sets forms a weakly faulted wedge overlying the faults. It is interpreted here that this represents rapid progradation across a formerly mobile shale unit that had been rendered relatively stable by dewatering accompanying the initial phase of faulting. Active large depocentre faulting began during unit C times at a counter regional fault, where clinoforms are stacked in its hangingwall. This indicates prograding shelf-slope sediments encountered an area where overpressured, undercompacted Setap Formation was thick enough to again permit the formation of fault-related depocentres. Subsequently the focus of deposition shifted further offshore and gave rise to a series of prograding regional fault depocentres, the most northwesterly one remains active today. As deltaic deposition matures, seaward propagating fault systems catch up with the slower propagating, or non-propagating (toe) diapir-belt structures and so then nucleate on the sites of older diapirs, following the models of Bruce (1973) and Dailly (1976), (Fig. 11). When this happens, the faultbounded depocentres tend to change character. Pseudo counter regional faults may develop at shale
rollers (Van Rensbergen & Morley 2000; McClay et al 2000). Depobelts nucleated on diapirs are often bounded by convergent conjugate faults and the depocentre is narrower (<20 km wide) than the asymmetic depocentres bounded by single, large regional or counter-regional faults . These characteristics are illustrated in Figure 11, which shows a complex example of how the present day structure of a section through the offshore area of the Niger Delta is inferred to have evolved. The youngest phase comprises a series of offshore younging depocentres that began with a counter regional fault, then a regional fault, the regional fault depocentre was uplifted (due to mobile shale movement) and a counter regional fault depocentre developed on the eroded regional fault depocentre. The last fault-bounded depocentres developed on top of a shale diapir, and form conjugate fault sets, that developed during diapir collapse. It appears that the diapirs and associated compressional thrusts remained fixed and active for longer than the normal fault depocentres. The faulting style appears to have been influenced by the diapir development. The section illustrates a complex interplay between the faulting and the mobile shales. The kinematics of growth faulting are not similar
350
C. K. MORLEY
everywhere on the delta. Although there is usually a tendency for faulting to become younger offshore. In some places, regional faults are dominant, in others counter regional faults are dominant. The result is a series of rhombic or diamond shape fault bounded blocks, with internal reflection packages that alternate thickening in opposite directions. The alternations between landward and seaward expanding reflection packages indicate that the regional and counter-regional fault activity has switched with time. The precise origin of these differences described above has not been well documented. But likely reasons for the regional variation are: (1) thickness of the mobile shale (due to depositional changes, earlier flowage, influence of pre-existing basement fabrics); (2) sedimentation rate - lower sedimentation rates passing towards the delta margin or the edges of delta lobes, may permit an oceanward dipping detachment to develop. Hence differential loading in the central parts of the delta may be augmented by gravity sliding passing towards the margins; and (3) as in other scales of conjugate fault systems the region of conjugate faults may reflect the co-lateral transfer zone between a system of predominantly counter regional faults, and another system of predominantly regional faults. Effects of growth faulting on sedimentation Three key ways in which growth faults influence sedimentation are: (1) they subside rapidly enough to stall the migration of facies boundaries (causing them to stack vertically in one place as discussed above); (2) they create bathymetric highs and lows that influence sediment pathways from the shelf to deep water and; (3) they create sea floor depressions that trap sediments moving under gravity (Fig. 2f). Sometimes the location of the sedimentary facies is not affected, but prograding clinoform sets may thicken across a growth fault situated at the shelf edge (Hiscott 2001). The ability of faults to create sea floor depressions seems to be variable. An investigation of regional seismic lines across the Niger Delta revealed very few faults that significantly affect the sea floor topography. This is partly because fault activity at any one time is confined to a narrow depobelt; it also indicates sedimentation tends to keep pace with displacement. The best example of a fault created depression comes from a counter regional fault (Fig. 2f). Such a depression ponds sediments moving under gravity, like turbidites, in the hanging wall close to the fault. With the counter regional fault located oceanward of the delta topset-foreset transition, rotation of the sea floor by the fault enhances the regional offshore slope. Conversely, rotation on a regional fault works against the regional slope. Consequently, counterregional faults are more likely to create sediment trap depressions and larger depressions than re-
gional faults (see, for example, the active regional fault described by Hiscott 2001).
Toe thrust and fold provinces Toe thrust belts are widely associated with deltas on passive margins. The youngest toe thrusts tend to be developed on the slope or in deep water, consequently there are few regions where extensive seismic reflection data has been gathered. However with frontier exploration now investigating deep water plays, toe thrusts in deltas may become significant plays (e.g. Hooper et al. 2002). The presence of a contractional toe thrust belt, tens of kilometres wide at the oceanward edge of a delta is a common feature related either to lateral shortening accommodating differential loading by sediments on the shelf, or gravity sliding. On active margins, there is sometimes the problem of distinguishing whether the toe thrust belt is due to subduction-accretionary prism tectonics or delta gravity tectonics. For example, toe thrusts associated with Neogene deltas coincide with the Palawan Trough offshore NW Borneo (e.g. Hinz et al 1989; Sandal 1996). This extensive thrust and fold belt could well be driven by a mixture of gravity (i.e. delta) and lithospheric (i.e. subduction) related stresses. Toe thrusts tend to be characterized by arrays of regularly spaced imbricate thrusts which sole out in a major detachment and have asymmetric, offshore verging hangingwall anticlines (fault bend and fault propagation folds) between the imbricates (Fig. 12). Commonly a uniform thickness pre-kinematic section is present in the lower parts of the structures and a syn-kinematic section is present which thins and onlaps the growing folds. Small extensional faults are commonly found around the crests of anticlines as sediments slump and slide off the growing anticlinal highs and sediment repose angles are exceeded. Ponded basins, which are possible traps for turbidite sands form on the flanks of the growing anticlines (Fig. 12). This simple description of the toe thrust belt while representative of a common style, does not do justice to the considerable lateral and temporal changes in structural style that exist in this setting (Fig. 13). Evamy et al (1978) and Doust & Omatsola (1990) have previously described the general distribution of the main structural zones in the Niger Delta, which passing from onshore to offshore are as follows: (1) fault-dominated depo-belts; (2) diapir dominated depo-belts; and (3) fold and thrust belt. In the Niger Delta, the thrust and fold zones form two arcuate zones that die out laterally (Fig. 2). The highly developed portions of the zones can be up to 70 km wide and contain numerous imbricate thrusts (Fig. 2). Shortening is in the order of 30-40 km. The thrust faults tend to be of similar size (horizontal displace-
MOBILE SHALE RELATED DEFORMATION IN LARGE DELTAS
351
Fig. 12. Cross-section through toe thrust system, Niger Delta, based on 2D seismic data, (modified from Morley & Guerin 1996).
ments generally range between 2-5 km, exceptionally up to 10 km) and no well developed large overthrusts or duplexes are present (Fig. 12). Laterally the well developed imbricate zones pass into regions of low shortening (less than 15 km) where only a few imbricates or pop-up and triangle zones are developed (Morley & Guerin 1996; Hooper^ al 2002). Considerable changes in the fold and thrust belt geometry and amount of shortening appear to be linked with variations in thickness of the mobile shale substratum and the overall geometry of the main delta lobes. Prior to the late Miocene the delta had two separate, eastern and western depocentres (Evamy et al 1978; Doust & Omatsola 1990), this pattern is still evident in the present day structural zonation (Fig. 2). Most noticeable is the absence of diapiric deformation in the central part of the delta and the lobate nature of the deep water fold and thrust belts (Figs 2 and 14). The amount of shale diapir involvement in the thrust belt changes markedly around the Niger Delta. Fringing the external margin of the diapir belt is a zone of compressional diapirs. More external still is a zone of imbricate thrusting, where mobile shales are not present, but the decollement is probably an overpressured shale unit (Fig. 2). Only on the eastern margin of the Niger delta does a narrow, mobile shale-involved compressive belt (Fig. 2) exist which also marks the thrust front. Despite the delta being up to 450 km wide (onshore-offshore), the area of active tectonics, defined as the distance from the active fault bounded depocentre, to the most oceanwards young thrusts, is quite narrow, commonly around 60 km. However, in some parts of the delta the thrust front can be up to 140 km away. It is not clear how or why stresses are transmitted so far from the depobelts before causing thrusting. However, it is likely the answer lies in the distribution of pore fluid pressures around the detachment zone and stress concentrations around basement fault blocks.
The Niger delta is known to have prograded several hundred kilometres offshore during its evolution (Evamy et al. 1978). It is reasonable to assume that the structural belts have also migrated offshore with time. Yet despite the good quality of the seismic data there is little evidence of extensive abandoned compressive toes towards the interior of the delta that have been overlain by the normal fault depobelts. Individual examples of isolated thrusts and folds can be found, but nothing on the scale of the modern fold and thrust belt is imaged (e.g. Hooper et al 2002). An explanation may be found in the laterally variable nature of the mobile shale substratum. In the past when the major fault bounded depocentres lay further landwards, extension was accommodated mostly by vertical movements within the diapir belt and to a lesser degree by compressional features on the external margins of diapirs. As the diapir belt evolved, it dewatered somewhat, and was prograded over by the growth fault belt, compressional deformation shifted further oceanwards into the modern region of toe thrusts. At the modern toe thrust region the shale for some reason (e.g. thinner overburden, less organic material in the section to cause sustained overpressuring; slower burial, better drainage, lower overpressures) did not become diapiric and the more regular imbricated toe thrusts of the modern delta developed. Today the delta shows a mixture of active thrust belts, in some places located on the margins of the diapir belt, in other places lying beyond the diapir belt (Fig. 2). In the main part of the delta the base of the mobile shale gently dips landwards or is horizontal. However, passing towards the margins of the delta, sedimentation is less and there is a transition towards the more typical passive margin geometry where the base of the sediment wedge and the sea floor dip offshore. Near the shelf break a series of regional (oceanward dipping) faults are developed that sole out into a detachment zone at the base of the wedge
352
C. K. MORLEY
Fig. 13. Idealized sketches across a toe thrust belt illustrating some of the variability seen in toes thrust. This ranges from thrust belts with a sole thrust and no shale diapir involvement, to shale diapir-dominated thrust belts, or a mixture of the two.
(Fig. 2). The wedge is approximately 2-3 km thick. Down slope the basal slope becomes more gently dipping and coincides with a change in structural style from the extensional regional faults to folds and minor thrusts. Thus the classic compressive toe to a gravity slide can be found on the margins of the main delta. The zone of extensional faulting is up to about 75 km wide, with the entire deformed zone about 150 km wide.
Effect of diapir and thrusted diapir belts on sedimentation pathways The two kinds of diapir province cover approximately the slope environment on the Niger Delta (Figs 2 and 14) and help to define the nature of the continental slope. Many of the diapir ridges in the main diapir province trend sub-perpendicular to the margin (Figs 2 and 14). Hence, they tend to guide the location of channels and canyons between the ridges. Near the base of the slope the thrusted diapirs give rise to margin parallel ridges. Channels tend to be more strongly incised to cross these diapirs, or are deflected around the ridges (Fig. 14). Many of the canyons and channels are most incised near the base or partway up the slope and do not seem to link with channels on the shelf. Incision would appear to be linked with slope instability associated with diapir activity in these cases. This notion is supported by the central-eastern part of the delta where diapir activity is minimal. The bathymetry in this area shows a smooth slope with no pro-
nounced channels (Fig. 14). The slope is also narrower and of higher gradient. One notable exception to the channels not reaching the shelf is the Niger Canyon. One pronounced difference between the Niger Delta and the Baram deltaic province is that the slope in the Niger Delta is largely underlain by shale diapirs (Fig. 14), whereas the slope in the Baram province is composed of active-inactive toe thrusts (Sandal 1996). In both areas, the slopes are of similar width (50-80 km). Whilst active diapirs exist in the Baram province they do not form a broad belt and most of the diapirs lie on the shelf or at the shelf-slope transition. There are not enough examples of active vs. passive margin deltas to determine whether this is a distinctive difference between the two types. However, the Mahakam Delta is another example of an active margin delta where growth faults and folds dominate and diapir development is weak (McClay et al 1998). Tectonically forced uplift and progradation associated with the active margin could be the critical factor in restricting diapir development by the rapid forced progradation of fault-dominated depobelts over the region that on a passive margin would evolve into the diapir belt.
Comparison of the deltas on active and passive margins The nature of active and passive margins is sufficiently different that variations in structural style,
Fig. 14. Bathymetry of the slope area of the Niger Delta (interpreted and depth converted from 2D seismic reflection data), superimposed on the structure map (Fig. 2) showing the close relationship between slope gradient, location, spacing and intensity of channels/canyons and diapir style/intensity.
354
CK.MORLEY
kinematics and impact of structure on sedimentation are likely to result. Some of the factors that need to be considered are discussed below. (1)
(2)
The volume and location of sediment input. On active margins, much smaller area drainage systems can produce thick deltaic depocentres. For example the Niger delta on the west African passive margin extends about 450 km in a NE-SW direction and 550 km in a NW-SE direction (e.g. Doust & Omatsola 1990). In places the deltaic sediments and associated deeper water sediments have reached a thickness in excess of 12 km. Approximately 1.3-1.5 X 106 km3 of Tertiary sediments fill the basin. The Middle Miocene-Recent Baram basin is smaller than the Niger Delta and extends about 250 km parallel to depositional strike and 200 km onshore-offshore (Fig. 3). Its exact dimensions are difficult to define because older deltaic sedimentary rocks in the hinterland have been uplifted and eroded and some deltaic sediments extend further NE along the Sabah margin. However, there is clearly an isopach thick centred around the area that today is the Baram delta. The volume of Middle Miocene-Recent sediment is estimated to be about 2.3 X 105km3. The Niger delta built out as two lobes that have merged, as the delta associated with a large single river has undergone channel switching. The Baram deltaic basin is different, while some large rivers can be identified, there is no strongly lobate pattern and there appears to have been more of a linear front of deposition. However there are structural divisions within the basin (more dominance of regional faults to the SW more influence of counter-regional faults to the NE) that can be attributed to different river and delta systems feeding into the basin (Sandal 1996). The active tectonic setting created numerous drainage systems coming off the uplifted interior and the progressive uplift of the deformation front, which forced the coalesced deltas further ocean wards. The lower sediment volume Baram basin has much narrower structural belts than the Niger Delta. Nevertheless it is still possible to determine a growth fault belt, diapir belt and toe thrust belt (Fig. 3). One distinguishing feature of active margin diapirs from those in the Niger Delta is the presence of inversion anticlines (e.g. Sandal 1996; McClay et al 2000; Figs 2 and 3). These are distinguished from rollover anticlines by uplift of horizons above regional at the fold axis, whereas for rollover anticline
(3)
(4)
horizons at the fold axis lie below regional. These anticlines, with multiple phases of growth, are the main traps for hydrocarbons in Brunei (e.g. Champion, Seria, SW Ampa, Iron Duke, Magpie fields; Sandal 1996). Similar folds are present in the Mahakam Delta in Kalimantan (e.g. McClay et al. 2000). They are completely different from the classic rollover anticline traps of the Niger Delta (e.g. Evamy 1978), however more classic growth fault traps also exist in the Baram Delta. In the Baram basin there is marked interaction between NE-SW striking inversion anticlines and the older thrust-related trends (N-S). These lie oblique to the NE-SW coast-parallel gravity tectonics trends (Fig. 3) as stresses changed during the evolution of the delta (Morleyeffl/,2003). One effect of the growing structures, particularly the oblique N-S trends interacting with NE-SW trends, is to create embayments in the synformal areas between anticlines. These embayments can be broad features tens of kilometres across. They create protected regions (from wave processes) that enable tide dominated deposition to occur in parts of the delta, while on the open marine coastline mixed, tidal, fluvial and wave processes operate (e.g. Lambiase et al. in review).
Conclusions Increased seismic resolution combined with the results of physical modelling indicate that intrusions associated with shale diapirs can be classified into true shale diapirs, as well as fluid pipes and gas clouds. True shale diapirs show features similar to those described from salt tectonics (reactive, active, passive and collapse features). However, they are commonly modified by lateral and vertical fronts of fluids and shale intrusions into stratified sequences that add to the apparent size of the diapir. Very commonly (but not always) the major depocentres comprise an oceanward belt of predominantly counter regional faults and a landwards belt of regional faults. The duration of fault activity is dictated by the ability of the mobile shale to create accommodation space by evacuating beneath the fault depocentre and the sediment supply rate to the depocentre. The active counter-regional faults tend to lie oceanward of the delta topset-foreset transition, in predominantly mud-prone sediments. Commonly the counter-regional fault can influence sedimentation by stalling progradation and forcing aggradation and by creating sea floor hangingwall depressions, which act as sediment traps for (possibly sand-prone) gravity deposits.
MOBILE SHALE RELATED DEFORMATION IN LARGE DELTAS
Differences in active and passive margin settings for deltas are manifest in many different ways including: (1) active margins may display shorter drainage systems and much higher rates of erosion per unit area than passive margins; (2) active uplift of the hinterland is an additional factor to force delta progradation not seen on passive margins; (3) uplift causes erosion and recycling of older deltaic deposits which can cause comparatively fine-grained sandstone reservoirs (Sandal 1996); (4) growing structures in the Niger Delta may locally create depressions and create changes in the dip of slopes. In the Bar am basin growing inversion structures show evidence for sub-aerial erosion and have caused the development of protected embayments that create distinct changes in the dominant depositional (wave, tide fluial) process along the margin. The escalator regression model is an important part of delta tectonics but there is an implication that the progradation of depocentres is basically repetition of the same geometry and process as the delta progrades offshore. Here it is argued that the nature of the substratum over which the delta progrades changes considerably from onshore to offshore (from the early to the mature stages of the delta) and along depositional strike. Consequently it is emphasized that a deeper understanding of deltaic structure requires an appreciation of the important changes in structural style and evolution that occur spatially and in time, including: (1) (2)
(3)
(4)
The development of early-stage closely spaced faults that are later prograded over by the major depocentres (Fig. 3). Along depositional strike of the delta there may be a change from differential loadingdominated deformation in the main part of the delta, to regional fault imbricates developed above an offshore dipping detachment (gravity sliding). This lateral transition may also lead to counter-regional fault depocentres being dominant in the thickest regions of the delta, whilst passing towards the margins (where the offshore passive margin slope starts to develop) regional faults tend to be dominant. The fault bounded depocentres developed closed to the continent may generate the rise of reactive diapirs and be bounded on their oceanward margin by toe diapirs. With progradation, later fault depocentres may be located on the (older) toe diapirs (causing their collapse) and force the development of detachment-related folds and thrusts oceanwards of the thick, over pressured shale sequences. Different parts of the delta may be at different stages of development. The early stages of delta development may occur on thin to non-existent marine shales,
355
(e.g. onshore Baram Basin), consequently gravity-driven deformation does not occur. As the delta progrades offshore there is a transition to gravity-driven deformation, across an increasingly thick wedge of overpressured shales. Regional faults may be the first fault sets of develop where the overpressured shales are thin and give way to counter-regional faults further offshore, where the overpressured, mobile shales are thicker. Sediment pathways to deeper water are likely to be modified considerably as the delta evolves. (5) Seismic reflection data across the Niger Delta indicates that today sediment loading, fluid escape and shale diapirism has driven the mobile shales which underlay the depocentres into large and small isolated pockets, separated by touchdowns and welds. Consequently it is difficult to ascertain the original thickness and distribution of the mobile shales. However, despite this qualifier, in certain areas of the delta the absence of deformation and the absence of major deopcentres indicates that mobile shales never developed to any great thickness. The inference is supported by the passage of stratified (reflective) sequences laterally into chaotic, (inferred) mobile shale units. Sometimes the absence of mobile shales is associated with the presence of underlying basement fault block highs. Consequently, it appears wrong to assume large deltas prograde over a ubiquitous, thick mobile shale substratum. While mobile shales are clearly very thick and extensive in places, their distribution (or the distribution of overpressure) is somewhat patchy, which consequently affects deformation style. (6) Structural style appears to play a role in the nature of the slope (slope gradient and density and location of channels) and hence how sediment is transported to deeper water. The gradient of the slope changes across the Niger Delta according to structural style. On the sediment starved NW margins of the delta the slope is gradient is near 5°. In the main part of the delta the slope is lower (1°-1.3°) and more incised where shale diapirs are active and steeper (1.3°-1.5°) and smoother in the central area where shale diapirs activity is minor. The slope region in Brunei where shale diapir activity is minor and toe thrusts are present the slope approaches angles of 2°. I would like to thank my colleagues and workers at UBD, Elf Aquitaine, BSP Shell, TotalFinaElf and Amoco for numerous discussions about the structure and sedimentology over the years and for providing data that have greatly contributed to my understanding of deltaic provinces.
356
C. K. MORLEY
Hydrocarbon habitat of Tertiary Niger delta. American Association of Petroleum Geologists Bulletin, 62,1-39. FLOOD, R.D., PIPER, DJW. & Shipboard scientific party 1995. Introduction. In: FLOOD, R.D., PIPER, D.J.W. & KLAUS, A. Proceedings of the Ocean Drilling Program, initial report, 155,5-15 GE, H.X., JACKSON, M.P.A. & VENDEVILLE, B.C. 1997. References Kinematics and dynamics of salt tectonics driven by BACK, S., MORLEY, C.K., SIMMONS, M.D. & LAMBIASE, JJ. progradation. American Association of Petroleum Geologists Bulletin, 81, 398-423. 2001. Depositional environment and sequence stratigraphy of Miocene deltaic cycles exposed along the GIBSON, R.G. 1994. Fault-zone seals in siliciclastic strata of Jerudong anticline, Brunei Darussalam. Journal of the Columbus basin, offshore Trinidad. American Association of Petroleum Geologists Bulletin, 79, Sedimentary Research, 71, 915-923. BARBER, A.J., TJOKROSAPOETRO, S. & CHARLTON, T.R. 1372-1385. 1986. Mud volcanoes, shale diapirs, wrench faults and GRAUE, K. 2000. Mud volcanoes in deepwater Nigeria. Marine and Petroleum Geology, 17, 959-974. melanges in accretionary complex, eastern Indonesia. American Association of Petroleum Geologists GULIEV, I.S. 1992. A review of mud volcanism. Translation Bulletin, 70,1729-1741. of the report by: Azerbaijan Academy of Sciences BARKER, C. 1990. Calculated volume and pressure changes Institute of Geology, 65pp. during thermal cracking of oil and gas in reservoirs. HASKELL, N., NISSEN, S., HUGHES, M., GRINDHAUG, J., American Association of Petroleum Geologists DHANANI, S., HEATH, R., KANTOROWICZ, J., ANTRIM, Bulletin, 74,1254-1261. L., CUBANSKI, M., NATARAJ, R., SCHELLY, M. & BISHOP, R.S. 1978. Mechanism for emplacement of pierceWIGGER, S.T. 1999. Delineation of geological drilling ment diapirs. American Association of Petroleum hazards using 3-D seismic attributes. The Leading Geologists Bulletin, 46,233-245. Edge, March 1999,373-382. BROWN, K.M. 1990. The nature and hydrogeological sigHEDBERG, H.D. 1974. Relation of methane generation of nificance of mud diapirs and diatremes for accretionundercompacted shales, shale diapirs and mud volcaary systems. Journal of Geophysical Research, 95, noes. American Association of Petroleum Geologists 8969-8982. Bulletin, 58,661-673. BRUCE, C.H. 1973. Pressured shale and related sediment HINZ, K., FRTTSCH, J., KEMPTER, E.H.K., MANAF deformation: mechanism for development of regional MOHAMMAD, A., MEYER, J., MOHAMMAD. D., contemporaneous faults. AAPG Bulletin, 57, 878-886 VOSBERG, H., WEBER, J. & BENANVIDEZ, J. 1989. CARTWRIGHT, J., BOUROULLEC, R., JAMES, D. & JOHNSON, Thrust tectonics along the northwestern continental H. 1998. Polycyclic motion history of some Gulf margin of Sabah/Bomeo. Geologisch Rundschau, 78, Coast growth faults from high-resolution displace705-730. ment analysis. Geology, 26, 819-822. HISCOTT, R N. 2001. Depositional sequences controlled by COHEN, H.A., & McCLAY, K. 1996. Sedimentation and high rates of sediment supply, sea-level variations, shale tectonics of the northwestern Niger Delta front. and growth faulting: the Quaternary Baram Delta of Marine and Petroleum. Geology, 13, 313-328. northwestern Borneo. Marine Geology, 175,67-102. COLEMAN, J.M & PRIOR, D.B. 1980. Deltaic sand bodies. HOOPER, R.J., FITZSIMMONS, R.J., GRANT, N. & VENDEAmerican Association of Petroleum Geologists, VILLE, B.C. 2002. The role of deformation in controlContinuing Education Course Notes Series, 15, 171p. ling depositional patterns in the south-central Niger DAVIES, J.L. 1980. Geographical variations in coastal Delta, West Africa. Journal of Structural Geology, 24, development. Longman, London, 212p. 847-859. DAILLY, G. C. 1976. A possible mechanism relating progra- JACKSON, M.P.A. & TALBOT, C.J. 1986. External shapes, dation, growth faulting, clay diapirism and overthruststrain rates and dynamics of salt structures. ing in the regressive sequence of sediments. Bulletin Geological Society of America Bulletin, 97, 305-323. of Canadian Petroleum Geology, 24,92-116. JACKSON, M.P.A. & VENDEVILLE, B.C. 1994. Regional DOUST, H. & OMATSOLA, E. 1990. Niger Delta. In: extension as a geological trigger for diapirism. EDWARDS, J.D. & SANTOGROSSI, PA. (eds) Divergent/ Geological Society of America Bulletin, 106,57-73. passive margins, American Association of Petroleum JAMES, D.M.D. 1984. The Geology and Hydrocarbon Geologists Memoir, 48, 201-238. Resources of Negara Brunei Darussalam. Special DE GAMERO, M.L.D. 1996. The changing course of the Publication, Muzium Brunei and Brunei Shell Orinoco River during the Neogene: a review. PalaeoPetroleum Company Berhad. geography, Palaeoclimatology and Palaeoecology, KHALIVOV, N.Y., & KERIMOV, A.A. 1983. Origin of mud 123, 385-402. volcanism and diapirism. International Geology DUVAL, B., CRAMEZ, C. & JACKSON, M. P. A. 1992. Raft tecReview, 25, 877-881. tonics in the Kwanza Basin, Angola. Marine and KATZ, B., RICHARDS, D., LONG, D. & LAWRENCE, W. 2000. Petroleum Geology, 9, 389-404. A new look at the components of the petroleum EDWARDS, M.B. 1976. Growth faults in upper Triassic system of the South Caspian Basin. Journal of deltaic sediments, Svalbard. American Association of Petroleum Science and Engineering, 28,161-182. a Petroleum Geologists Bulletin, 60, 341-355. KEHLE, R.O. 1989. The origin of salt structures. In: EVAMY, B.D., HAREMBOURE, J., KAMERLING, P., KNAAP, SCHREIBER, C.B. (ed.) Evaporites and hydrocarbons. W.A., MOLLOY, F.A., AND ROWLANDS, PH. 1978. Columbia University Press, New York, 345-^-04. T. Dooley and S. Stewart are thanked for very helpful, thorough reviews of this manuscript. S. Stewart is also thanked for providing a better example of a gas cloud than my original one.
MOBILE SHALE RELATED DEFORMATION IN LARGE DELTAS KNOX, GJ. & OMATSOLA, E.M. 1989. Development of the Cenozoic Niger Delta in terms of the 'Escalator Regression' model and impact on hydrocarbon distribution. In: VAN DER LINDEN, W.J.M., CLOETINGH, S.A.P.L., KAASSCHIETER, J.P.K., & VAN DER GUN, J.A.M. (eds) Proceedings KNGMG symposiums coastal lowlands, geology and geotechnology. Kluwer Academic Publishers, Amsterdam, 181-202. LAMBIASE, J.J., DAMIT, A.R., SIMMONS, M.D., ABDOERRIAS, R. & HUSSIN, A. in press. Modern and ancient tidedominated deltas of NW Borneo. Society for Sedimentary Geology Special Publication. MADON, BIN HJ. & RENDZUAN BIN ABU HASSAN 1999. West Luconia Province. The Petroleum Geology and Resources of Malaysia, Petronas, Kuala Lumpur, 429-439 McCLAY, K.R., DOOLEY, T. & LEWIS, G. 1998. Analog modeling of progradational delta systems. Geology, 26,771-774. McCLAY, K.R., DOOLEY, T., FERGUSON, A. & POBLET, J. 2000. Tectonic evolution of the Sanga Sanga Block, Mahakam Delta, Kalimantan, Indonesian Borneo. American Association of Petroleum Geologists Bulletin, 84,765-786. MCCLAY, K., DOOLEY, T. & ZAMORA, G. 2003. Analogue models of delta systems above ductile substrates. In: VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, AJ. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216,411-428. MERKI, P. 1972. Structural geology of the Cenozoic Niger Delta. In: DESSAUVAGIE, T.F.J. & WHITEMAN, AJ. (eds) African Geology. Ibadan University Press, 635-646. MORGAN, J.P, COLEMAN, J.M. & GAGLIANO, S.M. 1968. Mud-lumps: diapiric structures, in Mississippi Delta sediments. In: BRAUNSTEIN, J. & O'BRIEN, G.D. (eds) American Association of Petroleum Geologists Memoir, 8,145-161. MORLEY, C.K. & GUERIN, G. 1996. Comparison of gravitydriven deformation styles and behaviour associated with mobile shales and salt. Tectonics, 15, 1154-1170. MORLEY, C. K., CREVELLO, P. & ZULKIFLI A. 1998. Shale tectonics and deformation associated with active diapirism: the Jerudong Anticline, Brunei Darussalam. Journal of the Geological Society of London,155,475-490. MORLEY, C.K., BACK, S., VAN RENSBERGEN, P., CREVELLO, P. & LAMBIASE, JJ. in press. Characteristics of repeated tectonic inversion events during the Miocene-Recent in a large delta province on an active margin, Brunei Darussalam, Borneo. Journal of Structural Geology. MORTON, R.A. & GALLOWAY, W.E. 1991. Deposition, tectonic and eustatic controls on hydrocarbon distribution in divergent margin basins - Cenozoic Gulf of Mexico case history. Marine Geology, 102, 239-263. OSBORNE, MJ. & SWARBRICK, R.E. 1998. Mechanisms for generating overpressure in sedimentary basins: a reevaluation. American Association of Petroleum Geologists Bulletin, 81,1023-1041. PATERSON, D.W. 1997. Petroleum Systems of the Kutei
357
Basin, Kalimantan, Indonesia. IPA Petroleum Systems Symposium, 709-726. PEREZ-BELZUZ, E, ALONSO, B. & ERCILLA, G. 1997. History of mud diapirism and trigger mechanisms in the Western Alboran Sea. Tectonophysics, 282, 399-422. RIDER, M.H. 1978. Growth faults in Carboniferous of Western Ireland. American Association of Petroleum Geologists Bulletin, 62, 2191-2213. SANDAL, S.T. 1996. The Geology and Hydrocarbon Resources of Negara Brunei Darussalam, (1996 revision). Brunei Shell Petroleum Company/Brunei Museum, Syabas Bandar Seri Begawan, Brunei Darussalam, 243 pp. SHAMSUDDIN, A.H.M AND ABDULLAH, S.K.M. 1997. Geologic evolution of the Bengal Basin and its implications for hydrocarbon exploration in Bangladesh. Indian Journal of Geology, 69, 93-121. SIBSON, R. 1996. Structural permeability of fluid-driven fault-fracture meshes. Journal of Structural Geology, 18,1031-1042. STANLEY, D.J & WARNE, A.G. 1998. Nile delta in its destruction phase. Journal of Coastal Research, 14, 794-825. STEWART, S.A. 1999. Seismic interpretation of circular geological structures. Petroleum Geoscience, 5,273-285. VAN DER WEERD, A.A. & ARMIN, R.A. 1992. Origin and evolution of the Tertiary hydrocarbon bearing basins in Kalimantan (Borneo), Indonesia. American Association of Petroleum Geologists Bulletin, 76, 1778-1803. VAN RENSBERGEN, P. & MORLEY, C.K. 2000. 3D seismic study of a shale expulsion syncline at the base of the Champion delta, offshore Brunei and its implications for the early structural evolution of large delta systems. Marine and Petroleum Geology, 17, 937-958. VAN RENSBERGEN, P., MORLEY, C.K., ANG, D.W, HOAN, T.Q. & LAM, N.T. 1999. Structural evolution of shale diapirs from reactive rise to mud volcanism: 3D seismic data from the Baram delta, offshore Brunei Darussalam. Journal of the Geological Society, 156, 633-650. VENDEVILLE, B. & COBBOLD, P. 1987. Synsedimentary gravitational sliding and listric normal growth faultsinsights from scaled physical models. Compte Rendus Acadamie de Science, 305,1313-1319. VENDEVILLE, B.C. & JACKSON, M.P.A. 1992a. The rise of diapirs during thin-skinned extension. Marine and Petroleum Geology, 9, 331-353. VENDEVILLE, B.C. & JACKSON, M.P.A. 1992b. The fall of diapirs during thin-skinned extension. Marine and Petroleum Geology, 9, 354-371. WARREN, J. 1999. Evaporites, Blackwell Science, Oxford, 438pp. WEVER, H.E. 2000. Petroleum and source rock characterization based on C-7 star plot results: examples from Egypt. American Association of Petroleum Geologists Bulletin, 84,1041-1054. WOOD, L. J. 2000. Chronostratigraphy and tectonostratigraphy of the Columbus Basin, eastern offshore Trinidad. American Association of Petroleum Geologists Bulletin, 84,1905-1928.
This page intentionally left blank
Pore pressure/stress coupling and its implications for rock failure RICHARD R. HILLIS National Centre for Petroleum Geology and Geophysics, Australian Petroleum Cooperative Research Centre, University of Adelaide, SA 5005, Australia (e-mail: rhillis @ ncpgg. adelaide. edu. au) Abstract: Clastic dykes and sills witness that subsurface sediment mobilization is often controlled by the brittle failure of units 'sealing' overpressured and liquidized sediments. Brittle failure also imposes a limit on the buoyancy pressure that can be exerted by hydrocarbon columns. Conventional understanding of brittle failure induced by increasing pore pressure (Pp) assumes that total minimum horizontal stress (o^) is unaffected by changes in pore pressure. However, total minimum horizontal stress increases from shallow, normally pressured sequences to deeper, overpressured sequences. Data from the Canadian Scotian Shelf, the North Sea and the Australian North West Shelf demonstrate such Pp/crh coupling, with the minimum horizontal stress increasing at approximately 60-80% of the rate of pore pressure (i.e., A<7h/APp = 0.6-0.8). Hence, a greater increase in pore pressure can be sustained prior to brittle failure of units sealing overpressured compartments than would be predicted by conventional, uncoupled failure models. Furthermore, because total vertical stress is not similarly coupled to pore pressure, differential stress (o-rcr3) reduces as pore pressure increases in normal fault regime basins. Thus, the mode of rock failure can not be inferred from differential stress in the stable state and Pp/crh coupling promotes tensile over shear failure.
Introduction The decrease in strength required for sediment liquidization is most commonly ascribed to overpressure development (Maltman & Bolton 2003). The mobilization of liquidized sediments requires the failure, often brittle, of units 'sealing' liquidized sediments, as, for example, in the case of shale and sandstone dykes and sills (Harms 1965; Cosgrove 2001; Jolly & Lonergan 2002). Shale and sandstone intrusions occur in association with both tensile factures and faults (Huang 1988; Morley et al 1998; Jolly & Lonergan under review). The theory of brittle failure has been extensively discussed (e.g., Engelder 1993; Sibson 1996 and references therein), and the critical role of fluids in the dynamics of fault and fracture initiation and reactivation has long been recognized (Hubbert & Rubey 1959). It is not total applied stress, but effective stress (difference between total applied stress and pore pressure) that controls rock deformation. Thus increasing pore pressure can result in rock failure (Fig. 1). The mode of rock failure that occurs in response to increasing pore pressure is dependent on the differential stress (01-03) at failure. Assuming a composite GriffithCoulomb failure envelope, failure occurs in (Sibson 1996; Fig. 1): •
shear if differential stress is relatively high:
•
tension if differential stress is relatively low:
•
hybrid tension/shear if differential stress is
Fig. 1. The influence of increasing pore pressure on rock failure assuming that changes in pore pressure do not affect total stresses. Increasing pore pressure reduces the effective normal stress (
The conventional understanding of failure induced by increasing pore pressure assumes that Mohr circle moves, with unchanged diameter, towards the failure envelope as pore pressure increases and the effective stress decreases (Secor 1965; Hobbs et al 1976; Cosgrove 1995; 2001; Fig. 1). This conventional approach implies that the differential stress at pore pressures at which the rock is stable is the same as that at elevated pore pressures associated with failure and thus that the mode of failure (shear, hybrid or tensile) can be predicted from the differential stress in the stable state (Hobbs et al 1976; Cosgrove 2001; Fig. 1). This assumes that total
Pp
Pore pressure (MPa) Maximum principal stress (MPa) Minimum principal stress (MPa) Differential stress (MPa) Effective maximum principal stress (o~{ — Pp) (MPa) Effective minimum principal stress (cr3 — Pp) (MPa) (Total) minimum horizontal stress (MPa) (Total) maximum horizontal stress (MPa) (Total) vertical horizontal stress (MPa) Effective minimum horizontal stress (o~h — Pp) (MPa) Effective vertical stress (o~v — Pp) (MPa) Change in minimum horizontal stress (MPa) Change in pore pressure (MPa)
—
Average gradient of minimum horizontal stress from surface to depth of interest (z) (MPa/km)
p -^
Average gradient of pore pressure from surface to depth of interest (z) (MPa/km)
A—
Change in average minimum horizontal stress gradient (MPa/km)
p A—2
Change in average pore pressure gradient (MPa/km)
Am, —— P k p |x T
Ratio of change in minimum horizontal stress to change in pore pressure, i.e., pore pressure/stress coupling constant (unitless) Effective stress co-efficient, an empirical constant given by ratio of o~h'/o~v' (unitless) Poisson's ratio (unitless) Co-efficient of sliding friction on fault (unitless) Tensile strength (MPa)
minimum horizontal stress is unaffected by changes in pore pressure. However, total minimum horizontal stress is coupled to (increases with) pore pressure (Breckels & Van Eekelen 1982; Lorenz et al 1991; Hillis 2001; Fig. 2). Pore pressure/stress coupling precludes inferring the likely mode of rock failure from differential stress in the stable state because differential stress may reduce as pore pressure increases. Pore pressure/stress coupling also leads to a significantly greater increase in pore pressure being sustained before rock failure than would be predicted by conventional, uncoupled models. This paper presents pressure data from three sedimentary basins demonstrating Pp/o-h coupling, and discusses the implications of Pp/crh coupling for rock failure due to increasing pore pressure. Pore pressure/stress coupling also has implications for hydrocarbon exploration in overpressured provinces. The 'window' between pore pressure and minimum horizontal stress is a major consideration for exploration in overpressured regions because it limits the height of hydrocarbon columns that can develop If a column exerts a buoyancy pressure greater than this window , seal failure by tensile fracturing may occur. Indeed, Gaarenstroom et al
(1993) defined the retention capacity of traps in the Central North Sea as the difference between prevailing pore pressure and minimum horizontal stress. Although useful locally with calibration, the retention capacity concept needs refinement for global application because it does not incorporate the possibility of seal breach due to shear fracturing/reactivation (Hillis 1998; Finkbeiner etal 2001), nor does it incorporate the influence of P/OH coupling. This paper also discusses the implications of Pp/o-h coupling for the concept of retention capacity, Relations between pore pressure and stress The ratio, (k), of the effective minimum horizontal stress, (crh — Pp), to the effective vertical stress, (crv - Pp), has been widely used to describe the stateof-stress in sedimentary basins:
Re-arranging Eq. 1 yields
PORE PRESSURE/STRESS COUPLING
361
stress with time and if the vertical stress is constant with such changes (weight of overburden), Eq. 3 yields:
Pressure data from sedimentary basins are collected over a range of depths and the relationship between pore pressure and minimum horizontal stress in basin-scale data primarily reflects the increase with depth exhibited by both (Fig. 2a). Hence, it is necessary to depth-normalize basin-scale data, as pressure gradients, in order to investigate Pp/crh coupling (Fig. 2b). Expressing Eq. 2 in terms of pressure gradients with respect to depth (z):
Similarly, assuming shallow, normally pressured sequences are representative of deep, overpressured sequences prior to overpressure development (Engelder & Fischer 1994), Eq. 4 can be re-written in terms of changes in pressure gradients with time for overpressure development in sedimentary basins (Fig. 2):
Fig. 2. Predicted Pp/o-h coupling based on Eq. 2. (a) Predictions on a pressure versus depth plot for an upper, hydrostatically pressured interval (Pp = 0.44crv), an intermediate depth, moderately overpressured interval (Pp = 0.7lo-v), and a deep, highly overpressured interval (Pp = 0.88o-v). (b) The same data plotted as pore pressure gradient versus minimum horizontal stress gradient. Each of the three compartments plots as a single point (for a given k).
or,
the former being equivalent to the standard fracture gradient relation (Traugott 1997). Considering changes in pore pressure and minimum horizontal
Eq. 6 assumes that the vertical stress gradient is constant with changes in pore pressure and stress. The assumption that shallow, normally pressured sequences are representative of the deeper, overpressured sequences prior to overpressure development is analogous to that made in basin modelling where present-day porosity-depth or porosityeffective vertical stress relations are used to describe the porosity evolution of a sediment with progressive burial. There have been a number of different approaches to determining the constant, k. It has been most widely determined based on the poroelastic response of rocks under uniaxial strain conditions, i.e., no lateral expansion (e.g. Engelder & Fischer 1994), whereby:
where v is Poisson's ratio. In Holbrook's (1997) solidity approach, the constant k is given by the complement of porosity, i.e., solidity. Alternatively, k has the value unity if the sediment deforms in the plastic domain (Schneider et al. 1999). Finally, the constant k can be determined based on the assumption that rock stresses are in equilibrium with those required to induce frictional failure (Zoback & Healy 1984; Finkbeiner et al 2001). Assuming that
362
R.R. HILLIS
there exist suitably oriented faults of no cohesion and a normal fault regime typical of passively subsiding sedimentary basins:
where JUL is the co-efficient of sliding friction on the fault. In Eqs 1-4, Biot's poroelastic constant (a) has been assumed be unity, as is commonly the case. However, the above can easily be adapted to cases where a is not unity by replacing Pp with oPp and APp with «APp. Herein k is considered a constant calibrated by measured pressure data. The extent to which k relates to Poisson's ratio, solidity, the co-efficient of rock friction, or indeed equals unity, depends on the extent to which the assumptions in the various models are satisfied. It is not the purpose of this paper to discuss these assumptions. However, the assumptions are not likely to be fully satisfied in any of the models, hence, as suggested by Mouchet & Mitchell (1989), values for k determined by calibration to measured pressure data should not be considered to yield the physical properties pertaining to the different models. The empirical approach favoured herein has been widely followed (e.g. Breckels & Van Eekelen 1982). Engelder & Fischer (1994) argued that pore pressure and minimum horizontal stress data from the UK North Sea and Canadian Scotian Shelf were consistent with the poroelastic model, but not with the frictional failure model. However, as pointed out by Zoback et al (1995), the predictions of frictional failure do indeed also fit these data, as could the predictions of Holbrook's (1997) solidity model. Given that all these models result in the relations presented in Eqs 1-4, they cannot be differentiated based on how well they fit measured pressure data in a normal fault regime basin. Herein it is simply noted that provided k is less than one, the models predict that changes in pore pressure are accompanied by changes in minimum horizontal stress.
Pressure data demonstrating pore pressure/stress coupling An increase in minimum horizontal stress from shallow, normally pressured sequences into deeper, overpressured sequences has been demonstrated by pressure measurements in the Scotian Shelf offshore eastern Canada (Bell 1990; Yassir & Bell 1994; Fig. 3a), and in the Central North Sea (Gaarenstroom et al. 1993; Fig. 3b). In both basins, minimum horizontal stress increases with pore pressure and in zones of extreme overpressure, minimum horizontal stress approaches the vertical stress (Fig. 3). An increase of minimum horizontal stress with pore pressure has
also been demonstrated in the US Gulf Coast and in Brunei (Breckels & Van Eekelen 1982). This section of the paper summarizes the type of pressure data obtained from oil exploration drilling that are used to demonstrate Pp/crh coupling, re-analyses the data presented by Bell (1990) for the Canadian Scotian Shelf and presents new data from the North Sea and the Australian North West Shelf that also demonstrate Pp/crh coupling. Hydraulic fracture tests, in which increasing the fluid pressure within an isolated section of a wellbore opens a tensile fracture, yield the most reliable determinations of minimum horizontal stress. The fluid pressure at which a hydraulic fracture closes provides a direct estimate of minimum horizontal stress, based on the assumption that the fluid is holding the fracture open against the least principal stress (e.g. Engelder 1993). Unfortunately, hydraulic fracture tests are not widely undertaken during exploration drilling. However, leak-off tests are routinely undertaken. In a leak-off test, the pressure required to open a fracture in a few metres of open hole immediately beneath a casing shoe is determined. Leak-off (or fracture) pressures are routinely determined because they give an indication of the maximum pressure of drilling mud that can be used without generating fractures into which drilling mud would be lost. Leak-off pressures do not yield as reliable estimates of minimum horizontal stress as fracture closure pressures, largely because the disturbed stress field at the wellbore wall controls the leak-off pressure, and because the leak-off pressure must overcome any tensile strength of the formation. Nonetheless, it is widely accepted that the lower bound to leak-off pressures gives a reasonable estimate of minimum horizontal stress (Breckels & Van Eekelen 1982; Gaarenstroom et al 1993; White et al 2002). Herein leak-off pressures are used as a proxy for minimum horizontal stress. Direct pressure measurements from wireline formation interval tests provide the most reliable, commonly available measurements of pore pressure. However, such are not generally available at the same depth as a leak-off test. Furthermore, rock permeability must be at least 1-10 mD to obtain a reliable pressure build-up inside the probe and thus reliable pore pressure measurement (Swarbrick & Hillis 1999). Given that reliable pore pressures from wireline formation interval tests are almost invariably significantly depth-offset from leak-off test pressures and that the transition zones from normally to overpressured intervals may be narrow (Swarbrick & Osborne 1996), such are not generally useful in determining pore pressure at the depth of a leak-off test. Mud pressures in the open hole are generally kept slightly in excess of formation pore pressures to prevent the entry of formation fluids into the well-
PORE PRESSURE/STRESS COUPLING
363
Fig. 3. Pressure versus depth plots for (a) Canadian Scotian Shelf, after Bell (1990) and (b) Central North Sea, after Gaarenstroom et al. (1993). LOP: leak-off pressure, which is taken as a proxy for minimum horizontal stress (see text for further discussion).
bore. Mud pressures are invariably raised when elevated pore pressures are encountered and hence reflect overpressure. Mud pressures may be increased for reasons other than elevated pore pressures (e.g. to maintain wellbore stability), but, in general, mud pressures significantly higher than formation pore pressure are avoided because of the resultant reduced rate of penetration, additional mud costs and potential for formation damage. The disadvantages of using mud pressures at the depth of the leak-off test to analyse Pp/a"h coupling are considered less than those of using depth-offset wireline formation interval tests pressures. Hence the pressure of drilling mud in the open hole at the depth of the leak-off test whilst drilling ahead has thus been used as an estimate for pore pressure in the vicinity of the leak-off tests (i.e. where minimum horizontal stress has been determined). Bell's (1990; Fig. 3a) data for the Scotian Shelf offshore eastern Canada comprises leak-off pressures and mud pressures, although in some cases for-
mation interval or drill stem test pressures were available at the depth of leak-off tests. Bell's (1990) data has been re-plotted as minimum horizontal stress gradient (MPa/km) against pore pressure gradient (MPa/km) in order to depth normalize the pressures and elucidate the nature of Pp/crh coupling (Fig. 4a). Linear regression of these data suggests that minimum horizontal stress increases at approximately 76% of the rate that pore pressure increases (i.e. Ac7h/APp~ 0.76). There is significant scatter, particularly at near hydrostatic pore pressure gradients of approximately 10 MPa/km, where most data points are available. However, this scatter is to be expected given that leak-off pressure is an approximation for minimum horizontal stress, and mud pressures an approximation for pore pressure. Considerable scatter is also likely to be induced by the variety of lithologies sampled. Different lithologies exhibit different material properties (e.g. v and fx) and are therefore likely to exhibit different Pp/o-h coupling
364
R.R. HILLIS relations. Larger datasets might enable lithologyspecific Pp/crh coupling relations to be determined for individual basins. Data for the Gannet/Guillemot fields of the North Sea (pers. comm. Van Eekelen 1996), which are based on leak-off and mud pressures, are also scattered but reveal Pp/crh coupling with A(jh/APp~0.60 (Fig. 4b). Pore pressures in Gaarenstroom et al!s (1993; Fig. 3b) regional compilation of Central North Sea pressure data are based on formation interval tests. However, these pore pressures are not available for the same depths as the leak-off pressures. In the absence of pore pressure and stress data from the same depth in the same well, although providing a broad indication of Pp/crh coupling, Gaarenstroom et al.'s (1993) compilation should not be used to quantify Pp/crh coupling. Leak-off and mud pressure data from the Australian North West Shelf show considerable scatter, again especially at the relatively low pore pressure gradients that dominate the dataset (there are relatively few leak-off pressures available in overpressured formations). Linear regression for the Australian NW Shelf dataset suggests that Aoi/APp«0.75 (Fig. 4c).
Implications of pore pressure/stress coupling for pore pressure-related rock failure
Fig. 4. Pore pressure gradient versus minimum horizontal stress gradient plots for the (a) Canadian Scotian Shelf; (b) Gannet and Guillemot Fields of the North Sea and; (c) Australian North West Shelf. Although there is significant scatter (discussed in the text), all plots show increasing minimum horizontal stress with increasing pore pressure.
Subsurface sediment mobilization is often controlled by the brittle failure of units 'sealing' overpressured and liquidized sediments and clastic intrusions may be associated with either shear or tensile fractures (Huang 1988; Morley et al 1998; Jolly & Lonergan under review). Similarly, although analysis of the fracture-related failure of hydrocarbon reservoir seals has focused on tensile failure (e.g. Watts 1987; Gaarenstroom et al 1993), accumulations may be breached in association with either tensile or shear failure (Hillis 1998; Finkbeiner etal. 2001). This section of the paper discusses the implications of Pp/crh coupling for brittle failure. Rock failure limits the maximum pore pressure that can develop within either liquidized sediments or hydrocarbon columns. Pore pressures can increase to a point where failure is induced, at which point pressure bleeds off through the resultant fractures. Pore pressure may increase again until the failure limit is again reached. Rock failure thus acts as a valve limiting the maximum pore pressure that can develop (Secor 1965; Sibson 1995). Rock failure induced by increasing pore pressure may either be in tensile, shear or hybrid tensile/shear mode (Sibson 1996). In order for pure tensile fractures to develop, Mohr circle must touch the failure envelope where shear stress is zero and the minimum principal effec-
PORE PRESSURE/STRESS COUPLING
365
tive stress is equal to the tensile strength (T) of the rocks (Fig. 1). Compressive shear failure occurs if Mohr circle touches the failure envelope where effective normal stresses are positive (Fig. 1). It follows from the shape of the composite GriffithCoulomb failure envelope that failure occurs in (Sibson 1996; Fig. 1): •
shear if differential stress is relatively high: (crr
•
tension if differential stress is relatively low:
•
hybrid tension/shear if differential stress is
The conventional view of the effect of pore pressure on rock failure assumes that the total stresses are independent of pore pressure (Fig 1). Figure 5 illustrates the influence of Pp/crh coupling on pore pressure-induced rock failure, assuming a normal fault condition, i.e. vertical stress is the maximum principal stress (i.e., crv = (T1) and minimum horizontal stress is the minimum principal stress (i.e., crh = cr3). Effective vertical stress (
Fig. 5. Influence of P/Oh coupling on overpressurerelated rock failure assuming normal fault regime (crv> crH>crh), with Acrh/APp = 0.7, and vertical stress unaffected by changes in pore pressure (i.e. coupled model).
Assuming that the minimum horizontal stress acting at the top of hydrocarbon columns similarly increases with pore pressure, these results imply that columns significantly greater than those which exert a buoyancy pressure equivalent to the difference between prevailing pore pressure and minimum horizontal stress may be retained. Gaarenstroom et al (1993) defined the retention capacity of traps in the Central North Sea as the difference between prevailing pore pressure and minimum horizontal stress. This definition does not incorporate the influence of Pp/crh coupling and hydrocarbon columns larger than those predicted by the difference between prevailing pore pressure and minimum horizontal stress may be trapped. The maximum shear stress is equal to the radius of Mohr circle or half the differential stress. Pore pressure/stress coupling causes a decrease in shear stress as pore pressure increases in a normal fault regime basin because the effective vertical stress decreases more rapidly than the effective minimum horizontal stress (Fig. 5). This has critical implications both for the limits to pore pressure that can be sustained and for the mode of rock failure. If neither total stress were coupled to pore pressure, or if both were similarly coupled, then shear stress would remain constant with increasing pore pressure, i.e. the radius of Mohr circle would remain constant as effective stresses decrease and it moves towards the failure envelope. Hence, in a case such as Figure 5, compressive shear failure would occur at significantly lower pore pressure than tensile failure and shear failure would provide the upper limit for pore pressure. The decrease in shear stress associated with pore pressure/stress coupling is critical in preventing shear failure at relatively low pore pressure (Fig. 5). This decrease in shear stress and resultant propensity for tensile rather than shear failure may account for the more common observation of clastic dykes and sills than of clastic intrusions along active fault planes. It may also account for the common occurrence of tensile fractures in overpressured rocks
366
R.R. HILLIS
(Tingay et al 2003) and indeed for the occurrence of randomly oriented tensile fractures considered to be indicative of tensile failure in an isotropic stress state, i.e. no differential stress (Cosgrove 1995). The conventional or uncoupled approach implies that the mode of failure (shear, hybrid or tensile) can be predicted from the differential stress in the stable state (Hobbs et al 1976; Cosgrove 2001; Fig. 1). Pore pressure/stress coupling precludes inferring the likely mode of rock failure from differential stress in the stable state without knowledge of A
Discussion The common association of elevated pore pressure and elevated minimum horizontal stress has led to a 'chicken-and-the-egg' debate on whether high horizontal stresses generate overpressure or vice-versa (Bell 1996). Overpressure may be a consequence of high horizontal stresses in tectonically active basins, due to pore fluid disequilibrium with horizontal stress-driven compaction (Goulty 1998), and/or due to undrained shear collapse of the rock matrix (Yassir & Bell 1994). On the other hand, the poroelastic model of the response of rocks to uniaxial strain and indeed the frictional failure model, indicates that elevated minimum horizontal stress may be a consequence of elevated pore pressure. In passively subsiding sedimentary basins, overpressure is likely to be the principal cause of high horizontal stresses. However, in tectonically active basins, horizontal stress is likely to be an important contributor to overpressure generation. It is critical to the arguments presented herein that the total vertical stress, which is given by the weight of the overburden, is unaffected by changes in pore pressure. This assumption is widely made (e.g. Teufel et al 1991; Engelder 1993) and in the context of the poroelastic model, can be justified given that the earth's surface is a free surface where strain is allowed to absorb any increase in pore pressure. In the cases of rock failure in the normal fault regime discussed herein, maximum horizontal stress (a-H) is the intermediate principal stress and hence has a limited role in rock failure. Such is not the
case in the strike-slip and reverse fault regimes. Unfortunately, maximum horizontal stress is difficult to measure and there is little information on how it varies with pore pressure. A full understanding of the influence of pore pressure/stress coupling on rock failure requires better knowledge of maximum horizontal stress variation with pore pressure. Yassir & Rogers (1993) suggest that maximum horizontal stress increases similarly to minimum horizontal stress with increasing pore pressure and that consequently the fault condition in the Jeanne D'Arc Basin offshore eastern Canada changes from transitional normal/strike-slip (crH~crv>o-h) to strike slip (OH > crv > orh) to reverse (crH > ah >
Conclusions (i)
(ii)
Brittle failure limits pore pressure in overpressured basin compartments, in liquidized subsurface sediments and in hydrocarbon columns. Clastic injection into 'sealing' units, or fracturerelated breach of hydrocarbon reservoir seals may be due to tensile, shear or hybrid tensile/shear failure. Changes in minimum horizontal stress are coupled to changes in pore pressure. Minimum horizontal stress increases at 76% of the rate of pore pressure moving from shallow, normally pressured sequences to deeper, overpressured sequence offshore eastern Canada. In the Gannet/Guillemot Fields of the North Sea, the
PORE PRESSURE/STRESS COUPLING ratio is 60% and in the Australian North West Shelf it is 75%. Total vertical stress is given by the weight of the overburden and is largely unaffected by changes in pore pressure. (iii) Effective horizontal stress decreases more slowly than pore pressure increases because total horizontal stress increases with pore pressure. Hence, a greater increase in pore pressure (or hydrocarbon buoyancy pressure), can be sustained prior to brittle failure than would be predicted by conventional, uncoupled failure models. Considering the tensile failure condition, pore pressure can increase by a factor of l/(l-Ao-h/APp) more than would be predicted if minimum horizontal stress is assumed to be unaffected by pore pressure. Total vertical stress provides an absolute limit on pore pressure that may be reached before effective horizontal stresses become negative. (iv) The coupled models of Pp/crh changes proposed herein imply that in a normal fault regime the radius of Mohr circle shrinks as it is displaced, with increasing pore pressure, towards failure. Hence, as pore pressure increases differential stress decreases. The propensity for shear failure thus decreases and the propensity for tensile failure increases, with increasing pore pressure.
367
Society, London, Special Publications, 92,187-196. COSGROVE, J.W. 2001. Hydraulic fracturing during the formation and deformation of a basin: a factor in the dewatering of low-permeability sediments. American Association of Petroleum Geologists Bulletin, 85, 737-748. ENGELDER, T. 1993. Stress Regimes in the Lithosphere. Princeton University Press, Princeton. ENGELDER, T. & FISCHER, M.P. 1994. Influence of poroelastic behaviour on the magnitude of minimum horizontal stress, Sh, in overpressured parts of sedimentary basins. Geology, 22,949-952. FlNKBEINER, T, ZOBACK, M., FLEMINGS, P. & STUMP, B.
2001. Stress, pore pressure, and dynamically constrained hydrocarbon columns in the South Eugene Island 330 field, northern Gulf of Mexico. American Association of Petroleum Geologists Bulletin, 85, 1007-1031. GAARENSTROOM, L., TROMP, R.A.J., DE JONG, M.C. & BRANDENBURG, A.M. 1993. Overpressures in the Central North Sea: implications for trap integrity and drilling safety. In: PARKER, J.R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference, Geological Society, London, 2, 1305-1313. GOULTY, N.R. 1998. Relationships between porosity and effective stress in shales. First Break, 16,413-419. HARMS, J.C. 1965. Sandstone dikes in relation to Laramide faults and stress distribution in the Southern Front Range, Colorado, Geological Society of America Bulletin, 76,981-1002. HILLIS, R.R. 1998. Mechanisms of dynamic seal failure in the Timor Sea and Central North Sea. In: PURCELL, D. Dewhurst (CSIRO) and D. Swarbrick (University of P.G. & PURCELL, R.R. (eds) The Sedimentary Basins Durham) are thanked for discussion on the concepts preof Western Australia 2. Proceedings of Petroleum sented herein and for constructive comments on this manuExploration Society of Australia Symposium, Perth, script. H. van Eekelen is thanked for pressure data from the Western Australia, 313-324. Gannet/Guillemot fields. P. Flemings is thanked for conHILLIS, R.R. 2001. Coupled changes in pore pressure and structive comments in review. stress in oil fields and sedimentary basins. Petroleum Geoscience, 7,419-425. HOBBS, B.E., MEANS, W.D. & WILLIAMS, P.P. 1976. An References Outline of Structural Geology. John Wiley and Sons, Singapore. ADDIS, MA., LAST, N.C. & YASSIR, N.A. 1996. Estimation HOLBROOK, P. 1997. Discussion of a new simple method to of horizontal stresses at depth in faulted regions and estimate fracture pressure gradients. SPE Drilling & their relationship to pore pressure variations. SPE Completion, March, 71-72. Formation Evaluation, March, 11-18. SPE 28140. HUANG, Q. 1988. Geometry and tectonic significance of BELL, J.S. 1990. The stress regime of the Scotian Shelf offAlbian sedimentary dykes in the Sisteron area, SE shore eastern Canada to 6 kilometers depth and impliFrance. Journal of Structural Geology, 10,453-462. cations for rock mechanics and hydrocarbon HUBBERT, M.K. & RUBEY, W.W 1959. Role of fluid presmigration. In: MAURY, V. & FOURMAINTRAUX, D. (eds) sure in mechanics of overthrust faulting, I: Mechanics Rock at Great Depth, 3. Balkema, Rotterdam, of fluid filled porous solids and its application to over1243-1265. thrust faulting. Geological Society of America BELL, J.S. 1996. In situ stresses in sedimentary rocks (part Bulletin, 70,115-166. 2): applications of stress measurements. Geoscience JOLLY, R.J.H. & LONERGAN, L. 2002. Mechanisms and conCanada, 23,135-153. trols on the formation of clastic intrusions. Journal of BRECKELS, LM. & VAN EEKELEN, HA.M. 1982. Relationthe Geological Society, London, 159 (5), 605-617. ship between horizontal stress and depth in sedimenLORENZ, J.C., TEUFEL, L.W. & WARPINSKI, N.R. 1991. tary basins. Journal of Petroleum Technology, 34, Regional fractures I: a mechanism for the formation 2191-2198. of regional fractures at depth in flat-lying reservoirs. COSGROVE, J.W. 1995. The expression of hydraulic fracturAmerican Association of Petroleum Geologists ing in rocks and sediments. In: AMEEN, M.S. (ed.) Bulletin,15,llU-ll31. Fractography: Fracture Topography as a Tool in MALTMAN, A.J. & BOLTON, A. 2003. How sediments Fracture Mechanics and Stress Analysis. Geological become mobilized. In: VAN RENSBERGEN, P., HELLIS,
368
R.R. fflLLIS
R.R., MALTMAN, A.J. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization, Geological Society, London, Special Publications, 216,9-20. MANDL, G. & HARKNESS, R.M. 1987. Hydrocarbon migration by hydraulic fracturing. In: JONES, M.E. & PRESTON, R.M.F. (eds) Deformation of Sediments and Sedimentary Rocks. Geological Society, London, Special Publications, 29, 39-53. MORLEY, C. K., CREVELLO, P. & ZULKIFLI HAn AHMAD, 1998. Shale tectonics and deformation associated with active diapirism: the Jerudong Anticline, Brunei Darussalam: Journal of the Geological Society, London, 155,475-490. MOUCHET, J.P. & MITCHELL, A. 1989. Abnormal pressures while drilling. Boussens, France, Elf Aquitaine. SCHNEIDER, F., BOUTECA, M. & SARDA, J.P. 1999. Hydraulic fracturing at sedimentary basin scale. Oil & Gas Science and Technology - Rev. IFP, 54,797-806. SECOR JR., D.T. 1965. Role of fluid pressure in jointing. American Journal of Science, 263,633-646. SffiSON, R.H. 1995. Selective fault reactivation during basin inversion: potential for fluid redistribution through fault-valve action. In: BUCHANAN, J.G., BUCHANAN, P.G. (eds) Basin Inversion. Geological Society, London, Special Publication, 88,1031-1042. SIBSON, R.H. 1996. Structural permeability of fluid-driven fault-fracture meshes. Journal of Structural Geology, 18,1031-1042. SWARBRICK, R.E. & HILLIS, R.R. 1999. The origin and influence of overpressure with reference to the North West Shelf, Australia. Australian Petroleum Production and Exploration Association Journal, 39, 64-72. SWARBRICK, R.E. & OSBORNE, M.J. 1996. The nature and diversity of pressure transition zones. Petroleum Geoscience, 2,111-116. TEUFEL, L.W., RHETT, D.W. & FARRELL, H.E. 1991. Effect of reservoir depletion and pore pressure drawdown on in situ stress and deformation in the Ekofisk Field, North Sea. In: ROEGIERS, J.C. (ed.), Rock Mechanics
as a Multidisciplinary Science. Balkema, Rotterdam, 63-72. TINGAY, M.R.P., HILLIS, R.R., MORLEY, C.K., SWARBRICK, R.E. & OKPERE, E.G. 2003. Pore pressure/stress coupling in Brunei Darussalau - implications for shale injection. In: VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, A.J. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization, Geological Society, London, Special Publications, 216,369-379. TRAUGOTT, M. 1997. Pore/fracture pressure determinations in deep water. Deepwater Technology, Supplement to August 1997 World Oil and Pipe Line & Gas Industry, 68-70. WATTS, N.L. 1987. Theoretical aspects of cap-rock and fault seals for single and two-phase hydrocarbon columns. Marine and Petroleum Geology, 4, 274-307. WHITE, A.J., TRAUGOTT, M.O. & SWARBRICK, R.E. 2002. The use of leak-off tests as means of predicting minimum in-situ stress. Petroleum Geoscience, 8, 189-193. YASSIR, N.A. & BELL, J.S. 1994. Relationships between pore pressure, stresses, and present-day geodynamics in the Scotian Shelf, offshore eastern Canada. American Association of Petroleum Geologists Bulletin, 78,1863-1880. YASSIR, N.A. & ROGERS, A.L. 1993. Overpressures, fluid flow and stress regimes in the Jeanne d'Arc Basin, Canada. International Journal of Rock Mechanics and Mining Science, 30, 1209-1213. ZOBACK, M.D. & HEALY, J.H. 1984. Friction, faulting and in situ stress. Annales Geophysicae, 2,689-698. ZOBACK, M.D., BARTON, C., BRUDY, M., CHANG, C., Moos, D., PESKA, P. & VERNIK, L. 1995. A review of some new methods for determining the in situ stress state from observations of borehole failure with applications to borehole stability and enhanced production in the North Sea. In: FEJERSKOV, M. & MYRVANG, A.M. (eds) Rock Stresses in the North Sea: Proceeding of the Workshop, 13-14 February, 1995, Trondheim. University of Trondheim, 6-21.
Pore pressure/stress coupling in Brunei Darussalam - implications for shale injection MARK R. P. TINGAY1, RICHARD R. HILLIS1, CHRISTOPHER K. MORLEY2, RICHARD E. SWARBRICK3 & EUGENE C. OKPERE4 ^National Centre for Petroleum Geology and Geophysics, Adelaide University, South Australia, Australia (e-mail: [email protected]) 2 University of Brunei Darussalam, Bandar Seri Begawan, Negara Brunei Darussalam ^Durham University, Durham, UK ^Brunei Shell Petroleum, Seria, Negara Brunei Darussalam Abstract: Shale dykes, diapirs and mud volcanoes are common in the onshore and offshore regions of Brunei Darussalam. Outcrop examples show that shale has intruded along both faults and tensile fractures. Conventional models of overpressure-induced brittle failure assume that pore pressure and total stresses are independent of one another. However, data worldwide and from Brunei show that changes in pore pressure are coupled with changes in total minimum horizontal stress. The pore pressure/stress-coupling ratio (Acrh/APp) describes the rate of change of minimum horizontal stress magnitude with changing pore pressure. Minimum horizontal stress measurements for a major offshore field where undepleted pore pressures range from normal to highly overpressured show a pore pressure/stress-coupling ratio of 0.59. As a consequence of pore pressure/stress coupling, rocks can sustain a greater increase in pore pressure prior to failure than predicted by the prevailing values of pore pressure and stress. Pore pressure/stress-coupling may favour the formation of tensile fractures with increasing pore pressure rather than reactivation of pre-existing faults. Anthropogenically-induced tensile fracturing in offshore Brunei supports this hypothesis.
Overpressured shale has migrated along both tensile and shear fractures in Brunei Darussalam (Morley et al 1998; Morley 2003). Fluids can flow along tensile fractures when open and along active shear fractures/faults (Sibson 1996; Losh et al 1999). Hence, the transmission of overpressured shale through fractures is controlled by the stress field, pore pressure and rock strength. In the petroleum industry, the conditions for rock failure are used to estimate the amount of pore pressure increase that would cause seal breach. It is commonly assumed that the maximum hydrocarbon column height that can be trapped is one that generates a buoyancy pressure equal to the difference between the prevailing pore pressure and minimum horizontal stress (e.g. Caillet 1993; Gaarenstroom et al 1993). More recently it has been recognized that seal breach may be initiated if the buoyancy of the hydrocarbon column causes tensile or shear failure (Finkbeiner et al 2001). However, models of pore pressure related failure need to incorporate the observation that changes in pore pressure are associated with changes in the minimum horizontal stress magnitude (crh). Measurements of virgin pore pressures and minimum horizontal stress magnitudes (i.e. those unaffected by depletion) indicate that minimum horizontal stress increases, from shallow normally pressured sequences into deeper, overpressured sequences, over and above that due to increasing depth (Bell 1990; Gaarenstroom et al 1993; Yassir
& Bell 1994). The changes in minimum horizontal stress that accompany changes in pore pressure are herein termed pore pressure/stress (Pp/o-h) coupling. The nature of the Pp/crh coupling relationship is a factor in determining the amount of Pp increase that can be sustained before failure and the mode (shear versus tensile) of rock failure that develops with increasing overpressure. Hence, the Pp/crh coupling relationship controls the injection mechanism of sediment injection features such as shale dykes (Morley et al 1998). We use mini-fracture tests and repeat formation tests in one field in offshore eastern Brunei to determine the Pp/crh coupling relationship. It is shown that Pp/(jh coupling allows the development of stress states in which both tensile and shear failure occurs. Both modes of failure are observed in outcrop in Brunei (Morley et al 1998; Morley this volume). However, we suggest that in the present-day stress regime it is more likely for mobile shale to be transmitted along tensile fractures than shear fractures. A 1979 internal blowout of a well is described as an example of this failure process and subsequent injection of overpressured fluids. Regional setting The onshore and offshore regions of Brunei Darussalam are composed of several Late Neogene
Fig. 1. Major deltaic fault trends and localities in onshore and offshore Brunei Darussalam. The present day minimum horizontal stress is oriented sub-parallel with fault strike (adapted from Koopman & James 1996&). rapidly-prograding delta systems built outwards from the Crocker-Raj ang accretionary complex and deposited adjacent to the NW Borneo active margin (Koopman & James 1996a; Fig. 1). This tectonic setting has led to fast deposition rates and exhibits complex interaction between gravity-driven deltaic and transpressional basement tectonics (Koopman & James 1996&). Deposition rates within delta systems have reached 3000 m/Ma (Koopman & James 19966). Rapid deposition of the fine-grained prodelta sediments has led to the development of widespread overpressures generated by disequilibrium compaction (Schreurs & Ellenor 1996). Overpressures may also have been generated by increasing horizontal stress magnitudes with depth (De Breeetal. 1993 ;Yassir & Bell 1994). Overpressured fluids have also been transmitted vertically along faults and fractures resulting in inflationary overpressures at shallow depths (Schreurs & Ellenor 1996). The highly overpressured prodelta shales of Brunei are the source of ancient and present-day shale diapirs, dykes and mud volcanoes (Koopman & James 19966). Middle Miocene-Early Pliocene shale injection features are exposed in outcrop within the Jerudong Anticline (Morley et al. 1998;
Morley 2003) and are observed on seismic sections within several offshore and onshore fields throughout the delta (Van Rensbergen et al 1999). Exposed shale intrusions (predominantly dykes) within the Jerudong Anticline exhibit shale injection along both tensile and shear fractures up to 60 cm wide (Fig. 2). Morley et al. (1998) suggest that stress and pore pressure conditions during the Middle Miocene-Early Pliocene were such that mobile shale could migrate along shear fractures and also initiate tensile fractures. There is little published data known to the authors concerning the present day stress field in Brunei Darusslam. Walters et al. (1999) suggest that the au direction in the Seria Field is approximately NW-SE. Breckels & Van Eekelen (1982) show that av > crh in the upper sections of the delta. However, the magnitude of the maximum horizontal stress (crH) is difficult to determine and the full stress tensor for Brunei is poorly constrained. Tertiary deltas generally exhibit a normal fault stress regime (crv >
a-h) due to the convex upward nature of the delta wedge (Yassir & Zerwer 1997). Indeed, De Bree et al. (1993) suggest that the Brunei delta systems exhibit a normal fault stress regime (crv > crH > (Th) in the normally pressured upper deltaic sequences, but trends into a strike-slip regime (orn> crv> ah) with increasing pore pressure and depth. Herein, we have assumed that a normal/strike-slip in situ stress regime ( crh) exists for all pore pressures and depths due to the difficulty in constraining
Pore pressure/stress coupling Pore pressure/stress coupling at the sedimentary basin-scale An increase in minimum horizontal stress, from shallow, normally pressured sequences into deeper, overpressured sequences, over and above that due to increasing depth, is herein termed basin-scale Pp/
PORE PRESSURE/STRESS COUPLING IN BRUNEI
371
Fig. 2. Shale dykes within the Jerudong Anticline within: (a) a tensile fracture; (b) a shear fracture and; (c) small scale shear and tensile fractures.
M.R.P.TINGAYCTAL.
372
Determining the pore pressure/stress coupling relationship A number of different models have been proposed to describe the variation of stress with depth and pore pressure: • • •
poroelasticity (Engelder & Fischer 1994); solidity (Holbrook 1997); frictional limit (Zoback et al 1995).
All models have the form:
and all predict Pp/crh coupling (Hillis 2003). It is not the purpose of this paper to discuss the assumptions made in these models nor their relative merits. Herein we favour the widely used empirical approach to determine the Pp/crh coupling relationship (Breckels & Van Eekelen 1982). Basin-scale estimation of Pp/
pressure versus root time records (more details on the mini-fracture test procedure and interpretation can be found in Enever 1993).
Pore pressure/stress coupling and sediment injection The injection of fluids and fluidized sediments (such as mobile shale) along faults and fractures is controlled by the stress field, pore pressure and rock strength. Tensile failure occurs when pore pressure exceeds or equals the minimum stress and tensile rock strength (e.g. P p >S 3 + T; Sibson 1996). Suitably orientated faults slip (and may transmit fluids) when the shear stress acting on the fault exceeds or equals a function of the effective normal stress on the fault, the frictional limit to sliding and the fault cohesion (T>S0 + jju(a-n-Pp); Jaeger & Cook 1969). Basin-scale Pp/crh coupling has significant implications for the mode of failure that occurs with increasing overpressure and hence, the formation of features such as the shale dykes observed in the Jerudong Anticline (Fig. 2). The conventional model for shear or tensile failure associated with pore pressure increase assumes that the total stresses are independent of pore pressure. Hence, with increasing pore pressure, the effective vertical (crv — Pp) and effective minimum horizontal (<7h-Pp) stresses decrease at the same rate. This causes Mohr circles to slide, with unchanging diameters, towards the failure envelope (Fig. 4a). In the conventional model, tensile failure can only occur if the differential stress (crj - cr3) is low (grey circles) and consequently, the Mohr circles are small (Fig. 4a). Assuming a composite Griffith-Coulomb failure envelope, differential stress must be less than four times the tensile strength for tensile failure to occur (crl — (T3<4T; Sibson 1996). However, the minimum horizontal stress is coupled with variations in pore pressure, and hence changes, whereas the total vertical stress, given by the weight of the overburden, is thought to be unaffected by Pp changes (Teufel et al 1991; Engelder 1993). Hence, the effective vertical stress decreases at the rate that pore pressure increases. Effective minimum horizontal stress decreases more slowly than pore pressure increases (or effective vertical stress decreases) because the total minimum horizontal stress increases with pore pressure (Fig. 4b). Hence, in a sedimentary basin that has a normal fault stress regime (crv< crH< crh), Pp/
PORE PRESSURE/STRESS COUPLING IN BRUNEI
373
Fig. 3. Example of how mini-fracture tests are used to determine crh from the fracture closure pressure. Mini-fracture test was performed at 3250 m depth in Field A.
overpressured stress state, the failure envelope and the A
Pore pressure/stress coupling in Field A Geological setting of Field A Pore pressure/stress coupling has been investigated in a major field in eastern offshore Brunei herein referred to as Field A (Fig. 1). The Field A structure is a large rollover anticline on the hanging wall side of a large down-to-basin growth fault (Koopman et al. 1996). A series of syn-depositional collapse grabens has developed over the crestal area of the rollover anticline. Structures in Field A have been further complicated by later transpressional deformation and uplift (Koopman et al 1996). Faults in Field A strike between 010-040°N and dip approximately 50° to the east and west. Pore pressures in Field A are originally hydrostatic in shallow reservoirs (< 1500 m depth) but can increase to pore pressure gradients of up to 22.0 MPa/km below 2900 m depth (Koopman et al. 1996).
fracture tests were performed between depths of 1350-3250 m with pore pressure gradients between 9.8-22.0 MPa/km. Pore pressures have all been obtained from RFT measurements within at most 50 m vertically from the mini-fracture test point. All pore pressure and mini-fracture measurements have been performed in undepleted units. The plot of Pp/crv versus crh /crv yields a A
Results
Pore pressure/stress coupling and rock failure in Field A
Minimum horizontal stress estimates from eight mini-fracture tests are used in this study. The rnini-
We can use the Acrh/APp coupling ratio determined in this study (Fig. 5) to investigate the likely mode
374
M.R.P.TINGAYCTAZ,
Fig, 4. Evolution of Mohr circles with increasing pore pressure, (a) Conventional model where stresses are independent of pore pressure; (b) model with pore pressure/stress coupling and; (c) pore pressure/stress coupling trend suggested by stress estimates in Field A normalized to 1500 m depth. The existing faults in Field A are not well oriented for failure. Tensile failure is more likely to occur with increasing pore pressure in the present-day stress field.
PORE PRESSURE/STRESS COUPLING IN BRUNEI
375
Fig. 5. Basin-scale pore pressure/stress (Pp/crh) coupling relationship for Field A in eastern offshore Brunei (Pp/crh ratio = 0.59). Pore pressure and minimum horizontal stress have been normalized with respect to the vertical stress magnitude. The vertical stress gradient varies with depth and hence the normalized hydrostatic gradient is a range. Oil field-scale Pp/crh coupling is also observed in Field A from the results of two minifracture tests performed in depleted reservoirs. These two minifracs suggest an oil field-scale Pp/crh coupling ratio that is higher (0.99) than the basin-scale Pp/o-h ratio. However, there is not enough data to provide a reliable estimate of the oil field-scale Pp/crh ratio.
(and orientation) of rock failure (and fluidized shale injection) with increasing pore pressure in Field A. Mohr circles from the stress estimates of the eight mini fracture tests in Field A are plotted in Figure 4c. The minimum horizontal stress is determined from the mini-fracture test data and the vertical stress magnitude has been calculated by integrating density logs to determine the weight of the overburden. The vertical and minimum horizontal stresses and the pore pressures have been normalized to 1500 m depth for comparison. Unfortunately there are no data known to the authors on the nature of changes in maximum horizontal stress magnitude (OH) with pore pressure. However, the assumption that crH~(Tv reduces the problem to a two dimensional one, allowing crH to be ignored. Figure 4c suggests that the Mohr circles reduce in diameter with
increasing pore pressure until the pore pressure,
376
M.R.P.TINGAYCTAL.
day crh and are not well orientated for reactivation in a normal/strike-slip faulting stress regime (Fig. 1). However, it is possible for non-optimally orientated faults to be reactivated rather than new fractures being created (Streit & Cox 2001). The likely mode of failure (and hence, transmission of mobile shale) that will occur with increasing pore pressure can be investigated by combining information on the existing fault plane orientations and the stress tensor (Mohr circle). Any plane can be represented on a Mohr circle in terms of the shear and normal stress that act upon it under a given stress field. Figure 4c shows that existing faults in Field A have the lowest risk of failure either in shear or in tension in the present day stress regime. The low differential stress at high pore pressures (such as Mohr circle A in Figure 4c) suggests that newly formed tensile fractures are the likely overpressure-induced failure mechanism in Field A. Hence, it appears more likely that in the present day stress field, increasing pore pressure will create tensile fractures approximately oriented in the direction of the maximum horizontal stress (130°N) rather than reactivating pre-existing faults.
The Field A 1979 internal blowout
Fig. 6. Examples of borehole breakouts on resistivity image logs (dark patches separated by approximately 180°) in Brunei. Breakouts are oriented in the minimum horizontal stress (
040°N (Walters et al 1999; Whiteley et al 1991). The authors' own interpretation of breakouts and drilling-induced tensile fractures (from 4-arm caliper and image logs in 20 wells in Brunei) also indicate a 040°N
The predicted propensity for tensile failure is supported by observations from the internal blowout that occurred within Field A in 1979. An internal blowout involves overpressured fluids being transmitted along the wellbore to shallower reservoir units rather than to the surface. The 1979 Field A internal blowout was associated with a seabed blowout that expelled large volumes of overpressured fluids for ten days (Koopman etal 1996). The internal blowout has involved deep overpressured fluids and sediments being transmitted to shallow reservoirs along the uncased section of borehole (Fig. 7). Pore pressures within the shallow reservoirs rapidly increased until the rock above fractured, causing the seabed blowout. High resolution seismic over the location of the surface blowout indicates that the overpressured fluids and sediments reached the surface along a vertical NW-SE striking tensile fracture, approximately perpendicular to ah (Whiteley et al 1991). The creation of a new tensile fracture oriented NW-SE, in response to increasing pore pressure, rather than the reactivation of existing fractures is entirely as predicted by the above coupled Pp/o-h model (Fig. 4c).
Implications for outcrop observations in the Jerudong Anticline The hydraulic fracturing and subsequent fluid expulsion during the 1979 internal blowout may be analo-
PORE PRESSURE/STRESS COUPLING IN BRUNEI
377
Fig. 7. Schematic diagram of the 1979 Field A internal blowout (adapted from Whiteley et al 1991).
gous to the Miocene-Pliocene shale dykes observed in the Jerudong Anticline (Morley et al. 1998; Morley 2003) and the ancient and modern mud volcanoes in Brunei (Van Rensbergen et al. 1999; Koopman et al. 1996). Middle Miocene shale dykes in the Jerudong Anticline were primarily intruded into shear fractures and to a lesser extent along tensile fractures. Morley et al. (1998) documents two phases of shale intrusion: (i)
Middle Miocene intrusions prior to folding that are closely related to fault orientation. (ii) Late Miocene-Pliocene intrusions post folding (faults rotated 60-90°) that have injected into new fractures independent of the older faults.
Middle Miocene tensile fractures in the Jerudong anticline were initiated both sub-vertically and sub-horizontally (Morley et al 1998; Morley 2003).
The presence of both horizontal shale sills/laccoliths and vertical shale dykes suggests that the minimum horizontal stress was approximately equal to the vertical stress magnitude (cr^cr^ and that pore pressure approached lithostatic during Miocene shale injection. Morley et al. (1998) suggest that during the Middle Miocene the minimum horizontal stress (
378
M.R.P.TINGAY£rAL.
Pore pressure/stress coupling at the oilfieldscale Pp/crh coupling is also observed with reservoir depletion in Field A. This is termed oil field-scale Pp/crh coupling (ffillis 2001; Fig. 5). At the oil field-scale, the Pp/
Conclusions Changes in pore pressure and minimum horizontal stress are coupled at the geological timescale at which overpressure develops in offshore Brunei Darussalam. Within Field A in the eastern offshore part of the basin, the A
during the Late Miocene-Pliocene where pre-existing faults were not favourably oriented for reactivation, shale dykes were intruded into both vertical and horizontal tensile fractures. We would like to thank G. Couples, D. Grauls and P. Van Rensbergen for their excellent and helpful reviews of this paper. The authors gratefully acknowledge the assistance of Brunei Shell Petroleum for supplying data used in this study and the Australian Research Council for providing funding. The authors also wish to thank Brunei Shell Petroleum and the Petroleum Unit for permission to publish this paper.
References ADDIS, M.A. 1997. Reservoir depletion and its effect on wellbore stability evaluation. International Journal of Rock Mechanics and Mining Sciences, 34, 3-4, paper 004. BELL, J.S. 1990. The stress regime of the Scotian Shelf offshore eastern Canada to 6 kilometres depth and implications for rock mechanics and hydrocarbon migration. In: MAURY, V. & FOURMAINTRAUX, D. (eds) Rock at Great Depth. Rotterdam, Balkema, 1243-1265. BELL, J.S. 1996. In situ stresses in sedimentary rocks (part 1): measurement techniques. Geoscience Canada, 23, 85-100. BRECKELS, I.M. & VAN EEKELEN, H.A.M. 1982. Relationship between horizontal stress and depth in sedimentary basins. Journal of Petroleum Technology, 34,2191-2198. CAILLET, G. 1993. The caprock of the Snorre Field, Norway: a possible leakage by hydraulic fracturing. Marine and Petroleum Geology, 10,42-50. DE BREE, P., HYDENDAAL, H.G. & VAN DE HOEK, P.J. 1993. Rock mechanical investigations for improved fracture closure gradient prediction, Champion Field, Brunei. Shell Internationale Research Report RKGR.93.030. Rijswijk, The Netherlands. ENEVER, J.R. 1993. Case studies of hydraulic fracture stress measurement in Australia. In: HUDSON, J.A. (ed.) Comprehensive Rock Engineering, 497-532. ENGELDER, T. 1993. Stress Regimes in the Lithosphere. Princeton University Press, Princeton. ENGELDER, T. & FISCHER, M.P. 1994. Influence of poroelastic behaviour on the magnitude of minimum horizontal stress, Sh, in overpressured parts of sedimentary basins. Geology, 22, 949-952. ERVINE, W.B. & BELL, J.S. 1987. Subsurface in situ stress magnitudes from oil-well drilling records: an example from the Venture area, offshore eastern Canada. Canadian Journal of Earth Sciences, 24, 1748-1759. FlNKBEINER, T, ZOBACK, M.D., FLEMINGS, P. & STUMP, B.
2001. Stress, pore pressure, and dynamically constrained hydrocarbon columns in the South Eugene Island 330 field, northern Gulf of Mexico. American Association of Petroleum Geologists Bulletin, 85, 1007-1031. GAARENSTROOM, L., TROMP, R.A.J., DE JONG, M.C. & BRANDENBURG, A.M. 1993. Overpressures in the
PORE PRESSURE/STRESS COUPLING IN BRUNEI Central North Sea: implications for trap integrity and drilling safety. In: PARKER, J.R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference. Geological Society, London, 1305-1313. HILLIS, R.R. 2001. Coupled changes in pore pressure and stress in oil fields and sedimentary basins. Petroleum Geoscience, 7,419-425. HILLIS, R.R. 2003. Pore pressure/stress coupling and its implications for rock failure. In: VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, A.J. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216,359-368. HOLBROOK, P. 1997. Discussion of a new and simple method to estimate fracture pressure gradients. SPE Drilling & Completion, 71-72. JAEGER, J.C. & COOK, N.G. W. 1969. Fundamentals of Rock Mechanics. Methuen and Co. Ltd, London. KOOPMAN, A. & JAMES, D.M.D. 1996a. Chapter 2: Regional Geological Setting. In: SANDAL, S.T. (ed.) The Geology and Hydrocarbon Resources ofNegara Brunei Darussalam. Syabas Publishers, Brunei Shell Petroleum and the Brunei Museum, Bandar Seri Begawan, 49-63. KOOPMAN, A. & JAMES, D.M.D. I996b. Chapter 3: Structure. In: SANDAL, S.T. (ed.) The Geology and Hydrocarbon Resources of Negara Brunei Darussalam. Syabas Publishers, Brunei Shell Petroleum and the Brunei Museum, Bandar Seri Begawan, 64-80. KOOPMAN, A., SCHREURS, J. &ELLENOR, D.W. 1996. Chapter 5: The Oil and Gas Resources of Brunei Darussalam The Coastal and Offshore Oil and Gas Fields. In: SANDAL, S.T. (ed.) The Geology and Hydrocarbon Resources of Negara Brunei Darussalam. Syabas Publishers, Brunei Shell Petroleum and the Brunei Museum, Bandar Seri Begawan, 155-192. LOSH, S., EGLINTON, L., SCHOELL, M. & WOOD, J. 1999. Vertical and lateral fluid flow related to a large growth fault, South Eugene Island Block 330 Field, offshore Louisiana. American Association of Petroleum Geologists Bulletin, 83,244-276. MATTHEWS, W.R. & KELLY, J. 1967. How to predict formation pressure and fracture gradient. Oil and Gas Journal, 65,92-106. MILDREN, S.D., HILLIS, R.R. & KALDI, J.G. 2002. Comparison of geomechanical predictions of fault trap seal breach and trap integrity in the Timor Sea. Australian Petroleum Production and Exploration Association Journal, 42,187-202. MORLEY, C.K. 2003. Outcrop examples of mudstone intrusions from the Jerudong anticline, Brunei Darussalam, and inferences for hydrocarbon reservoirs. In: VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, A.J. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216,335-357. MORLEY, C.K., CREVELLO, P. & AHMAD, Z.H. 1998. Shale tectonics and deformation associated with active diapirism: the Jerudong Anticline, Brunei Darussalam. Journal of the Geological Society, London,155,475-490. MOUCHET, J.P. & MITCHELL, A. 1989. Abnormal Pressures While Drilling. Boussens, France, Elf Aquitaine.
379
PLUMB, R.A. & HICKMAN, S.H. 1985. Stress-induced borehole elongation: A comparison between the four-arm dipmeter and the borehole televiewer in the Auburn Geothermal well. Journal of Geophysical Research, 90,5513-5521. SALZ, L.B. 1977. Relationship between fracture propagation pressure and pore pressure. SPE 52nd Annual Conference, Denver, Colorado, 9-12 October 1977. SPE Paper 6870. SCHREURS, J. & ELLENOR, D.W. 1996. Chapter 5: The Oil and Gas Resources of Brunei Darussalam Hydrocarbon Habitat. In: SANDAL, S.T. (ed.) The Geology and Hydrocarbon Resources of Negara Brunei Darussalam. Syabas Publishers, Brunei Shell Petroleum and the Brunei Museum, Bandar Seri Begawan, 147-154. SffiSON, R.H. 1996. Structural permeability of fluid-driven fault-fracture meshes. Journal of Structural Geology, 18,1031-1042. STREIT, J.E. & Cox, S.F. 2001. Fluid pressures at hypocentres of moderate to large earthquakes. Journal of Geophysical Research, 106,2235-2243. TEUFEL, L.W., RHETT, D.W. & FARRELL, H.E. 1991. Effect of reservoir depletion and pore pressure drawdown on in situ stress and deformation in the Ekofisk Field, North Sea. In: ROEGIERS, J.C. (ed.) Rock Mechanics as a Multidisciplinary Science. Balkema, Rotterdam, 63-72. VAN RENSBERGEN, P., MORLEY, C.K., ANG, D.W, HOAN, T.Q. & LAM, N.T. 1999. Structural evolution of shale diapirs from reactive rise to mud volcanism: 3D seismic data from the Baram delta, offshore Brunei Darussalam. Journal of the Geological Society, London.156,633-650. WAITERS, D.G., MASKALL, R.C., WARRILOW, I.M. & Lmw, V. 1999. A sleeping giant awakened: further development of the Seria Field, Brunei Darussalam, after almost 70 years of production. Petroleum Geoscience, 5,147-159. WHITELEY, S.L., YUK, B.P. & SCHAAFSMA, C.E. 1991. Blowout cratering and principal stress orientations in the Champion Field. Brunei Shell Petroleum Report PE91/3. YASSIR, N.A. & BELL, J.S. 1994. Relationships between pore pressure, stresses, and present-day geodynamics in the Scotian Shelf, offshore eastern Canada. American Association of Petroleum Geologists Bulletin, 78,1863-1880. YASSIR, N.A. & ZERWER, A. 1997. Stress regimes in the Gulf Coast, offshore Louisiana: data from well-bore breakout analysis. American Association of Petroleum Geologists Bulletin, 81,293-307. ZOBACK, M.D., BARTON, C., BRUDY, M., CHANG, C., Moos, D., PESKA, P. & VERNIK, L. 1995. A review of some new methods for determining the in situ stress state from observations of borehole failure with applications to borehole stability and enhanced production in the North Sea. In: FEJERSKOV, M. & MYRVANG, A.M. (eds) Rock Stresses in the North Sea: Proceedings of the Workshop. University of Trondheim, Trondheim, 6-21.
This page intentionally left blank
Outcrop examples of mudstone intrusions from the Jerudong anticline, Brunei Darussalam and inferences for hydrocarbon reservoirs C.K. MORLEY Department of Petroleum Geoscience, University of Brunei Darussalam, Bandar Seri Begawan, 2028, Brunei Darussalam Abstract: Mudstone intrusions in the Jerudong area represent natural hydraulic fractures developed above an inferred mobile mudstone diapir sourced from the Middle Miocene Setap Formation. Intrusions are manifest as dykes, sills and laccoliths. Intrusion geometries are strongly influenced by pre-existing weaknesses, in particular normal faults. Local inflation of country rock shale units also occurs. Dyke terminations include curved traces, splays and en echelon faults. The wall rocks tend to be smooth, or have numerous small, sub-horizontal ledges (mini jogs). Faults and bedding surfaces cause larger jogs in the dyke trend. Mudstone sills commonly change thickness at normal faults, and inflate the hanging wall so that the normal fault sense of motion is inverted, thrusting and fault bend folding are the result. An exposed mudstone laccolith passes laterally via v-shaped bedparallel intrusions into sandstones. The roof of the laccolith is arched upwards and near the intrusion, the roof is broken up by small and large intrusions. Existing models for diapir rise require the brittle shear strength of the roof sequence to be overcome. The Jerudong outcrops demonstrate diapir rise by hydraulic fracturing and sloping which limits the roof sequence strength to the minimum horizontal stress plus the tensile strength of the country rock or normal fault zones in the roof sequence. The crosscutting mudstone dykes, sills, mudstone intruded fault zones and associated cataclastic deformation form planar permeability barriers that can highly compartmentalize the reservoir rock and significantly reduce the area of reservoir rock that can be effectively produced of hydrocarbons. Intruded mudstones could easily be mistaken for stratified shales when interpreting and correlating well logs.
Mudstone diapirs, pipes and other related intrusions have been widely imaged and reported from seismic reflection data (e.g. Dailly 1976; Sandal 1996; Van Rensbergen et al 1999 Stewart 1999; Graue 2000). However, the effects of overpressured fluids, gas clouds, and seismic resolution tend to mask the details of mudstone intrusive features (Van Rensbergen et al 1999; Van Rensbergen & Morley 2000). Hence, outcrop examples of mudstone intrusions have the potential to provide important details about the geometries and processes involved in mud or shale diapirism. Mud volcanoes and breccia-filled feeder pipes have been reported from outcrop (e.g. Barber et al. 1986), yet there are no readily accessible published outcrop descriptions of other kinds of mudstone intrusions. Consequently, this paper aims to give a description of the geometries of the mudstone dykes, sills and laccoliths found in the Jerudong area of Brunei Darussalam, NW Borneo. The Jerudong anticline is a north to south trending major fold in Brunei Darussalam (Figs 1 and 2) that developed during the Middle-Late Miocene. The fold is developed in a shallow marine sequence of alternating sandstones and shales called the Belait Formation (Middle Miocene-Late Miocene). Underlying the Belait Formation is a shale rich sequence called the Setap Formation (Middle Miocene). In the Jerudong area the upper part of the Setap Formation is stratified
and folded along with the Belait Formation, the deeper part of the Setap Formation is thought to be composed of overpressured, mobile mudstones (Sandal 1996; Morley et al 1998; Fig. 2). Direct evidence for a mudstone diapir is lacking, because onshore seismic data reflection quality is poor, particularly around the core of the diapir. However, the presence of a modern mud volcano and older mudstone intrusions along the Jerudong anticline indicate the presence of overpressured, mobile mudstones at depth. Offshore seismic reflection data indicates that reactive mudstone diapirs and mudstone pipes exist regionally in the Bar am Delta province (Sandal, 1996; Van Rensbergen et al 1999; Van Rensbergen & Morley 2000). Rotation of the Belait Formation by folding caused steep, in places vertical, bedding inclinations. The resulting map pattern is effectively a section through a major regional growth fault and associated antithetic and synthetic faults (Fig. 3). In a small area of about 1 km diameter, in the region of growth faults, is a region of natural hydraulic fractures related to overpressured shales, manifest as mudstone dykes, mud volcanoes, sills and laccoliths (James 1984; Sandal 1996; Morley etal 1998). Morley et al (1998) described the geological setting of the Jerudong mudstone intrusions and identified two main periods of mudstone intrusion
Fig. 1. Geological map of Brunei and adjacent areas of Sarawak (modified and updated from Wilford 1960; James 1984).
OUTCROP EXAMPLES OF MUDSTONE INTRUSIONS FROM THE JERUDONG ANTICLINE
383
Fig. 2. Section across the northern onshore area of Brunei, based on the geological map of Wilford (1960), and modifications to the map by the author and Sandal (1996), see Fig. 1 for locations. The ramp-flat geometry within the Palaeogene-Early Miocene 'basement' is highly speculative, but consistently fits the data. Near the coast, the Setap Shale is thick enough under the Jerudong anticline-Belait syncline to cause the lower part of the Setap Shale Formation to be mobile and permit some detachment of underlying and overlying units. Rotated early normal faults and mudstone intrusions are present in the cores of the Jerudong anticline.
emplacement. An early stage of intrusion occurred before folding, concomitant with or post-dating east to west to NE-SW striking normal faults which developed in the Middle Miocene. At the end of an extended phase of folding (Middle-Late Miocene) another set of mudstone dykes was intruded, and more recently, some mud volcano activity has occurred up to the present day. This paper complements previous work by focusing on the geometries of the mudstone intrusions and the implications of these geometries for hydrocarbon exploration and production.
Mudstone dykes In excess of 50 mudstone dykes are present in an area of approximately 1 km2 exposed in road cuts. The dykes intrude tidal to shoreface sandstones and alternating shales. The dykes are up to 60 cm wide, but it is not possible to determine the other dimensions of the larger dykes, since they extend beyond the outcrops in which they are found. These dimensions extend at least 10 m vertically and 45 m in a strike direction (Fig. 4). The sandstone sidewalls to the dykes tend to be very smooth, sharp and lacking classic joint propagation features such as plumose and conchoidal structures. However, arrays of small steps (mini jogs with sub-horizontal steps a few millimetres wide) are present on some dyke walls (Fig. 5). In places closed sub-vertical fractures, with no apparent offset, up to 40-50 cm long lie at a highangle (60-70°) to the dyke wall (Fig. 6). The fracture
intersection with the dyke wall shows the fractures have a planar to slightly wavy cross-sectional profile. These fractures are not intruded by mudstone, but their proximity to the dyke suggests a hydraulic-fracture related origin. Thin sections of the wall rocks show that no modification of grain size or porosity occurred in the country rocks adjacent to the dyke. There was little invasion of mud into the formation. The intruded mudstone is homogeneous, plastic and soft, and clearly different in texture from the slightly more indurated, laminated bedded shales of the adjacent Belait Formation. Small-scale sedimentary features such as low-angle cross-beds, bedding surfaces and coaly units can be matched across the dykes. In many cases there is no vertical or lateral offset of the markers, indicating that the dykes opened under pure tension. In other examples there is lateral offset of markers and the dyke walls display slickensides and cataclastic seams with extensional offsets (Fig. 7). When the bedding has low dips the slickensides show dip-slip displacements, in areas of steep to vertical bedding the slickensides are sub-horizontal. Morley et al (1998) concluded that these dykes were intruded along pre-existing normal faults that were subsequently rotated within the western limb of the Jerudong anticline. Some jogs in dykes occur where a dyke following one fault trend intersects a conjugate fault, and follows its path for a short distance. The mudstone dykes are affected by lithological variations. For example, within the sandstones slight jogs in dykes occur at some bedding surfaces and sometimes the tips of dykes curve to lie parallel to
384
C.K.MORLEY up by small veins of mudstone. A further indication that bioturbation is not the origin is the presence of mudstone dykelets which extend downwards and upwards from the 'broken' beds.
Mudstone laccolith
Fig. 3. Geological map of the western flank of the Jerudong anticline (see Fig. 1 for location). The map is effectively a cross-section through a growth faulted sequence. Map based partially on Morley et al (1998).
The largest intrusion seen in outcrop is a mudstone body 10 m thick and 25 m wide (Fig. 12). The mudstone intrusion has a flat base and convex top, giving it a laccolith geometry. Inside the mudstone body are stoped blocks of limified, fissile shales with coal horizons. Near the lower and lateral margin of the laccolith, stoped sandstone blocks are also present. The cut of the slope and steep dip of the outcrop mean that the intrusion is apparently inverted (Fig. 12). A normal fault provides one lateral boundary to the mudstone intrusion, passing away from the normal fault the mudstone thins and passes abruptly laterally into a sandstone unit. The lateral passage is a series of narrow v-shaped intrusions that wedged open the sandstone parallel to bedding (Fig. 12c). The roof of the intrusion is arched upwards. In a narrow zone 1-2 m above the intrusion the sandstone overburden is highly deformed into irregular, rubbly blocks typically in the order of centimetres to tens of centimetres wide (Fig. 12d). These blocks are affected by intense, closely spaced, conjugate deformation bands and minor faults (millimetres to centimetres apart). Numerous mudstone dykelets intrude the roof. Passing further up section the deformation bands and dykelets disappear, but larger, more widely spaced dykes and sills are present. The structure of the laccolith is schematically illustrated in Figure 13.
Mudstone sills sub-parallel to bedding surfaces (Fig. 8). Some dykes terminate in splays and en echelon fractures (Figs. 4, 9 and 10). A summary of the main features of the dykes discussed above is illustrated in Figure 10. Dykes commonly crosscut small shale units up to 1-2 m thick but tend to terminate in thicker shale sequences. However, small dykelets up to 1-2 cm wide with decimetre-scale extent perpendicular to bedding may be sourced from the bedded shales. These geometries suggest an increase in overpressure within the bedded shale, caused by the dyke emplacement, led to local mobilization of the bedded shales and pressure release via minor hydraulic fracturing. Mobilized bedded shales, which contain sandstone layers, can develop a chaotic geometry where sandstones are broken into large and small blocks (Fig. 11). The resulting appearance is similar to an intensely bioturbated sand-shale unit, except some of the sandstone blocks are angular and broken
In certain localities (in particular in the roof sequence of the laccolith described above), tabular intrusions of mudstone parallel to bedding have produced geometries similar to igneous sills. Mudstone sills, like their igneous counterparts, are in places transgressive from one bedding surface to another. The location of transgressive segments is strongly influenced by discontinuities in the rock (Fig. 14), in particular where the sill meets normal faults. Sills also commonly change thickness where a normal fault intersects the sill. The sill is thicker on the hanging wall side of the fault, as a result of the intrusion raising the hanging wall upwards with respect to the footwall and reversing the original sense of normal motion (Figs 14 and 15). Ramp-flat dykefault geometries and fault bend folds can develop as a result (Fig. 14). The observed faults display only 10's of centimetres of displacement, however is seems quite possible similar but larger, features
OUTCROP EXAMPLES OF MUDSTONE INTRUSIONS FROM THE JERUDONG ANTICLINE
385
Fig. 4. Sketch of an outcrop in Jerudong cut by numerous dykes. Some of the dykes follow older normal faults, but most dykes are independent of the normal faults. Today the normal fault strikes are similar to the NNW-SSE strike of the dykes, however when the steep dip of bedding (ranging between 60° and 90°) is removed the original strike of the normal faults is approximately east to west. The dykes have steep dips in their present orientation. Upon rotation of bedding to horizontal, dyke dips become much lower. Hence assuming dykes were intruded subvertically, dyke emplacement is inferred to be post folding (Morley et al. 1998). Photographs show jogs and splays in mudstone dykes in sandstones, these dykes are natural hydraulic fractures and do not follow pre-existing faults.
could also exist. Indeed there is a striking similarity quent intervals (Fig. 14). Most of the dykes are between the geometry of the mudstone sill in Figure inclined at about 60-70° and follow small normal 14 with its transgressive 'wings' formed at normal faults. Even dykes which do not follow cataclastic faults and much larger-scale sand injection features seams are inclined (rarely vertical) and one dyke in described from the subsurface in the North Sea (e.g. Figure 14 has a strongly listric shape. These pseudofault like geometries give rise to the suspicion that Lonergan & Cartwright 1999). Small dykes extend off the roof of the sill at fre- the dykes follow faults out of the plane of section
386
C.K. MORLEY
Fig. 5. Photograph of the wall of a mudstone dyke, sub-horizontal ledges a few millimetres wide in the wall rock show up as linear, lighter areas, a coin rests on one such ledge, other ledges are highlighted by arrows.
and are seen where they have propagated beyond the extent of the fault plane.
Source for the mobile mudstones
Fig. 6. Closed joints lying at a 70° angle to a mudstone dyke.
The Belait Formation in the outcrops affected by mudstone intrusions is composed predominantly of tidal and shoreface deposits (Morley et al 1998). The shallow marine shales contain a distinctive microfaunal assemblage, including some brackish water forms compared with the deeper marine Setap Formation (Simmons et al 1999). The shales commonly contain coal layers, and have a relatively low gamma ray response compared with deeper water shales found in the Setap Formation. However, geochemically mudstones from the dykes and country rocks are similar, being composed predominantly of kaolin (36-54%) and illite (36-54%). Based on microfauna and gamma ray readings, the mudstone dykes appear to have originated from the Setap Formation. However, there are field examples of shallow marine shales having been inflated and mobilized by dykes and consequently there must be a contribution of shallow marine shales to some of the intrusive mudstones.
OUTCROP EXAMPLES OF MUDSTONE INTRUSIONS FROM THE JERUDONG ANTICLINE
387
Fig. 7. Dyke following pre-existing fault plane as indicated by the presence of sheared, cataclastic zones, and small normal fault extensional duplexes parallel to the walls of the dyke.
Fig 8 Example of the tip of a dyke being influenced by stratigraphy. The strands of the dyke curved to lie parallel to bedding Note the short bed-parallel section of the dyke is deflected by differential compaction around a coal clast. This relationship suggests the dyke was intruded early, shallow in the section and was deformed as the section became buried.
388
C.K.MORLEY
Fig. 9. En echelon mudstone-filled fractures.
The deep Setap Formation source for the mobile mudstones is supported by regional structural evidence. The southern Jerudong anticline is a broad anticline and the sediments on the flanks of the anticline show no evidence for growth faulting or mudstone diapirism and it is inferred that the Setap Formation is not mobile (Back et al 2001; Fig. 1). Passing further north the Jerudong anticline becomes a narrower, tighter fold, with higher dips in the core (Fig. 1). This change in geometry occurs where growth faults and mudstone intrusions affect the fold core. The tightening of the fold can be explained as a lift-off structure developed due to the presence of mobile mudstones within the Setap Formation (Fig. 2). Lift-offs are tight folds, developed off more open folds due to the presence of a weak horizon, which enables deformation in the overlying units to be detached from the underlying ones.
sions lie within a highly porous and permeable sandstone sequence, consequently, fluid loss into the formation is an important factor limiting the size of the intrusions. The fluid content of the mudstones at the time of intrusion was probably around 70%, subsequently, much of the fluid has been lost, leaving about 20-30% water in the mudstone dykes. Intrusion size will be limited by the volume of overpressured mud available to be pumped into the country rock, the permeability and connectivity of the country rock sandstones and the pore fluid pressure within the country rocks. However, processes within the dyke might limit the amount of fluid lost into the formation, in particular the build up of an impermeable clay cake on the walls of the intrusions following the initial loss of fluids. If the intrusions were inflated wider than they are today, then collapsed back as fluids were lost, considerable fracturing and micro faulting of the country rock might be expected to accommodate the collapse, this is generally not seen. Consequently, it is inferred that the maximum width of the intrusions is similar to that seen today. This in turn implies either a single episode of dyke emplacement where water in mudrich fluids pumped along the fracture and fluids was given enough time to be expelled without overinflating the intrusion, or the intrusions grew by multiple events. It cannot be discerned from the featureless plastic muds filling the dykes whether they filled the dyke in one or multiple episodes. The resistance of the overburden to diapir rise is thought to be controlled by the brittle shear strength of the roof rocks and subsequent slip is controlled by the frictional behaviour of the block bounding faults (Vendeville & Jackson 1992). However the Jerudong outcrops demonstrate that the overburden is broken up by intrusive complexes prior to rise of the main mudstone diapiric mass. Consequently, if mudstone diapir rise occurs by hydraulic fracturing and sloping via a network of sills, dykes laccoliths and other intrusions then the shear strength and frictional behaviour of the overburden are not the limiting factors. Instead the strength of the roof rocks is limited to their resistance to hydraulic fracturing, which occurs when the pore fluid pressure exceeds the minimum horizontal stress plus the tensile rock strength. Commonly, in the presence of pre-existing fault zones, it is not even the tensile strength of the rock that has to be overcome, but the (lower) tensile strength of normal fault zones.
Discussion Mudstone emplacement A high overpressure must be maintained for mudstone intrusions to propagate and develop. The intru-
Implications for hydrocarbon exploration and production The Jerudong outcrops indicate that a considerable distance (at least hundreds of metres) above the main mudstone diapir the effects of hydraulic fracturing
OUTCROP EXAMPLES OF MUDSTONE INTRUSIONS FROM THE JERUDONG ANTICLINE
389
Fig. 10. Summary diagram illustrating the main structural features of the mudstone dykes found in the Jerudong area.
Fig. 11. Photograph of a mudstone unit containing 'broken' sandstones, mudstone dykes pass from the shales into adjacent overlying and underlying sandstones. This unit is interpreted to be a stratified shale unit with thin sandstones that became overpressured and slightly mobile due to pressure inflation by the arrival of mudstone intrusions.
390
C.K. MORLEY
Fig. 12. Mudstone laccolith, (a) View of the outcrop, the dark mudstone mass in the centre has intruded the sandstonerich section (see the sketch B). The section youngs to the southern side of the photo, so effectively the view is upside down, (c) Detail of the lateral termination of the mudstone mass, showing v-shaped injection of the dark mudstones into light-coloured sandstones along bedding surfaces, (d) View onto the bedding surface that forms the roof of the laccolith. The roof rocks are broken up by short, curvi-linear dykes and dykelets, demonstrating intense hydraulic fracturing of the roof and the start of upwards stoping into the roof by the mobile mudstones. (e) Orientation data for some of the secondary structures in the roof of the dyke.
OUTCROP EXAMPLES OF MUDSTONE INTRUSIONS FROM THE JERUDONG ANTICLINE
(2)
(3)
Fig. 13. Schematic illustration of the main characteristics of a mudstone laccolith intrusion based on the intrusion in Fig. 12.
(4)
by overpressured, mud-rich fluids can occur. These intrusions are likely to impact hydrocarbon exploration in five ways:
(5)
(1)
Intrusions will provide discontinuous, high and low-angled impedance contrasts that are likely to cause diffractions, out-of-the-plane reflections and interfere with reflections from the stratified sequences. Hence, stratified sequences affected by intrusions will either be areas of weaker, more discontinuous reflectivity, or actually appear to be chaotic reflectivity (and hence, interpreted as regions composed entirely of mobile mudstone). An example of
391
the problems associated with interpreting seismic reflection data associated with mobile, overpressured mudstones in provided in Figure 16 (also see Van Rensbergen et al 1999 and Van Rensbergen & Morley 2000 for discussions of seismic interpretation). Reservoir units will be crosscut by mudstone dykes, mudstone intruded fault zones and associated cataclastic deformation. The intersection of dykes, faults, and stratified shales will result in a network of permeability barriers that highly compartmentalize the reservoir rock and significantly reduce the producible volume of reservoir. Networks of mudstone dykes and sills could render detailed correlation of sand and shale sequences from core and well logs virtually impossible since discrimination of stratified shales from injected, featureless mudstones is extremely difficult. The intrusion of mudstones up fault planes may, in places, provide a significant mechanism for enhancing the sealing potential of fault planes in addition to shale smear, cataclasis and diagenetic effects. Episodes of hydraulic fracturing pump fluids from fractures into the formation, if these fractures tap units with hydrocarbon source potential, or propagating through existing hydrocarbon traps, then enhanced episodes of hydrocarbon migration may be associated with mudstone intrusions. The fluid pumping potential of intrusive complexes is illustrated by the volumes of fluid moved up feeder pipes to mud volcanoes, which are in the order of 1-11 km3 over time periods c. 105 years (e.g. Guliev 1992; Graue 2000).
Fig. 14. Sketch of outcrop example of mudstone sill, Jerudong. The sill geometry is strongly influenced by the location of normal faults. Numerous small dykes sourced from the sill cut the roof of the dyke, most follow pre-existing normal faults.
392
C.K. MORLEY
Fig. 15. Detail of an inverted normal fault caused by intrusion of the sill shown in Fig. 14 (see Fig. 14 for location).
Fig. 16. Example of two counter-regional fault depocentres, which overlie a chaotic mobile mudstone core, Niger Delta. Some of the key features associated with interpreting the seismic data are: (1) an anomalously bright, high amplitude, steeply dipping reflection, possibly a diffraction, but also coincidental with a regional fault. The fault zone might be strongly reflective because it is filled by a hydrocarbon charged mudstone intrusion. (2) A ragged concave-up boundary, which separates highly reflective from weakly reflective growth-fault depocentre reflection packages. The concave upwards nature of the boundary is unusual for a growth fault termination, and the zone of weak reflections displays continuity with the more highly reflective package. Hence, the weakly reflective area between 3 and 2 is interpreted to be the hanging wall of the growth fault depocentre disrupted by mudstone intrusions similar to those seen in Jerudong. (4) Strongly reflective unit lying within the chaotic reflections though to represent a mudstone diapir. This reflective unit might be stratified non-mobile shales, or compacted formerly mobile shales that dewatered at a specific interval, possibly drained along the overlying counter-regional fault. (5) Touchdown geometry between the counterregional fault depocentre and the horizontal reflective package 4.
OUTCROP EXAMPLES OF MUDSTONE INTRUSIONS FROM THE JERUDONG ANTICLINE
393
Fig. 17. Schematic illustration of the early stages of fault and mudstone intrusion development based on the Jerudong outcrops. Initial shale intrusions associated with the reactive phase of diapirism, in response to growth faulting may just develop hydraulic fractures at the top of the intrusion (where the stresses opposing hydraulic fracturing are least). Subsequent tapping of overpressure by conjugate fault systems developed during shale diapir collapse may trigger further intrusions.
The relationship of faults to intrusions potentially is highly varied. The larger normal faults appear to cause reactive rise of mobile mudstones in the footwalls of the faults (e.g. Morley & Guerin 1996; Van Rensbergen et al 1999). Subsequent to the initial reactive rise, the overpressured mudstones may start to rise though networks of hydraulic fractures which commonly follow existing fault planes. As the active phase of intrusion develops, fluids will be lost via the hydraulic fractures into the formation, which is likely to lead to significant volume loss from the main region of overpressured mudstones (commonly the reactive diapir core). Consequently, the roof sequence is likely to display conjugate extensional faults related to the collapse of the underlying diapir. In the Niger Delta such collapse conjugate fault sets associated with mudstone diapirs are well developed (e.g. Morley & Guerin 1996). In the Jerudong area just in the vicinity of the mudstone intrusions conjugate fault sets are intensely developed (Morley et al 1998), their very localized development suggests they too could be related to volume loss in an underlying mobile mudstone mass. These conjugate faults may have propagated down into the
deflating core of the diapir and tapped pockets of remaining overpressured in the mudstones. The magnitude of the remnant overpressure was not enough to initiate new hydraulic fractures (by exceeding the magnitude of the minimum horizontal stress plus the tensile strength of the rock). However, upon arrival of the propagating fault at the top of the overpressure, hydraulic fracturing up the fault plane may only require the overpressure to exceed the magnitude of the minimum horizontal stress (if the fault plane has close to zero tensile strength). Consequently, the propagating faults may trigger mudstone intrusions and further loss of overpressure from the diapir core.
Conclusions The seismic expression of zones affected by mudstone intrusions (such as those outcropping in the Jerudong area) and associated overpressured fluids commonly appear as regions of chaotic data, or reduced reflectivity (e.g. Van Rensbergen et al. 1999). The mudstone intrusions are natural hydraulic
394
C.K. MORLEY
tensional fractures. Key features of mudstone intrusions are dyke, sill and laccolith geometries, intrusion along pre-existing weaknesses, in particular normal faults, local inflation of country rock shale units and occasional reverse and inverted normal faults due to mudstone intrusion. The source of the mobile mudstones is mostly from the Setap Formation, augmented by local pressure inflation and mobilization of bedded shales in the Belait Formation country rock. Diapir rise aided by hydraulic fracturing and stoping indicates that the roof sequence is weaker and diapir rise easier than for models (based on salt tectonics) that assume diapir rise must overcome the brittle shear strength of intact rock forming the roof sequence. The crosscutting mudstone dykes, mudstone intruded fault zones, and associated cataclastic deformation form planar permeability barriers that can highly compartmentalize the reservoir rock and significantly reduce the area of reservoir rock that could be effectively produced of hydrocarbons. More positively for hydrocarbon exploration, mudstone intrusions along faults or into breached top-seals will enhance their sealing capacity and intrusions in general may be responsible for opening up shortlived hydrocarbon migration pathways in basins. I would like to thank G. Ingram and R. Johns for helpful constructive reviews of an earlier version of the manuscript. Several colleagues and students are acknowledged for helpful discussions and work concerning the injected mudstone features including P. Van Rensbergen, Z. Hj Ahmad, A. Hurst, J. Warren, R. Hillis, D. Swarbrick, M. Tingay and P. Kengmar.
References BACK, S., MORLEY, C.K., SIMMONS, M.D. & LAMBIASE, J.J. 2001. Depositional environment and sequence stratigraphy of Miocene deltaic cycles explosed along the Jerudong anticline, Brunei Darussalam. Journal of Sedimentary Research, 71, 913-921. BARBER, A.J., TJOKROSAPOETRO, S. & CHARLTON, T.R., 1986. Mud volcanoes, shale diapirs, wrench faults and melanges in accretionary complex, eastern Indonesia. American Association of Petroelum Geologists Bulletin,!^ 1129-1141. DAILLY, G.C., 1976. A possible mechanism relating progradation, growth faulting, clay diapirism and overthrusting in the regressive sequence of sediments. Bulletin of Canadian Petroleum Geology, 24,92-116. GRADE, K. 2000. Mud volcanoes in deep water Nigeria. Marine and Petroleum Geology, 17, 959-974.
GULIEV, I.S. 1992. A review of mud volcanism. Translation of the report by Azerbaijan Academy of Sciences Institute of Geology, 65pp. JAMES, D.M.D. 1984. The Geology and Hydrocarbon Resources of Negara Brunei Darussalam. Special Publication, Muzium Brunei and Brunei Shell Petroleum Company Berhad. LONGERGAN, L. & CARTWRIGHT, J.A. 1999. Polygonal faults and their influence on deep-water sandstone reservoir geometries, Alba Field, United Kingdom central North Sea. American Association of Petroelum Geologists Bulletin, 83,410-432. MORLEY, C.K. & GUERIN, G. 1996. Comparison of gravitydriven deformation styles and behaviour associated with mobile shales and salt. Tectonics, 15, 1154-1170. MORLEY, C.K., CREVELLO, P. & ZULKIFLI AHMAD 1998. Shale tectonics-deformation associated with active diapirism: the Jerudong Anticline, Brunei Darussalam. Journal of the Geological Society of London, 155, 475-490. SANDAL, S.T. 1996. The Geology and Hydrocarbon Resources of Negara Brunei Darussalam, (1996 revision). Brunei Shell Petroleum Company/Brunei Museum, Syabas Bandar Seri Begawan, Brunei Darussalam, 243 pp. SIMMONS, M.D., BIDGOOD, M.D., BRENAC, P., CREVELLO, P.D., LAMBIASE, J.J. & MORLEY, C.K. 1999. Microfossil assemblages as proxies for precise paleoenvironmental determination - an example form Miocene sediments of north-west Borneo. In: JONES, R.W. & SIMMONS, M.D. (eds) Biostratigraphy in Production and Development Geology. Geological Society, London, Special Publications, 152,219-241. STEWART, S.A. 1999. Seismic interpretation of circular geological structures. Petroleum Geoscience, 5,273-285. VAN RENSBERGEN, P. & MORLEY, C.K. 2000. 3D seismic study of a shale expulsion syncline at the base of the Champion delta, offshore Brunei and its implications for the early structural evolution of large delta systems. Marine and Petroleum Geology, 17, 937-958. VAN RENSBERGEN, P., MORLEY, C.K., ANG, D.W., HOAN, T.Q. & LAM, N.T. 1999. Structural evolution of shale diapirs from reactive rise to mud volcanism: 3D seismic data from the Baram delta, offshore Brunei Darussalam. Journal of the Geological Society of London, 156,633-650. VENDEVILLE, B.C. & JACKSON, M.P.A. 1992. The rise of diapirs during thin-skinned extension. Marine and Petroleum Geology, 9, 331-353. WILFORD, G.E. 1960. The geology and mineral resources of Brunei and adjacent parts of Sarawak with description of Seria and Miri oilfields. Memoir 10, The Geological Survey Department, British Territories in Borneo, 2nd edition, Brunei State.
Re-evaluation of mobile shale occurrences on seismic sections of the Champion and Baram deltas, offshore Brunei PIETER VAN RENSBERGEN1 & CHRISTOPHER K. MORLEY2 l
Renard Centre of Marine Geology, Universiteit Gent, Krijgslaan 281-S8, 9000 Gent, Belgium (e-mail: pieter_vanrensbergen@yahoo. com) Department of Petroleum Geosciences, Universiti Brunei Darussalam, Bandar Seri Begawan 2028 Brunei Darussalam Abstract: 3D seismic data in the Baram and Champion delta provinces offshore Brunei show that regions thought to be occupied entirely by chaotic seismic data and conventionally interpreted as shale diapirs, are regions of dimmed, but coherent reflectivity. Such data indicate shale diapir masses are not present, instead dimming can be attributed to sediment intrusive complexes, overpressured fluids and gas clouds, or processing artefacts. In this way significant delta structures are masked on 2D seismic data, which are important to interpret delta tectonic evolution. The Middle Miocene-Recent Champion and Baram deltaic provinces are characterized by typical gravity tectonics-related structures. However, being situated on an active margin they are also affected by episodic development of contractional structures, which are located on older reactive shale bulges and result in inversion of motion on some growth faults. The emplacement of shale pipes, gas clouds and intrusive complexes is generally relatively late (Pliocene) in comparison with the underlying reactive diapirs (Late Miocene) and their emplacement events may be separated in time by several million years. Late overpressured systems may be related to phases of pore fluid pressure increase during or following periods of inversion tectonics, which resulted in phases of enhanced fluid migration in the basin, where fluids were either expelled laterally oceanwards, or vertically.
Current models of shale diapirism in a deltaic setting (Morgan et al 1968; Bruce 1973; Evamy et al 1978; Doust & Omatsola 1990; Cohen & McClay 1996; Morley & Guerin 1996; McClay et al 2003) assume ductile flow of 'mobile shales' from below the delta wedge to a frontal pro-delta bulge. In conventional seismic interpretation, these 'mobile shales' can be defined as overpressured shale characterized by a chaotic seismic facies that occurs in ridges, vertical intrusions or dome-shaped forms. Overpressured shale can also occur as a stratified seismic facies, in which case it is not considered to be mobile (Morley & Guerin 1996). Although overpressure can be measured in wells and in some cases be detected by seismic velocity analysis (Musgrave & Hicks 1968), shale mobility is an interpretation, partly based on the similar occurrence, shape and seismic facies between shale and salt diapirs and the similarities of the associated stratal patterns (e.g. divergent reflections in withdrawal basins). But, where salt is characterized by a strong top-of-salt reflection 'mobile' shale is often not; 'mobile' shale is not confined to a stratigraphic interval (Westbrook & Smith 1983; Morley & Guerin 1996). The chaotic facies has no clear outline and strongly depends on the resolution of the seismic tool. There are significant differences in the mechanical behaviour of salt and shale. Salt will deform as a high-viscosity fluid dependent on the temperature
and pressure. The mobility of fine-grained sediments is a function of compaction and overpressure of pore fluids (Hedberg 1974; Bruce 1984; Brown 1990; Doust & Omatsola 1990; Morley & Guerin 1996). Morley and Guerin (1996) suggest that overpressure increase may convert some stratified, non-mobile, shale units to 'mobile' shale. Dewatering of 'mobile' overpressured shale will cause the shale to collapse (e.g. Graue 2000) and will halt diapiric rise. The initial properties of permeability and density cannot be re-instated after collapse and unlike salt, shale mobilization can probably not be repeated. These differences should be manifest as variations in the deformation styles associated with the two materials but to date detailed documentation of the similarities and differences of shale deformation styles with gravity tectonics with respect to salt is sparse. Most literature about shale diapirs and delta tectonics is based on the classical model of shale (or salt) diapirs. Only recently have observations based on 3D seismic data (Van Rensbergen et al 1999; Van Rensbergen & Morley 2000; Graue 2000; Cooper 2001) or detailed cases studies (Correggiari et al 2001; Sumner & Westbrook 2001) from various depositional environments suggested alternative interpretations for shale diapirs. All these authors suggest that the observed shale diapirs are related to fluid flow and sediment fluid content rather than ductile flow from shale. In the light of these observations the
Fig. 1. Structural map of Brunei redrawn from Sandal (1996).
validity of the term 'shale diapir' can be questioned. Sediment intrusions may be shale now, but were they emplaced as shale or as mud? In the literature, the words 'shale', 'mud' and 'clay' are used virtually interchangeably when discussing subsurface mobilization of argillaceous sediment. In common usage, 'mobile shale' may indicate any argillaceous material ranging from very high percentage water, fluid muds (>70% water), through plastic clays, to more consolidated shales. Since 'mobile shale' is often an unqualified interpretation or unspecific regarding material properties, it is, in this paper, used between quotation marks to indicate the difference with more specific usage of shale as a consolidated rock. In this paper, we combine 3D seismic case studies on the Brunei shelf with regional 2D shelf-to-basin sections (Fig. 1) to document and discuss the styles and mechanisms of sediment mobilization in a deltaic setting. The conventional interpretation of 'mobile shale' and 'shale diapirs' in the Baram and Champion deltas, offshore Brunei, is checked against the outcome of 3D seismic studies of the same structures. The results of this comparison are used to re-interpret 2D seismic sections of 'shale diapirs' in the delta toe region.
Geological setting The Late Miocene Champion delta and the Pliocene Baram delta (Fig. 1) largely form the offshore and onshore geology. The Champion delta forms the eastern part of offshore Brunei and has typical graben structures at the delta top, akin to those in the Niger delta (Fig. 2a). The Baram delta forms the western part of offshore Brunei and is characterized by parallel, down-to-basin growth faults (Fig. 2b). Both deltas are built on a substratum of marine shale, the Setap shale. The Brunei region was an active margin until the Miocene, but while active subduction may have ceased in the Early Miocene, important contractional episodes continued and their imprint remains in the modern stresses of the delta (Tingay et al 2003). Important contractional episodes occurred during the Middle Miocene, Late Miocene and Late Pliocene (James 1984). Uplift of the hinterland caused cannibalization of older delta sequences and westward migration of sedimentary depocentres and delta development (Sandal 1996). Changes in stress regime have played a major role in the deformation of the overpressured shale, especially at the inner part of the Brunei shelf. At the outer part of the shelf, the
RE-EVALUATION OF MOBILE SHALE OFFSHORE BRUNEI
397
Fig. 2. (a) Line drawing of a regional seismic section of the outer part of the Champion delta, eastern Brunei offshore area, (b) Line drawing of a regional seismic section of the offshore part of the Baram delta, western Brunei offshore area (after Sandal 1996).
delta structures are akin to those on passive margins and it is our opinion that the results of this study are significant not only for deltas developed along active margins, but also for those on passive margins. The Setap Formation The Setap Formation forms the substratum for the prograding Champion and Baram deltas (Sandal 1996). It is a shelfal to deep marine shale of widely varying age. Onshore it has a thickness in excess of 3 km and is of Early to Middle Miocene age. The Setap Formation is found in outcrop at the Jerudong Anticline where it consists of shales with thin sandstone beds (Morley et al 1998). Offshore, it consists of undercompacted shales (with generally thin interbedded sandstone layers) and ranges in age from Early Miocene to Quaternary with the top of the shales younging in a NW direction. In the offshore area, the Setap Formation is characterized by a chaotic seismic facies at the base of the stratified overburden. Since the Setap shale is the basinward equivalent of the deltaic formations, it is not always bounded by a coherent reflection at the top. In some cases, the top of the shale can be seen as
a zone of high-amplitude reflections with some smile-effects, probably related to over-migration and which may indicate lower acoustic velocities of the undercompacted shale. Such reflections probably image a facies change rather than stratigraphic horizons. The thickness of the Setap Formation in the offshore areas is not known, older shales (Temburong formation) and the Crocker-Raj ang accretionary complex probably underlie it. The Champion delta The Champion delta complex is up to 300 km wide, forms a large part of onshore Brunei, the eastern offshore shelf and continues far into offshore Sabah. The delta is actually more likely to be an amalgamation of several deltas than a single 'Champion river' delta (Lambiase pers. comm. 1999). The deltaic sequences prograded in a NW direction and accumulated over 10 km of sediment from the Middle to Late Miocene. The inner part of the delta has been tectonically deformed during the Late MiocenePliocene in a series of synclinal depocenters bounded by shale-cored inversion anticlines (Sandal 1996; Watters et al 1999; Morley et al 2003). The
398
P. VAN RENSBERGEN & C.K. MORLEY
most basinward inversion anticline is the Ampa anticline that separates the Champion depocentre from the younger Baram depocentre (Fig. 1). The structure of the undeformed, offshore part of the Champion delta (Fig. 2a) resembles, but is smaller in size than, the offshore structure of the Niger delta (Morley & Guerin 1996; Cohen & McClay 1996). The delta toe consists of a succession of toe thrusts, which steepen landwards (Fig. 2a; James 1984). Chaotic seismic zones traditionally interpreted as 'mobile shale' seem to affect some of the thrusts and become more common landwards. At the shelf edge, a series of counter-regional growth faults occur, of which the Perdana growth fault is the oldest and the largest. The Perdana growth fault (or Frigate fault in Sandal 1996) is over 100 km long and probably continues into the eastern part of the Ampa shale ridge. It bounds the Perdana depocentre that accumulated about 4.5 sec (about 6.75 km) of sediment since the Late Miocene and covers over 1400 km2. Part of the depocentre lies in the adjacent territory of Sabah. The Perdana depocentre is a typical withdrawal graben bounded by paired counter-regional and regional growth faults. At its SE margin, the depocentre is bounded by a series of en echelon regional growth faults (Amcott fault, Champion fault). Some of those faults, like the Champion fault, have been deformed in a series of compressional phases throughout the Pliocene (James 1984). Stratigraphically, the Perdana depocentre is characterized by two regressional sequences (Late Miocene and Plio-Pleistocene), separated by a transgression at the end of the Miocene. The Late Miocene sequence is strongly divergent to the NE and converges into the Peragam anticline to the south. The Peragam anticline is an asymmetric 'mock turtle back' anticline formed during the Late Miocene by subsidence at the Perdana depocentre in the North and the Champion fault in the South. The Plio-Pleistocene regressional sequence diverges into the Perdana fault to the north. The Champion growth fault at the landward margin of the Perdana depocentre formed at the end of the Miocene, slightly earlier than the Perdana growth fault. The stratigraphy in the hanging wall of the Champion fault is complicated by intense conjugate normal faulting (Sandal 1996) and is not discussed here. During the Pliocene, sediment supply shifts from the eastern Champion area to the west, where the rapidly prograding Baram delta develops. Progradation of the shelf edge in the eastern offshore is stalled by a series of counter-regional growth faults.
The Baram delta The Pliocene Baram delta prograded rapidly over the eastern margin of the Champion delta and accu-
mulated over 6 km of sediment. Its western expansion is contained by the Luconia platform in Sarawak, marked by the West Baram Line (e.g. Hutchison 1996). A maximum thickness of Baram delta sediments is reached in the hanging wall of the Outer Shelf Growth Fault, close to the present day shelf edge. The structure of the Baram delta is characterized by parallel, down-to-basin growth faults. The growth faults strike E-W to NE-SW, subperpendicular to the N-S to NW-SE direction of sediment supply by the Baram River (Fig. 1). At both the eastern and the western margin of the delta, the fault orientation curves towards the north, parallel to pre-existing structures, the Luconia platform margin and the Magpie ridge respectively. The Baram delta is underlain by the western, marginal part of the Champion delta complex. The delta toe has basically the same style as the Champion delta toe, a series of thrusts, steepening landwards. Near the shelf edge, the steep reverse faults bound narrow anticlines, with a chaotic core, interpreted as shale diapirs (Sandal 1996). Shale diapirs occur near the shelf edge but are less numerous than at the Champion delta. Stratigraphically, the Baram delta (including the Late Miocene sediments) is formed by a single regressive, aggradational sedimentary series. The transgression at the end of the Miocene is not noticed here, probably because it coincides with an increase in sediment supply and the development of the large Pliocene Baram delta. Progradation was rapid during the Early Pliocene but stalled at the end of the Early Pliocene when the Outer Shelf Growth Fault developed (Fig. 2b).
A short overview of the classical theory of shale diapirsm Following models for the development of salt diapirs (e.g. Vendeville & Jackson 19920,6; Morley & Guerin 1996), four phases in 'shale' diapir activity can be recognized. These are (1) reactive diapirism, which responds to differential loading within the prograding delta wedge; (2) active diapirsm, which occurs when the diapiric body lifts up, pushed aside and replaces its overburden; (3) passive diapirism, which occurs by downbuilding of the flanks of a diapir by lateral removal of mobile material at depth, and diapir collapse. Reactive diapirism is flowage of a mobile unit into a higher region without piercement of the overlying strata. Flowage is caused by movement from a region of higher pressure to lower pressure, which may be caused by thinning of the overburden by faulting, erosion, or local thickening of the overburden by uneven deposition. While most reactive diapirism is associated with faulting, differential
RE-EVALUATION OF MOBILE SHALE OFFSHORE BRUNEI
399
Fig. 3. Structural elements related to mobile shale in a deltaic setting based on section from the Niger delta by Doust & Omatsola (1990) and Morley & Guerin (1996).
loading by a delta can also create reactive diapirs. For example, expulsion of mobilized sediment by differential loading from below the delta front to the pro-delta zone, is interpreted to cause subsidence in an expulsion syncline (depobelt) at the delta front and bulging of 'shale' in the pro-delta area (Doust & Omatsola 1990; McClay etal 2003). On 2D seismic sections, the landward flank of the reactive shale bulge is often interpreted as a counter-regional growth fault, down-dip from a depobelt (Figs 3 and 4; Doust & Omatsola 1990). A younger regional growth fault often occurs at the oceanward flank of the reactive bulge. On 2D regional seismic data reactive bulges are characterized by a chaotic seismic facies and often occur as shale ridges between backto-back growth faults. These syndepostional structures occur at the base of the delta and are early structures in delta development. Active 'shale' diapirism (Fig. 5) occurs when the • shale mass uplifts and displaces its roof in response to pressure exerted by the diapiric body (Vendeville & Jackson 1992a,£>). As in salt diapirism, this pressure was previously thought to be buoyancy of lowdensity shale or isostasy in response to differential loading of a mobile substratum. Such diapirs are associated with syn-kinematic development of adjacent synformal withdrawal basins. Subsequent dewatering commonly leads to major collapse structures on diapir crests and large velocity push-downs if the base of the diapir is imaged (Morley 2003). Intrusion of fluidized mud as dykes and sills represent an active phase of 'mobile shales'. Such intrusions (including pipes) may form a halo around a 'traditional' active diapir core, or may be the only manifestation of active 'mobile shale' rise. Offshore the Niger delta, shale diapirs occur in an almost continuous belt at the delta front (Evamy 1978). The magnitude of sediment mobilization decreases towards the delta toe zone, where a chaotic seismic facies, conventionally interpreted as 'mobile shale', occurs in the core of the most landward toe thrusts.
Whereas reactive diapirs of 'mobile shale' are early syndepostional structures, active diapirs root in a deeply buried source layer of 'mobile shale' and are thus younger features in the delta evolution.
Re-evaluating the chaotic seismic facies in the Champion and Baram deltas, Brunei In this section, some examples of chaotic seismic facies, conventionally interpreted as 'mobile shale', are checked against the outcome of 3D seismic studies of the same structures. It appears that the chaotic seismic facies that characterizes 'mobile shale' is in fact due to a variety of reasons, not all geological. Some chaotic seismic facies occurrences on 2D seismic data appear stratified when imaged with 3D seismic data and the width of sedimentary intrusions, if present, appear much smaller than previously estimated. Therefore, the role of 'mobile shale' in the structural evolution the Baram and Champion deltas needs to be reconsidered. A chaotic seismic facies may correspond to a geological feature or may be an artefact. Large slumps, stacked irregular surfaces, densely faulted sediment layers, and the like will give a chaotic seismic facies on 2D seismic data. Hydrocarbon seepage through sandstones may create hydrocarbon-related diagenetic zones that appear as high-amplitude chaotic reflections on seismic sections (O'Brien & Woods 1995). Also reflection-free or low-amplitude seismic facies may appear chaotic due to amplitude scaling and addition of low-frequency migration noise. A reflection-free facies is caused by the lack of impedance contrasts within a depositional unit. Stratification may not have been present or may have been wiped-out by post-depositional effects (e.g. liquefaction of uncompacted muds). Increase of overpressure also reduces acoustic velocity differences between sand and shale and causes dimming of seismic reflections (Maucione & Surdam 1997).
400
P. VAN RENSBERGEN & C.K. MORLEY
Fig. 4. Main characteristics of reactive shale diapirs in a deltaic setting according to the classical theory (based on Doust & Omatsola 1990). 3D seismic data offshore Brunei show similar depobelts but indicate that the 'mobile shale' and the counter-regional growth faults appear to be acoustic artifacts on 2D regional seismic data. On 3D seismic data reflections continue from the depobelts into the chaotic seismic facies, their oceanward termination is not observed.
Amplitude wipeout effects also occur below strong vertical variations in acoustic impedance e.g. faults, fluid or gas pockets, that cast an acoustic shadow over normally stratified deposits. At the limit of the seismic penetration, signal attenuation decreases data quality and may result in a chaotic seismic facies, especially in low-reflective deposits. The resolution of 3D seismic data is much higher then the resolution of 2D seismic profiles with similar signal frequencies and often resolves subsurface structures unnoticed on 2D seismic profiles. One reason for this is the close line spacing, which makes it possible to achieve very detailed mapping. The 3D seismic data offshore Brunei have a line and trace spacing of 12.5 m. Another reason is that 3D migration reduces the Fresnel zone to a minimum (Brown 1999). 2D seismic profiles can only be migrated in one direction; the post-migration Fresnel zone is narrow in the direction of the profile but remains wide in the perpendicular direction. As a result, out of plane reflections are smeared onto the 2D seismic profile, which decreases the sharpness of reflections and adds irrelevant reflections and noise to the section. In delta provinces, chaotic seismic facies often occur at the base of the delta, between back-to-back faults and as vertical disruptions of the stratigraphic sequence in areas of mud volcanoes and hydrocarbon seeps. The seismic facies is, on 2D seismic profiles especially, not bounded by continuous or traceable reflections. In these cases the chaotic
Fig. 5. Main characteristics of active shale diapirs in a deltaic setting according to the classical theory (based on Morley & Guerin 1996). Again, 3D seismic data over piercing diapirs offshore Brunei indicate that the 'mobile shale' facies is an artifact on 2D regional seismic lines that exaggerates the size of vertical intrusions.
seismic facies is conventionally interpreted as 'mobile' shale and considered to behave as a viscous material. Figures 6 to 11 show examples of such chaotic seismic facies occurrences in the Baram and Champion deltas, offshore Brunei. These examples show that the chaotic seismic facies, especially when imaged on 3D seismic sections, is not always chaotic and cannot be interpreted as 'mobile shale'. The examples question the validity of the current shale diapir models that were developed on basis of 2D seismic data and that attribute large geological significance to zones without seismic information. Figure 6 is an example of a chaotic seismic facies between back-to-back faults in the Baram delta province. On 2D seismic lines, it is interpreted as a diapiric shale ridge in the footwall of the large Outer Shelf Growth Fault (McClay et al 1998). On 3D seismic data, we see that the shale ridge is in fact a horst block of stratified pro-delta clays. The landward flank consists of several, relatively small counter-regional faults. These counter-regional faults do not occur at the delta front transition from sand to shale as inferred at the Niger delta (Doust & matsola 1990) but occur in pro-delta deposits, oceanward of the delta front. They bound a syncline with densely faulted pro-delta deposits, overlain by an aggrading and prograding delta top succession. There is no indication for oceanward expulsion of mobilized sediment (Fig. 4), which prompted Van Rensbergen & Morley (2001) to suggest that expulsion synclines may be formed by prograding, localized compaction and expulsion of fluids rather then by lateral expulsion of mobile shale. Figure 7 is an example of a chaotic seismic facies
RE-EVALUATION OF MOBILE SHALE OFFSHORE BRUNEI
401
Fig. 6. Comparison of a 2D seismic section with a 3D seismic section over a shale ridge in the footwall of the Outer Shelf growth fault. This typical example of a reactive 'shale diapir' appears to be a stratified horst block on 3D seismic sections (see location on Fig. 2b).
at the base of the Champion delta, eastern offshore area. Continuous delta top reflections stop abruptly and the seismic facies changes into a chaotic seismic facies. Apparent landward dipping reflections (possibly caused by diffractions or migration of noise; Yilmaz 2001) emphasize the abrupt facies change and are conventionally interpreted as an early counter-regional growth fault bounding a 'mobile shale' ridge (e.g. Doust & Omatsola 1990). On 3D seismic images however, reflections continue from the syncline into the chaotic seismic facies without much, if any, offset. The interpreted counterregional fault appears to be a series of aligned bright spots (D-event). Below the bright spots, reflection continuity and amplitude decrease drastically. This may be due to the low impedance contrast within pro-delta shales but also may be the result of postdepositional diagenetic or catagenetic effects. Van Rensbergen & Morley (2001) attributed the D-event and the coincident abrupt decrease in reflection continuity and amplitude to a front of high overpressure related to hydrocarbon maturation and gas generation. Overpressure increase can cause reduction of the impedance contrasts; reflection continuity may decrease in bedded but fractured shale. Figure 8 is an example of a stock-like shale diapir, located at the oceanward flank of the Ampa inversion anticline. It consists of a vertical chaotic zone of about 2 sec height with a 5 km wide base and narrowing to less then 1 km wide to the top. Numerous bright spots occur near the top of the diapir. According to the conventional theory, a phase of reactive diapirsm (coincident with subsidence along the Late Miocene counter-regional fault) may have been followed by a phase of active vertical intrusion or passive downbuilding of the diapir flanks (coincident with the formation of the Pliocene regional
growth fault). The Ampa diapir is similar to tall diapirs found in diapir belts near the shelf break in delta provinces, e.g. the Champion and Niger deltas. When studied in detail, it is seen that Late Miocene reflections continue into the chaotic zone, but reflections become discontinuous and reflection amplitude decreases (Fig. 9). This zone of subdued reflectivity narrows upwards, in pace with the prograding shelf break. At the crest of this zone, at about 2.5 s TWT, two vertical seismic chimneys of about 500 m diameter occur (Fig. 10). The vertical pipes end in a wider head, interpreted as buried mud volcanoes or shallow sills of Late Pliocene age (Van Rensbergen et al 1999). Similar to the Perdana expulsion syncline (Fig. 7), the Late Miocene sequence at the base of the Ampa diapir is rotated into a migrating expulsion syncline with subdued reflectivity (Fig. 9). The depocentre is bounded oceanwards by a small shale bulge, possibly in the back of a counter-regional fault, but the data quality is too poor to resolve the fault or the internal structure of the shale bulge. The shale bulge and the expulsion syncline are syndepositional structures. As for the Perdana expulsion syncline, the subdued reflectivity is interpreted to be a postdepositional effect, probably caused by large pore fluid overpressure increase in pro-delta shales (Van Rensbergen et al 1999). Evidence that high pore fluid pressures occurred within the pro-delta shales is given by the occurrence of the two vertical chimneys at the crest of the 'chaotic' zone (Fig. 10). The two vertical chimneys most likely formed by hydrofracturing and fluid expulsion from the crest of a pressure compartment in the pro-delta shales. Outcrop examples of injected muds along hydrofractures are found onshore Brunei (Morley 2003). Van Rensbergen et al (1999) estimated that
402
P. VAN RENSBERGEN & C.K. MORLEY
Fig. 7. Comparison of a 2D seismic section with a 3D seismic section of the contact between stratified overburden with chaotic 'mobile shale' at the base of the Champion delta (see location on Fig. 2a). On the 2D seismic section, the contact is conventionally interpreted as a counter-regional growth fault. On a 3D section at the same location, it appears that subdued reflections continue into the chaotic seismic facies. A series of bright reflections, previously interpreted as a counter-regional fault, are attributed to an overpressured hydraulic front in pro-delta sediments (Van Rensbergen & Morley 2000).
Fig. 8. Comparison of the conventional and the alternative interpretation of the Ampa diapir on a 3D seismic line 1300. It appears that the Ampa diapir is not a massive intrusion of mobile shale but consists for over 90% of stable countryrock with subdued reflectivity, maybe due to overpressure increase in pro-delta sediments. At the crest of the lowamplitude zone a vertical seismic chimney rises to a mudsill or buried mud volcano.
mobilized sediment makes up less then 10% of the volume of the entire conventional 'shale diapir' and only occurs in the core of the vertical shale pipes and maybe in the Late Miocene shale bulge. Figure 11 shows chaotic seismic facies occurrences from a 2D regional seismic section in front of the shelf edge, at the eastern offshore part. The Figure shows sections from the delta toe towards the shelf edge. Patches of chaotic seismic facies occur at thrust anticlines and increase in size towards the shelf edge (Fig. 2). In conventional seismic interpretation, these patches are interpreted as 'mobile shale' in the core of thrust anticlines, passing landwards into the 'diapir depobelt' (Fig. 3; Morley & Guerin 1996). This interpretation is supported by analogue sand box models, using a silicon putty as
anologue for the 'mobile shale' substratum. Figure lib resembles the example in Figure 5, where divergent reflection patterns at the flanks of the chaotic zones are interpreted as passive downbuilding of the adjacent depocentre. The sections in Figure 11 can be interpreted differently when the chaotic seismic facies is considered a no-event zone, a zone without seismic information. No-event zones at the steep limbs of thrust faults can be a migration effect; steeply dipping reflections can only be imaged using appropriate migration algorithms that can handle steep dips. Normal migration algorithms will create a zone without seismic information in the flank of the structure (e.g. figure 4.6-36 in Yilmaz 2001). The correct imaging of these structures is also a function of the width of the struc-
RE-EVALUATION OF MOBILE SHALE OFFSHORE BRUNEI
403
Fig. 9. Detail of 3D seismic line 1300 showing the continuation of coherent reflections into the "chaotic" seismic facies (see location on Fig. 8)
ture and the trace interval; narrow structures need a shorter trace interval to be imaged correctly (Yilmaz 2001). In addition to these possible artefacts, numerFig. 10. Isochron map of the zone of the Ampa diapir. ous bright reflections occur near the crests of the The base of the diapir is a bell-shaped zone of subdued anticlines. A 'bottom-simulating reflection' (a pos- but coherent overburden reflections. At the crest of this sible indication of gas hydrate accumulation below zone two vertical seismic chimneys occur. the sea floor) at anticlines one and four bears witness to gas migration and accumulation at the anticlines. Signal absorption and blankening are likely to occur zone disrupts the stratigraphy and has a halo of bright in such places. For example, in Figure lib the spots, similar to the Ampa diapir. Likewise, the strucchaotic seismic facies in the core of anticline five ture may represent a chimney of injected fluidized is reflection-free below irregular high amplitude sediment, rather then a massive diapiric uprising of a reflections in the upper part but consists of medium- shale mass. amplitude, low-frequency noise at the base. This is possibly the combined effect of acoustic blanking in the upper part and addition of migration noise in the Discussion lower part. If sediment fluidization and expulsion of mud breccia accompanied earlier phases of fluid The seismic sections discussed above are examples expulsion, shallow irregular sedimentary bodies of of zones with a chaotic seismic facies that are conchaotic seismic facies will occur and possibly cast ventionally interpreted as 'mobile shale' in a delta setting. Our understanding of delta tectonics is acoustic shadows over the underlying geology. The divergent reflection pattern and anticlinal largely based on the presence of a mobile substratum structures in Figure 1 Ib resembles the syn-kinematic subject to expulsion from below the delta wedge and development of synformal withdrawal basins adja- accumulation in diapirs and shale bulges. The cent to diapirs shown in Figure 5. In this case however, current models are well supported by scaled anathe geometry may also result from displacement logue models that successfully scaled down and imialong steepened toe thrusts. The reflection pattern is tated delta structural evolution (McClay et al. 2003). asymmetrical at the base but gradually becomes more Yet, the above examples show that the conventional symmetrical to the top as the thrust faults become interpretation does not always apply when mobile more vertical. Steep reverse faults are preferential shale structures are imaged in detail, especially with sites for fluid flow and fault valve behaviour in accre- 3D seismic data. The chaotic seismic facies observed offshore tionary fold-and-thrust belts. High fluid pressure permits continued activity on reverse faults as Brunei can be attributed to: they progressively steepen and associated anticlines become more tightly folded (Sibson 1990). Accumu- • Acoustic shadows below high impedance contrasts at faults, hydrocarbon (mainly gas) acculation of fluids in shallow reservoirs and below submulations; surface hydrates may cause the amplitude effects observed near the crests of the anticlines. Figure lie • Subdued reflectivity due to post-depositional diagenetic or catagenetic effects; reduction of does not show any typical stratal patterns; the chaotic
404
P. VAN RENSBERGEN & C.K. MORLEY
Fig. 11. Details of chaotic seismic facies occurrences from 2D seismic data in front of the shelf edge. The chaotic seismic facies is conventionally interpreted as 'mobile shale' (see Fig. 5) but may also be due to processing artifacts of steeply dipping reflections, or narrow structures and acoustic wipe-out below shallow bright spots, (a) Toe thrusts with no-event zone where reflections are steepest; (b) further landwards stratal patterns characteristic of lateral shale withdrawal occur but may also be the result of continued movement along steepening thrust faults; (c) Wide vertical noevent zone with a halo of bright reflections may in fact be a narrow vertical fluid conduit (e.g. the Ampa diapir).
• • •
impedance contrasts due to overpressure increase, loss of reflection continuity in bedded but fractured shale; etc.; Buried mud volcanoes or shallow mud sills; Narrow vertical shale pipes or maybe clusters of shale pipes; Possible processing artefacts; apparent dipping reflections in noise, addition of low-frequency migration noise and zones without reflections of steeply dipping strata.
The examples indicate that the amount of 'mobile shale' is extremely overestimated, if 'mobile shale' is present at all. On the 3D seismic data that were available for this study, no evidence of mobile, massive shale diapirs, similar to salt diapirs, was found. Instead, the chaotic seismic facies masks important delta structures; early pro-delta synclines, growth faults, thrust faults and associated reflection patterns and narrow vertical intrusions of fluidized sediment. In general, we find again an important difference between early delta structures, related to differential loading of undercompacted pro-delta mud by a pro-
grading delta wedge (reactive structures) and active sediment intrusions that occur late in the delta evolution, after important pore fluid overpressure increase. In the Ampa diapir both structures are superimposed but active injection of fluidized sediment occurred about 9 myr after the pro-delta expulsion syncline formed. Migrating pro-delta synclines are found at the base of the Champion delta; examples from the Perdana depocentre and the base of the Ampa diapir are shown above, but the characteristic reflection pattern of rotated divergent reflections with clearly aggrading shelf break reflections is also a typical aspect in the landward flank of Brunei's inversion anticlines (Morley et al 2003). The position of the inversion anticlines in Figure 1 probably corresponds more or less with the positions of late Miocene pro-delta synclines, later inverted into the anticlines. The small syncline in the footwall of the Outer Shelf growth fault in the Baram delta province may be a younger equivalent of such pro-delta expulsion syncline. The pro-delta synclines are probably the equivalent of shale withdrawal basins but there is no evidence on the seismic data that they
RE-EVALUATION OF MOBILE SHALE OFFSHORE BRUNEI
405
Fig. 12. Detail of the depocentre bounded by the Perdana-Champion paired growth faults (see localization on Fig. 2(a) shows the sequence stratigraphic interpretation of the Plio-Pleistocene sequence in the hanging wall of the Perdana fault. The Perdana fault forms a trap for land-derived organic matter deposited in turbidites (base Pliocene sequence) and prograding delta front deposits (Upper Pliocene - Pleistocene sequence).
actually from by mass mobilization of pro-delta mud. Ductile deformation can occur in normally consolidated, overpressured mudstone (Bolton & Maltman 1998), but it is not documented if ductile flow of this material can be sustained over several kilometres without dewatering and collapse. Progressive localized compaction of undercompacted material may account for part of the subsidence in pro-delta synclines (Van Rensbergen & Morley 2000) but has not been quantified. On the seismic data, no evidence of large rising shale masses (active diapirism) has been observed, especially not on the 3D seismic data. Active intrusion of mobilized sediment is interpreted to occur by injection of fluidized sediment in hydraulic fractures and vertical pipes (Van Rensbergen et al. 1999). A cluster of mudstone dykes and associated sills and lacoliths have been documented onshore Brunei (Morley et al 1998; Morley 2003). Fluid expulsion and sediment mobilization along hydrofractures requires large pore fluid overpressures. At the Ampa diapir, sediment injection occurred at about the time of peak oil generation offshore Brunei, most likely during successive phases of fluid injection, probably preceded by gas leakage into the overburden (Van Rensbergen et al 1999). Repeated fluid injections require a rechargeable source of pore fluid overpressure in a re-sealing system (e.g. a thick shale sequence, Engelder & Leftwich 1997). Gas generation from mature organic material is probably the most important overpressure mechanism in this case (Barker 1990; see discussion in Osborne & Swarbrick 1997). Valving and gas exolution is probably a mechanism to maintain the pressure drive for upward fracture propagation (Brown 1990). Gas
leaking into the overlying sediments increases the pore pressure in the overburden. This may facilitate hydraulic fracturing (Miller 1995), and causes localized carbonate cementation in shallow sands (O'Brien & Woods 1995), which may help to create isolated pressure compartments along the vertical chimney. Buoyancy-driven propagation of methane filled fractures in geopressured sediments has also been suggested as a mechanism for vertical gas migration (Nunn & Meulebroek 2002), or vertical gas and fluid migration may occur by an upsurge of fluid pressure fronts, in the form of a solitary shock-wave, after hydrofracturation of the seal containing highly overpressured pore fluids (Revil 2002). Upward flow of overpressured fluids may fragment and mobilize rocks along their path (Lewis & Byrne 1996; Morley 2003) and form mud volcanoes or shallow sills. The vertical chimneys disrupt the stratigraphy but do not create divergent reflection patterns typical for diapirism. Only at shallow depths can mud intrusion cause small-scale updoming of the overburden in the immediate vicinity of the feeder pipe (Van Rensbergen et al 1999; Morley 2003). Fluid and sediment injection chimneys at Ampa, interpreted to be related to overpressure increase caused by gas generation, root in the Late Miocene expulsion syncline with subdued reflectivity. This interpretation suggests that the expulsion synclines, at least at the Ampa diapir, accumulated sediments with source rock potential. Extensive source rocks of mappable extent for Brunei's hydrocarbons, despite numerous efforts, have not been found. The source of most of Brunei's hydrocarbons is derived from gas-prone land plant organic material (type III, van Krevelen classification; Tissot & Welte 1978)
406
P. VAN RENSBERGEN & C.K. MORLEY
Fig. 13. Overview of the styles of shale mobility in relation in relation with pore fluid overpressure; early syndepositional structures can be distinguished from later post-depositional injections of mobilized sediment that occur after large overpressure increase.
deposited in fluvial or lagoonal environments or locally in open marine shelf environments and in deep marine environments by turbidity currents (Sandal 1996). Expulsion synclines are bounded at their oceanward side by an anticlinal bulge or a counter-regional growth fault and form a trap for turbidity currents down the delta slope. Seismic sections of Miocene expulsion synclines do not show any detail about the depositional processes within the syncline. However, the sediments are probably very similar to those accumulated in the hanging wall of the large Pliocene Perdana counter-regional growth fault (Fig. 4). A line drawing from the eastern regional 2D seismic line shows how turbidites are ponded against the Perdana growth fault between 2 s and 3 s TWT (about 2-4 km deep). As the delta prograded, the depositional environment changed from pro-delta to delta front and a rapidly aggrading sequence of delta-front deposits in a shallow marine basin accumulated in the subsiding depocentre. Both the turbiditic series as well as the delta front deposits may accumulate land-derived organic material. If the Miocene expulsion synclines consist of similar deposits, they may be good source rock for part of Brunei's offshore hydrocarbons. This suggestion is supported by results of a recent
geochemical study of hydrocarbons and potential source rocks onshore and offshore Brunei by Curiale et al (2000). This study suggests that offshore Brunei oils originated from source rocks deposited in a neritic pro-delta environment with predominantly allochtonous land-derived organic matter, but with admixtures of autochtonous algal derived organic matter from the water column directly above the site of deposition. The rate of allochthonous to autochthonous organic matter in the source increases to the NE. This may be related to an increase in the amount of contributed terrigenous organic matter in the NE areas of offshore Brunei, the main depocentre of the Miocene Champion delta. Miocene pro-delta synclines probably trapped the bulk of terrigenous organic matter that was transported over the shelf break; it is unlikely that during the Miocene large quantities of land-derived organic matter accumulated oceanward of the Champion pro-delta synclines. Yet, bright spots and BSRs bear witness of gas accumulations in the delta toe zone, especially at the crests of toe thrust anticlines. We suppose that these gasses also originate from source rocks below the main part of the delta. Similar to fold and thrust-belts, but on a smaller scale, lateral fluid flow from below the delta
RE-EVALUATION OF MOBILE SHALE OFFSHORE BRUNEI
to the delta toe may occur along fracture zones. Grauls & Baleix (1994) document a recent pulse of hydrocarbon fluid flow from depths more than 3200 m in the footwall of the Perdana counter-regional fault, relatively close to the shelf edge. Fluid migrating from below the delta can be expected to move up along reverse faults in the outer delta and pro-delta region. As discussed before, steep anticlines and fluid accumulation in shallow reservoirs are difficult to image on seismic data. If in addition vertical fluid chimneys occur near the shelf edge, they will add to the abundant occurrence of zones with a chaotic seismic facies.
Conclusions The study of delta tectonics is largely based on the interpreted occurrence of a 'mobile shale' substratum. 'Mobile shale' supposedly corresponds to differential loading by a delta by shale withdrawal from below a delta and shale accumulation in reactive shale bulges and active rising diapirs. This interpretation is based on characteristic reflection patterns in association with zones of chaotic seismic facies that occurs in bulges, diapirs, etc., conventionally interpreted as 'mobile shale', analogue to salt mobilization. As seismic resolution improves, there are increasing indications that 'mobile shale' deformation and salt diapirism differ significantly. In areas like the Niger Delta there are good examples of shale diapirs that follow the reactive-active-passive-collapse phases seen in salt diapirs. However, there are problems widely applying such models to other deltas, like the Baram and Champion Deltas. Much of the areas initially thought to be shale diapirs, characterized by poor data quality on seismic reflection data, appear to be regions of dimmed coherent reflectivity. The chaotic seismic facies is caused by a variety of effects that reduces the quality of the seismic data and is not always geologically relevant. Therefore, the amount of mobile shale is extremely overestimated and the chaotic seismic facies masks relevant delta structures that are important to understand and interpret the delta structural evolution. In the Baram delta, initial loading of the over pressured Setap Formation substratum by Middle Miocene deltaic deposits probably resulted in reactive diapirism. Much of the subsequent history has involved not the movement of large shale diapirs, but emplacement of intrusive sediment complexes and the development of pipes and gas clouds. The late overpressured systems may be related to phases of pore fluid pressure increase during or following periods of inversion tectonics, which resulted in phases of enhanced fluid migration in the basin, where fluids were either expelled laterally oceanwards, or vertically.
407
This paper is based on detailed case studies at the University of Brunei Darussalam for which the data was kindly supplied by Brunei Shell Petroleum with permission of the Petroleum Unit. PVR is funded by the Fund for Scientific Research - Flanders (FWO). Reviewers D. James and G. Westbrook made numerous useful suggestions to improve an earlier version of the manuscript.
References BARKER, C. 1990. Calculated volume and pressure changes during the thermal cracking of oil to gas in reservoirs. American Association of Petroleum Geologists Bulletin, 74,1404-1413. BOLTON, A. & MALTMAN, A. 1998. Fluid-flow pathways in actively deforming sediments: the role of pore fluid pressures and volume change. Marine and Petroleum Geology,l5,2%l-291 BROWN, A.R. 1999. Interpretation of Three-Dimensional Seismic Data. American Association of Petroleum Geologists Memoir, 42. BROWN, K.M. 1990. The nature and hydrogeological significance of mud diapirs and diatremes for accretionary systems. Journal of Geophysical Research, 95, 8969-8982. BRUCE, C.H. 1984. Smectite dehydration-its relation to structural development and hydrocarbon accumulation in northern Gulf of Mexico Basin. American Association of Petroleum Geologists Bulletin, 68, 673-683. COHEN, H.A. & McCiAY, K. 1996. Sedimentation and shale tectonics of the northwestern Niger Delta front. Marine and Petroleum Geology, 13, 313-328. COOPER, C. 2001. Mud volcanoes of Azerbaijan visualized using 3D seismic depth cubes; the importance of overpressured fluid and gas instead of non-existant diapirs. Abstract book 'Origin, processes, and Effects of Subsurface Sediment Mobilisation on reservoir to regional scale'. September 2001 Gent, Belgium, p. 71. CORREGIARI, A., TRIMCARDI, F., LANGONE, L. & ROVER, M. 2001. Styles of failure in late holocene highstand prodelta wedges on the Adriatic shelf. Journal of Sedimentary Research, 71 (2), 218-236. CURIALE, J., MORELOS, J., LAMBIASE, J., & MUELLER, W.
2000. Brunei Darussalam, Characteristics of selected petroleums and source rocks. Organic geochemistry, 31,1475-1493. DOUST, H. & OMATSOLA, E. 1990. Niger Delta. In: EDWARDS, J.D. & SANTOGROSSI, PA. (eds) Divergent/ Passive Margin Basins. American Association of Petroleum Geologists Memoir 48,201-238. ENGELDER, T. & LEFTWICH J.T. Jr., 1997. A pore-pressure limit in overpressured South Texas oil and gas fields, in Surdam, R.C. ed., Seals and traps, and the petroleum system: American Association of PetroleumGeologists Bulletin 67,255-267. EVAMY, B.D., HAREMBOURE, J., KAMERLING, P., KNAAP, W.A., MOLLOY, F.A. & ROWLANDS, PH. 1978. Hydrocarbon habitat of Tertiary Niger delta. American Association of Petroleum Geologists Bulletin, 62,1-39.
408
P. VAN RENSBERGEN & C.K. MORLEY
GRAUE, K. 2000. Mud volcanoes in deepwater Nigeria. Marine and Petroleum Geology, 17, 959-974. GRAULS, D. & BALEIX, J.M. 1994. Role of overpressures and in situ stresses in fault controlled hydrocarbon migration. Marine and Petroleum Geology, 11, 734-742. HEDBERG, H.D. 1974. Relation of methane generation to undercompacted shales, shale diapirs and mud volcanoes. American Association of Petroleum Geologists Bulletin, 58,661-613. McCLAY, K.R. DOOLEY, T. & LEWIS, G. 1998. Analogue modeling of progradational delta sytems. Geology, 26,771-774. McCLAY, K.R. DOOLEY, T. & ZAMORA, G. 2003. Analogue models of delta systems above ductile substrates. In: VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, AJ. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216, 381-394. JAMES, D.M.D. 1984. The Geology and Hydrocarbon Resources of Negara Brunei Darussalam. Special Publication, Muzium Brunei and Brunei Shell Petroleum Company Berhad. LEWIS, J.C. & BYRNE, T. 1996. Deformation and diagenesis in an anciant mud diapir, southwest Japan. Geology, 24, 303-306. MAUCIONE, D.T. & SURDAM, R.C. 1997. Seismic response characteristics of a regional scale pressure compartment boundary, Alberta Basin, Canada. In R.C. Surdam, Seals traps and the Petroelum system, American Association of Petroleum Geologists Memoir 67,269-281. MILLER, T.W. 1995. New insights on natural hydraulic fractures induced by abnormally high pore pressures. American Association of Petroleum Geologists Bulletin, 79,1005-1018. MORGAN, J.P., COLEMAN, J.M. & GAGLINO, S.M. 1968. Mudlumps: Diapiric structures in Mississippi delta sediments. In: BRAUNSTEIN, J. & O'BRIEN, G.D. (eds) Diapirism and Diapirs, American Association of Petroleum Geologists Memoir, 8, 145-161. MORLEY, C.K. 2003. Outcrop examples of mudstone intrusions from the Jerudong anticline, Brunei Darussalam, and interferences for hydrocarbon reservoirs. In: VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, AJ. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216,381-394. MORLEY, C.K., CREVELLO, P., & ZULKIFLI HAJI AHMAD. 1998. Shale tectonics and deformation associated with active diapirism: the Jerudong Anticline, Brunei Darussalam. Journal of the Geological Society of London, 155,475-490. MORLEY, C.K. & GUERIN, G. 1996. Comparison of gravitydriven deformation styles and behaviour associated with mobile shales and salt. Tectonics, 15,1154-1170. MORLEY, C.K, CREVELLO, P., BACK, S. VAN RENSBERGEN, P. & LAMBIASE, J.J. 2003. Characteristics of repeated tectonic inversion events during the Miocene-Recent in a large delta province on an active margin, Brunei Darussalam, Borneo. Journal of Structural Geology, 25,1147-1169. MUSGRAVE, A.W. & HICKS, W.G. 1968. Outlining shale masses by geophysical methods. In: BRAUNSTEIN, J. &
O'BRIEN, G.D. (eds) Diapirism and Diapirs, American Association of Petroleum Geologists Memoir, 8,122-136. NUNN, J.A., & MEULEBROEK, P. 2002. Kilometer-scale upward migration of hydrocarbons in geopressured sediments by buoyancy-driven propagation of methane-filled fractures. American Association of Petroleum Geologists Bulletin, 86,5,907-918. O'BRIEN, G.W., & WOODS, E.P. 1995. Hydrocarbon related diagenetic zones (HRDZs) in the Vulcan Sub-basin, Timor Sea: recognition and exploration implications. APEA JournalSS (1), 220-252. OSBORNE, M.J. & SWARBRICK, R.E. 1997. Mechanisms for generating overpressure in sedimentary basins: A reevaluation. American Association of Petroleum Geologists Bulletin, 81,1023-1041. REVIL, A. 2002. Genesis of mud volcanoes in sedimentary basins: a solitary wave-based mechanism. Geophysical research letters, 29,12,1-4. SANDAL, S.T. 1996. The Geology and Hydrocarbon Resources of Negara Brunei Darussalem, (1996 revision). Brunei Shell Petroleum Company/Brunei Museum, Syabas Bandar Sen Begawan, Brunei Darussalem. SIBSON, R.H. 1990. Conditions for fault-valve behaviour. In: KNIFE R.J. & RUTTER E.H. (eds) Deformation Mechanisms Rheology and Tectonics. Geological Society, London, Special Publications, 54,15-28. SUMNER, R.H. & WESTBROOK, G.K. 2001. Mud diapirism in front of the Barbados accretionaiy wedge: the influence of fracture zones and North America-South America plate motions. Marine and Petroleum Geology 18,591-613. TINGAY, M.R.P., HILLIS, R.R., MORLEY, C.K., SWARBRICK, R.E., & OKPERE, E.C. 2003. Pore pressure-stress coupling in the Baram basin, Brunei Darussalam - implications for shale mobilization. In: VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, AJ. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216,369-379. TISSOT, B.P. & WELTE, D.H. 1984. Petroleum formation and occurrence. New York, Springer-Verlag. VAN RENSBERGEN P. & MORLEY C.K. 2000. 3D seismic study of a shale expulsion syncline at the base of the Champion delta, offshore Brunei and its implications for the early structural evolution of large delta systems. Marine and Petroleum Geology, 17, 861-872. VAN RENSBERGEN, P., MORLEY, C.K., ANG, D.W., HOAN, T.Q. & LAM, N.T. 1999. Structural evolution of shale diapirs from reactive rise to mud volcanism: 3D seismic data from the Baram delta, offshore Brunei Darussalam. Journal of the Geological Society of London, 156, p. 633-650. VENDEVILLE, B.C. & JACKSON, M.P.A. 19920. The rise of diapirs during thin skinned extension. Marine and Petroleum Geology, 9, 331-353. VENDEVILLE, B.C. & JACKSON, M.P.A. 19926. The fall of diapirs during thin skinned extension. Marine and Petroleum Geology, 9, 354-371. WAITERS, D.G., MASKALL, R.C., WARRILOW, I.M., & Lmw, V. 1999. A sleeping giant awakened: further development of the Seria Field, Brunei Darussalam, after almost 70 years of production. Petroleum Geoscience, 5,147-159.
RE-EVALUATION OF MOBILE SHALE OFFSHORE BRUNEI WESTBROOK, O.K. & SMITH, MJ. 1983. Long decollements and mud volcanoes: Evidence from the Barbados Ridge Complex for the role of high pore-fluid pressure in the development of an accretionary complex. Geology, 11,279-283.
409
YILMAZ, 6.2001. Seismic data analysis: processing, inversion and interpretation of seismic data. Investigations in geophysics no 10, Society of Exploration Geophysicists, Tulsa, 2027 pp.
This page intentionally left blank
Analogue models of delta systems above ductile substrates KEN MCCLAY, TIM DOOLEY & GONZALO ZAMORA Fault Dynamics Research Group, Department of Geology, Royal Holloway University of London, Egham, Surrey TW20 OEX, UK Abstract: Delta systems developed above ductile substrates such as overpressured shales and salt have been modelled using layered sand-packs above ductile silicone polymer layers. Gravity spreading of progradational sedimentary wedges produces delta-top and upper delta-slope grabens linked to delta-toe contractional fold-thrust and diapir zones. The delta-top grabens are bound by both regional and counter-regional listric growth faults. A basinward-stepping sequence of regional, counter-regional followed by regional faulting is commonly developed. Polymer pillows and ridges commonly develop in the footwalls of the major listric extensional faults and may evolve into reactive diapirs. Successive progradational loads generate new delta-top or upper delta-slope graben systems on top of older contractional belts where the ductile polymer layer has been thickened significantly. The analogue model results in cross-section show many similarities to examples of natural deltas and differential sedimentary load systems such as offshore Angola, the Niger and Nile Deltas, Kutai Basin, Kalimantan, the Baram delta, Brunei and the Orinoco delta, Columbus basin and offshore Trinidad.
Introduction Large delta systems are formed by thick progradational clastic wedges that commonly overlie ductile, overpressured pro-delta shales. Classic delta systems include the Niger Delta (Cohen & McClay 1996; Doust & Omatsola 1990; Knox & Omatsola 1989), the Nile delta (Sestini 1989), the Baram delta, Brunei (James 1984; Koopman 1996; Schreurs 1997), the Mahakam Delta, Kalimantan (Ferguson & McClay, 1997; McClay et al 2000)and the Mississippi delta, Gulf of Mexico (Lopez 1990; Worrall & Snelson 1989). On some passive margins deltas are developed on top of ductile post-rift salt deposits e.g. Angolan margin (Anderson et al. 2000; Cramez & Jackson 2000; Marton et al 2000). Delta systems are usually characterized by regional (basinward dipping) and counter-regional (landward dipping) extensional growth faults that mainly form on the delta top (shelf) or upper delta slope together with a belt of imbricate thrusts and folds at the delta toe (Fig. 1). The Mississippi delta of the Gulf of Mexico displays well-documented growth faults both onshore and offshore as well as a deep-water contractional fold belt - the Perdido fold belt (Lopez 1990; Worrall & Snelson 1989; Rowan 1997). It is widely accepted that delta-top growth fault systems form as a result of differential loading of a ductile substrate by the progradational delta wedge and that this extension is mechanically linked to contraction at the delta toe (e.g. Blickwede & Queffelec 1988; Wu et al 1990; Morley & Guerin 1996; Rowan 1995,1997). The differential load of the delta wedge is thought to induce flow of the ductile, overpressured basal shales or salt toward the delta toe, the shales or salt becoming thickened in the contrac-
tional fold-thrust belt (e.g. Bruce 1973; Ge et al 1997; Cohen & Hardy 1996; Morley & Guerin 1996; Rowan 1995, 1997). Although the overall geometries of delta systems are broadly understood, seismic data across delta systems usually only clearly images the upper, shallow sections. Deeper, older structures are usually obscured, thus masking much of the detail that could reveal timing and structural geometries (e.g. Cohen & McClay 1996; Koopman 1996; Whiteman 1982). Numerical modelling by Cohen & Hardy (1996) showed how the thickness and viscosity of the basal ductile layer control the rate of flow and size of the contractional bulge at the front of a delta wedge. Numerical modelling, however, cannot simulate the development of the delta-top graben fault systems nor the frontal imbricated fold-thrust belt. This paper summarizes the results of a series of dynamically scaled analogue models designed to simulate the differential loading of a ductile substrate by a progradational delta wedge. These experiments have the advantage of being monitored through time-lapse photography thus showing the progressive evolution of the model delta and the reaction of the mobile layer to variations in progradation and aggradation. Instantaneous single and multi-load models were constructed and compared with models constructed with an initial instantaneous load followed by incremental progradation and aggradation during the model run. Models run with a basal slope generated raft structures in the upper delta whereas those without basal slopes successfully simulated the formation of regional and counter-regional growth faults on the delta top together with a contractional fold-thrust belt at the delta toe (cf. McClay et al 1998). Previous analogue
Fig. 1. Geoseismic sections through natural delta systems, (a) Section through the Baram delta, offshore Brunei (modified after Koopman et al 1996); (b) section through the Niger delta (after Cohen & McClay 1996).
modelling experiments of delta tectonics (e.g. Ge et al 1997; Szatmari et al 1996) failed to reproduce shelf growth faulting. Recent unconstrained gravity gliding model studies by Mauduit etal (1997) documented the effects of slope angle and syn-kinematic sedimentation rates on progressive extension due to gravity gliding but did not reproduce downslope contraction regimes or delta geometries. In many natural delta regimes the presence of delta toe contractional structures indicates a downslope constraint with the delta top extension, in part, compensated by a delta toe contractional belt. In the Gulf of Mexico and Campos Basin, Brazil the down-slope constraint is probably a result of thinning of the ductile salt detachment layer. In deltas developed on top of overpressured shales (e.g. Niger Delta; Mahakam Delta, Kalimantan; and Baram Delta, Brunei), the downslope constraint is most likely the basin-ward limit of overpressured shales (e.g. Ferguson & McClay 1997; McClay etal 2000).
Analogue Modelling Methodology In this paper four series of differential load experiments are presented. These models were constructed in a box 100 cm long and 60 cm wide (Fig. 2). A 10 mm thick basal layer of silicone polymer (SGM 36, Weijermars et al 1993) was used to simulate a ductile sequence of overpressured pro-delta shales. This was overlain by a 4-10 mm thick pre-kinematic layer of coloured, 190 jmm grain size, dry quartz sand (internal angle of friction 30°, McClay, 1990) in
order to simulate brittle, pre-differential loading, pro-delta sediments. On top of the pre-kinematic layer a 60 cm wide, 3 cm thick wedge of alternating, 2-3 mm thick coloured sand layers was mechanically sieved into one end of the deformation box (Fig. 2). This was done over a period of one hour in order to simulate the rapid progradation of a wedge of delta sediments over a ductile substrate. The models deform as a result of this differential loading by the sand wedge over periods of as much as 19 hours. During deformation extensional grabens that formed on the wedge top were sequentially infilled to simulate syn-kinematic sedimentation. Two-stage loading experiments consisted of a first phase of differential delta loading with deformation occurring over eight hours, followed by a second phase of rapid progradation of the delta-wedge sedimentation that covered the entire first phase system including the fold-thrust belt that formed at the delta-toe. Models run with incremental progradation/aggradation had an initial configuration similar to singleload models followed by infilling of extensional structures and a pre-determined progradationaggradation ratio. Photographs were taken of the top surface of the models and at the end of the experiments the completed models were impregnated with a gelling agent and serially sectioned at 2 cm intervals. Scaling and sedimentation parameters The modelling procedures used in the experiments described in this paper assume an almost instantaneous differential delta loading by rapid progradation of the wedge over the ductile substrate. Aggradation
ANALOGUE MODELS OF DELTA SYSTEMS ABOVE DUCTILE SUBSTRATES
413
Fig. 2. Schematic diagram showing a cross-sectional view of the experimental set up. Silicone polymer forms the ductile basal layer whereas white and coloured sand layers form the pre-kinematic layers and differential wedge layers 1 and 2.
delta shelf is characterised by a segmented or 'rafted' graben system countered by a major deltatoe fold-thrust belt breached by the polymer to form an emergent canopy (Figs 4b and c). Cross-section views through the sidewalls of the apparatus show the progressive evolution of the model (Figs 4d-f). These show highly extended, partially grounded rafts separated by major reactive and active diapir systems (Figs 4d-f). In the early evolution of the model a polymer bulge (shale ridge) developed at the delta toe at the position of the change in slope of Analogue model results the basement (Fig. 4d). With increased differential The results of four representative models are pre- loading extension increased within the delta-top sented in detail. They summarize the results of the graben systems with 'rollers' of ductile polymer major delta modelling programmes carried out at forming in the footwalls to the main regional growth Royal Holloway (e.g. Lewis 1997; McClay et al. faults (Fig. 4e). A contractional fold and thrust belt 1998; McClay et al 2000) and present new data formed at the delta toe. Continued evolution of the system resulted in reactive diapirism and separation summarized from Zamora (1999). of 'rafts' with the down-dip translation of the delta slope and outboard migration of the delta toe system (Fig.4f). Model one - raft model - Figure 4
rates from the Gulf of Mexico between 2-5 km/m.y. (Lopez 1990) indicate that it is reasonable in the models to add 2-5 cm in 1 hour given a thickness scaling factor of 10~5 (Fig. 3; McClay 1990) and a time scaling factor of 10~9 to 10~10 (e.g. Vendeville & Cobbold 1987; Childs et al 1993). All of the experiments shown in this paper have been repeated at least three times with similar results each time.
Model one consisted of an initial mobile layer and pre-load sand-pack that was placed on a 2° slope that changed to 0° on the right-hand side of the model. The differential load was added progressively and the model deformed for a total of 19 hours. The right-hand margin was unconstrained (Fig. 4). After an initial 1.6 cm differential load the hinged rig was placed on a 2° slope, dipping to the right in order to induce down-slope flow. Initial deformation above the polymer layer consisted of a linear delta-top graben system (Fig. 4a). With increasing load this delta top graben system developed in space and time to involve most of the delta shelf and the shelf-slope break was delineated by a major counter regional system (Fig. 4b). At the end of the experiment the
Model two - single differential load model Figure 5 Model two consisted of a single instantaneous differential load placed on top of a uniform pre-load sand-pack above the basal ductile layer. The model then deformed under the differential load and after eight hours the structure consisted of a delta-top graben system and a delta-toe fold-thrust belt (Fig. 5a). The delta-top graben system was formed by several segmented grabens bound by both regional and counter-regional growth faults. After 18 hours the final structure was composed of a wide delta-top graben system infilled with synkinematic sediments
414
K.MCCLAYETAZ,
Fig. 3. (a) Upper crustal strength profiles for sedimentary rocks, overpressured shales and salt (after Jackson & Vendeville 1994). (b) Scaling parameters for analogue models in this study.
together with a frontal fold belt and diapir system at the foot of the delta toe and into the pro-delta region (Figs 5b and c). Extension on the top of the delta wedge was accommodated by thinning of the ductile basal polymer layer, which flowed toward the front of the wedge to form a synchronous delta toe bulge and fold-thrust belt (Figs 5b and c). Serial crosssections through the final model show three delta-
top graben systems (1-3, Fig. 5d). These are bounded by listric regional and counter-regional growth faults that sole into the ductile polymer layer at the base of the model (Fig. 5d). Each half-graben is bound by a listric growth fault and a planar antithetic fault (Fig. 5d). Steeply-dipping, highlyrotated syn-kinematic layers occur in half-graben three. Half-graben one was the first to form, initially
ANALOGUE MODELS OF DELTA SYSTEMS ABOVE DUCTILE SUBSTRATES
with a regional growth fault, which, with increased extension, was replaced by a counter-regional growth fault (Fig. 5d). Upon further extension, halfgraben two developed in a similar fashion, first controlled by a regional growth fault and then by a counter-regional fault (Fig. 5d). A frontal polymer bulge formed at the toe of the progradational load together with a fold-thrust belt of tight, short wavelength thrust anticlines with thickened polymer in the cores (Fig. 5d). The greatest thickening of the ductile polymer is found at the edge of the progradational load, where a piercement diapir formed (Fig. 5d). Model three - two-stage progradational load model - Figure 6 Model three consisted initially of an identical set-up to that of Model two (Fig. 5). After the addition of the first differential load the model was allowed to evolve for six hours with no further additional loading. A complex, segmented delta-top graben system (Fig. 6a; similar to that in Model 2, Fig: 5) formed and was progressively infilled with synkinematic sediments. The delta-top graben was progressively infilled as the model evolved (Fig. 6a). A frontal bulge together with a wide frontal fold-thrust belt formed at the toe of the delta slope (Fig. 6a). At this stage the structure was very similar to that at the equivalent time in the single load models (Fig. 5a). The delta-top graben system was formed by two half-grabens bound by planar regional and listric counter-regional growth faults that soled out into the ductile basal polymer layer. As in Model two, the left-hand half-graben was the first to form, followed by the right-hand half-graben. A second differential progradational load was added after eight hours. This completely covered the first stage graben system as well as the frontal bulge and fold-thrust belt (Fig. 6b). The model then continued to deform under this second differential load for a further 10 hours (total deformation time 18 hours). A second-phase delta-top graben system formed directly over the old buried fold-thrust belt and a second-phase fold-thrust belt formed at the front of the new delta-slope (Fig. 6b). The buried, first-phase graben system was not reactivated and all of the delta-top extension was taken up by the new graben first-phase delta-toe fold-thrust belt. It consisted of two narrow graben above phase one fold anticlines, the cores of which were formed by thickened polymer (Figs 6b and c). In cross-section the final model shows the buried graben systems of the first phase of progradational loading (Stage A) and the second phase graben systems developed above the buried fold-thrust belt of the first stage of differential loading (Fig. 6d). The second phase fold-thrust belt
415
occurs at the foot of the second phase differential load (Fig. 6d). Model four - variable differential load delta model - Figure 7 Model four had a similar initial configuration as Models two and three but differed in that the differential load was progressively added from the lefthand side of the model. After an initial differential load and three hours of deformation a narrow deltatop graben system had formed near the shelf-slope break together with a segmented fold-thrust belt at the delta-toe (Fig. 7a). From this stage, over the next six hours, additional differential load was added with a 5:1 progradation/aggradation ratio (Fig. 7c). After nine hours the model showed that extension of the first graben (Fig. 7a) had continued with the development of a major counter-regional fault system (Fig. 7b). A major new graben system developed just outboard of the new shelf slope break, directly above the former delta-toe fold belt (Figs. 7a,b). In addition, smaller, new grabens developed both further downslope and also inboard of the shelfslope break (Fig. 7b). A wide, complex delta toe and pro-delta fold belt developed with thickened polymer in the anticline cores. Differential loading was continued for another three hours (to a total of 12.5 hours) (Fig. 7c). At this stage the outer shelf and upper slope portion of the model consisted of a wide, complex graben system with both regional and counter-regional fault systems (Figs. 7c and d). Earlier shelf-slope breaks (labelled 1 and 2 in Fig. 7d) corresponding to the stages shown in Figures 7 a and b, together with older graben systems were buried by the later progradation (Fig. 7d). Serial sections through the completed model revealed only limited along-strike changes in the structure and Figure 7e was chosen as a representative cross-section. Three well developed graben systems are shown in cross-section. System one graben formed early in the evolution of the model and was controlled by a major counter-regional fault system but became inactive as the delta was prograded out across the initial delta slope, thus moving the shelf-slope break further outboard rendering this graben inactive (Fig. 7e). Grabens two and three are dominated by two, major regional, listric growth fault systems that formed near the shelf-slope break at nine and 12.5 hours respectively (Fig. 7e). Both regional and counter-regional listric growth faults display large amounts of syn-kinematic sediments across them as well as large rotations of both prekinematic and syn-kinematic layers. The frontal, delta-toe, fold/thrust belt is characterized by detachment folds with thickened polymer cores as well as polymer ridges (Fig. 7e).
416
K.McCLAY#rAL.
ANALOGUE MODELS OF DELTA SYSTEMS ABOVE DUCTILE SUBSTRATES
417
Fig. 4 left and above. Model one. Raft model with partially tilted basement, (a) Photograph of the upper surface after 7.5 hours. Illumination is from the right. Light bands are extensional faults dipping to the right (regional faults) and dark bands are extensional faults dipping to the left (counter-regional faults). Two delta-top graben systems are developed, (b) Photograph of the upper surface after 19 hours. Illumination is from the right. Three delta-top graben systems are developed with emergent polymer diapirs in their centres. An emergent polymer diapir at the delta-toe has spread into a canopy system, (c) Line diagram interpretation of (b) showing the three delta-top and slope graben systems as well as emergent diapirs and the polymer canopy system, (d) Line diagram interpretation of model after 1.5 hours and a maximum differential load of 2.4 cm. Two delta-top graben systems are developed, (e) Line diagram interpretation of model after 3.0 hours and a maximum differential load of 3.0 cm. (f) Line diagram interpretation of model after 19.0 hours and a maximum differential load of 5.5 cm.
Discussion
tion at the delta-toe. Detachment folds and thrust faults together with polymer pillows and diapirs form at the base of the active slope (Figs 5-7). Underneath the extensional grabens on the delta-top, polymer Analogue models withdrawal occurs, leading to thinning of the mobile The analogue models illustrated in this paper graphi- layer and the pre-kinematic layer sand-pack (e.g. cally demonstrate the formation of delta-top graben Figs 5d, 6d and 7e). Where the polymer layer systems by ductile flow of a basal polymer under the becomes sufficiently thinned extension directly influence of a differential sedimentary load (Figs above this section ceases and is transferred basin4-7). The delta-top graben were characterized by ward in the model. Complete 'touchdown' by both regional and counter-regional listric growth polymer withdrawal is not necessary because when faults paired together with more planar antithetic the polymer layer becomes very thin, boundary layer faults (cf. Figs 5d and 7e). Within these half-graben shear forces prevent further migration of the ductile the pre-kinematic layers and syn-kinematic layers layer beneath the graben (cf. Waltham 1997). The models illustrate the synchronous developcan be strongly rotated as observed in many growth fault systems in delta terranes (e.g. Whiteman 1981; ment of delta-top graben systems and delta-toe comRowan 1995). In our models, because they were con- pressional structures (cf. McClay et al. 1998). The strained by an end wall, a polymer bulge with a ductile substrate clearly exhibits thinning beneath fold-thrust belt developed in front of the differential the delta-top graben systems and flows outboard load within the pro-delta section (Figs 4-7). This (basinward) towards the delta toe where it accumucontractional zone developed in all models including lates either as a frontal bulge or a series of detachthe raft models on a basal slope (Fig. 4). These results ment folds (cf. Cohen & Hardy 1996). Purely contrast with those of Maduit et al. (1997) whose progradation and mixed mode sedimentation results models were unconstrained downslope. Extension on in the migration of activity outboard towards the the delta-top is clearly dynamically linked to contrac- migrating shelf-slope break - the zone of maximum
418
K.McCLAYETAL
ANALOGUE MODELS OF DELTA SYSTEMS ABOVE DUCTILE SUBSTRATES
419
Fig. 5 left and above. Model two. Single differential load,(a) Photograph of the upper surface after 8 hours. The differential load was prograded from the left-hand side of the model. Illumination is from the right. Light bands are extensional faults dipping to the right (regional faults) and dark bands are extensional faults dipping to the left (counterregional faults). A single, wide, delta-top graben system is developed together with a fold belt at the delta-toe and in the pro-delta section, (b) Photograph of the upper surface after 18 hours. Illumination is from the right. A wide, delta-top graben system is developed with left-dipping, counter-regional faults dominant. The fold belt at the delta-toe / slope has an emergent polymer diapir at the foot of the delta-toe, (c) Line diagram interpretation of (b) showing the three deltatop and slope graben systems as well as emergent diapirs and the polymer canopy system, (d) Serial section through the centre of the completed model (above) and line diagram interpretation (below). Three delta-top - upper delta slope grabens are developed together with a fold thrust belt and diapir zone at the delta toe and into the pro-delta section.
differential load. Pre-existing depocentres demonstrate waning activity and eventual burial due to the combined effects of thinning of the basal mobile zone and the outboard migration of the shelf-slope during progradation and mixed mode phases (Figs 6 & 7). Rapid development of new graben systems above former delta-toe sites occurs due to the already over thickened ductile substrate. Models with instantaneous progradation above pre-existing delta-toe sites illustrate rapid development of a new graben system directly above the former toe-bulge (Figs 5 & 6). The development of counter-regional systems in single and double load models postdates regional fault formation and is closely linked to severe thinning of the ductile layer, beneath the regional fault system, as it flows outboard in response to the imposed differential load (cf. McClay et al 1998). McClay et al (1998) showed that complete grounding of the polymer is not necessary to generate counter-regional fault systems as the severe thinning beneath major delta-top graben systems effectively cuts off polymer supply from beneath developing graben. Thin-film forces prevent the total evacuation of the ductile layer from beneath the graben (cf. Waltham 1997). Models run with either multiple differential loads or with progressive progradation typically produce new delta-top or upper delta-slope graben systems on top of older contractional belts where the thickness of the ductile layer was significantly increased (Figs 6d and 7e). The basal pre-kinematic and early syn-kinematic sequences in these models were first folded and thrust in the contractional regime and then later extended during the development of the
grabens above. Therefore complex geometries and highly rotated layers are characteristic of these sequences (e.g. Fig. 7e). In models where there was significant aggradation as well as progradation, stable, long-lived delta top grabens developed with the regional listric growth faults concentrated at successive shelf-slope breaks (Fig. 7e). In particular, the serial cross-sections as well as the time lapse photography of the upper surfaces of the models show a sequence of delta-top and upper delta-slope growth faulting that evolves from regional (down to the basin) faulting that is superseded by large counter-regional faulting which is, in turn, cut by the next set of regional growth faults as the shelf-slope break steps outboard (Fig. 7). The dynamics of a delta system that builds outboard on top of an overpressured pro-delta shale sequence by both aggradation and progradation may be considered in terms of a tapered tectonic wedge model (cf. Elliot 1976; Chappie 1978; Davis et al 1983; Platt 1986). In particular Platt (1986) applied the critical Coulomb wedge model (Davis et al. 1983) to the exhumation and uplift of erogenic wedges that were largely in compression but also showed that, in thick wedges with high surface slopes, extension occurred in the hinterland of the wedges. Such a model may also be applied to an aggradational/progradational delta system developed on a very ductile substrate. The progressively increasing thickness of the delta top section together with a high delta slope angle and the low friction basal decollement unit produces delta top extension (lowers the wedge height) that is balanced by delta toe contraction (lowers the wedge taper).
420
K.McCLAYCTAL.
ANALOGUE MODELS OF DELTA SYSTEMS ABOVE DUCTILE SUBSTRATES
421
Fig. 6 left and above. Model three. Two differential loads, (a) Photograph of the upper surface after 5 hours. The differential load was prograded from the left-hand side of the model. Illumination is from the right. Light bands are extensional faults dipping to the right (regional faults) and dark bands are extensional faults dipping to the left (counterregional faults). A single, wide, delta-top graben system is developed together with a fold belt at the delta-toe and prodelta section, (b) Photograph of the upper surface after 8.5 hours after deposition of the second differential load. Illumination is from the right. A second delta-top graben system is developed on top of the first phase fold-thrust belt. The graben system is strongly segmented with both regional and counter-regional faults developed. A second phase fold belt has formed at the foot of the delta-toe, (c) Line diagram interpretation of (b) showing the second phase delta-top graben system as well as the second phase fold-thrust belt, (d) Serial section through the centre of the completed model (above) and line diagram interpretation (below). The two first phase delta-top graben are shown together with the second phase delta-top graben system formed above the first phase fold-thrust belt. The second phase fold-thrust belt is formed further outboard at the second phase delta toe and pro-delta region.
Such a wedge model is shown schematically in Figure 8 where the progressive evolution of a progradational/aggradational delta wedge is shown. The initial loading conditions (Fig. 8a) produce a thick, unstable wedge that undergoes extension with increased delta build up (Fig. 8b). At this stage the first 'down to the basin' regional growth fault forms together with a contractional fold-thrust belt at the delta toe (base of slope fold-thrust belt). The foldthrust belts may consist of contractional folds and thrusts as well as contractional diapirs of overpressured shales. With increased delta growth and sediment loading, delta-top extension continues with the formation of a 'counter-regional' growth fault as well as continued displacement on the regional growth fault system (Fig. 8c). The base of slope fold belt continues to develop and migrates in front of the prograding delta slope sediments. Continued delta aggradation and thinning of the ductile decollement layer beneath the extensional growth fault systems leads to effective grounding out of the ductile unit inhibiting further extension leading to the development of a second regional growth fault system (Fig. 8d). A second 'counterregional' growth fault system (Fig. 8e) is developed, cutting the earlier formed regional growth fault system. As the delta builds outboard, delta-top is extension is focused near the shelf-slope break and is balanced by the progressive development of the base of slope (delta-toe) fold-thrust belt. In this manner, successive extensional growth fault systems are formed as the delta aggradation/progradation proceeds with the older systems becoming inactive and buried (e.g. Figs 6 & 7).
This pattern of first-formed regional growth faults that are cut by later 'counter-regional' growth faults has been observed in other analogue models (cf. Maduit et al. 1997) and in many natural delta systems (Van Rensbergen & Morley 2000; Wood 2000). In many natural delta systems 'grounding' of the sedimentary section is not observed, nor is it expected, due to great thickness of the mobile shales. However, dewatering of these shales (diapir formation, etc.; see Morley 2003), substantial thinning beneath major grabens and half-grabens and a decrease in sediment supply may result in a waning of activity on the listric growth fault systems without the need for grounding. This may result in the infilling of these grabens and half grabens, thus reducing their effectiveness as sediment barriers and further pulses of sedimentation may bypass the existing graben systems and prograde outboard across the toe fore-bulge (cf. Morley 2003). Comparison "with natural examples 2D cross-sectional views of the scaled analogue models described in this paper can be compared with previously published examples of differential load tectonics such as offshore Angola (Fig. 9a; Marton et al 2000), the Niger Delta (Fig. 9b; Haack et al 2000) and the Columbus Basin (Orinoco delta, Fig. 9c; Wood 2000), as well as other delta systems such as the Baram Delta, Brunei (Fig. la, Koopman et al 1996; Van Rensbergen & Morley 2000) and the Mahakam Delta (Ferguson & McClay 1997; McClay eta/. 2000).
422
Fig. 7.
K.McCLAYCTAL.
ANALOGUE MODELS OF DELTA SYSTEMS ABOVE DUCTILE SUBSTRATES
Fig. 7 continued.
423
424
K.McCLAY ETAL
Fig. 7 continued. Model four. Variable differential load, (a) Photograph of the upper surface after 1.5 hours. The initial differential load was added from the left-hand side of the model. Illumination is from the right. Light bands are extensional faults dipping to the right (regional faults) and dark bands are extensional faults dipping to the left (counterregional faults). A single delta-top graben system is developed together with a fold belt at the delta-toe and pro-delta section, (b) Photograph of the upper surface after 9 hours after continued progradation of the differential load. Illumination is from the right. The delta-top graben system has widened and slope graben system has developed on top of the first formed fold-thrust belt. The fold belt at the delta-toe / slope has widened with complex, along-strike 3D fold geometries, (c) Photograph of the upper surface after 12.5 hours after continued progradation of the differential load. Illumination is from the right. The delta-top graben system has widened and slope graben system has developed on top of the first formed fold-thrust belt. The fold belt at the delta-toe / slope has widened with complex, along-strike 3D fold geometries, (d) Line diagram interpretation of c) showing the later-formed delta top - slope break graben system deltatop and slope graben system with dominant counter-regional faults together with later-formed slope-break graben systems and the fold-thrust belt at the delta toe. (e) Serial section through the centre of the completed model (above) and line diagram interpretation (below). These show the early-formed delta-top and slope graben system with dominant counter-regional faults together with later-formed slope-break graben systems dominated by regional fault systems.
Figure 9a shows a geoseismic section across offshore Angola where Cretaceous - Tertiary differential sediment loading from the Congo fan has generated down-slope gravity spreading of the postrift - drift section (Marton et al 2000). In the onshore and near shore parts (eastern) of the section regional listric growth faults sole out into highly thinned Aptian salt whereas in the central and western parts of the cross-section the salt is inflated and forms pillows and diapirs in the cores of contractional anticlines (Fig. 9a). These features are very similar in general morphology and form to the serial sections of the raft model depicted in Figure 4d. Cross-sections though progradational deltas above ductile and overpressured shales such as the Baram, Nile and Orinoco deltas (Figs 1, 9b and c) characteristically show up-dip (i.e. landward) listric growth faults with significant expansion of syn-kinematic sediments within individual half-grabens. Successive and younger basinward regional growth faults cut older counter-regional growth faults (Figs 9b and c). Similar geometries and fault sequences are found in many of the analogue models particularly where the ductile silicone polymer has thinned significantly such that extension is forced to migrate outboard and downslope of the active delta-top/ delta-slope graben system (Fig. 7e). Contractional delta-toe structures such as those found in the Baram
and Niger deltas (Figs 1 and 9b) are cored by mobilized ductile shales and are typically either detachment folds or thrust fault-propagation folds. Shale diapirs, pillows and ridges are characteristic of this part of the delta complex where the shale section is clearly thickened. Very similar geometries are found in all of the analogue models where the frontal sections are formed by detachment folds and thrust cored anticlines (Figs 4-7). The geoseismic crosssections of delta systems shown in Figures 1 a and 9 do not illustrate detailed structure at depth largely due to poor seismic resolution in the overpressured sequences. Cohen & McClay (1996) however discussed evidence that some of the extensional depobelts in the Niger Delta (Fig. Ib) were probably developed above older contractional fold-thrust belts as seen in the multi-load analogue models (Figs 6d and 7e). Cross-sectional views through the analogue models (Figs 4d, 5d, 6d and 7e) show a striking resemblance to many of the fault and fold architectures seen in natural delta systems and where differential sedimentary loading has induced downslope gravity spreading (e.g. Schultz-Ela 2000) together with contaction at the toe of the systems. The analogue models as yet however, do not incorporate variable pore-fluid pressures, ductile layer thicknesses and strain hardening phenomena (as one would
ANALOGUE MODELS OF DELTA SYSTEMS ABOVE DUCTILE SUBSTRATES
425
Fig. 8. Conceptual model for the development of successive counter-regional and regional faults in the analogue models, (a) Initial load deposited across the model, (b) Formation of first regional fault and development of delta-toe fold/thrust belt at the base of the slope, (c)As mobile layer thins beneath the tip of the first regional fault a counterregional fault forms marking the transition from shelf to slope, (d) Outboard migration of the shelf-slope break, infilling delta-top grabens generates new regional fault that cuts first counter-regional, (e) Continued delta growth results in outboard migration of the basal mobile layer from beneath the delta-top grabens inhibiting further extension and generating a second counter-regional fault system.
426
K.McCLAY£rAL.
Fig. 9. Comparative cross-sections through natural differential load systems, (a) Cross-section through offshore Angola showing raft tectonics, large listric growth faults and am highly thinned salt decollement layer at the eastern end of the section. A thickened salt section with diapirs and salt pillows has formed at the contractional toe region at the down-dip western end of the section (modified after Marton et al 2000). (b) Cross-section though the Niger delta. Extensional growth fault-bounded depobelts with both regional and counter-regional faults in the northern part of the section together with a down-dip contractional fold-thrust belt in the deep water offshore at the southern end of the section (modified after Haack et al 2000). (c) Regional cross-section through part of the Colombus Basin, offshore Trinidad. This section shows part of the delta top and slope region dominated by large listric regional growth faults that cut earlier counter-regional fault systems (modified after Zamora 1999).
expect to occur in overpressured shale units) within the ductile decollement layer. The models also do not account for crustal scale flexure resulting from the differential sedimentary loading as seen in the cross-section of the Niger Delta (Fig. 9b) - this will affect the dip of the basal decollement layer and hence the distribution of the gravity spreading in the system. Nevertheless, despite these limitations, the physical models do successfully illustrate the geometric and kinematic evolution of many of the features found in natural delta systems developed on ductile substrates such as overpressured shales and salt.
Conclusions The scaled analogue models presented in this paper show the progressive evolution of raft tectonics and also delta systems by differential sedimentary
loading above ductile substrates. Gravity spreading of progradational sedimentary wedges produces delta-top and upper delta-slope grabens linked to delta-toe contractional fold-thrust and diapir zones. The delta-top grabens are bound by both regional and counter-regional listric growth faults. A basinwardstepping sequence of regional, counter-regional followed by regional faulting is commonly developed in association with delta progradation and mobile layer migration. Polymer pillows and ridges commonly develop in the footwalls of the major listric extensional faults and may evolve into reactive diapirs. Successive progradational loads generate new deltatop or upper delta-slope graben systems on top of older contractional belts where the ductile polymer layer has been thickened significantly. The analogue model results in cross-section show many similarities to examples of natural deltas and differential sedimentary load systems such as offshore Angola, the Niger and Nile deltas, the Baram
ANALOGUE MODELS OF DELTA SYSTEMS ABOVE DUCTILE SUBSTRATES
delta, Brunei, the Mahakam Delta, Kalimantan and the Orinoco Delta and the Colombus basin, offshore Trinidad. Research presented in this paper was funded by the Fault Dynamics Research Group, Royal Holloway University of London. BG Group provided assistance with the modelling of the Colombus basin structures and K. McClay thanks bp Exploration for generous financial support. Initial modelling of delta structures was carried out during a research project funded by VICO, Indonesia. B. Adams and M. Craker are thanked for building and maintaining the analogue modelling apparatus. Fault Dynamics Research Group publication No. 120.
References ANDERSON, I.E., CARTWRIGHT, J., DRYSDALL, S.J. & VIVIAN, N. 2000. Controls of turbidite deposition during gravity-driven extension of a passive margin: examples from Miocene sediments in Block 4, Angola. Marine and Petroleum Geology, 17, 1165-1203. BLICKWEDE, JJ. & QUEFFELEC, T.A. 1988. Perdido foldbelt; A new deepwater frontier in the Gulf of Mexico [abs.]: Association of American Petroleum Geologists Bulletin, 72,163. BRUCE, C. 1973. Pressured shale and related sediment deformation: mechanism for development of regional contemporaneous faults: American Association of Petroleum Geologists Bulletin , 51, 878-886. CHAPPLE, W.M. 1978. Mechanics of thin-skinned fold-andthrust belts: Geological Society of America Bulletin, 89,1189-1198. COHEN, H. & HARDY, S. 1996. Numerical modelling of stratal architectures resulting from differential loading of mobile substrate. In: ALSOP, G.L, BLUNDELL, D.J. & DAVISON, I. (eds.), Salt Tectonics. Geological Society, London, Special Publications, 100,265-274. COHEN, H.A. & McCLAY, K.R. 1996. Sedimentation and shale tectonics of the northwestern Niger Delta front. Marine and Petroleum Geology, 13, 313-328. CHILDS, C., EASTON, S.J., VENDEVILLE, B.C., JACKSON, M.P.A. LIN, S.T., WALSH, JJ. AND WATERSON, J. 1993. Kinematic analysis of faults in a physical model of growth faulting above a viscous salt analogue. Tectonophysics, 228, 313-229. CRAMEZ, C. & JACKSON, M.P.A. 2000. Superposed deformation straddling the continental-oceanic transition in deep-water Angola. Marine and Petroleum Geology, 17,1095-1109. DAVIS, D., SUPPE, J. & DAHLEN, F.A. 1983. Mechanics of fold-and-thrust belts and accretionary wedges. Journal of Geophysical Research, 88,1153-1172. DOUST, H. & OMATSOLA, E. 1990. Niger Delta. In Divergent/ Passive margins. In: EDWARDS, J.D. & SANTOGROSSI, P.A. (eds) American Association of Petroleum Geologists Memoir. 48,201-238. ELLIOTT, D. 1976. The motion of thrust sheets. Journal of Geophysical Research, 81,949-963. FERGUSON, A. & McCLAY, K.R. 1997. Structural modeling within the Sanga Sanga PSC, Kutei Basin,
427
Kalimantan: Its application to paleochannel orientation studies and timing of hydrocarbon entrapment. In: HOWES, J.U.C. & NOBLE, R.A. (eds) Proceedings of an International Conference on Petroleum Systems of SE Asia and Australasia, Indonesian Petroleum Association.721-143. GE, H., JACKSON, M.P.A. & VENDEVILLE, B.C. 1997. Kinematics and dynamics of salt tectonics driven by progradation: American Association of Petroleum Geologists Bulletin, 81,398-423. HAACK, R.C., SUNDARARAMAN, P., DIEDJOMOHAR, J.O., XIAO, H., GANT, N.J., MAY, E.D. & KELSH, K. 2000. Niger Delta Petroleum Systems, Nigeria. In: MELLO, M.R. & KATZ, B.J. (eds) Petroleum Systems of South Atlantic Margins, American Association of Petroleum Geologists Memoir 73,213-233. JACKSON, M.P.A. & VENDEVILLE, B.C. 1994. Regional extension as a trigger for diapirism. Geological Society of America Bulletin, 106,57-73. JAMES, D.M.D. (ed.) 1984. The geology and hydrocarbon resources of Negara Brueni Dalussalam: Muzium Brunei and Brunei Shell Petroleum Company Berhad Special Publication, 8,2-21. KNOX, G. J. & OMATSOLA, E.M. 1989. Development of the Cenozoic Niger Delta in terms of the 'Escalator Regression' model and the impact on hydrocarbon distribution. In: VAN DER GRAFF, W.J.M., VANDENBERGHE, J. & VAN DER GUN, J.A.M. (eds) Proceedings KNGMG Symposiums Coastal Lowlands, Geology and Geotechnology: Amsterdam, Kluwer Academic Publishers, 181-202. KOOPMAN, A. 1996. Structure. In: SANDAL, S.T. (ed.) The geology and hydrocarbon resources of Negara Brunei Dalussalam (second edition): Muzium Brunei and Brunei - Shell Petroleum Company Berhad Special Publication, 64-80. LEWIS, G.J. 1997. The development of growth faults above ductile substrates. PhD Thesis. Royal Holloway (University of London) Unpublished. LOPEZ, J.A. 1990, Structural styles of growth faults in the U.S. Gulf Coast Basin. In: BROOKS, J. (ed.) Classic Petroleum Provinces. Geological Society, London, Special Publications, 50,203-219. MARTON, L.G., TARI, G.C. & LEHMANN, C.T. 2000. Evolution of the Angolan Passive Margin, West Africa, with Emphasis on Post-Salt Structural Styles. In: MOHRIAK, W. & TALWANI, M. (eds) Atlantic Rifts and Continental Margins. Geophysical Monograph 115, American Geophysical Union, 129-149. MAUDUITT, T., GUERIN, G., BRUN, J.-P. & LECANU, H. 1997. Raft tectonics: The effects of basal slope angle and sedimentation rate on progressive extension. Journal of Structural Geology, 19,1219-1230 McCLAY, K.R. 1990. Extensional fault systems in sedimentary basins. A review of analogue model studies. Marine and Petroleum Geology, 7, 206 - 233 McCLAY, K.R., DOOLEY, T. & LEWIS, G. 1998. Analog modelling of progradational delta systems. Geology, 26,9,771-774. McCLAY, K.R., DOOLEY, T., FERGUSON, A. & POBLET J., 2000. Tectonic evolution of the Sanga Sanga block, Mahakam delta, Kalimantan, Indonesia. Bulletin of the American Association of Petroleum Geologists. 84,6,765-786.
428
K.McCLAY£rAL
MORLEY, C.K. 2003. Mobile shale related deformation in large deltas developed on passive and active margins. In: VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, AJ. & MORLEY, C.K. Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216, 335-357. MORLEY, C.K. & GUERIN, G. 1996. Comparison of gravitydriven deformation styles and behaviour associated with mobile shales and salt. Tectonics, 15, 1154-1170. PLATT, J.P. 1986. Dynamics of orogenic wedges and the uplift of high-pressure metamorphic rocks. Geological Society of America Bulletin, 97, 1037-1053. ROWAN, M.G. 1995. Structural styles and evolution of allochthonous salt, central Louisiana outer shelf and upper slope. In: JACKSON, M. P. A., ROBERTS, D.G. & SNELSON, S. (eds) Salt tectonics, a global perspective. American Association of Petroleum Geologists Memoir 65,199-228. ROWAN, M.G. 1997. Three-dimensional geometry and evolution of a segmented detachment fold, Mississippi Fan foldbelt, Gulf of Mexico. Journal of Structural Geology, 19,663-680. SCHREURS, G. 1997. The petroleum geology of Negara Brunei Darussalem; an update. In: HOWES, J.V.C. & NOBLE, R.A. (eds) Proceedings of the IPA Petroleum Systems of SE Asia and Australasia Conference, Jakarta, Indonesia, May, 1997, Indonesian Petroleum Association, Jakarta, 751-766. SCHULT-ELA, D.D. 2001. Excursis on gravity gliding and gravity spreading. Journal of Structural Geology, 23, 725-731. SESTINI, G. 1989. Nile Delta: a Review of Depositional Environments and Geological History. In: WHATELY, M.K.G. & PICKERING, K.T. (eds) Deltas: Sites and Traps for Fossil Fuels. Geological Society, London, Special Publications 41,99-127. SZATMARI, P., GUERRA, M.C.M. & PEQUENO, M.A. 1996.
Genesis of large counter-regional normal fault by flow
of Cretaceous salt in the south Atlantic Santos Basin Brazil. In: ALSOP, G.I., BLUNDELL, D.J. & DAVISON, I., (eds) Salt Tectonics. Geological Society, London, Special Publications, 100,259-264. VAN RENSBERGEN, P. & MORLEY, C.K. 2000. 3D Seismic study of a shale expulsion syncline at the base of the Champion delta, offshore Brunei and its implications for the early structural evolution of large delta systems. Marine and Petroleum Geology, 17, 861-872. VENDEVILLE, B.C. & COBBOLD, PR. 1987. Glissements gravitaires synsedimentaires et failles normales listrique: modeles experimentaux: Paris, Acadenie des Sciences Comptes Rendus, 305,1313-1319. WALTHAM, D. 1997. Why does salt move? Tectonophysics, 282,117-128. WEIJERMARS, R., JACKSON, M.P.A. & VENDEVILLE, B.C. 1993. Rheological and tectonic modelling of salt provinces. Tectonophysics, 217, 143-174. WHITEMAN, A. 1982. Nigeria: Its Petroleum Geology, Resources and Potential. Graham and Trotman Ltd., London, 394p. WOOD, L. J. 2000. Chronostratigraphy and tectonostratigraphy of the Columbus basin, eastern offshore Trinidad. American Association of Petroleum Geologists Bulletin, 84,1905-1928 WORRAL, D.M. & SNELSON, S. 1989. Evolution of the northern Gulf of Mexico, with emphasis on Cenozoic growth faulting and the role of salt. In: BALLY, A. W. & PALMERS, A.R. (eds), The geology of North AmericaAn overview: Boulder, Colorado, Geology of North America, V.A., 97-138. Wu, S., BALLY, A.W. & CRAMEZ, C. 1990. Allochthonous salt, structures and stratigraphy of the north-eastern Gulf of Mexico, Part II, structure: Marine and Petroleum Geology, 1, 334-370 ZAMORA, G. 1999. 3D Delta Tectonics: Analogue modelling and structural interpretation of the Columbus Basin. MSc Thesis, Royal Holloway University of London, 159pp. (unpubl).
The role of shale deformation and growth faulting in the Late Cretaceous evolution of the Bight Basin, offshore southern Australia J.M. TOTTERDELL & A. A. KRASS AY Geoscience Australia, GPO Box 378, Canberra A.C.T. 2601, Australia (e-mail: Jennifer, totterdell @ ga. gov. au) Abstract: The Jurassic-Cretaceous Bight Basin of offshore southern Australia is a large, underexplored basin with little stratigraphic control. Sequence stratigraphic and structural interpretation of regional-scale 2-D seismic data has revealed the presence of two large progradational delta systems of mid-Late Cretaceous age, which are vastly different in terms of geometry and stratal architecture. The Cenomanian White Pointer Delta is characterized by shale deformation and growth faulting. Growth fault-controlled depocentres occupy a wide band across the Ceduna Subbasin of the Bight Basin that records the progradation of the delta across the palaeo-shelf during the Cenomanian. The growth faults are generally listric and basinward dipping (regional). Counter-regional faults are uncommon. The updip extensional features are accompanied downdip by a zone of diapirs that coincides with an outer basin high and, in deeper water at the toe of slope, a region of contractional deformation and toe-thrusts. In comparison, the younger (Late Santonian-Maastrichtian) Hammerhead Delta exhibits strongly progradational stratal geometries with little evidence of shale tectonics except in the SW part of the Ceduna Sub-basin. There, sustained progradation under a high sediment supply regime caused gravitational instability and the formation of listric growth faults at the palaeo-shelf margin with associated down-dip contractional deformation.
The Jurassic-Cretaceous Bight Basin is a large, mainly offshore basin that lies on the southern Australian margin in an area known as the Great Australian Bight (Fig. 1). Although the Bight Basin covers an area of over 250000 km2, it is a poorly explored region, with only nine wells drilled in the entire offshore portion of the basin. In 1999-2001, Geoscience Australia (formerly the Australian Geological Survey Organisation (AGSO)) undertook an integrated geological study of the Bight Basin, which resulted in the development of a new chronostratigraphic framework (Fig. 2) and an improved understanding of the tectonic and depositional history of the basin (Totterdell et al. 2000). A key component of this study was the sequence stratigraphic and structural interpretation of 8500 line kilometres of newly acquired, high quality, regional 2-D seismic reflection data. These investigations revealed that the Late Cretaceous evolution of the basin was dominated by the development of two large progradational delta systems in the Cenomanian and the Late Santonian-Maastrichtian. The architecture and areal extent of these deltaic systems is comparable with deltas of the West African passive margin, particularly the Niger Delta and the Orange Basin Delta.
Regional and tectonic setting The Bight Basin developed on the Australian southern margin during two successive periods of exten-
sion and thermal subsidence that commenced in the Middle-Late Jurassic (Totterdell et al 2000). The basin contains four main depocentres - the Ceduna, Duntroon, Eyre and Recherche sub-basins (Fig. 1). The two deltaic systems discussed in this paper are best developed within the northern portion of the largest and thickest of these, the Ceduna Sub-basin. The Ceduna Sub-basin underlies a broad bathymetric terrace (the Ceduna Terrace) in water depths ranging from 200 m to over 4000 m and contains a sedimentary section in excess of 12 km thick (Fig. 3). The Potoroo 1 well, which provides stratigraphic control in the northern Ceduna Sub-basin, intersects a thin mid-Late Cretaceous succession at the edge of the sub-basin; no wells have been drilled farther basinward. To the east of the main depocentres, a thin Bight Basin succession overlies Proterozoic basement of the Gawler Craton and deformed Early Palaeozoic rocks of the Kanmantoo Trough. To the north, the basement includes a variety of Proterozoic and older terranes. Basement trends have had a profound influence on the structural development of the Bight Basin, controlling the location and orientation of early basin-forming structures (Stagg et al 1990; Totterdell et al 2000). The Bight Basin is overlain unconformably by the dominantly cool-water carbonates of the Cenozoic Eucla Basin. To the south, the uppermost sequences of the Bight Basin onlap highly extended continental crust and rocks of the continent-ocean transition on the abyssal plain between Australia and Antarctica (Sayers et al 2001).
Fig. 1. Bight Basin location and structural elements map. Note location of the Potoroo 1 well on the NE margin of the Ceduna Sub-basin. Locations of seismic sections used in Figs 5, 6, 8, 9 and 11 shown in bold.
The Bight Basin formed within a tectonic framework dominated by the break-up of eastern Gondwana (Fig. 4). Basin development began in the Middle-Late Jurassic when extension along the southern margin of Australia formed one arm of a triple junction, with the other arms of the system along the incipient rifts between India and Antarctica, and India and western Australia (Norvick & Smith 2001; Fig. 4a). At this time, a convergent margin existed on the eastern margin of the continent. Rifting along the triple junction eventually resulted in sea-floor spreading between India and Australia/Antarctica, but the rift along the southern margin failed at that time. By the mid-Cretaceous, open ocean lay to the west and an arm of the sea extended into the Bight area along the failed rift (Fig. 4b). At around 90-100 Ma, subduetion ceased along the eastern margin, resulting in dynamic rebound of the cratonic platform (Waschbusch et al. 1999). This rebound is likely to have resulted in the development of a regional drainage gradient to the west. Kilometre-scale denudation took place along the eastern highlands (Raza et al 1995). It is possible that much of the sediment eroded from this area was transported west and SW towards the Bight depocentre (Veevers et al 1991; Raza et al 1995; Totterdell etal 2000). Very slow spreading between
Antarctica and Australia finally commenced in the Late Santonian around 83 Ma (Sayers et al 2001; Fig. 4c). This spreading rate continued until the Middle Eocene (Tikku & Cande 1999), when there was a dramatic increase in the rate of spreading and continental separation began in earnest (Fig. 4d). Because of the slow spreading, the seaway along the southern margin would have remained narrow until the Middle Eocene.
Tectonostratigraphy of the Bight Basin The tectonostratigraphic development of the Bight Basin can be described in terms of four basin phases (Fig. 2; Totterdell et al 2000). Basin Phase one (BP1) records the initiation of sedimentation during the Middle-Late Jurassic phase of intracontinental extension mentioned previously. This resulted in the formation of a series of extensional and transtensional half graben. Deformation appears to have been focused along pre-existing west-east and NW-SE striking basement trends (Fig. 1), although the extent of Jurassic age extensional structures beneath the thick Ceduna Sub-basin succession is difficult to determine. Half graben have been recognized along the eastern margin of the sub-basin but
SHALE TECTONICS OF THE BIGHT BASIN, AUSTRALIA
431
Fig. 2. Bight Basin sequence stratigraphy (from Totterdell et al 2000), showing the relationship with basin phases and the global sea level curve (modified after Haq et al. 1988).
could not be seismically resolved further west (Fig. 3). The extensional phase was followed by a period of slow thermal subsidence throughout most of the Early Cretaceous (BP2). Deposition during this time was largely non-marine. An abrupt increase in subsidence rate in the mid-Albian signalled the start of the third basin phase (BP3). This period of accelerated subsidence continued until the commencement of sea-floor spreading between Australia and Antarctica in the Late Santonian, which coincided with a period of rising global sea level (Fig. 2). This combination of factors resulted in a high rate of creation of accommodation, the first major marine flooding event in the basin and the widespread deposition of marine silts and shales of the Blue Whale Supersequence. The present-day distribution of the supersequence indicates that the seaway at that time extended along the southern margin from the open sea in the west towards the Otway Basin in the east. Progradation of deltaic sediments into this seaway (White Pointer Supersequence) commenced in the Cenomanian, following uplift and erosion along the eastern margin of the continent. Deposition was rapid, resulting in the development of overpressure in the underlying marine shales and a short-lived period of shale mobilization and growth faulting
throughout the northern half of the Ceduna Subbasin (Fig. 3). Continental break-up in the Late Santonian was followed by a period of thermal subsidence and the establishment of the southern Australian passive margin (BP4). It was during this phase that the second large progradational delta developed, represented by the Hammerhead Supersequence. In contrast with the earlier deltaic system, this sand-rich delta is characterized by strongly prograding stratal geometries and shows no evidence of widespread shale tectonics. A dramatic reduction in sediment supply at the end of the Cretaceous saw the abandonment of deltaic deposition and the development of a cool-water carbonate margin. The two deltaic phases feature very different stratigraphy, stratal geometry and structural architecture. They are separated by a dominantly aggradational marine succession and a regionally extensive, break-up-related unconformity underlies the later delta. For these reasons, we prefer to treat the two systems as separate deltas. In the absence of an extant river system, they are herein referred to informally as the White Pointer Delta and Hammerhead Delta, after the supersequences containing the deltaic successions.
432
J.M. TOTTERDELL & A. A. KRASS AY
Fig. 3. Geoseismic profile across the northern Ceduna Sub-basin. Jurassic-Early Cretaceous supersequences abbreviated as follows: SL Sea Lion; M Minke; SR Southern Right; BrW Bronze Whaler.
White Pointer Delta Stratigraphic and structural framework The White Pointer Delta comprises the deltaic sediments of the White Pointer Supersequence and the underlying marine silts and shales of the Blue Whale Supersequence. The few petroleum exploration wells drilled in the Ceduna and Duntroon sub-basins intersect the relatively thin, proximal parts of the basin succession. In these wells, the Blue Whale Supersequence generally consists of nearshore or restricted marine siltstone. The overlying White Pointer Supersequence contains dominantly fluvial to lagoonal siltstone and mudstone. The structural style of the White Pointer Delta is dominated by growth faults developed above a regionally extensive decollement within the mobile sediments of the Blue Whale Supersequence (Fig. 5). A limited set of depth-migrated seismic lines shows that the decollement has a very gentle seaward dip; in some distal areas of the basin it may dip landward. Growth faulting is interpreted to be accompanied by reactive shale diapirism. The reactive diapirs are triangular in cross-section and form the footwall to the adjacent growth faults (Fig. 5). In general, the height of the diapirs is equivalent to the throw on the fault. The seismic character of the mobile shale is typically chaotic, characteristic of overpressured facies and there is little seismic penetration beneath it. There is some seismic reflectivity at about the level of the decollement, which may mark the base of mobile facies. The listric growth
faults are dominantly regional, or basinwarddipping, faults. Counter-regional faults are uncommon and only a few have been recognized (Fig. 6). Throws on the regional faults are in the order of 1500-2500 m (approximate thickness herein based on depth-migrated seismic data or calculated using seismic stacking velocities). The growth faults have a generally arcuate trace and a NW-SE strike (Fig. 7). The regional faults dip to the WSW or SW. Throughout the northern Ceduna Sub-basin, the growth section, defined by reflections that diverge towards the bounding fault, is capped by a uniformly-thick aggradational unit (upper White Pointer Supersequence) that lacks synsedimentary growth (Figs 5, 6), indicating that faulting ceased during the late Cenomanian. The seismic character of the White Pointer Supersequence within the growth fault controlled depocentres (moderate to high continuity and amplitude) is suggestive of interbedded sandstone and mudstone. The presence of coal within the Supersequence is suggested by a band of high amplitude reflections along the eastern side of the Ceduna Subbasin. The region of growth faulting forms a 150 km wide band across the northern and central Ceduna Sub-basin (Fig. 7). Down-dip of this zone, across the continental slope is a zone of contractional deformation that features NE-dipping imbricate thrust fans (Figs 7, 8). These thrust faults sole out within the overpressured Blue Whale Supersequence. The thrust front most likely indicates the basinward limit of overpressure. Between these two zones is a structurally complex region (transitional zone) contain-
SHALE TECTONICS OF THE BIGHT BASIN, AUSTRALIA
433
Fig. 4. Plate tectonic reconstructions for the southern Australia-Antarctica conjugate margin. Modified after Norvick & Smith (2001).
ing zones of chaotic seismic reflection character and upright folds (Fig. 7). These folds could be simply shale-cored anticlines, but the absence of coherent seismic reflections beneath the crests of the anticlines and thickening of strata between them suggests diapiric movement of the overpressured shales (Fig. 9). This region, which underlies the seaward edge of the bathymetric terrace, has been termed the 'Outer Basin High' by previous workers (Stagg et al. 1990). Structural development The structural style of the White Pointer Delta is comparable to that of many other large progradational deltas around the world. Similar zones of growth faulting, accompanied farther basinward by zones of diapirism and contractional deformation, developed above a mobile substrate have been described from deltas such as the Niger and Baram (e.g. Doust & Omatsola 1990; Damuth 1994; Cohen & McClay 1996; Morley & Guerin 1996; McClay et al 1998; Van Rensbergen et al 1999), as well as the Gulf of Mexico (Buffler et al 1978; Winker &
Edwards 1983). Despite these structural similarities, the extent of active diapirism in the White Pointer Delta is significantly less than in the Niger and Baram deltas. Analogue modelling of large progradational delta systems has demonstrated that delta top extension is dynamically linked to delta toe contractional deformation and diapirism (McClay et al 1998; McClay et al 2003). Figure 10 compares the structural architecture of the White Pointer Delta in the northern Ceduna Sub-basin with Morley & Guerin's (1996) idealized profile of a progradational delta. Such deformation, where updip extension is compensated by downdip contraction, is driven by two processes that are end-members of a continuous series - differential loading and gravity sliding (Mandl & Crans 1981; Morley & Guerin 1996). Differential loading (due to lateral variations in load) of a weak substrate such as overpressured shale results in lateral pressure gradients and causes displacement of the mobile substrate beneath the load (Cohen & Hardy 1996). Gravity sliding involves the downslope displacement of material above a dipping ductile decollement (Jackson & Talbot 1991). Both processes can result in growth faulting and shale diapirism and it is likely that both contributed to the
Fig. 5. Seismic line from the northern Ceduna Sub-basin illustrating the typical structural style of the White Pointer Delta. Note growth section within the White Pointer Supersequence and listric faults soling out within the Blue Whale Supersequence. Note also the classic progradational geometry of the younger Hammerhead Supersequence. Ages are given (in italics) for the bases of the Blue Whale, Tiger and Hammerhead Supersequences. Location shown on Fig.l.
Fig. 6. Seismic line from the northern Ceduna Sub-basin showing a rare example of counter-regional faulting (CR). Ages are given (in italics) for the bases of the Blue Whale,, Tiger and Hammerhead Supersequences. Location shown on Fig.l.
436
J.M. TOTTERDELL & A. A. KRASSAY
Fig. 7. Structure map for the base of the White Pointer Supersequence, showing the distribution of growth faults and the farther basinward transitional and contractional zones.
development of the White Pointer Delta. The importance of gravity sliding as the dominant controlling mechanism is presented below. Deposition of the White Pointer Supersequence occurred rapidly, over a period of about 5 million years and rapid progradation of deltaic sediments across the undercompacted marine shales of the Blue Whale Supersequence led to the development of overpressure within the shales. The differential load provided by the prograding wedges of sediment triggered the reactive rise of shale diapirs and the commencement of growth faulting. The structural style of the delta provides an insight into the various factors that controlled deformation. In general, the normal faults are strongly listric and flatten out into a decollement within the Blue Whale Supersequence (Fig. 5). This style of faulting is similar to that in the South Texas region of the Gulf of Mexico (Winker & Edwards 1983) and is characteristic of gravity sliding-driven deformation (Morley & Guerin 1996). As would be expected of
such strongly listric faults, the heave-to-throw ratio of the growth faults is generally high. While fault throw is generally around 1500-2500 m, heave is on the order of 4000-8000 m, so heave to throw ratios range from 2:1 to greater than 3:1. According to Morley & Guerin (1996), this type of fault geometry indicates that downslope processes, that is gravity sliding, played a more important role than vertical processes in the development of the faults. Another striking feature of the White Pointer Delta is the dominance of regional over counter-regional faulting (Figs 5, 6). Such scarcity of counter-regional faulting is unusual in large progradational deltas. In the Niger and Baram deltas, for example, counterregional faults are a common feature of the outer parts of the shelf. Analogue modelling of progradational deltas has also shown the importance of counter-regional faulting (McClay et al 1998; McClay et al. 2003). This modelling suggests that in the development of delta-top graben systems, regional faults are the first to form and are then
SHALE TECTONICS OF THE BIGHT BASIN, AUSTRALIA
437
Fig. 8. Seismic line from the continental slope in northern Ceduna Sub-basin showing shale-cored anticlines and toethrusts of the outer White Pointer Delta. Location shown on Fig.l.
Fig. 9. Seismic line from the transitional zone between updip growth faults and downdip contractional structures, White Pointer Delta, northern Ceduna Sub-basin. Note the presence of possible diapiric features at 6.0-7.0 s (TWT), adjacent synformal strata, and the chaotic seismic character of much of the section, particularly beneath the crests of the anticlines. Location shown in Fig. 1.
438
J.M. TOTTERDELL & A. A. KRASSAY
Fig. 10. Comparison of White Pointer Delta and the idealized progradational delta of Morley & Guerin (1996). Note the similar distribution of extensional, diapiric/transitional and contractional structural zones.
superseded by generally larger counter-regional faults. Lewis (1997), in a study of the development of growth faults, found that counter-regional faults were less common in areas with thin substrates. The study showed that counter-regional faults only formed at high rates of extension and that the thickness of the mobile substrate was the key control on the rate of extension. Therefore, if the mobile unit was too thin, counter-regional faults could not develop. This may explain the faulting patterns of the White Pointer Delta, as the mobile substrate here is thin relative to other large deltas. For example, the Setap Shales of the Baram Delta are more than 3000 m thick (Van Rensbergen et al 1999) and the Akata Shales of the Niger Delta over 6000 m thick (Cohen & McClay 1996). By comparison, the maximum pre-deformation thickness of the Blue Whale was probably around 1500-2000 m. The relatively thin mobile shale in the Ceduna Sub-basin may also have provided a constraint on the amount of growth possible on the regional faults. Possible touchdown areas, where the base of the growth-faulted section impinges upon the decollement, have been recognized on several seismic profiles (Fig. 5). The cessation of growth faulting, marked by the change in sedimentary architecture from growth strata to simple aggradation, may therefore have been controlled by the vertical limit of mobile shale rather than dewatering. Shale deformation and growth faulting within the White Pointer Delta appears to be restricted to the northern half of the Ceduna Sub-basin. Thicker suc-
cessions of the Blue Whale and White Pointer Supersequences occur in the southern half of the basin, but there is no evidence of the type of shale tectonics that are evident to the north. The most likely explanation is that overpressure did not develop in the Blue Whale Supersequence in this part of the basin. This could have been related to either changes in lithofacies, or to the rate of burial. The White Pointer Supersequence in this area has a strongly aggradational stratal geometry, which suggests that it was away from the direct influence of the delta and the rapid progradation of deltaic sediments. Therefore, the sedimentation rate was not rapid enough to lead to the development of overpressure, and there was no differential loading to trigger shale mobility or diapirism.
Hammerhead Delta Stratigraphic and structural framework From the Turonian to the Late Santonian, in the lead up to seafloor spreading, deposition in the Bight Basin occurred in a rapidly subsiding, dominantly restricted marine environment. During this period, the marine and marginal marine sediments of the aggradational Tiger Supersequence accumulated (Fig. 2). Uplift associated with continental break-up resulted in widespread exposure and incision of the former shelf. A major unconformity separates the
SHALE TECTONICS OF THE BIGHT BASIN, AUSTRALIA
439
Fig. 11. Seismic line from the central Ceduna Sub-basin showing the gravity-driven deformation at the palaeo-shelf margin of the Hammerhead Delta. Location shown on Fig.l.
deltaic Hammerhead Supersequence from the underlying Tiger Supersequence. In places, the contact is an angular unconformity where gentlydipping tilt blocks of the Tiger Supersequence have been eroded. This relationship suggests at least local uplift, possibly related to the commencement of seafloor spreading, occurred prior to deposition of the Hammerhead Supersequence. The Hammerhead Supersequence represents a 19 million year period of sustained deltaic sedimentation in the Bight Basin. The Hammerhead Delta is a sand-rich shelf-margin delta complex that is subdivided into three internal sequence sets. The lower two sequence sets (Late Santonian to Campanian) exhibit strongly progradational geometries (Fig. 3). The upper sequence set (Maastrichtian) has a dominantly aggradational geometry indicating a major change in basin dynamics due to a gradually decreasing rate of sediment supply. Well data shows that in proximal parts of the basin, the Supersequence consists mainly of amalgamated fluvial channel sandstones. In distal parts of the basin the Hammerhead Supersequence comprises basinward-thinning wedges of marine shale at the toe of slope. Between these two end members of the deltaic depositional system, there are 14 other mappable seismic facies that have been interpreted to represent coastal and deltaic plain, shallow marine and shelf-margin to slope palaeoenvironments. In the NW Ceduna Sub-basin, the Hammerhead Delta is only affected by widely spaced, high-angle planar faults of Late Maastrichtian-Paleocene age. These faults are either
reactivated Cenomanian growth faults, or are new structures that have nucleated at the tips of the earlier growth faults, which were previously reactivated during Turonian-Santonian extension (Figs 3, 5). Fault density within the Hammerhead Supersequence increases to the southeast and west. The most striking structural feature of the Hammerhead Supersequence is the series of high-offset, listric growth faults that occur near the SW margin of the Ceduna Sub-basin across a narrow belt subparallel to the palaeoshelf-margin. The growth section on these faults reaches thicknesses of up to 3500 m. The extensional faults are accompanied downdip by contractional deformation in the form of imbricate thrust fans (Fig. 11). The deformation appears to have taken place in the Campanian, during the strongly progradational phase of deltaic development.
Structural development The marked difference in structural style between the Hammerhead and White Pointer deltas is evident on seismic profiles (Fig. 3). The most striking differences are the strongly progradational stratal geometry of the Hammerhead Delta and the absence of an underlying ductile and mobile substrate. Across the shelf, the underlying sediments show no evidence of overpressure and deformation. In fact, the clear evidence of erosion at the sequence boundary, such as large incised valleys, argues that the underlying sediments were not rapidly buried. In addition, there is
440
J.M. TOTTERDELL & A. A. KRASSAY
Fig. 12. Isopach map of the Hammerhead Supersequence (ms TWT)
no seismic evidence of the presence of thick coeval pro-delta or basinal muds. For the most part, particularly in the northern part of the basin, the Hammerhead Delta simply prograded across a broad shelf, with no syn-depositional faulting. However, this situation changed at the palaeoshelf margin. The time isopach map of the Hammerhead Supersequence (Fig. 12) shows that it thickens both basinward and towards the south, where it reached thicknesses in excess of 3000 ms TWT (over 5000 m). It is at the palaeo-shelf margin, where the supersequence is thickest, that gravitational instability resulted in localized deformation, with large growth faults compensated down-dip by folds and imbricate thrust systems (Fig. 11). This deformation is a classic example of gravity sliding at an unstable shelf margin. The decollement occurs within the basal part of the Hammerhead Supersequence and was probably provided by overpressured pro-delta shale or fine-grained slope facies. Deformation was accompanied by only limited shale mobilization, with relatively small triangular shale masses present beneath the large growth faults and detachment folds occurring within the transitional zone between growth faults and toe-thrusts (Fig. 11). Comparable examples of gravity-driven deformation, with the
development of updip listric growth faults and downdip contractional deformation, have been recognized from other basins in similar tectonic settings, including the west African passive margin (Duval etal 1992; Turner 1995). The structural and stratigraphic architecture of the Hammerhead Delta is markedly different from deltas that involve deformation of a mobile substrate, such as the earlier White Pointer Delta and the Niger and Baram deltas. It does, however, exhibit a broad similarity to the relatively coarse clastic progradational delta of the Orange Basin on the southwest African passive margin (Muntingh & Brown 1993). Both deltas feature strongly progradational to aggradational stratal geometries and load-induced gravitational failure at the shelf margin. The Orange River Delta, which is proving to be a successful petroleum province, provides an interesting analogue for understanding the reservoir and trapping potential of the Hammerhead Delta.
Petroleum implications Growth faulting and shale mobility are interpreted to have played important roles in the development of
SHALE TECTONICS OF THE BIGHT BASIN, AUSTRALIA
the Bight Basin and are directly related to its hydrocarbon prospectivity. The structural fabric developed during the evolution of the White Pointer delta has provided a strong control on later deformation. During the extensional phase prior to break-up in the Late Santonian and flexure of the margin in the Late Maastrichtian-Early Paleocene, the Cenomanian growth faults were selectively reactivated. These structures provide important potential fluid pathways for migrating hydrocarbons, and rollovers into the growth faults are potential structural traps. In addition, both the overpressured marine shale (Blue Whale Supersequence) and the prograding deltaic sediments (White Pointer Supersequence) contain potential source rocks (Struckmeyer et al 2001). The Blue Whale Supersequence has proven source potential in onshore wells (Boreham et al. 2001) and enhanced potential for the preservation of organicrich rocks within rapidly subsiding growth faulted depocentres makes the White Pointer Supersequence an attractive potential source rock interval. Growth faulting is also important for the petroleum prospectivity of the Hammerhead Delta. Synsedimentary growth within the large listric faults at the palaeoshelf-margin provides enhanced reservoir potential due to thickness increases and greater trap potential with the formation of large rollover structures within the larger rotated fault blocks.
Conclusions The Albian-Cenomanian White Pointer and Santonian-Maastrichtian Hammerhead deltas of the Bight Basin provide an example of two very different types of progradational delta. The White Pointer Delta is characterized by regionally extensive shale mobilization and shares many characteristics with other deltas developed above mobile substrates, such as the Niger and Baram deltas, although on a smaller vertical scale. A combination of differential loading and gravity sliding resulted in the development of a broad zone of basinward-dipping listric growth faults above a decollement formed within overpressured marine shales. These extensional structures were accompanied downdip by a zone of complex deformation, including possible diapirism, and farther offshore by a contractional fold and thrust belt. The thickness of the mobile substrate appears to have provided a constraint on both the style of faulting and the amount of growth. In contrast, the younger Hammerhead Delta is characterized by strongly progradational stratal geometries and a general absence of syndepositional shale deformation or faulting other than a narrow band of gravity driven deformation at the palaeo-shelf margin. The Hammerhead Delta has broad similarities with the coarse clastic progradational systems
441
of the Orange Basin and exhibits shelf-margin gravity tectonics similar to that seen along parts of the west African margin. We wish to thank B. Bradshaw, P. O'Brien, N. Lemon and M. de Jong for their thoughtful reviews and Fugro MCS for permission to use the seismic images. This paper is published with the permission of the Chief Executive Officer, Geoscience Australia.
References BOREHAM, C.J., KRASSAY, A.A. & TOTTERDELL, J.M. 2001. Geochemical comparisons between asphaltites on the southern Australian margin and source rock analogues of Early Cretaceous age. In: HILL, K.C. & BERNECKER, T. (eds) Eastern Australasian Basins Symposium. Petroleum Exploration Society of Australia, Special Publication, 531-541. BUFFLER, R.T., SHAUB, F.J., WATKINS, J.S. & WORZEL, J.L. 1978. Anatomy of the Mexican Ridges, southwestern Gulf of Mexico. In: WATKINS, J.S., MONTADERT, L. & DICKERSON, P.W. (eds) Geological and geophysical investigations of continental margins. American Association of Petroleum Geologists, Memoir 29, 319-327. COHEN, H.A. & HARDY, S. 1996. Numerical modelling of stratal architectures resulting from the differential loading of a mobile substrate. In: ALSOP, G.I., BLUNDELL, D.J. & DAVISON, I. (eds) Salt Tectonics. Geological Society, London, Special Publications, 100,265-273. COHEN, H.A. & McCLAY, K. 1996. Sedimentation and shale tectonics of the northwestern Niger Delta front. Marine and Petroleum Geology, 13, 313-328. DAMUTH, I.E. 1994. Neogene gravity tectonics and depositional processes on the deep Niger Delta continental margin. Marine and Petroleum Geology, 11,320-346. DOUST, H. & OMATSOLA, E. 1990. Niger Delta. In: EDWARDS, J.D. & SANTOGROSSI, P.A. (eds) Divergent/passive margin basins. American Association of Petroleum Geologists, Memoir, 48,201-238. DUVAL, B., CRAMEZ, C. & JACKSON, M.P.A. 1992. Raft tectonics in the Kwanza Basin, Angola. Marine and Petroleum Geology, 9,389-404. HAQ, B.U., HARDENBOL, J. & VAIL, PR. 1988. Mesozoic and Cenozoic chrono stratigraphy and cycles of sealevel change. In: WILGUS, C.K., HASTINGS, B.S., POSAMENTffiR, H., VAN WAGONER, J., ROSS, C.A. &
KENDALL, C.G. (eds) Sea-level changes: an integrated approach. Society of Economic Paleontologists and Mineralogists, Special Publication, 42,71-108. JACKSON, M.P.A. & TALBOT, C.J. 1991. A glossary of salt tectonics. Bureau of Economic Geology, Geological Circular, 91-4. LEWIS, G.J. 1997. The development of growth faults above ductile substrates. PhD thesis, Royal Holloway, University of London. MANDL, G. & CRANS, W. 1981. Gravitational gliding in deltas. In: McCLAY, K.R. & PRICE, NJ. (eds) Thrust and Nappe Tectonics. Geological Society, London, Special Publications, 9,41-54. McCLAY, K.R., DOOLEY, T. & LEWIS, G. 1998. Analog
442
J.M. TOTTERDELL & A. A. KRASSAY
modeling of progradational delta systems. Geology, 26,771-774. McCLAY, K., DOOLEY, T. & ZAMORA, G. 2003. Analog models of delta systems developed above mobile shale detachments. In: VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, AJ. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216,411-428. MORLEY, C.K. & GUERIN, G. 1996. Comparison of gravitydriven deformation styles and behaviour associated with mobile shales and salt. Tectonics, 15, 1154-1170. MUNTINGH, A. & BROWN, L.F. 1993. Sequence stratigraphy of petroleum plays, post-rift Cretaceous rocks (lower Aptian to upper Maastrichtian), Orange Basin, western offshore, South Africa. In: WEIMER, P. & POSAMENTIER, H.W. (eds) Siliciclastic sequence stratigraphy - recent developments and applications. American Association of Petroleum Geologists, Memoirs, 58,71-97. NORVICK, M.S. & SMITH, M.A. 2001. Mapping the plate tectonic reconstruction of southern and southeastern Australia and implications for petroleum systems. The APPEA Journal, 41, 15-35. RAZA, A., HILL, K.C. & KORSCH, RJ. 1995. MidCretaceous regional uplift and denudation of the Bowen-Surat basins, Queensland and its relation to Tasman Sea rifting. In: FOLLINGTON, I.L., BEESTON, J.W. & HAMILTON, L.H. (eds) Bowen Basin Symposium 1995 — 150 years on. Geological Society of Australia, supplement. SAYERS, J., SYMONDS, P., DIREEN, N.G. & BERNARDEL, G. 2001. Nature of the continent-ocean transition on the non-volcanic rifted margin of the central Great Australian Bight. In: WILSON, R.C.L., WHTTMARSH, R.B., TAYLOR, B. & FROITZHEIM, N. (eds) Nonvolcanic Rifting of Continental Margins: a Comparison of Evidence from Land and Sea. Geological Society, London, Special Publications, 187,51-77. STAGG, H.MJ., WILLCOX, J.B., NEEDHAM, D.J.L., O'BRIEN, G.W, COCKSHELL, C.D., HELL, A.J., THOMAS, B. & HOUGH, L.P. 1990. Basins of the Great Australian Bight region: geology and petroleum potential. Continental Margins Program Folio 5. Bureau of Mineral Resources, Geology and Geophysics, Australia, and Department of Mines and Energy, South Australia.
STRUCKMEYER, H.I.M., TOTTERDELL, J.M., BLEVIN, J.E., LOGAN, G.A., BOREHAM, C.J., DEIGHTON, I., KRASSAY, A.A. & BRADSHAW, M.T. 2001. Character, maturity and distribution of potential Cretaceous oil source rocks in the Ceduna Sub-basin, Bight Basin, Great Australian Bight. In: HILL, K.C. & BERNECKER, T. (eds) Eastern Australasian Basins Symposium. Petroleum Exploration Society of Australia, Special Publication, 543-552. TIKKU, A.A. & CANDE, S.C. 1999. The oldest magnetic anomalies in the Australian-Antarctic Basin: are they isochrons? Journal of Geophysical Research, 104, B 1,661-677. TOTTERDELL, J.M., BLEVIN, I.E., Struckmeyer, H.I.M., Bradshaw, B.E., Colwell, J.B. & Kennard, J.M. 2000. A new sequence framework for the Great Australian Bight: starting with a clean slate. The APPEA Journal, 40,95-117. TURNER, J.P. 1995. Gravity-driven structures and rift basin evolution: Rio Muni Basin, offshore Equatorial West Africa. American Association of Petroleum Geologists Bulletin, 79, 1138-1158. VAN RENSBERGEN, P., MORLEY, C.K., ANG, D.W., HOAN, T.Q. & LAM, N.T. 1999. Structural evolution of shale diapirs from reactive rise to mud volcanism: 3D seismic data from the Baram delta, offshore Brunei Darussalam. Journal of the Geological Society, London, 156, 633-650. VEEVERS, J.J., POWELL, C.McA. & ROOTS, S.R. 1991. Review of seafloor spreading around Australia. I. Synthesis of the patterns of spreading. Australian Journal of Earth Sciences, 38, 373-389. WASCHBUSCH, P., BEAUMONT, C. & KORSCH, RJ. 1999. Geodynamic modelling of aspects of the New England Orogen and adjacent Bowen, Gunnedah and Surat basins. In: FLOOD, P.G. (ed.) New England Orogen: regional geology, tectonics and metallogenesis. Earth Sciences, University of New England, Armidale, 203-210. WINKER, C.D. & EDWARDS, M.B. 1983. Unstable progradational clastic shelf margins. In: STANLEY, D.J. & MOORE, G.T. (eds) The shelfbreak: critical interface on continental margins. Society of Economic Paleontologists and Mineralogists, Special Publication 33,139-157.
Pliocene to Recent mud diapirism and related mud volcanoes in the Alboran Sea (Western Mediterranean) A.R. TALUKDER1, M.C. COMAS & J.L SOTO Institute Andaluz de Ciencias de la Tierra (C.S.I.C.-Universidad de Granada), Campus Fuentenueva s/n, 18002 - Granada, Spain (e-mail: [email protected]) Abstract: High resolution and multichannel seismic profiles depict the Pliocene to Recent evolution of the mud diapirism in the West Alboran Basin (WAB) and its relationship with the Miocene diapir province that occupies the WAB depocentre. During the early to middle Miocene period of basin extension (16 to 9 Ma), normal faulting triggered the diapirism from mobile overpressured shale containing olistostromes. Plio-Quaternary diapirism evolved as a second main stage of diapiric activity and developed throughout the subsequent contractive tectonic evolution of the basin (9 Ma to Holocene). Mud volcanoes, discovered to the south of the WAB, developed on the flank of Recent diapirs as a consequence of the rise of fluidized sediments through diapiric bodies and/or through fractures connecting with deeper diapirs. During the Pliocene to Recent, some diapirs stopped ascending, leading to the production of collapse structures on their tops due to lateral subsurface mud migration and/or fluid escape. Other cylindrical shaped diapirs continued rising and produced mud volcanoes on the sea floor. All the studied volcanoes seem to be currently inactive. Two major pulses of diapiric rise have been distinguished during the Pliocene to Recent contractive evolution of the basin.
Although mud diapirism and mud volcanism are geological phenomena that have long been known, in recent times more emphasis has been placed on tectonic history to explain their evolution and triggering mechanisms. It is generally accepted that the emplacement of diapirs is controlled not only by the physical properties of the rising materials, but also by geological events affecting the basin (e.g. Bishop 1978). Tectonic processes are commonly the main driving mechanism invoked for the development of mud diapirism and volcanism (e.g. Higgins & Saunders 1974; Barber et al 1986; Vendeville & Jackson 1992a, b\ Vernette et al 1992; Jackson & Vendeville 1994; Limonov et al 1996; Fowler et al 2000), in addition to high overpressure in sediments due to rapid sedimentation and gas generation (Hedberg 1974; Westbrook & Smith 1983; Brown 1990; Reed et al 1990; Hovland et al 1997). According to these authors, diapir evolution can be assumed to have four evolutionary phases (reactive, active, passive and diapir collapse) and the occurrence of any phase is the result of the interplay of regional tectonics, density ratios between diapiric materials and overburden, their relative sedimentary thickness, and the rate of sedimentation. These phases, originally described for salt diapirs, are nonetheless comparable to those encountered for mud diapir evolution (e.g. Morley & Guerin 1996; Van Rensbergen et al 1999). Mud diapirism and mud volcanism have been found in various tectonic settings. The majority have been reported in accretionary wedges, where the main tectonic forces are compressional, such as in the Mediterranean Ridge (Fig. 1) (Camerlenghi et al
1992;Citaeffl/. 1996; Robertson ef al 1996;Fusi& Kenyon 1996; Limonov et al 1997; Kopf et al 2000, among others), Eastern Indonesia (Barber et al 1986), Barbados (Biju-Duval et al 1982; Westbrook & Smith 1983; Martin & Kastner 1996) and Makran (Wiedicke et al 2001). In a comparative study between the Black Sea and the eastern Mediterranean-ridge mud volcanoes, Ivanov et al (1996) showed that mud volcanoes that have developed under an extensional regime and with a high terrigenous input (e.g. the Black Sea) have a very distinctive deep structure compared with those developing under subhorizontal compression (e.g. the Mediterranean Ridge). Therefore, the tectonic setting controls the mud-volcano structure. The Alboran Sea basin allows mud diapirism phenomena to be studied in a different tectonic setting, since mud diapirism has existed since the middle Miocene and developed through the Cenozoic geodynamic history of the basin, typified by alternating extension and compression. Unlike the eastern Mediterranean, detailed studies of the mud diapirism in the Alboran Sea are scarce and most of the previous works mainly refer simply to their occurrence in the basin (e.g., Bourgois et al 1992; Campillo et al 1992; Comas et al 1992, 1999; Jurado & Comas 1992; Maldonado et al 1992; Watts et al 1993; Chalouan et al 1997). Perez-Belzuz et al (1997), however, established an evolutionary model for the complete diapirism (Miocene to Recent) in the Alboran Sea basin based on the thin-skinned extension model of Vendeville & Jackson (1992a, b). This paper centres on an analysis of the Plioceneto-Recent evolution of the mud diapirism and on its
Fig. 1. Mud diapirs and mud volcanoes in a tectonic sketch of the Mediterranean Sea, showing Neogene extensional basins and the external front of the surrounding Alpine thrust belts (taken from Comas et al. 1999). GA = Gibraltar arc; HA = Hellenic arc; MR = Mediterranean Ridge; TA — Tyrrhenian arc. Arrows indicate motion of the African plate from the Oligocene to lower Miocene (a) and from the lower Miocene to Recent (b) (taken from Dewey et al 1989). Location of mud diapirs and mud volcanoes are as follows: Adriatic Sea: Hovland & Curzi (1989); Alboran Sea: PerezBelzuz et al (1997) and this work; Gulf of Cadiz (west of the Gibraltar arc): Gardner (2001); Hellenic arc: Cita et al (1981); Camerlenghi et al (1992); Ivanov et al (1996); Limonov et al (1996, 1997); Cronin et al (1997); Kopf etal (1998, 2000); and Sicily: Higgins & Saunders (1974).
relationship with the mud volcanoes recently discovered in the West Alboran basin (WAB). On the basis of the interpretation of seismic reflection profiles (both multi-channel and high resolution seismic profiles), we present a detailed contour map of the top of the Plio-Quaternary mud diapirs and of their intersections with key, basin-wide sedimentary unconformities. The geometric and genetic relationships between these mud diapirs and the mud volcanoes are analysed to infer the direction of subsurface mud migration during the Pliocene-to-Quaternary.
Geological setting The Alboran Sea basin, behind the Gibraltar Arc in the inner part of the Betic-Rif mountain chain (Fig. 1), is known to have formed by crustal extension in a general Eurasian-African plate convergence setting. The direction of plate convergence was north-south between the middle Oligocene to the late Miocene and changed to NW-SE from the latest Tortonian (9-8 Ma) to the present-day (Dewey et al 1989; Srivastava et al 1990; Mazzoli & Helman 1994; Morel & Meghraoui 1996) (see plate-kinematic
vectors in Fig. 1). Extension and crustal stretching in the basin were coeval with thrusting and shortening in the peripheral mountain belt. Extensional tectonics affecting the sedimentary cover was active from at least the early Miocene (about 22 Ma) to the late Tortonian (about 9-8 Ma). Since then, contractional tectonics have produced an inversion of previous normal faults, reverse and strike-slip faults, and folding (Ait Brahim & Chotin 1989; Morel 1989; Comas et al 1992; Garcia-Duenas et al 1992; Montenat et al 1992; Watts et al 1993; Meghraoui etal 1996; Martinez-Martinez eral 1997; AlvarezMarron 1999; Comas & Soto 1999; Comas et al 1999). The Alboran Sea is about 400 km long and 200 km wide, with a maximum water depth of 2 km. Three main sub-basins, the West Alboran Basin (WAB), the East Alboran Basin (EAB), and the South Alboran Basin (SAB) (Fig. 2), separated by several ridges, seamounts, and troughs, determine the seafloor morphology. The Alboran Ridge is the most prominent NE-SW linear relief in the Alboran Sea, emerging locally to form the volcanic Alboran Island. The Xauen Bank, the morphological prolongation of the Alboran Ridge towards the SW, is
MUD DIAPIRISM AND VOLCANISM IN THE ALBORAN SEA
Fig. 2. Structural map of the West Alboran basin and surrounding areas showing the diapir province and main sedimentary depocentres (taken from Comas et al 1999). AI = Alboran island; AR = Alboran ridge; EAB = East Alboran basin; SAB = South Alboran basin; SBB = South Balearic Basin; 976 = WAB = West Alboran Basin; and XB = Xauen bank. ODP Site 976; Commercial wells: Alb-Al = Alboran Al; EJ - El Jebha; Gl - Andalucia Gl. Inset map shows simplified bathymetry of the westernmost Mediterranean (Alboran and South-Balearic seas) (taken from Comas et al. 1999).
formed by folded sedimentary sequences, showing ENE-WSW trending close folds (Bourgois et al 1992; Comas etal 1999) (Fig. 2). Mud diapirism and associated mud volcanoes occurred only in the WAB, developing an extensive diapir province in the major Neogene sedimentary depocentre of the basin (Fig. 2). The sedimentary cover in the WAB is formed of early Miocene to Quaternary deposits up to 7-9 km thick, in which six lithoseismic units (labelled VI-I from bottom to top) have been recognized (Fig. 3) (Comas et al. 1992; Jurado & Comas 1992; Soto etal 1996; Chalouan et al 1997). According to data from the commercial well Alboran A1 (Fig. 3), the older deposits overlying the basement are marine sediments of latest Aquitanian (?)-Burdigalian age (Unit VI) and consist of sediments containing olistostromes. Sonic velocity, density and resistivity logging-data show the
445
Fig. 3. Correlation between sedimentary sequences drilled at ODP Site 976, and commercial wells Andalucia G-1 and Alboran A-1 in the WAB. Seismic stratigraphic units and major regional unconformities (R) are also shown. M = Messinian unconformity; MB = metamorphic basement; TB = top of basement. Inset map shows the location of commercial wells and ODP Site 976.
typical features of under-compacted shales in Unit VI (Alboran Al well) and also at the base of Unit V (Andalucia A-1 well) (Jurado & Comas 1992). Major basin-wide angular unconformities within the sedimentary cover occur at the top of Unit VI (Reflector R5), at the base of Unit III (Reflector R3), and at the base of Unit I (Comas et al 1992, 1999; Chalouan et al 1997). The base of Unit I (PliocenePleistocene), imaged in the seismic profiles as a prominent channeled, erosional unconformity, correlates with the top of the Messinian evaporite sequence recognized throughout the Mediterranean (the "M reflector" of Ryan, Hsu et al 1973; Cita & McKenzie 1986; Comas, Zahn, Klaus et al 1996). The Messinian deposits (Unit II) are depleted of salt and consist of marine siliciclastic and shallow carbonate facies, with occasional gypsum and anhydrite thin intervals (Jurado & Comas 1992; Comas, Zahn, Klaus et al 1996). The Pliocene-Quaternary sequence (Unit I) consists of pelagic marls, muddy turbidites, hemipelagic clays and rare silty-sand turbidites, that have been studied in detail from several holes drilled during ODP Leg 161 (Comas, Zhan, Klaus et al 1996). Based on seismic interpretation
446
A.R.TALUKDERCTAL.
Fig. 4. Location map of high resolution and multichannel seismic profiles used in this work. Bold lines correspond to seismic lines presented in figures. Commercial wells as in Fig. 2. Mud volcanoes: gmv = Granada mud volcano; mvl and mv2 = northern mud volcanoes (taken from Perez-Belzuz et al. 1997); mmv^ Marrakech mud volcano; smv = Southern mud volcano.
respectively, of the parallel passing through the Gibraltar Strait (Fig. 4). The study of the southern sector has been accomplished with high resolution seismic profiles and sidescan sonar images (OKEAN, 9 kHz) acquired during the B AS ACALB cruise (1999, TTR-9, Leg 3; R/V Professor Logachev). The sidescan sonar (OKEAN) images carried out during this survey revealed two mud volcanoes (Figs 4 and 5). The two volcanoes, named Granada and Marrakech, were also imaged in mud-penetrator profiles and were subsequently sampled by gravity coring. OKEAN profiles covering a mosaic of about 1120 km2 and accompanied by high-resolution seismic lines 70 km long, were also recorded during the BASACALB cruise (Fig. 4). These lines have been used for a detailed study of the mud diapirism and volcanism during the Pliocene and Quaternary. Commercial and scientific multi-channel seismic (MCS) lines crossing the area were also studied to correlate the lithoseismic units, differentiated by seismic interpretation, and to determine their relative ages. In the northern sector, multi-channel seismic (MCS) profiles with a maximum penetration of 6 sec (twtt), forming a grid of about 4 X 4 km2 (Fig. 4), were used to produce a structural contour map of the diapir culminations and of the M-unconformity (the base of Unit I). Lithoseismic correlation was established by tying the seismic lines with ODP Site 976 (Leg 161) and with the commercial well Andalucia Gl (Figs 3 and 4).
Mud volcanoes and stratigraphic data from ODP Site 976 (Shipboard Scientific Party 1996), Unit I (Fig. 3) was further subdivided into Unit la of Quaternary age and Units Ib and Ic of Pliocene age (Tandon et al. 1998; Comas et al. 1999). In accordance with the above seismic studies, mud diapirs in the WAB have their origins in the lowermost sedimentary unit (Unit VI) and probably also in the base of Unit V (Chalouan et al 1997; Perez-Belzuz et al. 1997). Micropalaeontological data from commercial wells indicate that Unit VI encompasses material of very different ages (from Cretaceous to Oligocene) and its age has been postulated as lower Miocene (Jurado & Comas 1992). Mud breccia samples from gravity cores in the top of the mud volcanoes also contain Eocene and upper Cretaceous clasts, but the age of the mud-volcanic matrix is lower Miocene (Sautkin 2000; Sautkin et al 2003).
Features of the mud diapirs and volcanoes This study has been carried out in two sectors of the Iberian and Morocco margins, north and south,
The morphology of the mud volcanoes in the sea bottom was analysed on side-scan sonar images (OKEAN). Mud volcanoes are observed as circular or semicircular features with irregular boundaries and a general high backscatter intensity compared with the general level of background backscatter in the southern sector (Fig. 5). Stronger backscatter can be caused by an irregular relief of the crater and cone and/or by the presence of mud breccia and rock clasts (sampled by gravity coring) immediately below the seafloor or cropping out at the seafloor surface (Ivanov et al 1996; Volgin & Woodside 1996). The Marrakech mud volcano has a diameter of c. 1 km and is located on the flank of a diapiric high with high backscatter intensity on its top and irregular boundaries. On the SE flank of this high, towards the Marrakech mud volcano, a series of patches with moderate backscatter are observed, which can be interpreted as mud flows from the volcano (Fig. 5). The Granada mud volcano has an outer ring (diameter c. 1.8 km) with moderate backscatter and an inner elliptical patch with high backscatter. This feature is offset towards the NE of the outer circle (Fig. 5).
MUD DIAPIRISM AND VOLCANISM IN THE ALBORAN SEA
447
Fig. 5. Sidescan sonar mosaic (OKEAN) of the seafloor in the mud volcano area (see location in Fig. 4), with detailed images of the Granada (a) and Marrakech (b) mud volcanoes. Bathymetry contour lines are in metres. Note that the OKEAN mosaic covers an area slightly smaller than the box titled 'Southern Sector' in Fig. 4.
The variation in backscatter intensity among the mud volcanoes and mud flows is probably caused by variations in the thickness of pelagic sediments overlying the mud breccia, which produce an attenuation of the sidescan sonar signal (e.g. Volgin & Woodside 1996). Moderate backscatter around the Granada mud volcano may also be due to the presence of a shallow mud diapir. Among the mud volcanoes discovered, the Marrakech volcano is the only one crossed by high resolution seismic profiling and the seismic track line crosses it off-centre. This mud volcano developed at the flank of a diapir high and has a negative seafloor expression. The Conrad 828 MCS line in the southern part of the southern sector shows a possible mud volcano, with a positive seafloor expression, which we termed the southern mud volcano (Fig. 6). It has an overall vertical, cylinder-like feeder channel with irregular vertical contacts probably due to interfingering with host rocks (Fig. 6). The feeder channel has a diameter of c. 1 km till 2 sec twtt, widening to c. 1.7 km in the source layer, at 4 sec twtt. Just above and below the M reflector (at about 1.6 to 2 sec twtt), the sedimentary sequence bends towards the volcano feeder channel, probably due to the development of collapse structures.
Between 1.2 to 1.4 sec twtt, within early Pliocene sediments, the channel shows a crater collapse structure and cone geometry on both sides. At 1.1 sec twtt, the channel appears to form lens-like bodies and the onlap geometry of the reflectors suggests mud extrusion rather than intrusion processes. The sediments overlying the channel (<0.9 sec twtt) are characterized by more transparent seismic facies than the host layered facies and the whole packet of reflectors is bent upwards.
Mud diapirs The diapiric mobile shale is identified in MCS seismic profiles as a low-velocity, chaotic to transparent facies, not always confined to the layered stratigraphic units (e.g. Higgins & Saunders 1974; Westbrook & Smith 1983; Cohen & McClay 1996; Morley & Guerin 1996). Our data confirm that the source of the mud diapirs and volcanoes belongs to the lithoseismic Unit VI, which is marked by high diffraction involving hyperbolic reflections and chaotic seismic facies. Unit VI corresponds to the lowermost sedimentary sequence in the WAB (Fig. 3). The absence of layering and the lack of laterally
448
A.R. TALUKDER ETAL
Fig. 6. MCS line CONRAD 828 showing the southern mud volcano in the mud volcano area (southern WAB). Note that the mud-volcano is rooted beneath the upper-to-middle Miocene sequence (Units II to V) and underwent several phases of activity from the late Miocene onwards. Location of seismic line is shown in Fig. 4. Seismic units and unconformities as defined in Fig. 3.
MUD DIAPIRISM AND VOLCANISM IN THE ALBORAN SEA
449
Fig. 7. High resolution seismic profile (Ms 170) across a mud-diapir in the southern WAB and draw-line interpretation. Note angular unconformities at the base of sub-units Ql and Q2 and the highly variable thickness of sub-unit Q2, denoting two major pulses of diapiric rise. The NW border of the diapir corresponds to a growth fault. Location of seismic line is shown in Fig. 4. Seismic units and unconformities as denned in Fig. 3.
continuous reflectors in the olistostrome sequence of Unit VI are notable. In high resolution seismic profiles, shallow Pliocene-to-Recent diapirs in the southern area are characterized by transparent to semi-transparent seismic facies with chaotic facies in their borders (Figs 7 and 8). Based on the 2D structures of the diapirs, they can be classified as domal or cylindrical types. Some diapirs reach the surface cropping out along fault escarpments (Fig. 8), while others are buried at shallow depth and in some cases developed collapse structures. Limits or contacts between the diapirs and the host rocks are diffused or sharp. Diffuse contacts are characterized by semi-transparent to chaotic seismic facies in the border of the diapirs. Sharp contacts, conversely, are in general subvertical and are commonly associated to high-angle faults (Fig. 7). The Pliocene-to-Recent host sediments consist of continuous reflections in a parallel-to-divergent, generally aggrading strata pattern. Lithoseismic units reach their maximum thickness in the marginal
troughs and pinch-out towards diapir highs (Figs 7 and 8). Maximum thickness is sometimes attained at the border of the diapirs, suggesting the development of growth faults (Fig. 7). In the marginal troughs, tbe reflectors present high to moderate amplitudes with parallel seismic facies that change laterally towards the diapirs to low amplitudes with semi-transparent facies. Near the diapirs, angular unconformities having onlap and/or toplap geometry change laterally to become conformities (Figs 7 and 8). Using local unconformities, the sedimentary deposits of late Pliocene to Quaternary age (Unit I) have been divided into several seismic sub-units, labelled PI and P2 (Pliocene) and Ql to Q3 (Quaternary) (Figs 7 and 8). The correlation between these sub-units (distinguished in the southern sector) and the sub-units established inside Unit I by Comas et al (1999) for the entire Alboran Sea, is shown in Figure 3. Folds observed in the high resolution seismic profiles are mainly related to the distribution of diapiric
450
A.R. TALUKDER ETAL.
Fig. 8. High resolution seismic profile (Ms 174) across a mud-diapir that includes the Marrakech mud-volcano, and draw-line interpretation. Note that the Marrakech mud volcano developed in the diapir flank, following a fault. Location of seismic line is shown in Fig. 4. Seismic units and unconformities as defined in Fig. 3.
highs, so that anticlines and synclines coincide with the diapiric highs and marginal troughs, respectively (e.g. Fig. 7). The upper Pliocene-Quaternary sequence presents a generally very gentle slope to the NW, decreasing congruently in thickness. MCS profiles image the complete characteristics of the middle Miocene to Recent mud diapirism in the WAB diapiric province. Plio-Quaternary diapiric structures occur as a formal continuation of the middle Miocene diapiric processes, on the top of former and deeper diapirs. MCS profiles show that mud volcanoes connect with deep diapir bodies from
Unit VI that developed on the top of the metamorphic basement (Comas & Soto 1999; Comas etal 1999). In MCS profiles, mud diapirs present chaotic seismic facies involving hyperbolic reflections and also show cylindrical (Fig. 9) and domal structures (Fig. 10) with a nearly flat or round top (see also Perez-Belzuz et al 1997). Cylindrical diapirs, in some cases, develop overhang geometries with inverted walls (Fig. 9). The Pliocene to Quaternary reflectors in the host sedimentary sequence show an onlapping, convergent strata pattern towards the diapirs. Figure 9
MUD DIAPIRISM AND VOLCANISM IN THE ALBORAN SEA
451
Fig. 9. MCS line ALB 27 across mud-diapirs in the northern WAB and draw-line interpretation. Collapse structures of different ages, differential diapir activation, and overhang associated with extension are observed. Location of seismic line is shown in Fig. 4. The location of J is shown in Figs 12 and 13. Seismic units and unconformities as defined in Fig. 3.
shows an asymmetric cylindrical diapir piercing the R3 reflector (base of Unit III, Tortonian in age). The basal Pliocene strata (just above the M reflector) are uplifted and deformed by the diapir in the southern flank, whereas in the northern one these sediments are undeformed and onlap the diapir residual high. Figure 10 shows an example of a domal diapir sealed by middle Miocene sediments, folding the
R3 reflector, and covered by undeformed PlioQuaternary sediments. Note that the M reflector may correspond to an erosion surface on top of the diapiric domes. The geometrical relationships between diapirs and host strata indicate that while some diapirs became inactive and are sealed by the R3 unconformity, others continued their ascension laterally, piercing the lowermost Pliocene sediments
452
A.R. TALUKDER ETAL
Fig. 10. MCS line EAS 30 across mud-diapirs in the northern WAB and simplified draw-line interpretation. Location of seismic line is shown in Fig. 4. The location of D is shown in Figs 12 and 13. Seismic units and unconformities as defined in Fig. 3. and intruding the overlying deposits. Collapse structures are common in many diapirs (often at various depths; see Fig. 6), suggesting that collapse processes most likely accompanied the entire diapir evolution in the WAB.
Shape of the mud diapirs The geometry of the top of the mud diapirs was mapped in both sectors on the basis of regularly spaced time slices from 200 ms downwards, using high resolution seismic and MCS profiles. In the southern sector, the geometry of the shallow diapirs (Fig. 11) piercing the Pliocene to Recent
sediments suggests that diapir rise induced noncylindrical folding, producing anticline culminations and encased synclines coinciding respectively with the diapir highs and troughs between them. The direction of these folds is approximately east-west (from ESE-WNW to ENE-WSW), with curvatures around the diapir highs and hinge lines plunging both east- and westwards (Fig. 11). In the northern sector, diapir culminations have two principal trends, which are broadly NW-SE and WSW-ENE (Fig. 12). The interference between these two perpendicular directions gave rise to an 'egg-box', non-cylindrical fold pattern. The domal diapirs have narrow and elongated culminations (C
MUD DIAPIRISM AND VOLCANISM IN THE ALBORAN SEA
Fig. 11. Contour map showing the 3D geometry of the Pliocene to Quaternary mud diapirs in the southern sector (southern WAB) and location of mud-volcanoes. Note that none of the mud volcanoes are situated in diapir culminations. Contour interval 200 msec twtt. Dashed lines indicate uncertain position of the contour line.
to G in Fig. 12a), whereas cylindrical diapirs have wider and irregular culminations (H to K in Fig. 12a). To analyse the Pliocene to Recent evolution of mud diapirism, a contour map of the M-reflector (Fig. 13) has been produced for the northern sector. It can be seen that none of the domal diapirs (C, F, and G in Figs 12 and 13) controls the current geometry of this unconformity. This is also particularly evident at point A, where a palaeo-channel coincides with a diapir culmination (Figs 12 and 13), indicating that this diapir became inactive before the Pliocene, being subsequently eroded during the lowermost Pliocene. The cylindrical diapirs nevertheless evidently remained active during the Pliocene and early Quaternary because they intrude in Unit I (point E; Fig. 13), also affecting the base of Unit la in places (point J in Figs 9 and 13). In these cases, diapirs were residual highs by the end of the Messinian and were affected by the erosion at the time of the M unconformity.
Discussion The mud diapir province developed in a region where there is a major sedimentary accumulation in the
453
Alboran Sea basin. The large diapir province involves sediments from the lowermost sequences of the sedimentary cover. Mud diapirism started in the middle Miocene and two distinctive major stages of activity have been identified, each one punctuated by several events of mud-flux activity (Perez-Belzuz et al 1991 \ Comas et al 1999; Comas et al 2000). The older, major stage of mud diapirism reflects the process of widespread crustal extension that affected the Alboran basin during the middle and late Miocene (between about 18 to 9 Ma ago), so mud diapirism occurred throughout the WAB depocentre for a long time. The limits of the mud diapir province run parallel to basement highs and are related to major extensional faults in the WAB (Fig. 2). Diapirism initiated on the bottom of basement halfgrabens. The main extension direction is perpendicular to the axes of the mud diapir province, consistent with a W-SW extension affecting the WAB at that time (Comas & Soto 1999; Comas et al 1999). The youngest stage of active diapirism, which is the specific subject of this paper, is post-Messinian and developed in concrete sectors of the diapir province. There is a significant asymmetry in the structural style between the southern and northern margins in the WAB hosting the mud diapir sectors studied here. Normal faulting and the development of progradational and aggradational units characterize the northern margin (Comas & Soto 1999). The southern margin, in contrast, is deformed up to the Recent by folding and strike-slip, normal and reverse faulting (Gensous et al 1986; Bourgois et al 1992; Maldonado et al 1992). During the late Miocene-toRecent, contractional deformation was more intense in the SW WAB, as shown by the close antiforms and strike-slip faults producing the Alboran ridge and Xauen bank (e.g. Bourgois et al 1992) (Fig. 2). The mud volcano features illustrate mud diapir mechanisms. The Marrakech mud volcano developed on the flank of a diapiric high following a fault (Fig. 7), indicating that it probably formed by fluid migration through the body of the diapir (Fig. 11). The Granada mud volcano shows, on the other hand, no direct relationship with any shallow diapir (Fig. 11). The existence of larger and abundant clasts on the crater of this volcano (seen on the underwater TV system during the B AS AC ALB cruise; Sautkin et al 2003) suggests that it was probably the result of the rise of diapiric material through faults from deeper diapirs. Mud volcanoes are commonly interpreted as the result of diapir extrusion primarily driven upwards by buoyancy forces or in response to the rapid flow of pore fluids up through a sedimentary mass, which becomes fluidized and entrained in the advecting fluids (Brown 1990; Milkov 2000). For the Granada mud volcano, in particular, the occurrence of a diapirically driven extrusion of a viscous flow capable of entraining larger
454
A.R.TALUKDERCTAL.
Fig. 12. (a) Contour map of the mud diapirs in the northern sector and structural interpretation (northern WAB; Fig. 4). Contour interval 200 msec twtt. (b) Shaded relief of the diapir surfaces. Labels (A to L) correspond to key diapir structures described in the text. Digital elevation of this surface has been illuminated from the N (000° azimuth and 43° of altitude or zenith distance) with a vertical exaggeration X 2. Same symbols as in Fig. 11.
MUD DIAPIRISM AND VOLCANISM IN THE ALBORAN SEA
455
Fig. 13. (a) Contour map of the M unconformity around diapirs in the northern sector and interpretation (northern WAB; Fig. 4). Contour interval 100 msec twtt. (b) Shaded relief of the M unconformity. Labels (A to L) correspond to key diapir structures shown in Fig. 12. Digital elevation of this surface has been illuminated from the N (000° azimuth and 43° of altitude or zenith distance) with a vertical exaggeration X 3. Other symbols as in Fig. 11.
456
A.R.TALUKDERCTAL.
and more abundant clasts can be advocated (e.g. Kopfetal. 1998). The southern mud volcano underwent several phases of activity from the Miocene to Recent, and periodic pulses (either intrusions or extrusions) producing interfmgering with the host sediments have mainly constructed it since the Messinian (Fig. 6). Continuous, undeformed reflectors in the overlying sedimentary sequence indicate the temporary plugging of this mud volcano. The positive seafloor expression could be produced by fluid seeps that are probably still active in the present day (Fig. 6). High backscatter intensity characterizes all the discovered mud volcanoes (Fig. 5), suggesting the absence of contemporary fluid escape or gas saturated mud-breccia (Ivanov et al 1996). In addition, most of the gravity cores taken in the Granada and Marrakech mud volcanoes were found covered by a thin drape of pelagic marls (Sautkin et al 2003). There is no evidence of fluid venting processes at the only gravity-core site lacking a pelagic blanket. All these circumstances suggest that the Granada and Marrakech mud volcanoes are most probably currently inactive. Mud diapir evolution occurred during the Pliocene to Recent simultaneously with sedimentation, determining the local uplift, subsequent erosion and thinning of the nearby coeval sedimentary units, producing local strata unconformities. Maximum thickness achieved at the border of the diapirs and the change in thickness around diapirs indicate that mud ascent followed existing fault surfaces. Sedimentary wedges, in the peripheral troughs, pinch out in opposite directions, evidencing the differential activation of the diapirs as well as their periodic ascent. The angular unconformities at the base of sub-units Ql and Q2 and the highly variable thickness of sub-unit Q2 (Figs 7 and 8) suggest that there were at least two major pulses of diapir rise in the southern sector during the late Pliocene and Quaternary. The studied data indicate that, in the northern sector, diapirism mostly ended during the lowermost Pliocene, evidenced by the fact that only two diapirs pierce the basal Pliocene strata (E and J in Fig. 13) and are sealed by upper Quaternary sediments (Fig. 9). Post-Messinian, cylindrical diapirs occurred in the interference between two sub-orthogonal diapir culminations (Fig. 12), thus indicating that subsurface mud migration is controlled by this structure and was driven towards maximum culminations (e.g. E in Fig. 12). The interference pattern reproduced by the shape of the diapir culminations suggests a triaxial strain formed by two coeval and orthogonal compressions during Plio-Quaternary diapirism. In the southern sector, the geometry of the Pliocene to Quaternary diapirs shows non-cylindrical folds with anticline culminations coinciding with the diapiric highs and encased synclines
between them, and is consistent with an approximately NW-SE compression during these times (Fig. 12). Regional data indicate that compression was also NW-SE during the late Tortonian, rotating to north-south until the middle Pliocene and henceforth to NNW-SSE during the rest of the Pliocene to Quaternary (Ott d'Estevou & Montenat 1985; Montenat et al 1987; De Larouziere et al 1988). This change of stress direction may have triggered the onset of active mud diapirism and volcanism in the WAB during the Pliocene to Quaternary. While some diapirs stopped rising in the PlioceneQuaternary and developed collapse structures in their tops, others resumed their ascent, with some reaching the seafloor. Volume loss by subsurface mud migration and fluid escape by extrusion are the most plausible causes for diapir collapse. Extension can be disregarded as the primary force for the Pliocene to Quaternary diapir collapse because this region was under compression at that time (see Comas et al 1999). Furthermore, the differential activation of the neighbouring diapirs supports the participation of subsurface mud migration in the diapirism which occurred roughly perpendicular (east-west; Figs 11 and 12) to the regional, maximum compressive stress (NNW—SSE). Different lithostatic pressures may also have caused the different Pliocene to Recent evolution of the mud diapirism noted in the two studied sectors. In the southern sector, near the deeper trough of the west Alboran Sea, the water column is 1200 m and the Pliocene to Recent sedimentary cover is around 1500 m thick, while in the northern sector they are about 800 m and 500 m, respectively. Based on these data and in addition to tectonic forces, a larger confining pressure would have enhanced greater development of the Plio-Quaternary mud diapirism in the southern WAB. Widespread collapse structures throughout the diapir evolution, recognized even within the older Miocene diapirs, indicate a continuous loss of fluidpressure in the diapiric system over time. The present overpressure of the diapiric material observed in logging data from Unit VI in commercial wells (Jurado & Comas 1992) suggests that active mud diapirs and mud volcanoes could occur at present. The WAB thus deserves to be surveyed in the future looking for possible active mud diapirs and volcanoes, especially at the central and southern areas of the WAB. We propose that pulses of contractional deformation, coexistent with continuous subsidence and sediment overburden in the basin, may have repeatedly triggered mud diapirism during the Pliocene to Recent. Pliocene to Quaternary mud diapirs and related structures are found to be generally consistent with the noticeable submeridian contraction and roughly east-west transtension in the Alboran basin,
MUD DIAPIRISM AND VOLCANISM IN THE ALBORAN SEA which was coeval to the general uplift at the basin margins (i.e. the Betics and the Rif mountain belt) from the Pliocene on.
Conclusions (1)
(2)
(3)
(4)
Multi-channel seismic (MCS) profiles image the complete characteristics of the middle Miocene to Recent mud diapirism in the West Alboran basin (WAB). Plio-Quaternary diapiric structures occur as a formal continuation of the middle Miocene diapiric processes, on top of former and deeper diapirs. Mud volcanoes connected with deep diapir bodies from Unit VI developed on the top of the metamorphic basement and initiated in half-grabens. During the Pliocene and Quaternary, NNWSSE regional compression was the main driving force triggering mud diapirism and volcanism in the WAB. During this period, two major pulses of diapiric rise are distinguished which resulted in active diapirism and mud volcanoes. Mud volcanoes, probably currently inactive, have been discovered in the SW WAB. They developed at the flank of diapir highs as a consequence of fluid migration and rose through the diapiric body (Marrakech mud volcano) and/or through faults and fractures possibly connected with deeper diapirs. During the Pliocene and Quaternary, diapir collapse coexisted in places with diapir ascent. Subsurface mud migration and fluid escape by extrusion produced a volume loss and subsequent collapse in the diapir. In the southern WAB and during the Quaternary, subsurface east-west mud migration was roughly perpendicular to the maximum compressive stress (NNW-SSE). In the northern WAB, subsurface mud migration seems to have been controlled by the interference between orthogonal anticlines induced by diapir rise.
Financial support for this work was provided by the Spanish Project REN2001-3868-CO3 (MCYT). We thank A. Kopf and L. Pinheiro for their thoughtful comments and criticism. A. Maltman is specially thanked for his editorial comments and advice. We thank also C. Laurin for her careful and detailed linguistic revision of the final manuscript.
References AIT BRAHIM, L. & CHOTIN, P. 1989. Genese et deformation desequilibrium bassins neogenes du Rif Central (Maroc) au cours du rapprochement Europa-Africa. GeodinamicaActa, 3,295-304.
457
ALVAREZ-MARRON, J. 1999. Pliocene to Holocene structure of the Eastern Alboran Sea (Western Mediterranean). In: ZAHN, R., COMAS, M. C. & KLAUS, A. (eds) Proceeding Ocean Drilling Program, Scientific Results, 161,345-355. BARBER, A.J., TJOKROSAPOETRO, S. & CHARLTON, T.R. 1986. Mud Volcanoes, shale diapirs, wrench faults, and melanges in accretionary complexes, Eastern Indonesia. American Association of Petroleum Geologists Bulletin, 70,1729-1741. BIJU-DUVAL, B., LE QUELLEC, P., MASCLE, V., RENARD, V. & VALERY, P. 1982. Multibeam bathymetric survey and high-resolution seismic investigation on the Barbados Ridge Complex (Eastern Caribbean): a key to the knowledge and interpretation of an accretionary wedge. Tectonophysics, 86,275-304. BISHOP, R.S. 1978. Mechanism for emplacement of piercement diapirs. American Association of Petroleum Geologists Bulletin, 62,1561-1583. BOURGOIS, J., MAUFFRET, A., AMMAR, A. & DEMNATI, A. 1992. Multichannel seismic data imaging of inversion tectonics of the Alboran Ridge (Western Mediterranean Sea). Geo-Marine Letters, 12,117-122. BROWN, K.M. 1990. The nature and hydrogeologic significance of mud diapirs and diatremes. Journal of Geophysical Research, 95, 8969-8982. CAMERLENGHI, A., CITA, M. B., HIEKE, W. & RICCHIUTO, T. 1992. Geological evidence for mud diapirism on the Mediterranean Ridge accretionary complex. Earth and Planetary Science Letters, 109,493-504. CAMPILLO, A.C., MALDONADO, A. & MAUFFRET, A. 1992. Stratigraphic and tectonic evolution of the Western Alboran Sea: Late Miocene to Recent. Geo-Marine Letters, 12,165-172. CHALOUAN, A., SAJI, R., MICHARD, A. & BALLY, A.W. 1997. Neogene tectonic Evolution of the Southwestern Alboran Basin as inferred from seismic Data off Morocco. American Association of Petroleum Geologists Bulletin, 81,1161-1184. CITA, M.B. & McKENzm, J.A. 1986. The terminal Miocene event. In: Hsu, K.J. (ed.) Mesozoic and Cenozoic Oceans. American Geophysical Union, Geodynamic Series, 15,123-140. CITA, M.B., RYAN, W.B.F. & PAGGI, L. 1981. Prometheus mud breccia. An example of shale diapirism in the western Mediterranean Ridge. Annales Geologicae des Pays Hellenides, 13, 37-49. CITA, M.B., ERBA, E., LUCCHI, R., POTT, M., VAN DER MEER, R. & LIETO, L. 1996. Stratigraphy and sedimentation in the Mediterranean Ridge diapiric belt. Marine Geology, 132,131-150. COHEN, H.A. & McCLAY, K. 1996. Sedimentation and shale tectonics of the northwestern Niger Delta front. Marine & Petroleum Geology, 13, 313-328. COMAS, M.C., & SOTO, J.I. 1999. Brittle deformation in the metamorphic basement at Site 976: implications for Middle Miocene extensional tectonics in the Western Alboran Basin. In: ZAHN, R., COMAS, M.C. & KLAUS, A. (eds) Proceeding Ocean Drilling Program, Scientific Results, 161,331-344. COMAS, M.C., GARCIA-DUENAS, V. & JURADO, M.C. 1992. Neogene tectonic evolution of the Alboran Basin from MCS data, in Alboran Sea. Geo-Marine Letters, 12, 157-164.
458
A.R. TALUKDER ETAL.
COMAS, M.C., ZAHN, R., KLAUS, A. et al 1996. Site 976. Proceeding: Ocean Drilling Program, Initial Reports, 161,1023pp. COMAS, M.C., PLAIT, J.P, SOTO, J.I. & WATTS, A.B. 1999. The origin and tectonic history of the Alboran basin: insights from Leg 161 results. In: ZAHN, R., COMAS, M.C. & KLAUS, A. (eds) Proceeding Ocean Drilling Program, Scientific Results, 161, 555-579. COMAS, M.C., SOTO, J.I. & BASACALB cruise (TTR-9 Leg 3) Scientific Party 2000. A tectonic overview of mud diapirs and related mud volcanoes in the Alboran Basin. In: COMAS, M.C. & AKHMANOV, G.G. (eds) Geological Processes on European continental margins, Intergovernmental Oceanographic Commission (UNESCO), Workshop Report No. 168, 29-30. CRONIN, B.T., IVANOV, M.K., LIMONOV, A.R, EGOROV, A., AKHMANOV, G.G., AKMETJANOV, A.M. & KOZLOVA, E. 1997. Shipboard Scientific Party TTR-5, new discoveries of mud volcanoes on the Eastern Mediterranean Ridge. Journal of the Geological Society, London, 154,173-182. DE LAROUZTERE, F.D., BOLZE, J., BORDET, P., HERNANDEZ, J., MONTENAT, C. & OTT D'ESTEVOU, P. 1988. The Betic segment of the lithospheric trans-Alboran shear zone during the Late Miocene. Tectonophysics, 152, 41-52. DEWEY, J.F., HELMAN, M.L., TURCO, E., HUTTON, D.H.W. & KNOTT, S.D. 1989. Kinematics of the Western Mediterranean. In: COWARD, M.P., DIETRICH, D. & PARK, R.G. (eds) Alpine Tectonics. Geological Society, London, Special Publications, 45,265-283. FOWLER, S.R., MILDENHALL, J., ZALOVA, S., RILEY, G., ELSLEY, G., DESPLANQUES, A. & GULIYEV, F. 2000. Mud volcanoes and structural development on Shah Deniz. Journal of Petroleum Science and Engineering, 28,189-206. Fusi, N. & KENYON, N.H. 1996. Distribution of mud diapirism and other geological structures from long range sidescan sonar (GLORIA) data, in the Eastern Mediterranean Sea. Marine Geology, 132, 21-38. GARCIA-DUENAS, V., BALANYA, J.C. & MARTINEZMARTINEZ, J.M. 1992. Miocene extensional detachments in the outcropping basement of the northern Alboran Basin (Betics) and their tectonic implications. Geo-Marine Letters, 12, 88-95. GARDNER, J. 2001. Mud volcanoes revealed and sampled on the Western Moroccoan continental margin. Geophysical Research Letters, 28, 339-342. GENSOUS, B., TESSON, M. & WINNOCK, E. 1986. La marge meridionale de la Mer d'Alboran: caracteres structuro-sedimentaires et evolution recente. Marine Geology,72,34l-310. HEDBERG, H.D. 1974. Relation of Methane generation to undercompacted shales, Shale diapirs, and Mud volcanoes. American Association of PetroleumGeologists Bulletin, 58,661-673. HIGGINS, G.E. & SAUNDERS, J.B. 1974. Mud volcanoes, their nature and origin. Verhandlungen der Naturforschenden Gesellschaft Basel, 84, 101-152. HOVLAND, M. & CURZI, P.V. 1989. Gas seepage and assumed mud diapirism in the Italian central Adriatic Sea. Marine and Petroleum Geology, 6, 161-169. HOVLAND, M., HILL, A. & STOKES, D. 1997. The structure
and geomorphology of the Dashgil mud volcano, Azerbaijan. Geomorphology, 21,1-15. IVANOV, M.K., LIMONOV, A.F., VAN WEERING, TJ. C.E. 1996. Comparative characteristics of the Black Sea and Mediterranean Ridge mud volcanoes. Marine Geology, 132, 253-27'1. JACKSON, M.P.A. & VENDEVILLE, B.C. 1994. Regional extension as a geological trigger for diapirism. Geological Society of America Bulletin, 106, 57-73. JURADO, MJ. & COMAS, M.C. 1992. Well log interpretation and Seismic character of the Cenozoic sequence in the Northern Alboran Sea. Geo-Marine Letters, 12, 129-136. KOPF, A., ROBERTSON, A.H.F., CLENNELL, M.B. & FLECKER, R. 1998. Mechanisms of mud extrusion on the Mediterranean Ridge accretionaiy complex. GeoMarine Letters, 18, 97-114. KOPF, A., ROBERTSON, A.H.F. & VOLKMANN, N. 2000. Origin of mud breccia from the Mediterranean Ridge accretionary complex based on evidence of the maturity of organic matter and related petrographic and regional tectonic evidence. Marine Geology, 166, 65-82. LIMONOV, A.F., WOODSIDE, J.M., CITA, M.B. & IVANOV, M.K. 1996. The Mediterranean Ridge and related mud diapirism: a background. Marine Geology, 132, 7-19. LIMONOV, A.F., VAN WEERING, TJ. C.E., KENYON, N.H., IVANOV, M.K. & MEISNER, L.B. 1997. Seabed morphology and gas venting in the Black Sea mud volcano area: Observations with the MAK-1 deeptwo sidescan sonar and bottom profiler. Marine Geology, 137, 121-136. MALDONADO, A., CAMPILLO, A.C., MAUFFRET, A., ALONSO, B., WOODSIDE, J. & CAMPOS, J. 1992. Alboran sea late Cenozoic tectonic and stratigraphic evolution. GeoMarine Letters, 12, 179-186. MARTIN, J.B. & KASTNER, M. 1996. Chemical and isotopic evidence for sources of fluids in a mud volcano field seaward of the Barbados accretionary wedge. Journal of Geophysical Research, 101, 20325-20345. MARTINEZ-MARTINEZ, J.M., SOTO, J.I. & BALANYA, J.C. 1997. Crustal decoupling and intracrustal flow beneath domal exhumed core complexes, Betics (SE Spain), Terra Nova, 9, 223-227. MAZZOLI, S. & HELMAN, M. 1994. Neogene patterns of relative plate motion for Africa-Europe: some implication for recent central Mediterranean tectonics. Geologische Rundschau, 83,464-468. MEGHRAOUI, M., MOREL, J.L., ANDRIEUX, J. & DAHMANI, M. 1996. Tectonique plio-quaternaire de la chame tello-rifaine et de la mer d'Alboran. Une zone complexe de convergence continent-continent. Bulletin de la Societe Geologique de France, 167, 141-157. MILKOV, A.V. 2000. Worldwide distribution of submarine mud volcanoes and associated gas hydrates. Marine Geology, 167, 29-42. MONTENAT, C., OTT D'ESTEVOU, P. & MASSE, P. 1987. Tectonic sedimentary character of the Betic Neogene basins evolving in a crustal transcurrent shear zone (SE Spain). Bull.Centres Rech. Explor-Prod. ElfAquitaine, 11-1,1-22. MONTENAT, C., OTT D'ESTAVOU, P., DE LAROUZIERE, F.D. & BEDU, P. 1992. Originalite geodynamique des bassins
MUD DIAPIRISM AND VOLCANISM IN THE ALBORAN SEA Neogenes du domain Betique orientale. Institut Geologique Albert de Lapparent, Notes et Memoires, 1-21,11-49. MOREL, J.L. 1989. Etats de contrainte et cinematique de la chaine rifaine (Maroc) du Tortonien a 1'actuel. GeodinamicaActa, 3,283-294. MOREL, J.L. & MEGHRAOUI, M. 1996. Goringe-AlboranTell tectonic zone: A transpresion system along the Africa-Eurasia plate boundary. Geology, 24,755-758. MORLEY, C.K. & GUERIN, G. 1996. Comparison of gravity driven deformation styles and behaviour associated with mobile shales and salt. Tectonics, 15, 1154-1170. OTT D'ESTEVOU, P. & MONTENAT, C. 1985. Evolution structurale de la zone betique orientale (Espagne) du Tortonian a 1'Holocene. Comptes Rendus de lyAcademic des Sciences de Paris, 300, 363-368. PEREZ-BELZUZ, E, ALONSO, B. & ERCILLA, G. 1997. History of mud diapirism and trigger mechanism in the Western Alboran Sea. Tectonophysics, 282, 399-422. REED, D.L., SILVER, E.A., TAGUDIN, I.E., SHILEY, T.H. & VROLIJK, P. 1990. Relations between mud volcanoes, thrust deformation, slope sedimentation and gas hydrate, offshore north Panama. Marine and Petroleum Geology, 1,44-54. ROBERTSON, A. & SHIPBOARD SCIENTIFIC PARTY 1996. Mud volcanism in the Mediterranean Ridge: Initial results of ODP Leg 160. Geology, 24,239-242. RYAN, W.B.F., Hsu, 1.K.etal. (eds) 197'3. Initial Reports of the Deep Sea Drilling Project, 13,1-514. Washington (US Government Printing Office). SAUTKIN, A. 2000. Micropaleontological investigation of rock clasts from mud volcanic deposits in the Alboran Sea. In: COMAS, M.C. & AKHMANOV, G.G. (eds) Geological Processes on European continental margins. Intergovernmental Oceanographic Commission (UNESCO), Workshop Report No. 168,20. SAUTKIN, A., TALUKDER, A.R., COMAS, M.C., SOTO, J.I. & ALEKSEEV, A. 2003. Mud volcanoes in the Alboran Sea: evidence from micropaleontological and geophysical data. Marine Geology, 195,237-261. SHIPBOARD SCIENTIFIC PARTY 1996. Site 976. In: COMAS, M.C., ZAHN, R. KLAUS, A. & SHIPBOARD SCIENTIFIC PARTY (eds) Proceeding: Ocean Drilling Program, Initial Reports, 161,179-297.
459
SOTO, J.L, COMAS, M.C. & DE LA LINDE, J. 1996. Espesor de sedimentos en la cuenca de Alboran mediante una conversion sismica corregida. Geogaceta (Geological Society of Spain), 20,382-385. SRIVASTAVA, S.P., SCHOUTEN, H., ROEST, W.R., KLITGORT, K.D., KOVACS, L.C., VERHOEF, J. & MACNAB, R. 1990. Iberian plate kinematics: A jumping plate boundary between Eurasia and Africa. Nature, 344,756-759. TANDON, K., LORENZO, J.M. & DE LA LINDE, J. 1998. Timing of rifting in the Alboran Sea basin- correlation of borehole (ODP Leg 161 and Andalucia A-l) to seismic reflection data: implications for basin formation. Marine Geology, 144,275-294. VAN RENSBERGEN, P., MORLEY, C.K., ANG, D.W., HOAN, T.Q. & LAM, N.T. 1999. Structural evolution of shale diapirs from reactive rise to mud volcanism: 3D seismic data from the Baram delta, offshore Brunei Darussalam. Journal of the Geological Society, London, 156,633-650. VENDEVELLE, B.C. & JACKSON, M.P.A. 1992a. The fall of diapirs during thin-skinned extension. Marine and Petroleum Geology, 9,331-353. VENDEVILLE, B.C & JACKSON, M.P.A. 19926. The rise of diapirs during thin-skinned extension. Marine and Petroleum Geology, 9,354-371. VERNETTE, G., MAUFFRET, A., BOBIER, C., BRICENO, L. & GAYET, J. 1992. Mud diapirism, fan sedimentation and strike-slip faulting, Caribbean Colombian Margin. Tectonophysics, 202, 335-349. VOLGIN, A.V. & WOODSIDE, J.M. 1996. Sidescan sonar images of mud volcanoes from the Mediterranean Ridge: Possible causes of variations in backscattering intensity. Marine Geology, 132,39-53. WATTS, A.B., PLATT, J.P & BUHL, P. 1993. Tectonic evolution of the Alboran Sea Basin. Basin Research, 5, 153-177. WESTBROOK, G.K. & SMITH, M.J. 1983. Long decollements and mud volcanoes: Evidence from the Barbados Ridge complex for the role of high pore-fluid pressure in the development of an accretionary complex. Geology, 11,279-283. WIEDICKE, M., NEBEN, S. & SPIESS, V. 2001. Mud volcanoes at the front of the Makran acretionary complex. Marine Geology, 172,57-73.
This page intentionally left blank
The role of shear stress in mobilizing deep-seated mud volcanoes: geological and geomechanical evidence from Trinidad and Taiwan NAJWAYASSIR Shell SIEP, Volmerlaan 8, Postbus 60, Rijswijk 2280 AB, The Netherlands Work carried out at University College London, Gower Street, London WC2, UK Abstract: Deep-seated mud volcanoes are observed in a variety of geological settings, which has led to considerable debate on their origin. This paper summarizes the geological features common to mud volcanoes around the world and possible mechanisms of their extrusion. Field and laboratory data from Trinidad and Taiwan are discussed to assess the possible sources and causes of the volcanoes. A close association between mud volcanoes and compressional tectonics leads to the conclusion that tectonic activity plays an important part in mud volcano development. Experimental data are presented to explain the role of tectonic activity and the association between shear stresses and mud volcanoes. It is demonstrated that shear stresses applied to low permeability sediments can produce a dramatic increase in pore pressure and can cause sediment flow. This is proposed as one possible contributing cause of mud volcanoes.
Deep-seated mud volcanoes are observed world-wide both onshore (see for example Jakubov et al 1971; McManus & Tate 1986; Dia et al 1999) and offshore (Milkov 2000; Kopf 2002). After over a century of study, they remain one of the most interesting and enigmatic sedimentary phenomena. They are a surface expression of overpressured sediments at depth, which has long made them recognized as a drilling hazard (Kugler 1933). They play an important role in dewatering thick young sedimentary sequences (Westbrook & Smith 1983). They are strongly associated with hydrocarbons, and the recent observations of gas hydrates around offshore mud volcanoes are attracting renewed interest in their origin (Milkov 2000). This paper looks at two onshore mud volcano areas: southern Trinidad and SW Taiwan, both of which occur in tectonically active regions, in association with thick, rapidly deposited Tertiary sediments. Field and laboratory data are presented here to investigate the properties of the ejected muds, and the role of shear stress in triggering the mud volcanic activity.
Origin of mud volcanoes: a brief review Deep-seated mud volcanoes around the world vary dramatically in shape and size (Shih 1967), from a pool of water and mud to a mountain-like cone, from a width of a few centimetres to several kilometres. However, they generally share several common geological features. (1)
The source is always a deep, thick (several kilometres), rapidly deposited regressive clay/ shale sequence. The clays are generally of
(2)
(3)
(4)
Tertiary age in most areas, although older examples are known (Jakubov et al 1971). They are associated with hydrocarbons and especially methane (Hedberg 1974). Where extruded gases have been analysed, methane has made up most of the gaseous emission of the mud volcano (Shih 1967; Higgins & Saunders 1974; Dia et al 1999; Deville et al 2003). The appearance of offshore mud volcano islands is always reported to be associated with vigorous bubbling of the seawater and massive gas explosions (see for example Kugler 1965; Higgins & Saunders 1967). Blocks are often found in the mud matrix of a volcano. They can be older than the matrix, which has led to different theories as to their origin. Some believe that the mud volcanic material is olistostromal in origin (Higgins & Saunders 1974). Other workers believe that blocks are incorporated from the overburden (see for example Barber et al 1986; Deville et al 2003). They have a very clear tectonic association. Deep-seated mud volcanoes onshore and offshore are found mostly at active margins (Higgins & Saunders 1974; Milkov 2000). They often erupt after a large earthquake, for example in New Zealand (Ridd 1970) and in Kobystan (Jakubov et al 1971). Mud volcanic islands are also known to appear suddenly after a major earthquake, for example in the Arabian Gulf (Sondhi 1947) and in Sabah (McManus & Tate 1986). However, several mud volcanoes are also reported at passive margins. Examples include the Gulf of Mexico (MacDonald et al 2000), the Niger Delta (Graue 2000) and Barents Sea (Hjelstuen et al 1999); see also
reviews by Milkov (2000) and Kopf (2002). Regardless of tectonic system, mud volcanoes generally occur along fold axes and are most commonly associated with faults. They are also often, but not always, associated with shale diapirs (Brown 1990) and sand dykes (Kuglerl933). In close association with the above factors, the source is always overpressured. Overpressuring is the condition whereby interstitial fluid pressure exceeds hydrostatic pressure. This can occur by any or a combination of the following factors: rapid sedimentary loading, a fluid source at depth (e.g. hydrocarbon generation, smectite dehydration, etc.) or tectonic shearing, each having an important effect on sediment rheology and porosity (Yassir & Addis 2002). There is general agreement that rapid loading and undercompaction is important in preserving the low density of mud volcano source sediments.
The above conditions are intricately inter-related: thick sedimentary sequences are deposited in active and passive margins and undergo significant deformations with basin evolution. Hydrocarbon maturation and possible clay mineral transformation at depth further contribute to the pressure in the sediment. Based on these conditions, several mechanisms for mud volcanism are proposed in the literature (see reviews by Bishop 1978; Milkov 2000; Kopf 2002). It is generally agreed that not one mechanism is responsible for all mud volcano occurrences around the world and that more than one mechanism can be active in one area. The three principal mechanisms, density inversion, gas emission and tectonism, are briefly outlined below. In the density inversion model, a high-density sediment is deposited over a low density sediment and the instability triggers intrusion of the lower density sediment into its overburden (Biot & Ode 1965; Chapman 1974). This is proposed by Morgan et al (1968) for the mudlumps of the Mississippi delta - shallow clay diapirs piercing a ductile overburden. In the case of deep-seated mud volcanism, low density is not enough to effect intrusion into a stiff overburden without the presence of a fracture (Bishop 1978). More realistically, the low density overpressured sediment escapes to the surface by fracturing its overburden or re-opening pre-existing faults and fractures. Kugler (1933, 1965) regarded methane as the driving mechanism for all sedimentary volcanism. It is possible to see the importance of methane in overpressuring and mobilizing sediments. Methane is not only an overpressured fluid at depth, it expands as it rises and comes out of solution, giving the explosive energy needed to transport the mud and
blocks upwards from great depth (Kugler 1965; Brown 1990). This mechanism cannot be ignored as an important controlling factor. The association between sedimentary volcanism and active margins leads many workers to suggest tectonic compression as a mechanism for mud volcanoes (see for examples Higgins & Saunders 1974; Westbrook & Smith 1983; Barber et al. 1986; Yassir 1987,1989). Active margins are associated with high sedimentation rates, subsidence, uplift, overthrusting, folding and faulting, in close association with temperature and fluid pressure increase and hydrocarbon maturation. One of the many important elements in such an environment is the high shear stresses generated by tectonic activity. These compress the soft sediments, which respond by flowing into structural highs and faults and/or 'hydraulically' fracturing the overburden. High shear stresses need not only be present in active margins. The thickness, shape and evolution of a sedimentary wedge on a passive margin will create significant shear stresses that can play an important role in sediment mobilization. Examples from Trinidad and Taiwan are used in this paper to address the influence of high shear stresses on overpressuring the sediment and triggering flow. The two mud volcano areas are described below. Experimental data are then presented to describe the geomechanical and mineralogical characteristics of the mud and its liquefaction potential under high shear stresses.
Trinidad mud volcanoes Trinidad is situated at the junction between the North and South American plates and the subducting Atlantic ocean floor (Bertrand & Bertrand 1985). The structure of the island trends ENE-WSW. The mud volcanoes are found south of the Central Range (Figs 1 and 2). Four are associated with the Naparima Fold Belt, made up of highly folded Tertiary sands and shales. The remaining volcanoes are found in the Southern Basin and the Southern Range Anticline, which is folded into small anticlines cut by faults (Figs 1 and 2). The SW of the island is offset by a major strike-slip fault, the Los Bajos Fault (Fig. 1), which offsets all pre-existing structures. The East Venezuelan mud volcanoes lie on the same structural trends extending westwards from Trinidad (Barr & Saunders 1965; Higgins & Saunders 1974; Carr-Brown & Frampton 1979; see also Devilled al. 2003). The sources of the Trinidad mud volcanoes are thought to be mid-Tertiary (Miocene), rapidly deposited thick shale sequences, the Nariva (3 km), Karamat (>1 km) and Lower Cruse-Lengua Formations (2 km). They all contain major olistostrome sequences, with massive blocks aged
THE ROLE OF SHEAR STRESS IN MUD VOLCANOES
463
Fig.l. Map of southern Trinidad showing structure and distribution of mud volcanoes. Numbered mud volcanoes are
referred to in the text (modified after Higgins & Saunders 1974).
Cretaceous to Lower Miocene and signs of slumping, interpreted to be caused by deposition on unstable structural flanks (Higgins & Saunders 1974). They are overpressured, showing some low density and velocity anomalies (Higgins & Saunders 1974). The Nariva and Karamat Formations are known to cause drilling problems: the Nariva clay sometimes oozes out like toothpaste, whereas the Karamat is highly slickensided and shattered (Frampton & CarrBrown, pers. comm. 1986). In fact, the Karamat Formation shows a variable thickness in the order of 100 m to over 1000 m, which has been attributed to flowage (Higgins & Saunders 1974). Geochemical analyses of the fluids emitted by the mud volcanoes suggest that they originate from depths exceeding 3 km (Dia et al. 1999), corresponding with the Miocene clays. Deville et al (2003) agree with this conclusion for the offshore mud volcanoes but suggest a Cretaceous origin for the onshore mud volcanoes. This is further discussed in a later section. The association between structure and mud volcanoes is evident in Trinidad. The majority of the mud volcanoes occur in the Southern Range Anticline area (Figs 1 and 2) which extends offshore into Erin Bay to the SW of the island. Mud volcanic islands have been reported in association with the anticline axis (Higgins & Saunders 1967, Deville et al this volume). Two or three mud volcanoes are associated with the Los Bajos Fault system onshore (Fig. 1). Large scale mud flows have been observed along this fault as well as other wrench faults seen on seismic sections taken offshore to the west of Trinidad. The
majority of mud volcanoes in Trinidad extrude along a thrust fault or a strike-slip fault (Fig. 1; Higgins & Saunders 1974); many are associated with sand and clay dykes (Kugler 1933). There is fossil evidence for mud volcanic activity in Trinidad. The Miocene Moruga Formation, which overlies the current overpressured zones, has a maximum thickness of 3 km and its deposition seems to be associated with almost uninterrupted sedimentary volcanism. Kugler (1933) describes 'great masses of mudflow interbedded between the Moruga beds' and 'numerous dykes along or parallel to cross-faults or irregularly injected into shattered zones'. The Forest Reserve Field mud volcano in southern Trinidad (numbered 7 in Fig. 1) shows mud flows containing large angular blocks interbedded with and injected into, Miocene sediments. The author visited and sampled six active mud volcanoes in Trinidad: (1), Piparo (2), Digity (3), Devil's Woodyard (4), Palo Seco (5), Anglais Point and (6) Lagon Bouffe (numbered in Fig. 1). These mud volcanoes have been extensively described in the literature (Higgins & Saunders 1974; Dia et al 1999; Deville et al 2003). The mud volcanoes show highly varied shapes (cones, pools) and sizes (centimetres to hundreds of metres), even within an individual mud volcano area. However, they all have an ambient temperature, a distinct 'oily' smell and where the crater is active, methane gas. Exotic blocks have been observed in some of the areas, most notably in Anglais Point where some small clasts are seemingly identical to the mud matrix this is analysed below.
464
N.YASSIR
Fig. 2. Schematic north to south section of southern Trinidad (modified after Barr & Saunders 1965).
Taiwan mud volcanoes Taiwan falls on an island arc/continent collision zone on the western edge of the Pacific plate, which is causing the island to rise at the rate of 4 cm/year (Hsieh 1987, pers. comm.). It is the site of geosynclinal deposition of over 10 km of Tertiary sediments on a metamorphic basement (Ho 1982). The structure of the island follows its longest axis, trending NNE-SSW and can be divided into three provinces, separated by long upthrust faults extending the whole length of the island: the Western Foothills, the Central Range (the 'backbone' of Taiwan) and the Coastal Range provinces (Ho 1982). There are two mud volcano areas in Taiwan: to the SE (north of Taitung) and to the SW (Tainan region), Figure 3. The majority of the mud volcanoes are situated in the Tainan area (Shih 1967). The SE mud volcanoes occur within the Coastal Range province, underlain by Neogene sediments. These are generally abundant in volcanic deriva-
tives, including poorly sorted elastics and melanges. The sediments have been involved in large-scale gravity sliding and low angle nappes. The source of the mud volcanoes in this region is the Lichi Formation - a chaotic sheared argillaceous melange of Plio-Pleistocene age, with ophiolite and sandstone blocks, reaching a thickness of at least 1600 m (Shih 1967). The mud volcanoes in the SW occur in the Western Foothills, where 8 km of sands and shales have been deposited in the Neogene (Ho 1982). In the SW coastal plain, these deposits are tilted gently eastward, becoming gently folded and breaking off along thrust faults to the east (Fig. 4). The source of the mud volcanoes is the Pliocene Gutingkeng Formation, which reaches a thickness of over 5 km in the Tainan area (Shih 1967; Chou 1971; Hsieh 1972). The Lower Gutingkeng Formation is a massive blue grey mudstone, with interbedded siltstone and greywacke showing graded bedding (Hsieh 1972). It is associated with olistostromes and
THE ROLE OF SHEAR STRESS IN MUD VOLCANOES
465
Fig. 3. Schematic map of the Tainan area showing the location of mud volcano areas and associated structure. The numbered mud volcanoes are referred to in the text. The insert map of Taiwan shows the structural provinces, and the location of the cross section in Fig. 4.
sandpipes (Chou 1971). It crops out east of the mud volcano area, forming 'badlands' - barren eroded topography caused by the high salt content in the shale; large blocks are observed in the shale in outcrop (Chou 1971). The Upper Gutingkeng Formation is characterized by bluish grey sandy mudstones and abundant foraminifera and molluscs (Chou 1971). There is uncertainty about which part of the Pliocene is the source of the mud volcanoes. Micropalaeontological studies of the muds show a variety of Pliocene aged microfossils, as well as some from the Quaternary (Lin 1965; Chou 1971). Geochemical analyses of the fluids in the muds suggest a deep origin (Gieskes et al 1992). The association between mud volcanoes and structure is again evident in Taiwan. All mud volcanoes occur along reverse fault lines parallel to the regional structure of Taiwan (Fig. 3), the fault movement being in a NW direction (Shih 1967). Mud volcanoes have also been located offshore to the SW, along the same deformation front of their onshore counterparts (Huang et al 1992). Both mud volcano areas, and especially the Coastal Range province in the SE, occur in a region of significant seismic activity (Ho 1982). Two diapiric structures in the Tainan area are composed of >3.5 km of Lower Gutingkeng shale, which shows no dip or seismic signature; the shale is associated with drilling problems (Hsieh
1972). These diapirs are both elongate and parallel to the regional structural trend. Four mud volcano areas were visited and sampled: Aoshenshui (1,2), Wushanting (3), Chienchuliao (4) and Hsiaokungshui (5,6)- all found in the Tainan area (Fig. 3). A similar variation in mud volcano shapes was observed here as in Trinidad, although no large examples are known in Taiwan. All the volcanoes showed vigorous gas emissions and oil, and recorded ambient temperatures; Chienchuliao was earlier reported to have a temperature of 41°C (Shih 1967).
Characteristics of mud volcanic material This section describes routine analyses of the sampled muds from both mud volcano areas, including grain size analysis, Atterberg limits, clay mineralogy and microscopic characterization of the grains. The results are briefly discussed in the light of the origin and in situ state of the mud. Grain size distribution The analysis was carried out on the <425jjim fraction using the British Standards method BS 1377 (1975). This limit was chosen because it separates
466
N.YASSIR
Fig. 4. Cross section through SW Taiwan, location in Fig. 3 (modified after Ho 1982).
out the larger particles, some of which might be incorporated overburden material. The majority of the samples contained between 0-4% of the larger fraction, with the exception of Piparo (11%) and Anglais Point (8%). The results of the grain size analysis are presented in Figure 5. The Taiwan samples are very homogeneous (T1-T6, locations numbered in Fig. 3). The Trinidad samples, on the other hand, show considerable variation. Significant differences were observed between the two Digity samples (Dig 1,2), but not so much with the three Palo Seco samples (PS 1, 2 and 3). The Lagon Bouffe (LB) sample was by far the most sand-rich volcano and the lowest in clay fraction. All samples and especially from Trinidad, have a grain size distribution which reflects poor sorting. This could be from the parent rock or could also reflect contamination of the mud by the overburden material. Also plotted in Figure 5 are the grain size analyses performed on two Lower Cruse Formation core samples (LCF1 = 1606 m, LCF2 = 1543 m) and one Nariva Formation core (NAR = 940 m) taken from three wells drilled in southern Trinidad (Kerr et al. 1970). The Lower Cruse Formation, thought to be the source of the Palo Seco and Anglais Point mud volcanoes, seems to be finer grained. On the other hand, the Nariva Formation, thought to be the source of Piparo, is somewhat coarser grained, with around 10% less clay fraction. None of the core samples contained sand. Atterberg limits The Atterberg limit tests are a geomechanical characterization tool used to assess the moisture content
required to allow a sediment to change from a brittle to a plastic state (plastic limit, PL) or from a plastic to a liquid state (liquid limit, LL). The plasticity index (PI) is the range of moisture contents in which the sediment behaves plastically, i.e. PI = LL - PL (see Lambe & Whitman 1979). There is a direct correlation between plasticity and grain size composition: the lower the grain size, the larger the grain surface area that attracts water. Plasticity therefore also correlates with sediment strength (Leroueil et al 1983; Yassir 1987). The Atterberg limits are therefore useful in classifying the behaviour of sediments with increasing water content. They were measured in the manner described in British Standards BS 1377 (1975). The results are presented in Figure 6. The water content of the field samples is included in the plot to show its proximity to the liquid limit of the sediment, bearing in mind that the surface water content in the field areas was variable and dependent on surface conditions. The Taiwan samples have comparable plastic and liquid limits, which would be expected from the similar grain size compositions observed (Figs 5, 6). Three published results for the formation samples (LCF1, 2, Nar, Kerr et al 1970) are added for comparison with the Trinidad mud volcano analyses. The Trinidad results tally well with the corresponding grain size analyses. The results for Piparo (Pip) and the Nariva Formation reflect the difference in clay content: Piparo has a much higher liquid limit. However, the results for Anglais Point (AP) and Palo Seco (PS 1&2) are in reasonable agreement with the Lower Cruse samples, but showing 10% lower liquid limits. The Lagon Bouffe material, as expected from its high sand/silt content, registered the lowest
THE ROLE OF SHEAR STRESS IN MUD VOLCANOES
467
Fig. 5. Grain size distribution of the <425jji fraction, Trinidad and Taiwan mud volcano samples. Included in the figure are formation core samples from Trinidad (Kerr et. al 1970). Sample names are referred to in the text.
fraction was separated and mounted on a ceramic slide using the vacuum, or suction technique (Shaw 1987). This method avoids differential settlement on the slide, which would skew the result in favour of the finer grained minerals. The X-ray diffraction measurements complied with the methodology in Brown & Brindley (1980). The mineral fractions were calculated according to method two in Pierce & Siegel(1969). The results for all samples are presented in Figure 7. Not shown in the figure is quartz, which showed up in some of the clay fraction in the Trinidad and in all the Taiwan samples. This is difficult to quantify as a small amount has a strong signal, skewing the results. It is estimated that quartz forms less than 10% of the clay fraction. The results should be treated as a semi-quantitative analysis of the clay minerals in the samples. The Taiwan samples seem to be homogeneous, showing little or no smectite/swelling clay. Published data for two wells, drilled less than 10 km Clay mineralogy west of the mud volcano area through the The clay mineralogy of the 2|xm fraction was ana- Gutingkeng Formation, show relative clay mineral lysed using the X-ray diffraction method. The clay proportions that are consistent with the mud
Atterberg limits, which reflects the readiness of the material to behave in a liquid manner at much lower water contents than its clay-rich counterparts. The activity of the samples, which is the PI divided by the percentage weight of the clay fraction (Skempton 1953) is also plotted in Figure 6. The activity of the muds varied between 0.5 (AP, PS2) and 1.1 (Digity) in Trinidad and around 0.5 in Taiwan (T3 recording 0.3). The values are largely too low for what would be expected from marine clays (1 + , Skempton 1953), considering the high clay content of many of the samples. Many gave values closer to sensitive clays (0.45) which have the ability to liquefy. This result could reflect the nature of the formations at depth. The mud volcano clays in both areas were determined to be thixotropic by Shih (1967) and Kerr et al (1970), as were the formation core samples tested by Kerr et al (1970).
468
N.YASSIR
Fig. 6. Atterberg Limits, moisture contents and activities of the Trinidad and Taiwan mud volcano samples. Included in the figure are formation core samples from Trinidad (Kerr et. al 1970). Sample names are referred to in the text.
volcano samples (Chou 1971). Trace amounts of swelling clays were detected in a core sample of siltstone from the Lower Gutingkeng Formation at depths of 1796—1976 m and in outcrop material from the same Formation (Chou 1971). Trace smectite was only observed in sample T6 but could possibly link the muds to the Lower Gutingkeng Formation. More detailed analysis is clearly required to verify this. The Trinidad samples all show a mixture of illite, smectite/swelling clay and kaolinite + chlorite, in variable proportions. A reasonable similarity was observed between different samples in the same area (compare Palo Seco PS 1, 2 and 3; and Digity Dig 1 and 2, sampled from different vents). Of particular interest in Figure 7 is the insert comparing the Anglais Point mud matrix (AP) with clasts within the matrix of seemingly identical composition (APC). The result gives the same relative clay mineral proportions (unfortunately, no measurement of clay content is available for APC). This result could suggest that the parent material has cohesive strength. Only comparison with the Lengua Formation from the Forest Reserve Core (FRC) is available (see insert, Fig. 7), but again, the percentage of the clay fraction is unknown.
The non-clay fraction and clasts Microscopic analysis of the silt and sand fractions in the matrix was carried out on all the samples. Quartz was always the dominant mineral (>50% of the total in Taiwan samples). All samples also contained feldspar, opaque iron-rich and lithic fragments. Some contained biotite and glauconite (Lagon Bouffe, Piparo, Devil's Woodyard). Microfossils were only found in Devil's Woodyard and Piparo samples and date at Miocene or younger (Banner 1988, pers. comm.). SEM work on the sand fraction showed very high angularity for all fragments, especially in Anglais Point and Piparo (Yassir 1989). Clasts were collected from two mud volcanoes: Anglais Point and Devil's Woodyard. None were found in any of the other volcanoes, although Piparo is known to have ejected an abundance of exotics interpreted to be from the olistostromal Nariva (Higgins & Saunders 1974), or incorporated overburden material (Deville et al. 2003). The sampled blocks were dated as Eocene sandstone, Miocene sandstone and clay (Lengua) and Cretaceous chert. The cherts were rounded pebbles and are more than likely to have been in younger formations. Other clasts were angular to sub-angular. An intact desert rose (gypsum) was retrieved from the Anglais Point
THE ROLE OF SHEAR STRESS IN MUD VOLCANOES
469
Fig. 7. Clay mineralogy of the <2fjum fraction, Trinidad and Taiwan mud volcano samples. The insert shows the relative proportions of minerals in the clay fraction of the Anglais Point mud, the Anglais Point clast and Forest Reserve core. SC refers to swelling clays. Sample names are referred to in the text.
mud volcano. Higgins & Saunders (1974) correlate blocks in the Anglais Point mud volcano with the Karamat, Herrera and Cruse Formations and in Devil's Woodyard with the Lengua, Tamana and Upper Cipero Formations (Fig. 2).
A note on the origin of the muds The homogeneity of results from the Taiwan mud volcanoes indicates that they have the same source and are not likely to be heavily contaminated by overburden material. Lin (1965) and Shih (1967) identify the source as the Gutingkeng Formation, but the depth is uncertain. The presence of greywackes, olistostromes and sandpipes in the Lower Gutingkeng, as well as its diapiric nature in the mud volcano area, seems to point to this part of the formation as the mud volcano source. This also ties in with the geochemical results of Gieskes et al. (1992), who suggest a deep origin for the mud volcano fluids. The results from the Trinidad volcanoes are more scattered, suggesting different sources and/or contamination by overburden material. There is a reasonable tally between the grain size and plasticity of the Lower Cruse clays and Anglais Point. A match in
relative clay mineral percentages between the Anglais Point matrix and mud clasts contained within the matrix suggests that it is the same material. These results do not give any conclusive evidence on the source of the parent bed, however. The presence of older Cretaceous blocks and microfossils in the matrix are a matter of debate. The occurrence of olistostrome horizons in the overpressured parent beds led Higgins & Saunders (1974) to suggest that the material was deposited on structural flanks, incorporating blocks from older formations. Chou (1971) makes similar observations with the Gutingkeng Formation. Deville et al (2003), however, suggest that the source of the onshore mud volcanoes in Trinidad is Cretaceous, which implies that the blocks are incorporated from all depths during the ascent of the mud. Their conclusions are based on geochemical, micropalaeontological and exotic block analyses.
The effect of stress on sediment mobilization The two areas studied show a clear association between mud volcanoes and compressional tectonics.
N.YASSIR
470
The mud volcanoes in both areas are in technically active zones, occur along structural axes and are associated with other sedimentary intrusions such as sand dykes. The overpressured sediments thought to be the source of the mud volcanoes have been deposited rapidly during tectonically active periods. Furthermore, evidence is presented to suggest that these sediments have cohesive strength in situ and disaggregate in the mud volcanic process. This has led the author to investigate the effect of high shear stresses on mobilizing the sediments.
Experimental results High-pressure triaxial undrained shear tests were performed to assess the effect of tectonic shearing on low permeability clay. Results from two types of test conducted on the remoulded Lagon Bouffe silts, Trinidad and the Taiwan silty clays are presented here; the details of the testing procedure are described in Yassir (1989): (1)
(2)
Anisotropic consolidation - undrained shear. The samples were consolidated anisotropically with a horizontal to vertical effective stress ratio of 0.6, which is a realistic ratio during consolidation of mudstone. They were then subjected to axial loading in an undrained state at a constant mean effective stress (pore pressures in the sample were constantly monitored). The axial loading represents tectonic shear stress in a geological setting. The consolidation effective stresses ranged between 15 and 65 MPa, which correspond roughly to consolidation depths of 1-4 km. Cyclic loading. One sample was subjected to cyclic loading, which involved isotropic consolidation to a certain mean effective stress, followed by a series of loading/unloading cycles to simulate the effect of earthquake activity.
The results are shown in the 'critical state', or p'-q, diagram in Figure 8, which is a plot of the mean effective stress (p' = (oj' + cr^ + cr3')/3) versus the differential stress (q = cr1' — cr3'). The critical state model predicts that a normally consolidated sediment will 'contract' or compact during shear. If shear occurs in an undrained state, then the pore pressure will increase, causing a reduction in p' (the stress path moves to the left); see Atkinson & Bransby (1978). All mud volcano samples show this effect (Fig. 8). In fact, with the exception of the Trinidad sample consolidated to the highest stress (65 MPa), they all show a 'flat' stress path. This means that shearing is accompanied by only a slight increase, or even a decrease, in q, or shear strength
(Fig. 8). This behaviour is indicative of a potential to flow during deformation (Fig. 8). The samples shear until they reach the 'critical state line' (CSL), which is unique for each sediment type (Fig. 8). If 'critical state' is reached, the material continues to deform with no further changes in effective stress (constant p'-q) (Atkinson & Bransby 1978). Again, all samples show this behaviour, with the exception of the Trinidad and Taiwan samples consolidated to the highest stress (65 MPa, Trinidad; 50 MPa, Taiwan). Grain crushing and increased grain contacts and angularity at the higher stresses were observed using SEM analyses of tested samples (Yassir 1989). This suggests that shearing is an important process in producing grain angularity, as observed in mud volcano sediments. Note that the Trinidad silts are stronger than the Taiwan clays (steeper CSL), which would be expected as silt has a higher friction angle. Figure 9 shows the effect of a cyclic loading/unloading test conducted on a remoulded Taiwan sample consolidated isotropically (i.e. equal all-round stress) to 30 MPa. It can be seen that, despite full consolidation to an equivalent depth of 2 km, a 'steady state' pore pressure of over 15 MPa was created in the sample during the test, even when axially unloaded.
Discussion: stress-induced flow and liquefaction The results confirm that all normally consolidated sediments can become overpressured when sheared under undrained conditions - even at depths of several kilometres. Earthquake activity produces a similar effect. Although the sediments in both areas have likely already been overpressured by a combination of rapid loading and fluid generation, tectonic shearing would have contributed to the overpressure. The effect of shear stress is not limited to active margins but also passive margins, particularly at the toe of a delta. The mud volcano samples display contractant behaviour during undrained shear, most reaching 'critical state'. This behaviour, observed up to high effective stresses, demonstrates the ability of the material to flow. The mud volcano materials tested in this study are loose sediments, remoulded during remobilization. It is more difficult to assess how the sediment in situ would behave. No usable samples are currently available, but some of the evidence presented here suggests that the parent bed is a competent rock. Examples include the presence of seemingly identical clasts in the otherwise remoulded Anglais Point mud and the descriptions of the overpressured parent beds (Higgins & Saunders 1974). If this sediment
THE ROLE OF SHEAR STRESS IN MUD VOLCANOES
471
Fig. 8. Plot of mean effective stress (p') versus differential stress (q) for undrained shear experiments conducted in this study on samples from Taiwan and Lagon Bouffe in Trinidad. Also included are results obtained from chalk (Leddra & Jones 1989) and mudstone (Ohtsuki et al 1981).
has a high in situ porosity, then it is likely that this material can become remoulded by shearing, displaying behaviour more akin to liquefaction. Liquefaction is an undrained shear phenomenon observed in loose sands and sensitive clays at the surface (Lambe & Whitman 1979). These sediments have an 'open' structure, which, if shaken or loaded beyond its strength, collapses. The effect of this collapse is to transfer the load onto the interstitial fluid, and the material liquefies. This occurs to varying degrees depending on the original porosity and points of contact between sediment particles. A 'loose' structure can be geologically maintained by cementation; it is commonly observed in high porosity chalks and sandstones, which have a tendency to liquefy under shear or cyclic loading (Leddra & Jones 1989). Figure 8 also shows results from undrained shear experiments on chalk (Leddra & Jones 1989) and mudstone (Ohtsuki et al 1981). The chalks at the consolidation mean effective stresses of 20 and 25 MPa liquefied, turning into slurry. The mudstone at 10 MPa showed comparable behaviour (Fig. 8). At the higher consolidation stress, the cementation of the chalk was destroyed during consolidation, altering the undrained shear behaviour of the sediment (dilation at failure). It is highly likely
the mechanism that produces sand sills and dykes common to mud volcano areas (Kugler 1933; Chou 1971) is liquefaction as observed in the samples at the lower effective stresses. Major high porosity turbiditic and deltaic sand bodies are associated with mud volcanic source beds and could potentially deform in this manner.
A note on stress control on mud volcano emplacement Deep-seated sedimentary volcanoes need to break through a brittle overburden. This process has to be driven by the fluid pressure in the rock and controlled by the in situ stresses and structure. For an overpressured sediment to actually hydraulically fracture its overburden, the pore fluid pressure in the overpressured unit (Ppopu) nas to exceed the total minimum stress in the overburden (orminob), as well as its tensile strength (Tob): In this case, the fracture will be oriented parallel to the maximum stress and open against the minimum stress.
472
N.YASSIR
Concluding remarks
Fig. 9. Plot of (a) differential stress (q) and (b) excess pore fluid pressure versus axial strain in the cyclic loading test conducted on a Taiwan silty clay sample.
If the overburden is already faulted, which is the more usual scenario, the tensile strength need not be exceeded. The mobilized sediment will usually find it easier to re-open pre-existing faults than to initiate new fractures, even if the faults are oriented normal to the maximum stress. Field evidence certainly seems to point to this. In Trinidad, mud volcanoes occur on the faulted crests of anticlines (Fig. 1), where the greatest amount of deformation and fracturing takes place. In Taiwan, the mud volcanoes were shown to extrude along reverse faults (Fig. 3). In both cases, the volcanoes are coming up against the current regional maximum stress. Mud volcanoes are also known to come up along fault intersections, for example in Trinidad (Yassir 1989) and in New Zealand (Ridd 1970). The mud volcanoes observed at offshore Louisiana (MacDonald et al 2000) and in the Niger Delta (Graue 2000) occur in the vicinity of listric normal faults. In these cases, the faults are oriented ideally with respect to minimum stress and the pore pressure need only exceed the value of the minimum stress to effect the escape of overpressured sediment. The pre-existing structure will therefore have a primary control on the mode of emplacement of a sedimentary volcano.
Deep-seated mud volcanoes are associated with a thick, overpressured source sediment, with gas, with structure (folds, faults) and usually with tectonic activity. This has led to conflicting opinions on their origin. Shear stress is identified as an important control on deep-seated mud volcano formation in Trinidad and Taiwan. The role of shear stress in overpressuring and mobilizing sedimentary rock is presented in this paper, based on laboratory results from high-pressure triaxial tests. It is shown that shear stresses are capable of generating enormous overpressures in sedimentary rocks, even when they are normally consolidated at depth. If the sediments are already overpressured and have a low density, shear stress can result in flow, or even 'classical' liquefaction. The environment of rapid deposition and high porosity overpressured sediments common to mud volcano areas renders them susceptible to this behaviour. These results are supported by observations of intact, competent parent rock from core and mud volcano clasts. This conclusion is particularly applicable to tectonic regions, but also applies to passive margins. The effect of gas emission on the formation of the mud volcanoes studied is not discussed in this paper. The relative roles of shear stress and gas in disaggregating and mobilizing sedimentary rocks is as yet poorly understood and should be further studied. Finally, it is concluded that pre-existing geological structure is more likely to control the path of a mud volcano than the formation of a new hydraulic fracture. This paper is dedicated to the memories of M. E. Jones, my PhD supervisor, who passed away in 2002 and G. Higgins, my mentor, who passed away in 1999. The work was undertaken as part of my PhD thesis whilst at University College, London. D. Dewhurst and A. Kopf kindly reviewed this manuscript.
References ATKINSON, J.H. & BRANSBY, PL. 1978. The mechanics of soils — an introduction to critical state soil mechanics. McGraw-Hill, Maidenhead 375p. BARBER, A., TJOKROSAPOETRO, S. & CHARLTON, T. 1986. Mud volcanoes, shale diapirs and melanges in accretionary complexes, Eastern Indonesia. American Association of Petroleum Geologists Bulletin, 70, 1729-1741. BARR, K. & SAUNDERS, J. 1965. An outline of the geology of Trinidad. Transactions of the 4th Carribean Geological Conference, Trinidad. Published 1968,1-10. BERTRAND, A. & BERTRAND, W. 1985. Plate tectonic evolution of the southeast Carribean. In: Transactions of the First Geological Conference of the Geological Society of Trinidad and Tobago, 242-260.
THE ROLE OF SHEAR STRESS IN MUD VOLCANOES BIOT, M. & ODE, H. 1965. Theory of gravity instability with variable overburden and compaction. Geophysics, 30, 213-227. BISHOP, R.S. 1978. Mechanism for emplacement of piercement diapirs. American Association of Petroleum Geologists Bulletin, 62,1561-1581. BROWN, K.M. 1990. The nature and hydrogeologic significance of mud diapirs and diatremes for accretionary systems. Journal of Geophysical Research, 95, 768-778. BROWN, G. & BRINDLEY, G. 1980. X-ray mineral analysis of clays. In: BRINDLEY, G. & BROWN, G. (eds) Crystal Structures of Clay Minerals and their X-ray Identification, 411-438. Mineralogical Society, London,. BRITISH STANDARDS BS 1377. 1975. Methods of test of soils for civil engineering purposes, BSI. CARR-BROWN, B. & FRAMPTON, J. 1979. An outline of the stratigraphy of Trinidad. In: 4th Latin American Geological Congress of the Geological Society of Trinidad and Tobago, 7-20. CHAPMAN, R.E. 1974. Clay diapirism and overthrust faulting. American Association of Petroleum Geologists Bulletin, 85,1597-1602. CHOU, J. 1971. A preliminary study of the stratigraphy and sedimentation of the mudstone formations in the Tainan area, southern Taiwan. Petroleum Geology of Taiwan, 8,187-210. DEVILLE, E., BATTANI, A., HEREIN, J-P, HOUZAY, J-P, PRINZHOFER, A. & MULLER, C. 2003. New insight for the origin and process of mud volcanism in Trinidad. In: VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, AJ. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization. Geological Society, London, Special Publications, 216,475-490. DIA, A.N., CASTREC-ROUELLE, M., BOULLEGUE, J. & COMEAU, P. 1999. Trinidad mud volcanoes: where do the expelled fluids come from? Geochimica et CosmochimicaActa63,1023-1038. GIESKES, J.M, You, C.-E, TYPHOON, L., Yui, T.-F. & CHEN, H.-W. 1992. Hydro-geochemistry of mud volcanoes in Taiwan. Acta Geologica Taiwanica 30,79-88. GRAUE, K. 2000. Mud Volcanoes in deepwater Nigeria. Marine and Petroleum Geology 17,959-974. HEDBERG, H.D. 1974. Relation of methane generation to undercompacted shales, shale diapirs and mud volcanoes. American Association of Petroleum Geologists Bulletin, 58,661-673. HIGGINS, G.E. & SAUNDERS, J.H. 1967. Report on the 1964 Chatham mud island, Erin Bay, Trinidad, W.I. American Association of Petroleum Geologists Bulletin, 51,55-64. HIGGINS, G. & SAUNDERS, J. 1974. Mud volcanoes - their nature and origin. Contributions to the Geology and Palaeobiology of the Carribean and Adjacent Areas. Verhandl. Naturfosch. Ges. Basel, 84,101-152. HJELSTUEN, B.O., ELDHOLM, O. FALEIDE, J.I. & VOGT, R. 1999. Regional setting of Hakon Mosby mud volcano, SW Barents Sea margin. Geo-Marine Letters 19, 22-28. Ho, C. 1982. Tectonic evolution of Taiwan. Published by the Ministry of Economic Affairs, Republic of China, Taipei, Taiwan, 153 p. HUANG, I.L., TENG, L.S. et al 1992. Structural styles of off-
473
shore southwestern Taiwan. Earth and Oceanic Science, 73,539. HsffiH, S.H. 1972. Subsurface geology and gravity anomalies of the Tainan and Chungchou structures of the coastal plain of southwestern Taiwan. Petroleum Geology of Taiwan, 10, 324-338. JAKUBOV, A., ALI-ZADE, A. & ZEINALOV, M. 1971. Mud volcanoes of the Azerbaijan, SSR. Publishing House of the Academy of Sciences of the Azerbaijan SSR, Baku, 256p. KERR, P.P., DREW, I.M. & RICHARDSON, D.S. 1970. Mud volcano clay, Trinidad, West Indies. American Association of Petroleum Geologists Bulletin, 54, 2101-2110. KOPF, A. J. 2002. The nature and significance of mud volcanism. Reviews of Geophysics, 40/1 (in press). KUGLER, H.G. 1933. Contributions to the knowledge of sedimentary volcanism in Trinidad. Journal of the Institute of Petroleum Geology of Trinidad, 19, 743-772. KUGLER, H.G. 1965. Sedimentary volcanism. In: Proceedings of the 4th Caribbean Geological Conference, 11-14. LAMBE, T.W. & WHITMAN, R.V 1979. Soil Mechanics, SI Version. John Wiley and Sons, New York, 553p. LEDDRA, M. & JONES, M. 1989. Steady state flow during undrained loading of chalk. In: Proceedings of the International Chalk Symposium, Brighton, 245-252. LEROUEEL, S. TAVENAS, F. & LE-BIHAN, J. 1983. Proprietes characteristiques des argiles de Test du Canada. Canadian GeotechnicalJournal, 20, 681-705. LIN, C.C. 1965. The naming of the Akungties Formation, with a discussion of the origin of the fossils in the mud ejected from the Kunshuiping mud volcanoes near Chiao-tou Kaosiunghsien, Taiwan. Petroleum Geology of Taiwan, 4,107-145. MACDONALD, I.R., BUTHMAN, D.B., SAGER, W.W., PECCINI, M.B. & GUINASSO JR., N.L. 2000. Pulsed oil discharge from a mud volcano. Geology 28 (10), 907-910. McMANUS, J. & TATE, A. 1986. Mud volcanoes and the origin of certain chaotic deposits in Sabah, east Malaysia. Proceedings of the Geological Society of Malaysia Bulletin 19,193-205. MILKOV, A.V. 2000. Worldwide distribution of submarine mud volcanoes and associated gas hydrates. Marine Geology 167,29-42. MORGAN, J., COLEMAN, J. & GAGLIANO, S. 1968. Mudlumps: diapiric structures in Mississippi river sediments. In: Diapirism and Diapirs. In: BRAUNSTEIN, J. & O'BRIEN, G. D. (eds) American Association of Petroleum Geologists Memoir, 8,145-161 OHTSUKI, H., NISHI, K, OKAMOTO, T. & TANAKA, S. 1981. Time dependent characteristics of strength and deformation of a mudstone. In: Proceedings of the Symposium on Weak Rock, Tokyo, 1,119-124. PIERCE, J. & SIEGEL, F. 1969. Quantification of clay mineral studies of sediments and sedimentary rocks. Journal of Sedimentary Petrology, 39,187-193. RIDD, M. 1970. Mud volcanism in New Zealand. American Association of Petroleum Geologists Bulletin, 54, 601-616. SONDHI, VP. 1947. Islands developed in the Arabian Sea during the Makran earthquake of 28th November 11945. Indian Minerals, 1,147-154.
474
N.YASSIR
SHAW, H. 1987. Clays and their effects on source and reservoir rocks. Course notes from the Joint Association for Petroleum Exploration Courses (UK), 61p. SfflH, T. 1967. A survey of the active mud volcanoes in Taiwan and a study of their types and the character of the mud. Petroleum Geology of Taiwan, 5,259-311. SKEMPTON, A. 1953. Soil mechanics in relation to geology. Proceedings of the Yorkshire Geological Society, 29, 33-62. WESTBROOK, G. & SMITH, M. 1983. Long decollements and mud volcanoes: evidence from the Barbados Ridge complex for the role of high pore fluid pressure in the development of an accretionary complex. Marine and Petroleum Geology, 4,71-80.
YASSIR, N. 1987. Mud volcanoes: evidence of neotectonic activity. Memoir of the Geological Society of China, 9,513-524. YASSIR, N. 1989. Mud Volcanoes and the Behaviour of Overpressured Clays and Silts. Unpublished PhD Thesis, University of London, UK. YASSIR, N. & ADDIS, M.A. 2002. Relationship between pore pressure and stress in different tectonic settings. In: HUFFMAN, A.R. & BOWERS, G.L. (eds) Pressure regimes in sedimentary basins and their prediction. American Association of Petroleum Geologists Memoir, 76, 79-88.
The origin and processes of mud volcanism: new insights from Trinidad E. DEVILLE1, A. BATTANI1, R. GRIBOULARD2, S. GUERLAIS1, J.R HEREIN1, J.P. HOUZAY3, C. MULLER,1 & A. PRINZHOFER1 1
Institut Frangais du Petrole, 1-4, av. de Bois-Preau, 92 852 Rueil-Malmaison Cedex, France (e-mail: eric, deville @ ifp.fr) 2 Universite de Bordeaux /, av. desfacultes, 33 405 Talence Cedex, France. 3 TOTALFINAELF-CSTJF, avenue Laribbau, 64018 Pau Cedex, France Abstract: The mobilized sediments expelled by the mud volcanoes in Trinidad correspond to liquefied argillaceous and sandy material in which the solid fraction is systematically polygenic and originating from several formations (Cretaceous to Pliocene). The mud is notably rich in thingrained quartz that is angular and frequently mechanically damaged related to shearing at great depth, during the sedimentary burial, and/or hydraulic fracturing processes. The exotic clasts are mostly fractured fragments from various formations of the tectonic wedge (mostly Palaeocene to Miocene). The origin of the solid particles of the mud is polygenic, including deep CretaceousPalaeogene horizons close to the decollement, and various materials from the stratigraphic pile pierced by the mud conduits. Moreover, the fluids expelled by the mud volcanoes have a deep origin and notably the gas phase is thermogenic methane generated probably below a depth of 5000 m. The effusions occur either during cycles of moderate effusion of mud and fluids (quiescence regime), or during catastrophic events responsible for the expulsion of huge volumes of mud, clasts and fluids (transient regime). Available subsurface data suggest that the deep structure of the mud volcanoes includes: (1) a focused deep conduit at depth in the zone of overpressure; (2) a mud chamber intruding the surrounding formations around and above the top of the abnormal pressure zone; and (3) a superficial outlet leading to the surface vents.
Convergent and transpressive margins are frequently affected by mud volcanism (extrusion of liquefied sediments) and shale diapirism (extrusion of plastic shale) processes, which influence the structural evolution and the fluid dynamics within the tectonic wedge. Mud volcanism and shale diapirism processes are both classically related to the development of overpressure at depth (Higgins & Saunders 1974; Brown & Westbrook 1988; Brown 1990 and many others), which contributes to sediment mobilization by reducing the strength within the overpressured layer (Hubbert & Rubey 1959) and which is necessary for the extrusion of the mud column. Our present knowledge about the origin mechanisms of mud volcanism is very qualitative and mostly restricted to a list of phenomena likely to generate the overpressure conditions (sedimentary loading, tectonic compaction, tectonic overloading, gashydrates occurrence in offshore areas which are likely to reduce permeability in the superficial levels). However, it is difficult to evaluate the relative importance between the different processes. Some are probably negligible, whereas others are of prime importance. The high deformation rates in convergent margins probably have an important role in the dynamic development of overpressure (typically non-static phenomena). Moreover, high temperatures induce the cracking of hydrocarbons in
thick prisms, which is an additional source for overpressure generation and reducing the density of the sediments (favourable factor for diapiric processes; Hedberg 1980). In offshore areas one may invoke destabilization of gas hydrate for the origin of the gas expelled by the mud volcanoes, but obviously this interpretation can not be generalized because many mud volcanoes develop in shallow water and onshore area, outside the field of stability of gashydrates. Also, the trigger of mud diapirism is unclear (liquefaction by seismic agitation of fluidsaturated sediments, hydrofracturing). It is not known exactly how the sediment is mobilized at depth, and little is known about the flows of expulsion and the dynamics of steady-state effusions versus catastrophic eruptions. In this paper, recent structural observations, biostratigraphic analyses, mineralogical and geochemical analyses from Trinidad are summarized. Trinidad is commonly considered as an onshore reference area to study mud volcanism processes (Barr 1953; Kugler 1965, Higgins & Saunders 1974; Dia et al 1999). It is a favourable area in order to provide new insight about the origin and the processes of mud volcanism. This work has been completed by a study on cores taken from the offshore mud volcanoes in the southern part of the Barbados accretionary prism, NE of Trinidad (Gonthier^a/. 1994).
Fig.1. Structural sketch-map of the southern part of the Barbados prism and Trinidad.
Geological setting of the mud volcanoes of Trinidad Sedimentary volcanism in Trinidad occurs in a context of plate boundary between the Caribbean plate and the South American plate, at the junction between the Barbados accretionary prism and the transform system of the northern Venezuela (Speed 1985; Stein et al 1988; Robertson & Burke 1989; De Mets et al. 1990). Within this compressional and transpressional system, a several hundred kilometres-long active belt of mud volcanoes and shale diapirs develops from the Barbados tectonic wedge to the thrust belt of Northern Venezuela (Fig. 1). In this system, the Trinidad mud volcanoes are the only emerged part of a widely developed phenomenon,
especially in the offshore area of the Barbados prism, which is an archetype of a thick accretionary wedge related to lithospheric convergence and subduction processes. In the NE offshore of Trinidad within the southern part of the Barbados prism, mud domes and volcanoes developed in different structural settings (BijuDuval et al 1982; Valery et al 1985; Brown & Westbrook 1987; Brown 1990; Figs 1 and 2). Some mud volcanoes are found in the Atlantic abyssal plain eastward of the tectonic front along transform faults (Henry et al 1996; Lance et al 1998). In the Barbados prism, the front of the tectonic wedge is characterized by an imbricate thrust system mostly devoid of active mud volcanism activity. The main province of active shale diapirs and mud volcanoes is
MUD VOLCANISM ORIGIN AND PROCESSES IN TRINIDAD
477
Fig. 2. Structural map of Trinidad and location of the mud volcanoes and the main oil and gas fields. Visited mud volcanoes: P. Piparo, T. Tabaquite, C. Cascadoux, DW. Devil's Woodyard, D. Digity, RD. Rock Dome, Ma. Marac, Mo. Moruga, LB. Lagon Bouffe, MD. Morne Diablo, PS. Palo Seco, A. Anglais Point, E. Erin group, CG. Columbus group, Cha. Chatham; PL. Pitch Lake (bitumen lake), Mah. Mahaica gas field.
found within the core of the tectonic prism. The highest activity of sedimentary volcanism is preferentially located along the major transfer zones (especially at the eastern extremity of the El Pilar fault in the Barbados prism; Valery et al. 1985; Griboulardefa/. 1991,1998). The western part of the Barbados prism is characterized by the occurrence of extension structures superimposed on thrust tectonics (Deville 2000) and the inner part of the prism is devoid of active mud volcanism. Nevertheless, in the inner part of the Barbados prism, fossil mud volcanic activity associated with hydrocarbon migration is known notably on Barbados island (Joe's River fm.; Senn 1940; Speeds al 1991). Laterally, toward the SW, the belt of mud volcanoes is becoming narrower and emerges in Trinidad within the fold-and-thrust belt of the southern part of the island. This thrust system develops in a transpressive regime generating transcurrent fold-andthrust structures in southern Trinidad. The deep structure of the southern part of Trinidad (central and southern range) is well constrained by numerous exploration wells (Kugler 1953, 1959, 1965; Tyson et al 1991a, b\ Payne et al 1991; Ahmad 1991; Higgins 1996). Three main groups of formations can be distinguished (Fig. 3): (1) Cretaceous platform
carbonates and shales (formations Cuche, Gautier, Naparima Hill and Guayaguayare); (2) Palaeocene to Early Miocene turbidites and hemipelagic shales (formations Lizzard Springs, Navet, Nariva, Cipero, Karamat); and (3) Miocene to present relatively shallow-water clastic series (formations Lengua, Cruse, Forest, Morne 1'Enfer, Cedros)(Kugler 1959; Barr & Saunders 1968; Carr-Brown & Frampton 1979). The mud volcanoes in outcrop, as well as the subsurface mud plugs, are located along the Naparima Hill fault system in the Central Range, along the trends of the transpressive thrust anticlines of the Southern Range and along the Los Bajos major transfer fault (Barr 1953; Kugler 1956; Birchwood 1965; Higgins & Saunders 1974; Yassir 1989; Fig. 2). Except for the northern mud volcanoes along the Naparima Hill fault system, in the Central Range (Piparo, Tabaquite), it is worth to note that the mud intrusions do not crosscut thrust-sheets, but they develop on tops of the buried CretaceousPalaeogene anticlines covered by the Neogene clastic series. An example of the deep structure of a mud intrusion is shown in Figure 4, in the area of the Forest Reserve oil field (Kugler 1959; Bower 1965; Pat 1960 published in Higgins 1996; Higgins & Saunders 1974). This section evidences different
478
E.DEVILLE£TAL.
Fig. 3. Simplified cross-section in the southern Trinidad area, (compiled from Kugler 1959, 1996; Tyson et al. 1991) (location in Figure 2).
parts of the mud intrusion: (1) a deep conduit (depth greater than 2000 m), which developed probably along a fracture system affecting Cretaceous and Palaeogene formations; (2) an intrusive mud chamber, which has a volume of several hundred millions of m3; this mud plug intrudes the argillaceous horizons (notably the Pliocene Lower Forest Clay) at depth between 1000-500 m; (3) a superficial conduit connecting the mud chamber to the surface.
of combustible gas. Gas bubbles have variable sizes from millimetric to metric. The expulsion is active either in the cones or in the mud pools and lakes. In some sites, the gas flows are relatively high, more than 10 m3 of gas per minute at Lagon Bouffe and Palo Seco. Generally, the mudflows are much more moderate compared to the gas flows (3 to 10 times lower). In some places, the mud volcanoes (notably in Marac and Erin) also expel important flows of oil. Also worth mentioning is the very famous Pitch Lake, which is a more than one kilometre wide crater expelling heavy oil and bitumen.
Processes of mud volcanism Superficial
effects
In the visited active sites in Trinidad, the mud volcanoes have different features; they are either: (1) fields (or tassiks) of small active metric to decimetric cones (griffons; Fig. 5) on the top of morphologic domes up to 150 m high, with a diameter up to 2 kilometres; or (2) mud pools and lakes with dimensions that vary from tens of centimetres (vertical chimneys of mud; Palo Seco, Erin) up to several hundred metres in diameter in Lagon Bouffe (Fig. 6) that shows several convective cells of mud. The morphological differences between cones and pools are directly dependant on the viscosity and density of the mud expelled (density variability between 1.2 g/cm3 in the mud pools to more than 1.7 g/cm3 in the griffons, the mean values being around 1.45 g/cm3). The surface temperatures of the expelled mud are relatively low and vary between 25-33 °C (the mean value being around 28°C, close to the mean annual temperature in Trinidad). The expulsion of the fluids and the solid fraction in the active mud volcanoes is not constant through time, but in most of the visited sites it is possible to identify phases of slow but permanent expulsion and more or less catastrophic events (Higgins & Saunders 1974; Yassir 1987,1989). Quiescence regime. In the visited active sites, we have noticed continuous venting of mud and bubbles
Catastrophic regime. Episodically, the mud volcanoes are responsible for eruption of mud containing polygenic blocks and breccias from various formations of the tectonic wedge (carbonates, sandstones, shales as well as mega crystals of calcite, pyrite nodules) and associated with high gas flows (Higgins & Saunders 1974; Wharton & Hudson 1995). A catastrophic eruption occurred in Piparo on February 22, 1997 (Fig. 7), which destroyed several houses. This eruption was associated with the development of a network of open fractures in the eruption area, which acted as conduits for the mud expulsion. Some of these newly formed fractures exhibited ENE-WSW dextral strike slip movements, with throws up to 15 cm, compatible with the present deformation regime along the central belt. An eruption occurred also in Devil's Woodyard on May 8, 1995. Short-lived islands associated with catastrophic eruptions appeared several times in the Columbus channel (southern offshore of Trinidad). The most famous site corresponds to the Chatham ephemeral island whose last eruption occurred on May 10, 2001. It was the fourth time that the mud volcano rose since 1911. The previous eruptions occurred on August 1, 1964 (Higgins & Saunders 1967), December 21, 1928 and November 3 and 4, 1911 (Arnold & MacReady 1956), i.e. a chronological spacing of 17, 36 and 37 years. The mud dome morphologies are mainly acquired during phases of catastrophic eruption and progressively rain run-off and progressive addition of small mudflows during phases of quiescence activity reshape these domes.
MUD VOLC ANISM ORIGIN AND PROCESSES IN TRINIDAD
479
Fig. 4. A. Cross-section of the Forest Reserve mud plug (compiled from Kugler 1965; Bower 1965; Pat 1960 published in Higgins 1996). Note the structure of the mud intrusion, which is constituted of a deep conduit, an intermediate mud chamber and a summital outlet. The mud chamber is intruded preferentially within the clay layers, especially the lower Forest clay. The sandstone reservoirs around the mud plug are oil charged. B. Depth map of the top of the mud chamber which shows a conical shape (from Bower 1965).
In the deep offshore multi-beam bathymetry, et al 1994). These structures correspond more deep-towed acoustic sonar and seismic data (2D and probably to piercing shale diapirs and are consid3D) acquired in the southern part of the Barbados ered different from the mud volcanoes described prism have illustrated that different types of mud onshore. volcanoes and piercing shale diapirs are spectacuGas bubbles are systematically expelled by the larly developed in this area (Valery et al 1985; mud volcanoes onshore Trinidad, whereas offshore Mascle et al 1990; Faugeres et al 1989, 1991;Trinidad and Barbados no bubbles were observed, Griboulard et al 1991; Gonthier et al 1994; despite the existence of hydrocarbon gas charged Rutledge & Leonard 2001). Some sedimentary effu- fluid expulsion at the top of mud volcanoes (Jollivet sive structures are volcano-shaped (slopes increas- et al 1990). The absence of gas bubbles in deep ing toward a top with a central crater). Some of them water is probably related to the higher solubility of show concentric structures and up to kilometric- methane under high pressure compared to atmoslong mud flows. Also, craters filled by convective pheric pressure. mud (diatremes) have been described in the Atlantic abyssal plain (Henry et al 1996; Lance et al 1998). Most of these offshore structures look very similar Origin of the solid fraction to mud volcanoes on Trinidad Island. Some structures appear on the seismic data as dome-shaped diapirs with tops showing a bumpy topography Mud punctuated by small pockmarks, from which cold fluids are expelled (Faugeres et al 1989, 1991; The mineralogy of the mud was determined by XJollivetefa/. 1990; Griboulard ef al 1991; Gonthier ray diffraction and Scanning Electronic Microprobe
480
E.DEVILLEETAL.
Fig. 5. Metric vents of mud and gas (griffon) in Devil's Woodyard.
Fig. 7. Fractured angular clasts from the catastrophic eruption of Piparo, which occurred on February 22, 1997.
Fig. 6. Gas sampling in the center of one of the convective cells of mud in the Lagon Bouffe lake.
Fig. 8. Structure of the coarse grain from the Erin mud volcanoes (SEM, EDS analysis).
(SEM) by Energy Dispersion Spectrometry (EDS) analysis. For the Trinidad mud volcanoes, X-ray diffraction analysis of the mud was carried out on samples from three sites (Moraga, Erin and Devil's Woodyard). In addition, a sample of mud taken from a mud volcano in the Erin Group was studied with a SEM. It appears that the studied samples consist of a wide range of particles (various sizes and compositions). Previous X-ray studies have shown that the mud is rich in kaolinite, illite, smectite (Kerr et al. 1970), but also quartz, feldspar and calcite (Dia et al. 1999), which is confirmed by this study. Indeed, combined X-ray diffraction and SEM results have shown that the solid particles within the mud are composed of clays (kaolinite, illite, smectite, vermiculite), chlorite and muscovite, but also abundant grains of quartz, feldspar (albite, K-feldspar), carbonates (calcite, dolomite, siderite) and titaniferous minerals (rutile, anatase), apatite, baryte and pyrite (Fig. 8). The grain size varies from less than 0.25 jxm to more than 200 jjim and the grains are supported within a very
thin matrix constituted by a mixing of various clays and micas sheets (Fig. 9). We can distinguish the following grain-types present in very variable proportions according to the site: (1)
Relatively large fragments of quartz, albite, and K-feldspar with sizes bigger than 50 fjum and smooth contours related to sedimentary transport. (2) Small fragments (less than 5 jjim) of quartz and albite with clearly angular shapes and internal microfractures (Fig. 10), especially in quartz. The mechanical damage probably results from shearing during compaction or mud volcanism eruptive processes. Such quartz grains can make up more than 90% of the solid fraction within the mud (notably in some vents at Moraga). (1) Agglomerates (less than 10 |xm) of titane oxides, apatite and barite. The mud is relatively rich in titane oxides (rutile, anatase) with respect to the sedimentary formations of southern Trinidad.
MUD VOLCANISM ORIGIN AND PROCESSES IN TRINIDAD
Fig. 9. Thin-grained matrix structure from the Erin mud volcanoes (SEM, EDS analysis).
481
Fig. 11. Microscopic structure (SEM, EDS analysis) of the mud from an offshore mud volcano in the Barbados prism (core KS87-117, NE of Trinidad).
Fig. 10. Thin-grained angular and damaged quartz from the Erin mud volcanoes (SEM, EDS analysis).
(4)
The matrix that supports the fragments, made up of illite, kaolinite, chlorite, mica.
In the studied cores from the Barbados prism, combined X-ray diffraction and SEM analysis has shown that the solid particles of the mud include clays (kaolinite, illite, smectite, vermiculite), but also abundant are very thin quartz grains, feldspar, calcite, Ca-Mg carbonate, siderite, pyrite, titane oxides (rutile and anatase) and gypsum (Appendix 1). The quartz, feldspar and carbonate particles have also very thin granulometry (<3 |xm; Fig. 12). Also, as observed in the mud volcanoes onshore Trinidad, quartz and feldspar have angular shapes and show frequently evidences of brittle internal damage.
Exotic clasts In the Trinidad mud volcanoes with a recent eruptive activity (Piparo, Devil's Woodyard, Columbus
Fig. 12. Histogram of the grain size of the quartz, albite and carbonate particles from the core KS 87-1 17, Barbados prism, NE of Trinidad (862 measures).
group, Anglais Point, Moruga), exotic clasts are found (mainly centimetric to pluri-decimetric; Fig. 7) issued from various formations from the tectonic wedge. The nature of the clasts is poly genie (several types of carbonates, various sandstones from thingrained to coarse, shales, calcite crystals, sulphur nodules). Some clasts have round shapes and they could correspond to ancient pebbles initially interbedded within Tertiary formations and mobilized during eruptions. However, most of the clasts show angular shapes resulting from intense fracturing. Fractures are filled with carbonate cements (Ca and Ca-Mg). Frequently, real breccias made up of
482
E. DEVILLE ET AL.
angular and initial joined elements are included within calcite crystallizations. Clearly, this type of element cannot be interpreted as sedimentary pebbles. In addition, numerous angular clasts are very soft shales that are unlikely to have preserved their shapes during a sedimentary transport. We interpret most of the angular clasts and the breccia as the result of hydraulic fracturing processes. The same interpretation has been proposed for similar clasts and breccia from mud volcanoes in other accretionary wedges (Behrmann 1991; Kopf et al 2000; Huguen et al 2001). This study also found sulphur nodules, notably in the Columbus group and in Anglais Point, with diameters up to 50 cm.
gests that the mobilized sediments in shale domes and mud volcanoes, are of Miocene-Pliocene age (zones NN15 to NN21) at the front of the mud volcanism zone. This is consistent with biostratigraphic analyses of foraminifers (Faugeres et al. 1989; Griboulard et al. 1991). However, various horizons probably including the Eocene and certainly including Oligocene, Miocene and Pliocene intervals (zones NP25 to NP21) have been mobilized in the inner part of the active subsurface sediment mobilization zone of the Barbados Ridge (cores KS90-106 and KS 90-122; Fig. 1 and Appendix 2).
Origin of the fluids Biostratigraphic results Water Using nannofossil determinations, it is possible to date very precisely the macroscopic blocks expelled by the mud volcanoes, providing they have not suffered high diagenesis or too much dilution. This study has carried out these analyses on the angular clasts that are interpreted as fragments detached from deep formations during the mud volcanism process and not on the reworked pebbles that were initially interbedded within the Tertiary series. According to the ages obtained, these individual macroelements belong to Tertiary formations, ranging from Palaeocene to Miocene (see Appendix 2 for some examples). Numerous Cretaceous clasts are expelled by the Piparo mud volcano in the Central Range (Fig. 2), the older elements being of Early Barremian age (see Appendix 2). But in this case, these clasts can come from the base of the Naparima thrust sheet (Fig. 3) and they do not necessarily have a deep origin. On the other hand, the mud systematically shows a mixing of species ranging from Cretaceous to Late Miocene (see Appendix 2). This suggests that the mud consists of a mixture of microscopic elements of various origins. These fragments come from Tertiary formations (as is the case for macroscopic exotics), but also from Cretaceous levels. Because Cretaceous nannofossils were found in the mud but not in the clasts (except in Piparo), we cannot exclude that these fossils could be reworked. The same observation (Cretaceous nannofossils in the matrix and not in the clasts) has been made in the Mediterranean ridge (Huguen etal. 2001). However, these data show that the zone of initiation of the mud volcanism is necessarily at least as deep as the Palaeogene and probably the Cretaceous, because, as mentioned above, the mud intrusions do not crosscut thrust sheets involving CretaceousPalaeogene formations, except for Piparo (Fig. 3). In the offshore, the study of nannofossils in cores collected from the south of the Barbados prism sug-
The results of the geochemical analysis of the water expelled by the Trinidad mud volcanoes provide evidence for a deep origin related to dehydration of the sediments (Dia et al. 1999; Barboza & Boettcher 2000). A contribution by a surface aquifer occurs locally, notably in some mud volcanoes in Trinidad, west of the Los Bajos fault (Dia et al. 1999) and in the eastern offshore of Trinidad, in the Columbus basin (Barboza & Boettcher 2000). Onshore Trinidad, estimates of equilibrium temperatures imply that to acquire the higher temperature (150°C; Dia et al. 1999) the mud volcanoes must have been partly fed by a reservoir located at more than 5 km deep (probably around 6 km), the mean geothermal gradient in Trinidad being close to 20°C/km (Leonard 1983; Rodrigues 1985, 1989). In addition, a deep origin of the expelled fluid from the offshore mud volcanoes in the Atlantic abyssal plain, at the toe of the Barbados prism, has also been deduced from interstitial water analysis (95-100+ 20°C; Dia et al. 1995; Martin et al 1996; Castrec et al 1996; Barboza & Boettcher 2000).
Gas In the onshore mud volcanoes of Trinidad, the gas is mainly methane associated with moderate concentrations of ethane, propane and carbon dioxide. This dry gas is characterized by a §13C of methane, which ranges between —52 and — 33%o (Prinzhofer et al 2000; Fig. 13). Such 5I3Cj values associated with very dry gases have been interpreted as a missing between a purely bacterial gas and a purely thermogenic gas by Van Soest (2000). Nevertheless, methane 513C can be affected by post-genetic phenomena (segregation during migration, chemical bacterial alteration) and it is possible to use the vs Cj/C2 diagram (Fig. 13) to distinguish
483
MUD VOLCANISM ORIGIN AND PROCESSES IN TRINIDAD Table 1. Gas analyses 8C3
8iC4
8nC4
3CO2
Mud vol.
8C1
5C2
Marac Palo seco Palo seco Los Iros Lagon bouffe Pitch Lake Pitch 2 Digity Piparo Devil's Woody ard Moruga Bouffe Rock Dome Colombus Colombus Cascadou Cascadou
some of these processes (Prinzhofer & Pernaton 1997). This suggests that only three of the analysed gas samples (one sample from the Columbus group, one sample from Devil's Woody ard and also the gas collected at Pitch lake, which is not a real mud volcano but a crater filled by an oil lake) could have suffered a slight bacterial contamination. In all the other cases, a mixing hypothesis between bacterial and thermogenic gas must be rejected because in these cases the bacterial end member would have methane 613C between —52--33 %o, which are too heavy values, incompatible with a bacterial origin. Therefore, it is considered that most of the analysed gas samples have a strictly thermogenic origin. The dryness of the gas would be due to a segregation process, which probably occurred during its migration from the depth to the surface (adsorption on the solid grains of the mud, and solubility processes). The concentration in C2+ is higher in the sites where eruptions occurred recently (Piparo and Devil's Woodyard, Columbus). It is assumed that adsorption occurs mainly during quiescence phases and that C2+ is released only during and after catastrophic eruptions. The maturity of the gas is difficult to define precisely because of the segregation processes mentioned above. This thermogenic gas could have been generated in the oil window. In the case of very recent (Neogene) gas generation, as observed in Trinidad, high flows of thermogenic gas could have been generated at temperature around 150°C, similar to the equilibrium temperature of the deep reservoir that partly has fed the mud volcanoes. The chemical and isotopic composition of the gas suggests a cogenetic origin with the hydrocarbon
fields of Trinidad, which both exhibit notably atypical heavy values of 513C (CO2). Some values are approaching 20%c, which is very unusual in potential sources of CO2 in sedimentary basins. It is now well established that the source rock of the hydrocarbon fields of southern Trinidad is of Cretaceous age (Gautier and Naparima Hill formations; Rodriguez 1988;Talukdar^a/. 1989,1990,1995; Lawrence et al 1991; Persad et al 1993; Heppard et al 1998, Requejo et al. 1994). The gas from the mud volcanoes being cogenetic with the gas of the HC fields, we also attribute to this gas a Cretaceous origin. From another point of view, the analysis of noble gas radiogenic isotopes has shown that the gas expelled from the mud volcanoes exhibits lower 40Ar*/20Ne, and 4He/20Ne ratios with respect to the gas within the deep HC reservoirs. This implies that the gas from the mud volcanoes has a shorter residence time than the gas associated with the oil fields (Battani et al. 2001). So, the gas of the mud volcanoes can not be issued from a direct leakage from the HC fields, but would come directly from deeper kitchens. Liquid HC The mud expelled by the Trinidad mud volcanoes is rich in liquid HC (up to more than 10 mg/g). These free HC probably have a similar origin to the oils of the southern Trinidad HC fields issued from Cretaceous source rocks, though the analysis of these products show that they are partly biodegraded. The close association between mud volcano area and the oil field distribution was noted a long time ago (Kugler 1959; Higgins & Saunders 1974). These observations indicate that the petroleum
484
E.DEVILLE£rAL.
Fig. 13. C2/Ci vs 813Cj diagram in which a mixing between two end members is characterized by a straight line. Only three samples have suffered a bacterial contamination, as outlined by the mixing straight line shown in this figure. In the other cases, a mixing process between bacterial and thermogenic gas cannot be explained by the analytical data because the 813C(Cj) bacterial isotopic end member would be too heavy. Most of the mud volcano HC gas have a purely thermogenic origin.
system in Trinidad is still active at the present time, allowing the segregation of HC into the mud volcanic processes.
Discussion and conclusions Classically, it has been suggested that the mud volcanism was initiated in relatively shallow horizons, especially within Miocene shales, either in Trinidad (Kugler 1953,1959; Higgins & Saunders 1974) or in the Barbados prism (Faugeres et al 1989; Henry et al. 1996). In Trinidad, nannofossils found in the mud suggest that the material is systematically polygenic and originating from several levels, from the Cretaceous to the Pliocene. As well, the exotic clasts and breccia extruded during eruptive phases are considered mostly as detached and hydraulic fractured fragments issued from various formations of the tectonic wedge (Palaeocene to Miocene) and not only reworked pebbles initially interbedded within Neogene formation, as proposed by Higgins & Saunders (1974). This suggests a deep origin of the mud volcanism process in Cretaceous and/or Palaeogene formations. This suggests also a progressive incorporation of solid particles issued from various Tertiary formations during the mud ascent. In the offshore studied cores, we did not find evidence of Cretaceous mobilized sediment and most probably, the mud origin is located within MiocenePleistocene sediment in the front of the prism and
Oligocene to Pleistocene horizons (possibly also within Eocene levels) in the centre of the prism. In all cases, this suggests a polygenic origin of the solid particles of the mud, involving deep horizons close to the decollement and the incorporation of various materials (thin grains and clasts) from the stratigraphic pile cross cut by the mud conduits. Moreover, in Trinidad, the fluids expelled by the mud volcanoes have a deep origin, whether the water phase (dehydration of the deeply buried sediments), or the gas phase (mostly thermogenic methane). If the water has effectively reached equilibrium temperature up to 150°C (Dia et al 1999), this means that it probably comes from a depth of up to 5000 m. The gas is issued from the thermogenic cracking of organic-rich levels, probably located within the Cretaceous formations. Indeed, this gas is cogenetic with the oil from the subsurface HC reservoirs of south Trinidad, though this gas exhibits shorter residence times than the gas associated with the oil fields. This means that it cannot come from leakage from the HC reservoirs and probably that it comes directly from deep kitchens. This gas cannot be generated from Neogene formations, which are immature in Trinidad. The depth of origin of the mud volcanic activity is clearly associated with the area where overpressure is expected to be the highest, either along the decollement or within the deep parts of the tectonic wedge. Where extension tectonics is active (PariaCaroni basin, Barbados crest), no active mud volcanic processes are observed. As such, the distribution of active/fossil mud volcanoes provides information about the evolution of the pressure regime within the wedge. The mud expelled by the mud volcanoes onshore Trinidad and in the Barbados prism corresponds to liquefied argillaceous and sandy material from initially undercompacted shales, but also sandstone horizons. Indeed, either on the Trinidad Island or in the cores studied from the offshore, the mobilized sediment includes abundant thin quartz grains (up to more than 90%). This thin-grained quartz is angular and frequently mechanically damaged, which can be related to shearing at great depth during the sedimentary burial and/or hydraulic fracturing processes during the mud volcanism activity. The mud volcanoes would correspond to chimneys of expelled fluids associated with a discontinuous re-equilibrium of the pressure regime within the tectonic wedge. In different tectonic wedges and especially in the Barbados prism NE of Trinidad, it has been proposed that the overpressure increase can in some cases approach or exceed the lithostatic pressure and reach the conditions of hydraulic fracturing (Brown 1990). Some authors have proposed that mud volcanoes have developed either along networks of hydraulic fractures, or along fracture zones
MUD VOLCANISM ORIGIN AND PROCESSES IN TRINIDAD
associated with thrust planes (Brown & Westbrook 1988; Brown 1990). This interpretation is indeed very consistent with the mode of fracturing of the ejected clasts, which have been wrenched out from the borders of the mud conduits. The simple occurrence of overpressure does not explain the initiation of the subsurface sediment mobilization and liquefaction, because overpressure commonly develops without inducing liquefaction. It is probably necessary to invoke other processes, such as fluid circulation (water input) and mechanical agitation susceptible to induce the liquefaction process. Notably, seismic activity could be a favourable factor for liquefaction. Gas segregation driven by capillary forces could also cause the initial disruption of the sediment (Henry et al. 1999). Well-constrained subsurface data suggest that the deep structure of the mud volcanoes includes: (1) a focused deep conduit at depth; (2) a mud chamber; and (3) a superficial outlet leading to the surface vents. The mud chambers intrude relatively shallow (above 1000 m) shaly horizons of the stratigraphic pile, i.e. close or above the top of the abnormal pore pressures (Jones 1965; Dyer & Cosgrove 1992; Heppard et al. 1998), whereas the deep conduits develop below within the zone of abnormal pressure. This probably corresponds to the fact that overpressured fluids at depth are focusing toward localized conduits to escape upward and that in the upper parts, the expulsion of the mud creates a pressure gradient between the mud column and the surrounding shallow zones of normal pressure. It is probably this pressure gradient that is responsible for the mud intrusion, the sill formation and fluid flows diverging from the mud chamber (Fig. 14). If it is so and if the depth of the mud chamber can be identified (the depth of maximum extension of the mud chamber would correspond to the top of the abnormal pressure), this suggests that subsurface structural data, which can be given simply by classic seismic data, could also indicate the spatial distribution of the normal and abnormal pressure zones at depth. The mud chamber is probably partly filled by convective cells of mud in the active mud volcanoes. The incorporation of fragments from the walls of the mud chamber is also probably related to the convection activity and obviously to the catastrophic eruptions. The surface mud and gas vents correspond to the outlets that allow the pressure to equilibrate between the mud chamber and the surface. The surface mud volcanism activity suggests that the pressure tends to equilibrate either during the cycles of permanent phases of moderate effusion of mud and fluids (quiescence regime), or during catastrophic events responsible for the expulsion of huge volumes of mud, clasts and fluids (transient regime). Gas flows observed during the quiescence phases of effusion can be up to several tens of m3 per minute locally and
485
Fig. 14. Conceptual representation of the fluid migration regime around a mud volcano. Pressure profiles are compared along two verticals that correspond to an area around a mud volcano (A) and to the mud volcano conduit (B). Al and Bl correspond to the points located on these verticals at the depth of the mud chamber. A2 and B2 correspond to the points located on these verticals at the depth of the deep conduit.
are in most cases higher than the mudflows. The catastrophic events probably take place when the mud expulsion is obtruded or not efficient enough to allow pressure equilibrium. A threshold effect when fluids are oversaturated is gas is also probable and would be responsible for the massive gas eruptions. Historically, the frequency of the catastrophic events is variable according to the sites, the shorter periods being in the range of several tens of years.
486
E.DEVILLECTAZ,
The fieldwork in Trinidad has benefited from numerous advice from representatives from the energy ministry of Trinidad and Tobago (S. Lashley, C. Roberts, H. Inniss and collaborators). We have also benefited from the very precious help of the oil companies producing in the onshore of Trinidad: Petrotrin (manager: K.C. Look-Yee) and Venture (manager: F. A. Khan), who have made possible the sampling of gas and oil in different major fields in several parts of the island. We have also profited from several discussions with specialists of the petroleum geology of Trinidad, notably C. Persad (director of KPA Group of Companies), M.M. J. Frampton and B. Carr-Brown (Biostrat). We thank also the reviewers N. Yassir and P. Henry for providing helpful comments and suggestions.
Columbia 1. sample rich in calcareous nannofossils with the following association: Coccolithus pelagicus, Coronocyclus nitescens, Cyclicargolithus abisectus, Discoaster deflandrei, D. exilis, Helicosphaera ampliaperta, H. carteri, H. Perchnielseniae, H. euphratis, Reticulofenestra pseudoumbilica, Sphenolithus belemnos, S. heteroAppendix 1 - X-ray diffraction morphus. Early Miocene (zone NN3). Columbia 2. Coccolithus pelagicus, Cyclicargolithus abisectus, C. floridanus, Discoaster deflanTrinidad drei, D. exilis, Helicosphaera carteri, H. perch-nielseniae, Reticulofenestra pseudoumbilica, Moruga (mud sample moruga 1, an. n° FOOJ0657): Sphenolithus belemnos, S. heteromorphus. Midquartz, muscovite and/or illite, albite, chlorite (cli- Miocene (zone NN5). nochlore?), traces of kaolinite, siderite, rutile, motColumbia 3. Coccolithus pelagicus, Cyclicargomorillonite. lithus abisectus, C. floridanus, Discoaster deflanMoruga (mud sample moruga 6, an. n° drei, Dictyococcites dictyodus, Erisonia fenestrata, FOOJ0658): quartz, muscovite and/or illite, albite, Helicosphaera euphratis, Reticulofenestra lockeri, chlorite (clinochlore?), kaolinite, Mg-siderite, rutile, Sphenolithus distentus, S. moriformis, S. predistenmotmorillonite. tus. Late Oligocene (zone NP24). Erin Group - Los Iros (mud sample Los Iros 3, an. Anglais point B6. Cyclicargolithus floridanus, n° FOOJ0659): Quatrz, muscovite and/or illite, C. abisectus, Helicosphera carteri, Sphenolithus albite, chlorite (clinochlore), kaolinite, siderite, heteromorphus, S. abies, Reticulofenestra pseudoanatase, montmorillonite. umbilica, Discoaster deflandrei, D. exilis, CycloDevil's Woodyard (mud sample devil's 6, an. n° coccolithus premacintyreL Middle Miocene (zone FOOJ0660): quartz, muscovite and/or illite, albite, NN5). chlorite (clinochlore?), kaolinite, siderite, anatase, Anglais point B12. Coccolithus pelagicus, calcite, montmorillonite. Fasciculithus tympaniformis, Discoaster multiradiatus, Toweius eminens, Ellipsolithus macellus, Cruciplacolithus tenuis, Sphenolithus anarhopus, Barbados accretionary prism (offshore) Ericsonia sp., Cyclococcolithus robustus. Upper Palaeocene (zone NP9). Core KS 87102 (510 cm). Quartz (abundant), muscoDevil's woodyard 3. Coccolithus pelagicus, vite, illite, albite, K-feldspar, chlorite (clinochlore?), Cyclicargolithus abisectus, C. floridanus, CycloBerthierine, kaolinite (abundant), calcite, siderite, coccolithus macintyrei, Discoasterbrouweri, D. rutile, anatase, montmorillonite, halite. deflandrei, D. exilis, Helicosphaera euphratis, Core KS 85069 (290 cm). Quartz (abondant), carteri, H. perch-nielseniae, Reticulofenestra muscovite et/ou illite, albite, feldspath potassique, pseudoumbilica, Sphenolithus abies, S. heteromorchlorite (clinochlore?), Berthierine, kaolinite (abon- phus. Mid-Miocene (zone NN5). dante), siderite, rutile, anatase, montmorillonite, Devil's woodyard Bl. Coccolithus pelagicus, halite. Reticulofenestra pseudoumbilica, Discoaster variCore KS 87117 (485 cm). Quartz (abundant), abilis, Discoaster hamatus, Discoaster brouweri, muscovite, illite, chlorite (clinochlore?), kaolinite Discoaster calcaris, Sphenolithus abies, Cyclo(abundant), calcite, siderite, rutile, anatase, mont- coccolithus macintyrei.Upper Miocene (zone NN9). morillonite, gypsum, halite. Devil's woody ard B7. Reticulofenestra pseudoumbilica, Sphenoliyhus abies, Discoaster variabilis, Discoaster challenged, Discoaster pentaradiatus, D. brouweri, Cyclococcolithus macintyrei. Upper Miocene-Lower Pliocene. Piparo B13, B21, B24, B28, B29. Nannoconus
MUD VOLCANISM ORIGIN AND PROCESSES IN TRINIDAD
colomii, N. kamptnerii, N. circularis, N. grandis, Calcicalathina oblongata, Watnaueria barnesae, W. communis, W. britannica, Cyclagelosphaera margerelli, Parhabdolithus embergeri, P. asper, Cruciellipsis chiastia, Polycostella senaria, Micranthilithus obtusus, Lithraphidites carniolensis, Manivitelle pemmatoidea. Lower Barremian. Piparo BIO. Watznaueria barnesae, Micula staurophora, M. murus. Upper Maastrichtian. Piparo 1. Chiamolithus solitus, Cyclococcolithus formosus, Discoaster barbadiensis, D. nonaradiatus, Retculofenestra cf. umbilica, R. unmbilica, Sphenolithus moriformis, S. radians. Mid-Eocene zone (zone NP14). Piparo Bl. Cyclicargolithus floridanus, C. abisectus, Coccolithus pelagicus, Helicosphaera euphratis, H. cf. recta, Sphenolithus moriformis, S. predistentus, Discoaster deflandrei, D. tani nofdifer. Late Oligocene (zone NP24). Piparo B22. Cyclicargolithus floridanus, C. abisectus, Coccolithus pelagicus, Sphenolithus moriformis, S. distentus, S. ciperensis, Helicosphaera euphratis, Reticulofenestra pseudoumbilica, Discoaster deflandrei, Dictyococcites dictyodus. Late Oligocene (NP24). Piparo 2. Coccolithus pelagicus, Cyclicargolithus abisectus, C. floridanus, Discoaster deflandrei, Dictyococcites dictyodus, D. tani nodifer, Helicosphaera compacta, H. euphratis, H. perch-nielseniae, Triquetrorhabdulus carinatus, Sphenolithus ciperoensis, S. moriformis, S. predistentus. Late Oligocene (zone NP25). Mud Devil's Woodyard. This mud sample shows a mixing of numerous species of nannofossils, mostly with associations of Early-Mid-Miocene age (zones NN4-5), with some Late Miocene species of the NN9 zone (Discoaster hamatus, Catinaster coalitus, Helicosphaera philippinensis, Reticulofenestra pseudoumbilic Discoaster calcaris, D. pentaradiatus), and some rares species of Cretaceous and Paleogene age. Cascadoux. This mud sample shows a mixing of numerous species of nannofossiles, mostly with associations of Early-Mid-Miocene age (zones NN3-5), with also typically Oligocene species (Helicosphaera recta, Triquetrorhabdulus carinatus).
Mud volcanoes from the offshore (Barbados Prism) Core KS 85-54, 390 cm. Nannofossils are rare and relate of a mixing of Pliocene-Pleistocene species.
487
Core KS 85-56,470 cm. Nannofossils are rare and show some traces of dissolution. The determined associations are the following '.Cyclococcolithus leptoporus, Gephyrocapsa oceanica, Rhabdosphaera claviger, Emiliania huxleyi, Syracosphaera pulchra, Helicosphaera carteri, Umbilicosphaera mirabiliis, Thoracosphaera heimii, Ceratolithus cristatus, Umbellosphaera tenuis which correspond to a Pleistocene-Holocene age (zone NN 21) Core KS 85-59, 280 cm. Nannofossils from the Late Pleistocene-Holocene (zone NN 21) are frequent with the following association: Gephyrocapsa oceanica, Syracosphaera pulchra, Umbilicosphaera mirabilis, Emiliania huxleyi, Rhabdosphaera claviger, Helicosphaera carteri, Umbellosphaera tenuis, Discolithina japonica, et Cyclococcolithus leptoporus. Core KS 85-69, 210 cm. This sample shows a mixing of Pliocene-Pleistocene species. Pliocene: Discoaster brouweri, D. pentaradiatus, D. surculus, Cyclococcolithus macintyrei, Pseudoemiliania lacunosa, Reticulofenestra pseudoumbilica, Sphenolithus abies (Early Pliocene, zone NN 15). Pleistocene: Helicosphaera carteri, Gephyrocapsa oceanica, Cyclococcolithus leptoporus, Ceratolithus cristatus, Discolithina japonica, Syracosphaera pulchra, Umbilicosphaera mirabilis, Emiliania huxleyi, (zone NN 21) Core KS 87-102, 430, 470 and 490 cm. Nannofossils are rare, with Early Pliocene forms (zone NN 15): Pseudoemiliania lacunosa, Reticulofenestra pseudoumbilica, Sphenolithus abies, Discoaster brouweri, D. pentaradiatus, D. surculus D. tamalis, D. variabilis. Also some rare forms from the Late Miocene were observed (zone NN 11): Discoaster calcaris, D. quinqueramus, Amaurolithus delicatus, Scyphosphaera globulus. Core KS 87-117, 300, 320, 360, 400, 420, 445 and 485 cm. Nannofossils are abundant and wellpreserved, with typical forms from the Early Pliocene (zone NN 15): Discoaster asymmetricus, D. brouweri, D. pentaradiatus, D. surculus, D. tamalis, D. variabilis, Cyclococcolithus macintyrei, C. rotula, Reticulofenestra pseudoumbilica, Sphenolithus abies, Pseudoemiliania lacunosa and Helicopshaera carteri. Core KS 90-106, 240, 270, 300 and 310 cm. This core is rich in Late Pleistocene forms (zone NN 20): Umbilicosphaera mirabilis, Syracopshaera pulchra, Scapholithus Jossilis, Ceratolithus cristatus, Gephyrocapsa ericsonii, G. oceanica, G.carribbeanica, Thoracosphaera heimii, Helicosphaera carteri, Cyclococcolithus leptoporus, Rhabdosphaera claviger, Helicosphaera inversa. Some rare forms from the Eocene, Oligocene and Early Miocene are also present. Core KS 90-122, 150-180 cm. These samples are
488
E. DEVILLE £7AL.
rich in well-preserved nannofossils from the Late Pleistocene-Holocene (zone NN 21) with the following associations: Emiliania huxleyi, Gephyrocapsa oceanica, G. aperta, Syracosphaera pulchra, S. mediterranea, Rhabdosphaera claviger, R. stylifer, Helicosphaera carteri, Discolithina japonica, Umbilicopshaera mirabilis, Cyclococcolithus leptoporus, Umbellosphaera tennis, Thoracosphaera heimii, Scapholithus fossilis. Some rare forms of Cyclicargolithus abisectus, Discoaster variabilis, and Sphenolithus abies from the Miocene-Early Pliocene are also present. The samples between 200 and 220 cm contain Late Oligocene nannoforms (zone NP 25) with Sphenolithus ciperoensis, S. moriformis, Dictyococcites dictyodus, Cyclicargolithus abisectus, C. floridanus, Discoaster deflandrei, Coccolithus pelagicus and Ericsoniafenestrata.
References AHMAD, R. 1991. Structural styles in Trinidad. Field-trip guide. In: GILLEZEAU, K. A. (ed.) Transactions of the second Geological Conference of the Geological Society of Trinidad and Tobago. The Geological Society of Trinidad and Tobago, (Port-of-Spain), 244-265. ARNOLD, R., MACREADY, G.A. 1956. Island-forming mud volcano in Trinidad, British West Indies. American Association of Petroleum Geologists Bulletin, 40, 2748-2758. BARBOZA, S.A. & BOETTCHER, S.S. 2000. Major and trace element constraints on fluid origin, Offshore Eastern Trinidad. Special Publication of the Geological Society of Trinidad and Tobago and the Society of Petroleum Engineers, TG03, lip. BARR, K.W. 1953. The mud volcanoes of Trinidad. Caribbean Quarterly 3(2), 80-85. BARR, K.W. & SAUNDERS, J.B. 1968. An outline of the geology of Trinidad. In: SAUNDERS, J.B. (ed.) Transactions of the Fourth Carribean Geolological Conference. Caribbean Printers, (Port-of-Spain), 1-10. BATTANI, A., PRINZHOFER, A., BALLENTINE, C. J. & DEVILLE, E. 2001. Identifying the relationship between mud volcanoes, grounwater flow and hydrocarbon reservoirs in Trinidad: a combined noble gas and geochemistry case study. Eos, 81(48), 442-443. BEHRMANN, J.H. 1991. Conditions for hydrofracture and the fluid permeability of accretionary wedges. Earth and Planetary Science Letters, 107, 550-558. Buu-DuvAL, B., LE QUELLEC, P., MASCLE, A., RENARD, V. & VALERY, P. 1982. Multi-beam bathymetric survey and high resolution seismic investigations on the Barbados ridge complex (Eastern Caribbean). Tectonophysics, 80, 275-304. BIRCHWOOD, K.M. 1965. Mud volcanoes in Trinidad. Institute of Petroleum Review, 19,164-167. BOWER, T.H. 1965. Geology of Texaco Forest reserve field, Trinidad, W.I. In: SAUNDERS, J.B. (ed.) Transactions of the Fourth Carribean Geolological Conference. Caribbean Printers, (Port-of-Spain), 75-86.
BROWN, K.M. 1990. The nature and hydrologic signifiance of mud diapirs and diatremes for accretionary systems. Journal of Geophysical Research, 95, B6, 8969-8982. BROWN, K.M. & WESTBROOK, G.K. 1987. The tectonic fabric of the Barbados ridge accretionary complex. Marine and Petroleum Geology, 4, 71-81. BROWN, K.M. & WESTBROOK, G.K. 1988. Mud diapirism and subcretion in the Barbados Ridge accretionary Complex: The role of fluids in accretionary processes. Tectonics, 7(3), 613-640. CASTREC, M., DIA, A.N. & BOULEGUE, J. 1996. Major- and trace element and Sr isotope constraints on fluid circilation in the Barbados accretionary complex. Part II: circulation rates and fluxes. Earth Planetary Science Letters, 145,487-499. CARR-BROWN, B. & FRAMPTON, J. 1979. An outline of the stratigraphy of Trinidad. Transactions of the fourth Latin American Geological Conference Field Guide, Printed by Trinidad & Tobago Printing and packaging limited, (Port-of-Spain), 7-9. DE METS, C., GORDO, R.G., ARGUS, A.F. & STEIN, S. 1990. Current plate motions. Journal of Geophysical International, 101,425-478. DEVILLE E. 2000. Evidence for extension tectonics in the crest of the Barbados accretionary prism. Special Publication of the Geological Society of Trinidad and Tobago and the Society of Petroleum Engineers, TC06. DIA, A.N., CASTREC, M., BOULEGUE, J. & BOUDOU, J.P. 1995. Major and trace element and Sr isotope constraints on fluid circulations in the Barbados accretionary complex. Part 1: Fluid origin. Earth and Planetary Sciences Letters, 134, 69-85. DIA, A.N., CASTREC-ROUELLE, M., BOULEGUE, J. & COMEAU, P. 1999. Trinidad mud volcanoes: where do the expelled fluids come from. Geochimica CosmochimicaActa, 63,1023-1038. DYER, B.L. & COSGROVE, P. 1992. Penal/Barrackpore field, West Indies: south Trinidad basin, Trinidad. American Association of Petroleum Geologists, Treatise of Petroleum Geological Atlas of oil and gas fields, 139-157. FAUGERES, J.C., GONTHIER, E., PONS, J.C., PARRA, M. & PUJOL, C. 1989. Les apports du diapirisme argileux dans la sedimentation d'un prisme d'accretion: la ride de la Barbade au sud-est des Petites Antilles. Compte rendus de Vacademie des Sciences, Paris, serie II, 308, 747-753. FAUGERES, J.C., GONTHIER, E., GRIBOULARD, R. & MASSE, L. 1991. Quaternary sandy deposits and canyons on the Venezuelan margin and South Barbados accretionary prism. Marine Geology, 110, 115-142. GONTHIER, E., FAUGERES, J.C., BOBIER, C., GRIBOULARD, R., HUYGHE, P, MASSE, L & PUJOL, C. 1994. Le Prisme d'accretion tectonique Sud-Barbade. Bilan des donnees recueillies au cours des missions Caracolante n, Diapicar, Diapisar et Diapisub. Revue Aquitaine Oceans, no. 1, 105 p. GRIBOULARD, R., BOBIER, C., FAUGERES, J.C. & VERNETTE, G. 1991. Clay diapiric structures within the strike-slip margin of the southern leg of the Barbados prism. Tectonophysics, 192, 383-400. GRIBOULARD, R., BOBIER, C., FAUGERES, J.C., HUYGHE, P., GONTHIER, E., DUNNE, F. & WELSH, R. 1998. Recent
MUD VOLCANISM ORIGIN AND PROCESSES IN TRINIDAD tectonic activity in the south Barbados Prism, deeptowed side-scan sonar imagery. Tectonophysics, 284, 79-99. HEDBERG H.D. 1980. Methane generation and petroleum migration. In: ROBERTS, W.H. Ill, CORDELL, RJ. (eds) Problems of the Petroleum Migration. American Association of Petroleum Geologists Bulletin, Stud. Geol. 10,179-206. HENRY, P., LE PICHON, X. et al. 1996. Fluid flow in and around a mud volcano field seaward of the Barbados accretionary wedge: results from Manon cruise. Journal of Geophysical Research, 101, B9, 20297-20323. HENRY, P., THOMAS, M. & CLENNELL, M.B. 1999. Formation of natural hydrates in marine sediments. Journal of Geophysical Research, 104, BIO, 23005-23022. HEPPARD, P.D., GANDER, H.S. & EGGERTSON, E.B. 1998. Abnormal pressure and the occurrence of hydrocarbons in offshore eastern Trinidad, West Indies. American Association of Petroleum Geologists Memoir,lQ, 215-246. HIGGINS, G.E. 1996. A history of Trinidad oil. Trinidad Express Newspapers Limited, (Port-of-Spain), Express Production House. 498p. HIGGINS G.E. & SAUNDERS J.B. 1967. Report on 1964 Chatham mud island, Erin Bay, Trinidad, West Indies. American Association of Petroleum Geologists Bulletin, 51,55-64. HIGGINS G.E. & SAUNDERS J.B. 1974. Mud Volcanoes. Their nature and origin. Verhandlungen Naturforschenden Gesselschaft in Basel, 84,101-152. HUBBERT, M.K. & RUBEY, WW. 1959. Role of fluid pressure in the mechanics of overthrast faulting, Geological Society of America Bulletin, 70,115-166. HUGUEN, C., BENKHELIL, J., GIRESSE, P., MASCLE, J., MULLER,
C.,
WOOSIDE,
J.,
ZlTTER,
T.,
& THE
MEDINAUT SCIENTIFIFIC PARTY. 2001. Echantillous Rouchex provenant de volcano de bouedle Mediterranee orientele. Oceanologica acta, 24 (4), 349-360. JOLLIVET, D., FAUGERES, J.C., GRIBOULARD, R., DESBRUYERES, D. & BLANC, G. 1990. Composition and spatial distribution of a cold seep comminity on the South Barbados accretionary prism: tectonic, geochemical and sedimentary context. Progress in Oceanography, 24,25-45. JONES, H.P 1965. The geology of the Herrera sands in the Moruga West oilfield of south Trinidad. In: SAUNDERS, J.B. Transactions of the Fourth Carribean Geolological Conference, (ed.), (Port-of-Spain), 91-100. KERR, P.P., DREW, I.M. & RICHARDSON, D.S. 1970. Mud volcanoes clay, Trinidad, West Indies. American Association of Petroleum Geologists Bulletin, 54, 2101-2110. KOPF, A., ROBERTSON, A.H.R & VOLKMANN, N. 2000. Origin of mud breccia from the Mediterranean Ridge accretionary complex based on evidence of maturity of organic matter and related petrographic and regional tectonic evidence. Marine Geology, 166, 65-82. KUGLER, H.G. 1953. Jurassic to recent sedimentary environments in Trinidad. Bulletin der Vereinigung
489
Schweizerisches Petroleum -Geologen und -Ingenieu, 20,27-60. KUGLER, H.G. 1959. Geological map of Trinidad (1/100000). Petroleum Association of Trinidad. KUGLER, H.G. 1965. Sedimentary volcanism. In: SAUNDERS, J.B. (ed.) Transactions of the Fourth Carribean Geolological Conference. Caribbean Printers, (Port-of-Spain), 11-13. LANCE, S., HENRY, P., LE PICHON, X., LALLEMANT, S., CHAMLEY, H., ROSTEK, R, FAUGERES, J.C., GONTHIER, E. & OLU, K. 1998. Submarine study of mud volcanoes seaward of the Barbados accretionary wedge: sedimentology, structure and rheology. Marine Geology, 145,255-292. LAWRENCE, S.R., SOULSBY, A. & CHELDS, V.H. 1991. A new framework for hydrocarbons exploration in Trinidad. In: GILLEZEAU, K.A. (ed.) Transactions of the second Geological Conference of the Geological Society of Trinidad and Tobago. The Geological Society of Trinidad and Tobago, (Port-of-Spain), 170-176. LEONARD, R. 1983. Geology and hydrocarbon accumulation, Columbus basin, offshore Trinidad. American Association of Petroleum Geologists Bulletin, 67, 1081-1093. MARTIN, J.B., KASTNER, M., HENRY, P., LE PICHON, X. & LALLEMANT, S. 1996. Chemical and isotopic evidence for sources of fluids in a mud volcano field seaward of the Barbados accretionary wedge. Journal of Geophysical Research, 101,20,325-20,345. MASCLE, A., ENDIGNOUX, L. & CHENNOUF, T. 1990. Frontal accretion and piggyback basin development at the southern edge of the Barbados Ridge accretionary complex. In\ MOORE J.C., MASCLE A. et al. (eds) Proceeding ODP, Scientific Results, 110, College Station, TX(ODP), 409-422. PAYNE, N. 1991 An evaluation of post-Middle Miocene geological sequences, offshore Trinidad. In: GILLAZEAU, K.A. (ed.) Transactions of the second Geological Conference of the Geological Society of Trinidad and Tobago. The Geological Society of Trinidad and Tobago, (Port-of-Spain), 70-87. PERSAD, K.M., TALUKDAR, S.C. & Dow, W.G. 1993. Tectonic control in source rock maturation and oil migration in Trinidad and implications for Petroleum Exploration. In: Proceedings, 13th Annual Research Conference of the Society for sedimentary Geology, published by the SEPM, 237-249. PRINZHOFER, A. & PERNATON, E. 1997. Isotopically light methane in natural gas: bacterial imprint or diffusive fractionation? Chemical Geology, 142,193-200. PRINZHOFER, A., MELLO, M.R. & TAKAKI, T. 2000. Geochemical characterization of natural gas: a physical multivariable approach and its applications in maturity and migration estimates. American Association of Petroleum Geologists Bulletin, 84,1152-1172. REQUEJO, A.G., SASSEN, R., WIELCHOWSKY, C. & KLOSTERMAN, M.J. 1994. Geochemical characterization of lithofacies and organic facies in Cretaceous organic-rich rocks from Trinidad east Venezuela basin. Organic Geochemistry, 22 (3-5), 441-459. ROBERSTON, R.P. & BURKE, K. 1989. Evolution of the southern Caribbean plate boundary vicinity of Trinidad and Tobago. American Association of Petroleum Geologists Bulletin, 73 (1), 490-509.
490
E.DEVILLE£TAL.
RODRIGUES, K. 1985. Thermal history modeling in petroleum exploration: examples from Southern Trinidad. In: Proceedings of the 1st Geological Conference of the Geological Society of Trinidad and Tobago: Geological Society of Trinidad and Tobago (Port-ofSpain), 237-241. RODRIGUEZ, K. 1988. Oil source bed recognition and crude oil correlation, Trinidad, West Indies. Organic Geochemistry, 13 (1-3), 365-371. RODRIGUEZ, K. 1989. Signifiance of geothermal gradients in petroleum exploration in Trinidad. Proceedings of the 12th Caribbean Geological Conference, (Sainte Croix), Edited by the Miami Geological Society, 444-452. RUTLEDGE, A.K. & LEONARD, D.S. 2001. Role of multibeam sonar in oil and gas exploration and development. Offshore Technology Conference 12956, 12 p. SENN, A. 1940. Paleogene of Barbados and its bearing on history and structure of Antillean-Caribbean region. American Association of Petroleum Geologists Bulletin, 24,9,1548-1610. SPEED, R.C. 1985. Cenozoic collision of the Lesser Antilles arc and continental South America and the origin of the El Pilar fault. Tectonics, 4(1), 41-69. SPEED, R.C., BARKER, L.H. & PAYNE, P.L.B. 1991. Geologic and hydrocarbon evolution of Barbados. Journal of Petroleum Geology, 14 (3), 323-342. STEIN, S., DE METS, C, GORDON, R.G., BRODHOLT J., ARGUS, D.F., ENGEL, J., LINDGREEN, P., STEIN, C., WERNS, D. & WOODS, D. 1988. A test of alternative Caribbean plate relative motion models. Journal of Geophysical Research, 93, 3041-3050. TALUKDAR, S., GALLANGO, W, CASSANI, R & VALLEJOS, C. 1989. Upper Cretaceous source rocks of northern and northwestern South America. American Association of Petroleum Geologists Bulletin, 73, 3,418. TALUKDAR, S., Dow, W, CASTANO, J. & PERSAD, K.P. 1990. Geochemistry of oils provides optimism for deeper exploration in Atlantic off Trinidad. Oil and Gas Journal 88 (46), 118-122. TALUKDAR, S., Dow, W, CASTANO, J. & PERSAD, K.P. 1995. Origin and evolution of oils and condensates from onshore and offshore Trinidad: Their exploration sig-
nificance. In: WINSTON, A., ANTHONY, P. & VICTOR YOUNG, O. (eds) Transactions of 3rd Geological Conference of the Geological Society of Trinidad and Tobago and 14th Carribean Geological Conference. Geological Society of Trinidad and Tobago (Port-ofSpain), 78. TYSON, L., BABB, S. & DYER, B. 19910. Middle Miocene tectonics and its effects on Late Miocene sedimentation in Trinidad. In: GILLEZEAU, K.A. (ed.) Transactions of the second Geological Conference of the Geological Society of Trinidad and Tobago. Geological Society of Trinidad and Tobago (Port-ofSpain), 26-40. TYSON, L. & WINSTON, A. 1991&. Cretaceous to Middle Miocene sediments in Trinidad. Field-trip guide. In: GILLEZEAU, K.A. (ed.) Transactions of the second Geological Conference of the Geological Society of Trinidad and Tobago. Geological Society of Trinidad and Tobago, (Port-of-Spain), 266-277. VALERY, P., NELY, G., MASCLE, A., BLTU-DUVAL, B., LE QUELLEC, P. & BERTHON J.L. 1985 Structure et croissance d'un prisme d'accretion tectonique proche d'un continent: la ride de la Barbade au sud de 1'arc antillais. In: MASCLE, A. (ed.) Geodynamique des Cara'ibes, edition Technip, (Paris), 173-186. VAN SOEST, M.C. 2000. Sediment subduction and crustal contamination in the Lesser Antilles Island Arc. The geochemical and isotopic inprint on recent lavas and geothermal fluids. PhD, University of Amsterdam. WHARTON, G. & HUDSON, D. 1995. Report on a recent eruption of the mud volcano Devil's Woodyard, South Trinidad. In: WINSTON, A., ANTHONY, P. & VICTOR YOUNG, A. (eds) Transactions of 3rd Geological Conference of the Geological Society of Trinidad and Tobago and 14th Carribean Geological Conference. Geological Society of Trinidad and Tobago, (Port-ofSpain), 85 YASSIR, N. 1989. Mud volcanoes and the behaviour of overpressured clays and silts. PhD, University of London. YASSIR, N. 1987. Mud volcanoes: evidence of neotectonic activity. Memoir of the Geological Society of China, 9,513-524.
Boron and boron isotopes as tracers for diagenetic reactions and depth of mobilization, using muds and authigenic carbonates from eastern Mediterranean mud volcanoes A. DEYHLE1, ACHIM J. KOPF1, & G. ALOISI2 1
SCRIPPS Institution of Oceanography, UCSD, 9500 Oilman Drive, La Jolla, CA 92093-0212, USA (e-mail: [email protected]) 2 GEOMAR Research Centre, University Kiel, Wischhofstrasse 1-3, 24148 Kiel, Germany Abstract: Authigenic carbonates and muds from six mud volcanoes in the eastern Mediterranean Sea were recovered during the French/Dutch MEDINAUT cruise utilizing the submersible Nautile in November 1998. The mud volcanoes are active seafloor vents in two areas at the plate boundary between the converging African and Eurasian Plates: the Mediterranean Ridge accretionary prism near Crete (Greece) and the Anaximander Mountains south of Turkey. B contents and 5UB signatures were measured with the aim of identifying the diagenetic processes and source depths of the material in the collision zone. B concentrations of the carbonate precipitates cover a range of 8-45 ppm and vary isotopically from +15.6 to +22.9%c (corresponding to a parent solution of 34.9-42.2%o at pH 7). Both the B-enrichment and a 6nB valve slightly lower than seawater suggest the mud domes originate from a moderately deep fluid source, with local admixture of seawater. B contents and 6nB of the mud show distinct differences between the areas: the Mediterranean Ridge mud domes have lower B contents and higher 5nB (average 3.9%c) compared to Anaximander Mountains mud volcanoes (5nB average — 0.6%c). These B results attest the release of structurallybound B from clay mineral lattices, probably due to stronger deformation near Turkey. These mudstones, which had previously been affected by deep-seated thrusting beneath the Antalya Complex, may have been liquefied and remobilized in their present setting. By contrast, the mud on the Mediterranean Ridge represents offscraped clay-rich strata that was incorporated into the large accretionary wedge.
Mud volcanism is a global phenomenon that occurs most frequently in compressional tectonic settings where rapidly deposited, clay-rich sediments accumulate (Higgins & Saunders 1974; Brown & Westbrook 1988; Moore & Vrolijk 1992). Some of these mud volcanoes may have exceptionally high fluid flux rates (Kopf el al. 2001) and mobilize both the mud and its fluids at depths inaccessible to sampling. Hence, the surface expression is a good "tectonic window" to study deep diagenetic processes. Mud volcanoes from two different settings in the eastern Mediterranean Sea were chosen for this study. The eastern Mediterranean Sea, where compression is associated with the initial stages of continent-continent collision of Africa with Eurasia, has long been known to have mud volcanoes on its most prominent feature, the Mediterranean Ridge accretionary wedge (e.g. Cita et al 1981; Camerlenghi et al 1995). Unlike many other accretionary prisms, the Mediterranean Ridge is characterized by mud domes far (>150 km) behind the deformation front, where the wedge is thrust over its continental northern backstop (Fig. 1; Kopf et al 1998; Mascle et al 1999). Of the two selected regions, the western one was the Olimpi Mud Volcano Field on the Mediterranean Ridge, the latter being a 200 km
wide accretionary prism along the Hellenic Arc (Fig. 1). Two of the mud volcanoes there have already been drilled during Ocean Drilling Program (ODP) Leg 160 (Robertson et al 1996) and have shed important light on their evolution through time. The eastern region at the junction between the Mediterranean Ridge accretionary wedge and the Florence Rise, where orthogonal subduction has switched to left-lateral strike slip (Le Pichon & Angelier 1979), is known as the Anaximander Mountains area south of the Isparta Angle (e.g. Woodside et al 1998). Here, a group of three independent seamounts of proposed continental origin are now caught up in the plate convergence between Africa and Eurasia (Aloisi et al 2000). Gas seepage and mud volcanism has been observed at the flanks of two of the seamounts. In this paper, boron isotopes and concentrations of authigenic carbonates and muds from several mud volcanoes of each region are investigated to identify the fluid sources that triggered mud volcanism. The results from authigenic carbonates can be related to their parent-solution compositions in order to characterize the pore fluids from which they precipitated (Deyhle et al 2001). Because B geochemistry has been shown to be a powerful tool in identifying diagenetic processes, the aim of this paper is to
Fig 1. Map of the Eastern Mediterranean, showing the main structural elements. The arrow indicates the main plate kinematic vector at present. Mud volcano domains visited for this study are highlighted. The line across the Mediterranean Ridge reflects the location of the seismic profile in Fig. 2a. See legend for explanation.
evaluate in which of the two areas muds and fluids 1500 km long, Mediterranean Ridge accretionary are mobilized from greater depth: the Mediterranean prism (see below). It comprises offscraped and accuRidge with orthogonal subduction-accretion, or the mulated sediments as a result of the ongoing converAnaximander Mountains, which involve oblique gence between Africa and Eurasia, which has strike-slip movements further east near the Florence reduced the Mediterranean Sea to a 150-250 km Rise (e.g. Robertson 1998). This approach may also wide basin between Libya and Crete (Fig. 1). The help to understand how deep faults root in the areas, subduction zone in the region south of mainland in relation to the local tectonic stress field in the Greece, the island of Crete and its eastern prolongation towards Turkey is generally referred to as the Hellenic and Cyprean Arcs. Hellenic Arc (e.g. Le Pichon & Angelier 1979). It terminates in the deepest part of the eastern Geological setting Mediterranean, the Rhodes Basin adjacent to the Anaximander Mountains (Fig. 1; see Woodside & The eastern Mediterranean Sea is located between Limonov 1997). East of this submarine mountain the converging continental plates of Africa and range, which is bounded in the south by a depresEurasia (Fig. 1). Allowing for the spreading of the sion, the Florence Rise occurs as the westernmost Aegean Sea (i.e. the back arc basin to the eastern extension of the Cyprean Arc further east (Fig. 1). Mediterranean), the net convergence between Africa Both the Hellenic and Cyprean Arcs are characterand Eurasia is about 30-40 mm/a (e.g. Le Pichon et ized by portions of oblique subduction and strikeal 1995). The Eastern Mediterranean's Ionian and slip (to transpressional) movement, evidenced by a Levantine Basins are possible relics of the Mesozoic series of depressions referred to as 'trenches'. While oceanic Tethys sensu stricto, which have been pre- features like the western Mediterranean Ridge or the served despite Alpine orogenesis (e.g. Hsu & Florence Rise have been accumulated roughly perBernoulli 1978; Dercourt et al 1986). By contrast, pendicular to the plate boundary due to convergence the Aegean Sea back arc basin developed only 13 (or parallel to the plate kinematic vector; Fig. 1), m.y. ago (Le Pichon & Angelier 1979). the Pliny and Strabo 'trenches' on the eastern The entire region is characterized by a complex Mediterranean Ridge are SW-NE striking and repplate kinematic and tectonic pattern. One of the most resent left-lateral strike-slip faults (Fig. 3b; prominent features is the arcuate, more than Robertson 1998).
B GEOCHEMISTRY IN MEDITERRANEAN MUD VOLCANOES
493
Fig 2. (a) Line drawing from seismic reflection profile across the central domain of the Mediterranean Ridge (from Kopf et al 2001). (b) Blow up of seismic reflection profile showing a small mud volcano juxtaposing a backthrust fault (from Mascle et al 1999).
Mediterranean Ridge The Mediterranean Ridge varies in width along strike and displays various styles and degrees of deformation. In the main collision zone between Libya and Crete (the outer forearc high of the overriding Eurasian Plate), the entire abyssal plain has undergone subduction. The prism is thrust onto the Libyan margin to the south and backthrust over the Cretan margin to the north (Mascle et al 1999). The progressive compression is accommodated not only by thrusting at the base of the prism toe- and backthrusting over the apex, but by internal conjugate faulting in the central domain (Chaumillon & Mascle 1997; Huguen et al 2001). The most prominent structural features are backthrusts that have caused the formation of a topographic escarpment separating the central domain from the Inner Ridge (Fig. 2a). The accentuation of collision in this area has been proposed to cause retardation of convergence in the central part of the Hellenic subduction zone, having started some 3-5 m.y. ago (Le Pichon et al 1995). Evidence for uplift of the entire Mediterranean Ridge is also provided by the absence of a large cover of evaporites from the latest Miocene (5-6 Ma; see Montadert et al 1978). The
area with the backthrusts in the landward part of the prism coincides with abundant mud volcanism (Kopf et al 1998; Figs. 2b and 3a). Further east, the several km thick hemipelagic sediments of the Herodotus abyssal plain have been subducted beneath a much wider prism than between Libya and Crete. The change from gentle folding at the toe to more enhanced deformation and faulting in the cental part of the prism is almost gradual and large mudflows have been reported previously (Huguen et al 2001; Kopf et al 2001). Here, transtensional tectonics have caused the offset of the Pliny-, Strabo-, and Matapan 'trenches' during the Neogene. The Finike Basin further east has evolved as an extensional feature due to the rifting of the Anaximander Mountains (Figs 1 and 3b; Woodside 1997). The area with the most abundant mud domes is located south of the island of Crete (Figs 1 and 3a). Here, domes of several km diameter and heights of several 100 m above the surrounding seafloor are commonly linked to subsurface reverse and backthrust faults (Camerlenghi et al 1995; Mascle et al 1999). The features have shown episodic, vigorous activity over >1 m.y. and are composed of polymictic mud breccia (Robertson & Kopf 1998a, b). For
494
A. DEYHLE £r/LL
this study, three of the domes have been sampled: Napoli, Milano, and Moscow mud volcanoes (Fig. 3b).
Anaximander Mountains The Finike Basin (Fig. 1) separates the Anaximander Mountains from the margin of southern Turkey. Fault scarps observed onshore, especially the prominent thrust between the Antalya Complex and the Bey Daglari Formation, may be traced southward on bathymetric charts of the seafloor (Fig. 3b; e.g. Woodside & Limonov 1997). The Turkish slope is characterized by an ENE-WSW trending fault scarp, which may reflect rifting of blocks away from Turkey now forming the Anaximander Mountains (Woodside et al 1998). The region is characterized by three topographic highs between the Rhodes Basin in the west, the Finike Basin in the north, the Antalya Basin in the NE and the Levantine Basin in the south (Figs 1 and 3b). The three seamounts have been named after natural philosophers from the area (from west to east): Anaximander, Anaximenes and Anaxagoras (Fig. 3b; Woodside & Limonov 1997). The westernmost mountain is Anaximander, where no mud volcanoes, but gas seepage and vents were observed. The southernmost mountain is Anaximenes with a height of approximately 670 m below sea level (bsl). To the east rises Anaxagoras to a height of 900 mbsl (e.g. Woodside etal. 1998). All three mountains are situated at the junction between the Hellenic Arc to the west (i.e. the PlinyStrabo Trench' system) and the Cyprean Arc to the east near the Isparta Angle (Fig. 1). As a result of the convergence of the sediment-laden Herodotus abyssal plain towards Eurasia, some sediment is thrust over the southern flanks of Anaximander and Anaximenes (Fig. 3b; Woodside & Limonov 1997). Anaxagoras lies on the prolongation of the Florence Rise, which marks the boundary between the African plate and the Anatolian microplate (McKenzie 1972). The Antalya Basin to the NE of the Florence Rise (Fig. 1) is actively subsiding and tilting to the north (Woodside 1977) under compression of Africa. Consequently Anaxagoras is pushed from the against the SW sediments in the Antalya Basin creating a fold belt. Some branches of the transpressive fault system may cut through the Anaximander Mountains and gravity data (Ivanov et al. 1992) indicate a major structural discontinuity between Anaximander and Anaximenes in the west and Anaxagoras in the east (Fig. 3b). This may coincide with the main suture between the Bey Daglari basement and the Antalya Nappe Complex further north in Southern Turkey (Fig. 3b; Collins & Robertson 1998, fig. 1). Furthermore, magnetic data suggests that the Anaximander Mountains are chiefly composed of sedimentary
rocks and ophiolites are not likely to be present, although they are widespread east and NE of the mountains (i.e. in the Lycian Nappe Complex; e.g. Woodside & Limonov 1997; Collins & Robertson 1998). The entire Anaximander Mountains region has been characterized by an intense tectonic activity since the end of the Miocene. Vertical tectonic movements are indicated by the absence of the thick Messinian evaporites throughout the region (Woodside et al 1998). Furthermore, tilted blocks of sedimentary sequences have been observed during the MEDINAUT submersible survey (J. Mascle, pers. comm. 1999). Tectonic movements, possibly together with gas vents, also seem responsible for a major landslide (termed the 'Great Slide' by Woodside & Limonov 1997; see Fig. 3b) in the centre between the three seamounts. Gas venting is also reported from the slopes of the seamounts, causing small pockmarks, small mounds and a number of mud volcanoes (Fig. 3b). In addition to the pore fluids, predominantly methane gas is believed to trigger mud ascent, during which some of the methane is fixated in the pore space as solid gas hydrate (Woodside et al. 1998). For this study, three mud domes have been sampled: Kula, Kazan and Amsterdam mud volcanoes (see Fig. 3b).
Methods Boron contents and isotopes Authigenic carbonate samples were dried and powdered. To avoid an influence of organic matter, the carbonates were bleached with sodium hypochlorite. Then they were rinsed and centrifuged with distilled water repeatedly to remove soluble salts and adsorbed B. After dissolution in 2N HC1, c. 5 ng B with 1 JJL! B-free seawater, the samples were loaded directly onto Re zone refined filaments and measured by negative thermal ionization mass spectrometry (N-TIMS), following the method outlined in Deyhle et al (2001). The external precision is ±0.5%o (2(Tmean). Measurements were accepted when the internal precision (2o-mean) was 0.1%c or better. Typically, an average internal precision of 0.05%o could be achieved (see Table 1). B from silica bearing muds was separated by HF digestion and a series of ion exchanges from washed and homogenized mud samples (Nakamura et al 1992). For positive TIMS, c. 100 ng of B were loaded onto Ta filaments and the isotope analyses were carried out following Deyhle (2001). The external precision is ±0.13%o (2crmean) and the internal precision (2
B GEOCHEMISTRY IN MEDITERRANEAN MUD VOLCANOES
495
Fig. 3. Detailed map showing the Olimpi Mud Volcano Field on the MedRidge (a) and the Anaximander Mountains area (b). Note that the grey heavy dashed line connecting the thrust between the Bey Daglari and Antalya Complex and the Florence Rise is part of the conceptual model explaining mud ascent and extrusion.
496
Boron in tectonic settings
A.DEYHLEErAL.
were found to be systematically c. 16.5%c lower than their corresponding parent-fluids at a pH of c. 8. Because of its highly volatile character and its wide Because of the pH sensitivity of B, the offset range of isotope ratios in geomaterials, boron is a between fluid and carbonate changes with varying unique geochemical tracer. Early work on shales pH (Palmer et al 1987). No pH information has been established that reversible adsorption of B at low published from the Anaximander Mountains mud temperatures occurs (e.g. Harder 1961; Couch & domes; however, as shown in the study of authigenic Grim, 1968; Keren & Mezuman 1981; Palmer etal carbonates from Ocean Drilling Program (ODP) Leg 1987). With increasing temperature and stress (e.g. 160, the average pH of modern mud volcano pore during burial or tectonic processes), adsorbed boron waters on the MedRidge is c. 7.8 (Deyhle & Kopf becomes enriched in the fluid phase and subse- 2001). Given that the release of CO2 after drill-core quently depleted in the clay (e.g. You et al 1996). recovery generally causes an increase in alkalinity, The adsorbed species has been suggested to have an in situ pore water pH of 7 seems a more realistic 8nB values of c. I5%c (Spivack et al. 1987), while assumption. The respective parent solution values at seawater has c. 39.5%c (Spivack & Edmond 1987). pH7 would be 19.3%o higher than the measured 8nB Therefore, fluids derived from tectonic dewatering of the carbonate. By contrast, at pH 8 the shift would of clay-rich sediments generally have B isotope be +16.5%o (Hemming et al 1995). According to ratios lighter than seawater. In addition to the simul- the variable isotope fractionation factor, all SnE taneous B enrichment in fluids by desorption from ratios reconstructed for the corresponding parent clay particles, processes such as mineral dehydra- solution are presented for a range of alkalinities tion reactions, increasing temperature and alteration values from pH 7 to 8 (see Table 1). Note that the of volcanic or igneous rocks are important control- minor changes in reconstructed fluid 8llB have no ling parameters for B mobilization and fractionation effect on the interpretation of this study. (see Palmer & Swihart 1996 and references therein). Among others, You et al. (1993, 1995) found that tectonically released, deep-seated fluids in subduc- Boron and Boron isotopes tion zones generally have elevated B contents and often show 8nB values significantly lower The B concentrations in authigenic crusts from the (20-3 5%o) than seawater. One aspect which compli- Mediterranean Ridge vary between 7.8 and 45.3 cates the behaviour of B in clays is the transforma- ppm, with an average of 26.4 ppm. The carbonates tion of smectite to illite when temperature and stress from the Anaximander Mountains fall in a similar increase (e.g. Colten-Bradley 1987). Hydrothermal range from 10.8 to 41.5 ppm, with a slightly higher experiments have demonstrated that some of the B in average concentration of 29.8 ppm (see Table 1). the pore fluid can be incorporated into the illite The isotope ratios measured are similar in the two mineral lattice (substituting Si), but may also be re- areas (Mediterranean Ridge average 8llB 20.3%o; adsorbed if confining pressures are sufficiently low. Anaximander Mountains average 8llB 18.4%o; for These laboratory results suggest that in a natural all data, see Table 1 and Fig. 4a), and correspond to mud volcano scenario, both muds and fluids may parent solutions with an average 8nB of 39.6%o and have high B concentrations. With progressive illit- 37.7%o, respectively (assuming pH 7; see Table 1). ization (i.e. greater mobilization depth of the mud), Our results are plotted as cross plots of 8UB versus the B content in the mud is expected to increase, the reciprocal B contents in Figure 4a. while 8nB of the mud decreases (e.g. Perry 1972; The muds from the two mud volcano fields You etal. 1996). show B contents similar to modern marine clays (c. 70-130 ppm; Ishikawa & Nakamura 1993). On the Mediterranean Ridge, an average 86.4 ppm are Results significantly lower than the average 131.9 ppm measured at the three mud domes in the AnaxiTen authigenic carbonates (five from the mander Mountains area (Table 1). Regarding the Mediterranean Ridge, five from the Anaximander isotope ratio, the ridge samples are higher (average Mountains) were analysed for B content and 8nB 3.9%c) than those from the Anaximander Mountains compositions (see Table 1). Authigenic carbonates (average — 0.6%o; for all data, see Table 1 and Fig. were composed of calcite, aragonite and dolomite 4b), which presumably reflects the more intense tec(Aloisi et al. 2000). The relationship between carbo- tonic deformation in the latter area (see discussion nate precipitates and the parent solution they precip- below). Again, these 8nB results range between itated from is mainly a function of pH and those from modern marine smectite (2.3-9.2%c) and composition of the parent pore fluid (Vengosh et al Miocene clayey shales (—5.5— 8.4%0; Ishikawa & 1991; Hemming et al 1995). In the study by Nakamura 1993). However, unlike the muds (Fig. Hemming et al. (1995) carbonate 8nB signatures 4b), the carbonate crusts are not significantly differ-
B GEOCHEMISTRY IN MEDITERRANEAN MUD VOLCANOES
497
Table 1. Boron contents and isotopes Authigenic carbonates Sample ID MedRidge MN17 MN5 MN4 MN16 MN18
Anaximander MN10 MN12 MN13 MN11 MN8
8nB(%0) (parent solution) at pH7
8nB(%o) (parent solution) at pH8
0.065 0.022 0.128 0.026 0.040 0.038
41.6
38.8 34.9 34.0 39.4 36.8 36.8
0.024 0.026 0.093 0.029 0.042
38.8 40.6 34.9 38.2 36.0
0.034
37.7
-2cTmean
B COtttCnt
MV name
SnB(%0)
(%o)
(ppm)
1/B
Milano Milano Napoli Napoli Moscow Average
22.3 18.4 17.5 22.9 20.3 20.3
0.05 0.07 0.02 0.09 0.03 0.05
15.4 45.3 7.8 38.4 25.3 26.4
Mountains Kazan Kazan Kazan Kula Amsterdam Average
19.5 21.3 15.6 18.9 16.7 18.4
0.04 0.07 0.08 0.02 0.03 0.05
41.5 38.9 10.8 33.9 24.1 29.8
MVname
8nB(%0)
±2amean <%o)
37.7 36.8 42.2 39.6 39.6
36.0 37.8 32.1 35.4 33.2 34.9
Mud
Sample ID MedRidge MN17 MN17 MN9
Anaximander MN12 MN11 MN14
Milano Milano Moscow Average Mountains Kazan Kula Amsterdam Average
B content (ppm)
2.6 4.8 4.3 3.9
0.02 0.10 0.04 0.05
90.2 98.0 7.1 86.4
-0.9 1.7 -2.5 -0.6
0.08 0.04 0.05 0.06
172.8 154.9 68.2 131.9
ent in the two mud volcano areas when 1/B concentration and 5nB are cross-plotted (Fig. 4a). Bearing in mind that the fluid (i.e. its precipitate) and the mud may have a different origin, this study's B results suggest that the fluid in either area may have a similar mobilization depth, while the mud in each area may originate at different depth levels (see discussion).
Discussion and conclusion The discussion sheds light on the B geochemistry in the diagenetic context of the mud volcano domains and will then place the results into the overall tectonic framework of the eastern Hellenic and western CypreanArcs. As a result of ODP drilling, the mud volcano province on the Mediterranean Ridge south of Crete
1/B
0.011 0.010 0.014 0.012
0.006 0.006 0.015 0.008
contains probably the most comprehensively studied and best understood features to date (see Fig. 3a and summary in Robertson & Kopf 19980). Clearly, the most obvious comparison of our subsurface data is to the wealth of results from studies on drillcores (Robertson et al 1996; Robertson & Kopf 19986; Deyhle & Kopf 2001). The mud breccias recovered during ODP Leg 160 yielded age records of episodic mud extrusive activity over c. 1 m.y. and longer (Robertson et al. 1996). Mobilization depth estimates ranged from 5-7 km (based on extrapolation of geophysical data; Camerlenghi et al. 1995) for the origin of the mud and clasts, but have been revised after studying the ODP drillcores. Based on both mathematical approaches (Kopf & Behrmann 2000) and low vitrinite reflectance of organic mud breccia compounds (Kopf et al. 2000), a shallower mobilization depth between 1-2 km has been inferred. So far, no estimates exist for the mud domes in the
498
A.DEYHLE£TAL.
Fig. 4. Boron results from mud volcano authigenic carbonate crusts (a) and sediment (b), plotted as 8llE versus reciprocal B content. Note that the 6nB ratios in Figure 4a are the calculated values for the parent solution to the authigenic crusts in order to illustrate the variation relative to seawater. See text for explanation.
B GEOCHEMISTRY IN MEDITERRANEAN MUD VOLCANOES
Anaximander Mountains area. However, given the tectonic setting, with penetrative faulting south of the Isparta Angle, the rifting of the Anaximander Mountains to a water depth of >2.5 km and welllithified clast lithologies within the mud breccia, mud volcanism has been inferred to be deep-seated in the region (Woodside et al. 1998; Aloisi et al 2000). Comparing the geochemical evidence from this boron study to previous work on B, C and O stable isotopes (Aloisi et al 2000; Deyhle & Kopf 2001), isotope fractionation and B concentration in the muds suggest that the depth of origin may be different in the two study areas (Fig. 4b). Clays having undergone burial and loss of adsorbed B (c. 15%o, Spivack et al. 1987) show a trend towards lower B isotope signatures (You et al 1993) than their nondiagenetic counterparts. This observation is supported by a study of marine sediments by Ishikawa & Nakamura (1993), where the 5nB of marine smectites ranges between 2.3-9.2%c, whereas wellconsolidated illite/chlorite Miocene shales show more negative values from —5.6——8.4%c. Therefore, we suggest that the more negative 5UB from Anaximander Mountain mud volcanoes represent rocks which were buried deeper and suffered more severe tectonic stresses than the accreted strata on the Mediterranean Ridge mobilized from 1-2 km subseafloor depth (Kopf et al 2000; Kopf & Behrmann2000). When an intercomparison is made between the Mediterranean Ridge and the Anaximander Mountains data from this study, the muds from the latter mud volcanoes have more negative 8nB values. The muds of the Mediterranean Ridge have an average 5UB of +3.9%c, while the average 5UB of the Anaximander Mountains mud volcanoes is -0.6%o (Table 1). This negative shift of the B isotopes of 4.5%o attests more pronounced tectonic stresses and a deeper level of mud mobilization. This explanation is supported by the different local tectonic framework in the two study regions and although there is no B fractionation model to directly relate this study's results to a particular mobilization depth, it is inferred that mud mobilization in the Anaximander Mountain region may occur at least twice as deep as on the Mediterranean Ridge. In contrast to the muds, the authigenic crusts have more similar 5nB compositions in the two areas. The fluid the crusts precipitated from has an average 8nE of 39.6%0 at the Mediterranean Ridge and insignificantly lower ratios (average 37.7%o) at Anaximander Mountains (Table 1). While the calculated parent solutions on the Mediterranean Ridge are similar to seawater (39.5%o), a shift in 8nE to slightly more negative values is observed in the Anaximander Mountains. Also, when comparing the present results from the two regions to the earlier
499
study on ODP core material, it seems that the authigenic crusts collected during the Medinaut survey have more positive 5nB ratios than the carbonates found in the ODP cores (Deyhle & Kopf 2001). There are two possible ways to explain this observation, as outlined below. As a first hypothesis, the difference may be a result of the selected sampling of fresh crusts of various carbonate compositions by submersible, while the drillcores mostly contained older clasts of different lithology with authigenic carbonate cements. In other words, the corresponding fluid to a crust recovered via submersible is directly related to near-seafloor precipitation (and possibly incorporation of Mediterranean bottom water), whereas the authigenic material in the ODP clasts (and what appeared to be crust fragments) in the core may actually be a product of cementation processes of (diagenetically altered?) pore fluids. The more positive §UB of the Mediterranean Ridge crusts from this study (average 39.6%o; see Table 1) compared to the ODP study (Deyhle & Kopf 2001) hence suggests that in certain parts of the mud volcano, different types of fluids are expelled, or that a different admixture of seawater has been occurred in differnt places (maybe due to variable venting from depth). While the ODP study has indicated a clear influence of deep-seated fluids (c. 5±1 km), the data from this study hint towards shallower fluids with near-seawater composition (Table 1). As an alternative explanation or second hypothesis, the calcareous cements in the mud breccia fragments could be the products of diagenetic dissolutionprecipitation processes within the accretionary wedge, but not a result of fluid flow from the plate boundary thrust. The carbonate cements may hence be a product of recrystallization and not of precipitation of the primary fluid. Furthermore, this interpretation does not exclude that the carbonates recovered in the mud breccias by ODP drilling had been precipitated prior to the initiation of mud extrusion some 1.25 m.y. before present. A enrichment of 10B in pore waters recovered from convergent margins has been reported previously (e.g. You et al. 1993, Deyhle et al. 2001) and has been attributed to the release of adsorbed B (5UB c. 15%o; Spivack et al 1987) at elevated temperatures. Data from hydrothermal experiments on clays imply that the mobilization and release of exchangeable B starts at very low temperatures of c. 60°C and that this process is finished around 120-150°C (You et al 1996). At temperatures in excess of 150°C, latticebound B (5UB c. —5%c) is released and causes B enrichment and lower 5nB values in the fluid (e.g. You et al 1996). In natural systems such as accretionary prisms, fluids recovered from both along the decollement (Barbados & Nankai: 5UB values c. 25-34%0; You et al 1993) and from mud volcanoes on out-of-sequence faults (5nB c. 20-37%o; Deyhle
500
A.DEYHLEETAL
& Kopf 2001) often have 8UE ratios considerably lower than seawater. B isotope values of this study, however, are only slightly more negative than seawater (Fig. 4a). The diagenetic signature of the fluid probably results from stress-driven (Fitts & Brown 1999) and temperature-induced (Colten-Bradley 1987) clay mineral dehydration reactions in the mud breccias. These fluids rapidly migrate upwards along highly permeable faults in the accretionary prism and hence preserve some of the inherited chemical "fingerprint" from depth. Apart from the B geochemistry, support for deep-seated mineral dehydration reactions is provided by the low chlorinity of Milano dome pore fluids (Emeis et al 1996). The smectite-to-illite transformation is also implied by the B concentration in the mud. As has been shown previously (e.g. Perry 1972), illite can incorporate about twice as much B into its mineral lattice than smectite (assuming sufficiently high B contents in the surrounding pore fluids). In this study, the altered mudstones from the Anaximander region have caused a minor, but significant enrichment in primary illite and chlorite (Zitter et al. 2001). The illite contents are 8% vol, which is 5-10 times higher than at Olimpi. Also, they are dominated by smectite, while the Olimpi muds contain largely kaolinite and hallyosite (Zitter et al. 2001). As a consequence, the Anaximander Mountains muds have approximately 50% more B than those from the Mediterranean Ridge. Consequently, it is proposed that this B has been incorporated during the formation of authigenic clay minerals when B was readily available in the deep-seated pore fluids. A second aspect that may have contributed to the higher B contents in the Anaximander Mountain muds is the rehydration of previously consolidated sedimentary rock. Liquefaction, which disaggregates rock fabrics and can cause a mudstone to become a fluidized mud again (e.g. Terzaghi 1947), has been shown to enhance the capacity of B uptake in clays (termed 're-wetting' in experiments by Keren & Gast 1981). The fact that the Anaximander Mountain muds undoubtedly have been liquefied after consolidation, while the Mediterranean Ridge muds may have kept an underconsolidated, poorly indurated state in the shallow accretionary prism, may well explain part of the differences in B content and isotope fractionation in the mud. Our overall results lie in the range of B geochemical results from earlier studies on mud volcanoes in subduction zones and fault belts (e.g. Taiwan: Gieskes et al. 1992; Trinidad: Dia et al. 1999). Thus, it is concluded that the Mediterranean Ridge muds have been mobilized at c. 1-2 km depth below the seafloor, while the Anaximander Mountain mud originates from c. 3-5 km subseafloor depth. Consequently, the B geochemistry suggests that in the area of orthogonal subduction (Mediterranean Ridge), faults are less pervasive and
deep-seated than in the strike-slip dominated Anaximander Mountains further NE (Figs 1 and 3). Based on the evidence available from earlier geophysical and field studies as well as the geochemical data, the following conceptual model is proposed. Regional tectonics in the Isparta Angle between the Hellenic and Cyprean Arcs have been dominated by right-lateral faulting in the Antalya Complex region together with westward thrusting over the Bey Daglari Formation (along the fault at c. E30.20 longitude in Fig. 3b). Although a direct link to the Anaximander Mountains offshore is obscured by the subsiding Finike Basin, seafloor topography as well as gravity data suggest that pervasive thrusting along an arcuate trace may have existed between the fault at c. E30.20 longitude and the northernmost limb of the Florence Rise. The fault trace would roughly cut NW-SE between Anaximenes and Anaxagoras (Fig. 3b), which is in agreement with major change in gravity in this region (Ivanov et al. 1992). Petrographic results on bottom samples (Woodside & Dumont 1997) attest the genetic relationship between the dredged rocks and their counterparts in Southern Turkey. In addition, the negative 8nB values of AM muds (—0.6%o, especially when compared to those from the MedRidge further west, 3.9%o; Fig. 4b) as well as clay mineralogical data (Zitter et al. 2001) support such an origin from diagenetically altered, buried sedimentary rocks. Given that there are mudstones and shales in the Bey Daglari Formation (e.g. Robertson 1998), liquefaction and remobilization at depth may have allowed them to ascend. If we further acknowledge the abundant fault scarps along the flanks of Anaxagoras and to a lesser extent, of the other two seamounts, it seems likely that the mud extruded along these zones of weakness in the faulted area (Fig. 3b). Parts of the clasts in the mud breccia may well represent fault breccia along the inferred suture (Fig. 3b). It remains unclear whether the fault zone connects the AC thrust with the Florence Rise (as indicated by the sub-vertical dashed line in Fig. 3b), or if it is a more local phenomenon. However, the B enrichment as well as the shift to more negative 6 U B ratios as low as — 2.5%o are in favour of a deep-seated feature. In summary, it is concluded from this study that boron is a good tracer to distinguish between the relatively shallow origin of the mud from Mediterranean Ridge mud volcanoes (1-2 km below seafloor) and the deeper remobilization depth of rocks from the Anaximander Mountains mud domes (3-5 km below seafloor). B enrichment in the Anaximander Mountains results most likely from a combination of progressive illitization of smectite and re-hydration utilizing deep-seated, B-rich fluids in the collision zone between Africa and Eurasia, but is not observed on the MedRidge. A deeper fluid is inferred from 8UB of pore water precipitates at the
B GEOCHEMISTRY IN MEDITERRANEAN MUD VOLCANOES
Anaximander Mountains when compared to the MedRidge and is in good agreement with earlier stable isotope studies on Mediterranean mud volcanoes (Aloisi et al 2000; Deyhle & Kopf 2001), and the complex tectonics of the Isparta Angle (Glover & Robertson 1998; Robertson 1998) and Anaximander region (Woodside et al. 1998). Quite possibly, the Anaximander Mountains have got disrupted by the complex interacting tectonic processes including left-lateral strike slip movements in the eastern Hellenic Arc (along the Pliny-Strabo 'Trenches') and southward rifting (Finike Basin) adjacent to the western Cyprean Arc (Florence Rise). We thank A. Eisenhauer, M.S. Farm, F. Hauff and R. Surberg for having made available the laboratories at GEOMAR, Kiel, Germany. We also thank shipboard scientists and crew of the MEDINAUT expedition for discussion and providing samples. G. de Lange and J. Woodside are thanked for their valuable criticism and detailed suggestions owing to which the consistency and clarity of the manuscript greatly benefited.
References ALOISI, G., PIERRE, C., ROUCHY, J.-M. FOUCHER, J.-R, WOODSIDE, J.M., & THE MEDINAUT SCIENTIFIC PARTY 2000. Methane-related authigenic carbonates of Eastern Mediterranean Sea mud volcanoes and their possible relation to gas-hydrate destabilisation. Earth Planetary Scientific Letters, 184, 321-338 BEN-AVRAHAM, Z., TIBOR, G.K., LIMONOV, A.F., LEYBOV, M.B., IVANOV, M.K., TOKAREV, M. Y. & WOODSIDE, J.M. 1995. Structure and tectonics of the eastern Cyprean Arc. Marine Petroleum Geology, 12 (3), 263-271. BROWN, K.M. & WESTBROOK, G.K. 1988. Mud diapirism and subcretion in the Barbados Ridge Complex. Tectonics, 7, 613-640. CAMERLENGHI, A., CITA, M.B., DELLA VEDOVA, B., Fusi, N., MIRABILE, L. & PELLIS, G. 1995. Geophysical evidence of mud diapirism on the Mediterreanean Ridge accretionary complex. Marine Geophysical Researches, 17,115-141. CHAUMELLON, E. & MASCLE, J. 1997. From foreland to forearc domains: new multichannel seismic reflection survey of the Mediterranean Ridge accretionary complex (Eastern Mediterranean). Marine Geology, 138, 237-259. CITA, M.B., RYAN, W.F.B. & PAGGI, L. 1981. Prometheus mud-breccia: An example of shale diapirism in the Western Mediterranean Ridge. Annales Geological Pays Helleniques, 30,543-570. COLLINS A.S. & ROBERTSON A.H.F 1998. Processes of Late Cretaceous to Late Miocene episodic thrust-sheet translation in Lycian Taurides, SW Turkey. Journal of the Geological Society 155,759-772. COLTEN-BRADLEY, V.A. 1987. Role of pressure in smectite dehydration - Effects on geopressure and smectite-toillite transformation. AAPG Bulletin, 71,1414-1427. COUCH, E.L. & GRIM, R.E. 1968. Boron fixation by illites. Clays Clay Minerals, 16,249-256.
501
DECOURT, J., ZONENHAIN, L.P., Ricou, L.E., KAZMIN, V.G., LE PICHON, X., KNIPPER, A.L., GRANDJACQUET, C., SBORTSHIKOV, I.M., GEYSSANT, J., LEPVRIER, C., PECHERSKY, D.H., BOULIN, J., SIBUET, J.C., SAVOSTIN, L.A., SOROKHTIN, O., WESTPHAL, M., BAZHENOV, M.L., LAUER, J.P. & BIJU-DUVAL, B. 1986. Geological evolution of the Tethys Belt from the Atlantic to the Pamirs since the Lias. Tectonophysics, 123,241-315. DEYHLE, A. 2001. Improvements of boron isotope analysis by positive thermal ionization mass spectrometry using static multi-collection of Cs2BO2+ ions. International Journal of Mass Spectrometry, 206, 79-89. DEYHLE, A. & KOPF, A., 2001. Deep fluids and ancient pore waters at the backstop: Stable isotope systematics (B, C, O) of mud-volcano deposits on the Mediterranean Ridge accretionary wedge. Geology, 29(11), 1031-1034. DEYHLE, A., KOPF, A. & EISENHAUER, A., 2001. Boron systematics of authigenic carbonates: A new approach to identify fluid processes in accretionary prisms. Earth Planetary Science Letters, 187,191-205. DIA, A.N., CASTREC-ROUELLE, M., BOULEGE, J. & COMEAU, P. 1999. Trinidad mud volcanoes: Where do the expelled fluids come from? Geochimical Cosmochimica. Acta, 63,1023-1038. EMEIS, K.-C. & SHIPBOARD SCIENTIFIC PARTY 1996. Proceedings ODP, Initial Reports, 160, College Station, TX (Ocean Drilling Program), 972 pp. FITTS, T.G. & BROWN, K.M. 1999. Stress induced smectite dehydration ramifications for patterns of freshening fluid expulsion in the N. Barbados accretionary wedge. Earth Planetary Science Letters, 172, 179-197. GIESKES, J.M., You, C.-F, LEE, T, Yui, T.F. & CHEN, H.-W. 1992. Hydro-Geochemistry of mud volcanoes in Taiwan. Acta Geological Taiwanica, 30,79-88. GLOVER, C. & ROBERTSON, A.H.F. 1998. Neotectonic intersection of the Aegean and Cyprus tectonic arcs; extensional and strike-slip faulting in the Isparta Angle, SW Turkey. In: ROBERTSON, A.H.F. & COMAS, M.C. (eds) Collision-related processes in the Mediterranean region. Tectonophysics, 298, (1-3), 103-132. HARDER, H. 1961. Beitrag zur Geochemie des Bors: III. Bor in metamorphen Gesteinen und im geochemischen Kreislauf. Nachr. Akademie der Wissenschaften Gottingen, II. Mathematisch-Psysikalische Acta, 1, 1-26. HEMMING, N.G., REEDER, R.J. & HANSON, G.N. 1995. Mineral-fluid partitioning and isotopic fractionation of boron in synthetic calcium carbonate. Geochim. Cosmochim. Acta, 59,371-379. HIGGINS, G.E., & SAUNDERS, J.B. 1974. Mud volcanoes Their nature and origin. Verhaudlungen der Naturfor SchendeuAessellschaft Basel, 84,101-152. Hsu, K.J. & BERNOULLI, D. 1978. Genesis of the Tethys and the Mediterranean. In: Hsu, K.J., MONTADERT, L., et al DSDP 42, Initial Results, Washington D.C. (US Gov. Print. Office) Pt. 1,943-950. ISHIKAWA, T. & NAKAMURA, E. 1993. Boron isotope systematics of marine sediments. Earth Planetary Scientific Letters, 117,567-580. IVANOV, M., LIMONOV, A. & WOODSIDE, J.M. 1992. Geological and geophysical investigations in the
502
A.DEYHLE£rAL.
Mediterranean and Black Seas. In: UNESCO Reports in Marine Science, 56, UNESCO, Paris. KEREN, R. & CAST, R.G. 1981. Effect of wetting and drying, and of exchangeable cations, on boron adsorption and release by montmorillonite. Soil Science Society of America Journal, 45,478^82. KEREN, R. & MEZUMAN, U. 1981. Boron adsorption to clay minerals using a phenomenological equation. Clays Clay Minerals, 29,198-204. KOPF, A., ROBERTSON, A.H.F., CLENNELL, M.B. & FLECKER, R. 1998. Mechanism of mud extrusion on the Mediterranean Ridge. GeoMarine Letters, 18(3), 97-114. KOPF, A., ROBERTSON, A.H.E & VOLKMANN, N. 2000. Origin of mud breccia from the Mediterranean Ridge accretionary complex using petrography and maturity of solid organic carbon. Marine Geology, 166, 65-82. KOPF, A. & BEHRMANN, J.H. 2000. Extrusion dynamics of mud volcanoes on the Mediterranean Ridge accretionary complex. In: VENDEVUXE, B.C., MART, Y. & VIGNERESSE, J.L. (eds) Salt, Shale and Igneous Diapirs in and around Europe. Geological Society, London, Special Publications, 174,169-204. KOPF, A., KLAESCHEN, D. & MASCLE, J. 2001. Extreme efficiency of mud volcanism in dewatering accretionary prisms. Earth Planet. Scientific Letters, 189, 295-313. LE PICHON, X. & ANGELEER, J. 1979. The Hellenic Arc and trench system: a key to the neotectonic evolution of the eastern Mediterranean area. Tectonophysics, 60, 1-42. LE PICHON, X., CHAMOT-ROOKE, N., LALLEMENT, S., NOOMEN, R. & VEIS, G. 1995. Geodetic determination of the kinematics of central Greece with respect to Europe: Implications for eastern Mediterranean tectonics. Journal of Geophysical Research, 100, 12675-12690. MASCLE, J. et al 1999. Images may show start of European-African Plate collision, EOS Transactions AGU, 80/37,421-428. MCKENZIE, D. 1972. Active tectonics of the Mediterranean region. Geophysical Journal of the Royal Astronomical Society, 30,109-185. MONTADERT, L., LETOUZEY, J. & MAUFFRET, A.
1978.
Messinian event: Seismic evidence. In: Hsti, K.J. & MONTADERT, L. etaL Deep Sea Drilling Project Initial Results, 42, Washington D.C. (US Gov. Print. Office) Pt. 1,1037-1050. MOORE, J.C. & VROLUK, P. 1992. Fluids in accretionary prisms, Reviews of Geophysics, 30,113-135. NAKAMURA, E., ISHIKAWA, T., BICK, J.L. & ALLEGRE, C.J. 1992. Precise boron isotopic analysis of natural rock samples using a boron-mannitol complex, Chemical Geology, 94,193-204. PALMER, M.R., SPIVACK, AJ. & EDMOND, J.M. 1987. Temperature and pH controls over isotopic fractionation during adsorption of boron on marine clay. Geochim. Cosmochimica. Acta, 51, 2319-2323. PALMER, M.R. & SWIHART G.H. 1996. Boron isotope geochemistry: An overview. In: Boron: Mineralogy, petrology, and geochemistry. Mineralogical Society of America, Reviews in Mineralogy, 33,709-744. PERRY, E.A. 1972. Diagenesis and the validity of the boron
paleosalinity technique. American Journal of Science 272,150-160. ROBERTSON, A.H.F. & SHIPBOARD SCIENTIFIC PARTY OF ODP Leg 160 1996. Mud volcanism on the Mediterranean Ridge: Initial results of Ocean Drilling Program, Leg 160, Geology, 24(3), 239-242 ROBERTSON, A.H.F. & KOPF, A. 1998a. Tectonic setting and processes of mud volcanism on the Mediterranean Ridge accretionary complex: evidence from Leg 160. In: ROBERTSON, A.F.H., EMEIS, K.-C., RICHTER, C. & CAMERLENGHI, A. Proceedings of ODP, Scientific Results, 160, College Station, TX (Ocean Drilling Program), 665-680. ROBERTSON, A.H.F. & KOPF, A. 1998&. Origin of clasts and matrix within Milano and Napoli mud volcanoes, Mediterranean Ridge accretionary complex. In: ROBERTSON, A.F.H., EMEIS, K.-C., RICHTER, C. & CAMERLENGHI, A. Proc. ODP, Sci. Results, 160, College Station, TX (Ocean Drilling Program), 575-596. ROBERTSON, A.H.F. 1998. Mesozoic-Tertiary tectonic evolution of the easternmost Mediterranean area: integration of marine and land evidence. In: ROBERTSON, A.F.H., EMEIS, K.-C., RICHTER, C. & CAMERLENGHI, A., Proceedings of ODP, Scientific Results, 160, College Station, TX (Ocean Drilling Program), 723-782. SPIVACK, A.J., PALMER, M.R. & EDMOND, J.M. 1987. The sedimentary cycle of the boron isotopes. Geochim. Cosmochimica. Acta, 51,1939-1949. SPIVACK, A.J. & EDMOND, J.M. 1987. Boron isotope exchange between seawater and oceanic crust. Geochim. Cosmochimica. Acta, 51,1033-1043. TERZAGHI, K. 1947. Shear characteristics of quicksand and soft clay. 7th Texas Conference on Soil Mechanics and Foundation Engineering, Houston, 1-10. VENGOSH, A., KOLODNY, Y., STARINSKY, A., CHTVAS, A.R., & MCCULLOCH, M.T. 1991. Coprecipitation and isotopic fractionation of boron in modem biogenic carbonates. Geochim. Cosmochimica. Acta, 55, 2901-2910. WOODSIDE, J.M. 1977. Tectonic elements and crust of the eastern Mediterranean Sea. Marine Geophysical Researches, 3,313-317. WOODSIDE, J.M. 1997. Mediterranean leg (ANAXIPROBE96); introduction to ANAXIPROBE-96 expedition. In: WOODSIDE, J.M., IVANOV, M.K. & LIMONOV, A.F. (eds) Neotectonics and fluid flow through seafloor sediments in the eastern Mediterranean and Black seas; Part I, Eastern Mediterranean Sea. Technical Series Unesco, Intergovernmental Oceanographic Commission, 48(1), 20. WOODSIDE, J.M. & DUMONT, J.F. 1997. The Anaximander Mountains are a southward rifted and foundered part of the southwestern Turkish Tauruss. Terra Nova (Suppl.), 9, 394. (Abstract) WOODSIDE, J.M., & LIMONOV, A.F. 1997. Mediterranean leg (ANAXIPROBE-96); general morphology and structure of the Anaximander Mountains . In: WOODSIDE, J.M., IVANOV, M.K. & LIMONOV, A.F. (eds). Technical Series - Unesco, Intergovernmental Oceanographic Commission, 48(1), 29-36. WOODSIDE, J.M., IVANOV, M.K., LIMONOV, A.F. & SHIPBOARD SCIENTIFIC PARTY OF THE ANAXIPROBE
B GEOCHEMISTRY IN MEDITERRANEAN MUD VOLCANOES EXPEDITIONS 1998. Shallow gas and gas hydrates in the Anaximander Mountains region, eastern Mediterranean Sea. In: HENRIET, J.P. & MIENERT, J., (eds) Gas Hydrates: Relevance to World Margin Stability and Climate Change. Geological Society, London, Special Publications, 137,177-193. You, C.-E, SPIVACK, A.J., SMITH, J.H. & GIESKES, J.M. 1993. Mobilization of boron at convergent magins: Implications for boron geochemical cycle. Geology, 21,207-210. You, C.-R, CHAN, L.H., SPIVACK, AJ. & GIESKES, J.M. 1995. Lithium, boron, and their isotopes in sediments and pore waters of Ocean Drilling Program Site 808,
503
Nankai Trough: Implications for fluid expulsion in accretionary prisms. Geology, 23(1), 37-40. You, C.-E, CASTILLO, PR., GIESKES, J.M., CHAN, L.H., & SPIVACK, A.J. 1996. Trace element behavior in hydrothermal experiments: Implications for fluid processes at shallow depths in subduction zones. Earth Planetary Scientific Letters, 140,41-52. ZITTER, T.A.C., VAN DER GAAST, S.J. & WOODSIDE, J.M. 2001. New information concerning clay mineral provenance in mud volcanoes. In: Proceedings of 36th CIESM congress, Monaco, Monte-Carlo, 23-28 September 2001, Rapp. Comm. Inter. MerMedit., 36, 46.
This page intentionally left blank
Remobilization of sand from consolidated sandstones: evidence from mixed bitumen-sand intrusions JOHN PARNELL & JASON KELLY Department of Geology and Petroleum Geology, University of Aberdeen, Aberdeen AB24 SUE, UK (e-mail: [email protected]) Abstract: Some sand in intrusive bodies has an origin in the disaggregation of consolidated sandstones. Evidence from the study of mixed bitumen-sand intrusions includes progressive spalling of individual grains from cemented sandstones and individual sand grains with adhering cements. Both physical forces related to rapid fluid movement and chemical dissolution may contribute to the disaggregation process. Rapid fluid flow due to hydrofracturing of an overpressured sequence is responsible for several examples of sand-bearing bitumen veins. As overpressuring is a widespread phenomenon, disaggregation of lithified sandstones may be more prevalent than hitherto appreciated. Sand may be entrapped in bitumen within vein systems cutting kilometres of stratigraphy. Fluid inclusion data from an example at Bentheim, Germany, shows that cementation of sandstone occurred at high temperatures before sand disaggregation and mobilization.
Sandstone dykes and sills are a widespread phenomenon in the geological record. The general assumption is that they develop before the source sand bed became consolidated (Lonergan et al 2000), implying a shallow depth of formation. This is probably true in most cases. However, there is evidence that some sandstone dykes develop subsequent to consolidation of the source sandstones, which implies greater possible depths of formation (Parnell et al I996b). Implications of sandstone dyke formation after cleavage development (Phillips & Alsop 2000) also indicate that the conditions of sand mobilization may be more varied than hitherto appreciated. The most unequivocal intrusions are dykes and sills of width up to metres, but larger intrusive structures up to tens of metres are evident on seismic records (MacLeod et al. 1999; Duranti etal 2002; Molyneux etal 2002). This report summarizes the evidence for sand mobilization following disaggregation of previously consolidated sandstones, in particular through the study of mixed bitumen-sand intrusions. Examples are taken from the Eocene Uinta Basin, Utah, USA; the Cretaceous of the Lower Saxony Basin at Bentheim, Germany; the Tertiary of Selenica, Albania; the Tertiary melange in Barbados; and the Cretaceous of the Junggar Basin, China (Table 1). All samples studied are from surface exposures. In each case, bitumen veins contain mobilized sand. We show that the mobilized sand consists of disaggregated grains, in comparison with the sand described in most examples of sand intrusions, which had not been consolidated at the time of intrusion. The bitumen dykes and sills are of comparable scale to sandstone dykes and sills (see Table 1). Several mechanisms have been proposed for sand mobilization and intrusion, including differential
compaction (Cosgrove & Hillier 2000), slumping (Newman et al 1993), diapirism (Dixon et al 1995) and release of overpressuring (Lonergan et al 2000), the latter also being an explanation for bitumen vein formation (see below). As sandstone dykes may be pathways for oil migration (Jenkins 1930), there is an important analogy between sandstone dykes and bitumen veins as conduits for fluid expulsion from deeper to shallower levels in sedimentary basins.
Sand in bitumen veins Veins of solid bitumen occur on a large scale, up to metres width (Fig. 1), in many parts of the world and are currently mined in several countries including the United States, Argentina, China and Albania (Parnell et al 1994a; Parnell & Carey 1995). They exhibit many similarities with igneous and sandstone dykes, including scale, occurrence in swarms, propagation normal to bedding and sharp contacts with country rock (Fig. 2). In most cases, bitumen veins are determined to have injected upwards, but rarely they extend downwards from the bitumen source (Parnell 1996), as is also deduced for some sandstone intrusions (e.g. Vitanage 1953). Many suites of bitumen veins contain sand entrained within the bitumen. The degree of entrainment varies from sparse isolated sand grains to abundant isolated grains to a grain-supported sand body with a bitumen matrix. Thus, there is a complete gradation from pure bitumen veins to bituminous sandstone veins. Many bitumen veins are sand-free and those which are effectively bituminous sandstones are uncommon. However, this is to be expected as
Table 1. Summary of bitumen veins containing disaggregated sand Junggar Basin, China
Lower Saxony Basin, Germany
Uinta Basin, Utah
Melange, Barbados
Selenica, Albania
Age
Cretaceous
Eocene 6m 20km +
Pliocene
1.2m 500m Sandstone
Cretaceous 0.5m 900m
Eocene - Oligocene
Max. vein width Max. vein length Host rocks
1.5m
>1 m
1km
?
Sandstone
Sandstone Mudrock
Sandstone Mudrock
Clasts in vein
Sandstone Mudrock
Sandstone Mudrock
Sandstone Mudrock
Sandstone Mudrock Siderite
Sandstone Conglomerate Mudrock Sandstone Mudrock
Sand grains
Present Calcite Iron oxide Clay
Present
Present Quartz Albite Calcite Chlorite Clay
Present Quartz
Adhering cements
Quartz Calcite Clay
Present Quartz Calcite Clay
many do not occur in sandstone sequences and the most extreme examples of sand entrainment reflect unusual circumstances (see below). Suites of bitumen veins tend to occur in the close vicinity of the hydrocarbon source rock and may even be rooted in the source rock as in the Tertiary Uinta Basin, Utah (Pruitt 1961). Fracturing of the source rock may be induced by the hydrocarbon generation process, whereby the increase in volume caused by conversion of solid organic matter into hydrocarbon fluids confers anomalous pressure which exceeds the fracture strength of the rock (du Rouchet 1981; Mann 1990). As such the veins are an aspect of primary hydrocarbon migration (release from the source rock), so they do not always traverse sandstone strata. However, where they do they commonly entrain sands within the bitumen. Bitumen veins do not normally occur rooted in reservoir rock, i.e. they are not a typical feature of remigration, although there is no reason why catastrophic faulting of an oil reservoir should not lead to bitumen veining. Hydrofracturing of an overpressured sandstone reservoir could yield a bitumensand mixture: this is probably the origin of mixed bitumen and sand veins at Bentheim, Germany Parnell et al. 1996b).
Petrography Sand grains in bitumen Fig. 1. Vein of solid bitumen with large slab of country rock entrained, Eocene, Uinta Basin, Utah. Vein width about 3 m.
Where bitumen veins are sand-rich, most notably in the Uinta Basin, Utah, it is commonly possible to
REMOBILIZATION OF SAND FROM CONSOLIDATED SANDSTONES
Fig. 2. Vein of solid bitumen (mined out cleft) through sandstone host rock, Cretaceous, Junggar Basin, China, showing zone of impregnation (dark) of sandstone by bitumen around vein.
observe the transition from isolated grains to consolidated sandstone, whereby grains appear to have spalled off the sandstone into the bitumen (Fig. 3). When the isolated grains are examined using scanning electron microscopy, they are found to have adhering traces of mineral cements, including quartz, carbonates and clay minerals, for example at Selenica, Albania (Fig. 4). Quartz grains and overgrowths may show evidence for breakage where they have become detached from adjacent grains. In rare cases (e.g. Uinta Basin, Selenica), the sand grains in bitumen are highly angular and non-spherical, rather than the rounded and spherical shapes which normally characterize grains in sandstones. The cements recorded on mobilized sand grains (quartz, calcite, clays) are the same as those recorded in normal sandstone beds: no distinctive styles of cementation have been recognized. Some bitumen veins show multiple stringers extending into the host sandstone. This is seen in the Tertiary of Barbados (Parnell et al I994b). The bitumen stringers may form interlinking networks that exhibit isolation of sandstone clasts and individual sand grains (Fig. 5). Although disaggregation of the wall rock is most
507
Fig. 3. Margin of bitumen vein showing: (a) bitumen (black) impregnation of cemented sandstone and isolated sand grains suspended in bitumen; (b) bitumen veinlets extending into sandstone, from which grains are partially detached. EB Vein, Eocene, Uinta Basin, Utah, scale bars each 600 microns.
easy to demonstrate in sandstones, abundant fragments of mudrock in bitumen veins in the Uinta Basin, Utah (Monson & Parnell 1992) and the Neuquen Basin, Argentina (Parnell & Carey 1995) show that other limologies are also affected. The lithified nature of mudrock fragments confirms that bitumen injection involved breakage of consolidated sediment. Commonly, the clasts in bitumen veins can be matched to the immediately adjacent wallrock, i.e. the clasts are locally derived. However, in some cases, including the Uinta Basin, sand grains occur in portions of bitumen veins which are hosted by mudrocks, indicating that grain movement along veins had occurred. Fossils in bitumen veins that are of a different age to the host rock also imply clast movement, as recorded in the Neuquen Basin (Parnell & Carey 1995).
Impregnation of sandstone wall rocks Some veins exhibit significant impregnation of sandstone wall rock by bitumen. In some cases,
508
J. PARNELL & J. KELLY
Fig. 4. Scanning electron micrographs of isolated sand grains within bitumen (B), showing traces of adhering cement, including quartz overgrowth (Q), calcite (C) and smectite-illite (S). Pliocene, Selenica, Albania, scale bar 40 microns.
impregnation is the only evidence for the passage of hydrocarbons along a fracture, as no bitumen solidified within it. In a good example at Dragon, Uinta Basin, Utah, two vertical fractures through sandstone have bitumen-impregnated wall rock, but no bitumen vein (Eldridge 1896), despite the close vicinity of the large Dragon bitumen vein. This indicates that the impregnation was an early event, while the bitumen was fluid enough to completely bleed away. Passage through the pore network of the host sandstone also implies fluidity. Petrographic study of samples from the bitumen-impregnated zones shows grains completely encased in bitumen (Parnell et al 1994#), which would be relatively easy to mobilize. Bitumen veins pass through other lithologies in addition to sandstone. Where they do, the host rocks exhibit brittle fracturing, and spalled off fragments become suspended in the bitumen. Even in crystalline basement rocks that have been penetrated by downward or lateral injection of bitumen, angular fragments are calved off into the bitumen (Fig. 6).
Paragenetic relationships Where bitumen veins contain clasts of sandstone in addition to individual sand grains, it is possible to determine the diagenetic sequence experienced by the sandstone before disruption and incorporation in the bitumen. Where the bitumen has migrated into sandstone wall rocks, it can be incorporated into a diagenetic sequence that commences before and continues after, bitumen emplacement. An example of aparagenesis (sequence of diagenetic and veining events) involving both bitumen and sand emplacement (Fig. 7) at Bentheim, Germany, is shown in Figure 8. The inclusion of emplacement episodes in such a sequence is valuable because we may be able to use some events in the sequence to constrain the timing/conditions of the emplacement. One approach to this is to obtain temperature constraints from fluid inclusions in cements or mineral veins. In the example from Bentheim, successive high temperatures can be distinguished from overgrowths on grains in bitumen, moderate temperatures from trails in the injected sand and further high temperatures in later calcite veining (Fig. 9). Kelly et al. (2000)
REMOBILIZATION OF SAND FROM CONSOLIDATED SANDSTONES
509
Fig. 5. Interlinked stringers of bitumen in sandstone, showing isolation of sandstone clasts. Microscopy shows many individual grains also isolated. Tertiary melange, Frizers, Barbados, scale bar 5mm.
Fig. 6. Bitumen vein through Precambrian metasedimentary host rock, containing calved off angular fragments of host within bitumen. Late Proterozoic (Longmyndian), Shropshire, scale bar 5 mm.
records another example of the use of fluid inclusion data to understand the conditions of bitumen emplacement through sandstones. In this example, from the Carboniferous of the Frontal Zone of the Ouachita Mountains, Oklahoma, bitumen veins occur in fractures, which had a history of primary formation associated with low-temperature aqueous fluids followed by reactivation by higher temperature (100-120°C) mixed aqueous-hydrocarbon fluids and bitumen precipitation. These data show that sand injection can occur after the succession has experienced temperatures that would be interpreted to represent either several kilometres burial or an anomalously hot fluid pulse. The very fact that sand grains exhibit quartz overgrowths in several examples implies temperatures of 80°C and higher, as almost all fluid inclusion measurements in quartz cements indicate a temperature of precipitation in the range 80-130°C (e.g. Walderhaug 1994).
Fig. 7. Photomicrographs of petrographic relationships in sand/bitumen vein, Cretaceous, Bentheim, Germany. (a) Breccia of sandstone (dark grey) and bitumen (black) fragments in matrix of calcite/sand (light grey) and vuggy sparry calcite (white); (b) Detail of calcite veining through sandstone, showing sand grains (grey) disaggregated from sandstone. Scale bars (a) 1.2 cm, (b) 1 mm.
Discussion Significance of petrographic features The overgrowths on the sand grains indicate cementation of the sand before incorporation in the bitumen. It might be envisaged that quartz grains could develop overgrowths while in the bitumen, given that there is widespread evidence for the growth of authigenic quartz crystals in bitumen elsewhere (Parnell et al 1996a). However, the dimensions of overgrowths are exactly what would be expected to grow in sandstone pores and they tend to be developed on particular grain facets. Large overgrowths are not observed, nor overgrowths that completely encapsulate grains, both features that could reflect unconstrained growth. However, other cementing materials are observed, including grain-coating clays and carbonates, which have a paragenesis similar to that in much sandstone. In an example from the bitumen deposits at Selenica,
510
J. PARNELL & J. KELLY
Fig. 8. Summary of paragenesis of cementation and bitumen and sand emplacement, Cretaceous, Bentheim, Germany (simplified from Parnell et al 1996&).
Fig. 9. Homogenization temperatures from three successive populations of fluid inclusions in injected sand, Cretaceous, Bentheim, Germany.
REMOBILIZATION OF SAND FROM CONSOLIDATED SANDSTONES
Albania (Abraham 1945), isolated sand grains in bitumen exhibit quartz overgrowths followed by calcite cement (Fig. 3). The instances of highly angular grains suggest that they have been subject to intragranular fracturing in addition to intergranular fracturing (see below). Comparable angular quartz grains have been recorded in the mud volcanoes of Trinidad, where they are attributed to hydrofracturing under high pore fluid pressure (Deville et al 2003). If the hydrofracturing, which allows bitumen emplacement, originates in the fine-grained, hydrocarbon source rock, the occurrence of much sand in the vein might be surprising, as the mechanical damage associated with the initial 'explosive' event would not affect the sandstones. However, it is clear that fracture propagation through sandstone was not always through a single fracture tip. Hence, much sandstone was affected by the bitumen injection and much sand could become mobilized.
Disaggregation oflithified rocks The implication of the petrographic observations is that bitumen emplacement was responsible for disaggregation of sandstones and incorporation of sand grains in the bitumen. As the morphology of bitumen veins and their relationship with the country rock suggests that they reflect a release of overpressure, probably related to hydrocarbon generation (Verbeek & Grout 1993; Parnell & Carey 1995; Parnell 1999; Cobbold et al 1999), we might infer that the physical forces involved were sufficient to disaggregate the sandstones. Fluid pressure is invoked as the cause of fracturing, rather than a reactivation of tectonically-induced fractures, because bitumen veins are commonly rooted in the fluid (i.e. hydrocarbon) source rock, bitumen infills fractures to their tips and in the Uinta Basin fracture density is greatest where the hydrocarbon yield of the source rocks is greatest (Monson & Parnell 1992). In some cases bitumen appears to replace a pre-existing calcite cement (Parnell et al 1994a), indicating that chemical processes were also involved. Organic acids possibly contributed to the progressive dissolution of calcite, to be replaced by bitumen. The evidence for progressive disaggregation of a lithified sandstone into its constituent sand grains indicates that the sandstone was being broken along the weakest points, i.e. the grain boundaries. If the degree of cementation in the original sandstone was patchy, which is commonly the case during the initial stages of lithification, the more weakly cemented portions would have disaggregated more readily, leaving the better cemented portions more or less intact. The grain boundaries in the sandstones should be easy to break: they are mostly point or
511
planar contacts and the sutured boundaries which characterize a high degree of lithification are not yet developed. The spalling of angular fragments of crystalline basement rocks and other non-sandstone lithologies is comparable with brecciation observed in many hydrothermal deposits. However, in the case of bitumen the fragments become suspended when the ambient fluid solidifies, rather than due to precipitation of a vein-filling mineral out of solution. The fracturing of highly coherent host rocks implies that disaggregation of sandstones along grain boundaries, which are sites of relative weakness, should be readily achieved. If hydrocarbon fluids are physically capable of disaggregating sand grains, it is just as likely that overpressured aqueous fluids should be able to do the same. Hence, the sand in some sandstone dykes may have originated from lithified sandstones rather than unconsolidated sands, i.e. the occurrence of sandstone dykes does not necessarily imply that the source was not consolidated. It is less easy to demonstrate that sand grains in a sandstone dyke were previously consolidated, as any observed cement is assumed to have precipitated within the dyke. However careful petrographical study could show evidence for a pre-injection phase of cementation, similar to that found in the bitumen veins. In some circumstances, an additional contributing factor to disaggregation could be from shear stresses. When porous sediments are subjected to shear stresses, they can experience matrix collapse and liquefy. This can occur at depth and is a potential mechanism for the formation of sandstone dykes (Yassir 2003). Lateral tectonic forces (Yassir 1997) may induce such shear stresses, but this is not a prerequisite. Both bitumen veins (Lebkuchner et al 1972; Haggan & Parnell 2000) and sandstone dykes (Winslow 1983) can occur in compressional terranes, but they do not appear to be anomalous in their clast content. It is commonly not possible to determine the depth from which entrained sand was mobilized. However, in some cases we can constrain the maximum depth possible. In the Uinta Basin, where the bitumen veins are rooted in the source rock and much maturation data is available for the source rock, the maximum burial was 2-3 km (Anders et al 1992). The possible source of sand grains must have been disaggregated at a level shallower than the source rock root zone.
Conclusions The distribution of sandstone and bitumen dykes and their petrography, in the studied examples provide several lines of evidence that indicate an origin subsequent to consolidation of the sandstone source:
512
(i) (ii) (iii)
(iv) (v)
J. PARNELL & J. KELLY
Bitumen-rich intrusions contain a range of clasts from cemented sandstone fragments to individual sand grains. Individual sand grains in bitumen exhibit adhering traces of the cements that normally characterize lithified sandstones. Bitumen veins cutting other host lithologies show that the bitumen engenders brittle fracturing. By analogy, disintegration of sandstones along grain boundaries would be readily achieved. Sand is entrapped in bitumen within vein systems extending upward through kilometres of stratigraphy. Fluid inclusions entrapped in sandstones before disaggregation show that the sandstone had experienced moderately high temperatures, not consistent with heating due to shallow burial alone.
The sand in the bitumen veins is distinct from that in most cases of injected sands, in exhibiting preinjection cements. As the process of disaggregation appears to reflect fluid movement due to overpressuring and consequent hydrofracturing and overpressuring is a widespread phenomenon in sedimentary basins (Swarbrick 1994; Osborne & Swarbrick 1997), the disaggregation of lithified sandstones could be a widely developed phenomenon, and it should not be assumed that all cases of mobilized sands are near-surface features. We are grateful to E. Deville and an anonymous reviewer for critical reviews which helped to improve the manuscript. Numerous workers collaborated in collection of the samples used in this study, including B. Monson, A. Ruffell, Z. Tang, A. Geng, A. Serjani and A. Dulaj.
References ABRAHAM, H. 1945. Asphalts and Allied Substances (5th edition). Van Nostrand, New York. ANDERS, D.E., PALACAS, J.G. & JOHNSON, R.C. 1992. Thermal maturity of rocks and hydrocarbon deposits, Uinta Basin, Utah. In: FOUCH, T.D., Nuccio, V.F. & CHIDSEY, T.C. JR. (eds) Hydrocarbon and Mineral Resources of the Uinta Basin, Utah and Colorado. Utah Geological Association Guidebook, 20, 53-76. COBBOLD, P.R., DIRAISON, M. & ROSELLO, E.A. 1999. Bitumen veins and Eocene transpression, Neuquen Basin, Argentina. Tectonophysics, 314,423-442. COSGROVE, J.W. & HILLIER, R.D. 2000. Forced-fold development within Tertiary sediments of the Alba Field, UKCS: Evidence of differential compaction and postdepositional sandstone remobilization. In: COSGROVE, J.W. & AMEEN, M.S. (eds) Forced Folds and Fractures. Geological Society, London, Special Publications, 169,61-71. DEVILLE, E., BATTANI, A., HEREIN, J.R, HOUZAY, J.R,
PRINZHOFER, A. & MULLER, C. 2003. New insight for the origin and process of mud volcanism in Trinidad. In: VAN RENSBERGEN, P., Hnxis, R.R., MALTMAN, AJ. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization, Geological Society, London, Special Publications, 216,475^90. DlXON, R.J.,
SCHOFIELD, K., ANDERTON, R., REYNOLDS,
A.D., ALEXANDER, R.W.S., WILLIAMS, M.C. & DAVIES, K.G. 1995. Sandstone diapirism and clastic intrusion in the Tertiary sub-marine fans of the BruceBeryl Embayment, Quadrant 9, UKCS. In: HARTLEY, AJ. & PROSSER, D.J. (eds) Characterization of Deep Marine Clastic Systems. Geological Society, London, Special Publications, 94,77-94. DURANTI, D., HURST, A., BELL, C., GROVES, S. & HANSON, R. 2002. Injected and remobilised Eocene sandstones from the Alba Field, UKCS: core and wireline log characteristics. Petroleum Geoscience, 8,99-107. Du ROUCHET, J. 1981. Stress fields, a key to oil migration. American Association of Petroleum Geologists Bulletin, 65, 63-73. ELDRIDGE, G.H. 1896. The Uintaite deposits of Utah. United States Geological Survey 17th Annual Report, Part 1,909-949. HAGGAN, T. & PARNELL, J. 2000. Hydrocarbon-metal associations in the Western Cordillera, Pern. Journal of Geochemical Exploration, 69-70,229-234. JENKINS, O.P. 1930. Sandstone dikes as conduits for oil migration through shales. American Association of Petroleum Geologists Bulletin, 14,411-421. KELLY, J., PARNELL, J. & CHEN, H.H. 2000. Application of fluid inclusions to studies of fractured sandstone reservoirs. Journal of Geochemical Exploration, 69-70, 705-709. LEBKUCHNER, R.F., ORHUN, F. & WOLF, M. 1972. Asphaltic substances in southeastern Turkey. American Association of Petroleum Geologists Bulletin, 56, 1939-1964. LONERGAN, L., LEE, N., JOHNSON, H.D., CARTWRIGHT, J.A. & JOLLY, R. J.H. 2000. Remobilization and injection in deep water depositional systems: implications for reservoir architecture and prediction. In: WEIMER, P., SLATT, R.M. & COLEMAN, J. (eds) Deep-Water Reservoirs of the World. GCSSEPM Foundation, 20th Annual Conference, Houston, 515—532. MACLEOD, M.K., HANSON, R.A., BELL, C.R. & McHuoo, S. 1999. The Alba Field ocean bottom cable seismic survey: Impact on development. The Leading Edge, 18,1306-1312. MANN, U. 1990. Sedimentological and petrophysical aspects of primary petroleum migration pathways. In: HELING, D., ROTHE, P., FORSTNER, U. & STOPPERS, P. (eds) Sediments and Environmental Geochemistry. Springer-Verlag, Berlin, 152-178. MOLYNEUX, S., CARTWRIGHT, J. & LONERGAN, L. 2002. Conical sandstone injection structures imaged by 3D seismic in the central North Sea, UK. First Break, 20, 383-393. MONSON, B. & PARNELL, J. 1992. The origin of gilsonite vein deposits in the Uinta Basin, Utah. In: FOUCH, T.D., Nuccio, V.F. & CHIDSEY, T.C. Jr. (eds) Hydrocarbon and Mineral Resources of the Uinta Basin, Utah and Colorado. Utah Geological Association Guidebook, 20, 257-270.
IMMOBILIZATION OF SAND FROM CONSOLIDATED SANDSTONES NEWMAN, M. St. J., REEDER, M.L., WOODRUFF, A.H.F. & HATTON, I.R. 1993. The geology of the Gryphon Field. In: PARKER, J.R. (ed.) Petroleum Geology of Northwest Europe: proceedings of the 4th Conference. The Geological Society, London, 123-133. OSBORNE, M. & SWARBRICK, R.E. 1997. Mechanisms for generating overpressure in sedimentary basins: A reevaluation. American Association of Petroleum Geologists Bulletin, 81,1023-1041. PARNELL, J. 1996. Alteration of crystalline basement rocks by hydrocarbon-bearing fluids: Moinian of Rossshire, Scotland. Lithos, 37,281-292. PARNELL, J. 1999. Petrographic evidence for emplacement of carbon into Witwatersrand conglomerates under high fluid pressure. Journal of Sedimentary Research, 69,164-170. PARNELL, J. & CAREY, P.F. 1995. Emplacement of bitumen (asphaltite) veins in the Neuquen Basin, Argentina. American Association of Petroleum Geologists Bulletin,19,H98-m6. PARNELL, J., CAREY, P.F. & MONSON, B. 1996a. Fluid inclusion constraints on temperatures of petroleum migration from authigenic quartz in bitumen veins. Chemical Geology, 129,217-226. PARNELL, J., GENG, A., Fu, J. & SHENG, G. 19940. Geology and geochemistry of bitumen vein deposits at Ghost City, Junggar Basin, northwest China. Geological Magazine, 131,181-190. PARNELL, J., GENG, A. & VEALE, C. I994b. Petrology of the bitumen (manjak) deposits of Barbados: hydrocarbon migration in an accretionary prism. Marine and Petroleum Geology, 11,743-755. PARNELL, J., RUFFELL, A.H., MONSON, B. & MUTTERLOSE, J. I996b. Petrography and origin of deposits at the Bentheim bitumen mine, north west Germany. Mineralium Deposita, 31,104-112.
513
PHILLIPS, C.A. & ALSOP, G.I. 2000. Post-tectonic clastic dykes in the Dalradian of Scotland and Ireland: implications for delayed lithification and deformation of sediments. Geological Journal, 35,99-110. PRUITT, R.G. 1961. The mineral resources of Uinta County, Utah. Utah Geological and Mineralogical Survey Bulletin, 71,1-101. SWARBRICK, R.E. 1994. Reservoir diagenesis and hydrocarbon migration under hydrostatic palaeopressure conditions. Clay Minerals, 29,463-473. VERBEEK, E.R. & GROUT, M.A. 1993. Geometry and structural evolution of gilsonite dykes in the eastern Uinta Basin, Utah. United States Geological Survey Bulletin, 1787, HH1-HH42. VITANAGE, P.W. 1953. Sandstone dikes in the South Platte area, Colorado. Journal of Geology, 62,493-500. WALDERHAUG, O. 1994. Temperatures of quartz cementation in Jurassic sandstones from the Norwegian continental shelf - evidence from fluid inclusions. Journal of Sedimentary Research, A64, 311-323. WINSLOW, M.A. 1983. Clastic dike swarms and the structural evolution of the foreland fold and thrust belt of the southern Andes. Geological Society of America Bulletin, 94,1073-1080. YASSIR, N. 1997. Geomechanical controls on overpressuring: implications for pore pressure prediction in compressional basins. In: HENDRY, J., CAREY, P., PARNELL, J., RUFFELL, A. & WORDEN, R. (eds) Proceedings of Second International Conference on Fluid Evolution, Migration and Interaction in Sedimentary Basins and Orogenic Belts, 235-238. Belfast. YASSIR, N. 2003. The geology and geomechanics of deepseated mud volcanoes. In: VAN RENSBERGEN, P., HILLIS, R.R., MALTMAN, A.J. & MORLEY, C.K. (eds) Subsurface Sediment Mobilization, Geological Society, London, Special Publications, 216,461-474.
This page intentionally left blank
Index
Page numbers in italic, e.g. 214, refer to figures. Page numbers in bold, e.g. 216, signify entries in tables. accretionary prisms/wedges 337, 340, 350, 443, 477, 482, 500 see Barbados accretionary prism; CrockerRaj ang accretionary prism; Hikurangi subduction margin; Mediterranean Ridge accretionary prism; Nankai prism aggradation 419, 431-432, 438, 439, 453 aggregative fluidization 125, 128 Alba Field, North Sea 123, 125-128, 133-134, 226, 263 Albian succession, SE France 51 Alboran Sea 5, 337, 443-459 Allen diagrams see horizon separation diagrams Ampa shale ridge, Brunei Darussalam 398, 402-405, 412 Anaximander Mountains 5, 491-503 Anglais Point, Trinidad 463, 466, 468-470, 477, 481-482 aquathermal effects 15, 15 artesian conditions 14, 75, 31 Atterberg limits 466 Australian NW Shelf 362, 364, 364, 367 Azerbaijan mud volcanoes 159,160 Balder Formation tuffs, North Sea 139, 143,146, 149, 152, 154-155, 266-267, 266, 271-272, 277, 273, 279 ball-and-pillow structures 2, 22-25, 22, 27-28, 27, 32, 32,90 Baram Delta, Borneo 336-339, 340, 346-348, 352-355, 381, 395-409, 411-412, 412, 421, 424, 426, 433, 436, 438, 440-441 Barbados accretionary prism 2, 13-15, 103, 443, 476-482, 484, 486-488 Barbados melange 505-507, 509 Bay of Biscay polygonal faults 300 Belait Formation, Brunei Darussalam 381-383, 386, 394 Belgium polygonal faults 227, 272 BGHS see gas hydrates stability zone, base of Bight Basin, offshore Australia 429-441 biostratigraphy 155, 175, 184, 277, 386, 468^69, 482, 484, 486-488 Blue Maris Formation, SE France 53-54, 53-54 Blue Whale Supersequence, Australia 429-441 Boom Clay, Belgium 4, 309-321 borehole data 126, 727 boron isotopes 5, 491-503 bottom current 326-327 furrows and grooves 323 Bottom Simulating Reflector (BSR) 176,177-179, 184, 191-207, 211, 217, 285, 403, 404, 406 boudins 10 bowl-like depressions 154-156 breccia 381,500 caused by overpressure 4,10 flint 298, 301-307 shale clast 124, 724, 135 brittle deformation 24, 110, 116, 118, 124, 724, 128, 223, 298, 301, 359, 366, 471, 508, 512 Brunei Darussalam 4-5, 362, 369-379 well blow-out 376-378
