Advances in Geosciences: Volume 13: Solid Earth (Se)

A d v a n c e s i n Geosciences Volume 13: Solid Earth (SE) ADVANCES IN GEOSCIENCES Editor-in-Chief: Wing-Huen Ip (...

Author: Kenji Satake

30 downloads 425 Views 37MB Size Report

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!

Report copyright / DMCA form

DOWNLOAD PDF

A d v a n c e s

i n

Geosciences Volume 13: Solid Earth (SE)

ADVANCES IN GEOSCIENCES Editor-in-Chief: Wing-Huen Ip (National Central University, Taiwan) A 5-Volume Set Volume 1: Volume 2: Volume 3: Volume 4: Volume 5:

Solid Earth (SE) ISBN-10 981-256-985-5 Solar Terrestrial (ST) ISBN-10 981-256-984-7 Planetary Science (PS) ISBN-10 981-256-983-9 Hydrological Science (HS) ISBN-10 981-256-982-0 Oceans and Atmospheres (OA) ISBN-10 981-256-981-2

A 4-Volume Set Volume 6:

Hydrological Science (HS) ISBN-13 978-981-270-985-1 ISBN-10 981-270-985-1 Volume 7: Planetary Science (PS) ISBN-13 978-981-270-986-8 ISBN-10 981-270-986-X Volume 8: Solar Terrestrial (ST) ISBN-13 978-981-270-987-5 ISBN-10 981-270-987-8 Volume 9: Solid Earth (SE), Ocean Science (OS) & Atmospheric Science (AS) ISBN-13 978-981-270-988-2 ISBN-10 981-270-988-6 A 6-Volume Set Volume 10:

Volume 11:

Volume 12:

Volume 13:

Atmospheric Science (AS) ISBN-13 978-981-283-611-3 ISBN-10 981-283-611-X Hydrological Science (HS) ISBN-13 978-981-283-613-7 ISBN-10 981-283-613-6 Ocean Science (OS) ISBN-13 978-981-283-615-1 ISBN-10 981-283-615-2 Solid Earth (SE) ISBN-13 978-981-283-617-5

ISBN-10 981-283-617-9 Volume 14:

Volume 15:

Solar Terrestrial (ST) ISBN-13 978-981-283-619-9 ISBN-10 981-283-619-5 Planetary Science (PS) ISBN-13 978-981-283-621-2 ISBN-10 981-283-621-7

A d v a n c e s

i n

Geosciences Volume 13: Solid Earth (SE)

Editor-in-Chief

Wing-Huen Ip

National Central University, Taiwan

Volume Editor-in-Chief

Kenji Satake

University of Tokyo, Japan

World Scientific NEW JERSEY

•

LONDON

•

SINGAPORE

•

BEIJING

•

SHANGHAI

•

HONG KONG

•

TA I P E I

•

CHENNAI

Published by World Scientific Publishing Co. Pte. Ltd. 5 Toh Tuck Link, Singapore 596224 USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601 UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library.

ADVANCES IN GEOSCIENCES A 6-Volume Set Volume 13: Solid Earth (SE) Copyright © 2009 by World Scientific Publishing Co. Pte. Ltd. All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means, electronic or mechanical, including photocopying, recording or any information storage and retrieval system now known or to be invented, without written permission from the Publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy is not required from the publisher.

ISBN-13 ISBN-10 ISBN-13 ISBN-10

978-981-283-610-6 981-283-610-1 978-981-283-617-5 981-283-617-9

(Set) (Set) (Vol. 13) (Vol. 13)

Typeset by Stallion Press Email: [email protected] Printed in Singapore.

December 12, 2008 16:12

AOGS 2007 - SE Volume

EDITORS Editor-in-Chief:

Wing-Huen Ip

Volume 10: Atmospheric Science (AS) Editor-in-Chief: Jai Ho Oh Editor: Gyan Prakash Singh Volume 11: Hydrological Science (HS) Editor-in-Chief: Namsik Park Editors: Joong Hoon Kim Eiichi Nakakita C. G. Cui Taha Ouarda Volume 12: Ocean Science (OS) Editor-in-Chief: Jianping Gan Editors: Minhan Dan Vadlamani Murty Volume 13: Solid Earth (SE) Editor-in-Chief: Kenji Satake Volume 14: Solar Terrestrial (ST) Editor-in-Chief: Marc Duldig Editors: P. K. Manoharan Andrew W. Yau Q.-G. Zong Volume 15: Planetary Science (PS) Editor-in-Chief: Anil Bhardwaj Editors: Yasumasa Kasaba Paul Hartogh C. Y. Robert Wu Kinoshita Daisuke Takashi Ito v

9in x 6in

b672-V13-fm

FA

December 12, 2008 16:12

AOGS 2007 - SE Volume

9in x 6in

This page intentionally left blank

b672-V13-fm

FA

December 12, 2008 16:12

AOGS 2007 - SE Volume

9in x 6in

b672-V13-fm

LIST OF REVIEWERS

The Editor of Volume 13 (Solid Earth) would like to thank the following referees who have helped review the papers published in this volume: Agustan Giulio Barbieri Jungho Cho Phil Cummins Takashi Furumura Richard Gross Martin Flower Nuraini Rahma Hanifa Hasanuddin Z. Abidin Takahiro Hatano Yukio Hayakawa Kosuke Heki Giulio Iovine Shuanggen Jin Yasuyuki Kano Teruyuki Kato Somboon Khositanont Hamzah Latief Kuo-fong Ma Norio Matsumoto

Ritsuko S. Matsu’ura Irwan Meilano Ki-Bok Min James Mori Hiroshi Munekane Ken T. Murata Danny H. Natawidjaja Jong Uk (James) Park Bill Petrachenko Marco Piras CP Rajendran Lucas Donny Setijadji D. Srinagesh Yuichiro Tanioka Hiroyuki Tsutsumi Yasushi Watanabe Pornsawat Wathanakul Moriaki Yasuhara Jiancang Zhuang Michael Zolensky

vii

FA

December 12, 2008 16:12

AOGS 2007 - SE Volume

9in x 6in

This page intentionally left blank

b672-V13-fm

FA

December 12, 2008 16:12

AOGS 2007 - SE Volume

9in x 6in

b672-V13-fm

CONTENTS

Editors

v

List of Reviewers

vii

Mineralization Characteristics and Ore Fluid of Huai Kham on Gold Deposit, Northern Thailand Somboon Khositanont, Khin Zaw and Prayote Ounchanum Formation of Hollow Concretions in Northeastern Thailand Prinya Putthapiban and Sutatcha Hongsresawat Investigations on Local Quartz Sand for Application in Glass Industry Pisutti Dararutana, Prukswan Chetanachan, Pornsawat Wathanakul and Narin Sirikulrat Geochemical and Sr–Nd–Pb Isotopic Study of Late Neogene Volcanic Rocks from the Arita–Imari Area (SW Japan): Evidence for Coexisting OIB-Like and Subduction-Related Mantle Sources Nguyen Hoang, Jun’ichi Itoh, Kozo Uto and Akikazu Matsumoto Landslide Hazard Zoning of the Muravera Hillside (Sardinia, Italy) Giulio Barbieri and Paolo Cambuli

ix

1

13

23

31

57

FA

December 12, 2008 16:12

x

AOGS 2007 - SE Volume

9in x 6in

b672-V13-fm

Contents

An Integrative Geoscience Data Model by Linking Field-Speciﬁc Data Models in Digital Geologic Map, Earth Resource, and Geo-Hazard Lucas Donny Setijadji and Koichiro Watanabe Crustal Deformation Monitoring by GNSS: Network Analysis and Case Studies Marco Piras, Marco Roggero and Maurizio Fantino

73

87

The Global Geodetic Observing System H.-P. Plag, M. Rothacher, M. Pearlman, R. Neilan and C. Ma

105

The International Laser Ranging Service Michael Pearlman, Carey Noll, Jan McGarry, Werner Gurtner and Erricos Pavlis

129

Earth Rotation Parameters from Very Long Baseline Interferometry and Ringlaser Observables P. J. Mendes Cerveira, H. Spicakova, H. Schuh, T. Kluegel, U. Schreiber and A. Velikoseltsev Toward a New VLBI System for Geodesy and Astrometry J¨ org Wresnik, Johannes B¨ ohm, Andrea Pany and Harald Schuh

155

167

Periodic Station Motion in Gothenburg Observed with GPS — Possibly Related to Hydrological Phenomena? R. Haas, N. Tangdamrongsub, H.-G. Scherneck and J. Johansson

181

Evaluation of the Coseismic Pore Fluid Pressure on a Thrust Fault Jeen-Hwa Wang

193

FA

December 12, 2008 16:12

AOGS 2007 - SE Volume

9in x 6in

b672-V13-fm

Contents

The Water Level Changes of the Aneuklaot Lake, Weh Island after the 2004 Sumatra–Andaman Earthquake Agustan, Djoko Nugroho, Lena Sumargana, Irwan Meilano, Mohd. Eﬀendi Daud, Fumiaki Kimata and Yusuf S. Djadjadihardja Eﬀect of Near-Source Trench Structure on Teleseismic Body Waveforms: An Application of a 2.5D FDM to the Java Trench Taro Okamoto and Hiroshi Takenaka Numerical Modeling of the 2006 Java Tsunami Earthquake Nuraini Rahma Hanifa, Irwan Meilano, Takeshi Sagiya, Fumiaki Kimata and Hasanuddin Z. Abidin Why Many Victims: Lessons from the July 2006 South Java Tsunami Earthquake Hasanuddin Z. Abidin and Teruyuki Kato Earthquake Potential in Myanmar Hla Hla Aung

xi

203

215

231

249

265

Looking into a Sandpile by Photo-Elasticity and Discrete Element Method Naoto Yoshioka and Hide Sakaguchi

281

Statistical Properties and Time Trend in the Number of Holocene Volcanic Eruptions A. N. Zemtsov and A. A. Tron

291

FA

This page intentionally left blank

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch01

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

MINERALIZATION CHARACTERISTICS AND ORE FLUID OF HUAI KHAM ON GOLD DEPOSIT, NORTHERN THAILAND SOMBOON KHOSITANONT Geological Sciences, Chiang Mai University, Chiang Mai, 50002 Thailand [email protected] KHIN ZAW CODES ARC Centre of Excellence in Ore Deposits, University of Tasmania, Hobart, TAS, 7005 Australia PRAYOTE OUNCHANUM Geological Sciences, Chiang Mai University, Chiang Mai, 50002 Thailand

Huai Kham On gold deposit is located within the Lampang-Phrae volcanic belt of the Sukhothai Fold Belt in northern Thailand. The gold deposit is hosted by Triassic andesitic tuﬀ and intercalated rhyolitic welded tuﬀ. The host sequence is overlain by sedimentary sequences including Triassic calcareous sandstone of the Wang Chin Formation and Middle Triassic limestone of the Kang Pla Formation. Gold nuggets and electrum were found in quartz-rich veins with associated pyrite, chalcopyrite, galena and bournonite cutting across the andesitic tuﬀ and underlying rhyolitic tuﬀ. Sulfur isotope analyses indicate that the gold-ore forming ﬂuids vary in sulfur isotopic values from −5.3 to −3.5 per mil. Fluid inclusion studies indicate that the ore-forming ﬂuids were typically enriched in CO2 which is evidenced by the occurrence of carbonic (CO2 (L)– H2 O(L)–CO2 (V)) inclusions in quartz adjacent to the gold-bearing sulﬁde minerals. Three types of ﬂuid inclusions including Type I (L-V) aqueous inclusions, Type II (L-L-V) aqueous–carbonic inclusions, and Type III (V) vapor rich inclusions, were classiﬁed in vein quartz. Results from a preliminary microthermometry study of ﬂuid inclusions in ore-bearing vein quartz indicate that the melting temperature of CO2 solid varies from −56.8◦ C to −56.6◦ C and that the homogenization of the carbonic phase varies from 28◦ C to 31◦ C suggesting that the carbonic phase contains pure CO2 . This interpretation is supported by Laser Raman Spectroscopy analyses. The homogenization temperatures of the carbonic inclusions in quartz vary from 280◦ C to 300◦ C Salinities of ore ﬂuids range from 1 to 7 wt.% NaCl equiv. 1

FA

December 10, 2008 15:45

2

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch01

S. Khositanont, Khin Zaw and P. Ounchanum On the basis of available data such as vein texture, alteration, and ﬂuid inclusion data, the Huai Kham On gold deposit is comparable to orogenic gold deposits.

1. Introduction The Sukhothai Fold Belt in northern Thailand has formed as a result of Indochina–Shan Thai collision.1−4 The formation of N–S trending orogenic belt in conjunction with volcano-plutonism in relation to subduction and collision has produced a large variety of precious and base metal deposits, e.g. tungsten, antimony, copper, and gold. A number of gold deposits which have been discovered within the Sukhothai Fold Belt are also located in the volcano-plutonic aureole including Huai Kham On deposits in Phrae Province. However, the gold mineralization at Huai Kham On deposit has not been well studied. This study reports gold mineralization and ﬂuid inclusion characteristics at Huai Kham On deposit in the Sukhothai Fold Belt.

2. Geology of Huai Kham On Deposit The Huai Kham On deposit is hosted within the NNE-SSW trending Triassic volcanic rocks, named as Lampang-Phrae volcanic belt, and the overlain by Triassic sedimentary sequences of the Kang Pla Formation and Wang Chin Formation of Lampang Group. The lower part of the volcanic sequence at the Huai Kham On deposit is composed mainly of andesite porphyry, andesitic tuﬀ, and breccia. Pyrite and other sulﬁde minerals are rarely observed in the andesite porphyry. The overlying andesitic tuﬀ contains breccia within a welded matrix. The andesite sequence is overlain by rhyolitic and welded rhyolitic tuﬀ. The rhyolite, composed of angular translucent to transparent quartz and subrounded pink feldspar, is set in the dark reddish brown ﬁne grained groundmass. Some rhyolites are composed of angular to subangular quartz phenocrysts with fragments of subhedral K-feldspar and plagioclase. Large pyrrhotite crystals with ﬁne-grained pyrite replacement are observed elsewhere in this rock. The uppermost successions of volcanic rocks at the Huai Kham On deposit are overlain by middle to late Triassic sedimentary rocks including limestone of Kang Pla Formation (Lst) and clastic rocks (Tr) of Wang Chin Formation.5

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Mineralization Characteristics and Ore Fluid of Huai Kham

b672-V13-ch01

3

3. Vein Mineralogy and Alteration Quartz veins are found only in the Triassic andesitic tuﬀ and intercalated rhyolite (Fig. 1). The veins are 50 cm to 3 m thick and delineate along

Fig. 1. Geological map showing major rock units and signiﬁcant geological features at Huai Kham On deposit, Wang Chin District, Phrae Province.

FA

December 10, 2008 15:45

4

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch01

S. Khositanont, Khin Zaw and P. Ounchanum

Fig. 2. Gold and galena association in quartz vein from Huai Kham On deposit. Au: native gold, Gn: galena, Gtz: quartz.

northwest strike. Quartz veins are enclosed within chlorite alteration zone. Most of vein quartz in the gold-bearing quartz veins show cataclastic texture suggesting that the gold-bearing quartz veins were formed in the orogenic environment. Pyrite is the most abundant sulﬁde mineral, whereas chalcopyrite, galena, and bournornite are much less abundant. Pyrite generally has a reaction rim with galena inclusions. Gold grains are specially associated with galena or included by quartz (Fig. 2). Results from Laser Ablation Inductively Couple Plasma Analyses indicate that pyrite also contains dissolved gold where chalcopyrite is present.

4. Sulfur Isotope Study Sulfur isotope study is applied to the sulﬁde mineral from ﬁve vein quartz samples of various depths at the Huai Kham On deposit. Pyrite grains were drilled from gold-bearing quartz veins. Sample preparation for sulfur isotope analyses was carried out using quantitative preparation of SO2

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch01

Mineralization Characteristics and Ore Fluid of Huai Kham Table 1. deposit.

5

Sulfur isotope analyses of pyrite from Huai Kham On

Sample ID

[email protected] [email protected] [email protected] [email protected] [email protected]

Mineral

Weight (mg)

d 34 SCDT (permil)

pyrite pyrite pyrite pyrite pyrite

14.2 31.0 12.7 14.0 12.3

−3.4 −4.6 −4.2 −5.5 −5.4

for 34 S/32 S analyses from sulﬁdes by combustion with cuprous oxide technique.6 The sulfur isotope analyses were performed with VG SIRA Series 2 triple-collector mass spectrometer at University of Tasmania. Sulfur isotopes of pyrite from the Huai Kham On deposit show a narrow range from −3.4 to −5.5 per mil (Table 1). The S isotope ranges for magmatic source generally show small positive values or close to 0 per mil and the isotope value for sedimentary source generally shows much greater positive value.7

5. Fluid Inclusion Study Fluid inclusions are observed in quartz. Primary ﬂuid inclusions are observed in crystal growth zones by polarizing microscope (Fig. 3) and by cathodoluminescence — SEM image.

5.1. Classification of fluid inclusions Fluid inclusions in quartz are classiﬁed into three types as follows: 1. Type I (L-V) aqueous ﬂuid inclusions are composed mainly of aqueous liquid with a relatively small vapor bubbles (Fig. 4). They generally occur as secondary origin along fractures that cut across the crystal boundaries of quartz. 2. Type II (L-L-V) aqueous–carbonic ﬂuid inclusions show aqueous and carbonic liquids and a gas bubble at the core of the carbonic liquid at a room temperature (Fig. 5). Carbonic gas and carbonic liquid phase separation is observed at temperature below 31.1◦ C (critical point of CO2 ). The presence of CO2 as a major component in the carbonic phase is also conﬁrmed by Laser Raman Spectroscopic Analyses. These

FA

December 10, 2008 15:45

6

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch01

S. Khositanont, Khin Zaw and P. Ounchanum

Fig. 3. Photograph of crystal growth zone (dashed lines) and ﬂuid inclusions in quartz at Huai Kham On deposit.

Fig. 4. Type I ﬂuid inclusions in quartz from Huai Kham On deposit; V = vapor bubble, L = aqueous liquid.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch01

Mineralization Characteristics and Ore Fluid of Huai Kham

7

Fig. 5. Type II (L-L-V) aqueous carbonic ﬂuid inclusions in quartz from Huai Kham On deposit. Low-density CO2 phases are squeezed in the middle, surrounded by aqueous liquid at rim.

inclusions are aligned along crystal growth zones in quartz crystals, suggesting a primary origin. 3. Type III (V-L) aqueous–carbonic inclusions generally show large volumes of carbonic phase including liquid and gas (Fig. 6). The CO2 liquid may not be observed at room temperature due to insuﬃcient CO2 in the ﬂuids. Laser Raman Spectroscopy analysis shows that the carbonic phases are composed only of CO2 . These inclusions are of primary origin due to their appearance along crystal growth zones.

5.2. Microthermometric measurement and P–T–V–X relation Microthermometric measurement was carried out using a Linkamcomputerized ﬂuid inclusion freezing/heating stage. The thermocouple was calibrated from −56.6◦C (melting point of pure CO2 ) to 374.1◦C (critical point of pure H2 O) using Fluid Inc synthetic ﬂuid inclusions. The results can be reproduced within 0.1◦ C for the freezing experiment and 1.0◦ C for heating experiment. The inclusions were cooled down to −120◦C while all phases were frozen. Then they were gradually heated up to −56.6◦ C to observe the

FA

December 10, 2008 15:45

8

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch01

S. Khositanont, Khin Zaw and P. Ounchanum

Fig. 6. Photograph of Type III vapor-rich inclusions in quartz from Huai Kham On deposit. Inclusions contain mainly CO2 gas bubble (dark part inside the inclusions).

melting point of CO2 for Type 2 (L-L-V) inclusions and Type III vaporrich inclusions. The heating experiment was carried on to 10◦ C in order to observe the ﬁnal melting temperatures of the solid phases including ice and clathrate (mixed CO2 –H2 O solid). Finally, all inclusions were heated up to observe the CO2 homogenization temperature in Type II (L-L-V) and Type III inclusions and the ﬁnal homogenization temperature. The melting temperature of carbonic ice in Type II (L-L-V) inclusions ranges from −59.2◦ C to −56.6◦C suggesting that CO2 is the major component in the carbonic phase for Type II (L-L-V) inclusions. The homogenization temperatures of CO2 phase range from 25◦ C to 31.1◦ C, which also support CO2 as a major component in the carbonic phase. The clathrate melting temperatures of Type II (L-L-V) inclusions and Type III vapor-rich inclusions range from 5.3◦ C to 9◦ C. It is noticeable that the depression of the clathrate melting temperature from 10◦ C is aﬀected by the amount of salt in the aqueous phase in the Type II (L-L-V) inclusions. The salinity of aqueous ﬂuid ranges from 1–7wt% NaCl eq. (Fig. 7) is obtained from the depression of clathrate melting temperature using the equation of Brown.10

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Mineralization Characteristics and Ore Fluid of Huai Kham

b672-V13-ch01

9

Fig. 7. Histogram shows salinity variation in Type II and Type III inclusions of quartz from Huai Kham On deposit. The majority of inclusions contain 2–3wt% NaCl equivalent.

Fig. 8. Histogram shows homogenization temperature variation of Type II and Type III ﬂuid inclusions from quartz of Huai Kham On deposit.

The majority of ﬁnal homogenization temperatures (Th ) of Type II (L-L-V) inclusions are 240◦ C to 310◦ C (Fig. 8), whereas the ﬁnal homogenization temperatures of Type III vapor-rich inclusions are slightly higher (320◦ C–340◦C). The diﬀerence between the ﬁnal homogenization temperatures of Type II and Type III inclusions may be due to a kinetic barrier during the formation of vapor and liquid phases of aqueous– carbonic ﬂuids. Since the majority of inclusions were homogenized at temperature ranging from 280◦C to 300◦ C. These temperature ranges

FA

December 10, 2008 15:45

10

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch01

S. Khositanont, Khin Zaw and P. Ounchanum

Fig. 9. Th -XH2 O plots for Type II and Type III inclusions. The data, which are scattered above 1 kb solvus, suggest that the inclusions were trapped at pressure above 1 kb.8,9

therefore, represented the homogenization of Type II and Type III ﬂuid inclusions. The similarity between the ﬁnal homogenization temperatures (Th ) of the liquid-rich and vapor-rich inclusions is, therefore, interpreted as the same trapped P –T conditions of 280◦ C to 300◦ C and 1 to 3 kb.

5.3. P–T–V–X relation Homogenization of Type II and Type III inclusions are plotted against molar volume of XH2 O (Fig. 9). The data plotted above the 1 kb solvus indicate that the pressure of entrapment of Type II inclusions and Type III inclusions were higher than 1 kb. It is also noticeable that the salinity– homogenization temperature plots form clusters rather than a mixing trend suggests that the ore-forming ﬂuids were derived from homogeneous ﬂuid (Fig. 10).

6. Gold Mineralization at Huai Kham On Deposit Gold grains that are present as cavity ﬁlling and as dissemination in quartz without observable fractures may suggest that they were formed simultaneously with the formation of quartz veins. Therefore, primary ﬂuid inclusions in quartz veins were trapped simultaneously with the formation of gold mineralization.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Mineralization Characteristics and Ore Fluid of Huai Kham

b672-V13-ch01

11

Fig. 10. Salinity — Homogenization plots showing composition of ﬂuid inclusions of vein quartz from Huai Kham On deposit.

The presence of Type II (L-L-V) aqueous–carbonic and Type III vapor rich ﬂuid inclusions in gold-bearing vein quartz in conjunction with microthermometric measurements indicate that gold-bearing quartz veins at the Huai Kham On deposit were derived from aqueous–carbonic ﬂuids, which are composed of 2–3wt% NaCl equivalent in the aqueous phase, whereas CO2 is a major component in the carbonic phases. Evidence from ﬂuid inclusion trapping temperature also indicates that the majority of oreforming ﬂuids were trapped at 280◦ C–300◦C at 1 kb–3 kb. These pressure and temperature ranges are similar to those in mesothermal and/or orogenic gold styles of mineralization. The presence of aqueous–carbonic (Type II and Type III) ﬂuid inclusions indicates that the precipitation of ore at the Huai Kham On deposit occurred by ﬂuid eﬀervescence process, which can be only formed under high pressure condition. Fluid eﬀervescence (phase separation of CO2 ) may have resulted in pressure reduction.

7. Conclusions 1. Sulfur isotope characteristics suggest that the ore-forming ﬂuids at Huai Kham On were formed either from magmatic ﬂuids. 2. Fluid inclusion studies indicate that the gold-bearing quartz veins at the Huai Kham On deposit were formed at 280◦ C–300◦C and 1 kb–3 kb. 3. The formation of gold-bearing quartz veins resulted from ﬂuid eﬀervescence which led to rapid pressure reduction in the hydrothermal system.

FA

December 10, 2008 15:45

12

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch01

S. Khositanont, Khin Zaw and P. Ounchanum

4. The pressure–temperature range of ore formation is similar to that of mesothermal or orogenic gold mineralization around the world.

References 1. S. Bunopas, Palaeogeographic history of western Thailand and adjacent part of Southeast Asia: A plate tectonic interpretation, PhD Thesis, Victoria University of Wellington, 1981, 810p. 2. S. Bunopas, P. Vella, H. Fontaine, S. Hada, C. Burrett, P. Haines, S. Potisat, Th. Wongwanich, P. Chaodamrong, K. T. Howard and S. Khositanont, Growth of Asia in the late Triassic continent–continent collision of Shan Thai and Indochina against South China, J. Gonwana Res. 4(4) (2001) 584–586. 3. S. M. Bar, S. A. Macdonal, G. R. Dunning, P. Ounchanum and W. Yaowanoiyothin, Petrochemistry, U-Pb (zircon) age and the palaeotectonic setting of the Lampang volcanic belt, northern Thailand, J. Geol. Soc. London 157 (2000) 553–563. 4. I. Metcalfe, Permian tectonic framework and palaegeography of SE Asia, J. SE Asian Earth Sci. 20 (2002) 551–566. 5. P. Chaodamrong, Stratigraphy, sedimentology and tectonic implications of the Lampang Group, central north Thailand, PhD Thesis, University of Tasmania, 1992, 230p. 6. B. W. Robinson and M. Kasakabe, Quantitative preparation of SO2 for 34 S/32 S analyses from sulphides by combustion with cuprous oxide, Anal. Chem. 47(7) (1975) 1179–1181. 7. H. Ohmoto and R. O. Rye, Isotope of sulphur and carbon, Geochem. Hydrothermal Ore Deposits (1979) 509–567. 8. E. Roedder, Fluid inclusion, Rev. Mineralogy 12 (1984) 646. 9. T. S. Bowers and H. C. Helgeson, Calculation of the thermodynamic and geochemical consequences of nonideal mixing in the system H2 O-CO2 -NaCl on phase relations in geologic systems: Equation of state for H2 O-CO2 -NaCl fluids at high pressure and temperatures, Geochim Cosmochim Acta 47 (1983) 1247–1275. 10. P. Brown, Flincor Software (1989).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch02

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

FORMATION OF HOLLOW CONCRETIONS IN NORTHEASTERN THAILAND PRINYA PUTTHAPIBAN∗ and SUTATCHA HONGSRESAWAT Geoscience Programme, Department of Physics, Faculty of Science, Mahidol University, Rama VI Road, Bangkok 10400, Thailand ∗ [email protected]

The mysterious rocks “Naka’s eggs” commonly found in Northeastern Thailand are hollow concretions derived from clastic rocks of the Khorat Group. The concretions appear in diﬀerent shapes, such as spheroidal, ellipsoidal, and irregular with sizes varying from a few cm up to 60 cm. Their dark brown outer shells are much harder than the hosted rocks, and the inner surfaces of the hollows are rugged and occasionally contain remnants of pyrite (FeS2 ) minerals indicating incomplete oxidation processes. The result of extensive examinations of these hollow concretions suggests that their formation involves subsurface water that penetrates through fractures of rocks and the boundaries of sand grains forming several species of iron solutions. Due to their exothermic nature, these solutions sieve outward to the region with lower temperature and pressure where chemical reactions can continue. When equilibrium is reached, reddish brown iron oxide sediments remained as hard shells of the concretions. The hollow is then created in situ as a result of these chemical processes. The size and shape of these hollow concretions clearly depend on the quantity of pyrite crystals and the morphology of the pyrite nodules. As an external erosion process subsequently takes place, the outer shells which are more resistant and have a smaller porosity due to the secondary cemented iron oxides survive with shapes of sphere, ellipsoid and others, whereas other sandy parts of the host were eroded away. Because it is evidently clear that the reddish-brown color of the clastic rocks in our study areas is secondary in origin, parts of chemical reactions discussed here are promising candidates for actual chemical alterations responsible for the reddish color of the Khorat Group red beds in Thailand.

1. Introduction The continental Mesozoic rocks of the Khorat Group in northeastern Thailand have been intensively studied by many researchers.1−5 These rock ∗ Corresponding

author. 13

FA

December 10, 2008 15:45

14

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch02

P. Putthapiban and S. Hongsresawat

sequences vary in age from Upper Triassic to Cretaceous-Tertiary.6−9 The rocks of the Khorat Group are subdivided into nine formations listed from the oldest to the youngest as follows: Huai Hin Lat, Nam Phong, Phu Kradung, Phra Wihan, Sao Khua, Phu Phan, Khok Kruat, Maha Sarakham, and Phu Thok Formations.10,11 In this study, we present evidences of the formation of mysterious rocks, “Naka’s eggs” or hollow concretions which are found in the low hills with gentle slopes and ﬂat land within the terrain covered by the clastic rocks of the Phra Wihan and Sua Khua Formations (Study Area 1) and of the Phu Kradung Formation

Fig. 1. Geologic map of Southeastern Khorat Plateau showing locations of the hollowed concretions Study Area 1 (ﬁlled circle) Phu Phrao near the Ban Chong Meg border town, Sirinthorn District, and Study Area 2 (open star in a ﬁlled circle) Ban Palai, Amphoe Po Sai, Ubon Ratchathani Province (ﬁlled star) Pak Moon Dam Site (modiﬁed after Sattayarak and Suteethorn, 1983).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Formation of Hollow Concretions in Northeastern Thailand

b672-V13-ch02

15

(Study Area 2), shown in Fig. 1. Consequently, we propose, for the ﬁrst time, the mechanisms responsible for the creation of the hollow concretions.

2. General Geology of the Study Areas The Khorat Group clastic rocks in the southeastern corner of the northeastern plateau are generally ﬂat lying beds, and slightly dip west toward the center of the Khorat basin (Fig. 1). The study areas predominantly contain three major rock formations. The ﬁrst formation is classiﬁed as the Middle to Upper Jurassic Phu Kradung Formation normally distinguished by brown to grayish siltstones and ﬁne-grained sandstones. This rock formation is mainly observed in the Study Area 2. Two additional formations observed in the Study Area 1 are Upper Jurassic to Lower Cretaceous Phra Wihan Formation containing mostly quartzitic sandstones and siltstones and the Lower Cretaceous Sua Khua Formation. This last formation is recognized as the rock unit where most vertebrate fossils were excavated in Thailand. Typical rocks in the Sao Khua ranging from most to least abundant are reddish brown sandstones, siltstones, conglomeratic sandstones, conglomerates, and intercalated shale, respectively.

3. Physical Descriptions of the Hollow Concretions Most spaces of hollow concretions’ cores are empty with minute amounts of loose sand, silt, or clay particles or detached hard lumps of clay materials resembling nuts. These concretions appear in various sizes and shapes. Their average sizes vary from a few cm up to 60 cm in lengths and greatly depend on the grain sizes of the pre-existing clastic rocks of the formation (Fig. 2). The typical shapes of these concretions are spheroidal, ellipsoidal, distorted cylindrical, and irregular (Fig. 3). While embedded in the strata, these hollow rocks do not show any uniform patterns of distribution. The outer shells are signiﬁcantly harder than the surrounding rocks in which they are buried. When viewing the cross section, the inner part near the hollow has a reddish brown color, and such color fades outward making the outermost crust relatively pale. The inner surface of the hollows is usually rugged showing distinct corroded features. In some cases, there are remnants of pyrite crystals (FeS2 ) observed at the inner surfaces. We will explain in the next section that several species of iron solutions react as additional cementing media which hold the mantle of hollow concretions tighter than the sand grains farther away in the rock formations.

FA

December 10, 2008 15:45

16

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch02

P. Putthapiban and S. Hongsresawat

Fig. 2. The sizes of the hollow concretions are larger in the coarser-grained clastic rocks comparing to those of the ﬁner-grained one, (a), (b), and (c) show the concretions from the Study Area 1 and (d) shows the concretions from the Study Area 2.

Fig. 3. The adopted shapes of the hollow concretions showing their dependence on the pre-existing pyrite nodules. (a) and (b) are from the Study Area 1, and (c) and (d) are from the Study Area 2.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Formation of Hollow Concretions in Northeastern Thailand

b672-V13-ch02

17

4. Formation Processes Extensive examinations performed on hundreds of samples of the hollow concretions collected from and observed in the ﬁelds lead to the conclusion that the most suitable processes responsible for their formation are sedimentary depositions followed by a series of chemical reactions and ended with weathering and erosion processes. From our present information, the best formation scenario is chronologically described as follows. 4.1. Lithified stage The ﬁrst stage is the formation of clastic sedimentary rocks, siltstone, sandstone, shale, and conglomerate present in the Phu Kradung, Phra Wihan, and Sao Khua Formations by meandering and braided stream deposits. The existence of pyrite (FeS2 ) occurring as individual crystals, clusters of combined crystals or pyrite nodules, the existence of coal jets and carbonaceous materials suggest that these rock formations were formed under signiﬁcantly reducing conditions. These crystalline pyrites distribute unevenly throughout the rock formations. The gray to greenish gray color of the rock formations also indicates reducing environment of depositions. This color can only be observed when the rocks are not subjected to prior chemical weathering processes, fresh rocks. 4.2. Subsurface chemical reaction stage The second stage involves activities of subsurface water which sieves through fractures and grain boundaries of these pyrite-bearing rocks upon being exposed above the water table. The eﬀectiveness of this water transmission is mainly regulated by the porosity and permeability of the rocks; therefore, properties such as grain size, primary rock structures, and the complexity of fracture system are important. The presence of water and oxygen gradually converts the system into oxidizing environment. As a result, several chemical reactions are triggered and often release heat and pressure to the system where reactions occur. The following four chemical equations show the general accepted sequence of pyrite reactions12,13 : + 2FeS2 + 7O2 + 2H2 O → 2Fe2+ + 4SO2− 4 + 4H

4Fe2+ + O2 + 4H+ → 4Fe3+ + 2H2 O 4Fe3+ + 12H2 O → 4Fe(OH)3 + 12H+ + FeS2 + 14Fe3+ → 15Fe2+ + 2SO2= 4 + 16H .

(1) (2) (3) (4)

FA

December 10, 2008 15:45

18

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch02

P. Putthapiban and S. Hongsresawat

In Eq. (1), FeS2 which is stable under the previous reducing condition starts to react with oxygen and water to form ferrous iron (Fe2+ ), sulfate (SO2− 4 ), and acid (H+ ). Later, Fe2+ is transformed to ferric iron (Fe3+ ), as shown in Eq. (2). Fe3+ ions can undergo two diﬀerent reactions, either hydrolyzed with water to form the solid ferric hydroxide (ferrihydrite) (Fe(OH)3 ) shown in Eq. (3) or further consume more FeS2 releasing additional + Fe2+ , SO2− 4 , and H , as shown in Eq. (4). These subsequent oxidation reactions of the pyrite crystals cause the textural failure; consequently, the individual sand grains (mainly quartz) previously held in the sedimentary rock matrix fall oﬀ becoming loose sand and silt particles. The Fe2+ /Fe3+ solution percolates outward seeking appropriate environments with lower temperature and pressure for chemical reactions to continue. This radiallyoutward percolation ceases at the outer shell of the concretion after the system reaches equilibrium leaving behind the rusty color of iron oxide. In the presence of alkaline minerals such as feldspars, the H+ and water from the above equations will be naturally neutralized by presumably, the following chemical reactions: 4KAlSi3 O8 + 4H+ + 18H2 O → 4K+ + 2Al2 Si2 O5 (OH)4 + 8Si(OH)4 2CaAl2 Si2 O8 + 4H+ + 2H2 O → 2Ca2+ + Al4 Si4 O10 (OH)8

(5) (6)

The end product of both Eqs. (5) and (6) is mainly a kaolinite clay. This clay mineral can further react with SiO2 and water yielding the solution of silica and silicic acid (H2 SiO4 ). These iron oxide and silicic solutions which disperse throughout the mantle additionally strengthen the crust of the hollow concretions. If carbonate minerals such as calcite are present, the neutralized process can be described by the following equations, and the chemical weathering product will be the mineral gypsum: CaCO3 + 2H+ + SO2− 4 + H2 O → CaSO4 · 2H2 O + H2 CO3

(7)

− CaCO3 + H+ + SO2− 4 + H2 O → CaSO4 · 2H2 O + HCO3

(8)

The embedded FeS2 crystals and nodules in the rock formations are evidently the essential starting materials for these chemical reactions to persist. Final size and shape of the hollow concretions are directly derived from the quantity of FeS2 and the orientation of the FeS2 nodules (Fig. 4).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Formation of Hollow Concretions in Northeastern Thailand

b672-V13-ch02

19

Fig. 4. Successive developments of the hollow concretions. (a) is gray to greenish gray, fresh sandstone with pyrite and carbonaceous materials of the Phra Wihan Formation excavating from the river bed near the Pak Moon Dam site; (b) shows the alteration of the similar rocks shown in (a) after being exposed to air for some years. Rusty spots and banded circles evidently ascertain the validity of chemical reactions caused by weathering; (c) shows the rugged inner surface of the hollow previously underwent severe chemical reactions; (d) shows trapped loose sand, silt, and clay particles which are always observed in the undisturbed hollows.

4.3. Post-chemical erosion stage The last stage is the prolongation of weathering and erosion processes undertaking well after maturation of chemical reactions of the previous stage. The portions of the rock farther away from the crust of the concretion are subjected to erosion process much more severely and are easily removed. Without surrounding materials, some hollow concretions become free and can be transported to new depositional sites by various means. The remaining partially free hollow concretions are left embedded in situ within the outcrop exposures (Fig. 5).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

20

9in x 6in

b672-V13-ch02

P. Putthapiban and S. Hongsresawat

Fig. 5. The outcrops containing partially free hollow concretions embedded in situ at Phu Phrao, the Study Area 1.

5. Conclusion The origin of hollow concretions can be summarized by the lithiﬁed stage, the subsurface chemical reaction stage, and the post-chemical erosion stage. The lithiﬁed process of the clastic rocks occurs under signiﬁcantly reducing condition as indicated by the existence of pyrite nodules. Later on, the environment becomes more oxidized due to the presence of subsurface water and oxygen. Several chemical reactions take place forming hollows and strengthening outer shells of the concretions. The ﬁnal post-chemical erosion stage explains the mobility and the distribution of the hollow concretions observed in the rock outcroups.

Acknowledgments Mahidol University, Faculty of Science and Mahidol University, Kanchanaburi campus are thanked for their support and encouragement. The authors are grateful for the provision of basic geological information

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Formation of Hollow Concretions in Northeastern Thailand

b672-V13-ch02

21

by The Royal Thai Department of Mineral Resources. Sirot Sulyapongse, Manop Raksaskulwong, Apichat Lumjuan, and Nopadol Chaikum are thanked for oﬀering valuable discussions. The provisions of research facilities and necessary supports from Nakorn Hamah, Vice President for Campus Development, Mahidol University and Tanakorn Osotchan, Director of the Physics Department, Faculty of Science, Mahidol University are appreciated.

References 1. T. Kobayashi, F. Takai and I. Hayami, On some Mesozoic fossils from the Khorat Series of East Thailand a note on the Khorat Series, Geol. Palaeontol. SE Asia 1 (1964) 119–133. 2. D. E. Ward and D. Bunnag, Stratigraphy of the Mesozoic Khorat Group in Northeast Thailand, Department of Mineral Resources of Thailand Report of Investigation 6 (1964) 95pp. 3. R. Ingavat and P. Taquet, J. Geol. Soc. Thailand 3 (1978) 1–6. 4. N. Sattayarak, Review of the continental Mesozoic stratigraphy of Thailand, Proc. Workshop on Stratigraphic Correlation of Thailand and Malaysia, ed. P. Nutalaya (Had Yai, Thailand, 1983), pp. 127–148. 5. E. Buﬀetaut and V. Suteethorn, The biogeographical signiﬁcance of the Mesozoic vertebrates from Thailand, in Biogeography and Geological Evolution of Southeast Thailand, eds. R. Hall and J. D. Holloway (Backbuys, 1998), pp. 83–90. 6. A. Meesook, V. Suteethorn and T. Wongprayoon, Early Cretaceous nonmarine bivalves of the Sao Khua Formation, Khorat Group, Northeastern Thailand, 3rd Symposium IGCP 350, Manila, Philippines, Program and Abstract Volume (1995), pp. 10–11. 7. A. Meesook, V. Suteethorn, P. Chaodumrong, N. Teerarungsigul, A. Sardsud and T. Wongprayoon, Mesozoic rocks of Thailand: A summary, Proc. Symposium on Geology of Thailand, eds. N. Mantajit and S. Potisat (2002), pp. 82–94. 8. E. Buﬀetaut, V. Suteethorn, H. Tong, Y. Chaimanee and S. Khunsubha, New dinosaur discoveries in the Jurassic and Cretaceous of Northeastern Thailand, Proc. Int. Conf. Stratigraphy and Tectonic Evolution of Southeast Asia and the South Pacific (GEOTHAI 1997), eds. P. Dheeradilok, C. Hinthong, P. Chaodumrong, P. Putthapiban, W. Tansathien, C. Uthaaroon, N. Sattayarak, T. Nuchanong and S. Techawan, Vol. 1 (1997), pp. 177–187. 9. A. Racey, J. G. S. Goodall, M. A. Love, S. Polachan and P. D. Jones, New age data for the Mesozoic Khorat Group of Northeastern Thailand, Proc. Int. Symp. Stratigraphic Correlation of Southeast Asia, eds. P. Angsuwathana, T. Wongwanich, W. Tansathien, S. Wongsomsak and J. Tulyatid (Department of Mineral Resources, 1994), pp. 245–252.

FA

December 10, 2008 15:45

22

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch02

P. Putthapiban and S. Hongsresawat

10. L. S. Gardner, H. F. Howarth and P. Na Chiangmai, Salt resources of Thailand, Department of Mineral Resources, Report of Investigation 11 (1967) 100pp. 11. N. Sattayarak and V. Suteethorn, Geological Map of Thailand 1:500,000 (Northeastern Sheet) (Geological Survey Division, Department of Mineral Resources, Thailand, 1983). 12. W. Stumm and J. J. Morgan, Aquatic Chemistry: An Introduction Emphasizing Chemical Equilibria in Natural Waters, 3rd edn. (John Wiley and Sons, New York, 1996). 13. K. B. Krauskopf, Introduction to Geochemistry, International Series in the Earth and Planetary Sciences (McGraw-Hill Book Company, 1967).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch03

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

INVESTIGATIONS ON LOCAL QUARTZ SAND FOR APPLICATION IN GLASS INDUSTRY∗

PISUTTI DARARUTANA The Royal Thai Army Chemical Department, Phaholyothin Road, Chatuchak, Bangkok, 10900 Thailand; The Glass and Glass Products Research and Development Laboratory, Institute for Science and Technology Research and Development, Chiang Mai University; The Graduate School of Chiang Mai University [email protected] PRUKSWAN CHETANACHAN National Institute of Health, Department of Medical Sciences, Nonthaburi 11000, Thailand PORNSAWAT WATHANAKUL Department of Earth Sciences, Faculty of Science, Kasetsart University, Bangkok 10900, Thailand NARIN SIRIKULRAT Department of Physics, Faculty of Science, Chiang Mai University, Muang, Chiang Mai, 50200 Thailand; The Glass and Glass Products Research and Development Laboratory, Institute for Science and Technology Research and Development, Chiang Mai University

Silica or glass sand is a special type of quartz sand that is suitable for glassmaking, because of its high silica content, and its low content of iron oxide and other compounds. In Thailand, deposits of quartz sand are found as the beach and the river sands in many areas; eastern, southern, northeastern and northern. In this work, grain-size distribution and chemical analyses were carried out on 10 sand samples taken from various localities in Thailand such as Chanthaburi, Trat, Rayong, Chumphon, Nakhon Si, Pattani, Phuket, Songkhla, Nong Khai, and Tak provinces. The geological resources show that most of them are the surface-to-near-surface glass sand deposits. The sand grains in most deposits were mainly angular-to-rounded, except in some areas of either angular or rounded grains. Chemical analysis showed that the sands

∗ Work partially supported by The Graduate School of Chiang Mai University and The Oﬃce of the National Research Council of Thailand.

23

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch03

P. Dararutana et al.

24

contained more than 95wt% silica and low content of Fe, Al, Ca, Mg, Na, and K. The concentration levels of these components in the samples conﬁrm with internationally acceptable standard for glass production. The quartz sand dressing plants that used the spiral classiﬁer to improve the properties of the quartz sands to meet the standard speciﬁcations are mostly located in the eastern area. It can be concluded that most of the quartz sand deposits in Thailand investigated show well-sorted grain-size with considerable purity, i.e. high-grade quality. The advanced works resulted in that these raw quartz sands can be used as raw material for fabrication of soda-lime, lead crystal, and lead-free high refractive index glasses. The colorless and various colored glass products have been satisfactorily used in the domestic art and glass manufactures.

1. Introduction Glass is one of the oldest artiﬁcial materials known to man. The major raw materials of the most common glasses are the same as the oldest known glass recipe, and the process of glass making was recorded on clay tablets from the Royal Library of Assur-bani-pal at Nineveh (Iraq) in the seventh century BC. Silica, lime, and alkali are the bases of oxide glasses, which are of great importance, both historically and technologically. The source of silica is quartz sand.1−3 Sand consists of small grains or particles of minerals and rock fragments. Although these grains may be of any mineral composition, the dominant component of silica sand is the mineral quartz. Glass sand is a special type of sand that is suitable for glass making because of its high silica content and its low conduct of iron oxide and other compounds. It may be produced from both unconsolidated sands and crushed sandstones.4 The glass sand standards that was ﬁxed by the US Bureau of Standard ﬁxed the Sand Class Conditions for glass making,5 and the British standard methods for sampling and analyzing of glassmaking sand6 (Table 1). In Thailand, deposits of glass sand are found to be of both the beach and the river sands; most of them are beach sand,8 located in many areas; eastern, southern, northeastern, and northern.7 The geological Table 1. Standard

USA British

The standard of glass sand. Composition (%wt)

SiO2

Fe2 O3

Al2 O3

CaO + MgO

95 (min) 98.5 (min)

1 (max) 0.30 (max)

4 (max) —

0.5 (max) —

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch03

Investigations on Local Quartz Sand for Application in Glass Industry

25

resources show that they are mainly the surface-to-near-surface glass sand deposits. The previous works indicated that the other local raw materials including zircon sand and basalt rock also met the requirements for the glass industrial process.9,10 The local quartz sand from Trat area was satisfactory for the lead crystal glass fabrication.11 In this work, the physical structure, grain-size distribution, and chemical analyses of quartz sand samples taken from various localities in Thailand were investigated.

2. Experiment The geological explorations of silica sand were carried out in 10 various localities in Thailand such as Chanthaburi, Trat, Rayong, Chumphon, Nakhon Si, Pattani, Phuket, Songkhla, Nong Khai, and Tak provinces. The sand samples of about 10–100 k from each locality were taken to analyze their properties. The samples were prepared using standard metallographic technique prior to other speciﬁc experimental investigations. The physical structure investigations of the samples were determined using scanning electron microscope (SEM); Jeol JSM-5910, operated at 20 kV. The chemical compositions were determined by the wavelengthdispersive X-ray ﬂuorescence spectrometer (WDXRF), Phillips MagixPro PW 2400, operated with LiF 200 crystal, scintillation and ﬂow proportional detectors, with the Rh-tube at 60 kV 125 mA. The neutron activation analysis (NAA) was used to determine the trace elements. The grain distributions of the sand samples from each locality were analyzed by using the particle size analyzer, Malvern Instrument Mastersizer, using polydisperse analysis, and operated with a beam length of 240 mm and range lens 300 RF mm.

3. Results and Discussion The geological explorations showed that most of the sand resources from 10 localities in Thailand (Chanthaburi, Trat, Rayong, Chumphon, Nakhon Si, Pattani, Phuket, and Songkhla provinces) were the surfaceto-near-surface glass sand deposits, except at Nong Khai site that was the subsurface deposit. Most of them were the beach sand, except that of Nong Khai site that was the river sand, and at Tak deposit was the by-product from the feldspar ﬂoatation plant. The beach sands were

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch03

P. Dararutana et al.

26

Fig. 1.

The micrographs of the local sands studied.

deposited as lenses, of more than 100 m2 and approximately 0.2–2.0 m thick each, covering the wide area possibly about 1–3 km2 in each locality. Scanning electron microscope investigation (Fig. 1) revealed that the grain shape of the older beach sands (Chanthaburi, Trat, and Pattani) were angular to rounded. The present beach sands of Rayong, Chumphon, and Nakhon Si, showed the angular grain, whereas the Songkhla and Phuket sites were the rounded and the angular-to-rounded grains, respectively. The river sand from Nong Khai site was the angular-to-round grain, mixed with

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch03

Investigations on Local Quartz Sand for Application in Glass Industry

27

calcium carbonate in composition. The smaller grain sizes and angular grain were found at Tak site. Most of the older beach sand showed ﬁne-grain, and also most of the present beach sand, except the Nakhon Si site showed ﬁne to coarse grain. The ﬁne grain was founded from Nong Khai and Tak sites. The color of the older beach sand was pink–gray, except the one from Rayong site that was orange–gray. Most of the present beach sands were pink–gray, but those at Phuket site were gray in color. Both the river sand from Nong Khai and the feldspar-ﬂoated sand from Tak sites showed white color. The origin, grain shape, and color of the local sands are shown in Table 2. Results of the grain-size analysis showed that the sizing of the sand particles was between 20 to 510 µm, as shown in Table 3, that fall within the ideal sand fraction range used for the glass production. Chemical analysis showed that the sands contain more than 95wt% silica and low content of Fe2 O3 , Al2 O3 , CaO, MgO, Na2 O, and K2 O, as shown in Table 4. Because of its origin of occurrences, the quartz glass sand from Tak site contained high silica and low impurities due to being the product of feldspar ﬂoatation. The concentration silica and the minor components in most of the samples conﬁrmed with the internationally acceptable standards for glass production. Moreover, the trace elements using neutron activation analysis showed the presence of Hf, Os, Ti, Yb, and Se in the beach sand. All of these diﬀerential properties may be caused from the chemical and the physical weathering, and the biotic eﬀect that erode the original source rocks. The quartz sand dressing plants that used the spiral classiﬁer to improve the properties of the quartz sands to meet the wanted speciﬁcations were mostly founded in the eastern area. Table 2. Sites Rayong Chanthaburi Trat Chumphon Songkhla Phuket Pattani Nakhon Si Nong Khai Tak

The origin, grain shape, and color of the local glass sands. Origin

Grain shape, texture

Older beach Older beach Present beach Present beach Present beach Present beach Old beach Present beach River By-product

Angular-to-rounded, ﬁne Angular-to-rounded, ﬁne Angular, ﬁne Angular, ﬁne Rounded, ﬁne Angular-to-rounded, ﬁne Angular-to-rounded, ﬁne Angular, ﬁne-to-coarse Angular-to-rounded, ﬁne Angular, ﬁne

Color Orange–gray Pink–gray Pink–gray Pink–gray Pink–gray Gray Pink–gray Pink–gray White White

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch03

P. Dararutana et al.

28 Table 3.

The distribution of local quartz sands. Grain size, µm

Sites D(V , 0.1)

D(V , 0.5)

D(V , 0.9)

132 138 168 150 140 201 155 54 185 20

191 250 248 219 222 331 241 224 252 38

264 395 344 308 322 510 349 477 423 139

Rayong Chanthaburi Trat Chumphon Songkhla Phuket Pattani Nakhon Si Nong Khai Tak

Table 4.

Speciﬁc area m2 /g

0.148 0.138 0.166 0.146 0.119 0.212 0.140 0.118 0.158 0.026

0.043 0.057 0.046 0.043 0.055 0.040 0.050 0.050 0.031 0.474

The composition of local quartz sands.

Sites

Rayong Chanthaburi Trat Chumphon Songkhla Phuket Pattani Nakhon Si Nong Khai Tak

Concentration %vol

Composition SiO2

Fe2 O3

Al2 O3

CaO + MgO

Na2 O + K2 O

97.5 98.8 98.8 99.1 96.9 97.0 96.0 95.2 97.6 99.2

0.31 0.05 0.05 0.03 0.12 0.37 0.10 0.70 0.14 <0.01

0.95 0.77 0.56 0.57 1.85 0.85 2.60 1.65 1.24 0.36

0.03 0.01 <0.01 <0.01 0.01 0.01 0.60 0.13 0.03 <0.01

0.14 0.22 <0.01 0.02 0.57 0.03 <0.01 1.35 <0.01 0.20

4. Conclusion It can be concluded that the investigations on the quartz sand deposits in Thailand were very useful; the results showed that most of them were of well-sorted grains with considerable purity with high silica content, i.e. high-grade quality. The consequent works have revealed that these raw quartz sands can be used as raw materials for fabrication of soda-lime, lead crystal, and lead-free high refractive index glasses for the colorless and various colored glass products, and found to be satisfactory for the art and domestic glass manufacturers.12−14

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch03

Investigations on Local Quartz Sand for Application in Glass Industry

29

Acknowledgments Uses of the scanning electron microscope and the wavelength-dispersive X-ray ﬂuorescence spectrometer were supported by the Faculty of Science, Chiang Mai University at Chiang Mai Province. The particle size analyzer was supported by the Northern Ceramics Development Center at Lumpang Province. The Glass and Glass Products Research and Development Laboratory, Institute for Science and Technology of Chiang Mai University jointly supported the experiments of glass fabrication in the laboratory scale.

References 1. F. L. Harding, Introduction to Glass Science (Plenum Press, 1972). 2. S. Frank, Glass and Archaeology (Academic Press, 1982). 3. K. P. Kursula, PIXE and SEM Study of Old Finnish and European Glass and European Oyster Ostrea edulis, University of Helsinki, 1999. 4. Baumgart et al., Process Minerology of Ceramic Materials (Elsevier, 1984). 5. Glass Production Guide: Production of the US Glass Association (1965). 6. BS 2975 British Standard Methods for Sampling and Analyze of Glassmaking Sand (1988). 7. P. Ratanajaroorak, Glass sands, Department of Mineral Resources of Thailand (2005). 8. B. Jarooglas, Glass sands, Dep. Mineral Res. Econ. Geol. Bull. 35 (1982) 1–60. 9. P. Pimkhaokham and R. Conradt, Potential utilization of diﬀerent local raw materials, Chulalongkorn University (1990). 10. P. Pimkhaokham et al., Raw material substitution in technical glass melting, Chulalongkorn University (1990). 11. K. Visetchat et al., Abstracts of the 29th Congress on Science and Technology of Thailand (2003), p. 202. 12. C. Iamjitkusol et al., Fabrication of glass for art’s work, Oﬃce of the National Research Council of Thailand (2004). 13. P. Dararutana and N. Sirikulrat, J. Micros. Soc. Thailand 20 (2006) 1. 14. P. Dararutana and N. Sirikulrat, KMITL Sci. J. 6 (2006) 2b.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

This page intentionally left blank

b672-V13-ch03

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

GEOCHEMICAL AND Sr–Nd–Pb ISOTOPIC STUDY OF LATE NEOGENE VOLCANIC ROCKS FROM THE ARITA–IMARI AREA (SW JAPAN): EVIDENCE FOR COEXISTING OIB-LIKE AND SUBDUCTIONRELATED MANTLE SOURCES NGUYEN HOANG∗ , JUN’ICHI ITOH† , KOZO UTO and AKIKAZU MATSUMOTO Geological Survey of Japan, AIST, Higashi 1-1-1 Tsukuba Central 7th, Tsukuba, Japan 305-8567 ∗ [email protected] † [email protected]

New geochemical and isotopic (Sr, Nd, Pb) data are reported for Neogene basaltic to high-K andesitic, dacitic, and rhyolitic lavas from four closely spaced volcanic centers in the back arc region of the Arita–Imari area, NW Kyushu, Japan. Two major eruptive groups are recognized: (1) 6.7 Ma small volcanoes such as Kuro-dake and Jinroku-yama comprise basaltic to andesitic stratocones that form a discreet volcanic ediﬁce; (2) the 2.7–2.3 Ma Koshidake and Arita volcanic complex consisting of basaltic to high-silica volcanic products, erupted in the form of multiple vents. Andesitic types comprise about 90 vol.% of the erupted lavas, overall. The basalts conform broadly to oceanic island basalt (OIB) type, characterized by relatively high MgO (ca. 6wt%), low SiO2 (ca. 48.5wt%), and “smooth” (high ﬁeld strength element-rich: HFSE-rich) mantle-normalized trace element patterns. They are isotopically enriched, with 87 Sr/86 Sr ranging 0.7040–0.7048, 143 Nd/144 Nd, 0.5128–0.5127, and Pb isotopes ranging between enriched mantle types EM1 and EM2like indicating small-scaled source heterogeneity. The Sr and Nd isotopic compositions of the andesite and rhyolite are even more enriched (0.7044– 0.7052, 0.51270–0.51267, respectively), however, within the range of northern Kyushu Cenozoic basalts. Chemically they fall in the calc-alkaline ﬁeld and show strong HFSE-negative anomalies in mantle-normalized trace element patterns. Two chemically distinct groups are identiﬁed among the rhyolites, associated with (a) Arita and (b) Koshi-dake. Distinguishing geochemical parameters of each are, respectively as follows: (La/Yb)N (a) between 12 and 14.5, and (b) < 7; Ba/Nb (a) from 25 to 40, and (b) < 15; Rb/Sr (a) < 0.5, and (b) > 4.5; 87 Sr/86 Sr (a) 0.7047, and (b) 0.7052. The combined presence of relatively high 87 Sr/86 Sr isotope ratios and negative HFSE anomalies may be interpreted in terms of at least three potential eﬀects: partial melting of, or magmatic assimilation by, lithospheric mantle and/or crustal material, contamination of the magmatic source by subducting slab-derived hydrous

31

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

N. Hoang et al.

32

ﬂuids, and the presence of refractory HFSE-rich phases during magma genesis and/or ﬂuid segregation from the slab. The geochemical and isotopic characters of the Arita andesite–rhyolite series strongly suggest a combined history of assimilation–fractionation crystallization (AFC). On the other hand, the Koshidake rhyolites, while showing even higher 87 Sr/86 Sr ratios, exhibit signiﬁcantly lower abundances of incompatible elements in general, and higher contents of heavy rare earth elements (HREE) as compared to their accompanying basalts, not readily explained via simple fractional crystallization or AFC processes.

1. Introduction Highly silicic magmas may be derived from mantle-equilibrated parent magmas through a variety of diﬀerentiation processes,1 including the assimilation of crustal material in association with fractional crystallization (AFC),2,3 or partial remelting of trapped andesitic or basaltic intravolcanic bodies in the middle or upper crust.4,5 Detecting the involvement of crustal materials is relatively easy given its distinctive, enriched character in terms of LILE, LREE, and radiogenic Sr and Pb isotopic characters.6 However, recognition of the more subtle distinctions between mantlederived signatures is handicapped by the superimposed eﬀects of such intracrustal fractionation eﬀects that may produce similar chemical and isotopic characteristics. In any case, the recognition of potential crustal contamination sources in magmas is an essential prerequisite to any successful petrogenetic model for SiO2 -rich magmas and, in turn, to interpreting the chemical evolution of continental crust through time. In northern Kyushu, southwestern Japan, Neogene intraplate volcanic activity produced widely scattered, mainly basaltic centers.7−11 The activity was sporadic, quiescent episodes being interpreted to correlate with renewed lithospheric extension triggered by plate-kinematic adjustments following the opening of the Japan Sea (20–14 Ma).12 Most workers have concluded that the dominant controls on the geochemistry of the intraplate basalts were related to their spatial and, to some extent, temporal distribution, reﬂecting both the geochemical evolution of their source regions and dominant intracrustal fractionation mechanisms.8,9,11 Abundant silicic (SiO2 > 75wt%) and andesitic eruptive products, also accompanying the basalts,7 either as a calc-alkaline continuum (NishiSonogi, Iki island, etc.) or as bimodal maﬁc — acidic associations (Saga– Futagoyama10). Early Pliocene to Pleistocene rhyolite and andesite complexes, along with subsidiary basalts, appear within an area of <100 km2 in the Arita– Imari area, in the eastern part of Kita-Matsuura, representing the largest

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

Evidence for Coexisting OIB-Like and Subduction-Related Mantle Sources

33

region of intraplate volcanism in northwestern Kyushu post-dating opening of the Japan Sea.7,13,14 Their volcanic products form distinctive mountain peaks, overlying, and rising above the Tertiary sedimentary basement. From north to south, these include a bimodal association of basalt and rhyolite (2.7–2.6 Ma), in the vicinity of Koshi-dake, basalt and andesite

Fig. 1. Location map of study area: (a) Geological scheme of the Arita–Imari area; (b) Simpliﬁed from Ref. 13. Numbers are Ar/Ar ages. Quaternary volcanic front (dashed line in (a)) is also shown for reference.

FA

December 10, 2008 15:45

34

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

N. Hoang et al.

(5 Ma), at Kuro-dake, Sera-san, in the Arita complex — comprising 2.4– 2.3 Ma andesite, rhyolite, pyroclastics, along with a minor amount of basalt, and association of basalt and basaltic andesite, aged 6.8 Ma, in the Jinroku-yama area15,16 (Fig. 1). With the exception of the large-volume Arita volcanic complex (3 km3 in area), the other bodies do not exceed 0.5 km3 in area. The acidic to intermediate rocks typically consist of biotite rhyolite, hornblende-clinopyroxene dacite, olivine clinopyroxene andesite, and a range of biotite-hornblende dacite to rhyolite. Also encountered are aphyric andesites, rhyolites, obsidian, and (subsidiary) perlite. Xenoliths of basalt and andesite commonly occur in the silicic lavas and pyroclastic of the Arita complex. Petrographic studies of these show that some of the former retain ﬂuid-like morphologies, along with the evidence of magma co-mingling and solid-melt reaction in the basalt inclusions and their host rhyolites (Fig. 2).

Fig. 2. Micro-photograph of a thin section showing shape and marginal contact of maﬁc inclusions in rhyolitic lavas in the Arita volcanic complex. (See text for detail.)

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

Evidence for Coexisting OIB-Like and Subduction-Related Mantle Sources

35

A large sample set, including the relatively rare basalts and more abundant andesites and rhyolites, was collected from four of the Arita– Imari centers, for K–Ar age dating, along with petrologic, geochemical, and Sr–Nd–Pb isotopic studies. The objective of the study is to better deﬁne regional magmatic evolution patterns and interpret these in terms of magmatic source heterogeneity and possible intracrustal fractionation eﬀects within a well-deﬁned spatial–temporal framework. 2. Analytical Techniques The analytical work was conducted at the Geological Survey of Japan (GSJ). Procedures for X-ray ﬂuorescence (XRF) major elements analyses have been described elsewhere,9,11 reported accuracies being better than ±1.5% based on repeated measurements of GSJ standard JB-1a. Trace elements were analyzed by inductively coupled plasma mass spectrometry (ICP-MS) on a Micromass Platform-ICP. Details of this procedure were reported by Hoang and Uto.17 ICP-MS running parameters are similar to those reported previously,18 reproducibility being better than ±4% (2σ) for rare earth elements, Rb, and Nb, and better than ±6% for other elements. For the analysis of Sr, Nd, and Pb isotopic compositions, 1–2 mm chipped samples were cleaned ultrasonically for about 30 mins, then dried and ground in an agate mill. All the acids and the water used during chemical work were certiﬁed TAMA-Pure AA-10 grade. Approximately, 50 mg aliquots of powder were dissolved in concentrated HNO3 and HF (ratio 1:2), re-dissolved in HNO3 before being evaporated. Sr-spec resin R from Eichrom was used in Sr and Pb extractions, following procedures described by Deniel and Pin.19 Preliminary extraction of rare earth elements (REE) was accomplished using conventional AG50W-X8 resin in small quartz columns, Nd being extracted using Ln-resin (Eichrom) and 0.25 N HCl as eluant, following procedures described by Pin and Zalduegui.20 Nd, Sr, and Pb isotopic ratios were measured in a multicollector VG Sector 54 thermal ionization mass spectrometer (TIMS) at the GSJ. Details of TIMS running parameters were reported by Hoang and Uto.9,21 The within-run precision (2σ) for 87 Sr/86 Sr was about ±0.000007 and about ±0.000008 for 143 Nd/144 Nd. The internal precision of Pb isotopic ratios (2σ) is less than 0.01%, and total blank is less than 80 pg. The data are shown in Table 1.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

b672-V13-ch04

N. Hoang et al.

36 Table 1.

9in x 6in

Chemical and Sr, Nd, and Pb isotopic compositions of the Arita–Imari lavas.

Sample Location

NAG02003 Koshi-dake

NAG02001 Koshi-dake

Age (m.y.) SiO2 TiO2 Al2 O3 FeO* MnO MgO CaO Na2 O K2 O P2 O5 Sum Mg # Zn Cu Ni Cr V Rb Sr Y Zr Nb Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta Pb Th U 87 Sr/86 Sr 143 Nd/144 Nd εNd 206 Pb/204 Pb 207 Pb/204 Pb 208 Pb/204 Pb

2.61 48.94 2.66 16.57 11.96 0.18 5.97 8.37 3.43 1.39 0.53 100 47.07 101.6 38 75.4 157.3 233.1 28.8 443.87 31.23 175.37 31.49 354.75 24.6 50.41 6.05 27.74 6.62 2.18 6.68 1.04 5.74 1.16 2.99 0.46 2.66 0.39 4.21 2.01 3.29 2.77 0.6 0.704111 0.51278 2.77 17.87 15.438 38.107

2.6 53.58 1.34 14.79 8.26 0.15 7.85 9.5 2.78 1.46 0.28 100 62.88 102.5 39.8 72 147.7 226.9 30.69 444.89 33.94 174.42 30.66 367.32 23.52 49.69 6.11 26.93 6.45 2.2 6.48 1.01 5.66 1.12 3.09 0.45 2.62 0.39 4.01 1.91 2.95 2.71 0.58 0.704113 0.512774 2.65 17.924 15.488 38.238

NAG02004A Koshi-dake 2.71 76.43 0.03 13.09 1.16 0.05 0.03 0.64 3.99 4.56 0.02 100 4.47 41.3 1.6 32.8 118.9 178.35 45.07 25.13 71.88 20.17 228.99 23.36 42.65 4 13.6 3.11 0.22 3.39 0.6 3.92 0.83 2.31 0.37 2.27 0.33 2.98 2.7 23.53 16.31 4.78 0.705249 0.512691 1.03 18.346 15.582 38.575

NAG02022 Arita 2.4 48.75 2.68 16.72 12.04 0.19 5.74 8.37 3.52 1.46 0.54 100 45.94 102.4 40.4 62.5 111.7 230.8 31.54 470.99 32.18 175.76 30.34 368.23 24.11 50.41 6.17 27.94 6.74 2.28 6.43 1.03 5.72 1.13 3.04 0.42 2.59 0.4 4.15 1.53 2.3 2.54 0.53 0.704122 0.51275 2.18 17.956 15.515 38.32

NAG02018 Arita 2.4 57.14 1.32 16.29 8.11 0.17 3.56 7.06 3.7 2.29 0.36 100 43.94 78.4 30.2 73.7 266.7 140.9 84.29 381.52 27.23 200.96 24.45 485.55 27.24 52.17 5.54 22.3 4.61 1.36 4.33 0.68 4.16 0.84 2.28 0.35 2.14 0.33 4.51 1.76 7.99 6.67 1.6 0.70458 0.512699 1.19 18.287 15.576 38.548 (Continued)

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

Evidence for Coexisting OIB-Like and Subduction-Related Mantle Sources Table 1.

37

(Continued)

Sample Location

NAG02013 Arita

NAG02014 Arita

NAG02007 Arita

NAG02021 Arita

Age (m.y.) SiO2 TiO2 Al2 O3 FeO* MnO MgO CaO Na2 O K2 O P2 O5 Sum Mg # Zn Cu Ni Cr V Rb Sr Y Zr Nb Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta Pb Th U 87 Sr/86 Sr 143 Nd/144 Nd εNd 206 Pb/204 Pb 207 Pb/204 Pb 208 Pb/204 Pb

2.4 58.53 1.13 16.26 7.3 0.13 3.46 6.74 3.72 2.45 0.26 100 45.82 76.8 24.4 88.7 396.2 138.3 69.45 345.91 24.14 215.43 22.41 487.2 30.21 54.87 5.64 22.29 4.54 1.41 4.48 0.71 4.06 0.84 2.42 0.37 2.23 0.35 4.88 1.62 10.54 7.49 1.61 0.704466 0.512706 1.32 18.288 15.58 38.546

2.4 58.44 1.16 16.33 7.28 0.13 3.55 6.79 3.65 2.41 0.27 100 46.53 79.7 24.3 89.3 413 139.5 70.21 357.54 24.43 219.48 23.61 500.01 29.45 53.93 5.66 22.04 4.57 1.42 4.27 0.67 3.8 0.8 2.37 0.32 2.09 0.33 4.79 1.25 8.11 6.57 1.59 0.704454 0.512691 1.04 18.285 15.57 38.53

2.4 60 1.23 17.21 6.71 0.14 2.27 5.31 4.5 2.25 0.38 100 37.65 91.7 15.6 52.1 228.2 125.6 61.79 515.32 30.36 263.09 25.99 543.64 41.19 77.7 8.92 32.37 6.32 1.81 5.65 0.85 4.81 0.95 2.63 0.38 2.48 0.39 6.05 1.34 10.22 5.7 1.26 0.704664 0.512678 0.77 18.341 15.583 38.563

2.4 61.63 1.34 16.74 4.87 0.1 2.5 5.73 3.94 2.75 0.39 100 47.75 94.8 26.5 31.9 118.7 134.5 76.1 588.63 22.89 193.87 24.72 467.05 36.89 75.31 8.2 32.42 6.35 1.61 5.05 0.74 3.74 0.73 1.96 0.29 1.85 0.29 4.86 1.4 11.97 8.06 1.71 0.705121 0.512682 0.86 18.347 15.576 38.555

NAG02015 Arita 2.4 63.37 0.89 15.56 5.34 0.1 2.39 5.18 3.8 3.16 0.22 100 44.36 71 16.3 52.9 228.5 96.9 88.16 282.75 23.48 198.49 21.59 512.75 31.67 55.01 5.51 20.9 4.08 1.21 3.94 0.62 3.63 0.74 2.12 0.33 1.99 0.3 4.33 1.68 13 10.18 2.19 0.704506 0.512715 1.51 18.307 15.578 38.54 (Continued)

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

N. Hoang et al.

38

Table 1.

(Continued)

Sample Location

NAG03202 Seira-san

NAG03211 Seira-san

NAG03207 Seira-san

NAG02009 Arita

Age (m.y.) SiO2 TiO2 Al2 O3 FeO* MnO MgO CaO Na2 O K2 O P2 O5 Sum Mg # Zn Cu Ni Cr V Rb Sr Y Zr Nb Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta Pb Th U 87 Sr/86 Sr 143 Nd/144 Nd εNd 206 Pb/204 Pb 207 Pb/204 Pb 208 Pb/204 Pb

2.4 59.38 1.28 17.32 7.41 0.09 1.86 5.52 4.63 2.12 0.39 100 30.85 100.4 19.5 6.4 128.5 6 52.71 529.55 28.65 250.97 24.36 537.97 41.16 78.23 8.84 34.16 6.39 1.88 6.15 0.91 5.02 0.99 2.82 0.42 2.61 0.39 5.89 1.68 10.79 6.23 1.26 0.704686 0.512667 0.56 18.34 15.579 38.561

2.4 66.05 0.82 15.1 5.39 0.06 1.67 4.01 3.69 3.07 0.12 100 35.57 74.3 14 9.6 78 119.7 101.14 201.35 27.71 137.7 15.06 442.12 35.38 54.12 6 22.33 4.38 0.93 4.63 0.74 4.15 0.83 2.41 0.36 2.24 0.33 3.41 1.45 15.27 10.65 2.2 0.704968 0.51267 0.63 18.392 15.615 38.687

2.4 68.62 0.54 15.72 3.88 0.04 0.67 2.6 4.13 3.68 0.11 100 23.61 55.5 12 10.2 50.8 14.2 120.56 225.49 28.37 214.7 18.73 632.85 57.01 91.21 9.31 32.49 5.56 1.21 5.06 0.76 4.2 0.85 2.46 0.39 2.28 0.35 4.44 1.71 18.65 14.52 2.98 0.704754 0.512667 0.57 18.358 15.572 38.576

2.32 71.22 0.44 16.25 1.39 0.01 0.17 1.93 4.12 4.33 0.14 100 18.04 43.8 6.7 28.3 106.9 16.9 148.28 214.08 23.13 306.61 23.96 708.08 43.81 75.66 7.51 27.03 4.89 1.08 3.98 0.6 3.45 0.75 1.86 0.27 2.07 0.23 4.68 2.08 18.63 15.77 3.63 0.704834 0.512672 0.67 18.379 15.59 38.606

NAG02011 Arita 2.4 76.55 0.05 13 1.14 0.04 0.02 0.7 3.71 4.77 0.02 100 3.17 35.9 0.6 7.6 22.5 — 133.54 62.37 18.21 85.94 13.83 496.77 34.33 60.42 5.52 18.2 3.16 0.36 2.65 0.42 2.6 0.55 1.5 0.24 1.61 0.25 2.99 1.9 20.91 16.35 3.53 0.705133 0.512674 0.7 18.361 15.581 38.574 (Continued)

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

Evidence for Coexisting OIB-Like and Subduction-Related Mantle Sources Table 1.

39

(Continued)

Sample Location

NAG02005 Arita

NAG02028 Kuro-dake

NAG02026 Kuro-dake

NAG02025 Kuro-dake

NAG9301 Kuro-dake

Age (m.y.) SiO2 TiO2 Al2 O3 FeO* MnO MgO CaO Na2 O K2 O P2 O5 Sum Mg # Zn Cu Ni Cr V Rb Sr Y Zr Nb Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta Pb Th U 87 Sr/86 Sr 143 Nd/144 Nd εNd 206 Pb/204 Pb 207 Pb/204 Pb 208 Pb/204 Pb

2.4 74.51 0.15 14.33 1.7 0.01 0.11 0.85 3.83 4.5 0.02 100 10.44 48.5 2.9 16.1 62.4 2.1 143.97 90.53 19.01 171.36 17.42 654.78 44.45 67.51 6.37 20.6 3.42 0.44 2.88 0.44 2.51 0.53 1.54 0.24 1.69 0.26 4.52 1.45 21.71 12.3 3.16 0.705139 0.512673 0.68 18.374 15.59 38.606

5 51.81 1.31 18.3 8.65 0.13 4.95 10.64 3.03 0.91 0.26 100 50.52 85.7 42.9 38 113.2 271.2 17.93 721.51 23.63 105.62 8.71 272.02 18.99 39.31 4.67 20.18 4.28 1.31 3.97 0.65 3.76 0.77 2.1 0.32 1.91 0.29 2.61 0.58 4.04 3.31 0.67 0.703903 0.512792 3 18.186 15.507 38.29

5 54.04 1.9 17.66 8.62 0.13 4.07 8.75 3.16 1.27 0.4 100 45.68 94.1 43 62.5 164 232.7 28.22 545.26 26.1 180.72 20.51 277.79 27.29 56.5 6.57 28.99 5.97 1.86 5.33 0.84 4.35 0.88 2.39 0.35 2.01 0.31 4.36 1.02 6.18 4.08 0.74 0.704871 0.512682 0.87 18.337 15.568 38.584

5 55.44 1.5 16.22 10.82 0.14 2.89 5.92 4 2.46 0.62 100 32.25 121.4 29.9 63.2 161 86 72.46 378.14 36.6 302.69 45.9 633.19 41.51 87.18 9.87 40.32 8.06 2.6 7.52 1.11 6.31 1.23 3.36 0.5 2.92 0.44 6.79 2.97 8.76 7.78 1.88 0.704917 0.512645 0.14 18.335 15.565 38.708

5 54.72 2.17 17.87 7.95 0.13 3.91 7.98 3.72 1.25 0.31 100 46.68 91 27 38.1 218 26.1 19 665 21 171 19 402 23.36 48.2 — 21.7 5.21 1.67 — 0.83 — — — — 1.85 0.23 4.08 1.34 — 3.02 1.19 0.70454 0.512738 1.95 18.362 15.566 38.572 (Continued)

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

N. Hoang et al.

40

Table 1.

(Continued)

Sample Location

NAG02024 Kuro-dake

NAG02032 Jinroku-yama

NAG02029 Jinroku-yama

Age (m.y.) SiO2 TiO2 Al2 O3 FeO* MnO MgO CaO Na2 O K2 O P2 O5 Sum Mg # Zn Cu Ni Cr V Rb Sr Y Zr Nb Ba La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta Pb Th U 87 Sr/86 Sr 143 Nd/144 Nd εNd 206 Pb/204 Pb 207 Pb/204 Pb 208 Pb/204 Pb

5 57.56 1.35 15.59 7.83 0.14 4.74 6.7 3.37 2.4 0.32 100 51.89 105.8 36.7 127.6 384.8 151.6 134.86 339.15 37.2 165.89 43.33 284.4 26.15 53.13 6 25.29 5.81 1.34 6.01 1.06 6.11 1.29 3.49 0.54 3.16 0.44 4.59 4.05 16.24 10.86 3.11 0.704638 0.512687 0.96 18.37 15.568 38.71

6.77 50.45 2.32 15.88 11.72 0.16 5.13 10.16 2.95 0.91 0.3 100 43.84 110.3 43.2 149.7 327.2 182.9 26.27 389.06 28.51 155.68 23.5 296.09 18.3 39.6 4.83 21.75 5.18 1.73 5.36 0.84 4.88 0.97 2.51 0.36 2.01 0.3 3.67 1.49 1.65 2.42 0.5 0.704125 0.512765 2.47 17.988 15.493 38.673

6.77 56.14 1.08 15.8 7.66 0.12 5.59 9.07 2.91 1.44 0.18 100 56.55 72.9 32.5 68.8 329.2 215.6 42.7 368.08 24.48 115.03 12.82 341.64 20.84 35.57 4.43 19.38 4.24 1.37 4.07 0.7 3.78 0.8 2.15 0.34 2.09 0.33 2.85 0.78 5.32 4.55 1.02 0.704439 0.512722 1.63 18.297 15.553 38.55

NAG02030 Jinroku-yama 6.77 56.69 1.12 14.78 8.41 0.13 7.06 7.4 2.97 1.29 0.15 100 59.94 78.4 34.4 139.9 503.6 155.6 38.82 260.41 21.07 108.8 9.96 283.25 14.13 27.45 3.15 13.07 3.22 1.04 3.53 0.56 3.42 0.72 1.95 0.31 1.8 0.28 2.63 0.82 4.39 4.36 0.94 0.704141 0.512784 2.85 18.289 15.55 38.525

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

Evidence for Coexisting OIB-Like and Subduction-Related Mantle Sources

41

3. Analytical Results 3.1. Major element compositions Phenocrysts in olivine-phyric basalt and andesitic basalt (SiO2 <57wt%) are 0.5–2 mm in size, showing multiple generations, accompanied by elongate (less than 1.5 mm) plagioclases (Pl) (<7 in vol.%) in an Ol-bearing matrix. Typical andesites (SiO2 from 57 to 63wt%) show orthopyroxene phenocrysts (up to 1 by 2 mm) and tabular plagioclase (ca. 1 by 2 mm), up to 15 vol.% in some cases, accompanied by small amounts of mica and hornblende (NAG02008). The dacites (SiO2 <70wt%) and rhyolites (SiO2 >70wt%) typically show phenocrysts of albitic plagioclase, quartz, and subsidiary mica, aphyric exceptions recorded from Koshi-dake (SiO2 >76wt%). The basalt samples show compositional diﬀerences both within and between each volcanic center (Table 1). For example, SiO2 contents range between 48 and 53 (wt%) and Mg-numbers ((100 × (Mg/(Fe2+ + Mg)) between 47 and 63 in the 2.6 Ma in the Koshi-dake basalts, while the 5 Ma Kuro-dake basalts show SiO2 contents of 52–54 (wt%) and corresponding Mg-numbers of 51–47. Basalts from both areas are distinct from the 6.8 Ma Jinroku-yama basalts which show SiO2 contents between 50.5 and 56 (wt%) and corresponding Mg-numbers of 44–60 (Table 1). However, representative basalt compositions from Arita resemble those of the low-SiO2 Koshidake basalt (Table 1). Overall, the mentioned diﬀerences are likely to indicate one or more of the following eﬀects: source heterogeneity24 and/or varying partial melting parameters — including pressure, temperature, and PH2 O .25,26 Plots of SiO2 vs K2 O in Fig. 3(a) illustrate the major element distinctions of basalts and andesitic basalts with K2 O contents <1.5 (wt%) that occupy a broad region within the so-called “medium-K” ﬁeld.22,27 The voluminous Arita andesites show similar major element variation, Mg-numbers ranging from 37 to 50 (MgO = 2.3 and 3.4, and FeO* = 6.7 and 6wt%, respectively) although TiO2 contents (0.9–1.4wt%) and K2 O (2.3–3.7wt%), are somewhat higher, conforming to the high-K ﬁeld (Fig. 3(a)). This is also the case for the dacites and rhyolites, for example, in Koshi-dake (SiO2 ranging 67–>76wt%, rhyolite) and K2 O, 2.5–4.5 (wt%) aside from calc-alkaline magmas of medium-K aﬃnity, such as those from SW Japan5,28 (Table 1, Fig. 3(a)). In contrast, much of the Jinroku-yama basalts and Arita andesites fall in the calc-alkaline ﬁeld as deﬁned by the plots of FeO*/MgO versus SiO2 wt%23 (Fig. 3(b)), and almost certainly reﬂect source heterogeneity.

FA

December 10, 2008 15:45

42

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

N. Hoang et al.

Fig. 3. Plots of SiO2 vs K2 O (a) with boundaries22 separating low-K, med-K, and high-K (a), and plots of SiO2 vs FeO*/MgO (b) with ﬁelds of tholeiite and calc-alkaline after Miyashiro.23 Representative southwest Japan subduction-related Daisen samples are from Tamura et al.5 Note that much of the Jinroku-yama basalt and Arita–Imari andesite are plotted in the calc-alkaline ﬁeld.

3.2. Trace element characteristics The combined variation of major and trace elements among the volcanic series, as exempliﬁed by the positive covariance of Rb and Th versus SiO2 over a wide range of contents, are consistent with fractional crystallization eﬀects, with or without crustal assimilation and/or mixing (AFC) processes (Table 1, Figs. 4(a) and 4(b)). However, plots of SiO2 against Ba (Fig. 4(e)),

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

Evidence for Coexisting OIB-Like and Subduction-Related Mantle Sources

43

Fig. 4. Plots of SiO2 (wt%) against Rb, Th, Sr, Nb, and Ba (ppm). Generally positive correlation between SiO2 and Rb, Th and negative covariance with Sr might reﬂect fractional crystallization relationship among the volcanic rocks. Very low contents of Ba in the Koshi-dake rhyolite and various contents of Nb, in contrast, suggest source heterogeneity. (See details in text.)

FA

December 10, 2008 15:45

44

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

N. Hoang et al.

as for lavas with low Ba contents such as the Koshi-dake rhyolite (220– 360 ppm), cf. with those of the Arita rhyolite (500–700 ppm), cannot be explained in terms of AFC eﬀects related to the accompanying basalt (350 ppm). Similarly, the negative correlation observed between SiO2 and Nb in the Koshi-dake basalts and rhyolites (Fig. 4(d)) also precludes an intracrustal genetic relationship. The contrasting Nb contents in these groups (<20 ppm cf. >40 ppm), for equivalent SiO2 contents, in the Kurodake basalts and andesites (Fig. 4(d)), appear to support this conclusion, especially given that fractionation of a Nb-rich phase is not expected during intravolcanic fractionation.29,30 The negative covariance from basalt to rhyolite observed in Fig. 4(d) may nonetheless indicate the possible involvement of Nb-depleted crustal material.6,31 Chondrite- and primitive mantle-normalized patterns of incompatible trace element (incl. REE) patterns for the basalts are typical of OIB (Figs. 5, 6).32,33 However, slight distinctions may be observed for each locality, especially in terms of high ﬁeld strength elements (HFSE) such as Nb, Ta, and Ti (Figs. 5(a), 6(a)). Signiﬁcant negative Eu anomalies observed in the Koshi-dake and some Arita rhyolites clearly reﬂect substantial plagioclase feldspar fractionation (Figs. 6(b), 7(b)). Compared to the Koshi-dake rhyolites, those from Arita show slightly more relative LREE enrichments and depletions in HREE and HREE. This observation again suggests that rhyolites from the two locations may tap distinct mantle source regions (Figs. 5(b), 6(b)). Rare earth element patterns in the Arita rhyolites, however, closely resemble the accompanying andesites (Figs. 5(c), 6(c)), suggesting that they may have genetic relationship although they are distinct from their accompanying basalts.

3.3. Sr–Nd–Pb isotopic compositions The basalt isotopic compositions, 87 Sr/86 Sr ranging 0.704–0.7048 and Nd/144 Nd ranging 0.5128–0.51268 (εNd = 2.5−0.9), are within the known range of Northern Kyushu Cenozoic intraplate basalts (Table 1, Fig. 7(a)).9,11 The rhyolites and andesites are the most enriched, with 87 Sr/86 Sr >0.7045–0.7052, corresponding to 143 Nd/144 Nd from 0.51269 to 0.51267 (εNd = 1−0.55) (Table 1). It is noteworthy that while the strontium isotopic ratios are high and variable in the rhyolites, their corresponding 143 Nd/144 Nd ratios vary within a much smaller range (0.5128–0.51267) (εNd = 2.75−0.7) (Table 1). Overall, the Sr and Nd isotopic compositions, 143

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

Evidence for Coexisting OIB-Like and Subduction-Related Mantle Sources

45

Fig. 5. Chondrite-normalized rare earth elements32 showing (a) diﬀerent patterns for basalts from diﬀerent locations, at least two chemically diﬀerent rhyolite types, (b) similarity in distribution conﬁguration between Arita andesite and rhyolite, (c) An Iwate andesite (NE Japan Quaternary island arc lava)16 is shown for reference.

FA

December 10, 2008 15:45

46

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

N. Hoang et al.

Fig. 6. Primitive mantle-normalized trace element33 showing chemically diﬀerent basalts (a), two rhyolite rock types with no genetic relation to basalt (b), and distribution pattern similarity between Arita rhyolite and andesite (c). Also shown is a 18 Ma tholeiite from Oki–Dogo island inﬂuenced by subduction-derived hydrous phases34 for reference. (See details in text.)

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

Evidence for Coexisting OIB-Like and Subduction-Related Mantle Sources

47

Fig. 7. Plots of 87 Sr/86 Sr vs εNd (a) and 207 Pb/204 Pb vs 206 Pb/204 Pb (b) showing samples trending between depleted (DM) and enriched mantle (EM2) (and CC: continental crust). Some basalts plot toward enriched mantel type 1 (EM1) ﬁeld similar to many Northern Kyushu Cenozoic intraplate basalts (crossed, data from Refs. 9, 11, and 17). DM, EM1, and EM2 are after Zindler and Hart.37 Symbols are as of Fig. 4.

including the most enriched rhyolite, are nonetheless within the range of Kyushu intraplate basalts.9,11,35 The Pb isotopic data (Fig. 7(b)) may be divided into two groups. Basalts from Arita, Koshi-dake, and a sample from Jinroku-yama, having low 206 Pb/204 Pb (<18.0) and 208 Pb/204 Pb (<38.45) plot in an EM1-rich ﬁeld (equivalent to the “depleted” quadrant in the Sr–Nd diagram) similar to the OIB-like northern Kyushu and Japan Sea basalts.11,36 The second

FA

December 10, 2008 15:45

48

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

N. Hoang et al.

group comprises andesites and rhyolites, and some basalts from Jinrokuyama and Kuro-dake with 206 Pb/204 Pb >18.2 and 208 Pb/204 Pb >38.5, plotting toward EM2-rich ﬁeld. These discrete Pb isotopic ﬁelds probably, at least in part, reﬂect distinct source regions.

4. Discussion 4.1. Origin of the isotopic enrichment Crustal contamination — whether by interaction between mantle-derived basaltic melt with wallrock via assimilation–fractional crystallization (AFC) processes or of their source regions, e.g., interaction with subductionderived ﬂuids or melts — is a potential factor contributing to negative HFSE anomalies (Fig. 6) and overall high 87 Sr/86 Sr ratios in the Arita– Imari magmas (Table 1, Fig. 7(a)). However, the isotopic compositions, including those of the most enriched andesites and rhyolites, are within the known limit of northern Kyushu Cenozoic intraplate, mantle-derived basalts.7,11,38,39 For example, a Kuro-dake basalt (Table 1: NAG02026), showing typical OIB-like trace element patterns (Figs. 5 and 6), has an even higher 87 Sr/86 Sr ratio (0.70487) than many other andesites and rhyolites (Table 1, Fig. 7(a)). On the other hand, three basaltic samples from the Jinroku-yama area, which are among the most Sr–Nd isotopically depleted (Table 1, Fig. 7(a)), show relatively high 206 Pb/204 Pb and 208 Pb/204 Pb and low HFSE as compared to the basalts (Figs. 6(a) and 7(b)). Besides indicating the possible assimilation of HFSE-depleted crustal of material at shallow levels, their geochemical and isotopic signatures may also reﬂect source mantle characteristics.

4.2. Eﬀects of assimilation–fractional crystallization While some of the intermediate to acidic lavas may be derivatives of parental maﬁc magmas via one or several intracrustal fractionation processes, the mantle-derived basalts described above are unlikely to be parental to the calcalkaline andesites, dacites, or rhyolites, given the signiﬁcant diﬀerences in their incompatible trace element (Figs. 4–6) and isotopic compositions (Figs. 7(a) and 7(b)). While crustal-like involvement in the magmatic evolution is clearly consistent with a strong positive covariance between SiO2 and 87 Sr/86 Sr (Fig. 8),2,40,41 the discrepancy in abundance levels of Ba in the Koshi-dake

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

Evidence for Coexisting OIB-Like and Subduction-Related Mantle Sources

49

Fig. 8. Plots of SiO2 vs 87 Sr/86 Sr shown with hypothetical binary mixing between a depleted source (87 Sr/86 Sr = 0.703, Sr = 40 ppm, SiO2 = 47wt%) and a SW Japan basement sediment (87 Sr/86 Sr = 0.7144, Sr = 85 ppm, SiO2 = 70wt%)35 to explain the chemical evolution of the volcanic series. The results, however, do not match with those of trace elements. (See details in text.)

basalt (380 ppm) and rhyolite (as low as 40 ppm) or Nb in the Kuro-dake basalt and andesite (referred to above) (Fig. 4) appear to preclude AFC processes as a dominant factor in explaining the compositional variation in the Arita–Imari volcanic rocks. Given that experimental data indicate that dacites and rhyolites can be generated by partial melting of, or equilibration with, hydrated basalt and/or andesite at pressures between ca. 7 and 2 kbar,42,43 they are clearly potential products of a combination of fractional crystallization, magma mixing, and assimilation and/or partial melting of lower or upper crust.3,43−47 Accordingly, Tamura and Tatsumi4 have proposed that dacites and rhyolites are most likely to have formed by partial melting of basalt and andesite in subduction-related settings. However, it is diﬃcult to distinguish the geochemical and (particularly) isotopic systematics associated with partial melting models for rhyolite genesis from those generated simply by fractional crystallization given that both processes are likely to produce similar eﬀects.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

50

9in x 6in

b672-V13-ch04

N. Hoang et al.

The FARM process (fractional crystallization, assimilation, replenishment, and mixing) proposed by Baker et al.44 was applied to explain rhyolite formation in northern California43 and Unzen volcano, NW Kyushu.9 The results of a simple mixing calculation between compositions of a relatively depleted mantle source (87 Sr/86 Sr = 0.703, Sr = 40 ppm, SiO2 = 47wt%) and an enriched (crust-derived) granitic sediment from northern Kyushu (87 Sr/86 Sr = 0.714, Sr = 85 ppm, SiO2 = 71wt%)5,35 showed that mixing of 7%–10% sediment could explain the complete Sr isotopic range of andesite, dacite, and rhyolite from the Arita–Imari area (Fig. 8). However, the trace element mixing between the most primitive Arita basalt (NAG02001) and the northern Kyushu sediment, respectively showing 30 and 10 ppm Nb, and 10 ppm Th, 360 and 500 ppm Ba is not consistent with the isotopic mixing model. For example, adding 10% sediment with the basalt would produce Nb/Th ratios of 7.5–10 and Ba/Nb ratios of about 12.5 cf. 1.5–3.5 and 15–38 for Nb/Th and Ba/Nb, respectively (Table 1). Because each of these elements has closely similar melt-solid distribution coeﬃcients48,49 (esp. Nb and Th) whose elemental ratios are essentially unaﬀected by fractional crystallization, magmatic AFC processes alone are not responsible for the variation in magmatic isotopic variation in this area.

4.3. Source heterogeneity The combination of relatively high 87 Sr/86 Sr isotope ratios and negative mantle-normalized Nb and Ta anomalies (Figs. 6–8) has been explained by several diverse mechanisms, including: (1) partial melting of or assimilation by lithospheric mantle and/or crust,2,40,42−44 (2) contamination of the magmatic source by subducting slab-derived hydrous ﬂuids and/or melts,22,34,50 and (3) the presence of relict HFSE-rich phases during magma genesis and/or slab-derived ﬂuid segregation.51−53 As a basis for evaluating potential geodynamic factors aﬀecting mantle variation beneath the Arita–Imari area, the role of the underlying Paciﬁc and Philippine Sea (PPS) — both subduting beneath the Japanese archipelago — plates should be evaluated. The volcanic front of the PPS is delineated by Aso Volcano28 in south central Kyushu (Fig. 1(a)) and Mts. Daisen and Sambe in southwest Honshu.50,54 The initial rifting phase of the Japan Sea (ca. 25–17 Ma; Ref. 55) was accompanied by low alkaline tholeiite (LAT) volcanism activity (e.g.) in the Oki–Dogo island area, Shimane region, in the Japan Sea, characterized by unusually high LILE and relatively low HFSE contents.34 These features

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

Evidence for Coexisting OIB-Like and Subduction-Related Mantle Sources

51

have been interpreted to present partial melts of shallow lithospheric mantle characterized by longstanding enrichment34 in hydrous, subduction-related, metasomatic minerals.56 Middle Miocene (ca. 11 Ma) LAT and evolved calcalkaline (andesite and rhyolite) eruptive products occur in the nearby Matsue Formation showing transitional arc — intraplate trace element and isotopic characteristics, viewed as products of depleted mantle, modiﬁed by slab-derived ﬂuid,51 because the leading edge of the PPS plate was not beneath the area during the Miocene50,54,57,58 attributed subductionrelated volcanism in the Shimane area to Paciﬁc plate subduction. Because the late Miocene Arita–Imari activity, located more than 100 km to the west of the volcanic front, is not directly related to PPS subduction,57,60 it is unlikely that the OIB-like basalts are parental to the andesite–dacite–rhyolite (Figs. 4–6), the latter being more plausibly ascribed to a distinct subduction-related (calcalkaline) tholeiitic parent.22,60 This implies the likelihood of coexisting OIB-like and subduction-related mantle sources beneath the Arita–Imari, analogous to that reported elsewhere.61−63 The presence of primitive OIB-like mantle, albeit a subsidiary component, beneath such volcanic arcs as those in Northeast Japan Arc64,65 and the Aleutians66 suggests coexisting with MORB-like mantle modiﬁed by subduction-derived metasomatic contaminants.67,68 Thus, in summary, calcalkaline volcanism in the Arita–Imari area, and elsewhere is SW Japan, is considered to result from the melting of the subcontinental lithosphere mantle aﬀected by Paciﬁc slab-derived contamination since before the Japanese archipelago separated from Eurasia.34,50,54,69

5. Summary Late Neogene magmatism in the Arita–Imari area, northern Kyushu, SW Japan comprising basalt–andesite–dacite and rhyolite, accompanied lithospheric extension during plate kinematic readjustments following the opening of the Japan Sea. The basalts show OIB-like geochemical aﬃnity, their isotopic compositions suggesting a range EM1-like asthenospheric sources and with superimposed eﬀects of EM2-like lithospheric mantle or crustal components; the relatively high LILE/HFSE ratios in the latter, suggest either the presence of crustal material in lithospheric source regions or its assimilation during magmatic ascent and emplacement. Both of the EM isotopic endmembers show a strong spatial dependence in the basalts, suggesting

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

52

9in x 6in

b672-V13-ch04

N. Hoang et al.

regional source heterogeneity in the lithospheric mantle in northern Kyushu to be a signiﬁcant factor.11,17 The (calcalkaline) andesite and dacite–rhyolite series show a range of similar incompatible element distributions, distinct from OIB-like patterns, but consistent with diﬀerentiation from maﬁc parent melts. It is unlikely that these magmas are products of the OIB-like basalts and more likely derived from a distinct calcalkaline tholeiitic magma. Derivative calcalkaline magmas probably evolved via a process combining the eﬀects of fractional crystallization, assimilation, replenishment, and mixing (FARM).44 The composition of the assimilant involved is consistent with the regional crustal granitic sediments whose geochemical character contrasts strongly with wallrock of transitional crustal magma chambers. The chemical and isotopic diversiﬁcation thus depends largely on the extent of wallrock melting and interaction with mantle-derived maﬁc magmas and their diﬀerentiation products. The occurrence of OIB-type basalts and calcalkaline-type andesites suggests that the two mantle sources may coexist beneath the Arita– Imari area and elsewhere in Southwest Japan — the ﬁrst from fertile asthenospheric reservoirs within the mantle wedge, the second generated from the mantle metasomatized by (Paciﬁc plate) subduction-dating before the separation of the Japanese archipelago from Eurasia.

Acknowledgments The manuscript was beneﬁted from intensive discussions with Isoji Miyagi (GSJ) and Martin Flower (UIC). Osamu Ishizuka and Naoko Honda are gratefully acknowledged for assistance in ICP-MS analysis. The authors thank an anonymous reviewer for helpful reviews. Kenji Satake is thanked for editorial handling of the manuscript. This research project has been conducted under the research contract with the Japan Nuclear Energy Safety Organization (JNES).

References 1. T. Takagi, Y. Orihashi, K. Naito and Y. Watanabe, Chem. Geol. 160 (1999) 425–445. 2. D. J. DePaolo, Earth Planet. Sci. Lett. 53 (1981) 189–202. 3. D. F. Parker, A. Ghosh, C. W. Price, B. D. Rinard, R. L. Cullers and M. Ren, J. Volcanol. Geotherm. Res. 139 (2005) 185–210.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

Evidence for Coexisting OIB-Like and Subduction-Related Mantle Sources

53

4. Y. Tamura and Y. Tatsumi, J. Petrol. 43 (2002) 1029–1047. 5. Y. Tamura, M. Yuhara, T. Ishii, N. Irino and H. Shukuno, J. Petrol. 44 (2003) 2243–2260. 6. S. R. Taylor and S. M. McLennan, Phil. Trans. Royal Soc. London 301 (1981) 381–399. 7. H. Kurasawa, Geological Survey of Japan, Report 217, Kawasaki, Japan, (1967). 8. K. Uto and Y. Tatsumi, The Island Arc 5 (1996) 250–261. 9. N. Hoang and K. Uto, Chem. Geol. 198 (2003) 249–268. 10. H. Mashima, J. Volcanol. Geotherm. Res. 131 (2004) 333–349. 11. K. Uto, N. Hoang and K. Matsui, Tectonophysics 393 (2004) 281–299. 12. T. Seno, The Island Arc 8 (1999) 66–79. 13. S. Imai, K. Sawamura and T. Yoshida, Geology of the Imari district, Geological Sheet Map of 1:50,000, Geological Survey of Japan (1958) (in Japanese with English abstract). 14. T. Yanagi and S. Maeda, Phys. Earth Planet. Interiors 107 (1998) 203–219. 15. K. Uto, O. Ishizuka, A. Matsumoto, H. Kamioka and S. Togashi, Bull. Geol. Surv. Japan 48 (1997) 23–46. 16. J. Itoh, K. Uto, A. Matsumoto, M. Sudo and N. Hoang, Japan Earth Planet. Sci. Meeting g017 (2004) 004. 17. N. Hoang and K. Uto, Earth Planet. Sci. Lett. 249 (2006) 229–240. 18. O. Ishizuka, R. N. Taylor, J. A. Milton and R. W. Nesbitt, Earth Planet. Sci. Lett. 211 (2003) 221–236. 19. C. Deniel and C. Pin, Analytica Chimica Acta 426 (2001) 95–103. 20. C. Pin and J. F. Santos Zalduegui, Anal. Chimica Acta 339 (1997) 79–89. 21. N. Hoang, K. Uto, H. Hoshizumi and K. Oguri, IUGG Meeting A-541 (2003) Sapporo. 22. J. B. Gill, Orogenic Andesite and Plate Tectonics (Springer-Verlag, New York, 1981). 23. A. Miyashiro, Am. J. Sci. 274 (1974) 321–335. 24. S. Turner and C. Hawkesworth, Chem. Geol. 120 (1995) 295–314. 25. K. Hirose and I. Kushiro, Earth Planet. Sci. Lett. 114 (1993) 477–489. 26. I. Kushiro, in Earth Processes: Reading the Isotopic Code, eds. A. Basu and S. R. Hart, Vol. 95 (American Geophysical Union, 1996), pp. 109–122. 27. R. J. Arculus, J. Petrol. 44 (2003) 929–935. 28. T. Nakada and H. Kamata, Bull. Volcanol. 53 (1991) 182–194. 29. D. Ionov and A. W. Hofmann, Earth Planet. Sci. Lett. 131 (1995) 341–356. 30. B. S. Kamber, A. Greig, R. Schoenberg and K. D. Collerson, PreCambriar Res. 126 (2003) 289–308. 31. D. Ben Othman, W. M. White and J. Patchett, Earth Planet. Sci. Lett. 94 (1989) 1–21. 32. N. Nakamura, Geochim. Cosmochim. Acta 38 (1974) 757–775. 33. W. F. McDonough, S.-S Sun, A. E. Ringwood, E. Jagoutz and A. W. Hofmann, Geochim. Cosmochim. Acta 56 (1992) 1001–1012. 34. K. Uto, E. Takahashi, E. Nakamura and I. Kaneoka, Geochem. Journal 28 (1994) 431–449.

FA

December 10, 2008 15:45

54

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

N. Hoang et al.

35. T. Hosono, T. Nakano and H. Murakami, Chem. Geol. 201 (2003) 19–36. 36. M. Tatsumoto and Y. Nakamura, Geochim. Cosmochim. Acta 55 (1991) 3697–3708. 37. A. Zindler and S. Hart, Ann. Rev. Earth Planet. Sci. 14 (1986) 493–571. 38. S. Nohda, Y. Tatsumi, Y. Otofuji, T. Matsuda and K. Ishizaka, Chem. Geol. 68 (1988) 317–327. 39. E. Nakamura, M. T. McCulloch and I. H. Campbell, Tectonophysics 174 (1990) 207–233. 40. I. J. Graham and W. R. Hackett, J. Petrol. 28(3) (1987) 531–567. 41. R. W. Carlson and W. K. Hart, in Continental Flood Basalts, ed. J. D. Macdougall (Kluwer Academic Publication, 1988). 42. J. S. Beard and G. E. Lofgren, J. Petrol. 32 (1991) 365–401. 43. T. L. Grove, J. M. Donnelly-Nolan and T. Housh, Contrib. Min. Petrol. 127 (1997) 205–223. 44. M. B. Baker, T. L. Grove, R. J. Kinzler, J. M. Donnelly-Nolan and G. A. Wandless, J. Geophys. Res. 96 (1991) 1819–1842. 45. S. P. Verma, Lithos 56 (2001) 141–164. 46. J. Chesley, J. Ruiz, K. Righter, L. Ferrari and A. Gomez-Tuena, Earth Planet. Sci. Lett. 195 (2002) 211–221. 47. M. Mortazavi and R. S. J. Sparks, Contrib. Min. Petrol. 146 (2004) 397–413. 48. A. W. Hofmann, Earth Planet. Sci. Lett. 90 (1988) 297–314. 49. D. McKenzie and R. K. O’Nions, J. Petrol. 32 (1991) 1021–1091. 50. P. A. Morris and H. Kagami, Earth Planet. Sci. Lett. 92 (1989) 335–346. 51. J. D. Woodhead, S. M. Eggins and J.A. Gamble, Earth Planet. Sci. Lett. 114 (1993) 491–504. 52. P. B. Kelemen, N. Shimizu and T. Dunn, Earth Planet. Sci. Lett. 120 (1993) 111–134. 53. J. M. Brenan, H. F. Shaw, D. L. Phinney and F. J. Ryerson, Earth Planet. Sci. Lett. 128 (1994) 327–339. 54. P. A. Morris, Geology 23 (1995) 395–398. 55. K. Tamaki, K. Suyehiro, J. Allan, J. C. Ingle Jr and K. A. Pisciotto, Proc. Ocean Drilling Program, Scientific Results, eds. K. Tamaki et al. (Ocean Drilling Program, College Station, Texas, 1992), pp. 1333–1348. 56. Y. Tatsumi, J. Geophys. Res. 94 (1989) 4697–4707. 57. I. Nakanishi, Tectonophysics 69 (1980) 1–35. 58. T. Seno and S. Maruyama, Tectonophysics 102 (1984) 53–84. 59. J.-I. Kimura, R. J. Stern and T. Yoshida, Geol. Soc. Amer. Bull. 117 (2005) 969–986. 60. I. Kushiro, J. Geophys. Res. 95 (1990) 15929–15939. 61. C. Chavel, A. W. Hofmann and P. Vidal, Earth Planet. Sci. Lett. 110 (1992) 99–119. 62. R. W. Carlson and A. J. Irving, Earth Planet. Sci. Lett. 126 (1994) 457–472. 63. A. W. Hofmann, Nature 385 (1997) 219–229. 64. S. Togashi, T. Tanaka, H. Yoshida, K. Ishikawa, A. Fujinawa and H. Kurasawa, Geochem. J. 26 (1992) 261–277.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch04

Evidence for Coexisting OIB-Like and Subduction-Related Mantle Sources

55

65. T. Sano, T. Hasenaka, A. Shimaoka, C. Yonezawa and T. Fukuoka, Earth Planet. Sci. Lett. 186 (2001) 187–198. 66. J. D. Morris and S. R. Hart, Geochim. Cosmochim. Acta 47 (1983) 2015–2030. 67. M. J. Hole, A. D. Saunders, G. Rogers and M. A. Sykes, Geol. Soc. Spec. Publ. 81 (1995) 265–285. 68. M. Gorring, B. Singer, J. Gowers and S. M. Kay, Chem. Geol. 193 (2003) 215–235. 69. Y. Miyake, Geochem. J. 28 (1994) 451–472.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

This page intentionally left blank

b672-V13-ch04

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch05

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

LANDSLIDE HAZARD ZONING OF THE MURAVERA HILLSIDE (SARDINIA, ITALY) GIULIO BARBIERI∗ and PAOLO CAMBULI Department of Territorial Engineering, University of Cagliari, Piazza d’Armi Cagliari, 09123, Italy ∗ [email protected]

Here a method for zoning landslide hazard is described adapting at slope scale the method proposed by the Italian Geological Service for creating the 1:50.000 in scale geological risk map related to slope instability. The approach consists of a preliminary zoning for roughly dividing the study area up into a discrete number of smaller areas, suﬃciently homogeneous as to their natural susceptibility to gravitational instability. This zoning takes into account lithological features, slope and plant coverage, all factors that have a direct inﬂuence on slope stability conditions, and are in the meantime easy to be measured and to be mapped. The landslide hazard zoning is obtained by assigning a numerical weight to each factor, in relation to the degree of instability determined therefrom and then summing for each area all the numerical weights produced by the factors examined. This procedure has been performed in GIS environment as it provides numerous calculation procedures for converting the vector data for the basic mapping information into raster data and their subsequent representation in ﬁnite elements using a numerical model. The proposed method has been applied to the built-up hillside of Muravera, in southeastern Sardinia, at the mouth of Flumendosa River.

1. Introduction The interventions planning aimed at mitigating the landslide risk in a certain area demands a preliminary evaluation of the landslide hazard, that, once all the elements potentially at risk have been identiﬁed and their vulnerability evaluated, allows to deﬁne the level of risk of the areas located in the territory and consequently to establish a scale of priority between the stabilization works relying on the diﬀerent levels of risk that the areas present. The evaluation of the landslide hazard, intended with Varnes1 as the probability that a certain landslide phenomenon may happen in a certain area and in a certain timecourse, is recognized as a very complex process, because the conditions of the slope stability depend on several factors as 57

FA

December 10, 2008 15:45

58

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch05

G. Barbieri and P. Cambuli

lithological, morphological, hydrological, geotechnical, hydrogeological, and those concerning the land use. Though, the need for the hazard evaluations on regional scale make the adoption of geotechnical quantitative impractical methods, because a physical and mechanical characterization of all the rocks involved in the stability evaluation of wide areas is impossible. Therefore, qualitative or semi-quantitative methods are usually employed, such as the Overlay Mapping,2 that produces an evaluation of the landslide hazard through a weighted overlapping of all the diﬀerent factors that aﬀect the slopes stability. In order to be feasible, this method takes into consideration those factors of landslide susceptibility that are already available and mapped at the regional scale such as lithology outcrops, hillside slope, and land use. While conferring variable numerical weights according to their real inﬂuence on the slope stability condition to the latter parameters (hillside slope and land use) is quite easy, the assignment to the lithological factor of a weight that truly corresponds to the real stability conditions is indeed arduous and problematic, because such conditions are not exclusively bound to the lithological characteristics of the rock formations (granite, basalt, clay, etc.), but depend on the physical and mechanical conditions of the formation (degree of weathering, joints frequency, compactness). In this work, a practical method for the correct evaluation of the lithological factor weights is described, exemplifying this process through the landslide hazard zoning of the county territory of Muravera, in the southeastern part of Sardinia (Italy).

2. Geological and Geomorphological Setting of the Muravera Hillside The built-up area of Muravera extends on the right bank of the Flumendosa river, a few kilometers from the Mediterranean sea, at the foot of a large hillside of Paleozoic metamorphites, that dominate and impend over it. The most ancient formation, the San Vito Sandstone, is composed of a thick, weakly metamorphosed sequence of sandstone, quartzites, and shales. These are overlain by metavolcanites, known as the Serrabus Grey and White Porphyrites. The volcanic formations are followed by metasandstones and metaconglomerates, the Punta Serpedd`ı Formation, by siliciﬁed limestones, the Tuviois Formation and last by metasediments of Silurian–Devonian and likely Carboniferous age, the Serra S’Illixi Group. The foothills are composed entirely of Quaternary clastic sediments, like talus and ancient and recent alluvium.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Landslide Hazard Zoning of the Muravera Hillside

b672-V13-ch05

59

The poor plant coverages and the anthropic interventions on the area make it exposed to possible phenomena of debris ﬂows, that may be activated by particularly rough meteorological events. Phenomena of falls and topples by detachment and rolling of etherometric blocks from the rocky subvertical slopes that rise in the upper part of the site are also possible. 3. Landslide Hazard Assessment Conforming to the speciﬁcations included in the Guidelines for mapping hydraulic and geomorphologic risk in the Sardinian Region,3 the analysis of landslide hazard has been carried out by evaluating and pondering, in every site potentially exposed to risk, the inﬂuence that the diﬀerent factors of landslide susceptibility (geological and technical factors, morphological, vegetational) have on the stability conditions of the slopes. Thus, for every factor of landslide susceptibility, a speciﬁc thematic map has been drawn: the landslide susceptibility factors have been indexed through the assignation of numerical weights proportional to the supposed incidence of the single factor on the possible development of landslide phenomena. Careful consideration of the single factors has been made by adopting the same range of weights provided for every factor in the abovementioned Guidelines. The map of landslide hazard has been obtained through the overlapping and the sum of all the weights assigned to all the landslide factors considered. The evaluation of the hazard has been at ﬁrst conducted on the basis of three fundamental factors of the landslide susceptibility: • the hillside slope, • the land use, • the lithology of outcropping formations.

3.1. Basic cartography In order to be able to analyze and weigh these factors, it has been necessary to survey and to adapt to the scale of detail chosen, 1:2000, the following basic cartography: • Map of the hillside slope, built dividing the area into 25×25 m meshes and employing the numerical 3D map on 1:2000 scale of the county territory of Muravera.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch05

G. Barbieri and P. Cambuli

60

• Map of the land use, built upon the Map of the Land use of the Sardinia Region, 2003 Edition, on 1:25000 scale, veriﬁed and corrected through direct surveys in the countryside. • Geological map, drawn according to the Geological Map F.549, Muravera Section II, on 1:25000 scale, of the Sardinia Region, veriﬁed and corrected through direct surveys in the ﬁeld (Fig. 1).

Fig. 1.

Geological map of the Muravera hillside.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch05

Landslide Hazard Zoning of the Muravera Hillside

61

3.2. Weighted cartography Using the basic cartography, the weighted maps of the landslide susceptibility factors have been built. Weighted map of the hillside slope, takes in account the intensity of the gravitational destabilizing component. The weighted map of the hillside slope has been built from the corresponding basic map, using the same weights speciﬁed in the aforementioned Guidelines (Table 1). Weighted map of the land use, taking into consideration an active, external, and time-variant parameter, gives a measure of the impact of the plant cover and of the anthropic activity on the conditions of hillside stability. The weighted map has been derived from the correspondent basic map, using the same weights and land use classes speciﬁed in the aforementioned Guidelines (Table 2). Weighted map of lithology, aims to represent, through the peculiar characteristics of the lithological formations, the extent of the resistant forces that are opposed to the potential landslide phenomena. Lithology is a complex factor, mainly evaluated from a geotechnical point of view, that Table 1. Slope classes α

Slope classes and relevant weights.

0%–10%

11%–20%

21%–35%

36%–50%

>50%

+2

+1

0

−1

−2

Weights

Table 2.

Land use classes and relevant weights.

Land use classes Residential areas, entertainment, and sport areas Commercial and industrial/handicraft establishments Build sites, sowable lands Artiﬁcial grasslands Orchards Woods Natural grazing lands Mediterranean scrub Canals and waterways Bared rocks

Weights 0 0 −1 −2 0 +2 0 +1 −2 −2

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch05

G. Barbieri and P. Cambuli

62

takes into account all the geological, structural, physical, and mechanical characteristics, in relation to the potential landslide phenomenon. In order to weigh correctly the landslide susceptibility of the rock formations described in the Geological Map, the following parameters, that altogether determine the geomechanic behavior of every single lithology, have been evaluated: for rock formations: • • • • •

weathering degree depth of the weathering spacing of joints attitude of joints compressive strength of the sound rock

for loose formations: • • • •

degree of cementation thickness of the scree deposit compactness permeability

These parameters have been carefully evaluated through direct on-site surveys and measurements taken in each point of a dense survey network, constituted by 117 stations distributed almost evenly on all the studied area (Fig. 2), in order to determine the weights of the single parameters and consequently, through the sum of the weights, the global weight of the lithological factor. The assessment of the parameters and the division into classes have been made following the speciﬁcations included in the Guidelines of the Italian Geological Service for carrying out a geologic hazard cartography.4 All the parameters considered have been divided into four classes, with an impedance degree to the landslide phenomenon decreasing from class I, the most stable, to class IV, the most unstable (Table 3). The total score given to the lithological factor in every survey station, calculated as the sum of the weights given to the single parameters taken for examination, varies between a minimum of 4 and a maximum of 20 for the rock formations and between a minimum of 4 and a maximum of 16 for the loose formations, and has been normalized in the range from 1 (minimum) to 9 (maximum) as the Guidelines for mapping hydraulic and geomorphologic risk in the Sardinian Region speciﬁc for the lithological factor.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Landslide Hazard Zoning of the Muravera Hillside

b672-V13-ch05

63

Weighted lithology map with network of survey stations. Fig. 2.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch05

G. Barbieri and P. Cambuli

64 Table 3.

Geomechanical parameter classes and relevant weights.

Geomechanical parameters Weathering degree Weathering deepness, z(m) Joints spacing, s(cm) Joints attitude Compressive strength, σ(kg/cm2 ) Cementation degree Scree thickness, d(m) Compactness degree Hydraulic conductivity, k(m/s) Weights

Class I

Class II

Class III

Class IV

None z<1

Moderate 1
High 3
Like soil z>5

s > 100 Horizontal σ > 40

30 < s < 100 Vertical 20 < σ < 40

5 < s < 30 Deep slope > α 10 < σ < 20

s<5 Deep slope < α σ < 10

High d<1 Very compact k > 10−3

Moderate 1
Light 3
None d>5 Loose k < 10−9

4

3

2

1

The indexed map of lithology has been realized for all the areas under study, distributing with geometrical criterion the exact values calculated in every survey station until the middle point between two contiguous stations and avoiding at the same time to cross the border of another lithology (Fig. 2).

3.3. Map of the landslide hazard by overlapping The overlapping of the weighted maps of hillside slopes (range −2 ÷ 2), land use (range −2 ÷ 2), and lithology (range 1 ÷ 9) produces at last a cartography of ﬁrst synthesis of the landslide hazard in the considered area (range −3 ÷ 13) in which every elementary area of the territory under study is characterized by a numerical index of hazard, obtained by arithmetical sum of the factor weights, that translates in a numerical value the inﬂuence that these three factor have on the conditions of the hillside stability. This numerical index range has been divided into four classes of even extent, with diﬀerent degrees of hazards, according to the scheme shown in Table 4. Table 4. Landslide hazard classes Intensity Weights h

Lanslide hazard classes.

Hg1 Moderate 9 < h < 13

Hg2 Medium 5
Hg3 High 1
Hg4 Very high −3 < h < 1

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Landslide Hazard Zoning of the Muravera Hillside

b672-V13-ch05

65

Landslide hazard map by overlay of lithology, slope, and land use. Fig. 3.

FA

December 10, 2008 15:45

66

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch05

G. Barbieri and P. Cambuli

The inﬂuence of the lithological and land use factors may in some cases lead to classify ﬂat territories as medium- or high-hazard areas, while these could be easily accounted as zones with low possibilities of landslide phenomena; so all the territories that present slopes lower than 20% have been automatically classiﬁed as class Hg1, without keeping in account their values in the lithology and land use (Fig. 3).

3.4. Map of the landslide hazard due to falls and topples The overlay mapping method, given that it performs an evaluation of the hazard on the orthogonal projection of the territory map, does not allow to evaluate the landslide hazard of subvertical rocky walls, typically aﬀected by potential phenomena of falls and/or topples. To keep these factors also into account, the class of landslide hazard due to falls and topples has been evaluated by making in the subvertical rocky sites the same kind of surveys made for the evaluation of the lithological factor. Once the areas potentially aﬀected by rock falls or topples had been identiﬁed, the extension of the area subject to landslides has been estimated by evaluating the possible paths and the limits of the landslides invasion, according to the typological characteristics of the gravitational movement and to the morphological and vegetational characteristics of the site under the rock face. In this way, for all the areas considered, a map of the landslide hazard caused by falls and topples has been drawn, divided into the same four classes of hazard (Fig. 4).

3.5. Map of the landslide hazard due to debris flows The hazard evaluation of debris ﬂows has been carried out through the following phases: • surveying and mapping of spatial distribution of scree deposits (Fig. 1); • assessment, for every scree deposit, of the overcoming of the stability limit of Takahashi5 for given return periods: 50, 100, 200, and 500 years; • evaluation of the extension and propagation of debris ﬂows, following lines of maximal slope, until the conﬂuence in the channel; • evaluation of the progressive attenuation of the debris ﬂow intensity, due to the reduction of the kinetic energy, caused by friction and by reduction of the hillside slope, with a consequent progressive granulometric segregation of the solid matrix.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Landslide Hazard Zoning of the Muravera Hillside

b672-V13-ch05

67

Landslide hazard map due to falls and topples. Fig. 4.

FA

December 10, 2008 15:45

68

AOGS 2007 - SE Volume

9in x 6in

G. Barbieri and P. Cambuli

b672-V13-ch05

Landslide hazard map due to debris ﬂows. Fig. 5.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Landslide Hazard Zoning of the Muravera Hillside

b672-V13-ch05

69

Final landslide hazard map of Muravera hillside. Fig. 6.

FA

December 10, 2008 15:45

70

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch05

G. Barbieri and P. Cambuli

In this way, a map of the landslide hazard caused by debris ﬂows has been drawn, divided into hazard classes according to the return period (Fig. 5).

3.6. Final map of landslide hazard The ﬁnal map of landslide hazard has been obtained by overlapping the weighted map of lithology, hillside slope and land use, the map of the landslide hazard due to falls and topples, and the map of the landslide hazard due to debris ﬂows (Fig. 6).

4. Conclusions In this study, the assessment of the landslide hazard has been mainly focused on the eﬀort of evaluating the lithological factor with a weight value correctly proportional to the actual impact that the geomechanical behavior of the lithological formations has on the condition of slope stability. Therefore, the main physical and mechanical parameters (deepness and degree of the weathering, compactness, spacing and attitude of the joints, compressive strength, permeability) of the lithological formations have been evaluated through direct on-site surveys and measurements taken in each point of a dense survey network, and each parameter has been attributed a numerical weight proportional to the relative impact on the landslide hazard. The distinctive weight of every lithological formation has then been determined through the algebraic sum of the weight attributed to every single factor. The method proposed in this study demands only on-site investigations and semi-quantitative surveys in the countryside, and consequently it allows evaluations of hazards at regional scale, or for wide areas, in circumstances where the adoption of quantitative and purely geotechnical methods are impractical, because of the impossibility to proceed with the physical and mechanical characterization of all the rock formations involved in the evaluation of the stability of wide areas.

References 1. D. J. Varnes and IAEG Commission on Landslides, Landslide Hazard Zonation — A Review of Principles and Practice (UNESCO, Paris, 1984). 2. P. Aleotti and R. Chowdhury, Landslide hazard assessment: Summary review and perspectives, Bull. Eng. Geol. Environ. 58(1) (1999) 21–44.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Landslide Hazard Zoning of the Muravera Hillside

b672-V13-ch05

71

3. D. Dovera, M. Mancini and M. Salis, Attivit` a di individuazione e perimetrazione delle aree a rischio idraulico e geomorfologico e relative misure di salvaguardia. Linee guida (Regione Autonoma della Sardegna, 2000). 4. M. Amanti, A. Carrara, G. Castaldo, P. Cosimo, G. Gisotti, M. Govi, G. Marchionna, R. Nardi, M. Panizza, M. Pecci and G. Vinello, Linee guida per la realizzazione di una cartografia della pericolosit` a geologica connessa ai fenomeni di instabilit` a dei versanti alla scala 1:50.000 (Presidenza del Consiglio dei Ministri–Servizio Geologico d’Italia, 1992). 5. T. Takahashi, Debris flow, IAHR Monograph (Balkema, Rotterdam, 1991).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

This page intentionally left blank

b672-V13-ch05

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch06

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

AN INTEGRATIVE GEOSCIENCE DATA MODEL BY LINKING FIELD-SPECIFIC DATA MODELS IN DIGITAL GEOLOGIC MAP, EARTH RESOURCE, AND GEO-HAZARD LUCAS DONNY SETIJADJI∗,†,‡ and KOICHIRO WATANABE∗ ∗Department of Earth Resources Engineering Kyushu University, 744 Motooka, Nishi-ku Fukuoka, 819-0395, Japan †Department of Geological Engineering Gadjah Mada University, Grafika 2, Bulaksumur Yogyakarta, 55281, Indonesia ‡ [email protected]

There are now several object-relational geoscience data models available, such as the digital geologic map and mineral exploration models. This study pursues an integrative model by integrating them and extending their contents with data from our research projects in Japan and SE Asia. Current result is the ﬁrst version of GeoSEA data model that can support diﬀerent kinds of geoscience projects in geological mapping, earth resources, and geo-hazards at a continental margin setting such as SE Asia.

1. Introduction The development of object-relational GIS in early 2000s1,2 makes possible the integration of GIS and object-relational databases into a customized GIS for speciﬁc applications, including geoscience. A properly designed database using object-relational technology enables modeling speciﬁc phenomena inside a database while it still supports an easy use of data analysis.1 Using this basic concept, several database model projects in geoscience have evolved, such as the North America Digital Geologic Map Data Model (NADM), the eXploration and Mining Markup Language (XMML), the Commission for the Management and Application of Geoscience Information (CGI) of the International Union of Geological Sciences (IUGS), and the ESRI Geology Data Model (EGDM). 73

FA

December 10, 2008 15:45

74

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch06

L. D. Setijadji and K. Watanabe

In this chapter we will describe our database modeling project for the purpose of studying geology of continental margin such as Japan and SE Asia. As the geology of continental margin is a complex interplay of diﬀerent geological processes, and as we are also interested in contributing to many research ﬁelds, our aim is to develop an integrative geosciences database model that can serve many geologic applications. The basic approach is by linking several existing database models, extending their current contents, and implementing results using a selected physical platform.

2. Geoscience Data Model Projects 2.1. Major geoscience data model projects Started almost simultaneously around year 2000, several international data model projects are now evolving in diﬀerent ﬁelds. Today, these initiatives might have been culminating at the IUGS-CGI data model project (https://www.seegrid.csiro.au/twiki/bin/view/CGIModel). The aim of CGI is to enable the global exchange of knowledge about geoscience information and systems. Its current product is the GeoSciML data model, that draws from many geoscience data model eﬀorts, and from these establishes a common suite of feature types based on geological criteria (units, structures, fossils) or artefacts of geological investigations (specimens, sections, measurements). A number of predecessor projects have a strong inﬂuence on the development of GeoSciML. These include activities by national geological surveys such as British, Japan, Canada, and the United States. The latest two countries have established the North American Data Model for geological maps project (NADM; http://geology.usgs.gov/dm/) that in the case of the United States has evolved into the National Geologic Map Database project (NGMDB; http://ngmdb.usgs.gov). This project focuses on developing standards on description, classiﬁcation, and interpretation of geological features in digital geologic-map databases. The conceptual design was conceptualized as the NADM-C1.0 data model.4 From the minerals industry sector, the XMML data model project (https://www.seegrid.csiro.au/twiki/bin/view/Xmml) works for mineral industry in order to support exchange of mineral exploration information. The ESRI’s EGDM project (http://support.esri.com/datamodels) focuses on implementation of the NADM model using the ESRI ArcGIS software. This project has produced the EGDM 812 data model,3 the ArcGIS implementation of the NGMDB Geologic Map Feature Class model,5 and

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch06

An Integrative Geoscience Data Model by Linking Field-Specific Data Models

75

the ArcGeology v.1.9 The Open Geospatial Consortium (OGC) has also been working on this initiative, and one important product is the OGC 05-087r4 on Observations and Measurements.8 All of these products, except for the XMML, are focusing on digital geological map. In other words, there are still many ﬁelds unﬁnished or even untouched. 2.2. About this project This project was initiated in 2003 when the senior author was in an internship program with ESRI at Redlands, California. With an initial support of ArcGIS software from ESRI, this data model project was started as part of the senior author’s doctoral thesis at Kyushu University, Japan, with the focus of modeling on island arc magmatism and earth resources.6,7 This initiative was then disseminated into the Asian University Network (AUN) program in the ﬁeld of geological engineering, and the scope was enlarged into other geoscience ﬁelds such as geo-hazards. Since the year 2006, this data model project is ﬁnancially supported by a Grant-in-Aid for Scientiﬁc Research from the Ministry of Education and Science, Japan. As the scale of this project is much smaller than other major projects mentioned above, at the core we merely take advantages from the development of other data model projects. Then, at parts that are not yet touched, we extend the model by self-modeling the data that are directly related with our research activities. Also, this project pursues an integrated model, in which one core geoscience data model can be used to serve diﬀerent ﬁeld-speciﬁc projects. This is done through linking the core geological map data model with other task-speciﬁc data models for earth resources and geo-hazards. 2.3. Case study Our research group is very active in conducting research studies on mineral resources in Japan and many Asian countries. However, for the case study on an integrative data model, we select the Yogyakarta region in central Java island, Indonesia. There are several reasons for this selection. First, this city is the center of geoscience education in Indonesia. Second is the existence of various geoscience research projects that we directly work here, such as the study on still very active Merapi volcano, the devastated earthquake 27 May 2006, the porphyry deposit prospect at Selogiri district, a small geothermal prospect at Parangtritis beach, and a major Dieng geothermal ﬁeld.

FA

December 10, 2008 15:45

76

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch06

L. D. Setijadji and K. Watanabe

3. Proposed Integrative Data Model 3.1. Modeling approach and platform Our model uses the conceptual model of NADM-C1.0 as the base for developing extended data models for earth resources and geo-hazards. The reason is that the NADM-C1.0 works on basic geological information, i.e. concepts related to information presented on digital geologic maps. Furthermore, the conceptual design of NADM-C1.0 has also been accepted by other data models. Following the database design concepts by Longley et al.2 and Zeiler,1 we came through three stages of database design: conceptual, logical, and physical models. In the conceptual model, all collected datasets are organized in order to deﬁne all object classes. We then deﬁned database objects and their relationships. A conceptual design was then drawn into a logical model using the computer-aided software engineering (CASE) tools using a uniﬁed modeling language (UML) diagram. The logical model was developed by matching the conceptual object types to speciﬁc platform used for the real implementation of databases. As we use the ESRI ArcGIS and Geodatabase platforms, several ArcGIS-proprietary standard classes were added for modeling spatial features and representation. Finally, physical database was implemented by populating the Geodatabase schema with real data. With directly doing research projects on geology, earth resources, and geo-hazards in Japan and SE Asian countries, we can now produce the ﬁrst version of integrative data model, i.e. the GeoSEA Data Model 1.0. This version is developed from our previous data model on earth resources6 through adaptation with new developments from other data model projects and from our projects on the Yogyakarta earthquake and Merapi eruption 2006.

3.2. GeoSEA data model structure The structural organization of data model is organized into packages, supergroups, groups, and classes. Packages are containers used to organize UML elements into speciﬁc projects. This is merely an organization issue as UML elements of diﬀerent packages can actually be shared with or moved into other packages. Within each package, UML elements are organized into supergroups, groups, and individual classes. In the current model, there are three packages created, i.e. the geologic map, earth resource, and geo-hazard

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch06

An Integrative Geoscience Data Model by Linking Field-Specific Data Models

Fig. 1.

77

Packages within the GeoSEA Data Model v.1.0.

models (Fig. 1). The geologic map package becomes a center of modeling as most object classes are modeled here. The earth resource and geo-hazard packages are used to organize UML elements which are very case-speciﬁc for respective purposes.

3.3. Detailed descriptions of object classes The contents of geologic map package as the center of data model are shown in Fig. 2. The top level object classes of geological map model consist of GeologicConcept, MetaData, and Feature classes that inherit from the ESRI Class: Object. The GeologicConcept is a supergroup class that compiles all geo-object names, which are then further grouped into several group classes. The EarthMaterial, GeologicProperty, GeologicStructure, GeologicUnit, GeologicProcess, GeologicRelation, and GeologicEvent group classes were adapted from the NADM-C1.0,4 while the EarthResource, GeologicSample, GeologicMeasurement, and GeologicVocabulary group classes were established on their own.6 The GeologicMeasurement has also been modeled at the OGC project together with the GeologicObservation.8 We insert the GeologicVocabulary group that compiles all geologic concept names using standard classiﬁcation, such as from the NADM Geological Map Unit Classiﬁcation Version 6.1 (http://geology.usgs.gov/dm/). Note that the EarthResource group class is part of the EarthMaterial group, while the Geo-hazardEvent is a member of the GeologicEvent group. The Feature is a spatial representation of data through a map and is a proprietary of the ESRI ArcGIS. At a more platform-free deﬁnition, the MappedFeature group class is often used in other projects (i.e. NGMDB, XMML, and CGI data models). We adapted the classiﬁcation of MappedFeature based on the NGMDB model,5 and add it with SamplingFeature and DrillHoleSegmentFeature. Especially for the drillhole

FA

December 10, 2008 15:45

78

AOGS 2007 - SE Volume

9in x 6in

L. D. Setijadji and K. Watanabe

b672-V13-ch06

Top level object classes of Geologic Map package. Fig. 2.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch06

An Integrative Geoscience Data Model by Linking Field-Specific Data Models

79

data, the reason for this addition is that we have developed a method to represent drillhole route as 3D line using linear referencing tool.6 The earth resource data model package (Fig. 3) is reﬁnement of our previous model6 with additional classes derived from the XMML data model especially related with industrial activities on metallic deposits. These classes include the mineral deposit classiﬁcation, commodity, quantity measures, mining activity, and their products. From our side, we have been working on two earth resource types, i.e. metallic deposits6 and geothermal.7 Both the earth resources are typically associated with the hydrothermal alteration zones (such as our case studies at Dieng geothermal ﬁelds and Selogiri porphyry prospect), so that we add a class related with hydrothermal alteration. The most intensive modeling so far done, which is also becoming the most signiﬁcant contribution originated from this project is on modeling the GeologicMeasurement with a focus on analytical works during the studies of metallic deposits, either in the exploration or in the evaluation stages (Fig. 4). However, as many of these analytical works are also applicable in other geosciences ﬁelds, such as radiometric dating and structural measurement, we put this group class within the Geologic Map package rather than in the Earth Resource. The geo-hazard package contains classes as seen in Fig. 5. Geo-hazards are parts of geological events that occur during particular periods. Based on the real cases in our study area, there are four types of geo-hazards that have been modeled, i.e. earthquake (Yogyakarta earthquake 27 May 2006), volcanic eruption (Merapi volcano eruptions 2006), tsunami (17 July 2006 tsunami along the south Java coast), and landslide (many landslides events were driven by 27 May 2006 earthquake). We model interrelationships among diﬀerent types of geo-hazards. For example, the 27 May 2006 Yogyakarta earthquake has triggered many landslides. There was also relationship between this earthquake and volcanic activity of Merapi. The 17 July 2006 tsunami was produced by a M 7.7 earthquake oﬀ the coast of Java. Finally, Merapi-style eruption is characterized by dome collapse that can be associated with landslide events. We model geo-hazard products into two kinds: earth materials (tephra, lahar, lava, debris deposit, and tsunamistone) and casualties. Earth material products can either be mapped for geological map or hazard map establishment. On the other hand, casualties are only used for generating geo-hazard map. In terms of geo-hazard mitigation, investigation on geo-hazard events can consist of observation and measurements (either

FA

December 10, 2008 15:45

80

AOGS 2007 - SE Volume

9in x 6in

L. D. Setijadji and K. Watanabe

b672-V13-ch06

Object classes within the Earth Resource package. Fig. 3.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

81

b672-V13-ch06

An Integrative Geoscience Data Model by Linking Field-Specific Data Models

Object classes within the GeologicMeasurement group. Fig. 4.

FA

December 10, 2008 15:45

82

AOGS 2007 - SE Volume

9in x 6in

L. D. Setijadji and K. Watanabe

b672-V13-ch06

Object classes within the Geo-hazard package. Fig. 5.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch06

An Integrative Geoscience Data Model by Linking Field-Specific Data Models

83

by taking samples or not). Most kinds of measurement methods have actually been modeled inside the geologic map package (Fig. 4), so that the geo-hazard package just makes a link to the related geological measurement classes.

4. Demonstration of Implementation The design model has been implemented using the ArcGIS 9.1 Geodatabase platform (MS Access format) in the Yogyakarta region, Indonesia. Figure 6

Fig. 6. Implementation of data model in the Yogyakarta region, central java, Indonesia using ArcGIS. The base map is a 3D view of digital geologic map of Yogyakarta region, while three overlying maps (Merapi, Selogiri, and Yogyakarta) represent three diﬀerent research projects in volcanology, mineral exploration, and earthquake hazard that share the same core database model. An example of relational database structure linked to maps is also shown.

FA

December 10, 2008 15:45

84

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch06

L. D. Setijadji and K. Watanabe

demonstrates how four involving and on-going research projects can share one integrative GIS data model. These projects are: (1) digital geological map project of the Yogyakarta region; (2) earthquake hazard study related with Yogyakarta earthquake 27 May 2006; (3) volcanology study of Merapi eruption; and (4) mineral deposit study on porphyry Cu–Au prospect at Selogiri district. In order to serve these four diﬀerent projects, we created one MS Access database for each project. Actually, we can create only one, big database for this purpose, especially if we use an “enterprise” database such as Oracle. However, to keep the uniqueness of each project and to maintain the performance of an MS Access database, we decide to make one database for each project. The concept of integrative database model here is more emphasized on its conceptual-design level rather than its physical implementation level. Each data model package of Geologic Map, Geo-hazard, and Earth Resources (Fig. 1) is then implemented into one or more physical databases. In this case, the Geologic Map package is implemented as the Yogyakarta Geologic Map database, the Geo-hazard package as the Yogyakarta Earthquake 2006 and the Merapi volcano databases, and the Earth Resources package as the Selogiri porphyry Cu– Au database. The majority of datasets are included in the Yogyakarta Geologic Map database, as it covers almost all geological concepts conceptually (Fig. 2) and many diﬀerent projects can share these generic data. For example, a fault map is a part of Geologic Map database rather than Geo-hazard database as it is used for many diﬀerent purposes. However, the event of earthquake on 27 May 2006, its aftershock events, and casualties map are included in the Yogyakarta Earthquake 2006 database. The same approach applies for Merapi and Selogiri projects. Figure 6 is a visualization of interaction among four GIS databases as described above. Each project has its own database, but they also share one generic database to be used in all projects, i.e. the geologic map.

5. Conclusions This data model project has successfully implemented an integrative approach to link application-speciﬁc data models (i.e. earth resource and geo-hazard) with the core digital geologic map data model. This method gives invaluable beneﬁts for managing our projects as we can use one geospatial information system to support diﬀerent geoscience

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch06

An Integrative Geoscience Data Model by Linking Field-Specific Data Models

85

applications and to synergize contributions from each geoscience ﬁeld for more comprehensive understandings on geological phenomena.

Acknowledgments This study is ﬁnancially supported by a Grant-in-Aid for Scientiﬁc Research of the Ministry of Education and Science, Japan (18404020). ESRI granted the ArcGIS 8.3 for the senior author for initiating this project, while the current software version of ArcGIS 9.1 is kindly supported by Prof. Tetsuro Esaki from Environmental Geotechnology Laboratory, Kyushu University. Authors thank two anonymous reviewers for their constructive suggestions.

References 1. M. Zeiler, Modeling Our World: The ESRI Guide to Geodatabase Design (ESRI Press, Redlands, 1998), 198 p. 2. P. A. Longley, M. F. Goodchild, D. J. Maguire and D. W. Rhind, Geographic Information Systems and Science (John Wiley and Sons, West Sussex, 2001), 454 p. 3. S. Grise and B. Brodaric, http://support.esri.com/datamodels (2003). 4. North American Geologic-map Data Model Steering Committee, http:// geology.usgs.gov/dm/ (2003). 5. S. M. Richard, D. R. Soller and J. A. Craigue, http://support.esri.com/ datamodels (2005). 6. L. D. Setijadji and K. Watanabe, Res. Geol. 55 (2005) 267 . 7. L. D. Setijadji, K. Watanabe, R. Wahyuningsih and D. Wintolo, Proc. World Geoth. Cong. (2005). R 05-087r4 (2006). 8. S. Cox (ed), OGC Best Practices OGC 9. G. L. Raines, J. T. Hastings and L. A. Moyer, http://support.esri.com/ datamodels (2007).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

This page intentionally left blank

b672-V13-ch06

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch07

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

CRUSTAL DEFORMATION MONITORING BY GNSS: NETWORK ANALYSIS AND CASE STUDIES MARCO PIRAS∗ and MARCO ROGGERO DITAG, Politecnico di Torino C.so Duca degli Abruzzi 24, Torino, 10129, Italy ∗ [email protected] MAURIZIO FANTINO NAVSAS, Istituto Superiore “M. Boella” C.so Castelfidardo 30/A, Torino, 10129, Italy

GNSS have been widely used to monitor crustal deformation with a high precision. In this chapter, the workﬂow of deformation monitoring by continuous GNSS observations is addressed. First, the permanent GNSS network characteristics and structure are described. Second, the GNSS observation data are processed by the Bernese software and corresponding solutions of GNSS coordinate time series are analyzed in detail at the time domain. Finally, the GNSS solutions are used to detect the deformation, and some case studies are presented.

1. Introduction The thin and deformable Earth’s crustal layer is subject to endogenous and exogenous agents, dominated by gravitation. Land deformation involves phenomena characterized by the long-term crustal movements and local short-term displacements. From a practical point of view, the study of the long-term phenomena requires that the geodetic observations are not aﬀected by local displacements. Moreover, for engineering applications, monitoring of local deformations is required. Global Navigation Satellite Systems (GNSS) techniques can be used in both cases, jointly with other geodetic techniques. If GNSS systems are the useful tools to investigate the Earth’s deformation, they are also employed as monitoring systems for local deformation. The goal of local phenomenon monitoring above all is to detect movements and then to estimate the entire deformation and direction. 87

FA

December 10, 2008 15:45

88

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch07

M. Piras, M. Roggero and M. Fantino

The local site can be instrumented with GNSS systems, and a control center saves raw data and estimates the solution in near-real time or postprocessing mode. As an alternative solution to the continuous operating network is set up, the site can be monitored by GNSS survey campaigns repeated at ﬁxed intervals; this solution is most frequently chosen in the case of small areas, such as landslide monitoring. In the last several years, a number of permanent stations has increased, although not uniformly, on the local, regional, and global network scale. GNSS’ continuously operating networks have diﬀerent purposes, both geodetic and geodynamic or, more often at the national or local level, which are designed for supporting topographic operation also in real time. Owing to recent densiﬁcation of GNSS arrays in boundary zones of tectonic plates, there is an increasing demand for computational programs incorporating realistic theoretical models. Recent studies include surface deformation stemming from dislocation at depth, crustal movements due to various loads, including jumps, periodic components and transient phenomena, incorporation of visco-elasticity, combination of data from diﬀerent satellite geodesy techniques, Laser Ranging, VLBI, and DORIS. Such studies are followed by several steps, like GNSS time-series analysis, tide gauge, linear regression, identiﬁcation of jumps, search set of faults or inﬂation sources, inversion of complicated slip distribution, temporal evolution of slips, modeling post-seismic transients, drawing diagrams, and so on. Herein, in this chapter, attention will be focused on the GNSS data production and analysis, following the workﬂow from the network setup, up to the coordinate time series analysis. Several case studies will be presented with permanent GNSS networks and repeated survey campaigns.

2. GNSS Monitoring Network and Data Processing The aim of GNSS techniques for monitoring crustal deformations is to distinguish from geodetic signals, which are generated by diﬀerent sources and mixed in the coordinate time series. To realize that, we need: • • • •

to set up a GNSS network, carefully designed and controlled; to use a reliable connection between the observation and the coordinates (correct datum deﬁnition and realization, models, and conventions); to check for clean and coherent results (remove outliers, systematic observation eﬀects, datum eﬀects); to detect and remove time series discontinuities, long-term linear trend and seasonal eﬀects.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Crustal Deformation Monitoring by GNSS

b672-V13-ch07

89

Monitoring networks should be analyzed according to three criterias: accuracy, reliability, and sensitivity. Accuracy describes the quality of the network in terms of random errors. Reliability is deﬁned as the ability of the network to sense and identify gross errors in the measurements, as the undetected inﬂuence of gross errors in a monitoring network can be easily interpreted as a deformation, thus resulting in an undesirable situation. The sensitivity of a monitoring network is deﬁned as its own capacity to detect and measure movements and deformations in the area covered by the network itself. The geological phenomenon is monitored and its parameters are used to analyze the sensitivity of a network. In a deformation area, the possible deformation model can be predicted according to information derived from geological and seismic surveys and geodetic results. The network sensitivity can be deﬁned with respect to the probability to detect a geological phenomenon; if it occurs, it can be detected with the probabilities α and β (α being the level of signiﬁcance and β the test power, as speciﬁed by the Baarda’s reliability theory).

2.1. GNSS continuously operating network setup The sensitivity of the network, deﬁned as its capacity to detect and measure movements and deformation in the said area, is a crucial criterion. It is therefore suggested that the network should be designed according to the geological phenomena to be monitored. The GNSS baselines conﬁguration design may be carried out mainly by the sensitivity criterion. With prior information, it is possible to design the network according to a deﬁned deformation model. This conﬁguration will then contain the vectors with the most eﬀective contribution to the network sensitivity. The proposed site location must be explored afterward, in order to ﬁnd the suitable location for station installation, where a number of other technical and logistic factors have to be considered (bedrock type, sky view, background noise, safety, accessibility, power availability, and data transmission). Preliminary GNSS campaign and analysis of the observations’ quality control (QC) are the necessary steps for evaluating the candidate sites (Fig. 1), at ﬁrst selected on the basis of geophysical, geological, or geotechnical requirements. This operation is performed collecting raw data on the candidate site for 48 h and then analyzing the observations following the proper quality standards (UNAVCO).

FA

December 10, 2008 15:45

90

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch07

M. Piras, M. Roggero and M. Fantino

Fig. 1.

Skyplot of multipath.

UNAVCO, a non-proﬁt, membership-governed consortium, supports and promotes Earth science by advancing high-precision techniques for the measurement and understanding of deformation. Most of UNAVCO’s members are geophysical geodesists who study deformation. The primary tool supported by UNAVCO has been GPS. However, UNAVCO is moving toward including support for other techniques useful for studying deformation. Borehole strainmeters, Interferometric Synthetic Aperture Radar (InSAR), and Light Detection and Ranging (LIDAR) are expanding the spatial and temporal signals that can be investigated with geodetic techniques. At the same time, GPS is ﬁnding applications in a frequency range that used to be the sole provenance of seismology, as GPS moves from one solution per day to one solution per second. UNAVCO is also expanding its role in education. These changes in UNAVCO are part of a conscious strategy to meet the future needs of the science community supported by UNAVCO. During the past year, UNAVCO developed a Strategic Plan to guide it through the next 5 years.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Crustal Deformation Monitoring by GNSS

b672-V13-ch07

91

If the sites satisfy the quality requirements, some logistic aspects have to be evaluated for the power supply and data link, and ﬁnally the access for the station maintenance and also the security against vandalism acts. The power supply is usually realized easily by photovoltaic cells and accumulators, and the major technical problems are often related to the data link. GSM modems and radio modems are the most frequently practiced solutions, even though in local networks, such as in landslide monitoring, it can be more convenient to set up a wireless LAN. This is possible thanks to the capability of modern GNSS receiver to communicate through the Ethernet protocol. The monumentation has to be studied for providing a good stability minimizing the realization costs (Fig. 2). Setting up a permanent station for deformation monitoring, it is usually necessary to anchor the monument directly on the bedrock, with the exception of stations situated on sedimentary layers. However, the bedrock can be covered by soil layers, with a thickness of some meters, or can be fractured by faults. In such cases, preliminary geological and geotechnical studies are necessary for evaluating the site-supposed local stability and the necessary structural solutions to set up the station. Finally, the economical aspects can suggest to choose another site.

Fig. 2.

Monumentation’s map.

FA

December 10, 2008 15:45

92

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch07

M. Piras, M. Roggero and M. Fantino

A low data sampling rate (i.e. 15 s) is enough to support most of the applications for a deformation monitoring network, with the advantage of requiring a low bandwidth for the data transmission. Higher data sampling rate, up to 1 Hz, is desirable to support speciﬁc GPS studies, such as investigation of ionospheric scintillations and the water vapor tomography during occultations of Low Earth Orbiters (LEO) with GPS receivers on board. However, there is growing interest in collecting data with high rates also for a variety of geodynamic applications. GPS measurements can accurately track the propagation of earthquake dynamic motions both on the ground and in the atmosphere, providing complementary information to seismic observations and estimates of Earth structure.

2.2. Monumentation The international standards for permanent GPS stations involved in geodynamical studies, and the fact that tectonic displacements are often of the order of only a few millimeters per year, require to achieve the highest possible stability of the monument. Moreover, to detect crustal deformation, it is underlying to be able to distinguish the land deformation from the monumentation deformation (temperature excursion, concrete ﬂuage, wind, etc.), or to minimize them. Usually, the most used monumentation is the concrete pillar, preferably anchored directly on the bedrock. However, diﬀerent solutions have been proposed, the most diﬀused are: • • • •

concrete pillar (UNAVCO rebar and C-bar reinforced concrete pillar) metallic pillar (UNAVCO stainless steel) tripod (UNAVCO short-deep drilled braced) steel mast.

The very common practice of installing stations on building roofs is in this case not advisable, because the observations could be aﬀected by oscillations or seasonal building displacements related to thermal expansion. A research overview conducted in Italy, analyzing diﬀerent Italian GNSS networks designed both for topographic and geodynamic purposes, shows that the stations monumented directly on the ground are 43%, while 57% are monumented on top of buildings. The 77% of the station monuments on the ground are concrete pillars (33% of the total), 12% steel tripod, and 7% steel pole; the remaining 4% are steel mast.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Crustal Deformation Monitoring by GNSS

b672-V13-ch07

93

Apart from the diﬀerent opinions in the geodetic community about the better monumentation (concrete pillar or steel tripod), the following guidelines must be followed in permanent GNSS station setup. The monument must be well anchored into solid bedrock, and in order to deﬁne whether exposed rocks have optimal features, it is mandatory to perform inspections with a geologist. The anchorage between pillar and bedrock must be done by means of iron bars of suitable suitability section, which must be inserted into the ground to 2–3 m deep. The iron bars, which are usually ﬁxed to the bedrock by means of special glues, have to emerge from the ground to allow the coupling of the monument’s structure. The monument height should be about 150 cm, or in any case greater than the average registered snow height. The receiver antenna must be linked to the monument by means of an iron mount device, leveled upon installation.

2.3. Local control network Site eﬀects and monument stability at permanent GNSS stations are controlled by monitoring a local control network, formed by three or more GNSS and leveling points. These control points can be materialized on subhorizontal rock surfaces by means of steel geodetic markers. Where it is not possible to ﬁnd suitable surfacing bedrock, the control points can be set up on the foundations of massive and stable buildings. In some cases, only subvertical surfaces are available to set up the control points, or the only possible locations for the control points are not suitable for acquiring GNSS observations: in these cases the control points can be used only for leveling. The distance of the control points from the GNSS station depends on the wavelength of the local phenomena to be monitored. Moreover, this distance must be a compromise between diﬀerent factors such as: bedrock availability, GNSS satellite visibility, time necessary to reach the control point or to perform the leveling, chance to preserve the control point. Local control network setup and monitoring are of primary importance. However, the monitoring is quite expensive in terms of time and human resources in the ﬁeld, and a monitor survey campaign repeated once per year can be a good compromise. An additional very important role of the control network is represented by the possibility to determine on-site the previous location of the GNSS antenna, in case the pillar is damaged or destroyed by natural causes, accidents, or acts of vandalisms. If it would ever be necessary to substitute

FA

December 10, 2008 15:45

94

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch07

M. Piras, M. Roggero and M. Fantino

or relocate the monument, in fact, control points are the only way to link the new antenna location to the previous position and avoid wasting previous measurements. For referencing reasons, it is very important that the control points are also observed by GNSS.

2.4. Data processing Brieﬂy, the diﬀerent purposes of the data analysis could be summarized as follows: • • • •

integration of the monitoring network with the global network and with neighboring networks; data quality control, by means of automatic procedures; production of coordinate time series, to be analyzed for the deformation monitoring; production of ionosphere and troposphere estimates.

In order to determine station positions and velocities of the entire monitoring network, a unique and consistent reference system has to be used in all steps of data processing for coordinate determination. Position coordinates refer to a coordinate system that has to be deﬁned coherently, and velocities need a stable origin to which the motions refer. The reference system is realized in practice by a reference frame, which is a set of reference stations with coordinates according to the deﬁnition of the reference system. The reference frame is broadcasted to the user by station coordinates, ephemeris, and related products. Due to crustal deformation (tectonic plate motions and regional deformations), the coordinates of the stations change with time relative to the satellite orbits. Therefore, we have to transform the coordinates of the reference frame by the station velocities to the observation epoch in the data processing.1 The reference system for the coordinates of the satellites and of the ground stations must be the same, and needs a dense spacing in the region of the measurements. This is only fulﬁlled by the ITRF, at present in its realization ITRF2005, consisting of a table of station coordinates and velocities at a reference epoch t0 = 2000.0, under the hypothesis of constant velocities. A GPS-based ITRS realization series, IGS05, is managed by the IGS, and distributed via ephemerides (GPS and GLONASS) and Earth Orientation Parameters (EOP). In IGS05, the GPS receiver phase center variations are corrected according to the IGS conventions of 2006. Both ITRF2005 and IGS05 realizations are characterized by position mean

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Crustal Deformation Monitoring by GNSS

b672-V13-ch07

95

standard deviations at the millimeter level. Nevertheless, discrepancies between the ITRF and IGS (long-term and weekly solutions) realizations may be as large as several millimeters. IGS realization, even if not more accurate, is indeed more consistent with GNSS observations. For this reason it seems more rigorous to perform the network adjustment in the IGS datum, and all the standards used for the computation of the IGS global network must be used in the adjustment, if applicable. The connection with the ITRF is known and the transformation parameters are published by the IGS. It must be noted that the variances and covariances of the coordinates of the IGS weekly solutions are quite underestimated, and therefore it is not correct to use them in the stochastic model for the adjustment of the PN. Therefore, the weekly solutions must be used through a procedure that includes the estimates of their variances and covariances from the time series of the coordinates. In order to get a dense terrestrial reference frame, an ITRF/IGS densiﬁcation is necessary for the IGS network non-uniform distribution. The GNSS data-processing strategy depends on the software used, and can be deﬁned through models, products, and procedures. It is necessary to deﬁne satellite orbits and Earth rotation parameters, antenna phase center variation, ionospheric and tropospheric models, ocean tide loading, ambiguity resolution strategy, cut-oﬀ angle, datum, and so on. The ﬁnal products, including station coordinates, depend on all these deﬁnitions and models.4

3. Time Series Analysis The most important results of the data analysis are the daily coordinates of the GNSS Permanent Stations. The long-term time series of daily solutions are then analyzed to get estimates of the linear station velocities, which are a basic input for geodynamic interpretation. A detailed time-series analysis however must be performed before estimating the station velocities. The analysis of the noise characteristics and of the periodic signal allows a deeper insight into some processes in the network and at the station’s environment as well. An appropriate statistical analysis of the colored noise allows a more reliable velocity uncertainty estimation and may provide information on the monument stability. The periodic seasonal signals may come from real physical sources (e.g. atmospheric and hydrologic eﬀects), and also mismodeling or local environmental eﬀects (e.g. multipath, thermal eﬀects) may contribute to the periodic coordinate variations. Coordinate time-series

FA

December 10, 2008 15:45

96

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch07

M. Piras, M. Roggero and M. Fantino

analysis requires the computation of a cumulative (multi-year) solution resulting in clean time-series plots, a catalog with the station coordinates and velocities, and a catalog with the coordinate oﬀsets and outliers. Then it is possible to determine realistic coordinate and velocity uncertainties by estimating the spectral index of the noise of the coordinate time series, and ﬁnally the harmonic analysis of the coordinate time series can be performed by estimating the seasonal periodic term and providing de-trended, cleaned time-series plots. The harmonic analysis of GNSS coordinate time series shows seasonal variation of the station coordinates, usually with an annual or semi-annual period. The distribution of the amplitudes and phases is often fairly random and hard to interpret as a real geophysical signal. Rather, they reﬂect the modeling shortcuts of antenna phase center variations (PCV) and environmental impacts (multipath, thermal eﬀects). Discontinuities in the time-series analysis can also occur in case of Reference Frame change or models change. Even though coordinate transformation is applied with proper parameters, residual discontinuities can be present after datum shift compensation. The residual discontinuities are further estimated and removed from the time series. The ﬁrst results of the comparison between IGS05 (with absolute PCV) and IGb00 solutions (with relative PCV) largely support the importance of the introduction of the absolute PCV models and then the general reprocessing of the whole GPS observation material using the new ITRF2005 frame and the new PCV models (Fig. 3). Other documented jumps are caused by hardware change, such as of antenna, receiver, or site modiﬁcations, but undocumented jumps remain. The main problem during the time-series analysis is to detect and remove these undocumented jumps, in order to obtain reﬁned time-series suitable for reliable velocity estimates. Diﬀerent algorithms can be used to detect and estimate the discontinuities in time series as traditional method (i.e. Least Squares) or robust method (i.e. Huber method) or sequential approach (i.e. Kalman ﬁlter2,3 ). After the jumps (documented and undocumented) are removed, the station velocities can be estimated, and the long-term linear trend can be removed. This is fundamental for the frequency analysis, the purpose of which is to recover principal signals without a priori hypothesis on their period and amplitude. The sampling of the time series is usually non-uniform due to data gaps, so that it is not possible to realize the frequency analysis using the FFT.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch07

Crustal Deformation Monitoring by GNSS

Fig. 3.

97

Example of coordinate time series with some discontinuities.

However, other algorithms can be suitable, such as FAMOUS — Frequency Analysis Mapping On Unusual Sampling, by F. Mignard (OCA/Cassiopee). Figure 4 shows diﬀerent kinds of information on seasonal and annual eﬀects and the long-term linear trend. The vertical line indicates the reference frame changes, in particular from IGS97 to IGS00 (at the left) and from IGS00 to IGS05 (at the right).

4. Some Case Studies 4.1. The Sumatra–Andaman earthquake The ﬁrst case study concerns the drastic phenomenon of Sumatra–Andaman earthquake, which occurred in 2004. Long-term GPS time-series analysis shows that the Australian and Indian plates move toward Sumatra– Andaman at the velocities of 5 cm and 4 cm per year, respectively. These results are derived from a large network of about 200 permanent GPS stations, routinely analyzed for various research purposes, including the

FA

December 10, 2008 15:45

98

AOGS 2007 - SE Volume

9in x 6in

M. Piras, M. Roggero and M. Fantino

Seasonal and annual eﬀects estimated using the FAMOUS algorithm. Fig. 4.

b672-V13-ch07

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Crustal Deformation Monitoring by GNSS

b672-V13-ch07

99

monitoring of global sea level, climate change, and crustal deformation in Southeast Asia, Australia, the South Paciﬁc, and Antarctica. Diﬀerent displacements have been detected during the event of 2004 by GPS, and the entity and velocity of displacement vary with location. Displacements vary from 3 to 6 m in the Andaman and Nicobar Islands. An almost 28 cm displacement was detected at GPS site PHUK (Phuket Island, southern Thailand near northern Malaysia), decreasing gradually from the north to the south. In this analysis, the GPS data can detect coseismic, kinematic, and post-seismic deformation. Co-seismic displacements were computed using two combined seven-day solutions — one before and one after the earthquake. Displacements of the sites were calculated as the diﬀerences between the two solutions. The co-seismic deformation at GPS site ARAU (Perlis, northern Malaysia) are about 15 cm in the east and 3 cm in the north. The southeast sites seem not to have been impacted by the earthquake, which implies that the stress along the subduction zone plate interface of southern Sumatra was not released. This is the likely reason for the Simeulue–Nias earthquake on 28 March 2005. The determination of co-seismic deformation is very useful for further investigation of fault slip models and of other seismic features of the earthquake.6 Kinematic coordinate solutions, computed from stations near the earthquake every 30 s over 30-min period before and after the earthquake (0:59, 26 December 2004), show the progression of the rupture. For example, GPS site ARAU had a deformation of 10 cm. Deformation was detected when the surface waves began to hit the site 2 min after the earthquake; 4 min later, positions at the site were relatively stable again. Using a long-term GPS time series after the earthquake, we can also examine the post-seismic deformation process. An eastward deformation of more than 6 cm was determined during the 80-day period after the earthquake. Such post-seismic deformation information from all available GPS sites in the earthquake region can help scientists analyze likely elastic, poroelastic, and viscoelastic deformation, and plastic ﬂow of the Earth’s crust in the earthquake region, giving a better understanding of crustal relocation and redistribution after the earthquake. 4.2. Crustal deformation in Japan The Geographical Survey Institute (GSI) has established the GPS control network with about 1224 control points throughout the country, which is designed to monitor crustal movement in real time.

FA

December 10, 2008 15:45

100

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch07

M. Piras, M. Roggero and M. Fantino

Crustal movements are not only caused rapidly by earthquake swarm or big earthquake, but also slowly by plate motions or magma activities. It is necessary to detect ongoing crustal deformation with high precision for researches of earthquakes and volcanic eruptions. The Geographical Survey Institute (GSI) conducts baseline analysis every day using 24 h GPS observation data; therefore, it is possible to detect crustal deformation continuously. GPS data is expected to contribute to the prediction of Tokai Earthquake that is anticipated in near future, and to the studies of an earthquake mechanism.

4.3. Crustal deformation in the Alps Along the Alps there are two diﬀerent geodetic GNSS networks, devoted to monitoring the Alps deformations: the ﬁrst one is REGAL, installed in 1997 by G´eosciences Azur and the second one is FReDNet operated by The Istituto Nazionale di Oceanografia e di Geofisica Sperimentale (OGS), which monitors the deformation along the northeast boundary of the Adriatic microplate (∼5 mm/yr north–south convergence). Other permanent stations have been installed in the context of the RING network, managed by the Istituto Nazionale di Geofisica e Vulcanologia (INGV). There is, besides, a transnational network called GAIN (Geodetic Alpine Integrated Network) installed during 2004–2006, funded by the EU Alpine Space III-B Program, which involves several research and environmental institutions standing within the Alpine Space (Fig. 5). The aim of the project is to build up a high-performance space geodetic network of continuous GPS (CGPS) receivers in the Alps and a number of campaign subnetworks. The Geodetic Alpine Integrated Network consists of more than 35 CGPS and will be the ﬁrst ever installed transnational geodetic network across the Alps. It will delineate the seismogenic potential within the Alpine space, mountains, and surrounding foothills, where concentrated attractive European metropolitan areas and rapidly growing urban centers with extensive infrastructures are present. It will favor transnational know-how exchange between regional authorities and Alpine universities and research centers. It will reinforce the European Space geodetic and geoscience communities.5 This will open new research initiatives in earth and environmental sciences, therefore rising the value of the Alps as a natural laboratory. The direct result will be an improvement in the knowledge of earthquake

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Crustal Deformation Monitoring by GNSS

Fig. 5.

b672-V13-ch07

101

Geodetic Alpine Integrated Network (GAIN).

potential and hazard, and this will allow a better land use in terms of a safe living space. GAIN will give ground for a higher resolution space-based coverage of urban and mountain areas in the Alpine space, better resolve satellite imagery, and therefore oﬀers a robust tool for future infrastructure investment, land use harmonization, and industrial planning. GAIN will be integrated with highly resolving remote-sensing methods (e.g. InSAR) and will give new potentials to monitor and prevent environmental degradation and limit the impacts of natural disasters. First results of the GAIN network are presented, focusing on the problem of misinterpretation of too short time series. In the case of the Alps, the deformations are not as large as in the previous examples, and we need to observe very long time series. The dominant source of horizontal deformation is caused by plate tectonics. It is therefore natural to remove this signal for studying the remaining deformation. A well-known and generally accepted model for plate tectonics is the NNR-NUVEL-1A model, based on a geological and geophysical data. The remaining horizontal velocities are now much smaller and the diﬀerential deformation signals, which are not caused by the plate tectonics of the Eurasian plate, are visible. The actual estimation of these horizontal velocities does not show a certain pattern and is rather randomly distributed. But this should not be misinterpreted, since the time series

FA

December 10, 2008 15:45

102

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch07

M. Piras, M. Roggero and M. Fantino

of the position changes is rather short, at the maximum of 2 years. It is quite typical to see seasonal variations in the time series of the coordinate components. The origin of these signals is yet not well understood. It is therefore necessary to observe the stations for several years. Then, the seasonal signal averages out and does not aﬀect the velocities. 5. Conclusions GNSS nowadays is a reliable technique for monitoring deformations, but it still requires some improvements in data-processing and analysis. We underlined some considerations about the main activities and the problems still to be solved. Analyzing coordinate time series, we need to detect the non-tectonic motion (e.g. documented and undocumented jumps), the long-term linear trend (station velocity), and the periodic components (annual–semiannual). Known discontinuities must be reported in log ﬁles, but we cannot be sure that all changes are always documented and that all metadata are updated. Moreover, the source of undocumented time-series discontinuities must be investigated. The long-term linear trend represents the station velocity with respect to the reference frame, constant or not after an earthquake; the long-period signal represents the non-linear part of the deformation, and it needs very long time series to be detected (>10 years). The main diﬃculty is to verify the reliability of the estimated deformations, and to explain their physical causes (tectonic drift, seasonal eﬀects, etc.). Finally, the growing number of permanent GNSS monitoring networks with numerous international cooperation, is increasing the resolution of crustal deformation knowledge, allowing the development of more realistic geophysical models.

References 1. W. Baarda, A testing procedure for use in geodetic networks, Publ. Geodesy 2(5) (1968), Netherlands Geodetic Commission, Delft, ISBN-13: 978-90-6132209-2. 2. R. E. Kalman, A new approach to linear filtering and prediction problems, Trans. ASME J. Basic Engr. (1960) 35–45. 3. Y. J. Wang and K. K. Kubik, Robust Kalman filter and its geodetic applications, Manuscripta Geodetica 18 (1993) 349–354. 4. M. Crespi, A software package for the adjustment and the analysis of GPS control networks, Reports on surveying and geodesy (1996), pp. 237–264.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Crustal Deformation Monitoring by GNSS

b672-V13-ch07

103

5. G. Even-Tzur, Sensitivity design for monitoring deformation network, Bollettino Di Geodesia E Scienze Aﬃni 58 (4) (1999) 313–324. 6. J. Minghai, Crustal deformation from the Sumatra-Andaman earthquake, AusGeo News No. 80 (2005).

Web References GAIN http://www.alps-gps.units.it/ IGS http://igscb.jpl.nasa.gov/ EPN http://www.epncb.oma.be/ Japan http://mekira.gsi.go.jp/ENGLISH/

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

This page intentionally left blank

b672-V13-ch07

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch08

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

THE GLOBAL GEODETIC OBSERVING SYSTEM H.-P. PLAG∗ , M. ROTHACHER† , M. PEARLMAN‡ , R. NEILAN§ and C. MA¶ ∗University of Nevada, Reno, Nevada Bureau of Mines and Geology and Seismological Laboratory Mail Stop 178, Reno, NV 89523, USA [email protected] http://geodesy.unr.edu/ †GeoForschungZentrum

Potsdam Potsdam, Germany

‡Harvard-Smithsonian

Center for Astrophysics Cambridge, Massachusetts, USA

§Jet

Propulsion Laboratory, Pasadena, California USA ¶Goddard

Space Flight Center Greenbelt, Maryland, USA

The Global Geodetic Observing System (GGOS) was established by the International Association of Geodesy (IAG) in July 2003. In April 2004 the IAG, represented by GGOS, became a participating organization of the Group on Earth Observation (GEO) and in May 2006 GGOS was accepted as a member of the Integrated Global Observation Strategy Partnership (IGOS-P). GGOS is the contribution of geodesy to the Global Earth Observation System of Systems (GEOSS). It provides the reference systems and frames, which are crucial for Earth observing systems. GGOS is built on the IAG Services (IGS, IVS, ILRS, IDS, IERS, IGFS, etc.) and the products they derive on an operational basis for Earth monitoring, making use of space- and groundbased geodetic techniques such as Very Long Baseline Interferometry (VLBI), Satellite and Lunar Laser Ranging (SLR/LLR), Global Navigation Satellite Systems (GNSS), Doppler Orbitography and Radiopositioning Integrated by Satellite (DORIS), altimetry, InSAR (Interferometric Synthetic Aperture Radar), gravity satellite missions, and gravimetry, etc. All these observation techniques are considered integral parts of GGOS, allowing the monitoring of the Earth’s shape and deformation (including water surface), the Earth’s orientation and rotation, and the Earth’s gravity ﬁeld and its temporal variations with an unprecedented accuracy. The observed parameters give direct evidence of many global processes that have a crucial impact on human

105

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch08

H.-P. Plag et al.

106

society such as earthquakes, volcanism, ﬂoods, sea-level change, climate change, groundwater redistribution, mass balance of the polar ice sheets, etc. GGOS relies on the observing systems and analysis capabilities already in place in the IAG Services and envisions the continued development of innovative technologies, methods, and models to improve our understanding of global change processes. GGOS provides a framework that ranges from the acquisition, transfer, and processing of a tremendous amount of observational data to its consistent integration and assimilation into complex numerical models of the Earth system (including solid Earth, oceans, atmosphere, hydrosphere, cryosphere, and the interactions thereof). This is being achieved by an international eﬀort, and a close, multi-disciplinary cooperation with groups working in related ﬁelds such as geodynamics, geophysics, oceanography, hydrology, glaciology, meteorology, and climatology. In summary, GGOS provides essential contributions to an integrated Earth monitoring system to help us better understand global change and its impact on environment and society.

1. Introduction Earth is a restless planet.1 With its atmosphere, oceans, ice covers, land surfaces, and its interior, it is subject to a large variety of dynamic processes operating on a wide range of spatial and temporal scales, and driven by large interior as well as exterior forces. Many areas of the Earth’s surface are exposed to natural hazards caused by dynamic processes in the solid Earth, the atmosphere, and the ocean. Earthquakes, tsunamis, volcano eruptions, tectonic deformations, land slides, deglaciation, sealevel rise, ﬂoods, desertiﬁcation, storms, storm surges, global warming, and many more are typical and well-known phenomena that are expressions of the dynamics of our restless planet. In modern times these processes are inﬂuenced, as well, by anthropogenic eﬀects; to what extent is still largely unknown. Some of the many examples of anthropogenic eﬀects are carbon and methane emissions, changes in soil composition and erosion rates, regulations and diversions of rivers, deforestation, and extinction of species (see Ref. 2 for an overview of the anthropogenic impact on the Earth system over the last 300 years). A growing population has to cope with this restless, and ﬁnite, planet. Settlements are encroaching into areas of high risks from natural hazards with major infrastructure being built in potentially hazardous locations, thus increasing the vulnerability of society. Increasingly, valuable and crucial infrastructure is lost in natural disasters, aﬀecting the economy on national and global levels, and displacing large populations, with

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The Global Geodetic Observing System

b672-V13-ch08

107

severe social implications. The growing demands for access to food, water, materials, and space put stress on the ﬁnite resources of the planet. Earth system processes, whether natural or modiﬁed by humans, aﬀect our lives and the lives of future generations. Decisions made today will inﬂuence the well-being of future generations. In order to minimize the anthropogenic impact on Earth system processes and to preserve resources for future generations, a better understanding of Earth system processes and an eﬃcient and conservative organization of anthropogenic processes is required. Responsible stewardship of the planet is not possible without a profound understanding of the processes that shape the planet. Examples are mitigation of the potential impact of climate change on ecosystems, sustainable management of the oceans, preservation of water resources for humans and the biosphere, and preparing for a potentially devastating impact of sea-level rise on coastal communities. Living on a restless planet with ﬁnite resources and a limited capacity to accommodate the impact of the increasingly powerful anthropogenic factor requires careful governance. A number of World Summits have acknowledged that ﬁnding a way to ensure sustainable development is mandatory for realizing a stable and prosperous future for the anthroposphere. Although there are many other inﬂuential factors, understanding the Earth system and its major processes and its trends, is one of the prerequisites for success in this quest for sustainable development. A deeper understanding of the Earth system cannot be achieved without suﬃcient observations of a large set of parameters characteristic of Earth system processes. As emphasized by the Earth Observation Summits (EOS), there is an urgent need for comprehensive Earth observations (see the documents in the Appendices of Ref. 3). Earth observations are not only necessary for a scientiﬁc understanding of the Earth, but are also fundamental for most societal areas ranging from disaster prevention and mitigation, the provision of resources such as energy, water, and food, gaining an understanding of climate change, the protection of the biosphere, the environment, and human health, to the building and management of a prosperous global society. Geodesy provides mandatory reference frames as a foundation for Earth observation. Moreover, geodesy observes parameters related to the mass transport in the Earth system and the system dynamics. With this, geodesy is a cornerstone in Earth observation.

FA

December 10, 2008 15:45

108

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch08

H.-P. Plag et al.

2. Geodesy’s Contribution to Earth Observations and Society at Large The “three pillars” of geodesy are the Earth’s time-dependent geometric shape, gravitational ﬁeld, and rotation (Fig. 1). With its observational means (Table 1), geodesy has the potential to determine and monitor with utmost precision the geometric shape of land, ice, and ocean surfaces as a global function of space and time. The geometric methods, when combined with global gravity information and the geoid, allow us to infer mass anomalies, mass transport phenomena, and mass exchange in the Earth’s system. The variations in Earth rotation reﬂect mass transport in the Earth system and the exchange of angular momentum among its components. The geodetic observations of the “three pillars” provide the basis for the realization of the reference systems that are required in order to assign (time-dependent) coordinates to points and objects, and to describe the motion of the Earth in space (Fig. 1). For this purpose, two reference systems are basic in geodesy, namely the celestial reference system and the terrestrial reference system, which are dynamically linked to each other by the Earth’s rotation. The two most accurate reference systems currently available are the International Celestial Reference System

Fig. 1. Constituents of an integrated geodetic monitoring system. The “three pillars” of geodesy provide the conceptual and observational basis for the reference frames required for Earth observation. Moreover, these “three pillars” are intrinsically linked to each other as they relate to the same unique Earth system processes. Today, the spacegeodetic techniques and dedicated satellite missions are crucial in the determination and monitoring of geokinematics, Earth’s rotation, and the gravity ﬁeld. Together, these observations provide the basis to determine the geodetic reference frames with high accuracy, spatial resolution, and temporal stability. (From Ref. 5 modiﬁed from Ref. 6).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch08

The Global Geodetic Observing System

109

Table 1. Observing the “three pillars” of geodesy and maintaining the global geodetic reference frames. GGOS utilizes a large set of space-geodetic, airborne and in situ techniques. (For acronyms, see text and Table 2.) Component I. Geokinematics (size, shape, kinematics, deformation)

Objective

Shape and temporal variations of land–ice–ocean surface (plates, intraplates, volcanoes, earthquakes, glaciers, ocean variability, sea level) II. Earth rotation Integrated eﬀects (nutation, of changes in precession, angular polar motion, momentum variations in (mass changes LOD) in atmosphere, cryosphere, oceans, solid Earth, core/mantle; momentum exchange between Earth system components) III. Gravity ﬁeld Geoid, Earth’s static gravitational potential, temporal variations induced by solid Earth processes and mass transport in the global water cycle IV. Terrestrial Global cluster of frames ﬁducial points, determined at mm to cm level

Techniques

Responsible

Altimetry, InSAR, GNSS cluster, VLBI, SLR, DORIS, imaging techniques, leveling, tide gauges

International and national projects, space missions, IGS, IAS, future InSAR service

Classical astronomy, VLBI, LLR, SLR, GNSS, DORIS; underdevelopment: terrestrial gyroscopes

International geodetic and astronomical community (IERS, IGS, IVS, ILRS, IDS)

Terrestrial gravimetry International (absolute and geophysical and relative), airborne geodetic gravimetry, community (GGP, satellite orbits, IGFS, IGeS, BGI) dedicated satellite missions (CHAMP, GRACE, GOCE)

VLBI, GNSS, SLR, International geodetic LLR, DORIS, time community (IERS keeping/transfer, with support of absolute IVS, ILRS, IGS, gravimetry, gravity and IDS; BGI) recording

FA

December 10, 2008 15:45

110

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch08

H.-P. Plag et al.

(ICRS) and the International Terrestrial Reference System (ITRS), which are deﬁned by the International Earth Rotation and Reference Systems Service (IERS). These systems are conventional coordinate systems that include all conventions for the orientation and origin of the axes, the scale, and the physical constants, models, and processes to be used in their realization. Based on observations, these systems can be realized through their corresponding “reference frames.” The frame corresponding to the ICRS is the International Celestial Reference Frame (ICRF), which is a set of estimated positions of extragalactic reference radio sources. The frame corresponding to the ITRS is the International Terrestrial Reference Frame (ITRF), which is a set of estimated positions and velocities of globally distributed reference marks on the solid Earth’s surface. These two frames are linked to each other by the estimates of the Earth rotation parameters. ICRS, ITRF, and the Earth rotation parameters are provided by IERS. Today, the internationally coordinated geodetic observations collected and made available by the global geodetic station networks provide a continuous monitoring of the ITRF. This well-deﬁned, long-term stable, highly accurate, and easily accessible reference frame is the basis for all precise positioning on and near the Earth’s surface. It is the indispensable foundation for all sustainable Earth observations, in situ, as well as airborne and space-borne. Furthermore, the ITRF underpins all geo-referenced data used by society for many uses. All these digital geo-referenced data are crucial for many activities, including mapping, construction, land development, natural resource management and conservation, navigation — in fact all decision-making that has a geo-related component. Historically, geodesy was limited to determining the Earth’s shape, gravity ﬁeld, and rotation including their changes over time. With modern instrumentation and analytical techniques, the scope of geodesy can be extended to include the sources of the observed changes, that is, the dynamics of and mass transport within the Earth system.4 With this broader scope new pathways emerge in which geodesy can contribute to the scientiﬁc understanding of the Earth system as well as the development, functioning, and security of society in general. Ultimately, the observations in these “three pillars” are aﬀected by the same unique Earth system processes: all of them relate to mass redistribution and dynamics (Fig. 2). Thus, geodesy provides a unique framework for monitoring and ultimately understanding the Earth system. Modern space-geodetic techniques are well suited for observing phenomena on global to regional scales, and thus are an important complement to traditional in situ observation systems.

FA

December 10, 2008 15:45 AOGS 2007 - SE Volume 9in x 6in

The Global Geodetic Observing System

b672-V13-ch08

111

Fig. 2. Mass redistribution in the Earth system. All geophysical processes are associated with mass redistribution and changes in the dynamics, thus aﬀecting commonly the Earth’s gravity ﬁeld, geometry, and rotation. Consequently, geodesy with observations of the “three pillars” contributes to an observing system that allows the monitoring of mass transport in the Earth system. (From Ref. 7).

FA

December 10, 2008 15:45

112

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch08

H.-P. Plag et al.

Many scientiﬁc applications depend on detailed knowledge of the Earth’s shape, its gravity ﬁeld, and rotation. In the past, geodesy has, with ever-increasing accuracy, provided the necessary observations, although with many limitations in accuracy and spatial coverage. The fairly recent advent of space-geodetic techniques has brought about a rapid development in global geodesy, particularly during the last decade or so. The relative precision of the measurements is approaching the very impressive level of 1 part-per-billion (ppb) or even better. Today, geodetic techniques permit the measurement of changes in the geometry of the Earth’s surface with an accuracy of millimeters over distances of several 1,000 km. Over the last one and a half decades, the global geodetic networks have provided an increasingly detailed picture of the kinematics of points on the Earth’s surface and the temporal variations in the Earth’s shape. Among other applications, the observations have been used to determine improved models of the secular horizontal velocity ﬁeld (e.g. Refs. 8–10), to derive seasonal variations in the terrestrial hydrosphere,11 to study seasonal loading,12 to invert for mass motion,13 and to improve the modeling of the seasonal term in polar motion.14 Geodetic techniques provide the means to observe surface deformations on volcanoes,15 –17 in unstable areas,18 associated with earthquakes and fault motion,19 –21 or subsidence caused by anthropogenic activities such as groundwater extraction.22 Current developments indicate that geodetic observing techniques will be able to determine the magnitude of great earthquakes in near-real time and thus help mitigate the problem of low initial magnitudes estimated by seismic techniques.23 The space-geodetic techniques and methods also enable auxiliary applications that utilize the atmospheric disturbance of geodetic measurements (ionosphere, troposphere, magnetic ﬁeld) for non-geodetic applications. The distortions of geodetic microwave signals propagating through the atmosphere can be inverted and used for weather prediction,24–26 climate studies, and studies in atmospheric physics. Air temperatures retrieved from GNSS radio occultation technique provide new and near-complete coverage of the Earth’s atmospheric mass ﬁeld in the upper troposphere and stratosphere, complementing passive measurements from existing infrared and microwave sounders. Water vapor in the lower troposphere is relevant for forecasts of precipitation, while water vapor in the upper troposphere is the largest contributor to the atmospheric greenhouse eﬀect. Thus, the geodetic techniques provide observations relevant for numerical weather forecast as well as climate studies.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The Global Geodetic Observing System

b672-V13-ch08

113

To a large extent, geodesy is a “service science,” and as such, the development of the geodetic observing system should be guided by the requirements of its users.27 In the past, the main “customers” of geodesy came from the surveying and mapping profession, while today geodesy serves all Earth science, including the geophysical, oceanographic, atmospheric, and environmental science communities. Geodesy is also indispensable for the maintenance of many activities in a modern society. Traditionally, geodesy has served society by providing reference frames for a wide range of practical applications from regional to global navigation on land, sea, and in air, construction of infrastructure, to the determination of reliable boundaries of real-estate properties. Reference frames were, however, national or regional in scope, and they were suited for the determination of coordinates relative to a network of reference points. Thus, determination of precise point coordinates required simultaneous measurements at several points. Today, the Global Navigation Satellite Systems (GNSS) provide access to precise point coordinates in a global reference frame anytime and anywhere on the Earth’s surface with centimeter-level accuracy and without requiring additional measurements on nearby reference points. On the user side, this technological development has stimulated new applications demanding even greater accuracy and better access to geodetically determined positions. On local to regional scales, applications such as land surveying, monitoring of infrastructure, prevention and mitigation of impacts of environmental hazards, and numerous technical applications require more or less instantaneous access to geodetic positions in a reliable reference frame with centimeter-level accuracy or better.27 New developments and applications will lead to increasing dependence on the geodetic foundation, that is, the terrestrial geodetic reference frame including easy access to this frame in the form of accurate positions. Geodesy has the potential to make very important contributions to the understanding of the state and dynamics of System Earth, particularly if consistent observations of the “three pillars” can be provided on a global scale with a precision at or below the 1 ppb level, and with suﬃcient stability over decades. A prerequisite to exploiting the full potential of geodesy for Earth observation, Earth system monitoring, and many practical applications is a sophisticated integration of all geodetic techniques (spaceborne, airborne, marine, and terrestrial), processing models, and geophysical background models into one system. The integration of the “three pillars” will permit — as part of global change research — the

FA

December 10, 2008 15:45

114

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch08

H.-P. Plag et al.

assessment of surface deformation processes and the quantiﬁcation of mass anomalies and mass transport inside individual components (such as ocean, cryosphere, terrestrial hydrosphere, and atmosphere), and mass exchange between the components of the Earth’s system. These quantities serve as input to the study of the physics of the solid Earth, ice sheets and glaciers, hydrosphere, and atmosphere. They are of particular value for the study of complex phenomena such as glacial isostatic adjustment, the evolution of tectonic stress patterns, sea-level rise and fall, the hydrological cycle, transport processes in the oceans, and the dynamics and physics of the atmosphere (troposphere and ionosphere).

3. The Global Geodetic Observing System: The Organization and the System The international cooperation fostered by the IAG has led to the establishment of the IAG Services that provide increasingly valuable observations and products not only to scientists but also for a wide range of non-scientiﬁc applications (Table 2). With the recent developments in geodesy, Earth observations, and societal needs in mind, the IAG has established the Global Geodetic Observing System (GGOS) as the observing system of IAG. After a preparatory phase which included the IAG Symposium on Integrated Global Geodetic and Geodynamic Observing System in Munich in 1998,28 GGOS was initially created as an IAG Project during the IUGG meeting in 2003 in Sapporo, Japan. After devoting the ﬁrst two years to the deﬁnition of the internal organizational structure of GGOS and its relationship with external organizations (the “Design Phase”), the Executive Committee of the IAG at its meetings in August 2005 in Cairns, Australia, decided to continue the Project. In the “Implementation Phase” from 2005 to 2007, the GGOS Steering Committee, Executive Committee, Science Panel, Working Groups, and Web Pages were established, and the Terms of Reference were revised. Finally, at the IUGG meeting in 2007 in Perugia, Italy, IAG elevated GGOS to the status of a full component of IAG as the permanent observing system of IAG. It is important to note here that “GGOS” has two very distinct aspects, which should not be confused: (1) the “organization GGOS” consisting of components such as committees, panels, working groups, etc., and (2) the “observation system GGOS” comprising the infrastructure of many diﬀerent instrument types, satellite missions, and data and

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch08

The Global Geodetic Observing System

115

Table 2. The IAG Services. The IAG Services are the backbone of GGOS, providing the infrastructure for observations, data archiving and processing, and generation of products. Pillar

Acronym

Systems, Frames, Rotation

IERS

Geometry Solid Earth

IGS

IVS

ILRS IDS

Geometry Ocean PSMSL and Ice IAS

Gravimetry

IGFS

BGI IGeS ICET Standards

BIPM IBS

Service

Technique(s)

International Earth Rotation and Reference Systems Service International GNSS Service

Combination of all geometric techniques

International VLBI Service for Geodesy and Astrometry International Laser Ranging Service

Very Long Baseline Interferometry (VLBI)

International Doris Service

Doppler Orbitography and Radiopositioning Integrated by Satellites

Permanent Service of Mean Sea Level International Altimetry Service (in preparation) International Gravity Field Service

Tide gauges

Global Navigation Satellite Systems (GNSS), particularly Global Positioning System (GPS), Global Navigation Satellite System (GLONASS), and in future Galileo

Satellite Laser Ranging (SLR), Lunar Laser Ranging (LLR)

Satellite altimetry

Gravimetric in situ (absolute and relative), airborne, and spaceborne techniques

Bureau Gravimetric Gravimetric techniques International International Geoid Gravimetric techniques Service International Center for Earth tide gravimeters Earth Tide Bureau International des Poids et Measures IAG Bibliographic Service

analysis centers. While GGOS as an organization has established, and is extending, its structure from essentially new entities, the observational infrastructure for GGOS as the system is being largely provided by the IAG Services.

FA

December 10, 2008 15:45

116

AOGS 2007 - SE Volume

H.-P. Plag et al.

9in x 6in b672-V13-ch08

Fig. 3. Organizational links and relationships of GGOS. GGOS is being built on the scientiﬁc support from the IAG Commissions and the infrastructure of the IAG Services. GGOS integrates the work of the Services through a number of GGOS Working Groups and provides coordination and advice through its Committees. GGOS links these entities to the main programs in Earth observations, and provides a unique interface for GGOS users to the geodetic services. (Modiﬁed from Ref. 5).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The Global Geodetic Observing System

b672-V13-ch08

117

GGOS as an organization is a unifying umbrella for the IAG Services and an interface between the Services and the “outside world” (Fig. 3). Internally, the GGOS Committees, Science Panel, and Working Groups focus on cross-cutting issues relevant for all Services. By combining the “three pillars” into one observing system having utmost accuracy and operating in a well-deﬁned and reproducible global terrestrial frame, GGOS adds to these pillars a new quality and dimension in the context of Earth system research. The observing system, in order to meet its objectives, has to combine the highest measurement precision with spatial and temporal consistency and stability that are maintained over decades. The research needed to achieve these goals inﬂuences the agenda of the IAG Commissions and the GGOS Working Groups. Externally, GGOS provides the links between the IAG Services and the main programs in Earth observations and Earth science. It constitutes a unique interface for many (although not all) users to the geodetic Services. GGOS participates on behalf of IAG in large international programs focusing on Earth observations, in particular, the Group on Earth Observations (GEO, see below). According to the IAG By-Laws, GGOS works with the IAG Services and Commissions to provide the geodetic infrastructure necessary for the monitoring of the Earth system and global change research. This statement implies a vision and a mission for GGOS. The implicit vision for GGOS is to empower Earth science to extend our knowledge and understanding of the Earth system processes, to monitor ongoing changes, and to increase our capability to predict the future behavior of the Earth system. Likewise, the embedded mission is to facilitate networking among the IAG Services and Commissions and other stakeholders in the Earth science and Earth Observation communities, to provide scientiﬁc advice and coordination that will enable the IAG Services to develop products with higher accuracy and consistency, meeting the requirements of particularly global change research, and to improve the accessibility of geodetic observations and products for a wide range of users. The IAG Services, upon which GGOS is built, beneﬁt from GGOS as a framework for communication, coordination, and scientiﬁc advice necessary to develop improved or new products with increased accuracy, consistency, resolution, and stability. IAG beneﬁts from GGOS as an agent to improved visibility of geodesy’s contribution to the Earth sciences and to society in general. The users, including the national members of IAG, beneﬁt from GGOS as a single interface to the global geodetic observation system of systems maintained by the IAG Services

FA

December 10, 2008 15:45

118

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch08

H.-P. Plag et al.

not only for the access to products but also to voice their needs. Society beneﬁts from GGOS as a utility supporting Earth science and global Earth observation systems as a basis for informed decisions. As an organization, GGOS is challenged by the recent developments in global Earth observation. The Ten-Year Implementation Plan (TYIP) for the Global Earth Observation System of Systems (GEOSS), which was prepared by the Group on Earth Observation (GEO) between 2003 and 2005, and endorsed by the Earth Observation Summit III (EOS-III) in 200529 is likely to guide the development of global Earth observation programs over the next decade. GGOS as an organization needs to be integrated appropriately into the context of Earth observation and society, and GGOS as an observing system has to be developed in accordance with the strategies and methodologies of the global observing systems for the mutual beneﬁt of all. Earth observation and society at large will beneﬁt from the availability of geodetic observations and products, and GGOS will beneﬁt from an improved visibility and acknowledgment of the valuable service it provides. In order to facilitate the integration of GGOS into GEOSS, IAG is a Participating Organization in GEO (since 2004) and is represented there by the GGOS organization. GGOS is also a contributing system to the GEOSS, which is implemented by GEO. In the frame of GEO, GGOS carried out a strategy process (denoted as GGOS 2020) with the goals (1) to establish the relevant user requirements across the nine Societal Beneﬁt Areas (SBAs) of GEO (for a list of these SBAs, see Ref. 3), and (2) to provide the basis for the implementation of a geodetic observing system that will meet the requirements of the society at large and the SBAs of GEO in particular.30 Since 2006, GGOS is a member of the Integrated Global Observing Strategy Partnership (IGOS-P)31 and is integrating its work into the Integrated Global Observing Strategy (IGOS) which is an initiative of IGOS-P. Moreover, GGOS contributes to a number of global observing systems; steps are being taken to strengthen joint initiatives with government organizations and international bodies, including relevant United Nations authorities. These initiatives have already and will continue to enhance the visibility of geodetic activities in the context of Earth sciences, Earth observation, and practical applications.27 GGOS as an observing system is built upon the existing and future infrastructure provided by the IAG Services. It aims to provide consistent observations of the spatial and temporal changes of the shape and gravitational ﬁeld of the Earth, as well as the temporal variations of the

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The Global Geodetic Observing System

b672-V13-ch08

119

Earth’s rotation (see Fig. 1). In other words, it aims to deliver a global picture of the surface kinematics of our planet, including the ocean, ice cover, and land surfaces. In addition, it aims to deliver estimates of mass anomalies, mass transport, and mass exchange in the Earth system. Surface kinematics and mass transport together are the key to global mass balance determination, and an important contribution to the understanding of the energy budget of our planet.32 –34 Moreover, the system aims to provide the observations that are needed to determine and maintain a terrestrial reference frame of higher accuracy and greater temporal stability than what is available today.35 GGOS as a system (Fig. 4) exploits (and tries to extend) for this purpose the unique constellation of satellite missions relevant to this goal that are in orbit now and those planned for the next two decades, by integrating them into one measurement system. The backbone of this integration is the existing global ground network of tracking stations for the geodetic space techniques, Very Long Baseline Interferometry (VLBI), Satellite Laser Ranging (SLR), Lunar Laser Ranging (LLR), GNSS, and Doppler Orbitography and Radio Positioning Integrated by Satellite (DORIS). GGOS integrates these tracking networks with terrestrial gravity networks. GGOS will complement the space segment and global ground network by airborne and terrestrial campaigns that serve the purpose of calibration and validation, regional densiﬁcation, and reﬁnement. Assimilation of these observations into models of weather, climate, oceans, hydrology, ice, and solid Earth processes will fundamentally enhance the understanding of the role of surface changes and mass transport in the dynamics of our planet. Furthermore, through the analysis of the dense web of microwave radiation connecting the GNSS satellites with Low Earth Orbiters (LEO) and with the Earth’s surface, a powerful new technique emerges for probing the atmosphere’s composition. GGOS (the observing system) faces two types of scientiﬁc and technological challenges, namely an “internal” and an “external” challenge: The “internal” challenge to geodesy is to develop GGOS and the geodetic technologies so that they meet the demanding user requirements in terms of reference frame accuracy and availability, as well as in terms of spatial and temporal resolution and accuracy of the geodetic observations. Developing an observing system capable of measuring variations in the Earth’s shape, gravity ﬁeld, and rotation with an accuracy and consistency of 0.1–1 ppb, with high spatial and temporal resolution, and increasingly low time latency, is a very demanding task. Accommodating the transition of new

FA

December 10, 2008 15:45

120

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch08

H.-P. Plag et al.

Fig. 4. Infrastructure contributing to GGOS. The combined infrastructure allows the determination and maintenance of the global geodetic reference frames, and the determination of Earth’s gravity ﬁeld and rotation. The ground networks and navigation satellites (currently in particular GPS) are crucial for maintaining the reference frame required for high accuracy positioning. In particular, they allow the monitoring of volcanoes, earthquakes, tectonically active regions, and landslide-prone areas. The Low Earth Orbit (LEO) satellites monitor sea level, ice sheets, water storage on land, atmospheric water content, high-resolution surface motion, and variations in the Earth’s gravity ﬁeld. The latter are caused, to a large extent, by regional and global mass transport in the hydrological cycle.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch08

The Global Geodetic Observing System

121

technologies as they evolve in parallel to maintaining an operational system is part of this challenge. The “external” challenge is associated with the integration of the “three pillars” into a system providing information on mass transport, surface deformations, and dynamics of the Earth. The Earth system is a complex system with physical, chemical, and biological processes interacting on spatial scales from micrometers to global and temporal scales from seconds to billions of years. Therefore, addressing the “external” challenge requires a “whole Earth” approach harnessing the expertise of all ﬁelds of Earth science.

4. GGOS: An Observing System of Layered Infrastructure GGOS as an observing system has ﬁve major levels of instrumentation and objects (Fig. 5) that actively perform observations, are passively observed, or both. These levels are: Level 1: the terrestrial geodetic infrastructure including the ground networks of in situ instruments and space-geodetic tracking stations, as well as the data and analysis centers; Level 2: the LEO satellite missions; Level 3: the Middle/Geostationary Earth Orbiters (MEO/GEO) that is, the GNSS and the Lageos-type SLR satellites; Level 4: the planetary missions and geodetic infrastructure on Moon and planets; Level 5: the extragalactic objects. The ground networks and the GNSS are crucial in positioning. Highly accurate orbits for the LEO satellites are determined with the help of ground-based infrastructure as well as GNSS satellites. Level 4 is particularly important for the dynamical reference frame. The stable quasars of Level 5 provide the inertial reference frame ﬁxed in space. These ﬁve levels of instrumentation and objects, independent of whether they are active or passive, receivers or emitters or both, are connected by many types of observations in a rather complex way to form the integrated GGOS observing system. In this system, the major observation types at present are: (1) observations of the microwaves emitted by GNSS satellites at the ground and at the LEO satellites; (2) laser ranging to LEOs, dedicated laser ranging satellites, GNSS satellites, and the Moon; (3) microwave observation of extragalactical objects (quasars) by VLBI; (4) instrumentation onboard the LEO satellites measuring accelerations, gravity gradients, satellite orientation, etc.; (5) radar and optical observations of the Earth’s surface (land, ice, glaciers, sea level, etc.) from remote sensing satellite; (6) distance measurements between satellites

FA

December 10, 2008 15:45

122

AOGS 2007 - SE Volume

H.-P. Plag et al.

9in x 6in b672-V13-ch08

Fig. 5. The ﬁve levels of GGOS and their interactions with observations of various types. The infrastructure of GGOS consists of ﬁve distinct levels, depending on the distance to the Earth’s surface. See text for details.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The Global Geodetic Observing System

b672-V13-ch08

123

(K-band, optical, interferometry, etc.). In the future, new measurement techniques will evolve and be included into the system. Figure 5 also makes clear that diﬀerent parts of the overall system are cross-linked through observations and are interdependent. Moreover, all these techniques are aﬀected by the same Earth system processes and they measure the “output” of the same unique Earth system, that is, the various geodetic ﬁngerprints induced by mass redistribution and changes in the system’s dynamics (see Fig. 2). Therefore, consistency of data processing, modeling, and conventions across the techniques and across the “three pillars” is mandatory for maximum exploitation of the full potential of the system. In order to achieve this, GGOS pursues two main roads: (1) The individual parts (i.e. observation types) of the overall system are connected by co-location of diﬀerent instruments at the same site on the Earth’s surface or on the same satellite or object. This co-location of instruments and sensors is extremely important for the consistency and accuracy of the system and for the integration of the system to perform as one large “instrument.” Moreover, each of the techniques has its own strength and weaknesses, and through co-location, the strengths can be exploited and the weaknesses can be mitigated. (2) The diﬀerent parts of the Earth system are increasingly linked through consistent and comprehensive models, in order to capture all interactions and feedback between the various components of the Earth system as depicted in Fig. 2, leading eventually to an integrated Earth system model predicting the geodetic ﬁngerprints in the “three pillars” consistently as a basis for improved geodetic analyses. Enhanced modeling will also allow more realistic estimates of the uncertainties of individual observations and products and thus improve the integrability of the diﬀerent geodetic datasets. GGOS is one of the systems contributing to GEOSS. This requires interoperability of those products and services provided by GGOS to GEOSS with the other systems of GEOSS. While there has been an increasing focus on interoperability between the various IAG Services, achieving interoperability with GEOSS components is still an emerging challenge with the extent not yet fully identiﬁed.

5. Conclusions GGOS is IAG’s Observing System. As an organization, it is the main interface for the IAG Services particularly to the major international Earth observation programs, including GEO. As an organization, GGOS

FA

December 10, 2008 15:45

124

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch08

H.-P. Plag et al.

enhances the visibility of geodesy in general and the work of the IAG Services in particular in society and speciﬁcally the relevant scientiﬁc organizations and Earth observation system. GGOS links the IAG Services to major user groups. The organization has the main tasks to: (1) identify a comprehensive set of geodetic products and establish the requirements concerning the products’ accuracy, temporal and spatial resolution, latency, and consistency; (2) develop the strategy for GGOS, appropriate to meet these requirements; (3) identify the gaps in the system of IAG Services and develop strategies to close these gaps; (4) ensure the availability, consistency, reliability, accessibility, and interoperability of geodetic observations, products, and models. As an observing system, GGOS builds on the infrastructure of the IAG Services and utilizes signals and observations from relevant infrastructure maintained by governmental authorities and space agencies (e.g. NASA and ESA). Based on this infrastructure, GGOS allows the monitoring of: (1) the deformation of the Earth surface (solid Earth, ocean, and ice) and Earth rotation with submillimeter accuracy; (2) the global gravity ﬁeld and its time variations with unprecedented accuracy and resolution, particularly through dedicated gravity satellite missions; (3) the water vapor in the troposphere, tropopause height, and electron density in the ionosphere, thus monitoring the atmospheric processes relevant for global warming; (4) many parameters related to natural hazards and disasters, increasingly providing observations relevant for early warning systems. Since GGOS depends to a degree on infrastructure (particularly the satellite missions) provided by others, GGOS has to interact with these providers (particularly the space agencies) in order to ensure continuity of the required infrastructure and to work with them to improve the infrastructure continuously. In order to integrate the infrastructure of the IAG Services into a comprehensive and integrated geodetic Earth observation system of systems with highly accurate and consistent observations and products, GGOS faces main challenges in the combination and integration of all observation techniques into a consistent observing system, and in developing comprehensive modeling of the interactions in the Earth system across the “three pillars.” Already today, geodesy and particularly GGOS has the documented potential to facilitate new insights into the geophysical processes in the Earth system. If the challenges can be met, GGOS will provide a basis for a deeper scientiﬁc understanding of the Earth System and the future of our changing planet, as well as an improved basis for Earth observations and geodesy-dependent applications in society at large.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The Global Geodetic Observing System

b672-V13-ch08

125

Acknowledgments The paper has greatly beneﬁted from the work done by many colleagues in the frame of the GGOS 2020 process. The authors would like to thank all those who have contributed to the GGOS 2020 report, particularly the chapter lead authors R. Rummel, D. Sagagian, C. Rizos, J. Zumberge, R. Gross, T. A. Herring, and G. Beutler. The authors are also grateful to two anonymous referees, who provided helpful comments on the original version of the manuscript.

References 1. S. C. Solomon and the Solid Earth Science Working Group, Living on a Restless Planet (NASA, Jet Propulsion Laboratory, Pasadena, California, 2002), also available at http://solidearth.jpl.nasa.gov. 2. B. L. Turner II, W. C. Clark, R. W. Kates, J. F. Richards, J. T. Mathews and W. B. Meyer (eds.), The Earth as Transformed by Human Action: Global and Regional Changes in the Biosphere Over the Past 300 years (University Press, Cambridge, 1990), 713 pp. 3. GEO, Global Earth Observing system of systems GEOSS — 10-year implementation plan reference document — Draft, Tech. Rep. GEO 1000R/ESA SP 1284, ESA Publication Division (ESTEC, Noordwijk, The Netherlands, 2005), avaliable at http://earthobservations.org. 4. B. F. Chao, Eos, Trans. Am. Geophys. Union 84 (2003) 145. 5. H.-P. Plag, National geodetic infrastructure: Current status and future requirements — The example of Norway, Bulletin 112, Nevada Bureau of Mines and Geology, University of Nevada, Reno (2006), 97 pages. 6. R. Rummel, Global integrated geodetic and geodynamic observing system (GIGGOS), in Towards an Integrated Global Geodetic Observing System, eds. R. Rummel, H. Drewes, W. Bosch and H. Hornik, International Association of Geodesy Symposia, Vol. 120 (Springer, Berlin, 2000). 7. K. H. Ilk, J. Flury, R. Rummel, P. Schwintzer, W. Bosch, C. Haas, J. Schr¨ oter, D. Stammer, W. Zahel, H. Miller, R. Dietrich, P. Huybrechts, H. Schmeling, D. Wolf, H. J. G¨ otze, J. Riegger, A. Bardossy, A. G¨ unter and T. Gruber, Mass transport and mass distribution in the Earth system, Tech. Rep., GOCE-Projectb¨ uro Deutschland, Technische Universit¨ at M¨ unchen, GeoForschungsZentrum Potsdam (2005). 8. C. Kreemer and W. E. Holt, Geophys. Res. Lett. 28 (2001) 4407. 9. H. P. Kierulf, L. Bockmann, O. Kristiansen and H.-P. Plag, Foot-print of the space-geodetic observatory, Ny-˚ Alesund, Svalbard, Proc. Second IVS General Meeting, 4–6 February 2002, Tsukuba, Japan, eds. N. Vandenberg and K. Baver (NASA Goddard Space Flight Center, Greenbelt, MD). 10. C. Kreemer, W. E. Holt and A. J. Haines, Geophys. J. Int. 154 (2003) 8.

FA

December 10, 2008 15:45

126

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch08

H.-P. Plag et al.

11. G. Blewitt, D. Lavall´ee, P. Clarke and K. Nurutdinov, Science 294 (2001) 2342. 12. D. Dong, P. Fang, Y. Bock, M. K. Cheng and S. Miyazaki, J. Geophys. Res. 107 (2002) 2075, doi: 10.1029/2001JB000573. 13. X. Wu, M. B. Heflin, E. R. Ivins, D. F. Argus and F. H. Webb, Geophys. Res. Lett. 30 (2003) 1742, doi: 10.1029/2003GL017546. 14. R. S. Gross, G. Blewitt, P. J. Clarke and D. Lavall´ee, Geophys. Res. Lett. 31 (2004), doi: 10.1029/2004GL019589. 15. Z. Lu, C. Wicks, D. Dzurisin, W. Thatcher, J. Freymueller, S. McNutt and D. Mann, Geophys. Res. Lett. 27 (2000) 1567. 16. R. Lanari, G. De Natale, P. Berardino, E. Sansosti, G. P. Ricciardi, S. Borgstrom, P. Capuano, F. Pingue and C. Troise, Geophys. Res. Lett. 29 (2002), doi: 10.1029/2001GL014571. 17. A. Bonforte and G. Puglisi, J. Geophys. Res. 108 (2003) 2153, doi: 10.1029/2002JB001845. 18. A. Ferretti, F. Novali, R. B¨ urgmann, G. Hilley and C. Prati, Eos, Trans. Am. Geophys. Union 85 (2004) 317, 324. 19. P. Banerjee, F. F. Pollitz and R. B¨ urgmann, Science 308 (2005) 1769. 20. C. Vigny, W. J. F. Simons, S. Abu, R. Bamphenyu, C. Satirapod, N. Choosakul, C. Subarya, A. Socquet, K. Omar, H. Z. Abidin and B. A. C. Ambrosius, Nature 436 (2005) 201. 21. C. Kreemer, G. Blewitt, W. C. Hammond and H.-P. Plag, Earth Planets Space 58 (2006) 141. 22. T. Strozzi, L. Tosi, U. Wegm¨ uller, P. Teatini, L. Carbognin and R. Rosselli, Geophys. Res. Lett. 29 (2002) 345. 23. G. Blewitt, C. Kreemer, W. Hammond, H.-P. Plag, S. Stein and E. Okal, Geophys. Res. Lett. 33 (2006) L11309, doi: 10.1029/2006GL026145. 24. D. Jerrett and J. Nash, Phys. Chem. Earth 26 (2001) 457. 25. G. Elgered, H.-P. Plag, H. Marel, S. Barlag and J. Nash, COST Action 716 Exploitation of Ground-based GPS for Operational Numerical Weather Prediction and Climate Applications, COST European cooperation in the field of scientific and technical research, no. EUR 21639 in COST European cooperation in the field of scientific and technical research (European Commission, 2005). 26. P. Poli, P. Moll, F. Rabier, G. Desroziers, B. Chapnik, L. Berre, S. B. Healy, E. Andersson and F.-Z. El Guelai, J. Geophys. Res. 112 (2007) D06114, doi: 10.1029/2006JD007430. 27. H.-P. Plag, GGOS and it user requirements, linkage and outreach, in Dynamic Planet — Monitoring and Understanding a Dynamic Planet with Geodetic and Oceanographic Tools, eds. P. Tregoning and C. Rizos, International Association of Geodesy Symposia, Vol. 130 (Springer Verlag, Berlin, 2006). 28. R. Rummel, H. Drewes, W. Bosch and H. Hornik (eds.), Towards an Integrated Global Geodetic Observing System, International Association of Geodesy Symposia, Vol. 120 (Springer, Berlin, 2000).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The Global Geodetic Observing System

b672-V13-ch08

127

29. GEO, The Global Earth Observing System of Systems (GEOSS) — 10-Year Implementation Plan, available at http://earthobservations.org (2005). 30. H.-P. Plag and M. Pearlman (eds.), The Global Geodetic Observing System: Meeting the Requirements of a Global Society on a Changing Planet in 2020 — The Reference Document (Global Geodetic Observing System, 2007), available at http://geodesy.unr.edu/ggos/ggos2020/. 31. H.-P. Plag, G. Beutler, R. Forsberg, C. Ma, R. Neilan, M. Pearlman, B. Richter and S. Zerbini, Linking the Global Geodetic Observing System (GGOS) to the Integrated Global Observing Strategy Partnership (IGOS-P) through the Theme “Earth System Dynamics”, in Dynamic Planet — Monitoring and Understanding a Dynamic Planet with Geodetic and Oceanographic Tools, eds. P. Tregoning and C. Rizos, International Association of Geodesy Symposia, Vol. 130 (Springer Verlag, Berlin, 2006). 32. R. Rummel, H. Drewes and G. Beutler, Integrated Global Geodetic Observing System (IGGOS): A candidate IAG project, in Vistas for Geodesy in the New Millennium, eds. J. Adam and K.-P. Schwarx, International Association of Geodesy Symposia, Vol. 125 (Springer, Berlin, 2002). 33. R. Rummel, M. Rothacher and G. Beutler, J. Geodynamics 40 (2005) 357. 34. H. Drewes, The science rationale of the Global Geodetic Observing System GGOS, in Dynamic Planet — Monitoring and Understanding a Dynamic Planet with Geodetic and Oceanographic Tools, eds. P. Tregoning and C. Rizos, International Association of Geodesy Symposia, Vol. 130 (Springer Verlag, Berlin, 2006). 35. G. Beutler, H. Drewes, H.-P. Plag, C. Reigber, M. Rothacher and R. Rummel, IAG GGOS Implementation Plan, Tech. Rep., GeoForschungZentrum Potsdam, Germany (2005).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

This page intentionally left blank

b672-V13-ch08

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch09

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

THE INTERNATIONAL LASER RANGING SERVICE MICHAEL PEARLMAN Harvard-Smithsonian Center for Astrophysics 60 Garden St., Cambridge, MA 02138, USA CAREY NOLL NASA Goddard Space Flight Center, Code 690 Greenbelt, MD 20771, USA [email protected] JAN MCGARRY NASA Goddard Space Flight Center, Code 694 Greenbelt, MD 20771, USA WERNER GURTNER Astronomical Institute of Bern, Sidlerstrasse 5 CH-3012 Berne, Switzerland ERRICOS PAVLIS Joint Center for Earth Systems Technology University of Maryland, Baltimore County 1000 Hilltop Circle Baltimore, MD 21250, USA

The International Laser Ranging Service (ILRS) was established in September 1998 as a service within the IAG to support programs in geodetic, geophysical, and lunar research activities and to provide data products to the International Earth Rotation and Reference Frame Service (IERS) in support of its prime objectives. The ILRS develops the standards and speciﬁcations necessary for product consistency and the priorities and tracking strategies required to maximize network eﬃciency. This network consists of more than 30 Satellite Laser Ranging (SLR) stations, routinely tracking nearly 30 retroreﬂectorequipped satellites and the Moon in support of user needs. The Service collects, merges, analyzes, archives, and distributes satellite and lunar laser ranging data to satisfy a variety of scientiﬁc, engineering, and operational needs and encourages the application of new technologies to enhance the quality, quantity, and cost eﬀectiveness of its data products. The ILRS works with the global network to improve station performance, new satellite missions in the design and building of retroreﬂector targets to maximize data quality and quantity, and science programs to optimize scientiﬁc data yield. The ILRS Central Bureau maintains a comprehensive web site (http://ilrs.gsfc.nasa.gov) as the primary vehicle for the distribution of information within the ILRS community. 129

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch09

M. Pearlman et al.

130

During the last few years, the ILRS has addressed very important challenges. Several new SLR stations have enhanced the global coverage of the network, particularly in the Southern Hemisphere. Two Combination Centers have been established to provide Oﬃcial ILRS Data Products to the IERS for the process of maintaining the ITRF (International Terrestrial Reference Frame), and seven ILRS Analysis Centers have been established to provide the solutions for the combination process. Procedures have been introduced to track vulnerable satellites, such as ICESat and ALOS. Work continues at several institutions on the ground system development toward the next generation laser ranging system with the introduction of kilohertz pulse repetition rates, greater automation, and greater eye safety. Work also continues on the development of new lighter weight retroreﬂector arrays.

1. ILRS Organization The ILRS coordinates the activities of the international laser ranging community from data acquisition through the delivery of products to the users. The ILRS is organized into Tracking Stations and Subnetworks, Operations Centers, Data Centers, Analysis and Associate Analysis Centers, a Central Bureau, Working Groups, and a Governing Board (see Fig. 1, Refs. 1 and 2).

Fig. 1.

ILRS organization.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The International Laser Ranging Service 131

b672-V13-ch09

ILRS Tracking Network (as of July 2007). Fig. 2.

FA

December 10, 2008 15:45

132

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch09

M. Pearlman et al.

Stations in the ILRS Tracking Network range to the approved constellation of artiﬁcial satellites and the Moon with state-of-the-art laser ranging systems and transmit their data in near-real time to the ILRS Data Centers. The full network currently consists of more than 30 SLR stations, as shown in Fig. 2. For the most part, stations are designed by the local participating agencies using fairly standard components. Some examples of stations are shown in Fig. 3. The majority of ILRS stations have a co-located GNSS receiver that adheres to standards established by the International GNSS Service (IGS). Some are also co-located with VLBI, DORIS, and absolute gravity instrumentation. The ILRS has given strong encouragement to the development of Fundamental Reference Stations, where combination of several space geodetic techniques including SLR, VLBI, GNSS, DORIS,

Fig. 3. Selected ILRS network stations (clockwise from top left, Zimmerwald Switzerland; Greenbelt, USA; Maui, USA; Matera, Italy).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The International Laser Ranging Service

b672-V13-ch09

133

and absolute gravimetry are co-located to strengthen the reference system constraints and system synergy. The current network is very far from an “optimal” one and there are currently studies underway to design an improved “space geodetic” network of ﬁducial observatories where SLR and all other primary space techniques will be present. The void over most of Africa and Asia has major impact on the quality of the orbits determined with SLR and by implication, the geodetic products based on them, be it daily resolution EOP, gravitational variations, tidal parameters, etc. This lack of sites over vast expanses has also severe repercussions on the deﬁnition of the ITRF, with the deﬁnition of the origin and scale being the most important. A similar situation rises in the Paciﬁc area where in addition to the limited availability of SLR systems, we also face the lack of suitable landmass to locate any available systems. Yet, this area is very important in terms of providing ground control for precise orbits, since it is the largest oceanic area over which the oceanographic missions carrying radar altimeters require very high accuracy orbits for the study of secular changes in the mean sea level. The oceanographically complicated area surrounding Indonesia and the northern coast of Australia would beneﬁt tremendously from the future addition of an SLR system somewhere in the neighborhood of Darwin. ILRS Operations Centers collect and merge the data from the tracking sites, provide initial quality checks on these incoming data, reformat and compress the data if necessary, and relay the data to an ILRS Data Center. Two Global Data Centers archive all the ranging data and auxiliary data (e.g. station log ﬁles and satellite orbit predictions), make the data available to the ILRS Analysis Centers and external users of the data, and act as distribution centers for the primary ILRS products. The Analysis and Associate Analysis Centers routinely generate the oﬃcial ILRS products (station coordinates at one-week intervals, Earth Orientation Parameters at one-day intervals) as well as special products, such as satellite predictions, time-bias information, precise orbits for special-purpose satellites, or scientiﬁc data products of a mission-speciﬁc nature. Lunar Analysis Centers process ranging data to the targets on the Moon and produce lunar-speciﬁc data products. There are currently about 30 ILRS Analysis and Associate Centers. The Central Bureau (CB) is responsible for the daily coordination and management of the ILRS. The CB maintains the ILRS web site, a

FA

December 10, 2008 15:45

134

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch09

M. Pearlman et al.

source for all SLR- and LLR-related information including details about the organization and operation of ILRS, and also an entry point to the data and products stored at the Data Centers. The CB maintains the ILRS documentation, organizes meetings and workshops, and issues service reports. The ILRS has four standing and one ad-hoc Working Groups that provide the expertise necessary to make technical decisions, to plan programmatic courses of action and are responsible for reviewing and approving the content of technical and scientiﬁc databases maintained by the Central Bureau (see Fig. 1). The ILRS Governing Board (GB) is responsible for the general direction of the service and deﬁnes oﬃcial ILRS policy and products, determines satellite-tracking priorities, and develops standards and procedures. The 16member body interacts with other services and organizations and is selected from ILRS associates representing all components of the service.

2. Scientific Applications of Laser Ranging Laser ranging measurements provide a long-term stable time history of station positions and precision orbit determination.3 Data products support maintenance of the Terrestrial Reference Frame (Earth center of mass and scale), plate tectonics and crustal deformation, static and time-varying gravity ﬁeld, Earth Orientation (polar motion and length of day), precision orbit determination and calibration of altimetry missions, mass distribution studies, relativity, and lunar and space science.

2.1. The global terrestrial reference frame The terrestrial reference system is the basis through which we connect and compare measurements over space, time, and evolving technologies. It is the means by which we know that measured change over time is real and not corrupted with instabilities in our measurement system. One of the best-known scientiﬁc realizations of a global terrestrial reference system is the ITRF,4 updated every few years by the IERS. It is based on contributions from the four diﬀerent space geodetic techniques (SLR, VLBI, GNSS, and DORIS) consisting of solutions for the positions and velocities of all participating tracking stations in an Earth-ﬁxed geocentric coordinate system. The most important contributions of the

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The International Laser Ranging Service

b672-V13-ch09

135

laser ranging technique to the reference frame are the ﬁxing of its origin (deﬁned with respect to the center-of-mass of the Earth, including oceans and atmosphere) and its scale (deﬁned by the speed of light, realized mainly through the measurement of the time of propagation, i.e. the ranges to satellites).5 Origin and scale are crucial elements, not only for “classical” referencing purposes (i.e. crustal deformation studies), but also as providers of an absolute reference for investigations on sea level change, ice budget, etc. ILRS contributions come either as multi-year solutions based on ranges to the geodynamic satellites LAGEOS-1 and -2 or, within the current IERS Combination Pilot Project, as time series of weekly solutions.

2.2. The Earth in space The connection between the terrestrial and the celestial reference systems is given by the current position of the Earth’s body with respect to its current axis of rotation (polar motion), its angular rotation about its axis (UT1), or the temporal change of this angle (angular velocity, length-of-day), and the orientation of the axis of rotation in space (precession and nutation). One of the oﬃcial ILRS products is the time series of polar motion and length-of-day, submitted weekly to the IERS for combination with the time series generated by the other space geodetic techniques.6,7

2.3. The Earth’s gravity field Global gravity ﬁeld models are based on a combination of spaceborne and ground-based measurements.8 Dedicated gravity satellite missions including satellite-to-satellite tracking in-orbit observation of derivatives of the gravity ﬁeld and ocean-surface altimetry have provided high spatial and temporal resolution of the gravity ﬁeld. The lower-degree terms of this ﬁeld and their secular changes (e.g. of the dynamic ﬂattening of the Earth) are most accurately observed by laser ranging. In this way, SLR provides extremely valuable information on (changes in) the overall mass distribution of Earth.9,10

2.4. Precise orbit determination and verification SLR provides routine precise orbit determination for some missions and veriﬁcation and calibration of precise orbits determined with other tracking

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

136

9in x 6in

b672-V13-ch09

M. Pearlman et al.

techniques such as GNSS or DORIS for others.11 The high accuracy and unambiguous nature of SLR data makes it an independent source of quality control and calibration for other tracking techniques. In particular, SLR has been used in all of the recent ocean and ice topography missions to support altimeter measurements and for a number of special engineering missions (e.g. altimeter calibration).12 SLR, with its totally passive spaceborne reﬂectors, also acts as a backup for active tracking techniques. It has saved satellite missions (ERS-1, GFO-1, TOPEX/Poseidon, and Meteor-3M) after the failure of the primary tracking system. The ILRS continues to encourage new missions with high precision orbit requirements to include retroreﬂectors as a fail-safe backup tracking system, to improve or strengthen overall orbit precision, and to provide important intercomparison and calibration data with onboard microwave navigation systems.

2.5. Lunar ranging The two ILRS stations currently capable of routinely tracking the four lunar targets have a long history of providing LLR data: the McDonald Observatory in Texas has been in operation since the Apollo 12 mission (since 1985 with the current system), whereas the Grasse Observatory in France started lunar laser ranging in 1987. Several stations have demonstrated lunar capability while others have tracked the Moon for some periods of time (e.g. Maui, Hawaii, 1984–1990). A number of stations (e.g. Matera, Italy and Mount Stromlo, Australia) are planning to include lunar tracking in their future activities. Applications in gravitational physics include: testing of the Equivalence Principle; (limits for) timevariation of the gravity constant G; and the assessment of the geodetic precession.13 Applications in lunar science include the determination or the improvement of lunar ephemerides and rotation; dissipation-caused (negative) acceleration; and an assessment of the interior and the Lunar Love numbers.14

2.6. Role within other activities In early 2004, under its new reorganization, the International Association of Geodesy (IAG) established the Global Geodetic Observatory System

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The International Laser Ranging Service

b672-V13-ch09

137

(GGOS) project to coordinate geodetic research in support of scientiﬁc and applications disciplines. GGOS is intended to integrate diﬀerent geodetic techniques, models and approaches to provide better consistency, long-term reliability, and understanding of geodetic, geodynamic, and global change processes. Through the IAG’s measurement services (IGS, IVS, ILRS, IDS, and IGFS), GGOS will work to ensure the robustness of the three aspects of geodesy: geometry and kinematics, Earth orientation and rotation, and static and time-varying gravity ﬁeld. GGOS will identify geodetic products and establish requirements on accuracy, time resolution, and consistency. The project will work to coordinate an integrated global geodetic network and implement compatible standards, models, and parameters. A fundamental aspect of GGOS is the establishment of a global network of stations with co-located techniques, working together to deliver the most accurate and robust reference frame. The ILRS is one of the service participants in GGOS, bringing its unique strengths to the geodetic complex.

3. Satellites Tracked by ILRS Since its inception, SLR has tracked more than 50 satellites with retroreﬂectors. Currently, the ILRS tracks 25 satellites for geodynamics, remote sensing (altimeter, SAR, etc.), gravity ﬁeld determination, general relativity, validation of global navigation systems satellite orbits, and engineering tests (see Fig. 4). Altitudes range from a few hundreds of kilometers to geosynchronous altitudes (36,000 km). Two stations, Grasse, France and McDonald, USA, routinely range to four targets on the Moon. Satellites are added and deleted from the ILRS tracking roster as the GB approves new observation campaigns and older ones are completed. The ILRS assigns satellite-tracking priorities in an attempt to maximize data yield on the full satellite complex while at the same time placing greatest emphasis on the most immediate data needs. Nominally, tracking priorities decrease with increasing orbital altitude and increasing orbital inclination (at a given altitude). Priorities of some satellites are then increased to intensify support for active missions (such as altimetry), special campaigns, and post-launch intensive tracking campaigns. New missions added during the past year are ALOS, ANDE-RR, TerraSAR-X, GLONASS-102, ETS-8, and GIOVE-A. Missions scheduled for ILRS tracking support over the next year include GIOVE-B, PROBA-2,

FA

AOGS 2007 - SE Volume

M. Pearlman et al.

9in x 6in

Examples of satellites tracked by the ILRS network: LAGEOS-1, Jason-1, and GIOVE-A (photo courtesy of ESA).

December 10, 2008 15:45

138

Fig. 4.

b672-V13-ch09 FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The International Laser Ranging Service

Fig. 5.

b672-V13-ch09

139

Schematic of Reﬂector satellite.

Jason-2, LRO-LR, and so on. It is anticipated that the entire 30-satellite Galileo complex, to be launched by 2013 will require, at least intermittent, SLR tracking. The ILRS supports space engineering studies on some rather unique missions. The Russian Reﬂector satellite included retroreﬂectors over its nearly 1 12 m length (Fig. 5). Diﬀerences in the laser return time-of-arrival were used to interpret the orientation and dynamics of the satellite (Fig. 6). Another mission, the Naval Research Laboratory’s Tether Physics and Survivability satellite (TiPS) with retroreﬂector arrays on two satellites separated by a four-kilometer tether was tracked by SLR to study tether dynamics in space.

4. Network Performance Satellite Laser Ranging uses short pulse lasers and fast timing electronics to measure the two-way range from a ground station to retroreﬂector

FA

December 10, 2008 15:45

140

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch09

M. Pearlman et al.

Fig. 6. Range residual pattern on the Reﬂector satellite observed at Yarragadee (courtesy of N. Parkhomenko, RSA).

arrays on satellites and the Moon (Fig. 7). The measurement is the roundtrip travel time corrected for optical refraction and spacecraft centerof-mass. The prime data product from the ILRS stations are normal points, which are full-rate ranging data averaged over time intervals from 15 s to 5 min depending upon the satellite altitude. Absolute accuracy is typically subcentimeter. Data are archived and available to the user on a pass-by-pass basis. The space segment is totally passive, and the refraction model in the optical wavelength region is relatively simple. Laser ranging stations operate during both day and night time, and the data are available to the user in near-real-time through the data centers. Laser ranging provides centimeter orbital accuracy and brings a unique ability to detect small changes through the analysis of long-term stable time series of data. Most of the current laser tracking stations range 10 times per second during part or all of the satellite pass, with many stations interleaving passes from diﬀerent satellites (see Fig. 8). An example of a productive day at Yarragadee is shown in Fig. 9.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch09

The International Laser Ranging Service

Fig. 7.

Fig. 8.

Laser ranging conﬁguration.

Interleaved passes from Zimmerwald Switzerland station.

141

FA

December 10, 2008 15:45

142

AOGS 2007 - SE Volume

M. Pearlman et al.

9in x 6in

b672-V13-ch09

An example of a productive day of tracking at Yarragadee Australia. Fig. 9.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The International Laser Ranging Service

b672-V13-ch09

143

5. Analysis and Data Products The most important aspects of the SLR/LLR observations are absolute accuracy, and long, stable time histories. Accuracy approaches the level of a few mm for modern stations; time histories can be 25–30 years for a few core sites, or even longer for the high-altitude satellites (e.g. LAGEOS). Since its inception, the ILRS put the generation of oﬃcial analysis products high on its agenda. The Analysis Working Group has initiated several pilot projects, to address questions on the oﬃcial products, to deﬁne standards, to reach consensus on product deﬁnition, and ultimately, to arrive at reliable and high-quality operational products.

5.1. “Positioning and EOP” product The ﬁrst-ever Pilot Project focused on the computation of the best-possible ILRS product for station coordinates and Earth Orientation Parameters (EOPs). Various scenarios were deﬁned and tested for establishing the proper satellite mix, means of representing the results, computational strategies, etc. This resulted in the currently operational processing scheme. At this moment, seven diﬀerent Analysis Centers (ACs), (ASI/Italy, BKG/ Germany, DGFI/Germany, GA/Australia, GFZ/Germany, JCET/USA, and NSGF/UK) deliver weekly solutions on LAGEOS-1 and -2 and Etalon-1 and -2 for global station coordinates and EOPs on the Tuesday of each week. These solutions are merged into a combined solution by the two oﬃcial Combination Centers (CCs) at ASI and DGFI. Based on the contributions that the analysis groups made to the Pilot Project and on the evaluation of the quality of their results, the ILRS adopted ASI as its oﬃcial CC, with prime responsibility for the generation of combination solutions for external customers such as IERS. DGFI is the oﬃcial Backup ILRS CC with the same product generation schedule. Several other institutions such as GRGS/OCA, France, and NCL/UK are also well on their way in the development of their contributions to the oﬃcial ILRS product. Once they pass a standardized set of benchmarks in the near future, these groups become additional ACs. As an illustration of the quality of the individual solutions as well as that of the oﬃcial combination product (the quality of the primary and the backup solutions is eﬀectively identical), Table 1 gives a summary of the scatter of weekly solutions for the Cartesian component oﬀsets and

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch09

M. Pearlman et al.

144

Table 1. Scatter of similarity transformation parameters w.r.t. ITRF2000 for successive weekly ILRS solutions for 2006 (oﬀsets in mm, scale in ppb).

Individual ASI BKG DGFI GFZ JCET NSGF Combination ILRS-A

Tx

Ty

Tz

Scale

3.8 4.0 4.7 4.2 3.0 6.1

3.1 1.6 3.9 2.7 2.2 7.3

8.5 2.4 9.0 6.9 7.1 12.0

1.2 0.6 0.8 0.9 0.9 1.4

2.8

2.2

6.5

0.5

global scale from a similarity transformation with respect to ITRF2000. The combination solution gives naturally the best result; however, all of these products indicate the high quality that ILRS ACs and CCs provide. The combination solutions are used for a variety of purposes: the IERS Combination Pilot Project, the IERS/NEOS Bulletin A, etc. At the request of IERS, the ILRS AWG initiated a reprocessing of the older SLR data using the exact standards as currently adopted for the operational product. These products will be ﬁnalized by the end of 2007, and they will serve as inputs for a successor to ITRF2005 and other applications. At this moment, the reprocessing of older data focuses on the 1983–2007 timeframe, and for SLR measurements on LAGEOS-1 and -2, augmented with Etalon-1 and -2 data obtained in the years following 2001. The temporal variations of the weekly solution origin oﬀsets from the long-term mean location realized by ITRF2000 (approximating the “geocenter”) can be seen in Fig. 10. Note the dominant annual signal in the z-component, clearly a manifestation of mass redistribution in the Earth system, related to climate. Similarly, Fig. 11 shows the temporal variation of the scale of the weekly solutions with respect to that of ITRF2000.

5.2. “Benchmarking” process A Benchmarking Pilot Project has been established early on, to provide internal quality checks and quality control over the analysis process. Initially, this was used to scrutinize individual elements of the SLR observations, measurement corrections, and parameter solutions. Having

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch09

The International Laser Ranging Service

145

ILRS-A vs. ITRF2000 From ILRS-A 100

ORIGIN OFFSET Tx, Ty, Tz [mm]

Mean [mm] Std. Dev.[mm]

Tx -0.75 5.05

Ty -0.30 5.12

Tz -7.47 10.70

Tx Ty Tz

50

0

-50

 2006 Copyright JCET, UMBC.

-100 94/01

96/01

98/01

08/01/14

00/01 Date

02/01

04/01

06/01

08/01

Fig. 10. Time series of the origin oﬀsets after a similarity transformation of the weekly ILRS-A product with respect to ITRF2000 for 1993–2006.

ILRS-B vs. ITRF2000 From ILRS-B 10

Mean [ppb] Std. Dev.[ppb]

-0.14 0.68

SCALE

SCALE [ppb]

5

0

-5

 2006 Copyright JCET, UMBC.

-10 94/01 08/01/10

96/01

98/01

00/01 Date

02/01

04/01

06/01

08/01

Fig. 11. Time series of the scale diﬀerences after a similarity transformation of the weekly ILRSB product with respect to ITRF2000 for 1993–2006.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

146

9in x 6in

b672-V13-ch09

M. Pearlman et al.

reached a fully operational status, the Benchmarking Process is now an integral part of the AWG procedures used to assess the quality of new candidate contributors to the ILRS combination products and to identify possible errors.

5.3. New products In addition to the Station Coordinate and EOP Product, the AWG is currently working on the delivery of additional oﬃcial products including satellite orbits and a time series of temporal variations in the origin of the ITRF with respect to the origin of the weekly solution,15 the latter being the closest approximation to the “geocenter.” These products are expected to become routinely available by the end of 2007.

6. Modeling The precision of the travel-time measurement is now of the order of several tens of picoseconds, corresponding to a few millimeters in distance. Ranging accuracy is limited mainly by errors in modeling for refraction propagation and the extrapolation from the reﬂectors to the satellite center-of-mass.

6.1. Refraction Since the launch of LAGEOS-1 in 1976, laser ranging stations have used the Marini and Murray model16 for atmospheric propagation correction. The model works well at higher elevations, but degrades substantially below 20◦ , and ILRS recently (2002) “lowered” the minimum elevation cut-oﬀ to 10◦ above the horizon to improve the decorrelation of measurement biases and station height. A new model available since 200417 provides improved refraction correction at lower elevations, and it has been adopted by ILRS and the IERS as the new standard, along with the companion mapping functions that work for all wavelengths used in SLR today.18 Additional improvements are expected when ILRS implements modeling of horizontal refraction gradients due to circulation in the lower atmosphere. Experiments with 2 years of SLR data have shown a ∼20%–30% reduction in the residual variance when accounting for these errors.19

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The International Laser Ranging Service

b672-V13-ch09

147

6.2. Retroreflector array Early retroreﬂector designs, even those of the LAGEOS era, relied on multi-cornercube returns to maximize return signal strength. Determining the eﬀective reﬂection plane with respect to the satellite center-of-mass is a fundamental correction in SLR metrology. This correction is highly dependent upon return signal strength and the ranging station detector conﬁguration, neither of which were suﬃciently well considered in the early correction models. Errors could be as high as a centimeter, when the entire error budget of a normal point aside from this correction is at or below the millimeter for most sites. With SLR being the principal technique to establish unambiguously the scale of the ITRF, an error of that size amounts to almost 2 ppb, when we are striving for 0.1–0.2 ppb for the future ITRF realizations. This correction has proven to be one of the more signiﬁcant contributors in measurement biases today and there are coordinated eﬀorts such as theoretical studies20 and calibration measurements,21 addressing this error source under the aegis of ILRS. Models are now being implemented that use ground system characteristics to improve the modeled correction to the level better than a single millimeter.

7. Advances Underway A number of advances are currently being implemented that will substantially improve data productivity and quality, while at the same time reduce operational costs. Many of these advances that are now working in one or two stations are envisioned as general characteristics of the future SLR network.

7.1. Automation Stations are implementing increased automation to reduce personnel costs and facilitate data throughput. The new Mount Stromlo Australia station was designed from the beginning with around-the-clock fully automated, unmanned operations. The Zimmerwald laser station operates autonomously for periods of several hours each day. The fully automated NASA Next Generation SLR (NGSLR) system is currently under development.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

148

9in x 6in

b672-V13-ch09

M. Pearlman et al.

7.2. KHz ranging With higher-repetition-rate lasers, faster event timers, and better control software, SLR systems are now able to signiﬁcantly improve the ranging signal-to-noise conditions, while maintaining modest laser characteristics. The new Graz SLR station was the ﬁrst to successfully operate a 2 kHz laser system, increasing the data volume by up to two orders of magnitude.22 The NGSLR prototype is also being developed with this capability, as are upgrades at several other stations. 7.3. New, more powerful stations A number of new systems with large meter-class telescopes and state-of-theart optical and mechanical performance are now operational. This helps to give the tracking network a mix of capabilities to better match the range of targets that now appear on the ILRS roster. The Matera station in southern Italy and the remotely controlled Tanegashima station in Japan both use powerful lasers and large optics to achieve single-shot range precision of a few millimeters. 7.4. Two-color ranging Several groups are using two-wavelength ranging which provides a promising technique for developing better models for the refraction delay imposed by the atmosphere. Two-color ranging at 423 nm and 846 nm has been underway experimentally at Concepci´ on, Chile for some time. The Zimmerwald station began routine operations at these same wavelengths in 2002. Other stations (including GSFC and Matera) have also demonstrated dual-wavelength capability, some of them with superior accuracy using streak cameras. Key to the success of this approach for refraction modeling is the extreme accuracy in measuring the diﬀerential time-of-ﬂight of the two pulses at diﬀerent wavelengths. Present technology does not permit us to utilize this approach now; however, there are major advances in developing the required technology, so we are looking forward to a promising future for the improvement of the refraction models, and hence also the ILRS science products. 7.5. Improved satellite retroreflector array design Multi-cornercube arrays will spread the return pulse. Even with LAGEOS, the return signal is smeared over several centimeters, making measurements

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The International Laser Ranging Service

b672-V13-ch09

149

highly dependent upon signal strength and ground system properties. Most of the recently launched satellites are using standardized arrays with restricted cornercube view. The spherical satellite GFZ-1, Westpac, and LARETS experimented with special reﬂector geometries to limit access to a very few (or single) cubes. Using another approach, the Russian Space Agency has provided the ILRS with a spaceborne test Luneberg Sphere that gives the same array correction for a wide variety of aspect angles.23 An improved version is now being prepared for launch in a sunsynchronous low Earth orbit in 2008. Eﬀorts are also underway to improve the eﬃciency of the arrays for high-orbiting spacecraft to increase the eﬀective cross-section while reducing the payload weight and surface area, both of which are at a premium on the spacecraft. The recently launched ETS-8 synchronous satellite is using a newer design array with uncoated cubes (totally internally reﬂecting) with considerable success. Work is also underway on hollow cube options being proposed for possible arrays for the GPS III series of satellites (Fig. 12). Commercial cubes will undergo performance testing in late 2007 at NASA’s Goddard Space Flight Center and at the Laboratori Nazionali di Frascati (LNF) in Italy in a newly built environmental test chamber.21

7.6. Transponders Optical transponders for extraterrestrial ranging are currently under early development by several research teams. An optical transponder is a combination of a laser-ranging receiver and a separate laser pulse transmitter. As opposed to two-way ranging with retroreﬂector targets,

Fig. 12. Single hollow cube and hollow cube array conﬁguration (drawing courtesy of Armando Morell, NASA Goddard Space Flight Center).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

150

9in x 6in

b672-V13-ch09

M. Pearlman et al.

one-way ranging with a transponder would oﬀer the exciting opportunity of ranging to Mars, planetary moons or orbiters, and deep space missions.24 These transponders will also help to connect the terrestrial reference frame with reference systems used for planetary missions.

7.7. New missions The Ocean Surface Topography Mission (OSTM) will continue the oceanography program begun by the TOPEX/Poseidon and Jason-1 missions. An updated version of the Jason-1 satellite (Jason-2) scheduled launched in June 2008, will use an altimeter and a microwave radiometer to monitor global ocean circulation, study the tie between the oceans and atmosphere, improve global climate predictions, and monitor events such as El Nino conditions and ocean eddies. Jason-2 will also carry the Time Transfer by Laser Link (T2L2) time-transfer experiment. SLR will be used for orbit determination and instrument validation. The Galileo radio navigation system will ultimately have 30 satellites in orbit by 2013. The current mission plan is to have retroreﬂector arrays on all of the satellites for precision orbit determination and validation of the radio system. The ﬁrst system test satellite GIOVE-A is currently in orbit; the second, GIOVE-B, is scheduled for launch in April 2008. The role of these test satellites is to secure the Galileo frequency allocations by providing a signal in space, characterize the on-board clock, and test some of the on-board systems. PROBA-2, the second in a series of technology demonstration missions by the European Space Agency is scheduled for launch in December 2007. The series is intended to provide in-orbit demonstration and evaluation of new hardware and software spacecraft technologies, on-board system operational autonomy, and in particular, Earth observation and space environment instruments. PROBA-B will carry solar observation and plasma measuring instruments in addition to new spacecraft components for testing. GPS and SLR will provide orbit determination. The Lunar Reconnaissance Orbiter (LRO), scheduled for launch in October 2008, will orbit the Moon to measure the ground topography with laser altimetry and to study other lunar characteristics such as gravity ﬁeld, radiation conditions, lighting characteristics, and surface properties in preparation for future human landing. Laser ranging will be used to improve upon the radio tracking accuracy available with C-band. Laser ranging from the Earth will make one-way range measurements to a receiver

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The International Laser Ranging Service

Fig. 13.

b672-V13-ch09

151

One version of the LARES satellite design.

on the satellite; recorded time on the satellite clock will be transmitted to Earth by telemetry for decoding into range. The need for new, well-designed geodetic SLR targets has always been advocated by ILRS. In response to that need, a small group of Italian, US, and German scientists proposed the design and launch of a “mini-LAGEOS” satellite named LARES (after LAser RElativity Satellite).25 The primary goal for this mission is to improve the accuracy of the space-geodetic general relativity test for frame-dragging.7 Although at a lower orbit than LAGEOS, the new design (Fig. 13) aims to improve orbital accuracy with a signiﬁcantly smaller area-to-mass ratio, better optical, mechanical, and thermal properties, and careful calibration and testing prior to the launch (∼2009).

8. Conclusions Laser ranging has proven to be a fundamental component of the spacegeodetic complex, oﬀering a straightforward, conceptually simple, highly accurate, and unambiguous observable. It provides essential contributions to geosciences, space sciences, and fundamental physics. It will play an important role in the GGOS project. Current and future challenges lie in the improvement of the accuracy, reliability, and availability of the data, and in the long-term support of the network. Many of the technological building blocks for the next generation

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

152

9in x 6in

b672-V13-ch09

M. Pearlman et al.

of laser ranging have already been demonstrated. Their comprehensive implementation will bring dramatic improvements to the capability of the technique.

Acknowledgments The authors would like to acknowledge the support of the International Laser Ranging Service and its contributing organizations.

References 1. M. Pearlman, J. Degnan and J. Bosworth, Adv. Space Res. 30(2) (2002) 135–143. 2. W. Gurtner, R. Noomen and M. Pearlman, Adv. Space Res. 36(3) (2005) 327–332. 3. M. Watkins and R. Eanes, Adv. Space Res. 13(11) (1993) 251–255. 4. Z. Altamimi, X. Collilieux, J. Legrand, B. Garayt and C. Boucher, J. Geophys. Res. 112 (2007) B09401, doi: 10.1029/2007JB004949. 5. M. Watkins and R. Eanes, Geophys. Res. Lett. 24(17) (1997) 2231–2234. 6. M. K. Cheng and B. D. Tapley, J. Geophys. Res. 109 (2004) B09402, doi: 10.1029/2004JB003028. 7. I. Ciufolini and E. C. Pavlis, Nature 431 (2004) 958–960. 8. J. Ries, R. Eanes, C. K. Shum and M. Watkins, Geophys. Res. Lett. 19(6) (1992) 529–531. 9. R. Ray, R. Eanes and B. Chao, Nature 381 (1996) 595–597. 10. B. Chao and R. Eanes, Geophys. J. Int. 122 (1995) 755–764. 11. C. Urschl, G. Beutler, W. Gurtner, U. Hugentobler and S. Schaer, Adv. Space Res. 39(10) (2007) 1515–1523. 12. G. Metris, D. Vokrouhlicky, J. Ries and R. Eanes, J. Geophys. Res. 102(B2) (1997) 2711–2729. 13. J. G. Williams, S. G. Turyshev and D. H. Boggs, Phys. Rev. Lett. 93 (2004) 261101, doi: 10.1103/PhysRevLett.93.261101. 14. J. G. Williams, S. G. Turyshev, D. H. Boggs and J. T. Ratcliﬀ, Adv. Space Res. 37(1) (2006) 67–71. 15. E. C. Pavlis, Vistas for geodesy in the new millennium, Proc. 2001 IAG, eds. J. Adam and K.-P. Schwarz (Springer-Verlag, NY, 2002), pp. 36–41. 16. J. Marini and C. Murray, GSFC Report X-591-73-351 (1973). 17. V. B. Mendes and E. C. Pavlis, Geophys. Res. Lett. 31(14) (2004) L14602. 18. V. B. Mendes, G. Prates, E. C. Pavlis, D. E. Pavlis and R. B. Langley, Geophys. Res. Lett. 29(10) (2002) 1414, doi: 10.1029/2001GL014394. 19. G. C. Hulley and E. C. Pavlis, J. Geophys. Res. 112 (2007) B06417, doi: 10.1029/2006JB004834. 20. T. Otsubo and G. M. Appleby, J. Geophys. Res. 108 (2003) B4, 2201.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

The International Laser Ranging Service

b672-V13-ch09

153

21. G. Delle Monache, The INFN-LNF Space Climatic Facility for LARES and ETRUSCO, Proc. 15th Int. Laser Workshop, Canberra Australia, Oct. 2006, CD-ROM (2007). 22. G. Kirchner, W. Hausleitner and E. Cristea, IEEE Trans. Geoscience and Remote Sensing 45(1) (2007) 201–205. 23. V. Burmistrov, N. Parkhomenko, Y. Roy, V. Shargorodsky and V. Vasiliev, NASA/CP-2003-212248 (2003). 24. J. Degnan, Geodynamics, Special issue on laser altimetry 34 (2002) 551–594. 25. A. Bosco et al., Int. J. Mod. Phys. D 16 (2007) 2271–2285.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

This page intentionally left blank

b672-V13-ch09

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch10

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

EARTH ROTATION PARAMETERS FROM VERY LONG BASELINE INTERFEROMETRY AND RINGLASER OBSERVABLES P. J. MENDES CERVEIRA∗ , H. SPICAKOVA and H. SCHUH Institute of Geodesy and Geophysics Advanced Geodesy, Vienna University of Technology Gusshausstrasse 27-29, Vienna, 1040, Austria ∗ [email protected] T. KLUEGEL Bundesamt f¨ ur Kartographie und Geod¨ asie BKG, Fundamentalstation Wettzell, Sackenrieder Str. 25 93444 Koetzting, Germany U. SCHREIBER and A. VELIKOSELTSEV Forschungseinrichtung Satellitengeod¨ asie FESG Fundamentalstation Wettzell, Sackenrieder Str. 25 93444 Koetzting, Germany

The primary goal of this chapter is to provide the mathematical formalism for the combination of Very Long Baseline Interferometry (VLBI) and ringlaser observations. First, we derived the partials of a simpliﬁed VLBI observation equation, w.r.t. nutation, sidereal time, universal time, length of day, and geodetic polar motion parameters. All the same, the partials for the ringlaser technique, compatible to VLBI parameterization, are presented. A combination of both techniques seems primordial, as the ringlaser technology may in the near future contribute eﬀectively to the determination of subdiurnal Earth rotation variations, given its uniqueness in sensitivity to the instantaneous Earth rotation vector and its near real-time data acquisition. We describe the theoretical compatibility of both types of observables.

1. Introduction Very Long Baseline Interferometry (VLBI) can accurately measure the geodetic parameters associated with the shape of the Earth and its orientation in inertial space. This includes universal time variation (UT1– UTC), polar motion, and nutation/precession, i.e. the full set of Earth orientation parameters (EOP).5,8,13,17,20 In addition, VLBI is the only 155

FA

December 10, 2008 15:45

156

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch10

P. J. Mendes Cerveira et al.

technique that provides access to the quasi-inertial reference system through observations of the extragalactic radio sources that form the Celestial Reference Frame (CRF). Throughout this chapter, we mean by geodetic polar motion the motion of the Celestial Intermediate Pole (CIP) w.r.t. a conventional Terrestrial Reference Frame (TRF),21 whereas the geophysical polar motion represents the motion of the instantaneous Earth rotation axis w.r.t. the same conventional TRF. Today, the geodetic polar motion and universal time observations with a temporal resolution as short as semi-hourly are obtained from VLBI, with the potential to reveal short period and episodic events with signatures below 100 µas (7.5 µs) level. Eight-hours variations in geodetic polar motion in the order of 40 µas15 have been reported during the VLBI campaign CONT02, with eight stations continuously observed from 16 to 31 October 2002. Today, data from a rather new technology, called ringlaser gyroscope is available.7 Ringlasers such as at Wettzell fundamental station (Germany) sense uniquely the signatures of the instantaneous Earth rotation vector in an Earth-ﬁxed frame.2,5,7,14 The current situation is the following: up to now, the nontidal variations in Earth rotation at subdiurnal periods have remained undetected. In the subdiurnal band, one diﬃculty arises in the observation equation of VLBI because of the ambiguity between the geodetic polar motion and the nutation/precession parameters.1,14,17,18 In order to resolve this ambiguity, the scientiﬁc community adopted in the IERS Conventions 2003 a choice to separate both types of parameters depending solely on their frequencies, i.e. conventionally.9 As long as no external observations are able to distinguish between the physical meaning of both phenomena, this adoption is completely justiﬁed. In this respect the ringlaser technology could, in the near future, take over an important position in the determination of subdiurnal Earth rotation variations. For this reason, we will describe the theoretical compatibility of both types of observables (VLBI and ringlaser) enabling a combined determination of subdiurnal polar motion and universal time variations. Special attention has been dedicated to the application of identical reduction models.

2. Kinematics of Earth Orientation The VLBI technique is sensitive to the complete rotation matrix from a conventional TRF21 to a conventional CRF. Here, rotation means the temporal derivative of the orientation. A simpliﬁed “classical” parameterization based on the equinox will be used throughout this chapter.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch10

Earth Rotation Parameters from VLBI and Ringlaser Observables

157

Although an alternative formulation based on the Non Rotating Origin (NRO) has recently been recommended by the International Astronomical Union (IAU), the scientiﬁc community usually still decides to apply the “classical” equinox-based theory. The latter is from the mathematical point of view equivalent to the NRO theory. The barycenter of the solar system and the center of mass of the Earth are the most appropriate origins for any celestial and terrestrial reference system, respectively. The ecliptic and the equator, aﬀected by the complexities of the structure of the Earth and solar system, are fundamental planes directly related to the dynamics of the solar system.

2.1. Theoretical considerations Let Q be the transformation matrix from some conventional TRF21 to some conventional CRF.9 Then, any vector R in the TRF can be transformed to CRF, leading to the vector r, so that r = Q · R = P · N · U · X · Y · R,

(1)

where P, N, U, X, and Y account for precession, nutation in longitude ∆ψ and obliquity ∆ε, sidereal time (∆h), and geodetic polar motion (xp , yp ), respectively. Now, we assume that the orientation matrix Q0 is based on some set of excellent a priori Earth orientation models,3,4,9,12,16 so that only the uniform rotation, about a constant axis, described by U0 , is not an identity matrix. Then, to ﬁrst order, we obtain18 : ∆Q = Q0 − Q = (∆P + ∆N ) · U0 + ∆h · ∆U0 + U0 · (∆X + ∆Y ),

(2)

with h = h0 + ∆h,

(3)

being the hour angle of the true equinox of date, and   cos h0 − sin h0 0 U0 =  sin h0 cos h0 0  , 0 0 1   − sin h0 − cos h0 0 ∆U0 =  cos h0 − sin h0 0  , 0

0

0

(4)

(5)

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch10

P. J. Mendes Cerveira et al.

158



 1 ∆ψ · cos ε ∆ψ · sin ε , ∆N =  −∆ψ · cos ε 1 ∆ε −∆ψ · sin ε −∆ε 1   1 0 −xp ∆X + ∆Y =  0 1 yp  , xp −yp 1

(6)

(7)

where ε is the mean obliquity of the ecliptic. In the following, we assume that a deﬁciency in the variational precession matrix ∆P will be absorbed by the variational nutation matrix ∆N .

2.2. VLBI Since nearly three decades, VLBI has been successfully applied for measuring variations in all ﬁve components of Earth’s orientation. The basic VLBI observable is the delay τ in the time of arrival of a radio wavefront between two VLBI antennas. The geometric distance L, corresponding to the delay τ multiplied by the speed of light in vacuum c, can be related to the following approximate observation equation: L = c · τ = kT · Q · B,

(8)

where k(αδ) represents the Cartesian vector of the source position with right ascension α and declination δ in the CRF, and B(Bx , By , Bz ) the baseline vector in the TRF:   cos δ · cos α k =  cos δ · sin α  . (9) sin δ Assuming that all rotational motions aﬀecting k and B have been properly reduced, the linearized version of the reduced observation equation reads to ﬁrst order: ∆L = kT · ∆Q · B.

(10)

The partial derivatives of this reduced equation w.r.t. the ﬁve Earth orientation parameters can be derived by using Eq. (2): ∂(∆L) = sin δ · sin ε · [−Bx · cos h0 + By · sin h0 ] ∂(∆ψ) + cos δ · sin ε · cos α · Bz ,

(11)

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch10

Earth Rotation Parameters from VLBI and Ringlaser Observables

159

∂(∆L) = − sin δ · [Bx · sin h0 + By · cos h0 ] + cos δ · sin α · Bz , ∂(∆ε)

(12)

∂(∆L) = Bx · sin δ − Bz · cos δ · cos hs , ∂(xp )

(13)

∂(∆L) = −By · sin δ − Bz · cos δ · sin hs , ∂(yp )

(14)

∂(∆L) ∂(∆L) 1 = − cos δ · [Bx · sin hs + By · cos hs ] = · , ∂(∆h) f ∂(dUT1)

(15)

∂(∆L) = −f · cos δ · [Bx · sin hs + By · cos hs ], ∂(dUT1)

(16)

where f is the conversion factor from solar to sidereal time, i.e. 1.0027379093. The partial derivative of the equation of equinoxes w.r.t. the nutation parameter ∆ψ has also been taken into account in Eq. (11). The consequence is a zero mathematical correlation between the parameters ∆ψ and ∆h. The latter parameter is a function of the variation of universal time dUT1. The Greenwich apparent sidereal time hs has been used, i.e. hs = h0 − α. In view of a possible combination with ringlaser data, the required partials w.r.t. the time derivatives of polar motion and universal time are also set up: ∂(∆L) ∂(xp ) ∂(∆L) ∂(∆L) = · = · ∆t, ∂(x˙ p ) ∂(xp ) ∂(x˙ p ) ∂(xp )

(17)

∂(∆L) ∂(∆L) ∂(yp ) ∂(∆L) = · = · ∆t, ∂(y˙ p ) ∂(yp ) ∂(y˙ p ) ∂(yp )

(18)

∂(∆L) ∂(dUT1)

∂t

∂(∆L) = · ∂(dUT1)

∂(dUT1) ∂(dUT1)

∂t

=

∂(∆L) · ∆t, ∂(dUT1)

(19)

as well as partial w.r.t. length-of-day variations ∂(∆L) ∂(∆L) ∆t =− · . ∂(∆lod) ∂(dUT1) T0

(20)

In the last four equations we introduced the time interval ∆t. The nominal length of day is T0 , i.e. 86400 s SI.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch10

P. J. Mendes Cerveira et al.

160

From the theoretical point of view, there is no problem to estimate corrections w.r.t. EOP with an amplitude of 20 mas to an accuracy of 1 µas without resorting to iterations, assuming no noise in the VLBI data. This restriction of 20 mas arises due to the linearization of the presented observation model, i.e. Eq. (2), and of the subsequent approximations for the nutation and polar motion incremental angles. However, this restriction is not applicable to current VLBI software packages, e.g. OCCAM,10 as numerical diﬀerentiations are applied to the nonlinear observation equation.

3. Ringlaser The Wettzell ringlaser “G”, located at the geographical position 49.145◦ North and 12.876◦ East, is an inertial rotation instrument, which uses the Sagnac eﬀect that arises when two counter-rotating laser beams lead to a beat frequency upon a rotation variation.2,7 The Sagnac frequency ∆Sf relates the enclosed area A of the instrument, its perimeter (optical pathlength) Pr , the optical wavelength λopt , the unit normal vector n to the ringlaser area, formed by the beam path, and the rotation vector Ω w.r.t. a prescribed TRF21 : ∆Sf =

4·A · nT · Ω Pr · λopt

(21)

with 

cos ϕ · cos λ



  n =  cos ϕ · sin λ  sin ϕ



− cos ϕ0 · sin λ0 · ∆λtop − sin ϕ0 · cos λ0 · ∆ϕtop



  = n0 + ∆n = n0 +  cos ϕ0 · cos λ0 · ∆λtop − sin ϕ0 · sin λ0 · ∆ϕtop , cos ϕ0 · ∆ϕtop (22) where n0 is the reference unit normal vector with ϕ0 and λ0 being the nominal geographic latitude and longitude of the ringlaser position, respectively, and ∆ϕtop and ∆λtop their respective topocentric deﬂections of the local vertical. The instantaneous angular velocity vector of the Earth Ω can be parameterized by using a perturbation ∆Ω w.r.t. the reference

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch10

Earth Rotation Parameters from VLBI and Ringlaser Observables

161

angular velocity vector Ω0 : 

 0  Ω = R1 (my ) · R2 (mx ) ·  0 Ω0 + ∆Ω = Ω0 + ∆Ω   −Ω0 · mx ≈ Ω0 +  Ω 0 · m y  , ∆Ω

(23)

with Ω0 = 0 0

Ω0

T

=

0 0

2π T0

T ,

(24)

where R1 and R2 represent the rotation matrices (counterclockwise being a positive rotation) about the x- and y-axis of the TRF, taking the geophysical polar motion angles my and mx into account. The Sagnac frequency ∆Sf itself is rarely used. The relative change of the Sagnac frequency ∆Sr is preferable, and is related to the relevant parameters my , mx , and ∆Ω by the following equation: ∆Sr =

∆Sf − ∆Sf 0 nT · Ω − nT 0 · Ω0 = ∆Sf 0 nT · Ω 0 0

=

(n0 + ∆n)T · (Ω0 + ∆Ω) − nT 0 · Ω0 nT · Ω 0 0

≈

∆nT · Ω0 + nT 0 · ∆Ω nT · Ω 0 0

≈ cot ϕ0 · [−mx · cos λ0 + my · sin λ0 + ∆ϕtop ] +

∆Ω , Ω0

(25)

where the last term can be linked to a variation of length-of-day ∆lod or universal time dUT1 through −∆lod ∂(dUT1) ∆Ω . = = Ω0 T0 ∂t

(26)

Equation (25) shows that, to ﬁrst order approximation, the ringlaser is not sensitive to longitudinal tilt variations ∆λtop . For an elastic Earth, the local topocentric latitudinal deﬂection ∆ϕtop , if produced by the solid Earth tides, is related to the ﬁrst order geocentric

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch10

P. J. Mendes Cerveira et al.

162

latitudinal variation ∆ϕ by ∆ϕtop ≈ −∆ϕ ·

1 + k2 − h2 l2

≈ −8.1163 · ∆ϕ,

(27)

where k2 = 0.29525, h2 = 0.6078, and l2 = 0.0847 are the nominal degree-2 Love and Shida numbers. Equation (27) allows to calculate topocentric latitudinal tilt variations due to the solid Earth tides from displacements on Earth’s surface as obtained from the most recent models,11 accounting for latitudinal and frequency-dependent Earth’s response. However, if the latitudinal variation ∆ϕ is caused by a surface load, such as an ocean tide load or atmospheric pressure load, then the latitudinal deﬂection of the local vertical ∆ϕtop is ∆ϕtop ≈ −

∞ 1 ∂Pn (cos ϑ) (1 + kn − hn ) ME n=1 ∂ϑ

∞

≈ −∆ϕ ·

(cos ϑ) + kn − hn ) ∂Pn∂ϑ ∞ ∂Pn (cos ϑ) n=1 ln ∂ϑ

n=1 (1

≈ −∆ϕ · G(kn , hn , ln , ϑ),

(28)

where the degree-dependent load Love numbers kn , hn , and ln are used as well as the partials of the Legendre polynomials Pn (cos ϑ) w.r.t. the spherical distance ϑ from the point mass load, and ME is Earth’s mass.22 The load Love numbers can be computed for speciﬁc Earth models in speciﬁc center of frames, e.g. the center of frame of the solid Earth or the center of mass frame of the Earth’s system including the load. Usually, the degree n goes up to 10,000, leading to a so-called Green’s function G(kn , hn , ln , ϑ) in Eq. (28), which can be computed once for all spherical distances ϑ from the point mass load, and used for the computation of ∆ϕtop if the latitudinal variation ∆ϕ is available from latitudinal surface deformations induced by surface loads. Finally, the reduced relative Sagnac frequency ∆SRLG (corrected for tilt variations ∆n of the unit normal vector n in latitudinal direction) is ∆SRLG = ∆Sr − cot ϕ0 · ∆ϕtop = cot ϕ0 · [−mx · cos λ0 + my · sin λ0 ] −

∆lod , T0

(29)

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch10

Earth Rotation Parameters from VLBI and Ringlaser Observables

163

and further ∆SRLG

1 1 ≈ − cot ϕ0 · xp − y˙ p · cos λ0 + yp + x˙ p · sin λ0 Ω0 Ω0 −

∆lod . T0

(30)

In Eq. (30), a substitution of the geophysical polar motion parameters was applied: to ﬁrst order approximation, the geodetic polar motion parameters obtained from space geodesy, xp and yp w.r.t. the CIP, are related to the geophysical polar motion parameters mx and my through the following formula1,14,17 :   1 y˙ x −  p Ω0 p  mx . ≈ (31)   1 my −yp − x˙ p Ω0 Equation (29) obviously shows that each ringlaser provides one observation for each epoch. However, there are three unknowns (mx , my , ∆lod) assuming that local position variations can be accurately monitored by external instruments, e.g. tiltmeters, or modeled theoretically. Thus, for one epoch, a minimum of three ringlasers at diﬀerent geographic positions observing simultaneously could unambiguously determine these three unknowns. Equivalently, one ringlaser having three independently oriented beam circuits (i.e. three ringlasers oriented orthogonally at one site) would solve for the three unknowns. An interesting case is available when a ringlaser is mounted horizontally at the North Pole: it would be primarily sensitive to the length-of-day variations, as cot(90◦ ) = 0. From the practical point of view, Eq. (30) allows to calculate the relative Sagnac frequency variation from geodetic polar motion and ∆lod data, while Eq. (29) is applicable for theoretical models, such as eﬀects accounting for solid Earth tides or Oppolzer terms.19,23 Both equations are needed for the reduction of ringlaser observations due to known eﬀects, such as solid Earth tides, eﬀect of ocean tides, and geodetic polar motion. Finally, we obtain the partial derivatives for the ringlaser w.r.t. geodetic polar motion (plus their respective rates) and ∆lod: ∂(∆SRLG ) = − cot ϕ0 · cos λ0 ∂(xp )

(32)

∂(∆SRLG ) = − cot ϕ0 · sin λ0 ∂(yp )

(33)

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

164

9in x 6in

b672-V13-ch10

P. J. Mendes Cerveira et al.

1 ∂(∆SRLG ) ∂(∆SRLG ) = · ∂(x˙ p ) Ω0 ∂(yp )

(34)

1 ∂(∆SRLG ) ∂(∆SRLG ) =− · ∂(y˙ p ) Ω0 ∂(xp )

(35)

∂(∆SRLG ) 1 =− . (∆lod) T0

(36)

The main advantage of the ringlaser compared to space geodetic techniques is the high temporal data acquisition in quasi-near real-time and its direct relation to the instantaneous Earth rotation vector Ω. Equations (26) and (31) can also be used as condition equations for the combination of VLBI and ringlaser observations. Nutation parameters always have to be ﬁxed, if subdiurnal geodetic polar motion is estimated. 4. Discussion and Conclusions We set up the mathematical formalism to connect VLBI and ringlaser observations, considering the diﬀerence between geodetic and geophysical polar motion. On the one hand, the detection limit of subdiurnal polar motion eﬀects from VLBI reaches presently an accuracy of better than 3 mm on the Earth’s surface. With the upcoming VLBI2010 system,6 enabling an increased station network and providing antennas with faster slewing rates we are optimistic to signiﬁcantly go beyond this detection limit. On the other hand, the subdiurnal relative Sagnac frequency variation obtained from the Wettzell G ringlaser, projected to Earth’s surface, today attains an accuracy of better than 3 cm. The gain of one order of magnitude within only a few years, makes this emerging ringlaser technology a serious complementary system to VLBI w.r.t. the determination of subdiurnal ERP variations. The combination of both systems for common and unique parameters of the Earth is one of the goals of the Global Geodetic Observing System (GGOS) initiated by the International Association of Geodesy (IAG). This chapter puts up succinctly a foundation for such a successful combination. Acknowledgments The authors are especially grateful to Richard Gross for the review of the manuscript. The ﬁrst author is particularly indebted to the German Science Foundation (DFG, Deutsche ForschungsGemeinschaft) for funding

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Earth Rotation Parameters from VLBI and Ringlaser Observables

b672-V13-ch10

165

this work within the Research Unit FOR584 “Earth Rotation and Global Dynamic Processes.”

References 1. A. Brzezinski and N. Capitaine, J. Geophys. Res. 98 (1993) 6667. 2. B. F. Chao, Trans. Am. Geophys. Union 72 (1991) 550. 3. B. F. Chao, R. D. Ray, J. M. Gipson, G. D. Egbert and C. Ma, J. Geophys. Res. 101 (1996) 20151. 4. C. Bizouard, R. Eanes, R. Ray, P. Brosche, P. Defraigne, S. Dickman, D. Gambis and R. Gross, IERS Conventions 2003 92 (2004). 5. H. Schuh, R. Dill, H. Greiner-Mai, H. Kutterer, J. Mueller, A. Nothnagel, B. Richter, M. Rothacher, U. Schreiber and M. Soﬀel, Mitteilungen des Bundesamtes f¨ ur Kartographie und Geod¨ asie 32 (2003). 6. J. Wresnik, J. Boehm, A. Pany and H. Schuh, Submitted to Ad. Geosci. 7. K. U. Schreiber, A. Velikoseltsev, M. Rothacher, T. Kl¨ ugel, G. E. Stedman and D. L. Wiltshire, J. Geophys. Res. 109 (2004). 8. M. Rothacher, G. Beutler, T. A. Herring and R. Weber, J. Geophys. Res. 104 (1999) 4835. 9. N. Capitaine, P. M. Mathews, P. Wallace, P. Bretagnon, R. Gross, T. Herring, G. Kaplan, D. McCarthy, B. Richter and P. Simon, IERS Conventions 2003 33 (2004). 10. O. A. Titov, V. Tesmer and J. Boehm, AUSLIG Technical Reports 7 (2001). 11. P. M. Mathews, V. Dehant and J. M. Gipson, J. Geophys. Res. 102 (1997) 20469. 12. P. M. Mathews, T. A. Herring and B. A. Buﬀett, J. Geophys. Res. 107 (2002). 13. P. J. Mendes Cerveira, J. Boehm, E. Tanir, J. Wresnik, H. Schuh and V. Tesmer, Geowissenschaftliche Mitteilungen 79 (2007) 209. 14. P. J. Mendes Cerveira, R. Weber and H. Schuh, VGI 2 (2007) 113. 15. R. Haas and J. W¨ unsch, J. Geodynamics 41 (2006) 94. 16. R. D. Ray, D. J. Steinberg, B. F. Chao and D. E. Cartwright, Science 264 (1994) 830. 17. R. S. Gross, Treatise Geophys. 3 (2007). 18. T. M. Eubanks, Geodynamic Series 24 (1993). 19. T. R. von Oppolzer, W. Engelman (2nd ed.), Leipzig 1 (1882) 152. 20. W. Schl¨ uter, E. Himwich, A. Nothnagel, N. Vandenberg and A. Whitney, Adv. Space Res. 30 (2002) 145. 21. Z. Altamimi, X. Collilieux, J. Legrand, B. Garayt and C. Boucher, J. Geophys. Res. 112 (2002). 22. W. E. Farrell, Rev. Geophys. 10 (1972) 751. 23. A. Brzezinski, Manuscripta Geodaetica 11 (1986) 226.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

This page intentionally left blank

b672-V13-ch10

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch11

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

TOWARD A NEW VLBI SYSTEM FOR GEODESY AND ASTROMETRY ¨ ¨ JORG WRESNIK∗ , JOHANNES BOHM, ANDREA PANY and HARALD SCHUH Institute of Geodesy and Geophysics, Vienna University of Technology Gusshausstr. 27-29, Vienna, 1040, Austria ∗ [email protected]

Simulations are carried out at the Institute of Geodesy and Geophysics (IGG), Vienna, to support the design of a new geodetic Very Long Baseline Interferometry (VLBI) system, VLBI2010. The main part of these simulations is a Monte Carlo simulator which produces artiﬁcial group delays by modeling the stochastic processes caused by the station clocks, the wet zenith delays, and additional system errors. The clocks are simulated with a random walk plus integrated random walk; the wet zenith delays are derived from a turbulence model; and the system errors are represented by white noise. For the evaluation of the Monte Carlo simulator, comparisons with the continuous VLBI observations of CONT05 have been carried out. They show that the simulator is performing realistically including the turbulence model that produces values for the atmospheric path delays. For the analysis of the simulated data, the Monte Carlo simulator is implemented in a modiﬁed version of the software package OCCAM. Because of limitations due to the huge number of observations in the OCCAM Gauß-Markov algorithm, the Kalman Filter approach was applied. Diﬀerent variance rates for the stochastic parameters in the Kalman Filter solutions are tested, and the best values are used for the antenna slew speed test. Diﬀerent schedules with antennas of diﬀerent slew speeds (from 1.5◦/s to 12◦/s in azimuth and 0.7◦/s to 3.1◦/s in elevation) are compared w.r.t. baseline length repeatabilities. The investigation shows that there is hardly any improvement with antennas faster than 4.5◦/s in azimuth and 2.1◦/s in elevation, in using the present scheduling strategies.

1. Introduction Within the context of the International Association of Geodesy’s (IAG) project, Global Geodetic Observing System (GGOS), it has become clear that the modern space geodetic techniques should try to provide station coordinates and/or baseline length time series with an accuracy of better than 1 mm and velocities of clearly better than 1 mm/year. Only then, 167

FA

December 10, 2008 15:45

168

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch11

J. Wresnik et al.

subtle eﬀects such as non-linear station motions or sea-level rise can be detected. There has been a lot of discussion in recent years on how Very Long Baseline Interferometry (VLBI) could exploit its present resources more eﬃciently and how future VLBI networks should look to achieve this general goal of sub-mm accuracy. In October 2003, the International VLBI Service for Geodesy and Astrometry (IVS) installed Working Group 3 (WG3) “VLBI 2010” to examine current and future requirements for geodetic VLBI systems. Based on the ﬁnal report of WG33 and in particular on the requests deﬁned by the WG3 subgroup on “Observing Strategies” and to encourage the implementation of the recommendations of WG3, the IVS established the VLBI2010 Committee (V2C) as a permanent body of IVS.4 A new set of criteria to specify the next generation geodetic VLBI system was established by the V2C, mainly based on • the recommendations for future IVS products given in the IVS Working Group 2 Report (WG2)5 and the requirements of the Global Geodetic Observing System project (GGOS) of the International Association of Geodesy (IAG); • the science-driven geodetic goals outlined in the NASA Solid Earth Science Working Group Report (SESWG). The IVS V2C investigated the various facets that the new system needs to fulﬁll the requirements. The major focus is on reducing the main error sources stemming from the atmosphere, the instrumentation, and the structure of the radio sources. Furthermore, the following strategies are investigated: • a reduction of the random component of the delay measurement error, e.g. the stochastic properties of the clocks and the unmodeled variation in the atmosphere; • a reduction of systematic errors, e.g. the thermal and gravitational deﬂection of the antenna, drifts of the electronics, and radio source structure; • an increase of the number of antennas for geodetic VLBI and an improved geographic distribution; • an increase of observation density, i.e. number of observations per unit time; • a reduction of susceptibility to external radio-frequency interference. At the Institute of Geodesy and Geophysics (IGG), Vienna University of Technology, diﬀerent simulations are performed to evaluate new observing

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch11

Toward a New VLBI System for Geodesy and Astrometry

169

strategies and schedules, to improve the modeling of troposphere refraction and clocks, to ﬁnd the best antenna design and to optimize the network geometry. A sequence of software programs is used for the simulations. After scheduling the observations with SKED,6 they are transformed to the socalled NGS format and then used as input to the VLBI analysis software package OCCAM,2 which was adapted for our simulations. The main part of the simulation studies is the so-called Monte Carlo simulator which creates the artiﬁcial observations based on realistic properties of the wet zenith delays and clocks. The observed group delay minus computed group delay (o − c) can be described as follows: o − c = (wzd 2 · mfw 2 (e) + cl2 ) − (wzd 1 · mfw 1 (e) + cl1 ) + wnBsl .

(1)

In Eq. (1), wzd1,2 are simulated wet zenith delays based on the turbulence model;1 cl1,2 are simulated clock values modeled as random walk plus integrated random walk7 at stations 1 and 2 of each observation; and mfw1,2 (e) are the wet mapping functions for the elevation angle e, which are assumed to be errorless in our studies. For each baseline observation, an additional white noise wnBsl is added to model the instrumental errors of stations 1 and 2. The Monte Carlo simulator, implemented in OCCAM, imports wet zenith delay values from the turbulence model, creates clock values for each station and epoch, and adds white noise for each observation. The following criteria are used to evaluate the potential of the VLBI system: baseline length repeatabilities, the root mean square (rms) of the 3D station position residuals, formal errors and standard deviations of the Earth Orientation Parameters (EOP), and the standard deviation between the simulated stochastic processes (troposphere delays, clocks) and their estimates. In this chapter, we are mainly focusing on the baseline length repeatabilities and the rms of the 3D station position to evaluate and compare diﬀerent observing strategies and scenarios. A comparison between simulated and real data is carried out ﬁrst to evaluate the Monte Carlo simulator itself and the reliability of the simulated group delays.

2. Evaluation of the Monte Carlo Simulation The Monte Carlo simulator creates group delays for each baseline observation as described in Eq. (1). An evaluation of the simulator is important to get an idea about the reliability of the whole simulation

FA

December 10, 2008 15:45

170

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch11

J. Wresnik et al.

Fig. 1. Station map with 11 VLBI sites for the continuous VLBI campaign CONT05 in September 2005.

work and the conclusions and/or decisions drawn from the investigations. Therefore, the Monte Carlo simulator is compared against real observations from CONT05, a two-week campaign of continuous VLBI observations from 12 September 2005 until 27 September 2005. The observed network consists of 11 stations (Fig. 1): Algonquin Park (Canada), Gilmore Creek (Alaska, USA), HartRAO (South Africa), Kokee Park (Hawaii, USA), Ny ˚ Alesund (Norway), Onsala (Sweden), Svetloe (Russia), TIGO Concepcion (Chile), Tsukuba (Japan), Westford (Massachusetts, USA), and Wettzell (Germany). The results of the CONT05 analysis are generally considered the best VLBI contribution to date. For the simulation, the schedule of the ﬁrst 24 hours of CONT05 is taken, and wet zenith delays, clocks, and white noise are simulated for 15 days, to be as similar as possible to the real observations. The simulation of the variable wet zenith delay follows the theory of Treuhaft and Lanyi,8 which is based on the Kolmogorov turbulence theory.9 The refractive index structure function constant (Cn ) and the eﬀective height of the wet troposphere, which are the most important input parameters for the turbulence model, are assumed to be constant over the 15 days for each station but wind speed and wind direction vary with time. Thus, the simulated delays will vary as a function of observed direction and as a function of time.1 This approach promises to provide very realistic wet path delays through a turbulent atmosphere. The clocks are simulated with a random walk plus integrated random walk corresponding to an Allan standard deviation (ASD) of 1 · 10−14 @50 min, which is a high-performing maser system representative

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch11

Toward a New VLBI System for Geodesy and Astrometry

171

25 CONT05 real data CONT05 simulated data

length repeatability [mm]

20 clocks: 1*10-14@50min wn: formal errors of NGSfile wzd: turbulence model 15

10

5

0 0

2

4

6 8 10 baseline length [1000 km]

12

14

Fig. 2. Baseline length repeatabilities of the real VLBI CONT05 sessions (ﬁlled circles) and the Monte Carlo simulation (ﬁlled squares). There is good agreement, especially for baselines shorter than 6,000 km.

for the present VLBI sites. The added white noise was taken from the formal errors of real observation ﬁles (NGS ﬁles) from CONT05. The results are presented in Fig. 2, where the circles show the baseline length repeatabilities of the CONT05 sessions and the squares correspond to the baseline length repeatabilities derived from the Monte Carlo simulation. The agreement for the shorter baselines is high. However, there is disagreement for a few baselines longer than 6,000 km, where the diﬀerence between the repeatabilities from real and simulated observations lies in a range of ±1 cm. For baselines shorter than 6,000 km, these diﬀerences are less than ±2 mm (Fig. 3). Generally, it can be concluded that the simulation of atmospheric path delays using the turbulence model creates realistic VLBI group delays for the Monte Carlo simulator. In parallel, similar studies are carried out at NASA/GSFC, Greenbelt, USA10 by using the VLBI analysis software SOLVE.

3. Comparison of Gauß–Markov LSM and KF in OCCAM The VLBI analysis software OCCAM is used by various IVS Analysis Centers for routine analysis. For the tasks described above, the oﬃcial version of the software package was modiﬁed to implement the Monte Carlo

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch11

J. Wresnik et al.

172 8

CONT05 real data - CONT05 simulated data

diff. of length repeatability [mm]

6 4 2 0 −2 −4 −6 −8 −10 −12

0

2

4

6 8 10 baseline length [1000 km]

12

14

Fig. 3. Diﬀerences of the baseline length repeatabilities from the analysis of the real CONT05 data and the Monte Carlo simulation. For the simulation, the wet zenith delays are taken from the turbulence model, clock errors are simulated with an ASD of 1 · 10−14 @50 min, and for the white noise, the formal errors are taken out of the observation ﬁles (NGS ﬁles).

simulator. Due to very dense schedules with large numbers of observations, it was not possible to solve the normal equation system of the Gauß–Markov model as implemented in OCCAM. Thus, for the proposed simulation, we applied the Kalman Filter approach, which is also implemented in OCCAM.2 To compare the two diﬀerent approaches, a test data set of 120 baselines was analyzed. The data set results from an observing schedule for a 16-station network with 9,678 observations (11 observations/hour/ station, 7 min and 20 s maximum time lag between two observations). The baseline length repeatabilities for the 120 baselines (Fig. 4), analyzed both with the Gauß–Markov and the Kalman Filter, show diﬀerences (Fig. 5) which are in the range of ±4 mm and to 67% positive in the sense Gauß–Markov minus Kalman Filter with a maximum reduction of variance of 40% in some cases. The reason for the Kalman Filter being superior to the Gauß–Markov method in most of the cases might be that the Kalman Filter estimates the wet zenith delays and clocks at every observation epoch, while in the Gauß–Markov method, wet zenith delays and clocks are estimated as continuous piecewise linear functions with time intervals of 30 min. This is a particular advantage of the Kalman Filter for

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch11

Toward a New VLBI System for Geodesy and Astrometry

173

12 Gauß Markov Kalman Filter

length repeatability [mm]

10

8

6

4

2

0 0

2

4

6 8 10 baseline length [1000 km]

12

14

Fig. 4. Baseline length repeatabilities of a test data set with 120 baselines, both analyzed with the Gauß–Markov method (ﬁlled circles) and with the Kalman Filter approach (ﬁlled squares).

4 Gauß Markov - Kalman Filter

diff. of length repeatability [mm]

3 2 1 0 −1 −2 −3 −4

0

2

4

6 8 10 baseline length [1000 km]

12

14

Fig. 5. Diﬀerences of baseline length repeatabilities of the test data set with 120 baselines, both analyzed with the Gauß–Markov method and with the Kalman Filter approach.

FA

December 10, 2008 15:45

174

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch11

J. Wresnik et al.

the simulation, because the resolution of wet zenith delays and clocks only depends on the observation density of the analyzed schedule and does not have to be adjusted for each schedule as in the case of the Gauß–Markov model.

4. OCCAM Kalman Filter Analysis with Diﬀerent Variance Rates for Wet Zenith Delays and Gradients The troposphere wet zenith delay and gradient parameters are considered as random walk stochastic processes11 in the Kalman Filter method. The gradients describe the azimuthal asymmetry in the refractive index of the atmosphere. In the following, the variance rates for estimating wet zenith delay and gradients are varied to assess their inﬂuence on the solution. For this analysis, a schedule for a virtual 16-station network (Fig. 6) was produced. The virtual stations are chosen in a way to get good global station distribution. The only criterion for the additional sites which is required is an existing IGS (International GPS Service) station at this position. For the schedule, antennas with slew speeds of 12◦/s in azimuth and 3.1◦/s in elevation are used. The average number of observations/hour/station is 131. Figure 7 shows the baseline length repeatabilities, once without using gradients and twice with gradients estimated, where two diﬀerent values for the variance rates of wet zenith delay and gradients were used: 0.1 ps2/s for wet zenith delays and 0.015 ps2/s for gradients (tight constraints), and

Fig. 6. Virtual 16-station test network for VLBI2010 simulations with very good global distribution.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch11

Toward a New VLBI System for Geodesy and Astrometry

175

7 no gradients wzd:0.1, grd:0.015 wzd:0.7, grd:0.5

6

clocks: 2*10-15@15min

repeatability [mm]

5

wn: 4psec / bsl wzd: turbulence model

4

3

2

1

0 3

4

5

6

7 8 9 10 baseline length [1000 km]

11

12

13

Fig. 7. Baseline length repeatabilities for the OCCAM Kalman Filter analyses with diﬀerent values for variance rates of wet zenith delays and gradients.

0.7 ps2/s for wet zenith delays and 0.5 ps2/s for gradients (loose constraints). The repeatabilities for the solution without gradients are in a range of about 7 mm for baselines longer than 12,000 km, and the improvement when using gradients can be seen very clearly and also the enhancement for loose constraints compared to tight constraints. The repeatability for the longest baseline improves by about 2 mm for the loose constraints. The baseline length repeatability is a good criterion to show the performance of the whole VLBI system, but it does not provide a single number to compare diﬀerent solutions easily. Thus, the median of the root mean square (rms) values of the 3D station position residuals is calculated Table 1. Median rms of the 3D station position residuals analyzed with diﬀerent values for the variance rates of wet zenith delays and gradients. Wet zenith delays [ps2/s]

Gradient [ps2/s]

Median rms [mm]

0.1 0.1 0.7 0.1 0.7

No 0.015 0.5 0.5 0.015

2.37 1.29 1.05 1.19 1.21

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch11

J. Wresnik et al.

176

6 no gradients wzd:0.1, grd:0.015 wzd:0.7, grd:0.5

4

median:2.3678

1.2902

1.0461

3 2

HOBA

KERG

MAS1

FORT

TIGO

HART

KWJ1

WEST

GILC

TAHI

KOKE

TSUK

BADA

−1

NYAL

0

BAN2

1

WETT

rms of 3D Position [mm]

5

Fig. 8. Rms values for the 3D station position for each single station with diﬀerent analysis strategies. The ﬁrst bar at the station shows the analysis without gradients, the second with variance rates of 0.1 ps2/s and 0.015 ps2/s for wet zenith delays and gradients, respectively, and the third with variance rates of 0.7 ps2/s for wet zenith delays and 0.5 ps2/s for gradients.

(Table 1). Also, the rms values of the 3D station position residuals are plotted (Fig. 8) to reveal station-dependent behavior in the estimation.

5. Slew Speed Studies One of the main goals of the simulations is to obtain information about antenna speciﬁcations for the new VLBI2010 system by testing diﬀerent slew speeds. The test values used range from 1.5◦/s to 12◦/s in azimuth and from 0.7◦/s to 3.1◦/s in elevation. The schedules for the analysis were produced by J. Gipson, NASA/GSFC, Greenbelt, USA. To get a dense schedule, about 100 of 230 radio sources were taken from a recently compiled and improved list of geodetic sources, and the on-source time was reduced to a maximum of 5–10 s. Statistics for the schedules are summarized in Table 2. As can be seen, the schedule performance improves steadily from the 1.5/0.7◦/s case to the 6.0/2.1◦/s case. But, improvement is negligible from the 6.0/2.1◦/s case to the 12.0/3.5◦/s case. This is because the on-source time and

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch11

Toward a New VLBI System for Geodesy and Astrometry

177

Table 2. Number of observations/hour/station (obs/h/st) and number of observations in total for diﬀerent antenna slew speeds in azimuth and elevation. Slew speed az [◦/s]

el [◦/s]

1.5 3.0 4.5 6.0 12.0

0.7 0.7 2.1 2.1 3.5

Avg obs/h/st

Min obs/h/st

Max obs/h/st

No. of obs.

47 61 101 120 131

39 46 76 89 96

55 76 121 144 155

59392 83149 134134 159088 173831

accelerate/decelerate phases begin to dominate over the maximum slew speed phase of the observing cycle. More investigations on optimizing scheduling strategies have to be carried out. For the Monte Carlo simulations, the wet zenith delays are simulated using the turbulence model; the clocks are simulated with an ASD of 2·10−15@15 min, and the white measurement noise is simulated using a 4 ps 1-sigma Gaussian random variable. The very small white noise corresponds

6 az:1.5˚/s, el:0.7˚/s az: 3˚/s, el:0.7˚/s az:4.5˚/s, el:2.1˚/s az: 6˚/s, el:2.1˚/s az: 12˚/s, el:3.5˚/s

repeatability [mm]

5

4 clocks: 2*10-15@15min wn: 4psec / bsl

3

wzd: turbulence model 2

1

0 3

4

Fig. 9.

5

6

7 8 9 10 baseline length [1000 km]

11

12

Baseline length repeatabilities for diﬀerent slew speeds.

13

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch11

J. Wresnik et al.

178

6 az:1.5˚/s, el:0.7˚/s az: 3˚/s, el:0.7˚/s az:4.5˚/s, el:2.1˚/s az: 6˚/s, el:2.1˚/s az: 12˚/s, el:3.5˚/s

repeatability [mm]

5

4

3

2

1

0 3

Fig. 10.

4

5

6

7 8 9 10 baseline length [1000 km]

11

12

13

Linear ﬁt of the baseline length repeatabilities shown in Fig. 9.

to that predicted for the new VLBI2010 antenna systems. Regarding the baseline length repeatabilities (Fig. 9), a very clear improvement from slow antennas to antennas with a slew speed of 4.5◦/s in azimuth and 2.1◦/s in elevation can be seen. For a better illustration, the linear ﬁt of the baseline length repeatabilities is plotted separately in Fig. 10 which clariﬁes that the biggest improvement is gained when using antennas with the highest slew speed of 12◦/s in azimuth and 3.5◦/s in elevation, but the enhancement compared to slower antennas with a slew speed of 4.5◦/s or 6◦/s in azimuth and 2.1◦/s in elevation is relatively small. 6. Conclusions and Perspectives For the VLBI2010 committee, simulation studies are of very high interest because decisions about the next generation VLBI system will be mainly based on these results. Therefore, the Monte Carlo simulator has to be able to reproduce real observations. Hence a comparison with the continuous CONT05 VLBI campaign has been carried out, which shows a good agreement between real and simulated observations. For baselines up to

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Toward a New VLBI System for Geodesy and Astrometry

b672-V13-ch11

179

6,000 km, the diﬀerences between the baseline length repeatabilities are in the range of ±2 mm, while for longer baselines, the range is ±1 cm. Since wet zenith delays have the main inﬂuence on the simulation results, they are simulated as realistically as possible. Thus, applying the turbulence model is very important for the Monte Carlo simulator. In the VLBI analysis software OCCAM, which was modiﬁed to analyze simulated observations, the Gauß–Markov model as well as the Kalman Filter is realized. In the comparison of both models, the Kalman Filter shows its advantages for simulation studies. With it, the resolution of wet zenith delays and clocks only depend on the observation density of the analyzed schedule and do not have to be adjusted for each diﬀerent schedule as is the case for the Gauß–Markov analysis. Further investigations on the variance rates for wet zenith delay and gradients have been carried out to ﬁnd the most suitable values. As a result, the repeatabilities are the best (2 mm for the longest baselines) when using variance rates of 0.7 ps2/s and 0.5 ps2/s for wet zenith delay and gradients, respectively. These can be called “loose constraints.” For comparison, the baseline length repeatabilities for solutions without gradients are in the range of about 7 mm, a factor of 3.5 worse. Furthermore, the median of the rms of the 3D station position residuals was compared for the three diﬀerent solutions with a median of 2.37 mm for the solution without gradients, 1.29 mm for the solution with tight constraints, and 1.05 mm for the solution with loose constraints. Consequently, these loose variance rates were used for the determination of ideal slew speeds for the VLBI2010 antennas. Baseline length repeatabilities show a big improvement for antennas with a slew speed up to 4.5◦/s in azimuth and 2.1◦/s in elevation, but for faster antennas the enhancement is hardly signiﬁcant. To proﬁt from very fast moving antennas, more investigations on scheduling (e.g. reducing the idle time of the antennas) have to be carried out. All these simulations consist of only the geometry and the stochastic situations, but neglect other important facts such as mechanical, structural, and geological stability of an antenna or costs.

References 1. T. Nilsson, R. Haas and G. Elgered, Proc. 18th EVGA Working Meeting (2007), p. 175. 2. O. Titov, V. Tesmer and J. B¨ ohm, Proc. 3rd IVS General Meeting (2004), p. 267.

FA

December 10, 2008 15:45

180

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch11

J. Wresnik et al.

3. A. Niell, A. Whitney, B. Petrachenko, W. Schl¨ uter, N. Vandenberg, H. Hase, Y. Koyama, C. Ma, H. Schuh and G. Tuccari, Report of Working Group 3 to the IVS Directing Board (2005). 4. D. Behrend, J. B¨ ohm, P. Charlot, T. Clark, B. Corey, J. Gipson, R. Haas, Y. Koyama, D. MacMillan, Z. Malkin, A. Niell, T. Nilsson, B. Petrachenko, A. Rogers, G. Tuccari and J. Wresnik, Submitted to IUGG Proc. 2007. 5. H. Schuh, P. Charlot, H. Hase, E. Himwich, K. Kingham, C. Klatt, C. Ma, Z. Malkin, A. Niell, A. Nothnagel, W. Schl¨ uter, K. Takashima and N. Vandenberg, International VLBI Service for Geodesy and Astrometry 2001, Annual Report NASA/TP-2002-210001 (2002), p. 13. 6. N. Vandenberg, SKED: Interactiv/Automatic Scheduling Program, Program Reference Manual, Goddard Space Flight Center (1999). 7. J. B¨ ohm, J. Wresnik and A. Pany, IVS Memorandum 2006–013v03 (2007). 8. R. N. Treuhaft and G. E. Lanyi, Radio Science 22(2) (1987) 251. 9. V. Tatarskii, Jerusalem Israel Program for Scientific Translations (1971). 10. D. S. MacMillan, Proc. 18th EVGA Working Meeting (2007), p. 163. 11. T. A. Herring, J. L. Davis and I. I. Shapiro, J. Geophys. Res. 95(B8) (1990) 12561. 12. L. Petrov, Proc. 18th EVGA Working Meeting (2007), p. 141.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch12

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

PERIODIC STATION MOTION IN GOTHENBURG OBSERVED WITH GPS — POSSIBLY RELATED TO HYDROLOGICAL PHENOMENA? R. HAAS∗ , N. TANGDAMRONGSUB, H.-G. SCHERNECK and J. JOHANSSON Department of Radio and Space Science Chalmers University of Technology, SE-413 26 G¨ oteborg, Sweden ∗ [email protected]

We analyze several years of GPS data recorded with roof-top stations in Gothenburg (Sweden) and permanent stations of the Swedish national reference network in the larger Gothenburg region. The derived station positions of several of the roof-top stations in Gothenburg reveal periodic horizontal motions with amplitudes of several millimeters and periodic vertical motions with amplitudes of up to 1.5 mm. The observed motions have primarily an annual period. We compare the GPS-derived station motions with hydrological information (ground water and river level) and meteorological information (air temperature and pressure). The results indicate that the GPS-detected motions might be related to hydrological phenomena.

1. Introduction The city of Gothenburg is located at the Swedish west coast. This region is geologically dominated by bedrock, sediments, and thick layers of postglacial clay and is classiﬁed as a region of moderate- to high-risk for ¨ river runs through the city landslides. The southern branch of the G¨ ota Alv center and transports the run-oﬀ into the sea. The central and oldest parts of the city are located close to the river. Here the buildings are constructed on sediments and postglacial clay with high porosity. Many of the outer parts of the city on the other hand are constructed on bedrock with only a thin or no soil cover at all. There are several ground water observation stations and a river level observing station in the Gothenburg area. In the South, East, and North of Gothenburg, there are observing stations of the national network of permanent reference stations for GPS in Sweden, SWEPOS.1 181

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch12

R. Haas et al.

182

The SWEPOS stations are equipped with geodetic monuments that are founded in bedrock. GPS observational data are continuously recorded within the SWEPOS network since several years. During the last years, additionally a number of roof-top stations for real-time kinematic services have been established in Gothenburg and are operated by the municipality of Gothenburg. Figure 1 shows a map of the Swedish west coast with three SWEPOS stations (VANE, BORA, ONSA) in the surrounding of Gothenburg depicted as black triangles enclosed by circles, ﬁve roof-top GPS stations in the city (BISK, FROL, GOTA, KORT, TRAK) depicted as black inverted triangles enclosed by circles, three ground-water-level stations (GW52, GW53, GW54), and the river-level station in Gothenburg harbor (RL),

LAKE VÄNERN

VANE

58˚30'

GÖTA ÄLV RIVER

58˚00' GW53 KORT BISK RL FROL

GW54

GOTA TRAK

BORA

GW52

57˚30' ONSA

12˚30'

12˚00'

11˚30'

THE SEA

Fig. 1. Map of the Swedish west coast showing the location of the GPS stations used for this work (triangles — SWEPOS stations, inverted triangles — roof-top stations), and the ground water and river level stations. (See text for further explanations.)

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch12

Periodic Station Motion in Gothenburg Observed with GPS Table 1. Site name

183

Geological and ground water situation at the GPS stations. Geological situation

Possible ground water withdrawal (l/h)

Soil cover

Bedrock

BISK

None or thin quaternary deposit

Oldest granite partly aged

600–2000

BORA

None or thin quaternary deposit

Orthogneiss partly aged

600–2000

FROL

None or thin quaternary deposit

Oldest granite partly aged

GOTA/TRAK

Sand, gravel postglacial clay

Oldest granite partly aged

600–2000

KORT

None or thin quaternary deposit

Oldest granite partly aged

600–2000

ONSA

None or thin quaternary deposit

Gabbro, diorit amﬁbolit

VANE

None or thin cover quaternary deposit

Oldest granite partly aged

<600

<600 600–2000

the latter depicted as black dots enclosed by circles. The sites GOTA and TRAK are located on the roof of the same building and were operated in parallel for some time. Table 1 summarizes the geological situation of the GPS stations in terms of soil cover and bedrock, and the amount of ground water that can be withdrawn. This information is taken from the interactive map service of the Geological Survey of Sweden, SGU.2

2. Hydrological and Meteorological Observations Figure 2 shows time series and amplitude spectra of ground water observations at three sites in the larger Gothenburg area, together with observations of the river level (see Fig. 1 for the locations of these sites). The ground water observation sites are located several kilometers apart from the GPS stations. Unfortunately, we do not have access to any ground water observation sites close to the GPS stations in the city center of Gothenburg and directly at the SWEPOS stations. Furthermore, we do not have any detailed information on the screen depth of the three sites. The only information that we have is that the aquifers are shallow.

FA

GW52

2

0.5

1 GW53

0.4 GW52

0

0.3

−1 GW52

Amplitude (m)

0.6

0.2

−2

RL

RL −3 1999 2000 2001 2002 2003 2004 2005 2006 2007

1

0.1 2

10

3

10

10

04 10

Period (days)

10 10

−10

1050

−20 1000

Pressure (hPa)

0

T 5

Amplitude (hPa / deg. C)

20

P

9in x 6in

Temperature (deg. C)

15 T

R. Haas et al.

30

AOGS 2007 - SE Volume

Ground water and river level (m)

3

0.7

GW54

December 10, 2008 15:45

0.8 GW52

4

184

5

P 950 1999 2000 2001 2002 2003 2004 2005 2006 2007

1

10

2

3

10

10

04 10

Fig. 2. Top left: Time series of ground water and river level observations. Mean values are subtracted and the time series are oﬀset from each other to improve readability. See Fig. 1 for the location of the sites. Top right: Corresponding amplitude spectra. Bottom left: Time series of temperature and pressure observations at ONSA. Bottom right: Corresponding amplitude spectra.

b672-V13-ch12

Period (days)

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch12

Periodic Station Motion in Gothenburg Observed with GPS

185

Table 2. Amplitudes and phases of annual signals in ground water (GW), river level (RL), temperature (T ), and pressure (P ). Station GW52 GW53 GW54 RL T P -ONSA

Amplitude (Aannual )

Phase (Pannual )

26.4 ± 2.1 cm 32.0 ± 4.8 cm 20.0 ± 2.8 cm 8.5 ± 0.5 cm 8.9 ± 0.1 C 0.9 ± 0.3 hPa

234 ± 5 doy 228 ± 9 doy 251 ± 8 doy 298 ± 3 doy 210 ± 1 doy 145 ± 18 doy

(Aug. 22) (Aug. 16) (Sep. 08) (Oct. 25) (Jul. 29) (May 25)

The river-level station is located in Gothenburg harbor and is aﬀected not only by changes in river level but also by tidal eﬀects from the sea. We are aware that the missing information and the quality of the ground water and river level sites restrict our current study. The time series cover 1999 through 2006, where we also have GPS data for most of the sites in the network. The dotted vertical line in the amplitude spectra indicates the annual period. Clearly, annual periodicities are visible from all four data sets. The results of a least-squares ﬁt of annual signals in the time series are given in Table 2 as amplitude in centimeters and phase as day of the year (doy). Additionally, Fig. 2 shows time series and amplitude spectra of temperature and pressure at ONSA that can be regarded as representative for the region. The ground water levels peak in August and September, while the river level peaks in October. Temperature peaks already in July, while the pressure record has a low seasonal signal that peaks in May.

3. GPS Data Analysis and Results The GPS data were analyzed with the Bernese GPS software Version 53 using the network analysis strategy. The orbit and earth rotation products provided by the International GNSS Service (IGS)4 were used for the analysis, and station positions and velocities were a priori modeled in the IGS2000 reference frame. Ocean tide loading for all sites was modeled with parameters derived from the automated ocean tide loading provider6 using the GOT00.2 ocean tide model.7 Cycle slips were detected and removed, and the Quasi-Ionosphere Free (QIF) strategy was used to resolve the ambiguities using the IGS ionospheric model products. Tropospheric parameters were estimated and ionospheric free daily normal equations were

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

Median weighted root-mean square (wrms) values of the GPS data analysis.

Site

BISK FROL GOTA KORT TRAK BORA ONSA VANE

b672-V13-ch12

R. Haas et al.

186 Table 3.

9in x 6in

Roof-top Roof-top Roof-top Roof-top Roof-top SWEPOS SWEPOS SWEPOS

North component (mm)

East component (mm)

Up component (mm)

0.4 0.4 0.4 0.4 0.4 0.3 0.4 0.4

0.3 0.3 0.3 0.3 0.3 0.2 0.2 0.2

1.3 1.3 1.3 1.3 1.3 0.8 0.8 0.9

created. In the last step of the processing, these daily normal equations were stacked and station positions and velocities were estimated. Finally, daily station residuals were determined by Helmert transformations with respect to the combined solution. The median weighted root-mean square values of these three-dimensional residulas are given in Table 3 for all the stations in the network. They are smaller than half a millimeter for the horizontal components and between 0.8 and 1.3 mm for the Up components. In a post-processing step we removed a common-mode following the approach described in Ref. 8. Figure 3 shows time series of the resulting station residuals. Some stations experience seasonal signals in particular for the horizontal components. The ﬁve roof-top stations show larger variations in station position than the three SWEPOS stations. Furthermore, the Up components do show larger scatter than the horizontal components. Figure 4 shows amplitude spectra for all sites and components. Annual periods are indicated with vertical dotted line. Most sites and components show clearly annual signatures, where the stations GOTA and TRAK show the largest amplitudes, in particular for the horizontal components. Table 4 lists the annual amplitudes and phases derived from a least-squares ﬁt to the time series of station residuals. Figure 5 visualizes the horizontal station residuals only in a two-dimensional representation. In particular, the two stations GOTA and TRAK show large residuals with a preferred distribution in the North–East direction. The stations BISK and FROL show North and North–West orientation of their horizontal residuals, while the other stations show smaller horizontal residuals only, and not so clear preferred orientations. The results of a least-squares ﬁt to the horizontal components in terms of amplitudes and azimuth directions are given in Table 4.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Periodic Station Motion in Gothenburg Observed with GPS

b672-V13-ch12

187

Fig. 3. Station residuals after common-mode removal. Shown are North, East, and Up components in (mm) in gray, and the corresponding 30 days running mean values in black. The left scale is relevant for the station residuals. To improve visibility, the residuals of all stations are oﬀset by multiples of 5 mm from the ONSA residuals. Additionally, time series of ground water level, river level, and temperature are shown. Mean values are subtracted and oﬀsets are introduced to improve visibility. The units are dm for the water levels, and deka ◦ C for the temperature. For the river level (RL) and temperature, again the running 30 days average values are shown in black.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch12

R. Haas et al.

188

Fig. 3.

(Continued)

4. Conclusion and Outlook The analysis of GPS data in the Gothenburg region reveals for several of the roof-top stations signiﬁcant horizontal station motions with predominantly an annual period. These periodic horizontal motions reach the order of several millimeters and are largest for the roof-top stations that are mounted on buildings in the city center that are constructed on a clay layer (Table 5). The roof-top stations mounted on buildings that are constructed on bedrock show signiﬁcantly smaller horizontal motions, and the three SWEPOS stations show the smallest amplitudes. The predominating direction of the horizontal motions are North–East for the stations GOTA and TRAK, North for the station BISK, and North– West for the station FROL. The two stations TRAK and GOTA, located on the same roof, show very similar, however not identical station motions. These two stations were operated at diﬀerent periods, and the overlap of their operation periods is only slightly more than a year. Thus, completely identical results cannot be expected. The vertical station components show much more scatter than the horizontal ones. The detected annual signatures in the vertical components are much smaller than the horizontal signatures and hardly signiﬁcant (regarding the wrms values of the original GPS data analysis, see Table 3).

FA

22

20

20

20

18

18

18

16

16

GOTA 12 BISK 10 FROL 8

TRAK

14

Up Amplitude (mm)

14

East Amplitude (mm)

North Amplitude (mm)

16 TRAK

GOTA 12 BISK 10 FROL 8

KORT

GOTA 12 BISK 10 FROL 8

KORT

6

KORT

6 BORA

6 BORA

BORA

4 VANE

4 VANE

2

VANE

2 ONSA

2 ONSA

2

3

4

10

ONSA 2

3

10 10 Period (days)

4

10

0 1 10

2

3

10 10 Period (days)

4

10

189

Fig. 4. Amplitude spectra of the North, East, and Up components for the stations GOTA, TRAK, KORT, BISK, FROL, ONSA, VANE, and BORA, bottom to top. To improve visibility, the residuals of all stations are oﬀset by multiples of 3 mm from the GOTA spectra. The vertical dotted lines indicate annual periods.

b672-V13-ch12

10 10 Period (days)

0 1 10

9in x 6in

4

0 1 10

14

AOGS 2007 - SE Volume

TRAK

December 10, 2008 15:45

22

Periodic Station Motion in Gothenburg Observed with GPS

22

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch12

R. Haas et al.

190

Table 4. Amplitudes and phases of annual signals in the North, East, and Up components. AN (mm)

PN (doy)

AE (mm)

PE (doy)

AU (mm)

PU (doy)

2.2 ± 0.1 0.3 ± 0.1 1.9 ± 0.1 0.2 ± 0.1 2.7 ± 0.1 0.4 ± 0.1 0.3 ± 0.1 0.7 ± 0.1

164 ± 1 83 ± 8 137 ± 2 213 ± 10 166 ± 1 325 ± 3 346 ± 3 153 ± 2

0.2 ± 0.1 0.9 ± 0.1 6.2 ± 0.1 0.4 ± 0.1 8.6 ± 0.1 0.2 ± 0.1 0.4 ± 0.1 0.2 ± 0.1

104 ± 7 325 ± 5 150 ± 1 350 ± 4 161 ± 1 172 ± 3 352 ± 2 164 ± 4

1.1 ± 0.1 1.1 ± 0.1 1.4 ± 0.1 1.3 ± 0.1 0.8 ± 0.1 0.3 ± 0.1 0.4 ± 0.1 0.1 ± 0.1

252 ± 5 276 ± 4 160 ± 5 255 ± 5 68 ± 7 245 ± 5 62 ± 6 247 ± 7

Site

BISK FROL GOTA KORT TRAK BORA ONSA VANE

North (mm)

ONSA

VANE

KORT

15

15

15

10

10

10

10

5

5

5

5

0

0

0

0

−5

−5

−5

−5

−10

−10

−10

−10

−15 −15 −10 −5

−15 −15 −10 −5

−15 −15 −10 −5

0

5

10 15

FROL

North (mm)

BORA

15

0

5

10 15

BISK

0

5

10 15

−15 −15 −10 −5

GOTA

0

5

10 15

TRAK

15

15

15

15

10

10

10

10

5

5

5

5

0

0

0

0

−5

−5

−5

−5

−10

−10

−10

−10

−15 −15 −10 −5 0 5 10 15 East (mm)

−15 −15 −10 −5 0 5 10 15 East (mm)

−15 −15 −10 −5 0 5 10 15 East (mm)

−15 −15 −10 −5 0 5 10 15 East (mm)

Fig. 5.

Horizontal station residuals.

The detected horizontal station motions might be related to hydrological phenomena, i.e. ground water and river level changes in Gothenburg. These hydrological parameters also show a clear annual periodicity. The ground water levels observed at the three ground water level stations in the Gothenburg area peak in September, and the river level peaks in October. The peak epochs for the detected horizontal station motions do not show any clear pattern. There is no direct simple relation to the peak epochs of ground water level or river level.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch12

Periodic Station Motion in Gothenburg Observed with GPS Table 5.

Horizontal amplitudes and azimuth directions.

Site

BISK FROL GOTA KORT TRAK BORA ONSA VANE

191

Roof-top Roof-top Roof-top Roof-top Roof-top SWEPOS SWEPOS SWEPOS

Horizontal amplitude (mm)

Azimuth direction (degree)

2.2 ± 0.1 0.9 ± 0.1 6.5 ± 0.1 0.5 ± 0.1 9.0 ± 0.1 0.5 ± 0.1 0.5 ± 0.1 0.7 ± 0.1

5±1 298 ± 3 73 ± 1 60 ± 4 73 ± 1 327 ± 3 52 ± 2 12 ± 1

Inﬂuence of meteorological parameters cannot be excluded since there are also seasonal signals in temperature and air pressure. However, these parameters act more regionally and can hardly explain diﬀerential periodic motions. It appears to be more likely that local changes in river level and ground water cause the detected motions. Our study is unfortunately restricted by the relatively poor knowledge on the hydrological parameters. So far, we have only ground water and river level data for areas far away from the GPS stations, and it is not clear what screens the aquifers represent. Thus, it is not clear to what extent the hydrological information is representative for the situation close to the GPS stations, and a detailed study as for example presented by Ref. 5 is not possible at this point. We therefore currently try to get access to more suitable hydrological information. A further step will then be to develop a model that predicts hydrologically induced deformations for the Gothenburg region. The model predictions can then be compared to the observed station motions which will lead to a better understanding of the hydrological processes in Gothenburg region. Also, a combination and comparison of the results derived by GPS with InSAR results seem to be an attractive approach.

Acknowledgments The authors thank the Gothenburg municipality (“Stadsbyggnadskontoret”) for providing the GPS data of the roof-top stations in Gothenburg. They acknowledge the use of the interactive map service of the Geological Survey of Sweden (SGU). Ground water level data were kindly provided by SGU, and the river level data by the Swedish Meteorological and

FA

December 10, 2008 15:45

192

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch12

R. Haas et al.

Hydrological Institute (SMHI). The authors thank the reviewers for their constructive comments.

References 1. SWEPOS, A National network of reference stations for GPS, http://swepos. lmv.lm.se/english/index.htm. 2. The interactive map service of the Geological Survey of Sweden (SGU), http:// www.sgu.se/sgu/en/service/kart-tjanst start e.html. 3. R. Dach, U. Hugentobler, P. Fridez and M. Meindl (eds.), Bernese GPS Software Version 5.0 (AIUB Astronomical Institute, University of Bern, 2007). 4. J. M. Dow, R. E. Neilan and G. Gendt, The International GPS Service (IGS): Celebrating the 10th anniversary and looking to the next decade, Adv. Space Res. 36(3) (2005) 320–326, doi: 10.1016/j.asr.2005.05.125. 5. H. Munekane, M. Tobita and K. Takashima, Groundwater-induced vertical movements observed in Tsukuba, Japan, Geophys. Res. Lett. 31 (2004) L12608, doi: 10.1029/2004GL020158. 6. H.-G. Scherneck and M. S. Bos, Displacements due to ocean tide and atmospheric loading, (eds.) N. R. Vandenberg and K. D. Baver, Int. VLBI Service for Geodesy and Astrometry 2002 General Meeting Proc. NASA/CP2002-210002 (2002). 7. R. D. Ray, A global ocean tide model from TOPEX/POSEIDON altimetry: GOT99.2, NASA Technical Memorandum 209478 (1999). 8. S. Wdowinski, Y. Bock and J. Zhang, Southern California permanent GPS geodetic array, spatial filtering of daily positions for estimating coseimic and postseismic displacements induced by the 1992 Landers earthquake, J. Geophys. Res. 102(B8) (1997) 18057–18070.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch13

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

EVALUATION OF THE COSEISMIC PORE FLUID PRESSURE ON A THRUST FAULT JEEN-HWA WANG Institute of Earth Sciences, Academia Sinica PO Box 1-55, Nangang, Taipei, 115 Taiwan [email protected]

Frictional heating would result in either localized melting or a high heat ﬂow. Observations show the absence of anomalous heat ﬂow at the ground surface. This is the so-called heat ﬂux paradox. One of the possible reasons to produce this paradox is the reduction of the frictional stress due to a high pore pressure on a fault plane. Therefore, it is necessary to seek evidence of showing the existence of the pore ﬂuid pressure on a fault during earthquake rupture. In this chapter, a way to evaluate the coseismic pore ﬂuid pressure will be introduced. Included also is an example of the evaluation of the coseismic pore ﬂuid pressure at a depth of 1111 m on the Chelungpu fault, along which the 20 September 1999 Ms 7.6 Chi-Chi, Taiwan earthquake ruptured. The evaluated pore-ﬂuid factor is 0.93, thus leading to a coseismic pore ﬂuid pressure of 25.3 MPa. This indicates that the fault plane was suprahydrostatic at this depth during the earthquake.

1. Introduction After an earthquake ruptures, the frictional stress, σ(t), which is a function of time and slip on a fault plane decreases from an initial σo to a dynamic σd , and ﬁnally becomes σf (see Fig. 1). In general, σd is equal to or smaller than σf .1 According to the slip- and rate-weakening frictional law, the frictional stress changes from σo to σd in a characteristic slip displacement, Dc .2,3 The static stress drop ∆σs = σo − σf and the dynamic stress drop ∆σd = σo − σd are usually used to specify the change of stresses on a fault. The strain energy, ∆E, per unit area, which results from tectonic loading, can be approximated by the area of a trapezoid underneath the linearly decreasing function of stress versus slip (Fig. 1). ∆E is transferred into, at least, three parts (see Fig. 1), i.e. the seismic radiation energy (Es ), fracture energy (Eg ), and frictional energy (Ef ), that is ∆E = Es +Eg +Ef . Es is the energy radiated through seismic waves. Eg is the energy used to extend the 193

FA

December 10, 2008 15:45

194

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch13

J.-H. Wang

Fig. 1. The stress–slip function: lines AC and CD represent slip-weakening friction, Dc = the characteristic slip displacement, Dmax = the maximum slip, σo = initial stress (or static frictional stress), σd = dynamic frictional stress, and σf = ﬁnal stress. The strain energy, ∆E, per unit area is the area of a trapezoid below line AD, Es = seismic radiation energy, Eg = fracture energy, and Ef = frictional energy.

fault plane. Ef , which results from the dynamic friction stress, can generate heat. Because of incomplete data, there are high uncertainties in measuring the energies, especially for Ef . Wang4,5 proposed ways to evaluate ∆E, Eg , and Ef . Several authors6−9 suggested that frictional dissipation might be high enough early in the rupture to melt the wall rock, thus reducing the frictional stress for the remainder of an event. Under the same tectonic stress, it would be harder to initiate an earthquake on a fault with a higher frictional strength than that with a lower frictional strength. A way to allow for a larger-sized event is that there is a large frictional strength on the fault plane. Under high dynamic friction, a large earthquake would generate a great amount of heat due to rubbing of two fault surfaces. Conventionally, it is assumed that this heat could not be easily dissipated in a relatively insulating earth and would accumulate under repetition of events. This would result in either localized melting or a high heat ﬂow at the ground surface. But, observations over the entire state of California, USA, showed the absence of anomalous heat ﬂow at the ground surface.10−13 This is the so-called heat ﬂux paradox.11 After the Chi-Chi earthquake, the feat ﬂow is also low in the source area.5,14 The low heat ﬂow should result from a remarkable decrease in the eﬀective frictional strength. To decrease the eﬀective frictional strength for the initiation of rupture

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch13

Evaluation of the Coseismic Pore Fluid Pressure on a Thrust Fault

195

on a fault with a high frictional strength or to reduce dynamic friction for maintaining rupture, diﬀerent mechanisms, including hydrodynamic lubrication,15 thermal pressurization,16−18 ﬂash heating,19 gel formation,20 melting lubrication21 etc., have been proposed. Except for ﬂash heating and melting lubrication, the pore ﬂuid pressure must be taken into account. Hence, the presence of a high pore pressure is commonly assumed to be a signiﬁcant factor in reducing the frictional stress, thus causing a low heat ﬂow. To resolve this problem, it is necessary to seek evidence to show the existence of ﬂuid pressure state on a fault plane. Wang5 proposed a two-dimensional (2D) model, in the presence of friction and a pore ﬂuid pressure, to link mechanical and thermal properties together for thrust faulting. Together with a one-dimensional (1D) heat conduction model,22 he evaluated the average coseismic pore ﬂuid pressure on the Chelungpu fault, along which the 1999 Chi-Chi earthquake ruptured, in central Taiwan23,24 from the source parameters inferred from seismograms and GPS data. In this study, an attempt is made to describe Wang’s model.5 In addition, an example to demonstrate the evaluation of pore ﬂuid pressure on the fault plane at a depth of 1111 m during the earthquake on the basis of this model using the measured temperature rise with thermal and mechanical parameters measured and inferred from the core samples of the TCDP hole cutting the fault plane by Kano et al.14

2. One-Dimensional Conduction Equation of Heat On a fault plane with an area of A and an average displacement D, the frictional energy caused by the dynamic friction stress, σd , is Ef = σd DA, which will result in a temperature rise, ∆T . Heat is assumed to be distributed within a layer of thickness h around the ruptured plane and thus average ∆T is given by1 ∆T = Ef /Cv ρAh,

(1)

where Cv and ρ are, respectively, the speciﬁc heat and density. Deﬁne Q = ∆T • h to be the strength of a heat source. From Eq. (1), we have Q = σd D/Cv ρ.

(2)

When the heat source with strength of Q is located at x = 0, the heat will conduct outward, and thus the temperature decreases. The temperature

FA

December 10, 2008 15:45

196

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch13

J.-H. Wang

rise at position x and time t is described by a 1D conduction equation22 : ∆T (x, t) = Q · exp(−x2 /4αt)/2(παt)1/2 ,

(3)

where α is the thermal diﬀusivity (in m2 /s).

3. Two-Dimensional Faulting Model Wang5 proposed a 2D faulting model to link the heat and stress ﬁeld with pore ﬂuid pressures on a fault plane. Here, only a brief explanation is given. As depicted in Fig. 2, the horizontal stress σ1 = ρgH + σT , where ρgH and σT are, respectively, the lithostatic pressure at a depth of H and an additional tectonic stress, and the vertical stress σ3 = ρgH. The normal and shear stresses are σn and σs , respectively, on the fault plane with a dip angle of θ. The relation between σn and σs is of the form: |σs | = µf (σn − pw ), where µf and pw are the frictional coeﬃcient and pore ﬂuid pressure, respectively. The hydrostatic pressure equals ρw gH, where ρw is the water density (=1000 kg/m3 ) and g = 9.8 m/s2 . Let pw be γρgH, where γ equals ρw /ρ and is called the pore-ﬂuid factor.25 At shallow depths, the ﬂuid gradient is hydrostatic and thus γ is 0.4. At depths, the pore ﬂuid pressure may become suprahydrostatic and thus γ > 0.4, because ρ is 2500 kg/m3 for crustal materials. When a fault breaks, the static frictional stress, σo , drops to the dynamic frictional stress, σd , and thus µf changes from the static to dynamic frictional coeﬃcient.

Fig. 2. The dipping fault with a dip angle of θ. The depth is denoted by H. The principal stresses along the horizontal and vertical axes are σ1 and σ3 , respectively. σn and σs , respectively, represent the normal stress and shear stress.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch13

Evaluation of the Coseismic Pore Fluid Pressure on a Thrust Fault

197

Based on Anderson’s theory of faulting for a thrust fault,26 σn and σs are, respectively: σn = ρgH[1 − µf (1 − γ) cos(2θ)]/[(1 + µ2f )1/2 − µf ],

(4)

σs = −µf ρgH(1 − γ) sin(2θ)/[(1 + µ2f )1/2 − µf ].

(5)

It is noted that the absolute value of σs is σo . In order to link Eqs. (2) and (5) together, a relation between σd and σo is needed. Wang5 assumed σd = ξσo , where ξ ≤ 1. The experimental result by Byerlee27 led to ξ = 0.83. The rate- and state-dependent friction law28 is: µv = µo + a ln(v/vo ) + b ln(ϕvo /Dc ), where µv and µo are the frictional coeﬃcients, respectively, at the sliding velocity, v, and a reference velocity, vo , and ϕ is a state variable. The two constants a and b represent, respectively, the direct and evolution eﬀects.3 For the two one-state-variable friction laws: dϕ/dt = −(ϕv/Dc ) ln(ϕv/Dc ) and dϕ/dt = 1 − ϕv/Dc , the ss steady-state frictional coeﬃcient, µss v , is µo + (a − b) ln(v/vo ). Hence, µv ss ss is µ1 = µo + (a − b) ln(v1 /vo ) at v = v1 and µ2 = µo + (a − b) ln(v2 /vo ) ss at v = v2 . This gives µss 1 − µ2 = (a − b) ln(v1 /v2 ). The average a − b is ss about −0.005 for frictional weakening,29 thus giving µss 1 − µ2 = 0.11 when −10 m/s, which is the long-term average sliding velocity v1 = 1.58 × 10 across the fault,30 and v2 = 1 m/s, which is the average peak ground velocity (PGV) in the northern segment.31 At low v, the dynamic frictional ss ss coeﬃcient is close to the static one. This suggests ξ ≈ µss 2 /µ1 ≈ (µ1 − ss ss ss 0.11)/µ1 = 1−0.11/µ1 . When µ1 = 0.55, which is the frictional coeﬃcient in the MSZ measured by Lockner et al.,32 we have ξ ≈ 0.8. Thus, the two results deﬁnitely suggest ξ ≈ 0.8. For comparison, ξ = 0.9 and 1.0 are also taken into account in the following calculations. Inserting the absolute value of Eq. (5) into Eq. (2) with σd = ξσo leads to Q = ξµf (1 − γ)gDH sin(2θ)/Cv [(1 + µ2f )1/2 − µf ].

(6)

The equality ∆T = Q/h gives ∆T = ξµf (1 − γ)gDH sin(2θ)/hCv [(1 + µ2f )1/2 − µf ].

(7)

Obviously, Eq. (7) links the heat represented by temperature rise and the stress ﬁeld with a pore ﬂuid pressure on the fault plane. Equation (6) leads to the eﬀective frictional coeﬃcient: µf (1 − γ) = (QCv /ξgDH)(1 + µ2f )1/2 [(1 + µ2f )1/2 − µf ],

(8)

FA

December 10, 2008 15:45

198

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch13

J.-H. Wang

because of sin(2θ) = (1 + µ2f )1/2 . The left-hand side of Eq. (8) is a linear function of µf with a slope of (1 − γ), while the right-hand side denoted by F (µf ) is a nonlinear function of µf . An example to show Eq. (8) is displayed below.

4. An Example On 20 September 1999, the Ms 7.6 Chi-Chi earthquake ruptured the Chelungpu fault, which is a ∼100-km-long and east-dipping thrust fault, with a dip angle of 30◦ , in central Taiwan.23,24 The epicenter and the surface trace of the fault are displayed in Fig. 3. Based on the lubrication model, Brodsky and Kanamori15 assumed that during the earthquake lubrication occurred on the northern segment, but not on the southern

Fig. 3. The epicenter (in a solid star), the Chelungpu fault (in a solid line), and the borehole site of TCDP (in solid circle).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch13

Evaluation of the Coseismic Pore Fluid Pressure on a Thrust Fault

199

one. Ma et al.33 inferred a low value of critical lubrication length, Lc , of the model associated with high lubrication during faulting. The results of the two groups of researchers strongly suggest the existence of ﬂuids on the fault plane, at least in the northern segment, during the event. Wang5 suggested that the whole fault plane was suprahydrostatic during faulting. In 2005, the Taiwan Chelungpu-fault Drilling Project (TCDP) was conducted, and two deep holes were drilled.34 The two deep holes were 40 m apart: Hole A with a depth of 2000 m and hole B with a depth of 1300 m. The solid circle in Fig. 3 displays the localities of the two holes. Continuously coring and geophysical well-loggings were made at the two holes. Kano et al.14 measured temperature, with a resolution of 0.003◦C, inside hole A in September 2005, 6 years after the earthquake, i.e. t = 1.9 × 108 s, and constructed a spatial distribution of temperature rise, ∆T , between −40 m and +40 m with reference to the fault plane, on which ∆T = 0.06◦ C is the maximum value. Tanaka et al.35 assumed that the ﬂuctuations of temperature measured by Kano et al.14 could be due to the variation in thermal conductivity between fault and wall rocks. This would reduce the representative of residual heat from faulting. Nevertheless, the temperature measurement made by Kano et al.14 is still taken to be an example to show how to evaluate the pore ﬂuid pressure from temperature rise, because the eﬀect proposed by Tanaka et al.35 is not big. Of course, there are some uncertainties in the evaluated result. On the basis of the 1D conduction equation as described above, they used a few modeled spatial distributions of ∆T to ﬁt the observed data. The optimum values of the heat strength, Q, and the thermal diﬀusivity, α, are 1.5◦ C m and 3.4 × 10−7 m2 /s, respectively. The values of D and Cv used by them are 5 m and 1700 J/kg◦C, respectively. From laboratory measurements, Tanaka et al.36 obtained α = 1.0 × 10−6 m2 /s in the MSZ. Obviously, α is about three times larger than that by Kano et al.14 From Eq. (3) at x = 0, Q is about 1.7 times larger than that inferred by Kano et al.14 Equation (8) leads to γ = 1 − Q{Cv (1 + µ2f )1/2 [(1 + µ2f )1/2 − µf ]/µf ξgDH}.

(9)

Substituting the values of related parameters into Eq. (9) leads to γ = 1 − 4.481 × 10−3 Q.

(10)

Equation (10) shows that the diﬀerence in γ caused by diﬀerent values of Q related to the diﬀerent values of α is δγ = 4.404 × 10−3 . Since this value is

FA

December 10, 2008 15:45

200

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch13

J.-H. Wang

small, the eﬀect of α on evaluation of γ is negligible. In addition, Eq. (10) leads to δγ = −(0.448%)δQ, thus indicating that for the present case to make a decrease in γ from 1 to 0.4, Q must increase from 0 to 134◦ C m. In other words, only a large increase in Q can result in a change of pore ﬂuid pressure on the fault from suprahydrostatic to hydrostatic. This suggests that the above-mentioned eﬀect on temperature proposed by Tanaka et al.35 cannot produce a total change of the state of pore ﬂuid pressure on the fault plane. Figure 4 shows the variations of µf (1 − γ) versus µf for γ = 0.1 − 0.9, with a unit of 0.1, and γ = 0.95. The variations are displayed with thin solid lines. The numbers of 0.4–0.95 denote the values of γ. On the other hand, the functions of F (µf ) are shown by three solid curves for ξ = 0.8, 0.9, and 1.0. The value of F (µf ) decreases with increasing ξ. The intersection points between µf (1 − γ) and F (µf ) are the solutions of Eq. (8). The dashed line denotes µf = 0.58 associated with a dip angle of 30◦ .14 From the intersection points between the dasehed line and the three curves for F (µf ), the value of γ should be between 0.9 and 0.95, and γ increases with ξ. This indicates that the coseismic pore ﬂuid pressure was suprahydrostatic. When ξ is taken to be 0.8, γ is 0.93, thus giving pw = 25.31 MPa. The hydrostatic pressure, ph , at 1111 m is 10.89 MPa. Obviously, pw is greater than ph . This indicates the existence of a suprahydrostatic state during faulting. This is consistent with the results inferred by Wang5 for the whole

Fig. 4. The variations of µf (1 − γ) versus µf for γ = 0.1 − 0.9, with a unit of 0.1, and γ = 0.95 with thin solid lines. The functions of F (µf ) are shown by three solid curves for ξ = 0.8, 0.9, and 1.0. The numbers, i.e. 0.4–0.95, denote the values of γ. The dashed line denotes µf = 0.58.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch13

Evaluation of the Coseismic Pore Fluid Pressure on a Thrust Fault

201

Chelungpu fault plane. The suprahydrostatic pressure could reduce the eﬀective frictional coeﬃcient and, thus, decreases frictional heating on a fault plane. This study suggests a direct way to evaluate the coseismic pore ﬂuid pressure and also provide signiﬁcant evidence to explain the cause of a low heat ﬂow on a fault plane. Thus, the present result seems to be able to resolve the heat ﬂow paradox.

5. Conclusions Wang5 proposed a useful method to evaluate the coseismic pore ﬂuid pressure for thrust faulting. On the basis of a 1D conduction equation and 2D faulting model, the heat strength and pore ﬂuid pressure on the Chelungpu fault plane at a depth of 1111 m are evaluated from the temperature rise measured in and thermal and mechanical constants inferred from the core samples of the TCDP hole cutting the fault plane. Results show that the heat strength is 1.50◦C m. The pore-ﬂuid factor is γ = 0.93, thus giving pw = 25.3 MPa. This indicates the existence of a suprahydrostatic state during faulting.

Acknowledgments The author would like to express his thanks to Dr K. Satake (Volume Editor) and two reviewers for valuable comments. This study was ﬁnancially supported by Academia Sinica and the National Sciences Council under Grant Nos. NSC94-2119-M-001-016 and NSC96-2116-M-001-012-MY3.

References 1. H. Kanamori and T. H. Heaton, Geocomplexity and Physics of Earthquakes, eds. J. B. Rundle, D. L. Turcotte and W. Klein, Geophysics Monograph, Vol. 120 (AGU, Washington DC, 2000), p. 147. 2. C. Marone, Ann. Rev. Earth Planet. Sci. 26 (1998) 643. 3. J.-H. Wang, Bull. Seism. Soc. Am. 92 (2002) 87. 4. J.-H. Wang, Geophys. Res. Lett. 32 (2004) L06316, doi: 10.1029/ 2004GL021884. 5. J.-H. Wang, J. Geophys. Res. 111 (2006) B11312, doi: 10.1029/ 2005JB004018. 6. H. Jeﬀreys, Geol. Mag. 79 (1942) 291. 7. D. P. McKenzie and J. Brune, Geophys. J. Roy. Astron. Soc. 29 (1972) 65. 8. P. G. Richards, Bull. Seism. Soc. Am. 66 (1976) 1.

FA

December 10, 2008 15:45

202

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch13

J.-H. Wang

9. R. K. Cardwell, D. S. Chinn, G. F. Moore and D. L. Turcotte, Geophys. J. Roy. Astron. Soc. 52 (1978) 525. 10. T. L. Henyey and G. J. Wasserburg, J. Geophys. Res. 76 (1971) 7924. 11. A. H. Lachenbruch and J. H. Sass, J. Geophys. Res. 85 (1980) 6185. 12. A. H. Lachenbruch and J. H. Sass, Geophys. Res. Lett. 15 (1988) 981. 13. A. H. Lachenbruch and J. H. Sass, J. Geophys. Res. 97 (1992) 4995. 14. Y. Kano, J. Mori, R. Fujio, H. Ito, T. Yanagidani, S. Nakao and K. F. Ma, Geophys. Res. Lett. 33 (2006) L14306, doi: 10.1029/2006GL026733. 15. E. E. Brodsky and H. Kanamori, J. Geophys. Res. 106 (2001) 16357. 16. A. Bizzarri and M. Cocco, J. Geophys. Res. 111 (2006) B05303, doi: 10.1029/ 2005JB003862. 17. A. Bizzarri and M. Cocco, J. Geophys. Res. 111 (2006) B05304, doi: 10.1029/ 2005JB003864. 18. Y. A. Fialko, J. Geophys. Res. 109 (2004) B01303, doi: 10.1029/ 2003JB002496. 19. J. R. Rice, J. Geophys. Res. 111 (2006) B05311, doi: 10.1029/2005JB004006. 20. D. L. Goldsby and T. E. Tullis, Geophys. Res. Lett. 29 (2002) 1844, doi: 10.1029/2002GL015240. 21. J. G. Spray, J. Geophys. Res. 98(B5) (1993) 8053. 22. C. B. Oﬃcer, Introduction to Theoretical Geophysics (Berlin, Springer, 1974). 23. K.-F. Ma, C.-T. Lee, Y.-B. Tsai, T.-C. Shin and J. Mori, Eos, Trans. AGU 80 (1999) 605. 24. T.-C. Shin, Terres. Atmos. Ocean. Sci. 11 (2000) 555. 25. R. H. Sibson, Tectonophysics 211 (1992) 283. 26. D. L. Turcotte and G. Schubert, GEODYNAMICS — Applications of Continuum Physics to Geological Problems (Wiley, 1982). 27. J. D. Byerlee, J. Geophys. Res. 72 (1967) 3639. 28. A. Ruina, J. Geophys. Res. 88 (1983) 10359. 29. G. H. Scholz, Nature 391 (1998) 37. 30. J.-H. Wang, Geophys. Res. Lett. 31 (2004) L10613, doi: 10.1029/ 204GL019417. 31. J.-H. Wang, Terres. Atmos. Ocean Sci. 17 (2006) 179. 32. D. A. Lockner, C. Morrow, S.-R. Song, S. Tembe and T.-F. Wong, Eos, Trans. AGU 86(52), Fall Meet. Suppl. (2005) T43D-04, F1825. 33. K.-F. Ma, E. E. Brodsky, J. Mori, C. Ji, T.-R. Song and H. Kanamori, Geophys. Res. Lett. 30(1244) (2003) doi: 10.1029/2002GL015380. 34. K.-F. Ma, S.-R. Song, H. Tanaka, C.-Y. Wang, J.-H. Hung, Y.-B. Tsai, J. Mori, Y.-F. Song, E.-C. Yeh, H. Sone, L.-W. Kuo and H.-Y. Wu, Nature 444 (2006) 473. 35. H. Tanak, W. M. Chen, K. Kawabata and N. Urata, Geophys. Res. Lett. 31 (2007) L10613, doi: 10.1029/204GL019417. 36. H. Tanaka, E. Chen, K.-F. Ma and C.-Y. Wang, ABSTRACT 2006 Western Pacific Geophysics Meeting (2006).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch14

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

THE WATER LEVEL CHANGES OF THE ANEUKLAOT LAKE, WEH ISLAND AFTER THE 2004 SUMATRA–ANDAMAN EARTHQUAKE AGUSTAN∗,†,‡ , DJOKO NUGROHO† , LENA SUMARGANA† , IRWAN MEILANO∗,‡ , MOHD. EFFENDI DAUD∗ , FUMIAKI KIMATA∗ and YUSUF S. DJADJADIHARDJA† ∗ Research Center for Seismology Volcanology and Disaster Mitigation (RSVD), Nagoya University, Japan † Center

of Technology for Natural Resources Inventory (PTISDA) Agency for the Assessment and Application of Technology (BPPT), Indonesia

‡ Geodesy

Research Division, Bandung Institute of Technology, Indonesia ‡ [email protected]

The phenomenon of water level decrease dramatically happened in Aneuklaot Lake, Weh Island, Aceh Province, Indonesia until June 2005. The water level is continuously decreasing with the rate of 1 cm per day in June–July 2005. A combination measurement consisting of GPS observation, geoelectrical observation, and hydroclimate observation is conducted around the lake. From observation and data analysis, it is found that the water level changes may be aﬀected by some factors, including climate changes, hydrogeology, and post-seismic deformation after the giant 2004 Sumatra–Andaman earthquake. This investigation shows that the post-seismic deformation can be detected by GPS observation and ﬁts with logarithmic decaying model. The horizontal displacement varies from 14 mm to 18 mm in 69 days, whereas there is almost no displacement in vertical components. Moreover, from geoelectrical data analysis, it is found that there are at least three minor faults detected around the lake which is indicated by the extremely diﬀerent resistivity around the location that was observed. This investigation ﬁnds that climate changes and hydrogeology process changes caused by post-seismic deformation, contribute to the water level changes of the Aneuklaot Lake.

1. Introduction The Aneuklaot Lake is located in Weh Island, a small active volcanic island to the northwest of Sumatra, Aceh Province, Indonesia (Fig. 1). It is located approximately 36 km in straight line to the north direction from 203

FA

December 10, 2008 15:45

204

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch14

Agustan et al.

Fig. 1. The location map of the Weh Island, Aneuklaot Lake, and Sumatran Fault. Locations of the fault is referred to in Ref. 5.

Bandaaceh, the capital city of Aceh Province. This island is categorized as a staratovolcano with more than 10 km in diameter.1 As a freshwater resource in Weh Island, the Aneuklaot Lake is a very important asset for local people who lived in the island, especially for Sabang City, the capital of Weh Island. The area of Weh Island is approximately 153.6 km2 and the lake is approximately 0.476 km2 according to a survey conducted in 2001. However, from the second survey based on satellite image, it is found that the lake’s area is reduced to approximately 0.397 km2 . A few months after the giant 2004 Sumatra–Andaman earthquake incident, the water level of Aneuklaot Lake decreased, surprising local people. The local government initiated the observation of water level changes from May 2005 to June 2006, and found that the rate of change reached 1 cm a day (Fig. 2). To investigate this phenomenon, an integrated survey and data analysis were conducted. This investigation was initiated by the local government and funded by the Agency of the Rehabilitation and Reconstruction for the Region and Community of Aceh and Nias (BRR).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Water Level Changes of the Aneuklaot Lake

b672-V13-ch14

205

Fig. 2. The water level changes observed 142 days after the 2004 Sumatra–Andaman earthquake.

The Center of Technology for the Natural Resources Inventory — Agency of the Assessment and Application of Technology (PTISDA-BPPT) led the research collaboration between Geodesy Research Division — Bandung Institute of Technology (ITB) and Research Center for Volcanology, Seismology, and Natural Disaster Mitigation — Nagoya University (RSVD). The research collaboration mainly assessed the impact of the 2004 Sumatra–Andaman earthquake, especially in detecting the postseismic deformation. For that purpose, 3 GPS campaigns were conducted in Weh Island. Some experts have studied the relation of earthquake to the water level changes, e.g. Roeloﬀs4 studied the Landers earthquake 1992 and concluded that there is a persistent water level change due to earthquakes; Montgomery and Manga3 summarized that the water level changes are related to the earthquake. Furthermore, they also illustrated that the earthquakes trigger crustal stress and ground shaking; the crustal stress then can generate aquifer strain and consolidation, whereas ground shaking, in addition to consolidation can generate fracturing and fracture clearing also. The aquifer strain, consolidation, and fracturing ﬁnally can inﬂuence the stream ﬂow and water level or pore pressure. These studies found that the earthquake’s rupture could inﬂuence the geological structure underneath or upper-crust and that they had relation to the aquifer system.

FA

December 10, 2008 15:45

206

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch14

Agustan et al.

However, the studies were based on well-planned observation system using monitored well. In addition, the deformation detected by Global Positioning System (GPS) observation due to the 2004 Sumatra–Andaman earthquake has been studied by Vigny et al.7 and Subarya et al.6 One of the important points that can be derived from these studies is that the deformation occurring over a period of a year might be dominated by poroelastic rebound, and thereafter by visco-elastic or plastic ﬂow from low-viscosity shallow earth layers. The basic assumptions are that the climate changes, hydrogeological processes, and ground deformation have a contribution for the changes of the water level. This paper will describe the ﬁeld observation activities and data analysis to understand the water level changes in Aneuklaot Lake.

2. Methods To investigate the water level changes in the lake, the water balance was used as the mainframe of the analysis. The water balance can be written as Input = Output, where input includes run oﬀ, precipitation including direct input such as rain and river charge; and output includes daily consumption, irrigation, evaporation, and evapotranspiration. To calculate the water balance, the integration of terrestrial and extraterrestrial method was needed. The remote sensing technique was used to monitor the land cover change and then would be analyzed to describe the eﬀect of the land cover change to the catchment area. The remote sensing techniques utilized the Ikonos and Quickbird images to monitor the land cover changes around the catchment area. Image from 2001 acquisition was compared to the image acquired in 2005. The deformation of the geometry of the lake could also be observed by using this technique. The hydroclimate analysis was used to investigate the input–output system of the groundwater for the lake, and mainly utilized the rainfall data around the lake from 1990 to 2006, sunshine duration data, temperature, relative humidity, wind gust, evaporation, cloud cover, and hydrology observation. This data was collected from direct ﬁeld observation as primary data and the historical data was the data recorded by the local meteorological station.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Water Level Changes of the Aneuklaot Lake

b672-V13-ch14

207

The geological data analysis also used to investigate the geological structure around the lake, completed with the geoelectric survey and data analysis to investigate the fault structure. The geodetic survey and data analysis were used to investigate the point movement on the surface around the lake. The geological data analysis mainly utilized the existing geological map to determine the aquifer system and the impermeable zone around the lake. This analysis was also completed with geomorphological mapping. Other terrestrial surveys were the 2D geoelectric resistivity in 7 lines and 3 times GPS campaign observation. The ﬁrst GPS campaign was conducted in October 2006 to develop the new benchmark, and reobservations were again carried out in November 2006 and February 2007. GPS data from continuous station at Syah Kuala University has been processed to support the investigation in post-seismic deformation. The illustration of the Weh Island setting, the geoelectric lines survey, and GPS stations conﬁguration can be seen in Fig. 3.

Fig. 3.

The conﬁguration of the GPS stations and the lines of geoelectric survey.

FA

December 10, 2008 15:45

208

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch14

Agustan et al.

In order to investigate the post-seismic deformation, after slip model by Marone et al.2 was used as a frame to model the deformation. This method was chosen since the post-seismic deformation would diminish in the logarithmic decaying model through the time: Up (t) = α ln((β/α)t + 1) + Vot + Uref ,

(1)

where Up (t) = the post-seismic deformation time series (in meter); t = time (in days); α = the parameter of the velocity-strengthening layer; β = is the co-seismic velocity; Vo = steady state velocity (in meter per day); and Uref = oﬀset of displacement at the start of observation (in meter).

3. Results Satellite image comparison between 2001 and 2005 shows, that in 4 years the land cover changes around the lake mainly from forest to bush or from forest to settlement. However, the changes occurred in a small area. Therefore, the impact of the land cover change to the catchment area could be neglected. The hydroclimate analysis gives the quantiﬁcation of the water balance components. From 16 years’ data analysis, the rainfall tends to decrease, especially from 2000 to 2006. On the other hand, the evaporation tends to increase, and this condition has direct impact on the water balance (Fig. 4).

Fig. 4. Observed monthly rainfall since 1990, evaporation since 2000 in Sabang and calculated monthly water balance from the rainfall and evaporation since 2000.

FA

December 10, 2008 15:45 AOGS 2007 - SE Volume 9in x 6in

Water Level Changes of the Aneuklaot Lake

b672-V13-ch14

209

Fig. 5. Resistivity models along the lines of geoelectric surveys: (a) along line 2 from West to East direction; (b) along line 4 from South to North direction, and (c) along line 6 from North to South direction.

FA

December 10, 2008 15:45

210

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch14

Agustan et al.

In quantiﬁcation, there was a deﬂation of rainfall by approximately 22% from 2000 to 2005. However, this phenomenon cannot be categorized into climate change, because this is just rainfall annual variability which is common in tropical region. The phenomenon of the decreasing rainfall from 2000 to 2005 also occurred during 1990–1996. The results of the sun duration, temperature, relative humidity, and wind gust data analysis also show that the trend is increasing. The intensity of the sun radiation reﬂects the potential energy that is available for evapotranspiration process. The intensity of sun radiation is also directly related to the temperature and humidity. In addition, the consumption of water from the lake for daily life has also been increasing from time to time. In general, the hydroclimate analysis shows that the potential of input to the lake tends to decrease. From geoelectric survey around the lake, it was found in three lines that there was an extreme diﬀerence in resistivity value of the structure underneath. Low resistivity region is suspected to be caused by the water content of the structure, and it might come from the lake through the small fracture. This fact supports the previous study3 that said that the ground shaking from an earthquake might lead to fracture. These lines were located on the northern part (line numbers 2 and 4) and southwest of the lake (line number 6). The results of the geoelectrical survey can be seen in Fig. 5. There were six new benchmarks developed and observed using GPS techniques; however, due to some limitation factors, only two GPS stations (SBGN and SBGS) could be reobserved and processed. Dual-frequency of GPS receiver was used, and the data was processed using Bernese 5.0 GPS data processing software. The results of GPS data processing show that station 1 has a horizontal displacement of 14 mm to the northwest direction, whereas station 2 has a horizontal displacement of 18 mm to the southwest in 69 days (Fig. 6). The vertical displacement for both stations is almost zero and is not countable.

4. Discussions A comprehensive and integrated data analysis shows that the hydroclimate factor is the major impact of the water level changes in Aneuklaot Lake. However, results from geoelectric and geodetic observation support the assumption of the subsurface structure that is aﬀected by the 2004 Sumatra–Andaman earthquake. The small fracture which was detected by

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Water Level Changes of the Aneuklaot Lake

b672-V13-ch14

211

Fig. 6. The horizontal displacements referred to Bandaaceh site based on GPS measurements in the period of 69 days since November 2006 to February 2007.

the geoelectric method, however, is still not clear whether it is caused by the earthquake event or had been there long time before the event. The post-seismic deformation could be modeled by using the continuous GPS data from the Syah Kuala University, Bandaaceh. The result of continuous GPS data processing shows that the post-seismic deformation decay ﬁts with the afterslip model. The post-seismic deformation from this model shows the relation to the water level change in the time range reported (Fig. 7). As additional information, the Aneuklaot Lake is located approximately 350 km away from the 2004 Sumatra–Andaman earthquake’s epicenter. There is a relation between the epicenter location and magnitude to water level. For distance in hundreds to thousands of kilometers from the epicenter (far-ﬁeld) there may be a delayed response in water level changes,3 such as the situation that had occurred in this case. The water level

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

212

9in x 6in

b672-V13-ch14

Agustan et al.

Fig. 7. The post-seismic deformation observed by GPS measurements in Bandaaceh GPS Continuous Station. The vertical dash line represents the time range of water level observation in Aneuklaot Lake.

changes were detected, 146 days after the earthquake. However, according to Montgomery and Manga,3 the cause of the delay response was not well understood.

Acknowledgments The authors would like to express their gratitude to the research staﬀ of PTISDA-BPPT for their contribution to this research, Prof. Teruyuki Kato, members of SE-19 AOGS 2007, Yasuyuki Kano, and an anonymous reviewer for giving comments on this paper.

References 1. D. H. Natawidjaja, PhD Thesis, California Institute of Technology, Pasadena, California, 2002. 2. C. J. Marone, C. H. Sholz and R. Bilham, J. Geophys. Res. 96 (1991) 8441. 3. D. R. Montgomery and M. Manga, Science 300 (2003) 2047. 4. E. A. Roeloﬀs, J. Geophys. Res. 103 (1998) 869. 5. K. Siehand and D. H. Natawidjaja, Map (California Institute of Technology, Pasadena, California, 2004).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Water Level Changes of the Aneuklaot Lake

b672-V13-ch14

213

6. C. Subarya, M. Chlieh, L. Prawirodirdjo, J.-P. Avouac, Y. Bock, K. Sieh, A. J. Meltzner, D. H. Natawidjaja and R. McCaﬀrey, Nature 440 (2006) 46. 7. C. Vigny, W. J. F. Simons, S. Abu, R. Bamphenyu, C. Satirapod, N. Choosakul, C. Subarya, A. Socquet, K. Omar, H. Z. Abidin and B. A. C. Ambrosius, Nature 436 (2005) 201.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

This page intentionally left blank

b672-V13-ch14

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch15

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

EFFECT OF NEAR-SOURCE TRENCH STRUCTURE ON TELESEISMIC BODY WAVEFORMS: AN APPLICATION OF A 2.5D FDM TO THE JAVA TRENCH T. OKAMOTO Department of Earth and Planetary Sciences Tokyo Institute of Technology, 2-12-1 Ookayama Meguro, Tokyo 152-8551, Japan [email protected] H. TAKENAKA Department of Earth and Planetary Sciences, Kyushu University 6-10-1 Hakozaki, Higashi-ku, Fukuoka 812-8581, Japan [email protected]

We study strong eﬀect of near-source structure on teleseismic body waveforms from two well-recorded aftershocks of the 2006 Java tsunami earthquake. We assume a two-dimensional, heterogeneous model of the Java trench structure on the basis of recent seismic surveys. We compute synthetic waveforms by an eﬃcient “2.5-dimensional” (2.5D) ﬁnite-diﬀerence method. A “waveform relocation technique” which combines a non-linear inversion of source parameters with a grid search procedure is applied in order to correct possible systematic bias in hypocentral parameters. Simulation of inversion shows that the resolution of the source position obtained by using 2.5D Green’s functions is better than that for 1D Green’s functions. The synthetic waveforms obtained by the inversion of true data with 2.5D Green’s functions reproduce well the observed large later phases, and retrieved moment tensors are similar to those of Global CMT. Thus, the eﬀect of the near-source structure can be reproduced by 2.5D model, and with the 2.5D synthetics we can obtain improved source parameters at the trench regions where only teleseismic data are available.

1. Introduction Teleseismic body waveform analysis is important for subduction zone earthquakes such as the 2006 Java tsunami event because of the availability of the data: near-ﬁeld seismic stations on land are often scarce, and ocean bottom seismometers are usually rare in the regions. At the oceanic 215

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch15

T. Okamoto and H. Takenaka

216

trench regions, however, large eﬀects of laterally heterogeneous structure are expected to appear on the teleseismic body waveforms1–4 : thick water layer, dipping ocean bottom, and thick sediments near the source distort ray paths to teleseismic stations and often cause large later phases on the teleseismic body waveforms. This eﬀect must be evaluated carefully before a detailed source process analysis is applied to large earthquakes. The purpose of this chapter is thus to study the strong eﬀect of the near-source structure on the teleseismic body waveforms. We select the Java trench because there occurred two, Mw 7.7 “tsunami” earthquakes in 1994 and 20075–9 that have relatively low energy release at high frequencies10,11 (see Fig. 1 and Table 1 for the 2006 event). Also, recently obtained detailed crustal structure12 is available at the trench.

−2

−8º

A1 PDE

200 2)

−2

Ko pp et al. (

−2

−9º

4 −6

A2 PDE

−10º

A2 WR

A1 WR

−6

−6

6 km

0

−4

−11º

50 100

106º

107º

108º

109º

110º

Fig. 1. Aftershock distribution of the 2006 Java tsunami earthquake (USGS NEIC PDE). Star: main shock. Triangle: aftershock A1. Diamond: aftershock A2. See Table 1 for focal parameters. Gray circles: other aftershocks within 24 h from the main shock. “WR” (Waveform Relocation) denotes epicenter determined in this study. Global CMT of the main shock is also shown. Contour interval of ocean depth is 1 km.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch15

Eﬀect of Near-Source Trench Structure on Teleseismic Body Waveforms Table 1.

Mc A1 A2

Focal Parametersa of main shock and aftershocks. Date

Time

Latitude

Longitude

hb

2006/07/17 2006/07/17 2006/09/19

08:19:26.6 15:45:59.8 13:58:56.8

9.284 S 9.420 S 9.902 S

107.419 E 108.319 E 107.346 E

20 21 12

a USGS

217

NEIC PDE; b Depth in km; c Main shock.

For this purpose, we apply an extended version of “waveform relocation technique”13 to two well-recorded aftershocks (A1 Mw6.1, A2 Mw5.9; Table 1) and to some synthetic data. The method combines a source parameter inversion with grid search, and we are able to study the eﬀect of the near-source structure on the waveforms and its impact on the source parameter estimation. Relocation is necessary because scarce station coverage near the oceanic trench and the presence of subducting plate with high seismic wave velocity often result in systematic bias in hypocenters determined with teleseismic travel times.14,15

2. Data We collected teleseismic P- and SH-waveform data of world-wide broadband network through the IRIS Data Management System. The instrumental response was deconvolved from the raw data to obtain components of displacement, and a butterworth band-pass ﬁlter16 with corner frequencies 0.01 Hz and 0.5 Hz was applied. The focal parameters of the two analyzed aftershocks are listed in Table 1 along with those of the main shock.

3. 2.5D Computation We incorporate the eﬀect of the near-source structure by a “2.5dimensional” procedure. In this method, we assume a 2D model structure and compute its response to an incident plane wave. The response strain is then converted to far-ﬁeld displacement or the Green’s function due to 3D point source (not 2D line source) buried in the 2D structure by applying a reciprocal algorithm.17 Thus, we call the model structure “2.5D model”. We use an eﬃcient 2.5D FDM18,19 : we can compute the response to an incident plane wave with an arbitrary incident azimuth by a 2D ﬁnitediﬀerence grid. The computational memory and time are less than about

FA

December 10, 2008 15:45

218

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch15

T. Okamoto and H. Takenaka

10−4 times of those that would be necessary in full 3D FDM. Also, the required number of FDM simulations is reduced to the number of stations because of the reciprocal algorithm.17,20 –22 We apply a correct ﬂuid–solid boundary scheme23 in our FDM code. The scheme has O(h) accuracy with h being the space increment provided that (1) the boundary (i.e. zero rigidity) is placed through the shear stress (τxz ) grid points and (2) the standard second-order centered equation with averaged density is applied to the normal velocity component at the grid points on the boundary. We use spatial increment of 250 m and time increment of 0.0141 s. The memory is about 10–20 giga-byte and computational time is about 25–55 h per FDM computation for a single station on a 16-processor SGI Altix3700 system. Material properties of the Java trench 2.5D model (Fig. 2(a)) are allowed to vary with respect to the trench-perpendicular axis, while they are assumed to be invariant with respect to the trench-parallel axis. For this model, we refer to the detailed seismic surveys conducted near the aftershock area by Kopp et al. (Ref. 12; see Fig. 1 for survey line) and simplify their results (e.g. our model has uniform velocity in each layer while the original crustal model has velocity gradient). Also, we select materials with nearly the same P -wave velocities as those of Kopp et al.12 from the global models,24 –26 and adopt their S -wave velocities and densities to our structural model. Following the standard 1D teleseismic wave computations, we incorporate mantle attenuation by choosing t∗ = 1.0 for P -waves and 4.0 for SH -waves, and do not include an elastic attenuation in our FDM computations that evaluate near-source response. We compare the synthetic waveforms computed for the 2.5D model with those for 1D or ﬂat-layered models in Fig. 2(b). The thickness and material parameters of layers of the 1D model are sampled from the 2.5D model along vertical line pass through the assumed source position. 1D Green’s functions are computed by the method of Kroeger and Geller.27 The 2.5D synthetics for shallow sources near the trench (S1) and near the epicenter of the main shock (S2) have prolonged, large amplitude later phases. They appear irrespective of station azimuth, and are not reproduced by 1D model. For deep source (S3) the eﬀect becomes minor. The comparison clearly shows the large eﬀect of the heterogeneous structure on the body waves from shallow sources. We need to incorporate such eﬀects in analyzing the body waveforms not only from aftershocks but also from the main shock, as the major part of the source region represented by the 24-h aftershock area is considered to be quite shallow (Fig. 1).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch15

Eﬀect of Near-Source Trench Structure on Teleseismic Body Waveforms

(a)

Java Trench

Vp

N26E

Depth(km)

0.0

20.0

S1 S2

S3

40.0 0.0

50.0

100.0

150.0

200.0

250.0

300.0

219 Vs

8.04 7.1-7.2 6.0-6.6 5.0 4.3 3.5-3.9 2.0-2.3 1.5

4.47 3.9-4.0 3.4-3.7 2.5 2.3 1.8-2.1 1.0-1.1 0.0

Distance(km)

MA2

(b)

VNDA

S1

ANTO

2.5D

2.5D

2.5D

1D

1D

1D

2.5D

2.5D

2.5D

1D

1D

2.5D

2.5D

2.5D

1D

1D

1D

S2

S3

0

35 Time(s)

70

0

35 Time(s)

70

0

35 Time(s)

70

Fig. 2. (a) Cross section of the assumed 2.5D near-source structure. VP and VS (km/s) are shown in gray scale. Solid triangle denotes projected epicenter of the main shock. Also plotted are the assumed point source positions (S1–S3) for comparison of synthetic waveforms. (b) Synthetic waveforms with no ﬁlter applied. Each pair shows waveforms for 2.5D (top traces) and for 1D models (bottom traces). The indices of source positions are attached to the left of the corresponding row of pairs. We assumed the best double couple of the Global CMT solution of the main shock. The station azimuths are 21◦ at MA2, 169◦ at VNDA, and 312◦ at ANTO.

4. Waveform Relocation Technique We newly develop a non-linear point source inversion method and combine it with a grid search procedure to ﬁnd best location, source time function, and mechanism. The source time function is represented by a series of triangular unit pulses, each having a basal width of 4 s. The synthetic

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch15

T. Okamoto and H. Takenaka

220

waveform u(t) is u(t) =

6 N k −1

Mij Ak Gij (t − k∆t),

(1)

i,j=1 k=0

where Mij is the moment tensor, Gij (t) is the Green’s function for a point source with a unit pulse, Ak is the amplitude of the kth unit pulse, and ∆t = 2 s is the time oﬀset between successive Green’s functions. We impose a zero-trace condition on Mij . In order to avoid spurious oscillations in the source time function, we impose a non-negative condition Ak ≡ µ2k , and use µk as the inversion parameters. This non-negative method with squared parameter is simple but provides a stable result. The non-linear part of the inversion is solved by the algorithm proposed by Marquardt.28 The point source thus obtained with grid search nearly corresponds to the centroid of a source.29 The square residual minimized in the inversion is NW Ti [Ui (t) − ui (t)]2 1 dt, (2) S= T i=1 0 a2i and synthetic waveforms, where Ui (t) and ui (t) are ith observed Ti respectively, ai ≡ (1/wi ) (1/Ti ) 0 [Ui (t)]2 dt is RMS amplitude of observed waveform, Ti is data length, wi is weight, NW is the number W of waveforms, and T ≡ N i=1 Ti . We set wi = 1 for P -waves and wi = 2 for SH -waves. 5. Simulation of Inversion Here we simulate inversion to compare the results obtained by using diﬀerent Green’s functions. We generate “synthetic data” by superposing 2.5D Green’s functions from sets of three-point sources (TEST1 and TEST3 in Fig. 3). The time oﬀset between successive Green’s functions is 2 s with that for the deepest source being the ﬁrst one. We generate 12 P-waves and 3 SH-waves for the same stations in the analysis of aftershock A1 (Fig. 6). We then invert these synthetic data by using 2.5D and 1D Green’s functions. The 1D model structure has standard crust overlain by water and sediment layers (Table 2). The water depth is varied from 2.0 to 5.5 km to see sensitivity to the water depth: the water depth may be used as a proxy of the distance from the trench axis. The source parameters are mostly retrieved with the 2.5D or correct Green’s functions (Figs. 3 and 5). The best point sources are very

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch15

Eﬀect of Near-Source Trench Structure on Teleseismic Body Waveforms

TEST1

(a)

0.28

Depth(km)

0.00

0.24 10.00

0.20 0.16

20.00 100.00

0.12 120.00 Distance(km)

140.00 0.08

TEST2

(b) 0.00 Depth(km)

221

0.33

0.31 10.00

0.27 0.23

20.00 100.00

0.19 120.00 Distance(km)

140.00 0.15

Fig. 3. Simulated residual distributions with 2.5D Green’s functions plotted onto the cross section of the 2.5D model: (a) TEST1; (b) TEST2. Open circles denote the source positions used to generate the synthetic data. Open star indicates the best point source position. See Fig. 5 for assumed moment tensor and source time function.

Fig. 4. Simulated residuals with 1D Green’s functions plotted with respect to water and source depths: TEST1 (left) and TEST2 (right). Open star indicates the best source and water depths. Residuals larger than 0.63 are plotted in gray (right). Note that the water depths of the 2.5D model at 110, 120, 130, 140, and 150 km are 5.4, 4.5, 4.1, 3.6, and 3.4 km, respectively.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch15

T. Okamoto and H. Takenaka

222

Table 2.

(a)

1D model structure.

Layer

VP (km/s)

VS (km/s)

ρ (kg/m3 )

Thickness (km)

1 2 3 4 5

1.50 2.10 5.80 6.50 8.04

0.00 1.00 3.36 3.75 4.47

1.02 2.10 2.60 2.90 3.37

2.0–5.5 3.0 15.0–11.5 15 —

(b)

TEST1 SYN

TEST2 SYN

GUMO

T

SYN

P

GUMO

T

SYN

P

1D

2.5D

1D

2.5D T

T

LBTB

P

MBAR

P

SYN

1D

SYN

1D T

T

1D

1D P

P

0 35 Time(s)

0

35 Time(s)

70

0 35 Time(s)

0

35 Time(s)

70

Fig. 5. Assumed and retrieved moment tensors, source time functions, and waveforms in the simulations. (a) TEST1. Left: moment tensor. “SYN” denotes that used to generate synthetic data, “2.5D” the solution with 2.5D Green’s functions, and “1D” that with 1D Green’s functions. Middle: source time functions. Right: “SYN” denotes the synthetic data, and “1D” denotes those reproduced by 1D Green’s functions. The 2.5D Green’s functions reproduce nearly perfect waveforms so that they are omitted. The ratio of the inverted scalar moment to the assumed scalar moment is 1.11 for 2.5D case and 5.22 for 1D case. (b) Same as (a) but for TEST2. The ratio of the inverted scalar moment to the assumed scalar moment is 1.16 for the 2.5D case and 5.54 for the 1D case.

close to the assumed distributed sources. On the other hand, the source time functions are distorted when we apply 1D Green’s functions. Also, ambiguity remains in interpreting the results to infer the horizontal position of the source because of the same best water depths obtained in both TEST1 and TEST2 (Fig. 4). We note that large scalar moments are obtained for 1D Green’s functions (Fig. 5). They are mostly the results of spurious, prolonged source time functions and diﬀerence in rigidity. Indeed, the ratios of the retrieved scalar moments to the assumed ones become 2.0 (TEST1) and 2.3 (TEST2), respectively, when we restrict the number of unit pulses to

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch15

Eﬀect of Near-Source Trench Structure on Teleseismic Body Waveforms

Depth(km)

(a) 0.00

0.750 0.710 GCMT

15.00

0.670 0.630

PDE

30.00 120.00

0.590

160.00

0.550

Distance(km)

(b) MA2 P

223

X= 140.0km DEPTH= 13.0km RESIDUAL= 0.550 4.4

VNDA P

3.0

MBAR P

1.7

VNDA SH

2.8

DATA

KURK

ULN

MA2

OBN

2.5D SYN

PET

T MBAR

PET P

4.6

SUR P

4.0

OBN P

4.0

LBTB SH

GUMO

9.7

P LSZ

HNR LBTB

SUR

VNDA

STF GUMO P

4.7

LBTB P

3.0

KURK P

5.5

KURK SH

7.3

(C)

HNR P

3.9

LSZ P

2.6

ULN P

5.8

0

35 70 Time(s)

RMS Error (s)

1.48 1.44 1.40 1.36 1.32 -20 -10 0 10 Trench Parallel Distance (km)

Fig. 6. (a) Residual distribution of the grid search for aftershock A1. Open star indicates the best point source position. Locations of Global CMT (triangle) and PDE (diamond) are also projected. The contour interval is 0.02. (b) Observed (top) and 2.5D synthetic (bottom) waveforms. Attached number denotes the maximum amplitude of the observed waveform in µm. Also plotted are the source time function (STF) and the focal mechanism. A time window of 70 s after the onset (indicated by vertical lines) is used for inversion. The moment tensor components in unit of 1017 Nm are: Mrr = −1.81, Mθθ = 4.63, Mφφ = −2.81, Mrθ = 11.1, Mrφ = −2.04, Mθφ = 0.67, which yield a scalar moment of 1.19×1018 Nm (Mw 6.0). (C) RMS error in travel time analysis plotted versus the distance with respect to trench-parallel axis (positive toward N116◦ E with the origin placed on the cross section through the PDE epicenter). Most of the travel times listed in USGS NEIC Monthly Earthquake Data Report are used.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch15

T. Okamoto and H. Takenaka

224

three (i.e. assumed number). We can then compare the potencies in order to correct the eﬀect of the diﬀerence in rigidity: the ratios of retrieved to assumed potencies become 1.1 (TEST1) and 1.3 (TEST2), respectively. These are comparable to those obtained for 2.5D Green’s functions.

6. Results of Waveform Relocation Figure 6(a) shows the residual distribution for the aftershock A1. The grid spacing for grid search is 2 km horizontally and 1 km vertically. In Fig. 6(b), the 2.5D synthetics are compared with the observations: in most

Depth(km)

(a) 0.00

1.083 PDE

1.043 GCMT

15.00

1.003 0.963

30.00 40.00

(b) ULN P

0.923

80.00 Distance(km)

0.883 OBN

X= 64.0km DEPTH= 21.0km RESIDUAL= 0.883 1.8

PET P

2.6

LBTB SH

1.6

ANTO ANTO P

ULN

2.1 MBAR

MA2

DATA

P 2.5D SYN

ULN SH

1.5

LSZ

HNR P

4.1

LSZ P

1.8

OBN P

LBTB

2.5

PET

T HNR VNDA

STF

(C)

MA2 P

2.1

VNDA SH

0.9

MBAR P

1.6

0

35 70 Time(s)

RMS Error (s)

1.40 1.36 1.32 1.28 1.24 1.20 -10

0

10

Trench Parallel Distance (km)

Fig. 7. Same as Fig. 6. but for aftershock A2. The moment tensor components in unit of 1017 Nm are: Mrr = −8.51, Mθθ = 5.98, Mφφ = 2.53, Mrθ = −1.21, Mrφ = 0.27, Mθφ = −4.82, which yield a scalar moment of 9.06 × 1017 Nm (Mw 5.9).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch15

Eﬀect of Near-Source Trench Structure on Teleseismic Body Waveforms

225

of the stations, peaks and troughs in the observed later arrivals are well reproduced by the synthetics. The best position (Fig. 6(a)) and the mechanism (Fig. 6(b)) of the obtained point source are close to those of the Global CMT. On the other hand, the PDE location is remote from the best position by about 30 km. It is not likely that the PDE location indicates the initial rupture, because the distance (30 km) is too large for a Mw 6 event. It is rather likely that the discrepancy is the result of the systematic bias in the teleseismic travel times caused by the high-velocity subducting slab.14,15 For aftershock A2, we again observe good agreement between the 2.5D synthetics and the data (Fig. 7): large later phases on the observed P-waves at about 40 s from the onset are reproduced in the 2.5D synthetics. In this case, however, the observed waves are richer in short period components than the synthetics are. Also, the width of the observed initial pulses are narrower than those of the synthetics. This is because, due to computational limitations in the FDM simulations, we used a unit pulse with a width of 4 s which might be slightly wider than appropriate for this event. We also analyze aftershock A1 by using the 1D Green’s functions. The obtained moment tensor and source time function (Fig. 8) are similar to those for the 2.5D Green’s functions (Fig. 6). The scalar moment for 1D Green’s function is larger than that for 2.5D Green’s function because of the diﬀerence in the assumed rigidity. The source depth of 10 km is comparable to that of 2.5D Green’s function (13 km). The resolution of the horizontal position is, however, poor in the case of 1D Green’s function.

7. Discussion The advantage of our 2.5D procedure is that improved source parameters can be obtained at trench regions where only teleseismic data are available. However, “absolute” position with respect to the trench-parallel axis cannot be constrained by our method. (Note that it is possible to determine the “relative” positions of subevents with respect to the trench-parallel axis by our 2.5D method.) Therefore, we here attempt to combine travel-time analysis in order to supplement our procedure. We ﬁx the source depth and the trenchperpendicular distance to those obtained by the waveform relocation. We then compute travel-time error by varying the epicenter along the trenchparallel axis to ﬁnd the best position that minimizes the error. (We note

FA

MA2

P

LBTB DATA

1D

1D

GUMO

MBAR DATA

0 35 Time(s)

DATA

1D

1D 0

70

Fig. 8. Result of inversion of aftershock A1 with 1D Green’s function. The same stations as in Fig. 6 are used. (left) Residuals. Open star indicates the best source and water depths. (middle) Moment tensor and source time function obtained for the best source. (right) Observed and synthetic waveforms. The moment tensor components in unit of 1018 Nm are: Mrr = −0.415, Mθθ = −0.036, Mφφ = 0.451, Mrθ = 3.17, Mrφ = −0.324, Mθφ = −0.118, which yield a scalar moment of 3.22 × 1018 Nm (Mw 6.3).

9in x 6in

35 Time(s)

T. Okamoto and H. Takenaka

STF

AOGS 2007 - SE Volume

DATA

December 10, 2008 15:45

226

T

b672-V13-ch15 FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch15

Eﬀect of Near-Source Trench Structure on Teleseismic Body Waveforms

227

that the primary slab eﬀect is advancing teleseismic travel times at stations in the direction of the subduction,14 which results in systematic shift of the hypocenters to the nearly trench-perpendicular direction.15 This primary eﬀect is already corrected through our waveform relocation.) We use HYPOCENTER 3.230 in the travel-time analysis. The results are shown in Figs. 1, 6(c), and 7(c). It is noteworthy that PDE hypocenter of A1 is remote from the waveform-relocated source position, while that of A2 is close to the waveform-relocated position. This suggests a spatially variable slab eﬀect on the teleseismic travel times: it is large for sources near the thrust zone but minor for sources near the trench. This variability has been pointed out by Engdahl et al.15

8. Conclusions We apply a “waveform relocation technique” to teleseismic body waveforms from two aftershocks of the 2006 Java tsunami earthquake as well as to two sets of synthetic data. We assume a 2.5D model of the Java trench structure, and the synthetic waveforms are computed by a very eﬃcient 2.5D ﬁnite-diﬀerence method in order to incorporate the strong eﬀect of the near-source structure on the teleseismic body waveforms. Simulations of inversion indicate that, with 2.5D Green’s functions, we can constrain the source location both vertically and horizontally. The inversions of observed data show that synthetic waveforms reproduce well the observed large later arrivals, and the inverted moment tensors are very similar to those of Global CMT solutions. Thus, the eﬀect of the near-source structure can be reproduced by the 2.5D model, and with 2.5D synthetics, we can obtain improved source parameters at the trench regions where only teleseismic data are available. Acknowledgments The non-negative inversion with squared parameters is suggested by Yosihiko Ogata. Discussions with Naoki Kobayashi were very helpful. The authors are grateful to an anonymous reviewer for his/her constructive comments. The authors used computers from GSIC (Tokyo Institute of Technology) and ERI (University of Tokyo). The facilities of the IRIS Data Management System, and speciﬁcally the IRIS Data Management Center, were used for access to waveform and metadata required in this study. The authors used focal parameters available online at NEIC (U.S. Geological

FA

December 10, 2008 15:45

228

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch15

T. Okamoto and H. Takenaka

Survey), and the Global CMT catalog available at http://www.globalcmt. org, and used the GMT package.31 This research was supported by the Japan Society for the Promotion of Science, Grants-in-Aid for Scientiﬁc Research (C), 18540418, 2006.

References 1. D. A. Wiens, Eﬀects of near source bathymetry on teleseismic P waveforms, Geophys. Res. Lett. 14 (1987) 761–764. 2. D. A. Wiens, Bathymetric eﬀects on body waveforms from shallow subduction zone earthquakes and application to seismic processes in the Kurile Trench, J. Geophys. Res. 94 (1989) 2955–2972. 3. T. Okamoto and T. Miyatake, Eﬀects of near source seaﬂoor topography on long-period teleseismic P waveforms, Geophys. Res. Lett. 16 (1989) 1309–1312. 4. S. Yoshida, Waveform inversion of rupture process using a non-ﬂat seaﬂoor model: Application to 1986 Andreanof Islands and 1985 Chile earthquakes, Tectonophysics 211 (1992) 45–59. 5. R. E. Abercrombie, M. Antolik, K. Felzer and G. Ekstr¨om, The 1994 Java tsunami earthquake: Slip over a subducting seamount, J. Geophys. Res. 106 (2001) 6595–6607. 6. C. J. Ammon, H. Kanamori, T. Lay and A. A. Velaso, The 17 July 2006 Java tsunami earthquake, Geophys. Res. Lett. 33 (2006) L24308, doi: 10.1029/ 2006GL028005. 7. Y. Fujii and K. Satake, Source of the July 2006 West Java tsunami estimated from tide gauge records, Geophys. Res. Lett. 33 (2006) L24317, doi: 10.1029/ 2006GL028049. 8. T. Hara, Magnitude determination using duration of high frequency energy radiation and displacement amplitude: Application to tsunami earthquakes, Earth Planets Space 59 (2007) 561–565. 9. S. L. Bilek and E. R. Engdahl, Rupture characterization and aftershock relocations for the 1994 and 2006 tsunami earthquakes in the Java subduction zone, Geophys. Res. Lett. 34 (2007) L20311, doi: 10.1029/2007GL031357. 10. H. Kanamori, Mechanism of tsunami earthquakes, Phys. Earth Planet. Inter. 6 (1972) 246–259. 11. J. Polet and H. Kanamori, Shallow subduction zone earthquakes and their tsunamigenic potential, Geophys. J. Int. 142 (2002) 684–702. 12. H. Kopp, D. Klaeschen, E. R. Flueh and J. Bialas, Crustal structure of the Java margin from seismic wide-angle and multichannel reﬂection data, J. Geophys. Res. 107 (2002), doi: 10.1029/2000JB000095. 13. T. Okamoto, Location of shallow subduction zone earthquake inferred from teleseismic body waveform, Bull. Seism. Soc. Am. 84 (1994) 264–268. 14. N. H. Sleep, Teleseismic p-wave transmission through slabs, Bull. Seism. Soc. Am. 63 (1973) 1349–1373.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch15

Eﬀect of Near-Source Trench Structure on Teleseismic Body Waveforms

229

15. E. R. Engdahl, J. W. Dewey and K. Fujita, Earthquake location in island arcs, Phys. Earth Planet. Int. 30 (1982) 145–156. 16. M. Saito, An automatic design algorithm for band selective recursive digital ﬁlters, Geophys. Exploration (Butsuri-Tanko) 31 (1978) 112–135. 17. T. Okamoto, Teleseismic synthetics obtained from three-dimensional calculations in two-dimensional media, Geophys. J. Int. 118 (1994) 613–622. 18. H. Takenaka and B. L. N. Kennett, A 2.5-D time-domain elastodynamic equation for plane-wave incidence, Geophys. J. Int. 125 (1996) F5–F9. 19. H. Takenaka and T. Okamoto, Teleseismic waveform synthesis for oceanbottom stations using a new, very eﬀective 2.5-D ﬁnite diﬀerence technique, Proc. Int. Workshop on Scientific Use of Submarine Cables (1997), pp. 23–26. 20. L. Eisner and R. W. Clayton, A reciprocity method for multiple-source simulations, Bull. Seism. Soc. Am. 91 (2001) 553–560. 21. R. W. Graves and D. J. Wald, Resolution analysis of ﬁnite fault source inversion using one- and three-dimensional Green’s functions 1. Strong motions, J. Geophys. Res. 106 (2001) 8745–8766. 22. T. Okamoto, Full waveform moment tensor inversion by reciprocal ﬁnite diﬀerence Green’s function, Earth Planets Space 54 (2002) 715–720. 23. T. Okamoto and H. Takenaka, Fluid-solid boundary implementation in the velocity-stress ﬁnite-diﬀerence method, J. Seism. Soc. Japan 57 (2005) 355–364. 24. C. Bassin, G. Laske and G. Masters, The current limits of resolution for surface wave tomography in North America, EOS Trans. AGU 81 (2000) F897. 25. B. L. N. Kennett and E. R. Engdahl, Traveltimes for global earthquake location and phase identiﬁcation, Geophys. J. Int. 105 (1991) 429–465. 26. A. M. Dziewonski and D. L. Anderson, Preliminary reference earth model, Phys. Earth Planet. Int. 25 (1981) 297–356. 27. G. C. Kroeger and R. J. Geller, An eﬃcient method for computing synthetic reﬂections for plane layered models, EOS Trans. AGU 64 (1983) 772. 28. D. W. Marquardt, An algorithm for least-squares estimation of nonlinear parameters, J. Soc. Indust. Appl. Math. 11 (1963) 431–441. 29. A. M. Dziewonski, T.-A. Chou and J. H. Woodhouse, Determination of earthquake source parameters from waveform data for studies of global and regional seismicity, J. Geophys. Res. 86 (1981) 2825–2852. 30. B. R. Lienert, E. Berg and L. N. Frazer, HYPOCENTER: An earthquake location method using centered, scaled, and adaptively damped least squares, Bull. Seism. Soc. Am. 76 (1986) 771–783. 31. P. Wessel and W. H. F. Smith, Free software helps map and display data, EOS Trans. AGU 72 (1991) 441.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

This page intentionally left blank

b672-V13-ch15

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch16

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

NUMERICAL MODELING OF THE 2006 JAVA TSUNAMI EARTHQUAKE NURAINI RAHMA HANIFA∗,†,‡ , IRWAN MEILANO∗,† , TAKESHI SAGIYA∗ , FUMIAKI KIMATA∗ and HASANUDDIN Z. ABIDIN† ∗ Research Center of Seismology, Volcanology and Disaster Mitigation Graduate School of Environmental Studies, Nagoya University Furo-cho, Chikusa-ku Nagoya, 464-8601, Japan † Geodesy

Research Division, Institut Teknologi Bandung, Jl. Ganesha 10, Bandung, 40132, Indonesia ‡ [email protected]

A M7.8 occurred oﬀ the coast of Pangandaran, Central Java, Indonesia at 15:24 (08:24 UTC) on 17 July 2006, and excited a deadly tsunami of average 3–8 m that inundated the southern coast of Java. This event is considered as a tsunami earthquake based on the fact that the earthquake generated a much larger tsunami than expected from its seismic waves, its unusual long rupture, and its source mechanism. We modeled the tsunami propagation by solving a non-linear shallow water equation. The initial condition of the co-seismic deformation was computed from Okada’s formula,23 using several diﬀerent source parameters. We simulated the tsunami propagation with a ﬁnite-diﬀerence numerical model using a system of multiple grids with one and two minutes grid sizes. The best model parameters to reproduce the observed tsunami heights are seismic moment M0 = 1.7 × 1021 Nm, average slip D = 12 m, and rupture area S = 14,000(70 × 200) km2 , with the rigidity of 10 GPa. The computed tsunami heights are in fair agreement with the runup observations along the south coast of Java. However, the same fault model results in a southward displacement of about 25 mm at BAKO station in west Java, which is too large compared with the observed value of 4 mm. This discrepancy can be attributed to two eﬀects. First, if the source process was accompanied by splay faulting, vertical displacement of the sea ﬂoor becomes larger and caused massive tsunami. Another possibility is an eﬀect of structural heterogeneity. The source region of the 2006 Java earthquake is near the trench, and a very small rigidity (10 GPa) is appropriate. For the calculation of a far-ﬁeld static displacement at BAKO, however, we need to assume the same amount of seismic moment (1.7 × 1021 Nm) with a normal rigidity (30 GPa). Then, the average slip is reduced to 8 m and the calculated displacement matches the observation. This example demonstrates the importance of taking structural heterogeneity into account in modeling crustal deformation, and an easy way to deal with a heterogeneous structure is implied. With this methodology, it is also possible to estimate rigidity of the source region by 231

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch16

N. R. Hanifa et al.

232

adjusting the slip amount so that the same seismic moment can reproduce both the tsunami data and far-ﬁeld displacement data.

1. Introduction On 17 July 2006, at 08:24 UTC or 15:24 in local time, a magnitude 7.8 earthquake occurred oﬀ the coast of Pangandaran, south of Java Island, Indonesia. The earthquake involved thrust faulting in the Java trench and excited a deadly tsunami of 3–8 m in height that inundated the southern coast of Java. This is a very rare event, especially because previous research showed that the Java trench is an aseismic subduction.22 Ammon et al.,2 Kato et al.,15 and Fujii and Satake7 concluded that this event was a tsunami earthquake as it generated more massive tsunami than expected from the seismic waves.11 The characteristics of tsunami earthquake are summarized as follows: 1. Slow rupture process.11 2. Long rupture duration, about 100 s from seismological studies. This is investigated from the 1992 Nicaragua tsunamic earthquake, which is the ﬁrst tsunami earthquake recorded on modern broadband seismic instruments.12 3. Have similar seismic waveforms, which are diﬀerent from normal (tsunamigenic) earthquakes.14 4. Occur at the plate boundaries where the plate coupling is weak.29 5. Source area is located within the shallow sedimentary layers8,24 or on the plate boundary near the trenches.33 Tsunami earthquakes are rare events and only 11 examples of tsunami earthquake events are known so far since the late 19th century. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

1896 1946 1960 1963 1975 1992 1992 1994 1996 1998 2006

Meiji Sanriku Tsunami Earthquake11,37 Aleutian Tsunami Earthquake11,19 Peru Tsunami Earthquake26 Kuriles Tsunami Earthquake8,25,26 Kuriles Tsunami Earthquake8,26 Nicaragua Tsunami Earthquake12,31 Flores Tsunami Earthquake3 Java Tsunami Earthquake1,27,28,36 Peru Tsunami Earthquake25 Papua New Guinea Tsunami Earthquake35 Java Tsunami Earthquake2,7,15

FA

December 10, 2008 15:45 AOGS 2007 - SE Volume 9in x 6in

Numerical Modeling of the 2006 Java Tsunami Earthquake 233

b672-V13-ch16

Fig. 1. Tectonic setting and historical earthquake along Sunda Trench in Indonesia. The hypocenters of the earthquakes are by USGS and focal mechanisms are by Harvard CMT solution. Convergent rates of the India–Austrian plate are by Sella et al.36 Ages of ocean bottom are shown as greyscale squares.

FA

December 10, 2008 15:45

234

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch16

N. R. Hanifa et al.

Very complicated plate-convergences consisting of subduction, collision, back-arc thrusting, and back-arc are ongoing in and around Indonesia. As the result of this complexity, the region is considered as one of the most tectonically active areas in the world.17 In Indonesia, at least 460 earthquakes of M > 4.0 occur every year.10 Based on the data compiled during the period 1600–1999, 105 tsunamis have been recorded in Indonesia.17 95 events (90%) of them were caused by earthquakes in shallow region at subduction and plate boundaries, 9 (8%) by volcano eruptions, and 1 (1%) by a landslide.17 It is suggested that these tsunamis were generated by tsunami earthquakes though no precise data is uncovered. Oblique subduction of the India–Australian plates beneath Sumatra has a rate of 5.6 cm/y.36 The rate of subduction in the West Java trench where the 17 July 2006 earthquake occurred is about 6.4 cm/y.36 In further east along the East Java trench, the rate of subduction is about 6.9 cm/y. Near New Guinea, the subduction rate increases to as much as 10.7 cm/y (Fig. 1). Newcomb and McCann22 reviewed the seismic history and seismotectonics of the Sunda Arc. They concluded that there are more frequent and larger earthquakes along Sumatra, where a younger plate is subducting beneath Sumatra with a shallow dip angle, indicating a signiﬁcant seismic coupling. The entire length of the plate boundary has a potential to produce great thrust earthquakes. Less frequent and smaller events occur along Java where subduction of older sea ﬂoor takes place relatively aseismically. However, there are needs for further seismological investigation along Java after three deadly earthquakes since 1987. The ﬁrst one was the M7.7 Java tsunami earthquake on May 1994. Other two earthquakes of M6.4 and M7.8 hit Java in 2006, the Yogyakarta earthquake in May, and Pangandaran tsunami earthquake in July. We describe the result of the 2006 Java numerical modeling using ﬁnite diﬀerence to understand the mechanism of the 2006 Java tsunami earthquake. The numerical modeling used several scenarios to ﬁnd the closest fair agreement with the continuous GPS observation and observed tsunami run-up heights.

2. Java 2006 Tsunami Earthquake The earthquake was centered at 9.295◦ S 107.347◦E, with a depth of 6 km below the mean sea level38 (Fig. 2). The source mechanism was a low angle

FA

December 10, 2008 15:45 AOGS 2007 - SE Volume 9in x 6in

Numerical Modeling of the 2006 Java Tsunami Earthquake 235

b672-V13-ch16

Fig. 2. Hypocenter distribution of main shock and aftershocks of July 2006 Java earthquake in the period within 2 days after the mainshock by USGS. Focal mechanism of the main shock by Harvard CMT solution. Observed tsunami heights by Geodesy Research Division of ITB and BMG are also shown.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch16

N. R. Hanifa et al.

236

thrust indicating that the event occurred on the plate interface between the Australia and the Sunda plates along the Java trench (http://www. globalcmt.org/). The strike, dip, and rake angles of the fault slip are estimated to be 287◦ , 10◦ , and 95◦ , respectively. On this part of their mutual boundary, the Australia plate moves north–northeast with respect to the Sunda plate at about 64 mm/y.36 The Australian plate thrusts beneath the Sunda plate at the Java trench, south of Java, and is subducted to progressively greater depths beneath Java and north of Java. The earthquake occurred on the shallow part of the plate boundary, about 250 km north of the Java trench. Based on the report of BMG Indonesia,4 the tsunami inundated the southern Java coastline by about 236 m on average. The tsunami waves destroyed around 90 houses, 62 hotels, 5 oﬃce buildings, 56 cars, 97 motorbikes, 190 boats, and 29 traditional transportations. At least 378 deaths occurred, 272 injured, and 77 were missing. The area destruction can be seen clearly by the remote sensing image. Center for Remote Imaging, Sensing and Processing of the National University of Singapore had provided Satellite image of Spot in the Java tsunami area. Tsunami run-up height was measured by Geodesy Research Group of Bandung Institute of Technology in cooperation with The University of Tokyo. BMG also conducted the tsunami run-up height measurement separately. The run-up height was about 3–8 m along the coast from Pameugpeuk to Kebumen (Table 1, Fig. 2), while the tsunami maximum height measured by the International Tsunami Survey Team (ITST) was 21 m in Nusa Kambangan Islands.6

Table 1. Point

1 2 3 4 5 6 7 8 Ngantik Kisik Ayah Kebumen Parangtritis Benoa

Observed run-up tsunami height.

Data source

Long.

Lat.

Heights observed (m)

ITB ITB ITB ITB ITB ITB ITB ITB BMG BMG BMG

107.690722 108.059917 108.402750 108.442250 108.497583 108.535944 108.612028 108.651667 109.079970 109.394100 110.334690 115.216667

−7.668306 −7.764667 −7.815556 −7.819500 −7.746833 −7.692194 −7.683667 −7.693778 −7.689440 −7.724510 −8.025970 −8.766667

5.2 3.74 3.67 5.9 2.12 5.34 7.67 4.69 4.92 5.12 2.57 0.243

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Numerical Modeling of the 2006 Java Tsunami Earthquake

b672-V13-ch16

237

Kato et al.15 concluded this event as tsunami earthquake based on their observation of run-up height, GPS observation, and numerical modeling. Their GPS results suggested that the motion of the Java Island is due ESE, almost perpendicular to the direction of the plate subduction, and does not show any inﬂuence of the plate convergence of the Australian plate that subducts underneath the Java Island. They inferred that plate coupling at the subducting plate interface must be very weak, that is a characteristic of tsunami earthquakes’ potential zone. They also mentioned that people along the coast felt weak ground shaking only, but suﬀered from unexpectedly high tsunami. The considerably weaker ground shaking than expected from its magnitude (Mw 7.8) strongly indicates that the slip velocity on the fault would be slow enough not to radiate much seismic waves. Based on numerical simulation, Koshimura16 suggests that if the earthquake was on a low angle thrust fault, it generates highest tsunami of about 3.5 m in the west of Pangandaran area, whereas high angle faults generate only about 3 m of maximum tsunami. In both cases, their tsunami simulation does not reach the east of Cilacap. Latief et al.18 also have run the tsunami numerical modeling and found tsunami height of 3–4 m along the coast from Pameungpeuk to Kebumen. Mori and Park20 suggested that the rupture velocity of the earthquake was about 1 km/s in average, based on the back-propagation analysis of P-wave seismograms. This rupture velocity was much slower than normal earthquakes that usually show rupture speed of 2–3 km/s. Ammon et al.2 stated that the size of the 2006 Java earthquake varies signiﬁcantly with seismic wave frequency band analyzed: analysis of very long-period signals of around 300–500+ s indicates a seismic moment of 6.7 × 1020 Nm (Mw = 7.8), while the surface wave magnitude is MS (∼20 s) = 7.2, the body-wave magnitude is mb (∼1 s) = 6.2, and shaking intensities (3–10 Hz) were ≤MMIV. According to them, the rupture time of the 2006 Java earthquake was abnormally long, continued for about 185 s, and propagated slowly at about 1.0–1.5 km/s. These attributes are common with other tsunami earthquakes. They mentioned that the Mw 7.8 2 June 1994 Java earthquake, located about 600 km east–southeast of the 2006 event, had similar strong tsunami excitation (15 m maximum run-up height), and both events had aftershock sequences dominated by normal faulting, suggesting relatively complete stress release on the interplate thrust. They showed that the 2006 Java earthquake involved 5–6 pulses of moment release superimposed on a smooth rupture, indicating a compound frictional environment likely inﬂuenced by weak material properties related to sediment subduction or the presence of ﬂuids.

FA

December 10, 2008 15:45

238

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch16

N. R. Hanifa et al.

Fujii and Satake7 have inverted the tsunami waveforms and revealed that the tsunami source was about 200 km long. The largest slip of about 2.5 m for instantaneous rupture model was located about 150 km east of the epicenter. Most of the slips occurred on shallow parts of the fault, which is also a common characteristic to tsunami earthquakes. The estimated slip distribution yields a total seismic moment of 7.0 × 1020 Nm (Mw = 7.8). Cummins et al.5 have constrained the slip distribution of the 2006 Java earthquake using long-period surface waves. Their model obtained a maximum slip of 2 m, with the concentration of the slip in the shallow part of the fault plan. They reconciled this slip model with the GPS and tide gauge data. GPS measurements of co-seismic displacement at Christmas Island, Australia and the IGS station BAKO on Java both suggest about 6 mm of horizontal displacement, which is about half that predicted from their slip model. They suggest that this inconsistency occurs, because the usage of surface waves ﬁltered to pass periods of 300–800 s is insuﬃcient to resolve the total slip. About this inconsistency, we will propose an alternative interpretation in discussion.

3. GPS PPP Continuous Observation Continuous GPS station, the BAKO IGS station (in West Java), was being operated during the 2006 Java earthquake. We computed horizontal displacement from this continuous GPS station using Bernese GPS software version 5.0 with the PPP (Precise Point Positioning) method. The results of GPS PPP data processing at BAKO show co-seismic displacement due to the 2006 Java tsunami earthquake as small as 4 mm in the southward direction (Fig. 3).

4. Tsunami Modeling Method We conducted model calculation of crustal deformation and tsunami propagation assuming various fault models to study the earthquake source process to check if they satisfy our observations of GPS and tsunami height. We do linear and non-linear tsunami modeling of tsunami propagation using a ﬁnite diﬀerence code by Nakamura.21 Seaﬂoor displacement as the initial condition of tsunami propagation was computed using Okada’s formula.23 In the simulation of tsunami propagation, we can apply the long-wave (shallow water) approximation when the tsunami wavelength is much larger

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch16

Numerical Modeling of the 2006 Java Tsunami Earthquake

Fig. 3.

239

GPS PPP result of BAKO-IGS station in the year 2006.

than the water depth and the tsunami height is negligible compared with the water depth. The wavelength of tsunami depends on the rupture length, which must be 20 times as large as the water depth to satisfy such a condition. In the case of the 2006 Java tsunami, the water depth is 6 km and the rupture length is about 200 km,2 more than 30 times larger than the water depth so that the long-wave approximation is validated. Under the long-wave approximation, velocity of tsunami propagation is expressed as follows: v=

g · d,

(1)

where v is propagation velocity in m/sec, g is the gravitational acceleration (9.8 m/s2 ), and d is the water depth in meters. Tsunami wave propagates faster in deepsea, with a small wave height. Approaching shallower area, propagation becomes slower, then the wavelength decreases and the wave height is ampliﬁed. Relation between the water depth and the wave height is illustrated in Fig. 4. For water deeper than 50 m, we can assume that the tsunami amplitude is much smaller than the water depth, and then the linear equation under

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch16

N. R. Hanifa et al.

240

Run-up height

η

η v

Fig. 4.

h

h

D

Relation between water depth and wave height (based on Ref. 20).

the long-wave approximation can be applied30 : du/dt = −g∇h.

(2)

For water depth less than 50 m (shallow water), non-linear eﬀects have to be properly taken into account and the equation of tsunami propagation can be written as30 du/dt + (u · ∇)u = −g∇h, v=

gλ 2π

2πd tanh . λ

(3)

(4)

From the diﬀerence in water height, the ﬂux change is computed using the momentum equation. Then from ﬂux change, diﬀerence in water height is calculated again using mass conservation equation, and repeated. This is the equation of motion and continuity of tsunami wave propagation used in this study. Mass conservation equation is written as21 ∂M ∂N ∂η + + = 0, ∂t ∂x ∂y

(5)

and the momentum equation is written as21 ∂ M2 gn2 ∂M ∂ MN ∂η + + 7/3 M M 2 + N 2 = 0, + + gD ∂t ∂x D ∂y D ∂x D (6) 2 ∂N ∂ MN gn2 2 ∂ N ∂η N M + N 2 = 0, + + + gD + ∂t ∂x D ∂y D ∂y D7/3 where η is the vertical displacement of water surface above the still water level, D is total water depth (D = η + h), g is gravitational acceleration, x and y are horizontal coordinates, t is time, M and N are discharge ﬂuxes in the x- and y-directions respectively, and n is coeﬃcient of bottom friction.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Numerical Modeling of the 2006 Java Tsunami Earthquake

b672-V13-ch16

241

If η/h is close to zero (wave height is very small compared to water depth), the non-linear term is negligible. The momentum equation then becomes ∂η ∂M + gD = 0, ∂t ∂x

∂N ∂η + gD = 0. ∂t ∂y

(7)

We use ETOPO2 (http://www.ngdc.noaa.gov/mgg/global/relief/ ETOPO2/ETOPO2v2-2006) for 2-min grid bathymetric data and GEBCO (http://www.ngdc.noaa.gov/mgg/gebco/grid/1mingrid.html) for 1-min grid bathymetric data. Initial condition of the tsunami propagation is given by co-seismic displacement calculated with Okada’s23 formula. We tested various fault parameters compiled from the information of USGS,39 BMG,4 Ammon et al.,2 and Fujii and Satake,7 as shown in Table 2 to check if the generated tsunami reproduces the run-up height observation. On the other hand, based on the same fault parameters, we also calculate displacement at BAKO GPS station to compare with the observation. Fault location is assumed to be the same for all the cases based on NEIC moment tensor solution of USGS39 which is 9.295S, 107.347E. Some of the model parameters tested are given in Table 2. For model 1, we tested fault parameters based on USGS-NEIC moment tensor solution,39 with ETOPO2 as bathymetric data. The fault area is estimated based on the aftershock distribution by USGS, with a rupture length of about 200 km and the width of 40 km. The slip is computed using the deﬁnition of seismic moment: M0 = µDS,

(8)

where Mo is seismic moment, µ is rigidity, D is rupture area in square km, and S is slip in meter. We use rigidity value of 10 GPa based on Ammon et al.2 This small rigidity is consistent with the sedimentary material within the accretionary wedge. The result showed a maximum wave height of 0.531 m. For model 2, we revised the value of seismic moment. We computed the seismic moment from moment magnitude log M0 = 1.5Mw + 9.1.

(9)

We got Mw of 4.5 × 1020 Nm, and so the slip is estimated to be 5.6 m. The result gave the maximum wave height of 2.928 m.

FA

Unit

Reference

USGS 7.7 4.5E+20 1.0E+10 297 6 93 6 5.63 200 40 2.928 30 4.6

Model 3 Fujii and

Satake7

7.8 6E+20 3.0E+10 289 10 95 6 2.50 200 40 1.386 13 4.5

Model 4 Ammon et

al.2

7.8 6.7E+20 1.0E+10 289 10 95 6 8.38 200 40 4.596 40 4.3

Model 9 al.2

Ammon et max. slip 8.1 2.1E+21 1.0E+10 289 10 95 8 15.00 200 70 8.665 83 3.2

Model 11 Ammon et al.2 avr. slip 8.1 1.E+21 1.0E+10 289 10 95 8 12.00 200 70 6.2907 25 3.4

9in x 6in

Pa degree degree degree km m km km m mm m

USGS modiﬁed 7.2 8.1E+19 1.0E+10 297 6 93 6 1.01 200 40 0.531 5 4.6

Model 2

N. R. Hanifa et al.

Mw Mo Rigidity Strike Dip Rake Depth Slip Length Width Max Tsunami BAKO displacement Tsunami height RMS

Model 1

AOGS 2007 - SE Volume

Column 1

Fault models and calculation results.

December 10, 2008 15:45

242

Table 2.

b672-V13-ch16 FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Numerical Modeling of the 2006 Java Tsunami Earthquake

b672-V13-ch16

243

Model 3 uses the parameters from Fujii and Satake.7 They did inversion calculation of tsunami waveforms from tide gauge data to get the tsunami source. They suggested a rupture length of 200 km, a slip of 2.5 m, and a seismic moment of 6.0 × 1020 (equivalent to Mw 7.8). They assumed the rigidity of 3.0 × 1010 N/m2 . The result gave a maximum wave height of 1.386 m. This result is in fair agreement with the tsunami waveforms recorded at tide gauges, but much smaller than the coastal observation. For the next model, we used the source information by Ammon et al.2 They suggested a seismic moment of 6.7 × 1020 . The variable slip ranges 8–15 m on the fault plane, the fault width of 75 km. They assumed the rigidity of 1.0 × 1010 N/m2 . For model 4, we used an average slip over the entire fault plane (200 km × 40 km) based on Ref. 2. We computed the slip as 8.3 m. The result gave a maximum wave height of 4.596 m. For model 5, we modiﬁed the rupture width to 70 km. The result gave the maximum wave height as 5.02 m. The chosen rupture area of 200×70 km covers all the aftershock distribution, so that we keep this rupture area in the following models. In model 6, we tried the maximum slip of 15 m from Ammon et al.,2 and get the maximum wave height of 8.832 m. In model 7, we modiﬁed the fault depth to 8 km, and used the slip of 15 m. Model 7 produces the maximum tsunami height of 8.997 m from the result of Ammon et al.2 It seems that the fault depth of 6 km does not give a signiﬁcant diﬀerence, but the increase of slip from 8 to 15 m causes signiﬁcant diﬀerence. Model 7 reproduces the pattern of the tsunami wave shown in Fig. 2 fairly well. We compared the usage of 2-min data of ETOPO with 1-min data grid of GEBCO. The maximum parameter based on Ammon et al.2 is 15 m of slip, rupture of 200 × 70 km, depth 8 m generating a maximum wave of 8.665 m with maximum height in the coastline of 5.821 m using non-linear equation (model 9). In model 11, we tried the average slip of 12 m based on Ammon et al.2 with 1 min grid bathymetry data. We gained tsunami maximum height of 6.291 m near the coast and displacement in BAKO of 25 mm. Illustration of results is given in Fig. 5 for model 11. The use of 1-min grid gives much improvement than 2-min grid. This is fair with the result of Satake.32 In summary of tsunami simulation, we conclude that the best model for the 2006 Java earthquake we obtained so far is model 11, assuming a fault area of 200 km × 70 km, an average fault slip of 12 m, with the tsunami

FA

December 10, 2008 15:45

244

AOGS 2007 - SE Volume

N. R. Hanifa et al.

9in x 6in

b672-V13-ch16

Tsunami height of model 11. Fig. 5.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch16

Numerical Modeling of the 2006 Java Tsunami Earthquake

245

propagation calculated with 1-min grid bathymetry data. This model reproduces the coastal tsunami observation, but the GPS displacement at BAKO is calculated as 25 mm using Okada’s formula. This discrepancy will be interpreted in the discussion.

5. Discussion and Conclusion The 2006 Java earthquake occurred near the Java trench where soft sediments form accretionary wedge. So, a small rigidity is appropriate for the source region. Seismologically estimated fault model (model 1) satisﬁed the GPS data but the calculated tsunami height was too small. On the other hand, in order to reproduce the observed tsunami height, seismic moment becomes as large as 2.1 × 1021 Nm, which is equivalent to Mw 8.1 (model 9). However, if we use Okada’s formula, the same fault model results in a southward displacement of about 83 mm at BAKO station in West Java, which is far too larger than the observation data of 4 mm. Model 11 reproduces 6 m tsunami height with displacement in BAKO of 25 mm, equivalent with the moment magnitude of Mw = 8.1 and seismic moment of 1.7 × 1021 Nm. Comparing the results of the models with the tsunami height and GPS displacement in BAKO station, the most suitable model is model 11, with small RMS of tsunami height compared with the tsunami run-up height data, with a displacement of 25 mm at BAKO. This discrepancy can be attributed to two eﬀects. First, if the source process was accompanied by splay faulting, vertical displacement of the sea ﬂoor becomes larger and caused massive tsunami (Fig. 6). Another possibility is an eﬀect of structural heterogeneity.

Fig. 6.

Illustration of case possibility of the 2006 Java earthquake mechanism.

FA

December 10, 2008 15:45

246

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch16

N. R. Hanifa et al.

The source region of the 2006 Java earthquake is near trench. A rigidity of the source region should be as low as 10 GPa.2 On the other hand, if we look at the source region from BAKO GPS station, we do not know the rigidity of the source region, but we can only infer a mechanical eﬀect by the seismic source, which can be represented by moment tensor density t ∂ dτ Gip (x, t; ξ, τ )mpq (x, τ )dξ. (10) ui (x, t) = −∞ v ∂ξq Here, mpq is moment tensor density, containing rigidity of the source region, and Gip is Green’s tensor, surface displacement response to a unit force in the pth direction at the source, containing rigidity of the medium. In Okada’s formula, displacements are calculated under the semiinﬁnite elastic body; rigidity of the surrounding medium is assumed to be the same and cancel out each other. In the case of the Java earthquake, rigidity around epicenter and at BAKO should be diﬀerent. However, for the calculation of a far-ﬁeld static displacement at BAKO, we need to assume the same amount of seismic moment (1.7 × 1021 Nm) with a normal rigidity (30 GPa). As a result, we have to reduce the slip amount in accordance with the ratio between two rigidities. If we assume the ratio to be 1/3, displacement is calculated as 8 mm at BAKO, which is more consistent with the GPS observation. With this assumption, model 11 shows to best ﬁt both the tsunami height and the horizontal displacement at BAKO GPS site. From our discussion, it makes clear that the structural heterogeneity is accounted to the modeling based on crustal deformation. It also implies an easy way to deal with a heterogeneous structure in calculating crustal deformation. With this methodology, it is also possible to estimate the rigidity of the source region by adjusting the slip amount so that the same seismic moment can reproduce both the tsunami data and the far-ﬁeld displacement data.

Acknowledgment Deep acknowledgment is given to Prof. M. Nakamura who pleasantly oﬀered the program of tsunami numerical model; Prof. H. Kanamori, Prof. Y. Fujii, and Prof. K. Satake for the reprints; Prof. Y. Tanioka who gave proper advice to tsunami numerical modeling; and Thuy Ba Ngyuen, Nazri, and D. Muto for their assistance.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Numerical Modeling of the 2006 Java Tsunami Earthquake

b672-V13-ch16

247

References 1. R. E. Abercrombie, M. Antolik, K. Felzer and G. Ekstr¨om, J. Geophys. Res. 106(B4) (2001) 6595–6608. 2. C. J. Ammon, H. Kanamori, T. Lay and A. A. Velasco, Geophys. Res. Lett. 33 (2006) L24308. 3. J. Beckers and T. Lay, J. Geophys. Res. 100(B9) (1995) 18,179–18,193. 4. BMG Indonesia (2006). 5. P. R. Cummins, M. Jia, R. Mleczko, D. Burbidge, H. K. Thio and J. Polet, Eos Trans. AGU, 87(52) (2006), Fall Meet. Suppl., Abstract S14A-08. 6. H. Fritz, J. Goﬀ, C. Harbitz, B. McAdoo, A. Moore, H. Latief, N. Kalligeris, W. Kodjo, B. Uslu, V. Titov and C. Synolakis, Geophys. Res. Lett. 34 (2007) L12602. 7. Y. Fujii and K. Satake, Geophys. Res. Lett. 33 (2006) L24317. 8. Y. Fukao, J. Geophys. Res. 84(B5) (1979) 2303–2314. 9. T. C. Hanks and H. Kanamori, J. Geophys. Res. 84 (1979) 2348–2350. 10. G. Ibrahim, M. A. Untoro and R. Hendrawan, Technical Report, Bandung Institut of Technology, Indonesia, 1989. 11. H. Kanamori, Phys. Earth Planet. Inter. 6 (1972) 346–359. 12. H. Kanamori and Kikuchi, Nature 361 (1993) 714–716. 13. H. Kanamori and E. E. Brodsky, Reports Progress Phys. 67 (2004) 1429–1496. 14. H. Kanamori, Lecture presentation in Nagoya University (2006). 15. T. Kato, T. Ito, H. Z. Abidin and Agustan, Eos Trans. AGU, 87(52) (2006), Fall Meet. Suppl., Abstract S21A-0120. 16. S. Koshimura, http://www.tsunami.civil.tohoku.ac.jp/hokusai2/disaster/ 06 Java/July17 Java.html (2006). 17. H. Latief, N. Puspito and F. Imamura, J. Nat. Disaster Sci. 22(1) (2000) 25–43. 18. H. Latief, A. Riadi and D. Zulkarnaen, www.ocean-partners.org/documents/ News/2006 West Java tsunami.pdf (2006). 19. A. M. Lopez and E. A. Okal, Geophys. J. Int. 165(3) (2006) 835–849. 20. J. Mori and S. Park, Eos Trans. AGU, 87(52) (2006), Fall Meet. Suppl., Abstract S21A-0126. 21. M. Nakamura, Lecture presentation in Nagoya University (2006). 22. K. R. Newcomb and W. R. McCann, J. Geophys. Res. 92(B1) (1987) 421–439. 23. Y. Okada, Bull. Seism. Soc. Am. 75 (1985) 1135–1154. 24. E. A. Okal, Natural Hazard 1(1) (1988) 67–96. 25. E. A. Okal and A. V. Newman, Physics Earth Planet. Inter. 124 (2001) 45–70. 26. A. M. Pelayo and D. A. Wiens, J. Geophys. Res. 97(B11) (1992) 15321–15337. 27. J. Polet and H. Kanamori, Geophys. J. Int. 142 (2000) 684–702. 28. J. Polet and H. K. Thio, Geophys. Res. Lett. 30(9) (2003) 1474. 29. L. Ruﬀ and H. Kanamori, Phys. Earth Planet. Inter. 23 (1980) 240–252. 30. K. Satake, Zisin Second Series 44 (1991) 99–112. 31. K. Satake, Geophys. Res. Lett. 21(23) (1994) 2519–2522. 32. K. Satake, Pure Appl. Geophys. 144 (1995) 455–470.

FA

December 10, 2008 15:45

248

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch16

N. R. Hanifa et al.

33. K. Satake and T. Tanioka, Pure Appl. Geophys. 154 (1999) 467–483. 34. K. Satake, International Handbook of Earthquake and Engineering Seismology, Vol. 81A (2002), pp. 437–450, Ch. 28. 35. K. Satake and Tanioka, Pure Appl. Geophys. 160(10–11) (2003) 2087–2118. 36. G. F. Sella, T. H. Dixon and A. Mao, J. Geophys. Res. 107(B4) (2002) 2081. 37. T. Tanioka and K. Satake, Geophys. Res. Lett. 23 (1996) 1549–1552. 38. T. Tanioka, Lecture presentation in Nagoya University (2007). 39. USGS, http://earthquake.usgs.gov/eqcenter/recenteqsww/Quakes/usqgaf. php (2006).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch17

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

WHY MANY VICTIMS: LESSONS FROM THE JULY 2006 SOUTH JAVA TSUNAMI EARTHQUAKE HASANUDDIN Z. ABIDIN∗,‡ and TERUYUKI KATO† Research Division, Faculty of Civil and Environmental Engineering Institute of Technology Bandung, Jl. Ganesha 10, Bandung, Indonesia

∗Geodesy

†Earthquake

Research Institute, University of Tokyo, Japan ‡[email protected]

A tsunami earthquake at a plate boundary occurred south of Java Island on 17 July 2006 at around 08:20 UTC (15:20 local time). Its magnitude was 7.7 (USGS) and caused a signiﬁcant tsunami along the southern coast of West Java. About 650 people were found dead and 33 were missing. Several reasons can be attributed to the relatively large number of victims caused by this distant earthquake, namely, the unexpected occurrence of tsunami on the southern coast of West Java, its nature of tsunami earthquake which did not give proper warning to the people, occurrence of tsunami with heights reaching up to about 7 m, existence of populated lowland areas and popular recreation areas along the coast, the less eﬀectiveness of emergency response due to the tsunami occurrence just about 2.5 h before the sunset, and the relatively weak capacity of district agencies in hazard mitigation system. This chapter discusses the above reasons along with several recommendations.

1. Introduction The 17 July 2006 South Java earthquake (Mb6.1, Ms7.7, and Mw7.7, Harvard global CMT) occurred about 200 km south of western Java Island (Fig. 1). It occurred on Monday, 17 July 2006 at 3:19:28 pm local time (8:19:28 UTC). The USGS recorded at least 22 aftershocks south of Java ranging between the magnitudes 4.6 and 6.1. Two largest aftershocks measured 6.0 and 6.1 Mw.1 According to Ref. 1, the earthquake occurred as a result of thrust faulting on the boundary between the Australian plate and the Sunda plate. On this part of their mutual boundary, the Australian plate moves north–northeast with respect to the Sunda plate at about 59 mm/y. The Australian plate thrusts beneath the Sunda plate along the Java trench, 249

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch17

H. Z. Abidin and T. Kato

250

Fig. 1.

Earthquake location (after USGS1 ).

south of Java, and is subducted to progressively greater depths beneath Java and north of Java. The earthquake occurred on the shallow part of the plate boundary, about 50 km north of the Java trench. This earthquake generated a tsunami with heights varying from place to place and ranged between 2 and 7 m.2 It is noteworthy to mention that tsunami of higher than 5 m was recorded in the coastal region of about 300 km in length. The generated tsunami severely damaged coastal communities along the southwest and south-central Java provinces in which hundreds of people were killed. Most of the fatalities occurred around the resort area of Pangandaran in Ciamis district (about 260 km southeast of the capital Jakarta) as well as nearby Cilacap and Tasikmalaya districts. The Ministry of Health (MOH) of Indonesia reported that approximately 668 people died, 65 were missing, and 9,299 were in treatment because of the disaster (Table 1). At least three non-Indonesian nationals were among the dead, and the conﬁrmed victims were travelers from the Netherlands, Pakistan, and Sweden.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch17

Why Many Victims Table 1. Province

District

West Java

251

Victims of south Java 2006 tsunami.3 Dead

In-patient

Missing

IDPs

Evacuation site

Tasikmalaya Ciamis Banjar Garut Subtotal

65 413 15 1 494

6124 2741 244 30 9139

2 15 0 0 17

982 4190 0 0 5172

16 9 0 0 25

Central Java

Cilacap Kebumen Banyumas Subtotal

157 10 1 168

104 22 24 150

15 33 0 48

0 0 0 0

0 0 0 0

Yogyakarta

Gunung Kidul Bantul Subtotal Total

3 3 6 668

10 0 10 9929

0 0 0 65

0 0 0 5172

0 0 0 25

IDPs = Internally Displaced Persons.

A report by the National Coordination Agency for Disaster Mitigation of Indonesia on property damage issued on 1 August 2006, stated that 1,986 buildings (including hotels, residential, and government buildings) were destroyed, 497 buildings suﬀered minor damage, 2,992 ﬁshing boats were destroyed and/or missing, 578 ﬁshing boats sustained damage, and approximately 30,000 ﬁshing nets owned by local ﬁshermen were lost. In this case, as many as 63 hotels, more than 50 restaurants, and 150 small shops were destroyed. The tsunami following the earthquake also caused destruction and damage of some 20,000 m2 of road. Moreover, there were at least 15,000–20,000 people whose livelihoods were directly aﬀected. Considering the size and location of the July 2006 earthquake, the number of victims and damages caused by earthquake-triggered tsunami was quite signiﬁcant. In the following sections, factors that contributed to this result will be discussed and analyzed.

2. Unexpected Earthquake and Tsunami Many people and also researchers did not anticipate a tsunami in the southern part of West Java. It is an unfortunate fact that about a year before tsunami struck the area, few scientists made a statement that Pangandaran coast is relatively safe from tsunami.4 Therefore, the central, provincial, and district governments were not prepared for the attack of

FA

December 10, 2008 15:45

252

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch17

H. Z. Abidin and T. Kato

Fig. 2. Previous large earthquakes in the area.5 In these ﬁgures, dashed lines are the boundaries of the earthquake-aﬀected regions.

tsunami, mainly in terms of public education, coastal planning and land use policy, and also infrastructures preparedness for tsunami hazard mitigation. Historically, two large earthquakes in the area were reported by Newcomb and McCann5 on 28 March 1875 and 11 September 1921 (Fig. 2). The 1875 earthquake aﬀected a broad region with moderate intensity (M M = V–VII). No tsunami was reported for this event. The 1921 earthquake (M = 7.5) generated a tsunami that aﬀected about 275 km long of the coastal area.5 Although this 1921 tsunami hit similar areas as the July 2006 event, there is no story about this event in the local community. The most recent tsunami-genic earthquake located in the southern oﬀshore of Java Island was the 1994 East Java earthquake, located about 600 km east– southeast of the 2006 South Java earthquake. This 2 June 1994 earthquake (Ms7.2, Mw7.6) generated tsunami that killed over 250 people and had run-up heights reaching 14 m.6,7 Learning from this July 2006 earthquake and tsunami, anticipative research and mitigation are necessary for other regions prone to tsunami earthquakes. In this regard, researchers should pay more attention to this topic, and government should provide better research funding. 3. Unfelt Earthquake and No-Tsunami Warning Based on interview and other reports, many people along the coast, from Pameungpeuk in West Java to Parangtritis in Yogyakarta, did not feel the earthquake. For those who realized it, only weak ground shake was felt. Therefore, nobody expected the coming of tsunami, which arrived about 30 min after the earthquake. Unfortunately, the Indonesian government ministries and agencies also issued no-tsunami warning to the threatened

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Why Many Victims

b672-V13-ch17

253

areas despite a Local Tsunami Warning Bulletin was issued by the Paciﬁc Tsunami Warning Center (PTWC) at 8:36 UTC (17 min after the earthquake) for the Indian Ocean coast of Indonesia and Australia. Scientiﬁcally, this “unfelt” earthquake that generates larger tsunami is termed a tsunami earthquake. The term “tsunami earthquake” is deﬁned as source characteristics that excite a larger tsunami than expected from seismic wave radiation.8 Based on inversion analysis of tsunami waveforms recorded at six tide gauges (i.e. Rodrigues, Benoa, Christmas Island, Cocos, Broome, and Hillarys stations), Fujii and Satake9 suggested that the 2006 South Java earthquake was a “tsunami earthquake.” Based on the analysis of Rayleigh and body waves of the July 2006 event, Ammon and Kanamori10 found that the rupture was unusually long (about 185 s) and propagated slowly (1.0–1.5 km/s), attributes shared with other tsunami earthquakes. Tsunami earthquakes occur generally at plate boundaries where the plate coupling is weak.11 GPS results given by Refs. 12 and 2 suggested that the motion of the Java Island is ESE and do not show any inﬂuence on the plate convergence of the Australian plate that subducts underneath the Java Island. The velocity ﬁeld at Java shows that the motion is mostly perpendicular to the plate motion, and it indicates that the plate coupling at subducting plate interface is very weak. If this is the case, the Java trench along the south of Java is a potential source of tsunami earthquakes. It is noteworthy to mention that the 1994 East Java earthquake that occurred east of the 2006 South Java earthquake was also thought to be a tsunami earthquake.7 In the case of tsunami earthquake such as the July 2006 earthquake, where coastal communities in general do not feel the ground shaking, timely tsunami warning from the external early warning system becomes crucial and necessary. Moreover, this event suggested that rapid magnitude estimation of earthquake needs to be improved for tsunami-genic earthquakes, so that the reliable tsunami early warning can be properly issued.

4. No Clear Precursory Sign of Residing Sea Level The earthquake occurred at the time of low tide in the aﬀected area (Fig. 3). Precursory sign of tsunami in terms of initial ocean withdrawal did not show too apparently to people around the coast, as people used to see the residing sea level during this time. Therefore, they did not expect the

FA

December 10, 2008 15:45

254

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch17

H. Z. Abidin and T. Kato

Fig. 3. Tide gauge records at Cilacap station, courtesy of the National Coordinating Agency of Surveying and Mapping of Indonesia (BAKOSURTANAL).

attack of tsunami. It was also reported that diﬃculty in recognizing the initial ocean withdrawal might also be contributed by large wind waves breaking at the coast which masked most of the recession of the water at the shoreline that preceded the tsunami.13 Figure 4 shows the situation in the east coast of Pangandaran beach at about the same time of tsunami occurrence 2 days later, i.e. at 15.30 WIB on 19 July 2006. This ﬁgure shows that during this low tide period, the water front is quite far from the shoreline. In this case of “unfelt” ground shaking and also no clear precursory sign of residing sea level, timely tsunami warning from the external early warning system is compulsory.

5. Relatively High Tsunami A relatively high tsunami occurred that reached up to 7–8 m above the sea level (Fig. 5). According to Ref. 13, there was even a pronounced peak on the south coast of Nusakambangan (see the location in Fig. 5) where the tsunami impact carved a sharp trimline in a forest at elevations up to 21 m high and 1 km inland.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

Why Many Victims

9in x 6in

b672-V13-ch17

255

Fig. 4. The east coast of Pangandaran beach at about the same time of the tsunami occurrence.

Fig. 5.

Measured tsunami heights along the coastline (after Refs. 2, 14, and 15).

According to Ref. 16, three main waves were recognized by eyewitnesses. Their ﬁeld survey also found that the maximum height of the second wave ranged from 4.2 to 8.6 m before its breaking; maximum ﬂow depth after the wave’s breaking reached 5 m, and maximum run-up heights

FA

December 10, 2008 15:45

256

AOGS 2007 - SE Volume

H. Z. Abidin and T. Kato

9in x 6in

b672-V13-ch17

Damages caused by the July 2006 tsunami. Fig. 6.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Why Many Victims

b672-V13-ch17

257

reached 15.7 m. The tsunami inundation varied spatially, and in general ranged from about 80 m to 460 m. Tsunami aﬀected about 300 km of the southern coast stretching from the Garut district (e.g. Pameungpeuk beach) of West Java to the Bantul district (e.g. Parangtritis beach) in Yogyakarta. Many houses and buildings along the coast aﬀected by tsunami could not resist the tsunami (Fig. 6).

6. Spontaneous Unsupervised Way of Facing Tsunami The July 2006 tsunami was a real surprise for the coastal communities. Although some people along the beach noticed the incoming tsunami and correctly identiﬁed it, they had a very short time (i.e. about tens of seconds) prior to the tsunami impact. It is therefore understandable that most people panicked and spontaneously tried to save their lives and families as quickly as possible. Unfortunately, most of them have little knowledge about tsunami and the best way to face it. At the time of tsunami, they did not seem to have proper understanding on where the nearest tsunami-safe location was, what the best way to reach that location was, how high of an elevation would provide safety, and how long to stay there. When they saw the high waves approached, most of them just ran, tried to be as far as possible away from the beach. Sometimes, they forgot the trees, buildings, and higher grounds nearby that can be used to escape from the tsunami. Some men managed to climb trees or ran to higher grounds and buildings nearby. In this chaotic situation, women and children were the most vulnerable targets of disaster. For example, in Pangandaran area, which has the highest tsunami victims, the majority of the victims were women and children, i.e. 205 and 78 lives, respectively. The July 2006 tsunami suggested that public education and outreach about tsunami and how to face it are really important. Evacuation drills for the people living in tsunami-prone areas should be conducted on a regular basis.

7. Many Inhabitants Along the Lowland Coastal Areas Along the approximately 300 km long coast aﬀected by the July 2006 tsunami, about 40% to 50% are lowland areas with elevations of less than 10 m. Since the main roads generally run along the coast, many communities are established along the coastal regions. Along these lowland coastal areas

FA

December 10, 2008 15:45

258

AOGS 2007 - SE Volume

H. Z. Abidin and T. Kato

9in x 6in

b672-V13-ch17

Houses and tourist resorts along the coast. Fig. 7.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Why Many Victims

b672-V13-ch17

259

many people live near the beach (many houses are located less than 100 m from the shoreline) without structural or natural protections from tsunami (Fig. 7). Most houses were also built with simple and modest structure. Therefore, it was not surprising that when the 5–7-m tsunami suddenly struck, many houses were damaged and many victims fell. The southern coast of West Java also has many beautiful beaches. Therefore, several tourist resorts are established along the tsunami-aﬀected coast, of which the largest and most well known is Pangandaran beach. Unfortunately, these tourists’ resorts are usually located just on and very close to the beach (Fig. 7). Several domestic and foreign tourists were victims of the July 2006 tsunami. Fortunately, the tsunami hit on Monday afternoon after a four-week long holiday, when there were less tourists on the beaches compared to the preceding Sunday. The dead toll could have been much higher if the earthquake occurred 24 h earlier when the beach was full of holidaymakers. Learning from the July 2006 tsunami, the district government actually has urged communities and tourist resorts to move at least about 300 m away from the shoreline. Unfortunately, based on our latest ﬁeld survey on 21 May 2007, most houses and buildings damaged by tsunami were reconstructed on the same location. Only some of them obeyed the new land use planning policy.

8. Short Period for First Emergency Response The July 2006 tsunami happened just about 2.5 h before the sunset. Only after about half an hour of chaotic period, the ﬁrst eﬀective emergency response could be started. However, about 2 h later, the sunset hampered the eﬀectiveness of emergency response. The disfunction of electrical networks in the aﬀected area and tsunami debris that covered some roads also slowed down the emergency response process. Rescue workers have warned that many people might be buried under the rubble left in the wake of the tsunami. Unfortunately, a relatively weak capacity of district agencies in tsunami hazard mitigation system could not do much in helping these potentially still-alive victims during the ﬁrst night after the tsunami occurrence. Few people were found alive after 2 days, trapped inside the houses and buildings. A well-known Indonesian NGO concluded that in the case of the July 2006 tsunami, the state institutions showed their unsystematic disaster

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

260

9in x 6in

b672-V13-ch17

H. Z. Abidin and T. Kato

management.17 In this case, the District Disaster Management Committee (or SATLAK PB in Indonesian term) and the Provincial Coordinating Unit for the Management of Disaster (or SATKORLAK PBP) with lack of capacity and experience in managing tsunami emergency conditions, could not eﬀectively face the disaster conditions that require rapid actions. The July 2006 tsunami suggested that the capacity of district agencies responsible for disaster (tsunami) mitigation system should be continuously built, in terms of equipment, knowledge, and skills for handling the disaster condition. Evacuation drills for their staﬀs should be conducted on a regular basis.

9. Closing Remarks Relatively many victims of the July 2006 tsunami were caused by several inter-related factors. These factors are: unexpected earthquake and tsunami in the region, hardly felt earthquake by the coastal communities, notsunami warnings given to the aﬀected coastal communities, no clear precursory sign of residing sea level, relatively high tsunami of up to about 7–8 m, spontaneous unsupervised way of people when facing tsunami, many inhabitants along the lowland coastal areas, and a short period for the ﬁrst emergency response. Learning from the July 2006 earthquake and tsunami, there are several lessons that should be taken into account by the researchers, governments, and the coastal communities. They are as follows: 1. Systematic and anticipative tsunami-related research and mitigation program should be properly prepared and conducted for other regions in Indonesia, which are prone to tsunami earthquakes. 2. The development of the Indonesian Tsunami Early Warning System (ITEWS) that is planned to be operational by 2009 should be established as reliable as possible. The most critical part of ITEWS is the development of an eﬀective rapid communications from national authorities to local authorities and from local authorities to numerous communities and large populations along all of the coastal regions of Indonesia. The July 2006 tsunami earthquake gave the lesson that this communication segment of ITEWS has not functioned properly. Moreover, ITEWS could not yet indicate the phenomena of tsunami earthquakes. In short, there are still gaps in science and technology

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

Why Many Victims

3.

4.

5.

6.

7.

9in x 6in

b672-V13-ch17

261

that need to be ﬁlled in the ITEWS. Quicker assessment and warning of potentially damaging tsunami event are needed, including better monitoring and initial assessments of tsunami earthquake and sea-level characteristics. Land use planning along the beach should be regulated based on possible tsunami scenario in the corresponding coastal areas. These land use regulations need to be passed by the government and should be strictly enforced. In order to be well accepted by the coastal communities, the socio-economic implications of these land use regulations should be considered properly and are managed in a win–win solution implementation. In case that the coastal land use regulation cannot be properly imposed, the coastal areas exposed openly to tsunami should be protected by wave breakers and wherever possible by bands of vegetations. The high-rise structures to escape from tsunami should also be built in these areas. Moreover, buildings along the beach should have strong pile foundations and the ground level is better to be left unoccupied and relatively open. In general, the tsunami-prone coastal communities in Indonesia are not well educated. Most of them are ﬁshermen, peasants, and farmers. Public education and outreach about tsunami earthquake, tsunami, and how to face them are critical. For most coastal communities in Indonesia, the best defense besides relying on the warning from ITEWS seems to be better understanding about tsunamis, their local warning signs, and ways to evacuate to safe ground. In this regard, much more awareness raising, education, information provision, and media preparation are needed to build community understanding and readiness for most tsunami-prone coastal communities in Indonesia. In the tsunami-prone coastal areas all over Indonesia, attention should be given to all building houses and infrastructure along the beaches, especially their strength and resilience to tsunami. This should include consideration of tsunami-related structural countermeasures and evacuation procedures, not just in the damaged areas, but also in other exposed and populated coastal zones of the country.18 The capacity of district agencies responsible for tsunami mitigation system in all tsunami-prone coastal regions of Indonesia should be continuously built, in terms of equipment, knowledge, and skills for handling the disaster condition. Evacuation drills for their staﬀs should be conducted on a regular basis.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

262

9in x 6in

b672-V13-ch17

H. Z. Abidin and T. Kato

Acknowledgments The ﬁeld survey in the areas aﬀected by the July 2006 South Java tsunami was made possible by the ﬁnancial supports from the Institute of Technology Bandung (ITB) and Earthquake Research Institute (ERI), The University of Tokyo, Japan.

References 1. USGS (United States Geological Survey), Earthquake Hazards Program, http://earthquake.usgs.gov/eqcenter/eqinthenews/2006/usqgaf/#details. 2. T. Kato, T. Ito, H. Z. Abidin and Agustan, Preliminary report on crustal deformation surveys and tsunami measurements caused by the July 17, 2006 South oﬀ Java Island Earthquake and Tsunami, Indonesia, Earth Planets Space 59 (2007) 1055–1059. 3. World Health Organization (WHO), Java Tsunami of 17 July 2006, Situation Report # 11, August 2006, http://reliefweb.int/rw/rwb.nsf/db900sid/ VBOL-6SCJ85?OpenDocument. 4. Pikiran Rakyat, Tsunami tak Akan Terjadi di Jabar Selatan Tim Ahli Pastikan Pangandaran Aman, Indonesian newspaper, Jumat, 04 Maret 2005, http://www.pikiran-rakyat.com/cetak/2005/0305/04/04p03.htm. 5. K. R. Newcomb and W. R. McCann, Seismic history and seismotectonics of the Sunda Arc, J. Geophys. Res. 92(B1) (1987) 421–440. 6. Y. Tsuji, F. Imamura, H. Matsutomi, C. E. Synolakis, P. T. Nanang, Jumadi, S. Harada, S. S. Han, K. Arai and B. Cook, Field survey of the east Java earthquake and tsunami of June 3, 1994, Pure Appl. Geophys. 144 (1995) 839–854. 7. R. E. Abercrombie, M. Antolik, K. Felzer and G. Ekstrom, The 1994 Java tsunami earthquake: Slip over a subducting seamount, J. Geophys. Res. 106(B4) (2001) 6595–6607. 8. H. Kanamori, Mechanism of tsunami earthquakes, Phys. Earth Planet. Interiors 6 (1972) 346–359. 9. Y. Fujii and K. Satake, Source of the July 2006 West Java tsunami estimated from tide gauge records, Geophys. Res. Lett. 33 (2006) L24317, doi: 10.1029/ 2006GL028049. 10. C. J. Ammon, H. Kanamori, T. Lay and A. A. Velasco, The 17 July 2006 Java tsunami earthquake, Geophys. Res. Lett. 33 (2006) L24308, doi: 10.1029/ 2006GL028005. 11. L. Ruﬀ and H. Kanamori, Seismicity and the subduction process, Phys. Earth Planet Inter. 23 (1980) 240–252. 12. Y. Bock, L. Prawirodirdjo, J. F. Genrich, C. W. Stevens, R. McCaﬀrey, C. Subarya, S. S. O. Puntodewo and E. Calais, Crustal motion in Indonesia from Global Positioning System measurements, J. Geophys. Res. 108 (2003) 2367, doi: 10.1029/2001JB000324.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

Why Many Victims

9in x 6in

b672-V13-ch17

263

13. H. M. Fritz, W. Kongko, A. Moore, B. McAdoo, J. Goﬀ, C. Harbitz, B. Uslu, N. Kalligeris, D. Suteja, K. Kalsum, V. Titov, A. Gusman, H. Latief, E. Santoso, S. Sujoko, D. Djulkarnaen, H. Sunendar and C. Synolakis, Extreme runup from the 17 July 2006 Java tsunami, Geophys. Res. Lett. 34 (2007) L12602, doi: 10.1029/2007GL029404. 14. Fachrizal, S. Pribadi and I. Hermawan, Laporan Survey Gempabumidan Tsunami Selatan Jawa Barat 17 Juli 2006 (Badan Meteorologidan Geoﬁsika, 2006). 15. Y. Tsuji, S. S. Han, Fachrizal and I. Gunawan, Field survey of the tsunami inundated heights due to the Java Tsunami (2006/07/17) along the coast on the Indian Ocean in Java Island, http://www.eri.u-tokyo.ac.jp/tsunami/ javasurvey/index e.htm. 16. F. Lavigne, C. Gomez, M. Giﬀo, P. Wassmer, C. Hoebreck, D. Mardiatno, J. Prioyono and R. Paris, Field observations of the 17 July 2006 Tsunami in Java, Nat. Hazards Earth Sys. Sci. 7 (2007) 177–183. 17. Indonesian Forum for Environment (WALHI), “Solidarity in Adversity,” A Relief Report of Jogjakarta and Central Java’s Quakes and West Java’s Tsunami, November 2006, http://www.walhi.or.id. 18. UN-ISDR (United Nations International Strategy for Disaster Reduction), Information Note on the 17 July 2006 Java Tsunami and the Performance of Early Warning Systems, Prepared on 21 July 2006 as a brieﬁng for ISDR partners by the ISDR secretariat and ISDR Platform for the Promotion of Early Warning, http://www.unisdr.org/ppew/tsunami/pdf/ Information-note-on-Java-tsunami.doc.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

This page intentionally left blank

b672-V13-ch17

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch18

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

EARTHQUAKE POTENTIAL IN MYANMAR HLA HLA AUNG Myanmar Earthquake Committee, MES Building Hlaing University Campus, Yangon, Myanmar [email protected]

Myanmar region is generally believed to be an area of high earthquake potential from the point of view of seismic activity which has been low compared to the surrounding regions like Indonesia, China, and Pakistan. Geoscientists and seismologists predicted earthquakes to occur in the area north of the Sumatra–Andaman Islands, i.e. the southwest and west part of Myanmar. Myanmar tectonic setting relative to East and SE Asia is rather peculiar and unique with diﬀerent plate tectonic models but similar to the setting of western part of North America. Myanmar crustal blocks are caught within two lithospheric plates of India and Indochina experiencing oblique subduction with major dextral strike-slip faulting of the Sagaing fault. Seismic tomography and thermal structure of India plate along the Sunda subduction zone vary from south to north. Strong partitioning in central Andaman basin where crustal fragmentation and northward dispersion of Burma plate by back-arc spreading mechanism has been operating since Neogene. Northward motion of Burma plate relative to SE Asia would dock against the major continent further north and might have caused the accumulation of strain which in turn will be released as earthquakes in the future.

1. Tectonic Setting Myanmar is characterized by an active tectonic setting. It represents an evolving continent of two diﬀerent crustal formation history consisting of Burma plate and Indochina plate. Burma plate is a unique plate, detached from northward moving India plate and accreted to Indochina plate during Late Cretaceous–Early Eocene. India plate is subducting obliquely beneath Burma plate along Sunda subduction zone consisting of four trenches: IndoMyanmar trench, Andaman–Nicobar trench, and Sumatra trench (Fig. 1). Burma plate acts as an individual plate since Neogene, but it is believed to have been more or less continuous with Sumatra prior to Neogene. Myanmar is composed of three lithotectonic provinces, each of which is signiﬁcantly diﬀerent in geologic history, stratigraphy, tectonic, and 265

FA

December 10, 2008 15:45

266

Fig. 1.

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch18

H. H. Aung

Schematic diagram showing present plate tectonic conﬁguration of Burma.

peleontologic indicators. The boundaries between them are major active faults trending nearly N–S Kyunthayar fault, Kabaw fault system, and Sagaing fault from west to east. The three lithotectonic entities from west to east are Rakhine coastal province, Rakhine western range province, and central Myanmar basin province (Fig. 2). Paleomagnetic data are not available to estimate paleolatitudes of each of the provinces. From the studies of micro- and macrofossils, all of these lithotectonic provinces were more or less in southern site than the present location. It is in agreement with constraints given by tectonic data. Eastern part of Myanmar, east of the Sagaing fault is Shan-Tanintharyi Province. It is part of Indochina Peninsula.

2. Historical Earthquakes and Recent Seismic Activities in Myanmar Historical earthquakes were not recorded instrumentally in ancient time, but some of the visiting geologists from GSI (Geological Survey of India) have described their own experience on earthquakes of Myanmar in a book The Geology of Burma by Chhibber.4 In this book, earthquake events were described year by year from 1762 to 1931. Visual observations of earthquakes felt during 1829–1958 and epicenters of instrumentally recorded earthquakes during 1906–1958 were described in a book

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Earthquake Potential in Myanmar

b672-V13-ch18

267

Fig. 2. Major tectonic lineaments and lithotectonic provinces of Burma. RC — Rakhine Coastal Province; RWR — Rakhine Western Range Province; CBB — Central Burma Basin Province.

The Problematic Seismotectonics and Seismic Zoning in the Territory of Union of Burma by Gorshkov.8 Most of the earthquakes having magnitudes of VII–VIII on Rossi-Forel scale were accompanied by some land deformation. The epicenters of large-scale earthquakes generally lie on or near the major faults (Fig. 3, Table 1). Remarkable earthquakes in the historical records are connected with folding and faulting movements. It

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch18

H. H. Aung

268

Fig. 3.

Epicenters of important earthquakes and general lines of weakness.4

is assumed that a land deformation takes place over an area under which the strain energy related to an earthquake is accumulating. Coggin Brown (1914) gave the following remarks on the Kyaukyan fault: The major axis of the elongated oval enclosed by the innermost isoseismal line coincides nearly with the Kyaukyan fault in the Northern Shan

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch18

Earthquake Potential in Myanmar Table 1.

Major historical earthquakes in Myanmar.4,8,11

Date

Location

Magnitude or brief description

2 April 1762

Kyaukphyu Earthquake of 6 Feb. 1843, 30 Oct. 1833, 3 Jan. 1848 Ava Earthquake of 23 March 1839

Earthquake of 24 Aug. 1858 Mandalay Earthquake of 16 Feb. 1871 Earthquake of 31 Dec. 1881

Rangoon Earthquake of 23 July 1884 Maymyo Earthquake of 23 May 1912 Pegu Earthquake of 5 July 1917

◦

◦

Near 19 12 N; 95 12 E Near Mandalay

◦

Near 19◦ N; 95 12 E

West of Andaman islands (Approximately 12◦ to 15◦ N; 89◦ E)

21◦ N, 97◦ E

Rangoon Earthquake of 17 Dec. 1927

Swa Earthquake of 8 Aug. 1929

269

◦

Near 90◦ N, 96 12 E

Its shaken vast area 16◦ –23◦ N; 87–94◦ E and felt near Chittagone in northern part of the eastern Bay of Bengal. Eruption of mud volcanoes and other damaging earthquakes followed. Terrible destruction, shattering of Pagoda of Mingun, one of the largest masses of solid brickworks in the world. Destructive, most severe near Thayetmyo and Pyay. No damage. M. 7, felt 5,800,000 sq km The epicenter — The Bay of Bengal large portion of Indian Peninsula, caused small tsunami, 1 m peak through in height, along the Indian and Arakan coasts. Moderate intensity, doors rattled, no damage. M. 8, 1,000,000 km2 disturbed. The umbrella of the famous Shwemawdaw Pagoda was shaken down, small pagoda destroyed. Severe, felt 15,000 sq km from Kyangin to Dedaye along lower reaches of Ayeyarwady. Bent railroad tracks, bridges, and culverts collapsed, loaded trucks overturned. (Continued )

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch18

H. H. Aung

270

Table 1. Date

Location

Pegu Earthquake of 5 May 1930

17◦ N, 96 12 E

Phyu Earthquake of 3 Dec. 1930 Htawgaw Earthquake of 16 Dec. 1929 Kamaing Earthquake of 28 Jan. 1931

18◦ N, 96 12 E

(Continued ) Magnitude or brief description

◦

◦

Near Burma–China border 25◦ 31 38 N, 96◦ 43 5 E

M = 7.3, Imax = IX, in the zone trending north– south for 70 km south of Pegu (therefore parallel to the Sagaing fault). M = 7.3, railroad tracks twisted. Cracks developed, landslides occurred. Big ﬁssures, wooden houses wrecked, pagodas destroyed.

State. A relationship is evident between the places where the maximum intensities were experienced and the greatest damage done and the known course and probable continuation of these faults.

Seismic activities along the Kabaw Fault System are shown in Fig. 4. Shallow earthquakes in southwestern part of Shan Plateau region, described in U.S.G.S. Earthquake Catalog (1990–1995) are as shown in Fig. 5. — Shallow earthquake was recorded in Southern Shan State, and it seems to be associated with the northern portion of Pawn Chaung Fault Zone trending N–S. — 6 December 1992 earthquake occurrence may be connected with NNW– SSE trending Panlang Fault Zone. — 15 June 1985, 2 June 1990, and 4 January 1991 earthquake events may be associated with the Labahku Fault Zone trending N–S. — 20 December 1985 and 17 February 1975 seem to be related with NW– SE trending Papun Fault Zone. Relationship between the major faults and epicenters of earthquake in Central Burma Basin (Data from U.S.G.S. Earthquake Catalog) is shown in Fig. 6. The epicenters for shallow earthquakes in the southern part of Shan Plateau region are closely associated with three major fracture zones: the Sagaing Fault Zone, Panlaung Fault Zone, and Papun Fault Zone.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch18

Earthquake Potential in Myanmar

Fig. 4.

271

Map of seismic activities along the Kabaw Fault System.

The most remarkable tectonic lines of N–S direction and NW–SE direction are identiﬁed as active faults. Fault plane solutions for shallow and intermediate earthquakes and east-dipping inclined zone of intermediate depth earthquakes suggest that a slab of oceanic lithosphere was subducted to the east under Burma. P -axes for fault plane solutions of intermediate shocks also trend N–S, parallel to

FA

December 10, 2008 15:45

272

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch18

H. H. Aung

Fig. 5. The map showing major lineaments, faults, and epicenters of shallow earthquakes in southwestern part of Shan Plateau region.

the strike of the subducted slab, and may indicate that the hanging slab is being dragged north by India through the surrounding asthenosphere. The Sagaing right-lateral fault accommodates most of the right-lateral slip of India past Indochina.11 It now appears that all the historical earthquakes recorded since 1762, recent earthquakes from U.S.G.S. earthquake catalog are shallow to

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch18

Earthquake Potential in Myanmar

Fig. 6.

273

Relation between major fault and earthquake events in Central Burma Basin.

intermediate earthquakes which are closely related to the major active faults of two structurally diﬀerent trends in N–S direction and NW–SE direction.

3. Structural Features Along the Sagaing Transform Belt Major tectonic lines which are directly connected with seismic activities and the deformation styles along major faults reveal that some of the faults are still active and these active processes have taken place by the ENE–WSW compressional and NNW–SSE extensional schemes since Neogene time. Their structural arrangements in lithotectonic provinces follow simple shear scheme. Simple shear strain results from the internal rotation of structural

FA

December 10, 2008 15:45

274

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch18

H. H. Aung

fabric elements within upper crustal layers operated upon by an external couple between the movement of two lithospheric plates. In this simple shear, crustal blocks are deformed within a couple operating parallel to the transform plate motion, or parallel to local strain on a segment of a fault. Folds trend obliquely to deeply buried movement zones or to a through-going fault, and associated faults can be classiﬁed into diﬀerent types: synthetic (R), antithetic (R ), extension, thrust, and P -shear faults. When simple-shear deformation has been suﬃcient, a conjugate system of vertical faults results (synthetic and antithetic) and extension faults with normal slip and dipping oppositely begin to form. All of these faults rotate as displacement accumulates and then en echelon, right-slip faults break through the surface and a rhomb-shape basin is formed. Pull-apart basins are rhombic in shape and have been termed rhomb-graben. These pull-apart basins have two sets of margins: the border parallel to the transform fault and those parallel to the extensional direction. The orientation of pull-apart margins is oblique to the trend of the major transform fault, and it may be controlled by the orientation of extension fault in the simple-shear system that is roughly perpendicular to the fold axes or extension direction. When a pull-apart basin grows to a critical size, mantle material rises into the pull-apart holes or cracks from depth, and spreading mechanism takes place. These simple-shear deformation patterns are the characteristics of the transform belt.5,16,21 The structural features west of the Sagaing fault include thrust faults, reverse faults, normal faults, strike-slip faults, and folds. Among these structures, the Sagaing fault is a transform plate boundary between Burma plate and Indochina plate. It is a long and straight major lineament across the entire length of Myanmar trending nearly N–S and dipping steeply. Sagaing fault extends northward from the Gulf of Mottama, then down into Andaman Sea with braided pattern and en echelon structure, and then connects with the northwestern plate edge in Andaman spreading center. The Kabaw fault is considered to be a reverse fault, but it has more likely accommodated considerable amounts of right-lateral strike-slip movement.12,23 The Kyunthaya fault trends NW–SE, and it has strike-slip component of displacement. The Ramree fault is located between Sittway and Thantwe. A series of basin formation with diﬀerent sizes and shapes can be found along the Burma plate both in Myanmar2 and in Andaman Sea. These basins from south to north for 1,100 km length in Myanmar are the Ayeyawady Delta Basin, Pyay Basin, Sittaung Basin, Salin Basin, Chindwin

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Earthquake Potential in Myanmar

b672-V13-ch18

275

Basin, and Hukawng Basin and the Mergui Basin, East Basin, and Central Andaman Basin in the Andaman Sea. Central Andaman Basin is one of the pull-apart basins which is ﬂoored by young basalts, covered by sediments shed by the Ayarwaddy, Sittaung, and Thanlwin Rivers (Fig. 7). The basins are separated from each other by uplifts, as the Basins and Ranges

Fig. 7.

Basin formations along the Burma plate.

FA

December 10, 2008 15:45

276

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch18

H. H. Aung

structures are controlled by normal faults trending ENE–WSW parallel to cross-fault and perpendicular to major fold axes and thrust sheets. The Central Andaman Basin is a pull-apart basin where crustal fragmentation and northward motion of Burma plate by back-arc spreading mechanism have been operating since Neogene. Spreading rate perpendicular to the spreading axis is 38 mm/y.7 GPS data gives 20–35 mm/y displacement rate along the Sagaing fault20 and spreading rate of 20–30 mm/y in the Central Andaman Basin7 accommodating migration of Burma plate since the opening of Central Andaman Basin in Neogene. NW- or SE-dipping cross-faults are common in Central Burma Basin along the ENE–WSW direction. The two sets of faults indicate the NNW–SSE-oriented simpleshear scheme. Series of en echelon folds associated with faults are found at 20◦ N latitude. Positive ﬂower structures with steep, north-striking reverse faults are also found in Southeastern Uplift. The Mann anticline is asymmetrical and formed above a steeply dipping reverse fault. Tertiary sequences reveal numerous anticlines and synclines, trending NW–SE, sometimes swinging to NNW–SSE direction. These folds are transected by E–W to ENE–WSE trending faults and N–S striking faults with vertical displacement over 300 m. Most of the anticlines are associated with thrust faults on their ﬂanks, trending NW–SE, dipping E or W. For example, the Gwegyo thrust fault extends for over 50 km along the NW–SE direction dipping west, and is associated with an anticline. The Yenangyat–Chauk thrust fault forms at the east margin of Salin Basin. The 60-km-long Yenangyat-Chauk anticline is strongly asymmetrical (Fig. 8). East-dipping thrust faults and east–north–east-striking cross-faults can be found throughout the west limb of Salin subbasin and in Central Burma Basin. Pressure ridges like Sagaing Ridge appear to originate between en echelon P -shears. The Pondaung and Ngahlaingdwin anticlines trend NW–SE near the Western Outcrops located at the western part of Central Burma Basin, then change to NNW–SSE direction. There is a graben-like N–S striking zone in the Western Outcrops near Myittha and Kabaw Valleys. Among the structures described above, the deformation styles and structural orientations are in accord with simple shear strain pattern, where ENE–WSW trending normal fault and cross-faults follow antithetic direction and splays from the Sagaing fault follow synthetic direction. Major fold trends are in accord with the simple-shear pattern, where the axes are oblique to the strike of the Sagaing fault in NW–SE direction, then swing to the NNW–SSE direction, and become sub-parallel to the Sagaing fault.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Earthquake Potential in Myanmar

b672-V13-ch18

277

Fig. 8. Major structural features in central part of Central Burma Basin (Modiﬁed from the map by Pivnic et al.14 ).

NW–SE trending thrust faults are associated with folds on their ﬂanks, dipping east or west. These active processes are the potential sources for earthquake in future. Most of the earthquakes studied over the world are preceded by the creation of tensile structures, the degradation of which coincides with the release of seismic energy. After that, tensional regimes begin to decay and gradually change into compressional ones.19 Northward motion of Burma plate relative to Indochina plate leads to continental

FA

December 10, 2008 15:45

278

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch18

H. H. Aung

Fig. 9. Schematic post-accretionary northward dispersion model of Burma plate. CAB — Central Andaman Basin; WAF — West Andaman Fault; SEU — Seuliman Fault; SFS — Sumatra Fault System; SF — Sagaing Fault.

collision further north, which may cause the accumulation of strain that will be released in a future earthquake (Fig. 9). 4. Discussion and Conclusions Normal faults and cross-faults strike in the ENE–WSW direction, thrustsheets extend along NW–SE direction, and splay faults run nearly N–S direction and subparallel to the Sagaing fault. Folds are along the NW–SE direction, sometimes swing to NNW–SSE. N–S striking horst and graben are associated with normal faults trending ENE–WSW.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Earthquake Potential in Myanmar

b672-V13-ch18

279

Series of pull-apart basin formations are the sites for sediments and hydrocarbon accumulation, some of which are ﬂoored with young basalt. When rhombic pull-apart basins have enlarged, the crustal plate at depth are moving laterally by the spreading mechanism like in the Central Andaman Basin where spreading accommodates northward rifting and drifting of Burma plate from the spreading center. Rate of spreading is 20–30 mm/y in the CAB.7 Slip rate of the Sagaing fault is 20–35 mm/y.20 Surface and subsurface structural patterns of folds and faults and smaller geomorphic forms indicate that the Sagaing Transform Belt has been experiencing NNW–SSE simple extensional scheme and ENE–WSW compressional deformation since Neogene. All the structural styles along the Sagaing Transform Belt follow the simple-shear scheme that resulted in rotation and translation of crustal blocks caught within the upper crustal layer by two diﬀerently directed movements of India and Indochina lithospheric plates. Seismic tomography and thermal structure of India plate along the Sunda subduction zone show that the India lithospheric plate under Andaman–Nicobar trench and Indo-Myanmar trench diﬀers in the thickness of the underlying plate, dip angle of Wadati–Benioﬀ zone, and the age and temperature of supra-slab mantle wedge under India plate, when compared with those of northern Sumatra. The subducted slab under Andaman and Myanmar region is very thick and dip steeply; therefore, it is negatively buoyant.17 Strong partitioning has been accommodated by the back-arc spreading process in the Central Andaman Basin, associated with displacement along the Great Sumatra fault in Sumatra and displacement along the Sagaing fault in Myanmar. Detailed and continuous studies along these major faults (described in Sec. 3) are greatly needed for the better estimation of earthquake probabilities in Myanmar.

References 1. Ba Than Haq, Metallogenic provinces and prospects of mineral exploration in Burma, Contrib. Burmese Geol. 1(1) (1981) 1–16. 2. F. Bender, Geology of Burma (Gebruder Borntraeger, Berlin Stittgart, Germany, 1983). 3. Chamot-Rooke et al., From partial to full strain partitioning along the IndoBurmese hyper-oblique subduction, Marine Geol. 209 (2004) 313–327. 4. J. C. Brown, Burma earthquake of may, Mem. Geol. Surv. Ind. Xlii (1914) 1–147.

FA

December 10, 2008 15:45

280

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch18

H. H. Aung

5. H. L. Chhibber, The Geology of Burma (Macmillan and Co. Limited, St. Martin’s Street, London, 1934). 6. H. Cloos, Experimente zur inneren Tektonik: Centralblatt fur Mineralogie, Geologie, and Palaontologie, Abteilung B (1928), pp. 609–621. 7. J. C. Crowell, The recognition of transform terrane dispersion within mobile belts, Principles Appl. Terrane Anal. A Guidebook, ed. J. C. Crowell, Univ. of Calif., Santa Barbara, pp. 51–61. 8. J. R. Curray, Tectonics and history of the Andaman Sea region. J. Asia Earth Sci. XX (2005) 1–42. 9. G. P. Gorshkov, The Problematic Seismotectonic and Seismic Zoning in the Territory of Union of Burma, Byull. Sovj. Seim. 12 (in Russ.) (1959). 10. H. H. Aung, Geological Study of Pathi Chaung Granite as an aid to tectonotratigraphic terrane analysis, M.Sc Thesis, University of Yangon (1987). 11. D. E. Karig, Accreted terranes in the Northern Part of the Philippine archipelago, Tectonics 2(2) (1983) 211–236. 12. A. Y. Le Dain, P. Tapponeir and P. Molnar, Active faulting and tectonics of Burma and surround regions, J. Geophys. Res. 89(13) (1984) 453–472. 13. H. Maung, Transcurrent movements in Burma–Andaman sea region, Geology 15 (1987) 911–912. 14. S. Ohya, Continued challenge in earthquake risk management, in Proc. 1st Bangladesh Earthquake Symp. (Dhaka, 14–15 December 2005). 15. D. A. Pivnic et al., Polyphase deformation in fore-arc/back-arc basin, Salin subbasin, Myanmar (Burma), AAPG Bull. 82(10) (1999) 1837–1858. 16. C. Rangin, Deformation of Myanmar, Results of GIAC Projects, 1997 GIAC Conf. (Yangon, Myanmar, 1997). 17. W. Ridel, Zur Mechanik geologischer Brucher-scheinungen: Centralblatt fur Mineralogie, Geologie, and Palaontologie, Abteilung B (1929), pp. 354–368. 18. M. H. Ritzwoller et al., Structural context of the Great Sumatra-Andaman Islands earthquake, Science, 2005. 19. Saw Ngwe Khaing: Structural study of the northwestern part of Kwingauk Area, Ingabu Township, Ayeyarwady Division, Master of Research, University of Yangon May, 2007. 20. G. S. Vartanyan, The HGD Monitoring System for Land-use Planning (Earthquake Prediction) Atlas of Urban Geology 8 (United Nations, New York, 1996). 21. Vigny et al., GPS network monitors on Sagaing fault, Myanmar, Trans. Suppl. 82 (2003). 22. R. E. Wilcox et al., Basic wrench tectonics, Bull. Am. Assoc. Petroleum Geol. 57 (1973) 74–96. 23. Win Naing, Cenozoic deformation in the southwestern Shan Plateau with special emphasis on the Ductile Shearing along the Papun Fault Zone, PhD Dissertation, University of Yangon (May 2006). 24. K. Zaw, Comment and reply on transcurrent movement in the BurmaAndaman sea region, Geology 17 (1989) 93–97.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch19

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

LOOKING INTO A SANDPILE BY PHOTO-ELASTICITY AND DISCRETE ELEMENT METHOD NAOTO YOSHIOKA Graduate School of Integrated Science, Yokohama City University Seto 22-2, Kanazawaku, Yokohama, 236-0027, Japan (Now at Fukada Geological Institute, 2-13-12, Honkomagome Bunkyo-ku, Tokyo, 113-0021, Japan, [email protected] and Japan Agency for Marine-Earth Science and Technology 3173-25, Showa-machi, Kanazawaku, Yokohama, 236-0001, Japan [email protected]) HIDE SAKAGUCHI Institute for Research on Earth Evolution (IFREE) Japan Agency for Marine-Earth Science and Technology (JAMSTEC) 3173-25, Showa-machi, Kanazawaku, Yokohama, 236-0001, Japan [email protected]

In sandpile experiments in which a conical sandpile is built on a circular disk, the behavior of avalanches (frequency-size distribution) is determined solely by the ratio of the grain size to the disk size. Depending on whether the ratio is less or larger than a critical value, the behavior becomes self-organized criticality type which obeys Gutenberg–Richter’s law or the characteristic earthquake type where large and periodic avalanches become dominant. In order to elucidate the underlying physics, we have performed a series of experiments using a photo-elastic material. A two-dimensional sandpile was built by dropping disks made of a photo-elastic material between two transparent walls. We have succeeded directly to see the force chains inside the sandpile. A numerical simulation was also performed using the discrete element method (DEM). The result well coincides with that of the physical experiments. A detailed observation of the results of the experiment and the simulation suggest a possible mechanism of avalanches: for a sandpile in a critical state, what determines the size of avalanches may be some kind of inhomogeneity. When the pile is small, the size of avalanches is small, because the uneven distribution of particles on the surface may be the dominant inhomogeneity of the system. When the pile becomes large, however, the deep inside of the pile is tightened and consolidated while the part near the surface is never tightened because of the repeated avalanches. The contrast in strength becomes the main factor of the inhomogeneity.

281

FA

December 10, 2008 15:45

282

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch19

N. Yoshioka and H. Sakaguchi

1. Introduction Since Bak et al.1 (referred to as BTW model) proposed the concept of Self-Organized Criticality (SOC), Gutenberg–Richter’s law (power law) has been regarded as a typical example of SOC. On the other hand, Schwartz and Coppersmith2 proposed the concept of “characteristic earthquakes” which occur periodically with a relatively narrow range of magnitudes near the maximum. The model implies a non-linear frequency–magnitude relationship that is dominated by the characteristic event and that has a low b-value (the slope of the frequency– magnitude plots) in the moderate-magnitude range. After the appearance of BTW model, a number of numerical studies have been performed (e.g. Refs. 3–7). The results have all shown that the frequency–size statistics satisfy the Gutenberg–Richter’s power law. However, the physical sandpile experiments using real sand yield both SOC-type behavior and characteristic earthquake (CE)-type behavior, depending on the experimental condition.8−10 The condition which determines the behavior of sandpile avalanches is the ratio of sand diameter to the size of the disk on which the sandpile is formed. The transition from the SOC-type to CE-type drastically occurs when the ratio of grain size to disk radius is reduced to about 0.02.10 A group from the University of Chicago proposed a simple explanation for the transition.11,12 They argue that there are two distinct angles of surface slope: the angle of repose θr to which the slope angle returns after a large avalanche occurs, and the maximum angle θm which is the angle just before a large avalanche occurs. Between the two angles, the slope is metastable. From their experiments using an open box with sand added from the top and a rotating semi-cylindrical drum,13 they found the diﬀerence angle δ = θm − θr

(1)

to be typically a few degrees. Denoting the length of the slope by L, the two angles can be distinguished only when the grain diameter d is small enough to satisfy the condition d < L · δ,

(2)

because if (2) is not satisﬁed, any grain added at any point on the slope will immediately exceed θm and the slope is always unstable. Thus, the

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch19

Looking into a Sandpile by Photo-Elasticity and Discrete Element Method

283

transition occurs when L

d . δ

(3)

For δ = 2◦ , we have L 30d or d/L 0.03 which roughly agrees with the experimental observation. However, this explanation is physically unreasonable, because a grain added at any point on the slope does not know how long the slope extends below itself. The stability of a grain must be determined by the local physics such as the interaction between a grain and its neighbors. In order to elucidate the underlying physics, we have been trying to “see” the force chain structure inside the sandpile with the aid of photoelasticity and numerical simulation by the discrete element method (DEM). This chapter describes the results both of the physical experiment using a photo-elastic material and of DEM simulation.

2. Photo-Elastic Experiment 2.1. Experimental setup and procedure Some people have already tried to “see” the inside of a sandpile with the aid of photo-elasticity.14 –16 We used GFP1200 Photoelastic Strain Analysis System (StressPhotonics Co. Ltd, USA). PSM-4 made by Vishay Micromeasurements USA was employed as a photo-elastic material. Disks with diameters of 12 mm and 9 mm were cut from a sheet of PSM-4. The ratio of the number of disks was 4:6 for 12 mm and 9 mm. The thickness of the disks is 6.35 mm (1/4 inches). A two-dimensional pile was created between two transparent walls with 7 mm width by dropping the disks one by one, which is shown in Fig. 1. It has been pointed out that the force chain structure inside a sandpile is strongly dependent on the formation procedure or history15,17 : a localizedsource procedure yields a dip (stress minima under the center of a pile), while a raining procedure does not. Although our interest in this chapter is not on the existence of a dip, the procedure employed in the experiments was a localized-source procedure. Under the assumption of the symmetry of the two-dimensional pile, only half a part of the pile was created in some experiments as shown in Fig. 1. The length of the bottom base was changed to see the eﬀects of size of the pile on the behavior of avalanches. The number of disks on the base was continuously counted. Two examples of the change in the number of disks

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

b672-V13-ch19

N. Yoshioka and H. Sakaguchi

284

Fig. 1.

9in x 6in

Dropping disks one by one between two transparent walls with a width of 7 mm.

are shown in Fig. 2: (a) a small base (30 cm) and (b) a large base (100 cm). The x-axis is the number of dropped disks. Avalanches sometimes occurred, and the number of disks on the base was reduced. The photo-elastic images were taken at the points shown by solid circles. A large base seems to yield, not completely, but periodic large avalanches, while a small base yields irregular small avalanches.

(a) 30cm (070118-1)

(b) 100cm (070208)

14 37

260 240 15 220 200

8 1 0

5 100

20 33 24 28 200

Number of drops

Fig. 2.

300

38

Number of disks on base

Number of disks on base

280

42

2800

48

69

56

29 18 2600

43 5

2400

62 49

30 19

1 0

57 72

6 1000

2000

Number of drops

Change in the number of disks on the base: (a) 30 cm base and (b) 100 cm base.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch19

Looking into a Sandpile by Photo-Elasticity and Discrete Element Method

285

2.2. Results An example of photo-elastic images is shown in Fig. 3 (taken at point 69 in Fig. 2(b)). The length of the bottom base is 100 cm. More than 2,800 disks are piled up on the base. This ﬁgure is a usual photograph with 60% transparency superimposed over a photo-elastic image taken by GFP1200, so that the uppermost boundary of the pile and individual disks can be seen. The dark colors (from red to blue) represent the magnitude of strain. A number of force chains are clearly seen in the pile. Observations of Fig. 3 suggest some structural features of the pile. First, only a portion of the total number of disks contributes to support the structure. Many disks between the developed pillars are not responsible for the support. Second, many arch-shaped pillars are built at the bottom of the pile. Third, a few pillars are developed parallel to the surface. All these structures are self-organizedly created. Next we focus on the change in structure by an avalanche. Figure 4(a) shows images of a small pile with the base length of 30 cm. The numbers on the ﬁgure corresponds to the points shown in Fig. 2(a). The top left is the image at the point 37 just before the avalanche and the top right is after the avalanche at the point 38. GFP1200 Strain Analysis System allows us to subtract an image from another image. The bottom ﬁgure was obtained

Fig. 3.

An example of photo-elastic images.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch19

N. Yoshioka and H. Sakaguchi

286

(a)

37

38

30cm

37-38

(b)

69

72

100 cm

69-72

Fig. 4. Change and diﬀerence in photo-elastic images: (a) 30 cm base (top: snapshots at 37 and 38 in Fig. 2(a), bottom: after subtraction of 38 from 37); (b) 100 cm base (top: snapshot at 69 and 72 in Fig. 2(b), bottom: after subtraction of 72 from 69).

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch19

Looking into a Sandpile by Photo-Elasticity and Discrete Element Method

287

by subtracting 38 from 37. The residual means the force chains which were lost or weakened by the avalanche between 37 and 38. About 60 disks were lost after the avalanche. The eﬀect of the avalanche reaches the bottom of the pile. Similarly, Fig. 4(b) is the case of a large base (100 cm) which corresponds to Fig. 2(b). A large avalanche occurred after the point 69 by adding one disk on the pile, resulting in the loss of nearly 400 disks. The subtraction shows that the avalanche between 69 and 72 aﬀects deep inside the pile. However, the deeper part of the pile was not aﬀected at all. This is important in considering the mechanism of avalanches, which will be discussed later.

3. Simulation by Discrete Element Method (DEM) 3.1. DEM specifications Discrete Element Method (DEM) was ﬁrst introduced by Cundall and Strack18 and has been widely used as an analytical method for granular materials. We already reported the results of numerical simulation using DEM for an analysis of a gouge layer under increasing shear stresses.19 Here we have performed a simulation by DEM for the analysis of inside structure of a sandpile. The simulation was two-dimensional. Each circular particle with diameter 6 mm or 10 mm was put onto the apex of the pile one by one. The friction between particles is 20◦ (coeﬃcient of friction, 0.36). The size of the table is 100 cm and 200 cm, which correspond to the base lengths of 50 cm and 100 cm, respectively, in the photo-elastic experiments.

3.2. Results Two examples are shown in Fig. 5: (a) 100 cm table and (b) 200 cm table. Color on the top of (b) represents kinetic energy of the particles, after one particle is added to the pile. The bottom ﬁgures of (a) and (b) show the force chains in the pile. It is noted that the features seen in a photo-elastic pile (see Fig. 3) are also seen in the force chain proﬁles: only a portion of the total number of disks contributes to support the structure; many arch-shaped pillars are built at the bottom of the pile and a few pillars are developed parallel to the surface. These ﬁgures are snapshots of a large number of successive ﬁgures. The dynamic change in force chains and the motion of each particle with time

FA

December 10, 2008 15:45

288

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch19

N. Yoshioka and H. Sakaguchi

Fig. 5. Examples of DEM simulation: (a) 100 cm table and (b) 200 cm table. Color on the top of (b) represents kinetic energy of the particles after one particle is added to the pile. The bottom ﬁgures show the force chains in the pile.

can be seen as an animation. From careful observations of the animation, we notice that the entire pile is aﬀected when an avalanche occurs on the small pile (Fig. 5(a)). This is consistent with the results of photo-elastic experiment (see bottom ﬁgure of Fig. 4(a)). When a large avalanche occurs on a large pile, it aﬀects a thick layer of the pile (Fig. 5(b)). However, the deepest part of the pile is left unaﬀected. This is also consistent with the results of photo-elastic experiment (see bottom ﬁgure of Fig. 4(b)).

4. Discussion and Conclusions We have succeeded in directly seeing the force chain structure inside the pile by using a photo-elastic material and the observation system. We have

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch19

Looking into a Sandpile by Photo-Elasticity and Discrete Element Method

289

Fig. 6. A schematic illustration of the diﬀerence between (a) a small pile and (b) a large pile. When the pile becomes large, the deep inside of the pile is tightened and consolidated while the part near the surface is never tightened because of the repeated avalanches.

also succeeded in seeing the dynamic movement of the particles and the force chain structure by DEM simulation. The results of the simulation well coincide with those of the physical experiment using a photo-elastic material. We now consider the diﬀerence in the behavior of avalanches between a large pile and a small pile, based on the results of the experiment and the simulation. For a sandpile in a critical state, what determines the size of avalanches may be some kind of inhomogeneity. When the pile is small, the size of avalanches is small because the uneven distribution of particles on the surface may be the dominant inhomogeneity of the system (Fig. 6(a)). When the pile becomes large, however, the deep inside of the pile is tightened and consolidated while the part near the surface is never tightened because of the repeated avalanches (Fig. 6(b)). This is conﬁrmed by the photo-elastic experiment (Fig. 4(b)) and the DEM simulation (Fig. 5(b)). The contrast in strength becomes the main factor for the inhomogeneity.

Acknowledgments The authors thank Dr Kenji Satake, an editor of this volume, for his kind suggestions and two anonymous reviewers for helpful comments. This work

FA

December 10, 2008 15:45

290

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch19

N. Yoshioka and H. Sakaguchi

was supported by the Grant-in-Aid for Scientiﬁc Research (B) #18340134 of Japan Society for the Promotion of Science, Japan.

References 1. P. Bak, C. Tang and K. Wiesenfeld, Phys. Rev. Lett. 59 (1987) 381. 2. D. P. Schwartz and K. J. Coppersmith, J. Geophys. Res. 89 (1984) 5681. 3. L. P. Kadanoﬀ, S. R. Nagel, L. Wu and S.-M. Zhou, Phys. Rev. A 39 (1989) 6524. 4. J. M. Carlson and J. S. Langer, Phys. Rev. Lett. 62 (1989) 2632. 5. J. M. Carlson and J. S. Langer, Phys. Rev. A 40 (1989) 6470. 6. H. Nakanishi, Phys. Rev. A 41 (1990) 7086. 7. H. Nakanishi, Phys. Rev. A 43 (1991) 6613. 8. J. Rosendahl, M. Vekic and J. Kelly, Phys. Rev. E 47 (1993) 1401. 9. J. Rosendahl, M. Vekic and J. E. Rutledge, Phys. Rev. Lett. 73 (1994) 537. 10. N. Yoshioka, Earth Planets Space 55 (2003) 283. 11. C.-H. Liu, H. M. Jaeger and S. R. Nagel, Phys. Rev. A 43 (1991) 7091. 12. S. R. Nagel, Rev. Mod. Phys. 64 (1992) 321. 13. H. M. Jaeger, C.-H. Liu and S. R. Nagel, Phys. Rev. Lett. 62 (1989) 40. 14. A. Drescher and G. de Josselin de Jong, J. Mech. Phys. Solids 20 (1972) 337. 15. L. Vanel, D. Howell, D. Clark, R. P. Behringer and E. Clement, Phys. Rev. E 60 (1999) R5040. 16. J. Geng, E. Longhi and R. P. Behringer, Phys. Rev. E 64 (2001) 060301. 17. T. Jotaki and R. Moriyama, J. Soc. Powder Technol. Jpn. 60 (1979) 184. 18. P. A. Cundall and O. D. L. Strack, G´eotechnique 29 (1979) 47. 19. N. Yoshioka and H. Sakaguchi, Advances in Geosciences, Solid Earth, Vol. 1 (World Scientiﬁc, 2006), p. 105.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch20

Advances in Geosciences Vol. 13: Solid Earth (2007) Ed. Kenji Satake c World Scientiﬁc Publishing Company

STATISTICAL PROPERTIES AND TIME TREND IN THE NUMBER OF HOLOCENE VOLCANIC ERUPTIONS A. N. ZEMTSOV Institute for the History of Science and Technology of the RAS Department of Earth Sciences, Staropansky Lane 1/5, Moscow, 109012, Russian Federation A. A. TRON Saint-Petersburg State University, Department of Geography 10-th Line, 33, Saint-Petersburg, 199178, Russian Federation [email protected]

The statistical estimations for the Smithsonian Institution Holocene volcanic eruptions catalog have been fulﬁlled. It is found that the general trend in the number of explosive eruptions follows the double-exponent model. The slow exponent (T ≈ 4000a ) reﬂects the cancellation of ancient data and the shortperiod exponent (T ≈ 200a ) reﬂects the rapid information grow-up in the recent centuries. The law of eruptions recurrence is established which follows the ≈ 1/2 exponent index, close to that one for earthquakes (the Gutenberg– Richter law). The global explosive productivity of major eruptions is estimated as 1–3 km3 /y of pyroclastic rocks.

1. General Trend in the Number of Eruptions Smithsonian Institution Holocene volcanic eruptions catalog1 allows for the general statistical analysis of their time trends. The term “Holocene” means here the total information about traces of eruptions preserved since the last glacial period. Catalog1 covers the period of 11,855 years and contains data for about 594 eruption events. We analyzed time density of events by building the histogram of number of events using data partitioning in 50, 100, 150, and 200 cells. Interval length was chosen as an optimal compromise between strong ﬂuctuations on minor lengths and representation and resolution. These data with the best-ﬁt estimation and 99% conﬁdence interval is shown in Fig. 1. Best-ﬁt model appeared to be as N (t) = A1 exp(−t/τ1 ) + A2 exp(−t/τ2 ), 291

(1)

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch20

A. N. Zemtsov and A. A. Tron

292

where N (t) is the number of events reduced to the center of interval t in years, and the parameters are presented in Table 1. We interpret the long period exponent mainly as a decay of information due to erosion of lava ﬂows and tephra ﬁelds burying, overlaying of geological structures. There are three long exponents in a given series. Short-range exponent is believed to be connected with the rapid growth of information about eruptions and increase in global covering. The fraction of unique eruptions, determined from the volcano names appearance within 4, 8, and 16 subintervals of the whole period, falls with time which indicates that relatively few new volcanic centers have been discovered during the last 200 years along with the growing resolution of events detection at the already known ones which corresponds to the results in Ref. 2. Figure 1 illustrates the data and model for the case of 100 cells. In Ref. 1, recent eruptions were studied (about 150 years), and quasilinear increasing trend in the annual number of eruptions was found. Its angular coeﬃcient ∼0.1 agrees with the derivative’s value for N (t) in expression (1) for the year 1900. Table 1.

Parameters of the best-ﬁt presentation of Holocene eruptions data.

No

Number of cells

Cell length, years

A1

τ1 years

A2

τ2 years

1 2 3 4

50 100 150 200

237.10 118.55 79.04 59.28

0.017881439 0.004377192 0.007114635 0.006164376

−217.436777 −202.014409 −223.572903 −228.086308

15.96926084 8.233002443 5.336156760 3.973717482

−4315.62897 −4104.07844 −4290.88940 −4303.88434

Fig. 1. Two-exponent best-ﬁt model of the Holocene eruptions data. Time in years, minus corresponds to dates BC.

FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch20

Holocene Volcanic Eruptions

293

2. Rank Statistics for the Number of Eruptions According to these data, cumulative number of eruptions as a function of erupted volume V (in m3 ) can be, as a ﬁrst approximation, expressed as 3.0 × 107 , V 0.57

(2)

dN 1.7 × 107 =− . dV V 1.57

(3)

N (V ) =

The volume of average eruption (median one) (5,018/2 = 2,509) is equal to 1.4 × 107 m3 which is 0.014 km3 , corresponding to a rather small eruption.

3. Explosive Volcanism Productivity Using the data of the “Global Volcanism Program,”1 it is possible to estimate contributions of eruptions of diﬀerent powers to the total erupted volume over the whole period of observations: The main contribution to the total erupted volume is produced by the most powerful eruptions with solid phase erupted volumes of tens of cubic kilometers. Such eruptions noticeably have inﬂuence on the Earth’s stratosphere. In our case, the total erupted volume of eruptions under study is approximately equal to 5,000 km3 of solid products. The total number of eruptions included in Table 2 is 5,018. There is no reliable estimation of the period of observations; so, if we take this period to be about 2,000 years of historical time, and taking into account that the data about later eruptions are incomplete, we arrive to estimate “explosive volcanic productivity” as 1–3 km3 /y. Very close value was obtained by authors earlier, using much shorter time eruptions records.2 Using these statistical data of erupted volumes, it is possible to obtain approximate analytical estimation of the relative contribution of eruptions with diﬀerent solid phase productivity — from parts to hundreds of cubic kilometers. Using the models (2) and (3), one can get the annual total erupted volume over the period of observations ∆T : Verupted

1.7 × 107 = ∆T

V max

V min

V

0.43 1.7 × 107 × Vmax dN dV ≈ . dV ∆T

(4)

The value of total erupted volume is determined mainly by strongest eruptions and time intervals between them.

FA

≥5 · 107

≥5 · 108

≥5 · 109

≥5 · 1010

≥5 · 1011

5,018

1,387

463

156

50

4

Table 3.

5 · 106 3,631 0.018 × 1012

5 · 107 924 0.046 × 1012

5 · 108 307 0.154 × 1012

5 · 109 106 0.53 × 1012

5 · 1010 46 2.30 × 1012

5 · 1011 4 2.00 × 1012

9in x 6in

Erupted tephra volume per event, m3 Number of events Total erupted volume, m3

Erupted volume shares for diﬀerent VEI classes.

AOGS 2007 - SE Volume

≥5 · 106

A. N. Zemtsov and A. A. Tron

Erupted tephra volume, m3 Number of events

Cumulative numbers of the eruptions of diﬀerent VEI classes.

December 10, 2008 15:45

294

Table 2.

b672-V13-ch20 FA

December 10, 2008 15:45

AOGS 2007 - SE Volume

9in x 6in

Holocene Volcanic Eruptions

b672-V13-ch20

295

The well-studied period with several most powerful eruptions is 19th century. In this century, three strong eruptions occurred: Tambora in 1812–1815, Cosiguina in 1835, and Krakatau in 1883. Taking their joint solid phase erupted volume to be 200 × 109 m3 and averaged time interval between explosions — (20 + 48)/2 = 48 years — as an estimation of repeating period of most strong eruptions, we can get a rough estimation of their productivity, assuming that three eruptions correspond to three periods, equaling 34 years each: 2.0 × 1011 /3 × 34 ≈ 2.0 × 109 m3 /y. From expressions (2) and (4) we get (1 − 3) × 109 m3 /y, which gives probably a higher value. Such estimation, additionally, allows us to make an assumption that strongest (powerful) eruption should be expected in the near decades.

4. The Law of Eruptions Recurrence Power law for the cumulative number of eruptions (2) is similar to the wellknown Gutenberg–Richter law in seismology which describes frequencies of earthquakes. Power index in Gutenberg–Richter law is estimated to be ≈0.5,3,4 close to the value in (2) equal to 0.57. This relation can be interpreted as frequency distribution of eruptions over their “strength” — solid phase erupted volumes. Data in an earlier study2 can also be interpreted in terms of “eruptions frequencies law.” In the early 1980s, the number of known eruptions with erupted volumes over 109 m3 was much less than that in the recent period (30 compared to ≈300 in the catalog under study). Statistics in Ref. 2 allows to consider the model N (V ) ∼ V −k , where 2/3 ≤ k < 1. Under k < 1, the total erupted volume and, thus, annual explosive volcanic productivity is determined by strong eruptions.

Conclusion Annual input of solid heated material to the Earth’s surface (∼170 × 106 km2 without ocean ﬂoor) is equivalent to the heat input of 3.0 × 1018 J which corresponds to the heat ﬂux of the order of 3.0 × 1018 J 2 ≈ 0.00056 Wt/m . 31.6 × 106 s × 170 × 1012 m2

(5)

Here the erupted material (tephra) density is taken to be ≈1 g/cm3 and speciﬁc heat about 1.0 J/(gram · K).

FA

December 10, 2008 15:45

296

AOGS 2007 - SE Volume

9in x 6in

b672-V13-ch20

A. N. Zemtsov and A. A. Tron

This value is about 0.01 of the observed heat ﬂux through the Earth’s crust. However, the annual energy output of volcanoes is compared at the order of magnitude with annual energy of earthquakes. To analyze the possible mechanism of magmatic energy transfer from the interior to the Earth’s surface, one could suppose that large volume of pyroclastic rocks is ejected on the land surface and into the atmosphere during the major volcanic eruptions, while suﬃcient amounts of thermal energy ﬁnds its way to the surface layers of the Earth’s crust through the global system of mid-ocean ridges. It could be expected that mid-ocean ridges volcanic thermal energy ﬂux exceeds the amount of pyroclastic heat of explosive eruptions.

References 1. T. Simkin and L. Siebert, Volcanoes of the World (Geoscience Press, Tucson, Arizona, USA, 1994), 350 p., electronic version at www.volcano.si.edu/world/ eruptioncriteria.cfm#VEI. 2. A. N. Zemtsov and A. A. Tron, A statistical analysis of catalogs of volcanic eruptions, Earth Sci. Sec. 285(3) (1987) 16–19 (Scripta Technika, Inc., 1987) (Translation of the Russian version, publ. in 1985). 3. M. A. Sadovsky, Selected Works (Moscow, Nauka Publ. House, 2004) (in Russian). 4. S. A. Fedotov, Long-Term Earthquake Prediction for the Kuril-Kamtchatka Arc (Moscow, Nauka Publ. House, 2005) (in Russian).

FA