Nat Hazards (2011) 57:726 DOI 10.

1007/s11069-010-9637-z ORIGINAL PAPER

Aftershock sequences of two great Sumatran earthquakes of 2004 and 2005 and simulation of the minor tsunami generated on September 12, 2007 in the Indian Ocean and its effect
R. K. Jaiswal A. P. Singh B. K. Rastogi T. S. Murty

Received: 5 December 2009 / Accepted: 28 September 2010 / Published online: 2 November 2010 Springer Science+Business Media B.V. 2010

Abstract We present the seismic energy, strain energy, frequencymagnitude relation (b-value) and decay rate of aftershocks (p-value) for the aftershock sequences of the AndamanSumatra earthquakes of December 26, 2004 (Mw 9.3) and March 28, 2005 (Mw 8.7). The energy released in aftershocks of 2004 and 2005 earthquake was 0.135 and 0.365% of the energy of the respective mainshocks, while the strain release in aftershocks was 39 and 71% for the two earthquakes, respectively. The b-value and p-value indicate normal value of about 1. All these parameters are in normal range and indicate normal stress patterns and mechanical properties of the medium. Only the strain release in aftershocks was considerable. The fourth largest earthquake in this region since 2004 occurred in September 2007 off the southern coast of Island of Sumatra, generating a relatively minor tsunami as indicated by sea level gauges. The maximum wave amplitude as registered by the Padang, tide gauge, north of the earthquake epicenter was about 60 cm. TUNAMI-N2 model was used to investigate ability of the model to capture the minor tsunami and its effect on the eastern Indian Coast. A close comparison of the observed and simulated tsunami generation, propagation and wave height at tide gauge locations showed that the model was able to capture the minor tsunami phases. The directivity map shows that the maximum tsunami energy was in the southwest direction from the strike of the fault. Since the path of the tsunami for Indian coastlines is oblique, there were no impacts along the Indian coastlines except near the coast of epicentral region. Keywords Aftershock Seismic characteristics Energy release Tsunami simulation Tsunami propagation Wave height Tide gauge BPR
R. K. Jaiswal National Geophysical Research Institute (NGRI), Hyderabad, Andhra Pradesh 500007, India A. P. Singh (&) B. K. Rastogi Institute of Seismological Research, Raisan, Gandhinagar, Gujarat 382009, India e-mail: apsingh07@gmail.com T. S. Murty Department of Civil Engineering, University of Ottawa, 140 Louis Pasteur Ottawa, Ottawa, ON K1N 6N5, Canada

123

Nat Hazards (2011) 57:726

1 Introduction The two largest earthquakes of the past 40 years ruptured a 1,600-km-long portion of the fault boundary between the Indo-Australian and southeastern Eurasian plates on December 26, 2004 (Mw 9.3) and March 28, 2005 (Mw 8.6). The rst event generated a tsunami that caused more than 2,83,000 deaths, fault slip of up to 15 m occurred near Banda Aceh, Sumatra but to the north, along the Nicobar and Andaman islands, rapid slip was much smaller. The March 28, 2005 event ruptured an adjacent 300-km-long portion of the plate boundary (Lay et al. 2005; Subarya et al. 2006). Though this great earthquake did not generate signicant tsunami, it caused damage to property and loss of lives in Sumatra islands of Indonesia. These two events are the largest to occur after the global deployment of digital broadband, high-dynamic range seismometers (Romanowicz and Giardini 2001; Butler et al. 2004; Natawidjaja and McCaffrey 2006), which recorded both the huge ground motions from ensuing free oscillations of the planet (Park et al. 2005; Stein and Okal 2004). A large number of aftershocks, including some strong ones, that occurred after the earthquakes enticed lot of interest. Study of these aftershocks is expected to throw light on tectonics of the region. The seismic energy, strain energy, frequencymagnitude relation (b-value) and decay rate of aftershock (p-value) for the complete AndamanSumatra sequences provide a much better understanding of the physical processes responsible for the continued aftershock activity. The 2004 tsunami event was followed by annual occurrence of ocean wide tsunami (i.e. those that inuence areas far from the generation region) in 2005, 2006, 2007 and 2009 [McCloskey et al. 2005; Ammon et al. 2006; Fujii and Satake 2006, 2008 United States Geological Survey (USGS)], and these tsunami have been recorded on sea level gauges in Sri Lanka except the 2006 tsunami of Java (Table 1). The fourth largest earthquake in this region occurred on September 12, 2007, with a magnitude Mw 8.4 and struck the western coast of Bengkulu, Sumatra, Indonesia. The epicenter of the earthquake was located 150 km SW of Bengkulu (4.520S, 101.374E), off west coast of southern Sumatra, and it generated a relatively moderate tsunami. The focal depth of this earthquake was 34 km, and origin time was 11:10:26 UTC (16:40:26 IST), (USGS). The seismic sequence continued for the next 2 days, with the biggest event of magnitude Mw 7.1 happening on September 13. This sequence took place in the same rupture zone where the historical earthquakes of 1797 and 1833 occurred which generated signicant tsunami (Nalbant et al. 2005). Sea level data indicated that a minor tsunami was generated on September 12, 2007. The maximum wave height registered by the Padang, tide gauge, north of the quakes epicenter was about 60 cm. The Padang tide gauge located at 0.9S, 100.4E. A maximum tsunami runup height of 90 cm has been reported. The travel time of the rst tsunami wave is estimated at about 20 min. The period of subsequent waves was about
Table 1 Details of Indian Ocean tsunami 20042007 Year Date Magnitude of earthquake 9.2 8.6 7.7 8.4 Epicenter location 3.32N, 95.854E 2.07N, 97.01E 9.22N, 107.32E 4.52N, 101.37E Maximum tsunami height in Sri Lanka 10.1 m Hambantota 2.7 m Kirinda 0.0 m 0.6 m Trincomallee

2004 2005 2006 2007

26th December 28th March 17th July 12th September

Source: NARA, Sri Lanka

123

Nat Hazards (2011) 57:726

28 min. Pacic Tsunami Warning Center (PTWC, NOAA) and Japan Meteorological Agency (JMA) issued ocean wide tsunami warning and watch, respectively, for the coastlines of Indian Ocean on September 12, 2007 (PTWC and JMA). Indian National Centre for Ocean Information Services (INCOIS), Hyderabad, issued alert for Andaman Nicobar region and watch for Indian mainland. Finally, INCOIS downgraded alert for AndamanNicobar region and watch for Indian mainland into all clear (no tsunami threat) after observing near-real-time minor sea level variations in Bottom Pressure Recorders (BPRs) and tide gauges deployed in the Indian Ocean by National Institute of Ocean Technology (NIOT), Survey of India (SOI) and Global Sea Level Observing System (GLOSS) (Nayak and Kumar 2008). This tsunami was recorded on about 22 sea level measurement networks (viz. tide gauges, Indian BPRs, NOAA Buoy 23401 for Thailand etc.) deployed by various authorities and Institutions in the Indian Ocean. This tsunami event helps tsunami modelers to quantify the size of tsunami and to validate the tsunami propagation and inundation models for the Indian Ocean as well as to nd the gaps in the Indian Ocean tsunami Warning Systems (IOTWS) and to ll them. In view of this, an attempt is made here to simulate the minor tsunami generated on September 12, 2007, using TUNAMI-N2 numerical code.

2 Data Aftershocks data for this study have been collected from USGS catalogs for the period up to May 20, 2006, covering the region between Lat. 0S to 15N and Long. 90 to 99E. Rupture of the 2004 earthquake propagated northward for 1,300 km up to Andaman (from Lat. 3N to 15N), while rupture of the 2005 earthquake propagated southward for 300 km (from 3N down to 0S). Aftershocks are considered to have occurred in the respective rupture zones. It is noticed that there were a few aftershocks of 2004 earthquake between latitude 2.5 and 3N. Hence, shocks between latitude 2.5 to 3N until March 28, 2005, are considered as aftershocks sequences of December 26, 2004, event, while shocks at and below latitude 3N after March 28, 2005 are considered as aftershocks of March 28, 2005 mainshock (Fig. 1). A similar pattern is shown by Ammon (2006). A total of 3460 aftershocks of December 26, 2004 mainshock are compiled in the magnitude range 3.57.5. On the other hand, 1835 aftershocks of March 28, 2005 earthquake are compiled in the magnitude range 3.66.9 (Fig. 2). For the December 26, 2004 earthquake, the two largest aftershocks were of magnitude, Ms 7.5. The rst one occurred a few hours after the mainshock near Nicobar Islands (Lat. 6.91E, Long. 92.96N). However, the second one occurred 150 km further NW on July 24, 2005 (Lat. 7.92E, Long. 92.19N). Analysis of the aftershocks activity reveals that the earthquakes are mostly conned to the top 50 km (Fig. 2). However, events with 33 km focal depth are the modelassumed depth, which are not properly resolved, as these are determined using stations located at regional and teleseismic distances. Since the catalog depths are found to have errors, the depth depends on the b-value was not attempted. Histogram of magnitudes of the original catalog compiled by USGS shows a maximum numbers of earthquakes occur in magnitude range of 4.04.9 (Fig. 3a, b) in both cases, which is a rst indication of value of magnitude completeness (Mc) of catalog (Fig. 4). We calculated the magnitude of completeness and eliminated all the events having magnitudes less than Mc (Fig. 5). The calculated Mc was found 4.2 in both cases using the maximum curvature method for the catalog. The estimated Mc is based on the assumption of a power-law GutenbergRichter relationship and taken as the magnitude where the rst

123

10

Nat Hazards (2011) 57:726

Fig. 1 Aftershocks of December 26, 2004, earthquake are shown by white patch of epicenters and a few red ones from Lat. 3N and northward, the earthquake of March 28, 2005, are shown by red epicenters about 3N and southward (source EMSC). Location of the earthquake of March 28, 2005, was about 190 km SE of the location of mainshock of December 26, 2004

derivative of the frequencymagnitude curve has its maximum. The magnitude from where curves became exponential is known as magnitude of completeness. Mc can be also dened as the lowest magnitude at which 100% of the events are detected in space and time volume (Woessner and Wiemer 2005). For the tsunami simulation of 2007 case GEBCO (General Bathymetric Chart of the Oceans), bathymetry data of 1 min arc (1.8 km) resolution with 5-km grid spacing has been used. The fault parameters are taken from USGS and Fujii and Satake (2007).

3 Tectonic setting The 2004 and 2005 earthquakes ruptured the boundary between the Indo-Australian plate, which moves generally northward at 4050 mm/year, and the southeastern portion of the Eurasian plate, which is segmented into the Burma and Sunda subplates (Lay et al. 2005). The eastern margin of the Burma microplate is dened by the NS trending and Sagaing transform fault, which separates the central low lands from the eastern high lands of Myanmar, and the fault continues as the Andaman sea rift system (known as the Andaman Spreading Ridge (ASR)). The Andaman Sea basin is considered to be a complex back arc spreading center, is categorized as a pull apart or rip off basin rather than a typical back-arc-extensional basin. Seismic reection studies across the trench slope indicate folding and thrusting in the accretionary prism, where a major component of convergence occurs normal to the trench axis. Among the series of thrusts/faults in the Andaman subduction zone, the west Andaman fault is most prominent. The thrust/fault appears to be

123

Nat Hazards (2011) 57:726

11

Fig. 2 Epicentral distribution of earthquakes used in this study is shown by solid circles; in cross sectional views with depth (left side shows distribution of events along latitude with depth, while lower shows the distribution along longitude with depth). Depth distribution of the used events (in crosses) in the latitude and longitude directions, respectively. Mainshock for 2004 and 2005 earthquakes are shown by red stars

continuous from the west of Sumatra in the south to Irrawady basin in the north, where it is buried beneath the sediment cover. The region between the AndamanNicobar Islands and the volcanic arc is the foredeep sedimentary trough (Dickinson and Seely 1979). Along the Sumatra subduction zone, plate convergence is partitioned into dip-slip and right lateral strike-slip components, the former being accommodated by slip on the subduction interface and the latter by the Sumatran fault. The present-day tectonic processes are controlled by three major fault systems, the most prominent being the subduction thrust, which outcrops in the Sunda trench. Inland, the trench-parallel Sumatra fault that runs through the entire length of the island from Banda Aceh to Sunda Strait, accommodates oblique convergence through strike-slip faulting. The Mentawai fault at the outer margin of the forearc basin is another important fault system in the Sumatra region (Sieh and Natawidjaja 2001). Subduction along AndamanSumatra trench system has given rise to a discontinuous belt of the submarine ridges and volcanic seamounts. The andesite volcanoes of the Barren and Narcondam Island are prominent among them; the Narcondam is now extinct, but the Barren is still marked by an active volcano; it erupted last in March, 1991 after lying dormant for about two centuries. Further south, this volcanic belt is represented by the Barisan range in Sumatra, and in the north, the trend is correlated with the central molasse basin of Burma.

123

12

Nat Hazards (2011) 57:726

(a) 1600
No. of Aftershock
1400 1200 1000 800 600 400 200 0
84 1265

1451

459

133 27 1 1 1 2 1

4.0 4.4

4.5 4.9

5.0 5.4

5.5 5.9

3.5 3.9

6.0 6.4

6.8

Magnitude Range

(b)
No. of Aftershock

900 800 700 600 500 400 300 200 100 0

45 171 32 10 1 2 2 1 767 742

4.9

3.9

5.4

4.4

5.9

6.4

6.6

6.7

6.9

7.5

4.5 -

3.6 -

5.0 -

4.0 -

5.5 -

Magnitude Range
Fig. 3 Magnitude distribution of the aftershocks of a December 26, 2004, and b March 28, 2005 earthquake

The September 12, 2007 earthquake was followed by a bigger aftershock of Mw 7.9 Bengkulu (2.50S, 100.906E, USGS), some 12 h after the onset of the mainshock, and this also generated a small tsunami that was recorded at Pointe La Rue and Rodrigues (Lorito et al. 2008). Both earthquakes caused a total of 25 fatalities and 161 injuries (USGS). The seismic sequence continued during the next 2 days, including a magnitude MW 7.1 earthquake on September 13, some 15 h after the mainshock. This sequence occurred in the same zone where the historical giant earthquakes of 1797 and 1833 generated major tsunami (Nalbant et al. 2005).

4 The b- and p-values of aftershock sequences The frequencymagnitude distribution mainly denes the relative amount of smaller to larger earthquakes through its slope-parameter b. This has been extensively studied in laboratory experiments, as well as for synthetic and real seismicity. There are three main natural factors, as inferred from laboratory studies that can cause signicant changes of the frequencymagnitude distribution from an average value of 1.0 (1) increased material

123

6.0 -

6.5

6.8

8.7

9.3

Nat Hazards (2011) 57:726

13

Fig. 4 Computational domain with bathymetry and topography data, locations of sea level measurement networks as well as focal mechanisms of recent great Sumatran earthquakes of 2004, 2005 and 2007. Color bar shows the data are in meters. Triangle shows locations of tide gauges and circle denotes the BPRs locations. Red stars shows recent (2004, 2005 and 2007) Sumatran earthquakes locations

heterogeneity increases the b-value; (2) increase in shear stress or effective stress decreases the b-value; and (3) increase in the thermal gradient may cause an increase in the b-value. High b-values are generally observed for highly cracked volumes like magma chamber, reservoir-induced or intraplate sites. However, small b-value is observed for asperities or probable zones for future large earthquakes like plate boundaries. Another important parameter to characterize an aftershock sequence is the p-value, which is a measure of the decay rate of aftershocks, as described by the Omoris (1894) law. Worldwide review of the p-value estimates suggests a range from 0.6 to 2.5, with a median around 1.1. Regional variation of the p-value has been related to the variation in heat ow (Mogi 1962). Variability of estimated p-values may be related to the structural heterogeneities, stress and temperature in the crust. Slope of b-value and p-value of aftershock sequences reveals the information about the heterogeneity of the rocks and stress conditions in the earthquake zones. The magnitudefrequency relation (b-value) is given by (Gutenberg and Richter 1944) Log N a b M; 1

where N is cumulative number of shocks, constant a depends upon level of seismicity and slope b reects upon the heterogeneity of the medium and level of stress. The b-values have been determined with least-square and maximum-likelihood methods. Using leastsquare method, for the sequence pertaining to the December 26, 2004 and March 28, 2005 earthquake obtained Log N = 8.871.21M (Fig. 5a) and Log N = 7.741.08M (Fig. 5b), respectively. The b-values are obtained 1.21 0.04 and 1.08 0.03 for the two mainshcoks, respectively, using least-square approach. Further, the b-value has been determined by maximum-likelihood method (Utsu 1961; Aki 1965) using the relation, which is given below

123

14

Nat Hazards (2011) 57:726

(a)
3

Log N = 8.87 - 1.22 M

Log N

(b)
3

Log N

Log N = 7.74 - 1.08 M

2

M
Fig. 5 The b-value of a December 26, 2004, and b March 28, 2005, earthquakes

123

Nat Hazards (2011) 57:726

15

b 0:4343=Ma Ms ;

where Ma is the average magnitude and Ms is the threshold value (lower limit) of the magnitude for which the dataset is complete. The lowest magnitude is 4.2 for both the sequences. From the above relation (2), the b-values show 0.92 and 1.07 for both the mainshocks, respectively. The values nearly 1 indicate normal values. The rate of decay of the aftershock sequence is empirically described with time and is given by the Omoris (1894) law. N t k=t cp ; 3

where N(t) is the frequency of aftershocks per unit time and k, c, p are constants. N(t) was calculated following the Ogata et al. (1993) approach for tting the cumulative number of aftershocks as a function of time after the mainshock. The k value is dependent on the total number of aftershocks during the rst time interval and c on the rate of activity in the earliest part of sequence. The decay in aftershock activity with time reects a decrease in stress in the region. The characteristics parameter p is determined from this relation and gives the rate of fall of aftershock activity. The frequency of occurrence of aftershocks n (t) at time t (weekly) after the mainshock is given as; N t 725t1:1 N t 281t0:9 R2 0:75 R2 0:68 December 26; 2004 sequence March 28; 2005 sequence 4 5

We estimated p-value 1.1 and 0.9 for Sumatran aftershocks (Fig. 6a, b). It shows a normal decay and decrease in stress with time in the region.

5 Energy and strain release in mainshocks and aftershocks Energy released during an earthquake is proportional to its magnitude (M). The energy release, E in ergs is estimated from the relation: Log E 9 1:8M 6

The values of E have been calculated for the mainshocks and their aftershock sequences. Cumulative energy release for the two sequences is shown in Fig. 7a, b. The energy released in the earthquake of December 26, 2004, was 5.49E ? 25 ergs, while the total aftershock energy was only 7.42E ? 22 ergs, which reveals only 0.135% energy of the mainshock was released in the aftershocks of earthquake of December 26, 2004. On the other hand, energy released in the earthquake of March 28, 2005, was 4.57E ? 24 ergs, while the total aftershock energy was only 1.67E ? 22 ergs. This indicates only 0.365% energy of the mainshock was released in the aftershocks of earthquake of March 28, 2005. The square root of energy for an individual earthquake is proportional to the strain release (Gutenberg and Richter 1956) and is given by Log E1=2 4:5 0:9M
1/2

Since energy unit is in ergs, strain release will be in the units of (ergs) . Cumulative strain release for the two sequences is shown in Fig. 8a, b. The major strain release in the

123

16

Nat Hazards (2011) 57:726

(a) 800
740 723

600

No. of Aftershock

400

n(t) = 725 t-1.1 R2 = 0.75

200

177 177

136

137

96 7676 64 63 49 47 3938 3838 252922302120 232027 21 2022 1816 17 17 1716 1816 15 13121210 10 9 11 7 14 5 6 1312141011 7 13 7 11 5 10 7 11 6 11 9 6 4 3 3 1 6 4 3

0 0

10

20

30

40

50

60

70

Time in week

(b)

600

514

No. of Aftershocks

400

n(t) 281 t -0.9 R2 = 0.68

200

185

137

85 7471 63 4748 4039 26 36

25292227

30 16 151317141719 1012 9 14 9 11 9 1011 17141313 6 6 1011 14 9 10 151314 7 14 9 6 1011 6 8 12 9 10 1 0 4 9 1816 6 11 8 6 4 2 3

Time in week
Fig. 6 Decay rate of aftershocks a December 26, 2004 and b March 28, 2005 earthquakes

aftershock sequence of December 26, 2004, earthquake is by two major earthquakes of Ms 7.5 on December 26, 2004, and July 24, 2005. The December 26, 2004, earthquake released strain of 7.41 9 1012 ergs1/2, while its aftershocks released 2.98 9 1012 ergs1/2, making 39% in the aftershocks. This gure is

123

Nat Hazards (2011) 57:726

17

(a) 5.504

5.502
Ms 7.5 (24July 2005)

E (10)25 (ergs)

5.5

5.498
Mw 9.3 & Ms 7.5 (26 Dec. 2004)

5.496

5.494 1 10 100

Day

(b) 0.3035

0.303

E (10)25 (ergs)

0.3025

0.302

0.3015

0.301 1 10 100

Day
Fig. 7 Cumulative energy release of a December 26, 2004, and b March 28, 2005, earthquakes

matching with the 30% slip in aftershocks for 1.5 months derived by Global Positioning System (GPS) and seismic methods by Subarya et al. (2006). The March 28, 2005, earthquake released strain of 1.74 9 1012 ergs1/2, while its aftershocks released 1.24 9 1012 ergs1/2, making it 71% strain release in aftershocks.

123

18

Nat Hazards (2011) 57:726

(a)

11

10

E1/2 (10)12 (ergs)1/2

7 1 10 100

Day

(b)

3.2

2.8

E1/2 (10)12 (ergs)1/2

2.4

1.6 1 10 100

Day
Fig. 8 Cumulative strain release of a December 26, 2004, and b March 28, 2005, earthquakes

123

Nat Hazards (2011) 57:726

19

6 Tsunami generation and propagation An attempt has been made here to simulate the September 12, 2007, tsunami, its generation and propagation in the Indian Ocean using TUNAMI-N2 numerical model. The TUNAMIN2 model basically takes the seismic deformation as input to predict the run-up heights and inundation levels at coastal regions for a given tsunamigenic earthquake (Imamura 2006; Imamura et al. 2006). The seismic deformation for the earthquake has been computed followed by Mansinha and Smylie (1971) formulation using the earthquake parameters namely location, focal depth, strike, dip and rake angles, length, width and slip of the fault plane. The model domain with bathymetry and topography data as well as locations of the tsunami recorded by sea level measurement networks are shown in Fig. 4. The great Sumatran earthquakes of 2004 (Mw 9.3), 2005 (Mw 8.7) and 2007 (Mw 8.4 and 7.9) show roughly similar focal mechanisms i.e. shallow dipping thrust faults (Fig. 4). The CMT focal mechanisms of these earthquakes using from Harvard program. In Fig. 4, triangles show locations of INCOIS, GLOSS and NIO (Goa) tide gauges and circles denote the INCOIS and NOAA BPRs (bottom pressure recorder) locations. INCOIS gets the tide gauge data from NIOT (National Institute of Ocean Technology), Chennai and SOI (Survey of India), Dehradun. Tsunami wave heights recorded by these tide gauges and bottom pressure recorders are given in Table 2. In order to compute co-seismic dislocation model (input for tsunami generation), the following fault parameters were used: location = (4.520S, 101.374E), focal depth = 34 km, strike angle = 327, dip angle = 12, rake angle = 85 (USGS), length of the fault = 160 km, width of the fault = 80 km and slip of the fault plane = 5 m (Fujii and Satake 2007). The dislocation model shows that the initial wave height of the tsunami was 1.5 m at the source of the earthquake and with a minimum was 0.5 m (Fig. 9). The maximum wave heights of tsunami were 2.27 m at Padang, 0.15 m at Christmas Island, 0.24 m at Cocos Island recorded by the Global Sea Level Observing System (GLOSS), 0.046 m on tsunami meter 23401 (8.9N, 88.54E) recorded by National Data Buoy Center (NOAA), 0.60 m at Trincomallee and 0.60 m at Colombo recorded by National Aquatic Resources Research and Development Agency (NARA), Sri Lanka. AndamanNicobar region was in the shadow zone of the earthquake rupture zone when compared to the Indian Mainland, the observed tsunami amplitude at Chennai and Port Blair was 20 and 10 cm, respectively. Our simulation results are in close agreement with the available observations. The tsunami propagation states at every one-minute interval are simulated. In this study, the simulation is carried out for duration of 600 min and to see the impact of the tsunami wave propagation as well as wave height at the eastern costal part of India. The maximum tsunami wave amplitude after the 600 min of tsunami is shown in Fig. 10. From the Fig. 10, it is found that the maximum tsunami energy was directed toward southwest direction from the strike of the northwest- and southeast-oriented fault plane. It is also found that the path of the tsunami for Indian coastlines is oblique. Therefore, there were no affect along the Indian coastlines except near the coast of epicentral region. The tsunami propagation snapshot at time (t) 0, 11, 31, 61, 93, 121, 151, 181, 211, 241, 283, 301, 361, 421, and 582 min is shown in Fig. 11ac. At different time intervals, the wave amplitude is shown in the bar with different colors, next to the simulation gure. The tsunami wave propagation could affect by the bathymetry data of the Indian Ocean (Jaiswal et al. 2009). Therefore, detailed bathymetry data are required for the accurate information about the tsunami propagation in the Ocean. Initially, tsunami propagated with amplitude of 1 m at

123

20

Nat Hazards (2011) 57:726

Table 2 Tsunami wave height recorded at different tide gauges and BPRs due to September 12, 2007, tsunamigenic earthquake
Location Padang, Sumatra (0.95S, 100.36E) Sibolga, Sumatra (01.75N, 98.77E) Cocos Island, Australia (12.11S, 96.89E) Sabang, Sumatra (5.83N, 95.33E) Tsunamimeter 23401 (8.9N, 88.54E) TB 10 (7.000N, 87.017E) TB 10 A (7.000N, 87.068E) TB 07 (9.001N, 85.517E) TB 03 (6.101N, 89.834E) Port Blair, Andaman, India (11.64N, 92.71E) Chennai, India (13.10N, 80.30E) Colombo, Sri Lanka (06.92N, 79.83E) Christmas I., Australia (10.29S, 105.67E) Cocos I., Australia (12.0S, 96.7E) Cillcap, Java (7.75S, 109E) Verem, Goa (15.50N, 73.81E) Kawaratti, Lakshadweep (10.57N, 72.63E) Gan, Maldives (00.7N, 73.2E) Male, Maldives (04.2N, 73.5E) Hanimaadhoo, Maldives (06.8N, 73.2E) Salalah, Oman (16.94N, 54.0E) Ponte La Rue, Seychelles (04.67S, 55.53E) Rodrigue, Mauritius (19.67S, 63.42E) Diego Garcia Is. (7.23S, 72.43E) Cochin, Kerala (09.96N, 76.26E) Trincomallee Djibout, Gulf of Aden Benoa, Java Approx. wave height (cm) 227 20 50 30 4.6 1 2 3 1 10 20 60 15 50 20 29 5 7 13 18 140 40 100 5 25 60 Source GLOSS GLOSS Nayak and Kumar (2008) Nayak and Kumar (2008) NDBC (NOAA) Nayak and Kumar (2008) Nayak and Kumar (2008) Nayak and Kumar (2008) Nayak and Kumar (2008) Nayak and Kumar (2008) Nayak and Kumar (2008) NARA, Sri Lanka GLOSS Nayak and Kumar (2008) GLOSS Prabhudesai et al. (2008) Prabhudesai et al. (2008) UHSLC Nayak and Kumar (2008) UHSLC UHSLC UHSLC UHSLC UHSLC INCOS NARA, Sri Lanka UHSLC UHSLC

GLOSS Global Sea Level Observing System (http://www.gloss-sealevel.org), NDBC National Data Buoy Center, USA (http://www.ndbc.noaa.gov), NARA National Aquatic Resources Research and Development Agency (http:// www.nara.ac.lk), INCOIS Indian National Center for Ocean Information Services, Hyderabad (http://www. incois.gov.in), UHSLC University of Hawaii Sea Level Center (http://www.uhslc.soest.hawaii.edu)

t = 0 min (Fig. 11a), and at different time interval, it started propagating in open sea into southwest direction from the source. At t = 93 min, tsunami wave is reecting back toward source region from the Cocos island (Fig. 11a). Tsunami was touching the Sri Lanka at 194 min, India and Lakshadweep Islands after 245 min, and the observed tsunami amplitudes were few centimeters in these areas. The recorded tsunami amplitude on tide gauge was 20 cm at the coast of Chennai (Table 2). The AndamanNicobar Islands were on the shadow zone or parallel to the strike of the fault when compared to the Indian Mainland; this could be the cause of a bit more observed tsunami amplitude at Chennai which was 20 cm when compared to Portblair which was only 10 cm, though Andaman Nicobar Islands were nearer to the source region and Chennai was far from the source.

123

Nat Hazards (2011) 57:726

21

Fig. 9 Deformation of sea oor at the source region estimated by Mansinha and Smylie (1971). The scale is in meters

Fig. 10 Directivity map of minor tsunami triggered due to September 12, 2007, earthquake. Tsunami amplitudes are in meters

From Fig. 11b, at t = 283 min, the wave is getting reected from Chagos-Laccadive ridge and due to presence of these ridges the west coast of Somalia would not get affected if tsunami intensity would have been great. After t = 283 min, wave is getting reected from Lakshadweep Islands and would have struck Kerala coast and recorded maximum trough to crest wave height at Cochin and Verem, Goa tide gauges were around 26 and 29 cm, respectively (Table 2). The maximum trough to crest wave height at Kavaratti (Lakshadweep) was around 5 cm. After t = 421, it can be seen that Madagascar was not

123

22

Nat Hazards (2011) 57:726

Fig. 11 a Tsunami wave travel time at t = 0, 11, 31, 61, 93, 121, 151 and 181 min. Tsunami amplitudes are in meters. b Tsunami wave travel time at t = 211, 241 and 283 min. Tsunami amplitudes are in meters. c Tsunami wave travel time at t = 301, 421, 361 and 582 min. Tsunami amplitudes are in meters

getting affected by tsunami due to Island of Pointe La Rue and Rodrigues (Fig. 11c). The maximum wave height of tsunami from trough to crest was around 2.27 m at Padang, Sumatra (0.95S, 100.36E), as Padang station was nearer to the Source (Fig. 4), and the second highest wave of tsunami was around 11.2 m at Rodrigue, Mauritius (19.67S, 63.42E), as this station was perpendicular to source into the southwest direction.

123

Nat Hazards (2011) 57:726

23

Fig. 11 continued

7 Results and discussions Occurrence of M 8.7 earthquake just after 3 months of M 9.3 earthquake in an adjacent area indicates that it was most likely triggered by change in static stress due to the rst earthquake (Nalbant et al. 2005; Pollitz et al. 2006). Aftershocks of the two earthquakes are conned to the respective rupture zones. The large shocks are conned upto Nicobar (9N). The aftershock zone at 1015N is narrow and conned to west of Andaman Islands along the zone of convergence. Due to occurrence of two M 7.5, some 30 of M 66.8 and about 600 of M 55.9 aftershocks of December 26, 2004, earthquake, it was perceived by many that it is an unusually strong aftershock activity with lot of energy released by them. The magnitude difference of the mainshocks and respective largest aftershocks is within two magnitude units, which is as expected for large earthquakes in the Indian region (Lay et al. 2005). The percentage of energy released by aftershocks due to two mainshocks is 0.135 and 0.365%

123

24

Nat Hazards (2011) 57:726

Fig. 11 continued

of the energy of the respective mainshocks, while the strain release in aftershocks of December 2004 and March 2005 earthquake is 39 and 71%, respectively, which are quite large amounts. The least-squares and the maximum-likelihood methods were used to determine the b-value. The maximum-likelihood method gives better estimates of the b-values than the least-square method, and the estimate is stable for the number of events exceeding 50. The b-value for the December 2004 aftershocks sequence was (Fig. 5a, b). Thus, the b-value for the December 2004 aftershocks zone is smaller than that for the aftershocks zone of the March 2005 event, suggesting a relatively greater probability of large earthquakes in the aftershock zone of the December 2004 earthquake in comparisons with zone of March 28, 2005. Using the Gutenberg and Richter (1944) relation, b-value is estimated for both the cases, as they varied between 0.9 and 1.0. This indicated that the rupture processes is basically governed by the material properties (Ramana et al. 2009). The p-value for the aftershock sequence of March 28, 2005, was 0.88 (Fig. 6a, b). Hence, it can be inferred that the estimated p-value of 0.88 or 0.99 for the Andaman Sumatra earthquakes is less than the global median of 1.1 (Utsu et al. 1995). The estimated low p-value for the aftershocks sequence of 2004 and 2005 events support the low decay of the aftershock activity. The 2007 earthquake did not rupture the whole source zone of the 1833 event nor released the likely accumulated moment since then. This poses a serious threat for a future big earthquake that will possibly occur in the unruptured area extending between South Pagai and the south of Siberut Island (Lorito et al. 2008). This earthquake did not generate ocean wide tsunami. Only minor tsunami generated in local areas near to the epicentral zone. The tsunami propagated in the southwest direction from the northwest- and

123

Nat Hazards (2011) 57:726

25

southeast-oriented fault plane (Srivastava et al. 2007). Tsunami warning and watches were issued for all the coastlines of Indian Ocean (NOAA, PTWC and JMA).

8 Conclusions Seismic characteristics of the aftershock sequences of giant/great Sumatra earthquakes of December 2004 and March 2005 have been studied. The 2004 earthquake has shown an unusually strong aftershock activity with lot of energy released by these aftershocks. It is found that the energy release by aftershocks is not much. The energy released due to aftershocks sequences of 2004 and 2005 earthquakes was 0.135 and 0.365% of the energy of the respective mainshocks, while the strain release in respective aftershock sequences was substantial being 39 and 71%, respectively. The b-value and p-value indicate normal values of about 1. The b-value and p-value indicate normal heterogeneity, stress distribution and normal rate of decay of aftershocks. It is also observed from p-value that the aftershocks sequence attenuation was slow as the magnitude was large. The 2007 earthquake did not generate destructive ocean wide tsunami may be owing to shallow water depth at the epicenter and small amount of upward slip of sea oor of about 2 m and rupture area of around 160 9 80 km. Tsunami generated by this earthquake only affected the coastal areas near to the epicentral zone. Though moderate intensity of tsunami generated into Indian Ocean and it was recorded throughout the Indian Ocean of order of few centimeters, the tsunami propagated into the southwest direction from the northwestand southeast-oriented fault plane. The maximum tsunami energy is perpendicular to the fault plane. The directivity map shows that the maximum tsunami energy was southwest direction from the strike of the fault. Since the path of the tsunami for Indian coastlines is oblique, there were no affect along the Indian coastlines except near the coast of epicentral region. The water depth below the epicenter was shallow (*135 m or so) also responsible for generation of moderate intensity of tsunami. To generate destructive Ocean wide tsunami, water depth at the epicenter of the earthquake should be more and focal depth would be shallow enough in order to displace the ocean oor.
Acknowledgments The rst author is thankful to Dr. V. P. Dimri, Director, NGRI, Hyderabad for his kind permission to publish this work. We also appreciate the comments of two anonymous reviewers, which greatly improved the manuscript.

References
Aki K (1965) Maximum likelihood estimate of b in the formula log N = a - bM and its condence limits. Bull Earthquake Res Inst Univ Tokyo 43:237239 Ammon CJ (2006) Megathrust investigations. Nature 440:3132 Ammon CJ, Kanamori H, Lay T, Velasco A (2006) The 17 July 2006 Java tsunami earthquake. Geophys Res Lett 33:L24308. doi:10.1029/2006GL028005 Butler R, Lay TK, Creager C, Earl P, Fischer KM, Gaherty JB, Laske G, Leith B, Park J, Ritzwoller MH, Tromp J, Wen L (2004) The global seismographic network surpasses its design goal. EOS Trans AGU 85:225229 Dickinson WR, Seely DR (1979) Structure and stratigraphy of forearc regions. Am As Pet Geol Bull 63(1):231 Fujii Y, Satake K (2006) Source of the July 2008 West Java tsunami estimated from tide gauge records. Geophys Res Lett 33:L24317. doi:10.1029/2006GL028049 Fujii Y, Satake K (2007) Tsunami source of the 2004 Sumatra-Andaman earthquake inferred from tide gauge and satellite data. Bull Seis Soc Am 97(1A):S192S207

123

26

Nat Hazards (2011) 57:726

Fujii Y, Satake K (2008) Tsunami waveform inversion of the 2007 Bengkulu, southern Sumatra earthquake. Earth Planets Space 60:993998 Gutenberg B, Richter CF (1944) Frequency of earthquakes in California. Bull Seism Soc Am 34:185188 Gutenberg B, Richter CF (1956) Earthquake magnitude, intensity, energy and acceleration. Bull Seism Soc Am 46:105145 Imamura F (2006) TUNAMI-N2 (Tohoku Universitys Numerical Analysis Model for Investigation of Near Field Tsunamis version 2) Manual draft. http://www.tsunami.civil.tohoku.ac.jp/hokusai3/J/projects/ manual-ver-3.1.pdf Imamura F, Yalciner AC, Ozyurt G (2006) Tsunami modeling manual, UNESCO Tsunami modeling course (UNESCO) Jaiswal RK, Singh AP, Rastogi BK (2009) Simulation of Arabian Sea tsunami propagated due to Makran earthquake and its effect on the western parts of India. Nat Hazard 48:245258 Lay T, Kanamori H, Ammon JC, Nettles M, Ward NS, Aster R, Beck LS, Bilek LS, Brudzinski RM, Butler R, DeShon RH, Ekstrom G, Satake K, Sipkin S (2005) The great Sumatra-Andaman earthquake of 26 December 2004. Science 308:11281133 Lorito S, Romano F, Piatanesi A, Boschi E (2008) Source process of the September 12, 2007, MW 8.4 southern Sumatra earthquake from tsunami tide gauge record inversion. Geophys Res Lett 35:L02310. doi:10.1029/2007GL032661 Mansinha L, Smylie DE (1971) The displacement elds of inclined faults. Bull Seism Soc Am 61(5): 14331440 McCloskey J, Nalbant SS, Steacy S (2005) Indonesian earthquake risk from co-seismic stress. Nature 434: 291 Mogi K (1962) Magnitude-frequency-relation for elastic shocks accompanying fractures of various materials and some related problems in earthquakes. Bull Earthquake Res Inst Univ Tokyo 40:831853 Nalbant SS, Steacy S, McCloskey J, Sieh K, Natawidjaja D (2005) Earthquake risk on the Sunda trench. Nature 435:756757 Natawidjaja DH, McCaffrey R (2006) Plate-boundary deformation associated with the great SumatraAndaman earthquake. Nature 440:4651 Nayak S, Kumar TS (2008) Indian tsunami warning system; the international archives of the photogrammetry, remote sensing and spatial information sciences. Part B4. Beijing XXXVII:15011506 Ogata Y, Utsu T, Katsura K (1993) Statistical features of foreshocks in comparison with other earthquake clusters. Geophys J Int 121:233254 Omori F (1894) On aftershocks of earthquakes. J Coll Sci Imperial Univ 7:111200 Park J, Butler R, Anderson K, Berger J, Benz H, Davis P, Hutt CR, McCreery CS, Ahern T, Ekstrom G, Aster RC (2005) Performance review of the global seismographic network for the Sumatra-Andaman megathrust earthquake. Seism Res Lett 76(3):332343 Pollitz FF, Banerjee P, Burgmann R, Hashimoto M, Choosakul N (2006) Stress changes along the Sunda trench following the 26 December 2004 Sumatra-Andaman and 28 March 2005 Nias earthquakes. Geophys Res Lett 33:L06309 Prabhudesai RG, Joseph A, Mehra P, Agarvadekar A, Tengali S, Vijay K (2008) Cellular-based and Internet-enabled real-time reporting of the tsunami at Goa and Kavaratti Island due to Mw 8.4 earthquake in Sumatra on 12 September 2007. Curr Sci 94:11511157 Ramana DV, Srinagesh D, Chadha RK (2009) Spatial analysis of the frequency magnitude distribution of aftershock activity of the December 2004 tsunamigenic Sumatra earthquake. Curr Sci 96(6):834837 Romanowicz BA, Giardini D (2001) The future of permanent seismic networks. Science 293(5537): 20002001 Sieh K, Natawidjaja D (2001) Neotectonics of the Sumatran fault, Indonesia. J Geophys Res 105(B12): 2829528326 Srivastava K, Swaroopa V, Srinagesh D, Dimri VP (2007) Could the 12 September 2007 earthquake of southern Sumatra, Indonesia, have generated a large Tsunami causing damage to the east coast of India. Curr Sci 93(9):12281229 Stein S, Okal E (2004) Speed and size of the Sumatra earthquake. Nature 434:581582 Subarya C, Chlieh M, Prawirodirdjo L, Avouac JP, Bock Y, Sieh K, Meltzner AJ, Natawidjaja D, McCaffrey R (2006) Plate boundary deformation associated with the great Sumatra-Andaman earthquake. Nature 440:4651 Utsu T (1961) A statistical study on the occurrence of aftershocks. Geophys Mag 30:521605 Utsu T, Ogata Y, Matsuura RS (1995) The centenary of the Omori formula for a decay law of aftershock activity. J Phys Earth 43:133 Woessner J, Wiemer S (2005) Assessing the quality of earthquake catalogs: estimating the magnitude of completeness and its uncertainty. Bull Seism Soci Am 95:684698

123