Seismotectonics of Andaman-Sumatra Region: Seismic Quiescence and B Value as Possible Precursors

Objectives: a) To analyze the long-term seismicity and understand the potential of seismic quiescence study to use it as a reliable seismic precursor b) To estimate B value and understand the variation in B value as earthquake precursor in seismically active region. Methods: The present study analyses the seismicity pattern from the Andaman-Sumatra region for a period of 1964 to 2018. The area has been divided into epicentral blocks. Earthquakes preceding and succeeding a major earthquake with diﬀerent seismic phases of quiescence and pattern of seismicity have been studied carefully. All quiescence periods are characterized by high b values and period of major shocks has a low b value. Main shock events for each epicentral block with diﬀerent phases of quiescence (Q1, Q2 and Q3) and active seismicity have been identiﬁed and analyzed. Findings: The study suggests that there is generally approximately 6-12 years of gap between major earthquakes. There are 28 years of quiescence before the major earthquake of the stature of Dec 26, 2004 mega earthquake, suggesting that long term quiescence leads to great earthquakes. The study also proposes that the area between 2 0 and 6 0 N latitudes shows a long-term quiescence which may lead to a major earthquake in the near future. A thorough analysis of long-term seismicity and seismic quiescence can be used as an earthquake precursor, though with limitations. The latest post seismic quiescence period Q2 after the greater events of 2004 and 2005, may yield an impending event of 6.5 M or even greater. Novelty: The present study deals with the precursor studies of one of the most important tectonic zone. Extensive analysis of 54 Years of data has been done and the scope of using seismicity and b value for the long-term prediction of earthquakes has been examined. The periodicity of seismicity has been used to quantify the precursors. Though various statistical studies have been attempted for this zone, methodology adopted here proved one of the reliable methods to check for the possible mega earthquakes of this seismically active zone, which can be applied to other seismically active areas as well.


Introduction
Sunda Arc is considered as a classic example of a volcanic island arc, where the Indo-Australian plate subducting under the Sunda and Burma plates (Figure 1). Subduction and active faults along with seismicity characterizes the arc and the tectonic deformation along this zone held responsible for the catastrophic Indian Ocean earthquakeof 26 December, 2004. The obliquity of the subducting plate make this zone a nodal area for the accumulation of strain energy (1) . Historic great earthquakes along this plate boundary occurred in 1797 (~8.4), 1833 (~9), 1861 (~8.5) and1907 (~7. 8). It is observed that large thrust earthquakes in 1847 (~7.5), 1881 and 1941 occurred on intermediate regions of the down-dip boundary areas that have been surrounded and probably incorporated into the 2004 rupture (2) . The Andaman Sea, which extends approximately 1,250 km from Myanmar to Sumatra lying east of Andaman-Nicobar is a back-arc spreading ridge (3)(4)(5). The fault systems divide Andaman Sea into Shallow fore arc and deeper back arc regions. The Andaman Sea has also experienced a number of earthquake swarms in 2014, 2015 and 2019 after the megathrust earthquakes a result of complexity in tectonic deformation (6) . West Andaman Fault (WAF) and the Sumatra fault system (SFS) form the boundary between Burma plate and Sunda plate south of the spreading ridge. The 1900-km-long Sumatran fault exhibits pure strike-slip faulting, which extends the entire length of the Sumatra Island. It is highly segmented and coincides geographically with the volcanic arc (7,8) . As per the historical and instrumental catalogs of Sunda arc, most of the major earthquakes are located in the Sumatran forearc and the deformation pattern suggest dominance of compressive stress in the area (9). West Andaman Fault is another major strike-slip fault connected to the Sumatra fault in the south and Sagaing fault to the north. The Andaman spreading center in the northern part controlled the release of seismic energy to the trench, but absence of such a structure in the south which can accommodate the extensional deformation allowed the seismic energy to transfer towards the trench (1). The earthquake activity appears to be starting from the trench location towards east, particularly in South Andaman (10) . The study on seismicity and the rate of seismicity plays a key role in understanding the stress distribution in the crust. Several studies have been done on spatial and temporal correlations interpreted as possible precursors of earthquakes, such as gas emissions, fluctuations in ground water level, fluctuation in temperature and electromagnetic field, ionospheric perturbations and statistical analysis (11) . Geophysical precursors include seismic quiescence, Accelerating Moment release (AMR) and changes in b value. It may be never possible to predict an earthquake absolutely, we still have scope for identifying precursors which can be useful as part of an early warning system. Seismic quiescence analysis has shown promising results in identifying precursory abnormalities (1,3,4,6,12,13) . The objectives of the present study are proposed as understanding precursors like https://www.indjst.org/ temporal quiescence, seismicity pattern, earthquake occurrence rate, and changes in temporal 'b' values to provide additional information regarding the seismic potentiality and future seismic hazard in Andaman-Sumatra region. Although researchers have attempted the quiescence study for Himalayas (14) , Burmese arc (4,15) based on Scholz (16) methodology, Andaman-Sumatra region were studied based on statistical Z test using a data base for 35 years (12,13) . Long term data helps to recognize the seismic potential of active zones. The present study deals with the precursor studies of one of the most important tectonic zone. 54 Years (1964-2018) of data has been used for the analysis and the scope of using seismicity and b value for the long term prediction of earthquakes. The periodicity of seismicity has been used to quantify the precursors. Though various statistical studies have been attempted for this zone, the methodology of Scholz (16) proved one of the reliable method in order to check for the possible mega earthquakes of this seismically active zone.

Methodology
ISC (International seismological Center) bulletin data has been used for the present study. Considering the fact that monitoring capability for smaller earthquakes (magnitude <4.5) is inadequate, we consider data above this threshold magnitude for the analysis. For the study purpose we divided the area into 6 epicentral blocks with 2 o overlap of latitudes so as to understand the continuity in seismic activity ( Figure 2). The study area is bounded by -4 o S-10 o N latitude and 88 o E-100 o E longitude. Around 16363 events of body wave magnitude greater than 4.5 are reported from this region within a period from 1964 to 2018. Bathymetry map of the study area prepared using GEBCO bathymetry data. Epicentral blocks considered for the study are shown as rectangles with 4 0 widths with 2 0 overlapping. Seismicity of the area is shown and the great earthquakes are marked as stars.

Seismic Quiescence
Scholz (16) has divided the seismic cycle of an active seismic domain into alternate domains of quiescence (Q1, Q2 and Q3) and active seismicity. "Q1 is defined as quiescence period between a major shock-aftershock sequence and a further increase in background seismicity. Q2 is defined as a lull period between the previously defined increase in background seismicity and further renewed seismicity. Q3 is defined as a short-term quiescence of seismicity prior to a major shock-aftershock sequence". Based on number of events, seismic occurrence rate and magnitudes seismic quiescence periods of each epicentral block are found out. Since the epicentral blocks 3, 4, and 5 are greatly influenced by two major 'intraplate' earthquakes in 2012 and the aftershocks and foreshocks associated with it, those earthquake events are carefully removed from the data set. https://www.indjst.org/

B-Value
'b-value' in the Gutenberg-Richter frequency-magnitude relation (17) is one of the basic seismological parameters used to describe an ensemble of earthquakes. Size distribution of earthquakes in a seismogenic volume can often be adequately described over a large range of magnitudes for different tectonic regions by a power law relationship. Gutenberg and Richter (17) is one of the well-fitted empirical relation in seismology, it represents the frequency of occurrence of earthquakes as a function of magnitude: where 'N' is the cumulative number of earthquakes with magnitude larger than 'M', a and b are constants. While the parameter a describes the heterogeneity of the medium, b denotes the slope of the linear frequency-magnitude plot. b-value represents properties of the seismic medium like stress and/or material conditions of the focal region (18) . The method of estimation of b value by regression is known as least squares method (LSQ). Alternatively, the maximum likelihood method (MLM) (19) is widely used for estimation of b value which is also used for study of epicentral blocks in this project. Generally, for b value estimation the MLM method is preferred over the least squares method (LSQ) (20) , since there are more uncertainties associated with LSQ. The b value represents the relative occurrences of small and large earthquake events (21) and a measure of ratio between them. It suggests the effective stress regime and tectonic character of the region (22,23) . Prior to major earthquake events, 'b' value decreases within the seismogenic volume that correlates with increasing effective stress levels (24) or an increase in applied shear stress /effective stress demonstrated that b-value varies for different styles of faulting. Highest b-values (~1.2) are associated with normal faulting, intermediate values (~1.0) are associated with the strike-slip events and lowest values (~0.90) show thrust events. Thus, 'b' acts as a stress meter in earth crust and depend inversely on the applied differential stress (3) .
In the present study, maximum likelihood method (19) has been adopted for the estimation of the b-value and it is estimated as b = (log 10 e) / (Mav -Mmin), where Mav is the mean magnitude above the threshold Mmin. The maximum-likelihood method provides the least biased estimate of b-value. Confidence limits of the b-value are inversely proportional to the square root of the number of events; thus, error is estimated by the formula, b / √ N. As b-value is dependent on data, earthquake data is treated separately as per techniques described by KulhÃ¡nek, Ota (18) to make the calculated b-value statistically strong and tectonically significant. The b-value here is calculated for different time period from 1964 to 2018 for each epicentral block.
Here also 2012 events are excluded from the data of epicentral blocks 3, 4, and 5.
The earthquake size distribution (Guttenberg -Richter relation) (17) follows the well-known power law designated as b value that is commonly used to designate the relative occurrences of large a n d small earthquake events (23) . The b values are calculated using the equation where x i is magnitude, y i = log N where 'N' is the number of earthquakes and 'n' is the number of magnitudes. A decrease in b value within the seismogenic volume indicate increasing stress levels prior to major shocks (24) . Several studies have been carried out to check the potential temporal changes in b value as a short term, medium term and long term precursor. Results show that large earthquakes are often preceded by a medium-term increase in b, followed by a decrease in the weeks-months before the earthquake. Overall seismicity pattern of the area shows increased seismicity in recent times especially after 2004 megathrust earthquakes, suggesting disturbances in the tectonic setup caused by these megathrust earthquakes lead to the increased tectonic activity in this zone.

Seismic Quiescence Periods and b-Value Estimates
The number of major shocks pertaining to epicentral blocks and b value calculated for each block has been carefully analyzed in the following session.

Epicentral block 1
Five major shocks have been identified for the epicentral block 1 during the time period 1964 to 2018. The first major shock happened in 15-7-1964 whose magnitude was 6.8. This main shock was followed by a post seismic quiescence Q1 which lasted two years (seismic occurrence rate = 4 events/year).1967 to 1968 marks a period of increase in background seismicity (seismic occurrence rate = 8.5 events/year). 1969 was a period of intermediate quiescence Q2where only 3 noticeable events happened in the year. This Q2 period was followed by main shock (mb=5.6, seismic occurrence rate = 20events/year) in 1970. The Q3 quiescence before this event lasted 2 hrs. The three other main shocks happened in 1982, 2004 and 2014, which follow the same pattern and is listed in the Table 1. Currently the area is in a period of post seismic quiescence.

Epicentral block 2
The epicentral block 2 has witnessed three major earthquakes for the period of 1964-2014. The first main shock occurred in 1967 with magnitude 6.1(seismic occurrence rate= 19 events/year), which was preceded by a Q2 quiescence period of two years (seismic occurrence rate = 5 events/year) and a Q3 period of three hours. Then followed a Q1 period of 2 years (seismic occurrence rate is 5 events/year). The second major event happened in 1982 followed by 2004. The seismic cycle of this area is shown in Table 2.

b Value estimates
The b value calculated for each time periods (quiescence periods active periods and period of increase in background seismicity) listed in the tables. Effective stress regime and tectonic character of the region decides on the parameter 'b' (22,23) . Lowering of 'b' within the seismogenic zones correlates with increasing effective stress levels prior to major shocks (24) . All quiescence periods are characterized by high b values and period of major shocks has a low b value. This pattern is followed in all epicentral blocks.

Conclusion
The seismicity pattern from the Andaman-Sumatra region has been studied for 1964 to 2018. The area has been divided into six epicentral blocks with 50% overlapping, in which major earthquakes have been identified. The study suggests that there are generally approximately 6-12 years of gap between major earthquakes. But combined analysis of blocks 3 and blocks 4 reveals that the area was comparatively quiet for a long period of around 28 years before the megathrust earthquakes of 2004, suggesting that long term quiescence leads to great earthquakes. Studies on long term quiescence before 2004 earthquake was also carried out by other researchers (25) . Magmatic pulsations can result in earthquake swarms in volcanically active areas such as Off Nicobar region (6) . Our analysis shows that the northern segments are comparatively quiet since the megathrust earthquake which can lead to a major earthquake in the near future. Bhatt et al (26) based on the mapping of coseismic ruptures of the Eastern boundary thrust of Andaman over the last 2000 years also suggest an increase in slip deficit which can lead to a https://www.indjst.org/ large magnitude earthquake in the Andaman-Nicobar region.
The study suggest that a proper study of the long term seismicity and seismic quiescence can be used an effective earthquake precursor. All quiescence periods are characterized by high b values and period of major shocks has a low b value. Such studies can help in preparing a mitigation plan of seismic hazard. The major conclusions of the study are • Combined analysis of epicentral blocks suggest that blocks 3 and 4 reveals that the area was comparatively quiet for a long period of around 28 years before the megathrust earthquakes of 2004, which was unusual. • Long term quiescence leads to great earthquakes and special attention may be given to such regions. • Changes in the temporal b-value, including decrease in b value can be used as a parameter for the prediction of an impending event. • Our analysis shows that the northern segments of Andaman region are comparatively quiet since the megathrust earthquake that may be leading to a major earthquake in the near future. • The study can be helpful in long term forecasting. However, the exact location and time of the event cannot be predicated from the earthquake data alone. Detailed information based on local seismic network and GPS studies can lead to more precise calculations on seismicity.