Seismic hazard estimation and medium-term earthquake precursor analysis of North East India: an assessment on large earthquake scenario

ABSTRACT The North-Eastern region of India has the distinction of being one of the most seismically active and tectonically complex regions in the world. The region has periodically jolted by a number of large earthquakes. The temporal and spatial frequency-magnitude distribution (FMD) of earthquakes in three major selected regions of North-East India, namely Shillong–Mikir Hills, Arunachal Himalaya, and Assam Foredeep region, are examined in this study. Temporal variation of ‘b’ shows a significant declination prior to a large earthquake, which strongly advocates for a medium-term (months-year) earthquake precursor. Similarly, spatial distribution map is prepared by estimating b-value at every grid using the nearest 150 events, which is vital in understanding the stress regime of a specific region. The present study critically demonstrates the higher b-value regions associated with minimum seismicity and vice-versa, especially focusing on large earthquake scenario. High anomalies of ‘b’ were found in the regions of Shillong–Mikir hills and Arunachal Himalaya, whereas the Assam-Foredeep region, free from seismic activity for a long period, shows a low. It may be due to the accumulation of stress energy within the region for a long time and could be an alarming sign for future large events.


Introduction
The North Eastern region (NER) of India exists in the zone V of seismic zonation map (BIS, 2004) of India, holds the distinction of being one of the most complex tectonic regimes in the world (Kayal, 2001(Kayal, , 2008;;Nandy, 2001).Over the past few decades, the area has been the site of numerous big earthquakes (M � 5:0).The persistent seismicity as well as the diverse geological features of the region have always piqued the interest of geoscientists, prompting them to investigate the region's tectonics thoroughly.
Earthquakes are one of the natural hazards that have always been important to human beings due to the serious threats they pose.Forecasting the major upcoming events has always been a common interest in the research field.Although prediction of the exact day and time of a future earthquake is not possible at this time, one can characterise earthquakeprone areas to reduce the devastating effect on human life.Understanding the statistical distribution of earthquake events can be used to calculate the probability of the occurrence of large events.The Gutenberg-Richter relation (Gutenberg & Richter, 1944), is one of the most popular approaches that can be used to assess the probability of significant events in near future from the available earthquake catalogue.However, the quality of the catalogue is crucial while determining the seismic hazards.Based on the capacity and coverage of seismic recording stations, earthquake events may vary throughout different time periods.At that time, where major focus stays with the magnitude of completeness (M c ) of the catalogue, which is always changing.It often gets smaller over time as instruments and processing techniques become more effective at detecting events.
The statistical distribution of sizes of group of earthquakes is very complex and has a power law distribution (Ishimoto & Iida, 1939).In view of these power law, the tectonic implications of seismically prone zones can be well investigated by studying the b-value of seismicity.The b-value obtained from Gutenberg-Richter frequency magnitude relation (Gutenberg & Richter, 1944) characterises the earthquakes over the observed range of magnitudes.The b-value which is related to the differential stress distribution of earth crust can be used for seismic hazard assessment study in a region.Depending on the tectonic setting and seismicity the value 'b' changes accordingly (C.Singh & Chadha, 2010;C. Singh et al., 2008;C. Singh, 2014;Pacheco et al., 1992;Wiemer & Wyss, 1997), however for a short time window it reaches to 1.0 for a particular CONTACT Debasis D Mohanty devlinkan06@yahoo.comGeoscience and Technology Division, CSIR-North East Institute of Science and Technology, Jorhat, India region.The mapping of b-value on a local to regional scale in various tectonic regimes results in a value between 0.4 and 2.0, (Wiemer, 2000).Study of b-value both in time and space can be used as medium-term (months-year) precursor for major future earthquakes (Nuannin and Kulhánek, 2012;Nuannin, 2005).Several researchers studied spatial variation of b-value (C.Singh et al., 2008Singh et al., , 2009;;C. Singh, 2014;Wiemer & Katsumata, 1999) in seismically active areas, which reflects locally the effective stress (Scholz, 1968).There are many factors that are responsible for the deviation of b-value from the normal value of 1.0.Generally high b-value is observed when material density or crack density increases (Mogi, 1962) and significant low values are observed in the areas of increased shear stress (Scholz, 1968) or effective stress (Wyss, 1973).In simple words, b-value can be used as a stress metre, which can be used to measure the accumulated stress in a tectonically active zones (C.Singh, 2014;Wiemer & Wyss, 1997).In the context of shear stress, a low b-value indicates that there is very high accumulated shear stress in a region whereas a high value indicates some major tectonic events that have already occurred in the past or the release of shear stress (Chan et al., 2012;Prasad & Singh, 2015).This inverse relation of b-value with the accumulated stress in and around epicentral region of major shocks can be used in estimating the possible future large earthquake.
The main goal of the present study is to conduct a thorough study on the evaluation of b-value with respect to both space and time for the selected three regions of North-East India, (Figure 1) in order to assess the high probability or expectation of large earthquakes in the NER.This study also aims to understand the present hazard scenario and the futuristic assessment of earthquake probabilities for a better social benefit of the general public with a constructive urbanisation of this region.The North-Eastern region of India tends to be a region of complex tectonics and seismically active zone from time to time.Many large earthquakes (M � 6) jolted the whole region in past and there is a high chance of occurrence of major events in near future.This active seismicity of the North-East India and the high probability of future earthquakes always being the major concern in the seismological research.In our present study, we have divided the whole region into three major entities as Shillong-Mikir Plateau, Arunachal Himalaya and Assam-Foredeep and try to estimate the b-value for the respective sub-regions to understand the present and future hazard scenario.For the estimation of b-value data collected from 1900 to 2022, including some major historical events are used in this study (Figure 1).

Geodynamics of NER
The North-Eastern region (NER) of India is one of the most tectonically active zones in the world, and is represented by India-Asia collision to the north and Indo-Burmese subduction to the east (Angelier & Baruah, 2009;Bilham & England, 2001;Bora et al., 2014).The region has experienced numerous big earthquakes ðM > 5:0Þ in the past few decades.The persistent seismicity in the region from time to time as well as diverse geological features has a direct impact on the region's complex tectonics structure.The region's significant geological features include the Shillong-Mikir Plateau, Brahmaputra valley, Indo-Burmese Subduction zone, and Eastern Himalayan Syntaxis (Mohanty & Mondal, 2019, 2020, 2021;Mohanty, 2023;Mondal & Mohanty, 2021).
The Shillong Plateau, which is a part of the Indian Shield (Evans, 1964), is notable for being a seismically active zone in the India.Numerous lineaments can be found in the Plateau and its surrounding areas as a result of complex tectonic forces from the Indo-Burmese subduction zone and the Himalayan collision zone (Kayal, 2001).The Shillong Plateau-Mikir hills extends from the Himalayas in the north to Bangladesh in the south.About 60 million years ago, when the Indian and Eurasian plates began to collide in a north-south orientation, the Shillong Plateau was formed.The Shillong Plateau is an ancient cratonic block in North-Eastern India, comprising roughly 47,614 km 2 (Mohanty & Mondal, 2020;Mohanty et al., 2021;Sarma, 2014), with regional heights of up to 2.0 km (Najman et al., 2016).The Shillong Plateau is composed of Precambrian gneissic rocks in the south, east, and west, with Proterozoic and Cenozoic sediments partially overlain (Sarma, 2014).There are numerous faults which encircle the Shillong plateau, and the plateau itself has numerous faults, namely Dhubri fault at the west and the Kopili fault at the east (Evans, 1964).The most prominent one is the Kopili fault which runs through the Shillong plains to the Mikir hills.This fault zone was the site of the majority of the large earthquakes in the past, including the Cachar earthquake (M w ~7.5) and the 1943 Assam Earthquake (M s ~7.2) (Baro & Kumar, 2015).The northern and southern boundary of the Shillong plateau is demarcated by E-W trending Brahmaputra fault and E-W trending Dauki fault respectively (Angelier & Baruah, 2009).Dauki fault also divides the plateau from the Bengal basin (Dasgupta & Nandy, 1982).The E-W trending Dauki fault had witnessed some of the prominent earthquakes in the past including 1923 Meghalaya earthquake (M s ~7.1) (Bilham & England, 2001).The plateau also has two significant thrusts, Barapani shear zone and the Dapsi thrust, which is the western extension of the Dauki fault, runs through the plateau in a NW-SE orientation and is 90 ~ 100 km long (Kayal, 1991).The NW-SE trending Kopili fault divides the entire massif into two parts Shillong Plateau and Mikir hills (Mohanty & Singh, 2021;Mohanty, 2022).The Kopili fault is 300-400 km long and 50 km wide fault and it may commence in Bomdila and stretch through the Assam valley across the east margin of the Mikir plateau to the Naga Hills (Abdel-Gawad, 1972;Baruah et al., 1997;Nandy, 1980).The depth variation of the past seismic events along the Kopili fault indicates most of the events have a depth range less than 70 km.The N-S trending Dhubri fault, which separates the plateau from the Indian subcontinent, lies to the west of the plateau.Other faults in and around the plateau include the Chedrang fault, the Samin fault, the Dudhnoi fault.The considerable seismic activity in and around the Shillong plateau complicates the geodynamic setting of the region.
The easternmost part of the Himalaya, between longitude 91.30'E and 96E occupied by Arunachal Himalaya.The eastern Himalayan syntaxis is also a part of it (Yin et al., 2006).This part of the Himalaya is located in the easternmost side of the Bhutan.Many pioneer researchers had done extensive studies in the region to study the geology and tectonic structure of the region.The Himalaya's exhumation and growth are a continual process caused mostly by reverse faults in which the rocks on the bottom surface of a fault plane move beneath relatively static materials on the top surface, a process known as under-thrusting of the Indian plate beneath its Eurasian plate.This process is fundamentally responsible for creating a significant seismic hazard in the Himalayan Mountain belt and surrounding areas by continuously changing the drainage patterns and landforms.The Himalaya connects with Indo-Burmese range taking a sharp southward bend and the zone is known as Tuting-Tidding Suture Zone (TTSZ) (Acharyya & Sengupta, 1998;Holt et al., 1991).A study carried out by Wadia Institute of Himalayan Geology, Govt. of India suggest that most of the low magnitude earthquakes are concentrated at depth 1-15 km and higher magnitude earthquakes ðM > 4Þ concentrated in the depth 25-35 km in the easternmost part of India.The fluid/partial melt zone is located at middle depth and is free from seismic activity.The crustal thickness in this area varies from 46.7 km beneath the Brahmaputra Valley to about 55 km in the higher elevations of Arunachal (A.Singh et al., 2017;Mohanty et al., 2016), with a marginal uplift of the contact that defines the boundary between crust and the mantle technically called the Moho discontinuity.This, in turn, reveals the underthrusting mechanism of Indian plate in the Tuting-Tidding Suture Zone.Extremely high Poisson's ratio was also obtained in the higher parts of the Lohit Valley, indicating the presence of fluid or partial melt at crustal depths.This detailed assessment of seismicity in this region will be helpful for planning any large-scale construction in this region in the future.
The Assam-Foredeep is a small valley that stretches from ENE-WSE and is bordered to the north and east by two young mountain ranges.The seismicity of the Assam-Foredeep region is quiet low for a long time.
There is an area called 'Assam gap' in the upper Assam valley (Khattri et al., 1983) of North-East India that has experienced essentially no seismic activity for the past 60 years (between Kopili fault and Mishmi thrust,92 o E À 96 o E).After such a long time, the region of Assam gap was jolted by an earthquake of magnitude, M b ~6.2 on 28 April 2021, followed by a series of aftershocks.The epicentre of this earthquake is located on the Kopili fault.This earthquake draws an immediate attention following a sudden burst of accumulated stress that has been building up for a long time, and it may also be a precursor to larger earthquakes in the near future (Mohanty et al., 2021).

Historical seismic events of North-East India
India's North-Eastern region and the surrounding areas are located in a tectonically unstable and seismically active area.The North-eastern region of India falls in the zone V of Seismic Hazard Zoning Map (BIS, 2004) which is more prone to earthquakes; being a highly populated area it draws the attention of the many researchers to understand the mechanism of seismic events, active faults, seismogenic sources and their effects.Using seismological data captured by the permanent and temporary analog networks between 1980 and 1990, the region has been well investigated (e, g, Khattri et al., 1983).The region is presently well equipped with various broadband seismological instruments, making it possible to fairly understand its seismotectonics.The Shillong Plateau and the areas around it have historically experienced a number of strong earthquakes.The great Assam Earthquake of 1897 (M w ~8.7), the Cachar earthquake of 1869 (M w ~7.5), the Dhubri earthquake of 1930 (M s ~7.1), the Meghalaya earthquake of 1923 (M s ~7.1), and the Assam earthquake of 1943 (M s ~7.2) are among the most powerful earthquakes ever recorded.
The Great Assam earthquake (M w ~8.7) was the most powerful earthquake to strike Assam in its history.The earthquake hit the Shillong plateau and its neighbouring areas on 12 th June, 1897 (as reported by Bilham & England, 2001).Many aftershocks were reported until July 25 th 1897.The damages caused by the earthquake were severe enough to cause fissure in the ground, cracks in construction, landslides and flooding in various sites of the Assam valley as well.The course pattern of river Brahmaputra also changed significantly due to the 1897 Assam earthquake.This earthquake shook various parts of the Indian subcontinent, causing major damage to structures in Kolkata and Myanmar (Baro & Kumar, 2015).Another strong earthquake (M w ~7.5) hit the entire North-eastern region on 10 January, 1869.The rupture zone of the earthquake was located within the Kopili fault.Severe damages were caused by this earthquake with influences ranging from Dibrugarh in the north to Manipur in the east, Patna in the west, and Kolkata in the south.This earthquake caused a number of ground fractures and sand vents in various regions.
Shillong plateau has the distinction of being the most seismically active region in Assam's history.The plateau had previously undergone two major earthquakes in the past of magnitudes 7.1 and 7.2.On 9 September,1923, a 7.1 magnitude earthquake struck the southern edge of the plateau.The epicentre of the earthquake is located on the Dauki fault.This earthquake had an impact on parts of Meghalaya, Assam, and West Bengal.The region was once again shocked by an earthquake of magnitude 7.2 on 23 October, 1943, causing fractures in the ground, tree falling, and structural damage.The stress pattern of Kopili fault could be the source of this earthquake, however very less information is available related to these earthquakes.The North-eastern region of India had experienced another historic event (M s ~7.1) on 2 nd July, 1930.According to Kayal (2008); main source of this earthquake was the Dhubri fault.Dhubri town which is near to the Dhubri fault was greatly affected with an intensity of IX on the Rossi-Forel scale (Olympa and Kumar, 2015).A number of aftershocks followed by the main shock are reported (Gee, 1934) till 5 July, 1930.

Data and methodology
In the present study we have selected the study area between longitudes 90-98° and latitudes 24-30°, (Figure 1).The event data that used in this study were collected from International Seismological Centre (ISC), United States Geological Survey (USGS), National Centre for Seismology (NCS) and CSIR-North-East Institute of Science and Technology (CSIR-NEIST) catalogue from the year 1900 to 2022.With the available data we try to estimate the b-value for the whole region and also studied the variation of 'b' with space and time.One essential parameter for mapping the spatial b-value is the magnitude of completeness (M c ).The magnitude of completeness (M c ) is calculated using the maximum curvature method from the Gutenberg-Richter relation (Wiemer, 2000).The magnitude of completeness (M c ) is an important parameter which is related to the quality of the data or the catalogue.M c can be defined as the smallest magnitude for which all events in a selected space-time volume are detected.It varies both spatially and temporally.Generally, M c decreases with time as the number of seismic monitoring stations increases and method improves continuously.In the present study events with magnitude (M � M c ) were selected for mapping the b-value (Figure 2).
The statistical distribution of size for a group of earthquakes is complicated and has a power law distribution.Several research on the frequency-magnitude relationship of earthquakes as a function of time, space, and depth have been conducted, establishing an empirical relationship between the occurrence-frequency and magnitude of earthquakes in a specific region within a finite time-window (Gutenberg & Richter, 1944;Ishimoto & Iida, 1939).The simplest earthquake frequency-magnitude relationship provided by Gutenberg and Richter (1944) which is as follows, Where 'a' and 'b' are constants and N is the number of earthquakes in the group with magnitudes greater than M. The constant 'a' relates to the length of the window under examination as well as its volume, and 'b' measures the ratio of strong to weak earthquakes and is thought to be related to the tectonic regime of the region in question.There are two possible methods available to estimate the b-value, least square methods (LS) and maximum likelihood methods (ML).
The maximum likelihood method (ML) is frequently cited as a more accurate estimation of b-value than the least-squares method (LS) (Aki, 1965;Hirata, 1989;Utsu, 1965).In the present study we have used the maximum likelihood method for mapping the b-value using the software package ZMAP (Wiemer, 2001).
Accordingly, b-value can be defined as, Where � M is the average magnitude and M o is the threshold magnitude used in the analysis, and log 10 e = 0.4343.
At first the entire database consisting of 7250 number of events were declustered using Reasenberg decluster algorithm (Reasenberg, 1985).The events with magnitudes scaled in M b -scale are only taken into consideration to avoid the asymmetry in scaling laws to analyse the seismicity and therefore b-values.The declustering of the catalogue is important because it removes the effects of dependent events (aftershocks) (Chan and Chandler, 2001).However, there is no direct procedure to identify the aftershocks, the mostly applied method is Reasenberg decluster algorithm (Reasenberg, 1985) for aftershocks identification or removal.In this method an interaction zone is created around each earthquake in the catalogue.Earthquakes that have occurred after a main shock and falling into the same interaction zone are classified as aftershocks and are considered to be dependent events.Several parameters have to be chosen for this procedure.The original database contains a total of 7250 events which is then limited to 6228 events after the declustering.This declustered database is then used for the estimation of b-value for the study region.The entire study area is divided into a 0:01 o � 0:01 o grid cells to map the variation of b-value using the declustered database consisting of 6228 events (Figures 1, 2 and Figure S1).For each node a minimum number of events (N min ) was assigned and the grid cell consisting of less than N min were removed.In the present study we have estimated b-value for the whole study area as well as for three main regions of interest namely Arunachal Himalaya, Shillong-Mikir Plateau and Assam-Foredeep.The obtained results of spatial b-value for these three regions give a better picture for interpreting the hazard scenario of the study area.

Results and discussions
The frequency-magnitude distribution (FMD) of the earthquakes in the North-Eastern region of India is analysed in the current study both in a temporal and spatial scale.The findings of analysis show that b-value significantly declines before the occurrence of big events, suggesting that the changes in the 'b' may be employed as a medium-term (month-to-year) earthquake precursor.During the studied period, no large earthquakes were observed in areas resulting in a high b-value.Mapping the b-value provides information about the stress regime of a specific region.Changes in the b-value with time following large earthquakes also suggest that b(t) can be used as a short-term (dayto-month) earthquake precursor for the aftershocks sequence.

b-value mapping
An overall b-value map is prepared using the available database collected from various seismological agencies as mentioned above.The database was declustered first and the map was prepared using the events, M � M c .The obtained result (Figure 3a)  3a).On the other hand, the higher b-value regions are associated with no large earthquakes (Figure 3a).

Resolution map
The geographical resolution is an important parameter in the study of b-value map.Resolution map is directly linked to the density of the earthquakes associated with each grid nodes where the resolution of the map is inversely proportional to the radii of the circle containing N min events (Nuannin & Kulhánek, 2012;Prasad & Singh, 2015).In our study we have taken N min (minimum number of events) equal to 20.In the present study the area under investigation shows a resolution of 20-100 km for the epicentral region, except for 1904 (M b ~7.6) earthquake which is slightly towards on higher sides (~40-150 km) of radii (Figure 3b).The region located to the north of the latitude 28° and west of longitude 94° shows a resolution of about 120-200 km, due to the low seismicity of the area.The resolution of the map decreases as we move away from the epicentre of the large earthquakes (Prasad & Singh, 2015), which can be seen from the prepared map (Figure 3b).

Temporal variation of b-value
The average b-values for three main regions of interest, the Shillong-Mikir Plateau, the Arunachal Himalaya, and the Assam-Foredeep were estimated for two time periods, the first period includes all events that occurred until 1999 and the second period includes events that occurred after 2000 (01.01.2000) till the year 2022.The corresponding average b-value are listed in Table 2.For Assam-Foredeep, the average b-value is presented only for the time period 1900-2022 period due to the low seismicity in this region.Hence, a comparative study for the two time periods could not be possible because of the less seismic events in this time period.
From Table 2, it can be seen that the b-value decreases significantly for the second period for all of the three regions.It may be due to the increasing in seismicity in the region of interests.The Figure 4 shows the temporal variation of b-value (b(t)) for the three regions.However, there are few numbers of major earthquakes (M b � 6) occurred in these three regions in the past.The occurrence time of large earthquakes (M b � 5) and the major drop in b-value are shown by arrow marks (Figure 4).It can be observed from the figure that there is a noticeable amount of decrease in b-value during such large earthquakes, after which the value gets increased significantly.
For Shillong-Mikir Plateau, a slight declination of the curve observed during the year 1900, which is due to the occurrence of many earthquakes (M b � 5) during that nearby period.In the next half a significant drop of b-value observed during the period 2006-2007.It may be due to the occurrence of a large earthquake of magnitude 5.5 in the year 2006 followed by a cluster of earthquakes (M b � 4:5) during that time period.The region once more experienced some significant earthquakes (M b � 5) in the years 2012, 2013, 2015 and 2016, which might be the cause of the rapid variation of 'b' between the years 2012 and 2017.The Shillong-Mikir Plateau region had experienced a large earthquake (M b ~6.2) after a long gap in 2021.It can be seen from the figure (Figure 4b) that the curve decreases after 2018 which indicates building up of high stress in this region.A very significant variation of b-value from (~0.6-1.4 units) can be observed in the region.The earthquakes (M b � 5:5) which occurred in the region Shillong-Mikir Plateau are listed in Table 3.
The temporal variation of b-value for Arunachal Himalaya also shows a significant anomaly (~0.6-1.5   moderate to large earthquakes in this region.The corresponding figures which show the declination of b-value prior to such large events are presented in figures 4c and 4d.The variation of b-value is very less for the region, Assam-Foredeep.No significant earthquakes (M b � 5:5) can be seen in the region during the study period.There is a very minimal amount of b-value anomaly (~0.5-0.7 units) and a very small amount of drop in b-value can be seen in this region (Figure 4e), which may be due to comparatively less seismic activities in this region since several years.

Spatial variations in seismicity and hazard estimation
The overall b-value map is prepared using the available database (Figure 3a), which shows a variation ranging from~0.4 to 2.4 units.The region North-East India had experienced many large earthquakes (M � 6) from time to time, which can be observed from the Figure 1.Hence, the b-value estimates the stress pattern and the hazard analysis on the basis of the earthquake occurrence in this region with respect to a spatial-temporal domain.
To make the analysis more precise, b-value map for the selected three regions of interest is prepared separately.Largest events (M b � 6) occurred during the study period, are marked with star (Figure 1).For the three regions: Shillong-Mikir Plateau, Arunachal Himalaya and Assam-Foredeep, the largest events occurred are listed in Tables 3 and 4.However, there is no large event (M b � 5:5) can be seen in the region Assam-Foredeep.From the Figure 5, it can be seen that all large events occur in the low b-value areas (b < 1; dark blue/blue colour), ranging from~0.5-1.It is important to note that the average b-value calculated for each case are different (Figure 5), where the same colour index is used to prepare the spatial distribution map for an easier interpretation purpose.The spatial distribution for the corresponding three regions is estimated for two time periods, as is the case with the temporal distribution.Events that took place prior to 1999 are included in the first period, while the events that occurred after 2000 and up to 2022 are included in the second period.The time period is divided only for the Shillong-Mikir plateau and Arunachal Himalaya List of large earthquakes (M ≥ 5.5) in the region Arunachal Himalaya.As temporal variation of 'b-value' varies depending on large earthquakes, list of large earthquakes is given.
region, whereas due to lower seismicity, the spatial distribution of b-value for Assam-Foredeep region is calculated only for the time period 1900-2022.
In Shillong-Mikir Plateau, the major changes in b-value observed in the two period roughly in between longitude 90-92.5°.The b-value map for second period preceding the earthquake (M b ~6.2, 2021) shows a low value (Figure 5b), indicating the increase of accumulation of stress in the region.Also, during the second period the region had experienced many large events (M b � 5:5) (Table 3), which can be observed in the low b-value areas, (blue colour) in figures as precondition of stress patterns before large earthquakes.
The region Arunachal Himalaya also shows an observable anomaly in the b-value between longitudes 95.5° and 98° and latitudes 28-29.5°.In the second period, the region had witnessed a large earthquake (M b ~6) in the year of 2005.The area near the epicentral region shows a low b value (Figure 5d), which indicates the accumulation of high stress in this region.No large earthquakes in the area of high b-value can be seen in this region, (Figures 5c and 5d).
No significant variation of b-value can be observed for the region Assam-Foredeep.No such large events (M b � 6) can be found in the region during the study period.The seismicity of the region is quiet low for a long time as compared to other two regions.The spatial distribution map of b-value shows that the majority of the region's areas are associated with low b-value ranging roughly from 0.5 to 1.0 (Figure 5e), which might be an indication of large-scale accumulation of stress in this region for quite a long time period.The spatial map for b-value, representing the lowest b-values (dark blue colour), suggests a possibility of future larger events/ earthquakes in this particular region associated with a large accumulation of stress energy.

Larger earthquake scenario on temporal distribution of b and hazard estimation
To strictly observe the variations in seismicity in the NER, specific research for the b-value concerning large earthquakes that occurred in the region is represented here.Analysis of earthquakes that occurred prior to 1980 is impossible due to the discontinuous nature of data.The analysis is performed only for the Shillong-Mikir plateau and the Arunachal Himalaya region, because of their high seismicity compared to the Assam-Foredeep region.For these two regions, we have selected some circular sub-regions (Table 5 & Figure 1) with a specific radius (120 km) from a central coordinate.While selecting the central coordinate and the sub-regions it was taken into consideration that, maximum number of earthquake data enters  into the computation of the b-value.Table 5 contains information on the earthquake database, including the central coordinate and the observed results, where only large earthquakes (M b � 5) were taken into consideration (Figure 6).Corresponding temporal distribution maps of the calculation (b-value variation) are shown in Figure 6, and the results are summarised in Table 5.Interestingly, the b-value dropped in the range of 0.016-0.7 units (Table 5) prior to the major events, corresponding to a significant increase in tension/ stress levels for the occurrence of large events.So, by measuring the temporal variations in the b-value, we can lead to characterise the hazard analysis or advent of major seismicity of a particular region of interest.
The observations manifest a drastic drop in the b-value at the time of occurrence of large events.
To observe the temporal variation of b-value for the earthquakes occurring within the Shillong-Mikir plateau region, we have chosen two coordinates as the central location of two circular sub-regions as; S region 1 (91°E, 26°N) and S region 2 (93°E, 26.10°N) (Figure 1).The corresponding results were summarised in Table 5.The observed results for S region 1 show a drastic drop in the b-value prior to major events M b � 5 ð Þ.The first drop of 'b' is observed between the periods 1986 to 1993 (Δb ~ 0.016 unit) when two large earthquakes (M b ~5 and M b ~5.1) occurred in the year 1992 (Figure 6a).After the occurrence of these events, there is a gap of about 13 years, during which the region had not gone through any major events and the corresponding b-value rose to a maximum value of 1.4 units during 2002-2003.This suggests a less tension scenario/period which has lasted for about 13 years after the two major earthquakes of 1992, as majority of the stress was already released in these events.After reaching the peak, it began to decrease once more (Δb ~ 0.25 unit) until 2006 (from 2003), when S region 1 was again shocked by another large earthquake of magnitude 5.5.The obtained results show a fluctuation of 'b' after 2010, which underlines the region's instability.During the period 2005 to 2006 the region had undergone two major events (M b ~5.5 and M b ~5) which leads to a major drop of b-value (Δb) roughly around 0.7 units.
The S region 2 (Figure 1), which is located very near the Kopili fault zone, has a notable amount of b-value anomalies.The Kopili fault is one of the most seismically active lineaments in the North-Eastern region.The obtained results for S region 2 show two significant deviations of 'b' from 1983 to 1991 and 2003 to 2012 (Δb ~ 0.6 and 0.65 units, respectively), supporting the tectonically active nature of the Kopili fault zone.The region experienced a series of earthquakes with magnitudes ranging from 5 to 5.4 between the period 1983 and 1991.The results reveal a steep decline in the value 'b' starting in the year 1982 when the area was hit by three large earthquakes (M b ~5.2, 5, and 5.3) during the period 1983 and 1984.It reached a low value again (~0.95 units) when a cluster of earthquakes occurred between the period 1986 and 1987.This decline of 'b' continued until it reached another minimum value (~0.9 units) when a cluster of earthquakes occurred in between 1989 -1991 (Figure 6b).After the occurrence of these events, the value 'b' rose exponentially, reaching a maximum value of 1.34 units, since there were no such significant events observed in the area during this period.This embarks the significant loss in the stress patterns after these series of earthquakes prior to 1996, which have released a major stress accumulated in this region responsible for these large events.After reaching its peak value, it started to fall once more until two significant events with magnitudes of 5 and 5.3 shocked the area in 1998 and 1999 corresponding to a drop of 'b' (Δb) by 0.21 unit.Afterwards, again an increase till the year 2003 has been seen, reaching a new peak of 1.26 units, following which there is a gradual decline of 'b' is observed that approached a minimum value (~0.61 unit) in the year 2012, indicating an increase in tectonic stress in the region.This led to again two large earthquakes of magnitudes 5.2 and 5.4 shocked the region in 2012, followed by a significant event (M b ~5) in 2009, corresponds to a significant drop in 'b' (Δb ~ 0.65 units).
To analyse the changes in b-value for the Arunachal Himalaya region within the time window considered, two sub-regions were selected (S region 3 & S region 4) (Figure 1) whose central point lies at the coordinates 92.52°E, 27.80°N and 95.50°E, 29.00°N.The temporal distribution map of b-value variation for the S region 3 (Figure 6c) shows three major deviations of 'b' (Δb) ranging from 0.3 to 0.51 units.During the periods 1985-1991, the first deviation is observed (Figure 6c) in the year 1985 where a cluster of four earthquakes with magnitudes ranging from M b ~5 to 5.7 occurred, causing the value of 'b' to fall from its peak value of 1.4 units.Following two more significant events (M b ~5 and 5.1) in the years 1987 and 1989, this declination continued and reached a minimum value of 1.1 units (Δb ~ 0.3 units).In the years, 1993 to 1998, a series of earthquakes with magnitudes of M b ~5 to M b ~5.5 jolted the area once again.'Figure 6c' shows how the value 'b' rose dramatically after 2013 and peaked at 1.22 units in 1993.When the region was again hit by a series of three earthquakes between the years 2000 and 2001, it corresponds to a declination in b-value once more, reaching a low value of 0.82 units in 1999 (M ~ 5.1, 5.2 &5.3).After these events, there was a 10-year gap during which no more significant events were seen, and the related b-value grows gradually until the year 2010, suggesting a low-stress regime and undergoing the accumulation period of stress.After the peak of 2010 (1.12 units), there was a sharp decline of 'b' that significantly manifests the 2012 event (M b ~5.2), reaching a minimum value of 0.61 units (Δb ~ 0.51) in 2013.
The second sub-region of Arunachal Himalaya (S region 4), likewise exhibits abrupt fluctuations in b-value (Figure 6d).The area has occasionally been the epicentre of numerous large earthquakes.A series of nine earthquakes with a magnitude of 5 to 5.5 struck the area during the years 1982 and 1988, triggering a significant drop of 'b' (Δb) up to 0.32 units.This declination was maintained following two clusters of earthquakes that occurred in the years 1992 and 1993, with magnitudes ranging from 5 to 5.6.It then dropped to a minimum value of 0.6 units (Δb ~ 0.48 units) until another significant event occurred in 1996 (M b ~5).Following the occurrence of these events, the b-value had a dramatic increase, peaking at 0.84 units in 2000.It again began to fall gradually during a series of earthquakes that occurred between 2002 and 2007, with magnitudes ranging from 5 to 6 and attaining a minimum value of 0.6 units (Δb ~ 0.24 units).When a series of five earthquakes with magnitudes ranging from 5 to 5.2 rocked the area in 2013, further declination 'b' (Δb) from that point caused it to fall even more, by 0.07 units.This abruptly manifests the significant loss in strain patterns after a series of major earthquakes.A large part of stress, which was stored for a gap of around 10 years was released in a series of events, not with a single stroke.After the serial advent of these cluster of earthquakes, the regime tries to gain the gradual stress pattern again, and the b-values keep gradually increasing again as suggested by the figure (Figure 6d).

Systematic discretisation of small earthquake clusters and coupling to 2019 Arunachal large earthquake (M b ~5.7)
In the present study, we try to establish a symptomatic discretisation of small earthquake clusters and relate their coupling with a major earthquake event in the north eastern region.A recent study carried out in the SW corner of Turkey (Toker, 2021) has revealed a significant coupling of clusters of events, which progressively contribute towards the outburst of a larger event.Our present study stressed upon this new ideology and fortunately mapped that the large M b ~5.7 (2019) earthquake in the Arunachal Himalaya is a consequence of a series of events, related to symptomatic clustering of events since 2015.The current findings are comparable to those in Toker (2021) in which the author used a micro-seismic cluster coupling procedure to analyse the 2017 Bodrum earthquake (M w ~6.6).The study identified several relocated events within the study area nearby SW corner of Turkey, which were then statistically analysed in both spatial and temporal forum to establish a symptomatic relation with the major Bodrum event of 2017.The term 'relocated events' refers to events that are repeated in each cluster and have the same seismic parameters (same hypo-central depths or focal mechanism solutions or similar magnitudes).The temporal and spatial distribution of these micro-seismic clusters were evaluated following the 26 November 2012 Bozburun mainshock (Mw 4.8) and the aftershocks thereafter in Bozburun, Ula, and Bodrum areas of the gulf (the 16 May 2013 Ula aftershock (Mw 4.6), the 25 March 2014 Bozburun aftershock (M w ~4.0), and the 1 May 2014 Bodrum aftershock (M w ~4.0)).The study considered the 2012 Bozburun earthquake (M w ~4.8) to be the first main shock and identified five clusters of events in the study area and their interaction was symptomatically established to evaluate the 2017 main shock of M w ~. 6.6 in Bodrum.The analytical model is closely related to Toker (2021) interaction coupling model, known as the 'chain reaction model', which is applicable to multiple-shock events as well as event clusters.However, there is a detectable connection over large distances, implying that the interaction extends even further and for a longer period.
We have attempted to analyse the interactions between micro-earthquake clusters in the Arunachal Himalayan region for the first time.An E-W crosssection of the Arunachal Himalaya region (Figure 1) is established and depth distribution of the events are mapped along this particular profile.The depth section reveals that the seismicity of the area extends up to a depth of 80 km.In a time, frame of 2014-2020, we have clearly noticed three major micro-seismic clusters whose epicentral locations and depth distribution are very close to one another (Figure 7).To evaluate the interactions of these clusters, we have closely analysed the clusters related to the event with M b ~4.8 (2015) and statistically established the relation of the clusters associated with this event towards the occurrence of a large event with M b ~5.7, 2019 through a symptomatic coupling of these cluster of events in this particular time frame.The first main shock in 2015 (M b 4.8) was associated with three large pre-shocks (M b ~4, 4.2 and 4.4; purple colour dots) and many minor shocks in the same cluster (Figure 7, represented by shadow regions).The incidence times of these pre-shock and main-shock series are fairly close.The 2015 (M b ~4.8) event had aftershocks that lasted for years, and led large shocks (4 � M b < 4:8) to occur quite frequently in the nearby area.These aftershocks had an E-W alignment with an upward migration which can be seen from the depth section (Figure 7), which lasted for almost three years.These aftershocks have three unique sequences, which are further classified as C A1 , C A2 , and C A3 based on their occurrence times.The aftershock sequence C A1 is related with the 2015 main shock (M b ~4.8), which occurred in 2015.Following the occurrence of these events, another series of events (C A2 ) is reported within few months which continued throughout 2016, with the epicentres migrated towards the east of the 2015 major shock with a few kilometres away.Continuing to this serial occurrence of events, a tight cluster of earthquakes (C A3 ) was observed near the epicentre of the 2019 event (M b ~5.7) in the time span of 2017-2019.It is significant to note that the aftershock series has a periodic migration to the east, as seen in Figure 7.These micro-cluster migrations indicate the transfer of stress energy in an E-W direction, which then accumulated in the epicentral region of the 2019 M b ~5.7 event.
This periodic build-up of stress energy led to a rapid rise in stress in this region, which is also active due to the collisional tectonic scenario, resulted in a sudden burst of energy that in the form of the 2019 event (M b 5.7).This experiment is a success example of the symptomatic interaction of micro-seismic clusters, leading to a rapid growth in the stress regime in this Arunachal Himalayan region and successfully explains the earthquake scenario associated with 2019, M b ~5.7 event.

NE hazard estimations and tectonic implications
The temporal and spatial variation of b-value were analysed for the North-Eastern region of India (between longitudes 95.5° and 98° and latitudes 28°-29.5°).The prime objective of the present study is to make a comparison of seismic b-value among the three main regions of Interest; Shillong-Mikir Plateau, Arunachal Himalaya and Assam-Foredeep.The entire database was divided into two periods for a robust analysis.The results from these two periods show observable changes in b-values for the three regions prior to and after large earthquakes.
Major changes of the b-value were observed in the region of Shillong-Mikir Plateau.Shillong plateau has a distinctive pop-up tectonics block of Northeast India due to its uplift dynamics, which are restricted by deep-rooted thrust faults (Bilham & England, 2001;Hossain et al., 2021;Mohanty & Singh, 2021, 2021).This distinct and complex plateau structure was most probably formed by a combination of N-S-directed stress from the Himalayan collisional zone and E-W-directed compressional force from the Indo-Burma subduction zone (Rao & Kumar, 1997).According to Bilham and England (2001), there is a blind reverse fault (Oldham fault) about 110 km long that dips steeply away from the Himalayas beneath the Shillong plateau.This blind reverse fault was believed to be the source of the 1897 great earthquake, which caused pop-up deformation of around 11 m of the plateau (Mohanty & Singh, 2021).The southern part of the plateau is defined by the Dauki fault, which is responsible for a portion of the shortening between the Himalayas and India (Bilham & England, 2001;Singh, Eken, et al., 2016).The plateau's uplift began in the mid-Cretaceous and reached a maximum during the Mid-Miocene epoch, which was followed by the India-Eurasia collision (Mitra et al., 2005).Its current elevation is around 1 kilometre (Bilham & England, 2001) and it exhibits a small positive bouguer gravity anomaly (Das Gupta & Biswas, 2000).The complex dynamics of the region makes the plateau seismically active and historically it has gone through many devastating and great earthquakes.The average estimated b-value for the region show a very low value (~0.4).The temporal variation in b is also in agreement in almost no such variations in the patterns in recent past.There is a high possibility of large-scale tectonic stress building up within the region, posing a high risk of future large earthquakes.
There are also noticeable b-value anomalies for the Arunachal Himalaya region.A significant drop of b-value was observed (~0.52 unit, (Table 2)) during the second time period .The Arunachal Himalaya encompasses the eastern Himalayan syntaxis and is located between longitudes 91° 30 E and 96° 00 E. This northeast region of the Indian plate is surrounded by major thrust zones (MBT -MCT to the north, Lohit -Mishmi thrusts to the northeast, Naga, Disang and Eastern Boundary thrusts to the east, front thrusts of the Arakan Yoma Belt to the southeast) (C.Singh et al., 2015;Singh, Eken, et al., 2016).In comparison to the central and western Himalayas, the eastern Himalayan orogeny has undergone significant crustal shortening (Yin, 2006).Because of the rapid rate of crustal shortening and tectonic deformation, the entire region attains a more activity in seismicity (C.Singh et al., 2015).The inversion of stress tensors from earthquake focal mechanism solutions indicates two broad dominating stress directions in the Arunachal Himalaya, ranging from NNW-SSE in the western and central regions to N-S in the eastern part (Angelier & Baruah, 2009).The obtained results of b-value from the current study also suggest the huge accumulation of tectonic stress beneath the region.Interestingly, for the last 6 years the strain pattern is gradually attaining the increment without any major fluctuations and may be a close signal for a major seismicity, as evidenced by temporal patterns of b-value in this region in the recent and further past periods.The Indo-Eurasian collisional tectonics with a significant crustal shortening is attributed towards this large scale stress accumulation in a temporal scale and may be a snoozing alarm for near future for a large event.
The region 'Assam gap' in the upper Assam valley (Khattri et al., 1983) of North-East India has experienced essentially no seismic activity for the past 60 years.The seismic activities in the region are quiet low.The unavailability of the earthquake data during the study period limits the stress analysis or the estimation of b-value for the region in the present study.However, the obtained results from the spatial distribution of b-value with the available data show that the majority of the region's areas are associated with low b-value (~0.5-1.0 units, blue colour).From this study, it can be inferred that the Assam Foredeep region might have accumulated a sufficient amount of stress for a long time, which may lead towards an intense and larger earthquake event in a near future.Again, jawed in between collisional tectonics towards north, enforced stress from Indo-Burmese Range from east, northward push from Bengal basin and overall absolute plate motion related strain guided by asthenosphere drag of Indian plate towards NE direction, really makes this geological entity a most complex tectonic domain in northeast Indian subcontinent.Hence, a complex strain pattern development, assisted by a near 60 years gap of major seismicity as suggested by lower b-value scenario, is indeed an alarming condition and a serious seismic issue of the present day, leads to a need of the hour research in a broader scale.

Conclusions
The frequency-magnitude analysis and systematic variations in b-values on a temporal and spatial scale allow us to decipher and understand the present seismic hazard scenario of the whole NER of the Indian subcontinent.These analyses allow us to understand the present stress regime, creation of resolution scale map and by logical reasoning the continuous variations in seismicity, herewith b-values, since the very past in a continuous scale.Further, the systematic spatial distributions of frequency-magnitude relations, allow us to clearly understand the hazard potential regions in the NER.Taking into consideration of the very complex tectonics and the critical geodynamic processes beneath the NER, along with the present scenario of stress patterns, this region is all set for large seismicity in near future.The need of the present hour is to analyse the complex geophysical processes with high resolution models to estimate the proper stress patterns and levels.A further study on characterisation of the most seismic potential zones, for a better preparedness in the near future is a real need of the hour.

Figure 1 .
Figure 1.Seismicity map of study area, representing three major regions (solid boundaries) of study; Shillong-Mikir Hills, Arunachal Himalaya and Assam Foredeep, where the probabilistic hazard analysis is made.The major geological structures (faults & folds) are represented by black dotted lines with abbreviations represents as.MBT: Main Boundary Thrust, MCT: Main Central thrust, MFT: Main Frontal Thrust, ITSZ: Indus-Tsangpo Suture Zone, KF: Kopili Fault, DF: Dudhnoi Fault, DAUKI.F: Dauki Fault, BF: Brahmaputra Fault, CMF: Churachandpur Mao Fault, DF: Dapsi Thrust, Du.F, Dudhnoi Fault, OF: Oldham Fault, BS: Barapani Shear Zone, LT: Lohit Thrust, MT, Mishmi Thrust, and WT: Walong Thrust.Circular sub-regions selected within the Shillong-Mikir Plateau (S region 1, S region 2) and Arunachal Himalaya (S region 3 and S region 4) for temporal distribution analysis of b-value are shown by red circles.The dashed line (black colour) represents the E-W cross-section (AB) drawn in the Arunachal Himalaya region for the microseismic clusters' interaction analysis.
shows a significant b-value anomalies in the entire study area.Epicenters of large earthquakes (M b � 6) are shown by yellow star marks in the map.There are a number of large earthquakes (M b � 6), found in various parts of the study area (Table 1), which are associated with the areas of low b-value, (low b-value areas are shown by blue colour in the map (Figure

Figure 2 .
Figure 2. Frequency-magnitude Distribution (FMD) of earthquakes database for the regions (a) Shillong-Mikir Plateau, (b) Arunachal Himalaya, and (c) Assam Foredeep.The straight line (red colour) is the best fit line by maximum likelihood analysis of events.The threshold magnitude, M c , is shown by the triangle.The average b-value is shown with the standard deviation.

Figure 3 .
Figure 3. a.The spatial distribution of b-value for the entire study area; b.The resolution map, the area with high resolution is shown by blue color.Epicenters of large earthquakes (M b � 6) are shown by yellow star marks.
units), as the region had undergone some major earthquakes in the past.The list of earthquakes (M b � 5:5) are given in Table4.An observable drop in the b-value was found in the region of Arunachal Himalaya for both time periods (Figure4c & 4d).In the first half (Figure4c) the first drop of the value 'b' was found between the time period 1982 to 1983; which may be due to the occurrence of a series of earthquakes (M b � 5) during that period of time.This decline of 'b' continued, and attained a minimum value when the region was again hit by some large earthquakes (M b � 5:5) in the year 1985 (Figure4c).During the next half when the region was once more jolted by a major event (M b 6) in 2005, an observable drop of b-value was noticed.Between the time periods 2012 and 2019, the region had undergone many earthquakes (M b � 5) causing an undulation in the b-value which can be seen in the figure (Figure4d).The Arunachal Himalaya region, which is located in the eastern part of the Himalaya is seismically very active.The accumulated stress from crustal shortening frequently releases huge amounts of energy, resulting in

Figure 4 .
Figure 4. Temporal variation of b-value for three regions; Shillong-Mikir Plateau, Arunachal Himalaya and Assam-Foredeep.Left side panel of the figure (a and c) shows the distributions for b-value for the period 1900-1999 and the right-side panel with figures (b and d) shows the distribution for the period 2000-2022 for the Shillong and Arunachal region, respectively.Due to low seismicity in the Assam-Foredeep region, temporal distribution is evaluated only for the period 1900-2022, without any divisions in time periods.The major drop in b-values are shown by arrow marks (red colour).

Figure 5 .
Figure 5. Spatial variation of b-values for the three regions; Shillong-Mikir Plateau, Arunachal Himalaya and Assam-Foredeep.The distributions for b-value for the periods 1900-1999 are shown in the left side panels of the figure (Figures 5a and 5c), whereas the distributions for the periods 2000-2022 are shown in the right-side panels of the figure (Figs.5b and 5d), respectively, for the Shillong-Mikir Plateau and Arunachal Himalaya regions.The temporal distribution for Assam Foredeep region is evaluated only for the period 1900-2022, with no divisions in time period, because of minimal seismic activities in the region.

Figure 6 .
Figure 6.Scenario of temporal variations of large events, where the analysis is performed around four coordinates within the study area and then determined the temporal variations of b-value for the events within a radius of ~120 km range.Figures (a) & (b) show the temporal distribution for Shillong-Mikir Plateau region (S region 1 and S region 2) and figures (c) & (d) are for Arunachal Himalaya region (S region 3 and S region 4), respectively.Large events (M � 5) are shown by star marks (yellow colour).The respective peak values of 'b' are marked by inverted triangles and the areas associated with major deviation of b-value are shown by red circle.The corresponding changes in b-values (δb) of clusters are marked by arrows.

Figure 7 .
Figure 7.An E-W Event-depth section over the Arunachal Himalaya region, across the profile AB (Figure 1).The red dot corresponds to the epicentre of 2015 (M b ~ 4.8) event (C M, the main event of a cluster) in the west.The yellow dot corresponds to the epicentre of the 2019 (M b 5.7) event in the east, which is the treated as the major event as a consequence of the cluster of previous events.The micro-seismic clusters under consideration are shown by rectangular shadow and the corresponding foreshocks (C F ) and aftershocks (C A1 , C A2 , C A3 ) are shown by purple and green colour, respectively.Notice three aftershock sequences that have moved from E-W direction, which took place in different time frames: C A1 (during 2015), C A2 (during 2016), and C A3 (from 2017 to 2019).

Table 1 .
List of large earthquakes (M ≥ 6) including some historical events and their epicenters (Lat, Lon) in the study area are presented below.

Table 2 .
'b-value for the consecutive two periods; first periodand second period.

Table 3 .
List of large earthquakes (M ≥ 5.5) in the region of Shillong-Mikir Plateau.As temporal variation of 'b-value' changes significantly due to large earthquakes, the scenario of these large events are represented below.

Table 4 .
List of large earthquakes (M ≥ 5.5) in the Arunachal Himalaya region.

Table 5 .
Temporal changes of b-value in four sub-regions of interest.