Kpseis 1

Physics and Chemistry of the Earth 123 (2021) 103029
Contents lists available at ScienceDirect
Physics and Chemistry of the Earth

journal homepage: www.elsevier.com/locate/pce
A hybrid probabilistic seismic hazard model for Northeast India and Bhutan
combining distributed seismicity and finite faults
Federica Ghione a, d, *, Valerio Poggi b, Conrad Lindholm c
a
NORSAR, Gunnar Randers Vei 15, 2007, Kjeller, Norway
b
Seismological Research Center, National Institute of Oceanography and Applied Geophysics (OGS), Via Treviso, 55, 33100, Udine, Italy
c
SeismoConsult, Fjellhamar, Norway
d
Department of Geosciences, University of Oslo, Norway
A R T I C L E I N F O A B S T R A C T
Keywords: In Probabilistic Seismic Hazard Analysis (PSHA), a widely used approach to model earthquake sources consists of
Northeast India using homogenous source zones. This approach suffers the limitation of assuming that the observed seismicity
Bhutan can occur anywhere with same probability over a specific area, which might lead to the potential undervaluation
Probabilistic seismic hazard analysis
of the predicted ground motion level due to an effect of smearing of seismic potential. To compensate that, a
Distributed seismicity
Finite faults
hybrid model is used, accounting both for distributed seismicity and localized seismogenic structures.
In this study, we perform PSHA for Northeast India and Bhutan, that are the most seismically hazardous re
gions on the planet. The region was partitioned into seismogenic source zones of supposedly homogeneous
seismic potential and seismotectonic characteristics. Earthquake recurrence parameters for each zone were ob
tained from direct magnitude-frequency analysis on a precompiled global catalogue. Seismogenic faults are
added to the model by converting slip rates from GPS velocity data to seismic activity. Complementary infor
mation was derived from the analysis of moment tensor solutions of large events and from a detailed literature
review.
Using this hybrid model, seismic hazard was calculated for a region bounded by lat/long 24.0◦ -28.8◦ N/88.0◦ -
94.5◦ E. Calculations were performed for Peak Ground Acceleration (PGA) and several Spectral Acceleration (SA)
periods for a Probability of Exceedance (POE) of 10% in 50 years, corresponding to 475 years return period, and
for a reference rock condition with Vs = 800 m/s. The results highlight significant acceleration levels (about
0.77g at PGA) in the Arunachal Pradesh region (Northeast India), due to the presence of the Himalayan Frontal
Thrust (HFT).
1. Introduction Historically, the study region is strongly affected by earthquakes

with large magnitudes: during the past 120 years, thirteen events with
Northeast India and Bhutan are one of the most active seismotectonic magnitude above 7.0 have occurred in the study area (Baro and Kumar,
regions of the world (Baro and Kumar, 2017; Baruah et al., 2016) and 2017; Kayal, 2008; see Fig. 1). The main purpose of this work is to
over the last decades, different kinematic models have been proposed to evaluate probabilistically the level of ground motion induced by large
explain the seismotectonics of the area. The area considered for the and potentially damaging future earthquakes in the densely populated
analysis lies between latitude 23◦ N to 30◦ N and longitude 85◦ E to 98◦ E region of Northeast India and Bhutan, which is fundamental to mitigate
(Fig. 1). From the geographical point of view, the study area is bounded the impact of future events on the population and therefore to reduce the
by the Himalayas to the north, the Indo-Burma ranges/Myanmar to the seismic risk. We use a Probabilistic Seismic Hazard Analysis (PSHA)
east, Bangladesh to the southwest and the Andaman Sumatra region to approach (Cornell, 1968; Esteva, 1967, 1968), which allows to define
the southeast. The geology of this region is complex with the interaction the probability of a specified ground motion level being exceeded at a
between the active north-south convergence along the Himalaya and the site or area of interest, as generated by any earthquake source poten
east-west convergence and subduction within the Indo-Burma ranges. tially expected for the region. The way the earthquake source is
* Corresponding author. NORSAR, Gunnar Randers Vei 15, 2007, Kjeller, Norway.
E-mail addresses: [email protected] (F. Ghione), [email protected] (V. Poggi), [email protected] (C. Lindholm).
https://doi.org/10.1016/j.pce.2021.103029
Received 2 November 2020; Received in revised form 12 March 2021; Accepted 10 May 2021
Available online 17 May 2021
1474-7065/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
F. Ghione et al. Physics and Chemistry of the Earth 123 (2021) 103029
modelled, therefore, plays a major role in the analysis and has a sig Kayal, 2003; Diehl et al., 2017; England and Bilham, 2015; Goswami,
nificant impact on the final results. 2005; Hauck et al., 1998; Hetényi et al., 2016; Kayal et al., 2006; Le
The most common approach to model the earthquake source is using Roux-Mallouf et al., 2015; Marechal et al., 2016; Nandy, 2001; Stevens
a distributed seismicity approach, where the study area is subdivided in et al., 2020; Stevens and Avouac, 2015; Velasco et al., 2007; Vernant
a number of homogeneous zones (from the seismotectonics point of et al., 2014; Yin, 2006).
view), where each point is assumed to have the same probability of As a final result, we compute ground motion for 10% Probability of
generating an earthquake. The seismic productivity (activity rates) is Exceedance (POE) in 50 years (corresponding to 475 years return
calibrated from the analysis of all the available observed events (e.g. period) for three target sites of particular significance for the area:
from an earthquake catalogue) on the area, assuming a given Thimphu, Guwahati and Shillong. The obtained results can be used in
magnitude-recurrence relation. This approach is convenient when little the future to design earthquake resistant civil structures and define the
knowledge about the seismic sources is available for the region, but earthquakes scenario for deterministic risk assessment.
might be limiting in case of known active lineaments with large seis
mogenic potential. A more recent and sophisticated approach consists 2. Geology and Seismotectonics
then in the direct modelling of the main active faults of the region
(Akinci et al., 2009; Carlton et al., 2018; Poggi et al., 2020; Rivas-Me Northeast India and Bhutan are among the seismically most active
dina et al., 2018; Valentini et al., 2017; Woessner et al., 2015; Yagh regions on earth. The high seismicity is due to the north-south collision
maei-Sabegh et al., 2018). As for the case of source zones, occurrence on between the Indian and the Eurasian plate along the Himalaya front, and
a fault can be calibrated by analysing the associated historical earth the east-west subduction along the Indo-Burmese Ranges (IBR) in the
quake record that, however, might be rather limited in time for the east. The convergence of the Indian and Eurasian plates at 40–50 mm/yr
specific lineament. To compensate for this, slip rates from geological or since their initial collision between 70 and 34 million years ago resulted
geodetical analysis (e.g. from strain rates) can also be used to model in the accumulation of widespread strain in the region. About 15–20
earthquake occurrences. Nonetheless, also this approach is affected by mm/yr of the convergence is accommodated in the Himalayan frontal
several limitations. Slip rates, for example, might be affected by large arc (Vorobieva et al., 2017). The convergence along the entire Hima
uncertainty and the assumption of stationarity of the deformation rate layan arc is directed perpendicular to the arc and increases eastwards
over long time periods might not be necessarily fulfilled. from 13.3 ± 1.7 mm/yr in the west to 21.2 ± 2.0 mm/yr in the east
Previous studies on the same area have used the classic area source (Stevens and Avouac, 2015).
model to compute the PSHA (Baro et al., 2020; 2018; Bhatia et al., 1999; Scientific debate is still ongoing about the identity and correlation of
Das et al., 2016; Giardini et al., 1999; Khattri et al., 1984; Nath and stratigraphic units, location and origin of the major contacts between
Thingbaijam, 2012; NDMA, 2011; Sharma and Malik, 2006; Shivamanth the different units, and the timing of the orogenic processes in the region
et al., 2016; Walling and Mohanty, 2009; Yadav et al., 2010); a recent (Greenwood et al., 2016). There are also substantial differences in the
study from Stevens et al. (2020) has performed a PSHA based on faults geology between the western, eastern and central part of the Himalayas
locations, slip-rates and paleoseismic earthquake data for Bhutan. (Kayal, 2008). Based on the geology and the seismotectonics, we have
The purpose of the present investigation is therefore to create a divided the area into seven main provinces, which are discussed below.
hybrid model for Northeast India and Bhutan, which comprises classical
area source zones and faults sources, in order to compensate the relative 2.1. Bhutan
limitations of the two approaches and thereby to obtain a more
comprehensive representation of the seismic hazard of the area. For that, The Bhutan region is characterized by the high topographic relief of
we make extensive use of the slip rate information available in literature the Himalaya, that is 2500 km-long along Himalayan belt and it is
for local studies (Barman et al., 2017; Bilham and England, 2001; De and associated to several thrust and faults. Along the Main Himalayan Thrust
Fig. 1. The Mw homogenized earthquake catalogue (New ISC-GEM Extended) for the study region. The stars show the cities (Thimphu, Guwahati and Shillong)
where the ground motion exceedance values are computed.
2
(MHT), which is a mid-crustal décollement (basal detachment fault), the junction that joins the Indian and Eurasian plates with the northern end
Indian plate underthrusts the Eurasian plate, causing the Himalaya of the Burma platelet. The Himalayan arc shows a sharp turn of about
mountain growth and associated deformation and high seismicity. From 90◦ and meets the Indo-Burma ranges. The crustal deformation due to
north to south, the Main Central Thrust (MCT), the Main Boundary plate motions is particularly complex and this syntaxis zone is also the
Thrust (MBT) and the Main Frontal Thrust (MFT) are the prominent location of high stress concentrations. To the southeast and east of the
north dipping thrusts with a geometry that is listric to the MHT (see Assam Syntaxis within east Burma and Yunnan (China), the relative
Fig. 2). The MFT, also known as Himalayan Frontal Thrust (HFT), is motion between India and southeast Asia is dextral; to the west and
considered as the surface expression of the MHT, the outermost front north of the Syntaxis, within the Eastern Himalaya and southeast Tibet,
toward south and it runs along the Bhutan/India border (Barman et al., the motion between India and Asia is convergent. The crustal thickness
2017). While the MFT is well defined in Nepal, in Bhutan it does not estimated in this region range from 50 to 60 km. This area was the place
have a continuous and clear morphological expression because it is of the August 15, 1950 Assam earthquake (Mw 8.6): it is considered as
buried under young alluvial sediments. the manifestation of the India-Asia convergence (Seeber et al., 1981).
Along the Himalayan arc, the seismic behaviour of the Bhutan region Armijo et al. (1989) supported the right-lateral strike slip solution on the
is uncertain, because the number of significant earthquakes in Bhutan is Po Qu fault zone in southeast Tibet, which wraps around the Eastern
much lower (Gahalaut et al., 2011; Vernant et al., 2014) than in other Syntaxis and connects with the right-lateral strike-slip Sagaing fault
zones (Fig. 1), in particular to the west of Sikkim. This can be due to zone.
continental shortening being cushioned by the Shillong Plateau (SP)
150 km farther south (e.g. Bilham and England, 2001; Gahalaut et al., 2.4. Indo Burma ranges
2011). The present-day deformation pattern is constrained only by
sparse GPS measurements in western Bhutan (Vernant et al., 2014) and a The Indo-Burma ranges (IBR) is located in a complex tectonic zone
single estimate of Holocene uplift rate along the Topographic Frontal with oblique subduction at its western boundary, a dextral transform
Thrust (TFT) in central Bhutan (Berthet et al., 2014). In central Nepal fault (Sagaing fault) on the eastern boundary, the Mishmi thrust in the
and Arunachal Pradesh (east Bhutan), Holocene slip rates along the MFT north, and the Andaman Spreading Ridge (ASR) to the south. The Bur
are about 21 ± 1.5 mm/yr (Lavé and Avouac, 2000) and 23 ± 6.2 mm/yr mese arc, NS oriented and 1100 km in length and 13–14 km in width, is
(Paul Burgess et al., 2012), respectively. The occurrence of the 1897 convex westward. Different interpretations have been proposed through
earthquake near the SP attests to the accommodation of the convergence years about the evolution of the Burmese arc (Curray et al., 1979; Le
in this region and has been proposed to increase the interval between Dain et al., 1984; Mitchell and McKerrow, 1975; Nandy, 1983; Tap
great earthquakes in the Bhutan Himalaya (Bilham and England, 2001). ponnier et al., 1982; Uyeda and Kanamori, 1979), but it is generally
Gahalaut et al. (2011) also proposed that the stress shadow caused by agreed that the IBR is a subduction zone characterized by high
this earthquake may be responsible for the low seismicity rate currently seismicity.
observed in Bhutan. According with instrumental and historical records,
Bhutan experienced no major earthquake in the past 200 years (Gaha 2.5. Shillong-Mikir Massif
laut et al., 2011).
The Shillong Plateau-Mikir Massif intraplate zone is a part of the
2.2. Northeastern himalaya Indian shield and moved to the east along the Dauki fault. The Shillong
Plateau is located between the Himalayan arc to the north and the
The northeastern Himalaya is subdivided into four major tectonic Burmese arc to the east; it was created by a “pop-up tectonics” in the
units by different thrusts. From south to north (see Fig. 2): the Outer Sub- Pliocene and it is bounded by active faults (Kayal et al., 2006; Saikia
Himalaya (that lies between the MFT and the MBT), mainly consists of et al., 2017). To the south, the Plateau is limited by the ~320 km long
Neogene and Quaternary molasse sediments; the Lesser Himalaya E-W Dauki fault, that separates the Precambrian basement of the SP to
(bounded by the MBT and MCT), consists in pre-Tertiary rocks that the north and the thick Tertiary sediments of the Bengal basin to the
underlie the central Himalaya. While the MBT is clearly defined all along south (Kayal, 2008; Sharma et al., 2017). The area between the SP and
the southern margin of the Himalaya, the position of MCT is debated, the Mikir hills is called the Kopili gap and they are separated through the
particularly in the northeastern part. NW-SE Kopili lineament (300–400 km long). The Kopili fault produced
The Central Himalaya is made by metamorphic and granitoid rocks two large intraplate earthquakes, the damaging 1869 Cachar earthquake
and higher structural units to the north of MCT. In the Trans-Axial (Mw 7.4), and another one in 1943 (Mw 7.2). In January 2016 a strong
Himalayas, a tectonic contact being defined by the Tethyan/Trans-Axial intraplate earthquake of Mw 6.9 occurred at the southeast end of the
Thrust or Indus Suture Thrust (IST) puts in contact Phanerozoic sedi Kopili fault in Manipur. To the west, the Plateau is bordered by the
ments and metasediments and the crystallines. Dhubri fault, that generated the 1930 Dhubri earthquake (Mw 7.1)
(epicenter at 25.8◦ N, 90.1◦ E and ~60 km depth). The northern bound
2.3. Eastern Himalaya Syntaxis ary of the SP is not well constrained and different authors proposed
different interpretation (Bilham and England, 2001; Kayal et al., 2006;
The Eastern Himalaya Syntaxis (“Assam syntaxis”) is a complex triple Rajendran et al., 2004). For this study, according to Bilham and England
Fig. 2. Simple model of how the Indian continent underplates the Eurasian continent which gives rise to the Himalayas with the major faults Main Frontal Thrust
(MFT), Main Boundary Thrust (MBT) and the Main Central Thrust (MCT) (Bungum et al., 2017).
3
(2001) based on GPS observations, we have considered that the SPis finally a quantification of known seismicity and geology into model
bounded by two reverse faults: the north dipping Dauki fault to the south recurrence parameters used as basis for the PSHA.
and WNW-ESE trending south dipping Oldham fault at the boundary
between the Plateau and the Brahmaputra valley. The Shillong
Plateau-Mikir massif and Assam valley intraplate zone produced many 3.1. Earthquake catalogues
large earthquakes, including the great Shillong earthquake, occurred in
June 12th, 1897. It is one of the largest intra-plate earthquakes in the An earthquake catalogue is an archive where all the past events are
world, with a magnitude Ms 8.7 (revised Mw 8.1 by Bilham and England, reported with information about magnitude, time and location, hypo
2001), and it was the first Indian earthquake for which instrumental central depth and other source-specific information. The earthquake
records, recorded outside India, are available (Oldham, 1899). It caused catalogues are, regrettably, inherently heterogeneous. From the early
widespread damages over the entire plateau and in NE India, causing days of instrumental seismology at the beginning of the twentieth cen
1542 casualties. Different interpretation (Baruah et al., 2016; Bilham tury, seismological networks have undergone many changes that are
and England, 2001; Imsong et al., 2016; Martin and Szeliga, 2010; reflected in the database of earthquake records in use today. One of the
Oldham, 1899) have been proposed to define the earthquake source. The challenges is related to the variety of scales used for recording earth
scenarios based on either the Chedrang Fault or the Oldham Fault are quake magnitude. In 1964 the International Seismological Centre (ISC)
more in line with the reported damage and shaking intensity reported catalogue shows a rapid increase in the reported earthquake globally.
from the 1897 event. The ISC-GEM Extended catalogue, updated in July 2017, was used as
one of the basic sources of information since it is based on reports from
2.6. Assam valley many international seismological networks and reporting agencies
(Weatherill et al., 2016). The ISC-GEM Extended is a global homoge
The Brahmaputra valley is divided into three parts: the eastern upper nized catalogue (Weatherill et al., 2016) that represent seismicity in a
Assam valley, the central Brahmaputra valley and the western lower common and comparable scale using the moment magnitude Mw scale.
Assam valley. The Brahmaputra plain is stretched nearly 800 km in It contains data from 1900 through 2016, and in the region of interest
length with an average width of 110 km and runs NE-SW, and it is the smallest magnitude reported is Mw = 3.93. Two other catalogues
composed by Mesozoic to early Tertiary sediments that they increase the were taken into consideration: the Global Centroid Moment Tensor
seismic hazard of the region. The Brahmaputra basin formed as a result Catalogue (Global CMT) (Ekström et al., 2012) and the USGS earthquake
of upliftment and subsidence of different blocks of Precambrian crys catalogue (see Data Availability). In particular, the Global CMT cata
talline deposition. The Oligocene sequence in Assam shelf is represented logue database covers a period from January 1976 to December 2017
by 600–1000 m thick sediments. The upper Assam valley is flat, about and it was used to evaluate the dominant rupture mechanisms of the
100 km in width and 300 km in length and is NE-SW oriented (Saikia area by means of ternary diagrams (Kaverina et al., 1996) (Fig. 3).
et al., 2017). The USGS earthquake catalogue was used, together with the ISC-
GEM Extended catalogue, for analyzing the hypocentral depth distri
2.7. Bengal Basin bution of the past events and a statistical analysis was performed to
understand how the ruptures are distributed along the depths.
The Bengal basin is one of the largest basins of the world, with a There are many uncertainties that affect the catalogues. These un
thickness of up to ~15 km (Kayal, 2008). The Bengal geosyncline shrunk certainties are basically related to the magnitudes, the precise hypo
with time in consequence of the eastward subduction of the Indian plate central locations and the depths of the events (Gulia et al., 2012; Panzera
below the Andaman-Burmese arc. Several stages have been observed: et al., 2016). Another problem is linked to the short period of observa
India collision with Eurasia, uplift and erosion of the Himalaya in the tion in comparison with the recurrence time of the large events. For
Tertiary-Quaternary, transport of the eroded materials by confluent example, the observation period of earthquake activity in the study area
Ganges and Brahmaputra rivers and deposition of the sediments in the covers around 116 years and this means that the used earthquake in
newly formed Bengal geosyncline. Due to continued spreading of the formation (reliably quantifiable) does not stretch longer back in history.
Indian ocean floor, the above stages are still ongoing. An important Since the average return period of large earthquakes is assumed much
tectonic feature is the Eocene Hinge Zone (500 km long, width varies
from 25 km to 110 km), which separates the continental shelf to the west
and the geosynclinal facies to the east. Curray et al. (1982) suggested
that the Hinge line represents the boundary between the continental
crust and the young oceanic crust that extends southwards into the Bay
of Bengal, and it intersects the Dauki fault near 92◦ E. The Bengal basin
has produced two large deeper (depth ≥ 20 km) intraplate earthquakes,
one in 1918 (Mw 7.1) at the Sylhet fault, and the other in 1923 (Mw 7.0)
at the junction of the Hinge zone and the Dauki fault.
3. Seismicity information
In order to completely represent source seismicity through a recur

rence relation, the annual rate of earthquakes above a minimum
magnitude is needed. There are basically two approaches to define ac
tivity rate of a given seismic source, either through the analysis of past
seismicity from an earthquake catalogue or through structural geolog
ical data, such as faults information. The procedure to identify potential
seismic sources in the studied region comprises an evaluation of the
historical and recent instrumental seismicity data, emphasizing that
these data are the primary empirical basis for conducting the seismic Fig. 3. The dominant rupture mechanisms for the entire study area are plotted
hazard analyses. Another information source is the evaluation of the in a ternary diagram, showing the main rupture kinematics (reverse, normal,
tectonic history based on available geological data and information and strike-slip).
4
longer than 116 years, it becomes clear that the uncertainty of any (around 25% of the events have been removed).
hazard estimate is significant.
3.3. Completeness of the catalogue
3.2. Declustering procedure
The magnitude completeness is the lowest magnitude above which
In order to perform a Poissonian analysis of the earthquakes in the 100% of the events in a space-time volume are detected; it is very
catalogue, all the events must be independent from each other. Usually, important because it tells us from which years all the events above this
after a larger earthquake, several events on a short time scale occur that threshold magnitude are detected and can be included in the statistical
are related to the previous one. These events, which are related to a analysis. The magnitude completeness of an earthquake catalogue is a
parent event, are called aftershocks. Later, the time interval between factor that profoundly affects the recurrence parameters, and it is a basic
earthquakes becomes longer. For what concerns the catalogue, it is requirement for the processing of input data for seismic hazard analysis.
important to perform the decluster procedure which removes the After some tests with the Stepp (1971) algorithm, a manual procedure
aftershock (and foreshock) events that are not allowed in the statistical was selected for the characterization of the magnitude completeness
analysis and violate the Poissonian assumption (Gardner and Knopoff, based on the analysis of the obtained rates and the bins related to the
1974; Hainzl et al., 2006; Taroni and Akinci, 2021; Zhuang et al., 2002). magnitude are fairly homogeneous and growing with magnitude. The
To remove the aftershocks and create a declustered ISC-GEM catalogue, time-magnitude plot (Fig. 4) demonstrates that the ISC-GEM catalogue
many tests have been done to understand the sensitivity of the classical can be regarded as being reasonably complete for M ≥ 5.0 only from
Gardner and Knopoff (1974) method. In this method, the earthquakes 1965, because below M < 5 there is an evident irregularity in the dis
within certain time and distance windows of the mainshock, the size of tribution of events, that cannot be attributed to a problem of a simple
which are defined by the mainshock magnitude, are considered depen incompleteness of the catalogue or to a real inhomogeneity of occur
dent events and removed from the catalog. These windows are depen rences. From 1925 for M ≥ 5.5 the figure again indicates complete
dent only on the magnitude of the event and were derived for California recording. For M ≥ 6.5, the catalogue has been considered complete
earthquake. The Gardner and Knopoff (1974) algorithm resulted not from 1900. Finally, for M ≥ 7.5 it’s very difficult to say with certainty
suitable for the area investigated in this work because it was removing when it started to record; however, the recording start year has been
more than 40% of the total events, reducing the activity rates of the estimated equal to 1850. It is worthy to note that the two major im
source zones. For this reason, a criterion based on a visual selection of provements in seismicity coverage occurred in the early 1960’s (the
the cluster events has been adopted. Different tests have been performed WWSSN network) and around 2000.
to evaluate the total number of events removed applying different time
and distance windows. At the end, it has been decided to identify and 3.4. Seismogenic faults
remove the aftershocks from the catalogue listings if they satisfy the
following assumptions: The modelling of all existing active faults as independent entities can
be considered as the most accurate source model for seismic hazard
- for events with magnitude equal or greater than 6.5, the distance assessment (Rivas-Medina et al., 2018; Stevens et al., 2020; Valentini
considered for the decluster procedure was set to 70 km and the post- et al., 2019). This idea is however quite idealistic, because many times
earthquake time set to two months; earthquakes do not occur on pre-existing faults, and many faults in the
- for main events with magnitude lower than 6.5, the distance was set sub-surface are simply not known in terms of size, geometry and po
to 50 km and the time set to one month. tential. A more realistic view includes only a limited number of active
faults that have the presumed highest seismic activity. The normal
Before the decluster procedure, the total number of the events were approach for mapping the faults is followed by different methods, for
4153 in the area between latitude 23◦ N to 30◦ N and longitude 85◦ E to example field observation, morphological evidences, using aerial and
98◦ E; after applying the decluster method, the total events are 3050 satellite photos. Since it is extremely difficult to determine the correct
Fig. 4. Time-magnitude plot for the ISC-GEM catalogue for the identification of magnitude completeness thresholds.
5
fault traces and their precise position, simplified fault geometries have fault traces and their locations, the slip rates values, the dip and depth of
been adopted in this work. In this way, the computational efforts were the faults. A way to overcome the problem of the correct fault traces and
reduced, without losing accuracy in the results. The faults information their locations can be comparing field data, morphological evidences,
was taken from the Geological Survey of India (see Data Availability) aerial and satellite images and try to define a simplified geometry. The
and they were also supported by a careful literature research (Baruah variations in geometry (both strike and dip and possible listricity) can be
et al., 2016; Bilham and England, 2001; Islam et al., 2011; Kayal, 2010). accommodated by assigning the different geometries with appropriate
The following seismogenic faults (Fig. 5) have been considered: HFT weights in the model, and in this way consider several plausible alter
(Himalayan Frontal Thrust), Yadong fault (strike-slip), Tista fault natives. The solution to overcome the uncertainties related to the slip
(strike-slip), Dhubri fault (left strike-slip), Kopili fault (right strike-slip), rate could be do different tests to extract the recurrence parameters
Oldham fault (reverse), Dauki fault (reverse fault). In section 2. Geology changing each time the value of the slip rate. In this way, a sort of
and Seismotectonics, three superficial expressions of the MHT have been balance between the recurrence parameters used and the percentage of
described: the HFT, the MBT and the MCT (Fig. 2). Since the MBT and the slip rate used can be created for building a model much credible and
MCT are possibly less important from the seismological point of view realistic as possible.
(there are no relevant events related to these thrust) than the under
plating main thrust (MHT – detachment zone), in this work only the HFT 4. The seismic source model
and the MHT have been considered for the hazard calculation. In Fig. 5
only the HFT is represented because the MHT starts at about 17 km deep The hazard calculation for this work is mainly performed using two
until ~50 km. The HFT is considered as the principal superficial types of sources: the area sources and finite faults.
expression of the MHT. The most important parameter for crustal
deformation is the slip rate (in units of mm/yr); the slip rate of a fault is 4.1. Area sources
the speed with which one side of the fault moves with respect to the
other. Slip rates are today most often estimated from GPS measure The earthquake catalogue is used to build the spatial and temporal
ments, but nevertheless often remain uncertain. Moreover, it has to be occurrence model that can be used as input for seismic hazard assess
considered that not all the slip rates value recorded by GPS can generate ment, providing essentially a prediction model for a number of expected
future earthquakes, since much of this slip is accommodated and con events at different magnitude levels and in different source regions.
verted in deformation (e. g. aseismic creep; Murray et al., 2014;
Thatcher and Pollitz, 2008). The percentage of the slip rates that takes 4.1.1. Seismic zonation, maximum magnitude, depth distribution and
account for future earthquakes is called seismic coupling coefficient and moment tensors
is usually a value between the 30% and 50% of the registered data. The area sources can be described by areal regions in which earth
When the slip rates are available, algorithms based on such estimates are quakes may occur anywhere and randomly within the defined zone and
considered to represent the most viable alternative provided that the these zones are assumed to have uniform source properties in both time
derived activity is always assessed in terms of the equivalent moment and space. The zonation has the purpose of dividing the seismicity into
release. Moreover, such approach assumes that the slip rate observed distinct source zones, and subsequently evaluating the seismic potential
can be considered constant back in past times. When the slip rates are in each zone based on past seismicity. This is clearly a simplification
not available, the earthquake activity on a fault has to be assessed by with respect to the more continuous seismicity distribution. A main goal
assigning to the fault a certain proportion of the seismic activity which is applied in a zonation process is to represent the seismicity and the tec
assessed for the region containing the fault. tonics as balanced as possible. In making the zonation, the basic prin
Regarding the faults, the main uncertainties are related to the correct ciples were used as guidelines: each zone should be large enough to
Fig. 5. The study area for hazard calculation is marked by a black dashed square (24–28.8⁰ N to 88–94.5⁰ E); and in the figure the main seismogenic faults of the area
are shown as modelled with a simplified geometry: in red the reverse faults, in green the strike-slip faults, in blue the normal fault. (For interpretation of the ref
erences to colour in this figure legend, the reader is referred to the Web version of this article.)
6
allow for a reasonably stable assessment of recurrence parameters and a probably very close to the maximum awaited. For this reason, the
homogeneous source zone should encompass regions of supposedly model Mmax has been put equal to observed Mmax 8.6;
uniform seismic potential and seismotectonic characteristics; the zones - Zone 6, called here India region, is represented by a low seismicity
should cover all areas where the seismicity could have some influence on distribution and the Mmax is 7.3;
the seismic hazard, which normally means 200–300 km around the site, - Zone 7 or Nepal region, is marked by the west part of the Himalayan
depending on activity level (this because also events that occur outside Frontal Thrust. The 1934 earthquake recorded a Mw 8 so the Mmax
the study area can be relevant for the hazard level of the region of in used is 8.5. In the area, half of the events indicate a reverse kine
terest); the zonation should, if required, allow for possible regional matic, and the other half is marked by both normal and strike-slip
differences in seismogenic conditions (focal depths, maximum magni mechanisms;
tudes and faulting mechanisms); the zonation should be consistent with - Zone 8, called Bhutan region, shows a reverse mechanism confirmed
the regional geology and tectonics. by the central part of the HFT. In this case the Mmax used is 7.0;
Based on the above tectonic knowledge, structural geology, main - Zone 9, named Arunachal Pradesh, is characterized by a reverse
active faults and earthquake distribution the division into ten distinct mechanism and through the east part of the HFT. The Mmax is 7.8;
seismic zones was made (Fig. 6). The division was based also on the data - Zone 10 comprises the China area and shows a low seismicity dis
processing of the whole catalogue regarding the seismicity, depth and tribution (Mmax used 6.5) with a normal kinematic.
the previous studies (Baruah et al., 2016; Bilham and England, 2001;
Islam et al., 2011; Kayal, 2010). The maximum magnitudes (Mmax) The hypocentral depth distribution has been analyzed using two
were defined, using the ISC-GEM Extended catalogue, for each of the different catalogues: the ISC-GEM Extended and the USGS earthquake
zones defined through a combination of the observed maximum catalogues (Fig. 7).
magnitude for each zone and the evaluated tectonic potential, where a
simple and commonly applied rule is that Mmax should be set to about 4.1.2. Quantification of the earthquake recurrence
0.5 magnitude units above the largest value. In each source zone, the temporal occurrence of events with
increasing magnitude is assumed to follow a power-law behaviour,
- Zone 1 represents largely the Burmese Arc subduction zone. It is a which we model using a double truncated Gutenberg-Richter recurrence
very active seismic region and the Mmax is been fixed equal to 8.2. In relation (or magnitude-frequency distribution, MFD) (Gutenberg and
this zone about half of the events show a reverse kinematic, and the Richter, 1956). Lower truncation is arbitrarily assigned to Mw 4.5
other half a strike-slip mechanism; (lowest magnitude threshold considered capable of generating damage)
- Zone 2 corresponds to the Bengal Basin with a Mmax equals to 7.7. In for all zones. Upper truncation is defined as the magnitude of the largest
this case, there are less events but they show both reverse and strike- earthquake assumed possible for an area (explained in the previous
slip kinematics; section and in Table 2). The ISG-GEM Extended catalogue was used to
- Zone 3 or the Shillong area, where the 1930 Dhubri earthquake compute the b-value. Since the b-value heavily affects the hazard
occurred with a Mw 7.1, and in this case the Mmax used is 7.6; calculation, and the uncertainty on b in some zones is very high, due to
- Zone 4, called the Mikir Massif region, shows a strike-slip behaviour the small number of events available in each zone, we decided to set this
and during 1943, the Kopili earthquake was recorded with a Mw 7.2. value equal to the b-value obtained for the total area. In a first round, the
So the Mmax used for the model is equal to 7.7; b-value has been computed for the total area using the linear regression
- Zone 5, named also Eastern Himalaya Syntaxis, is note for the great method, and then the rates are obtained for each zone by using the fixed
1950 Assam earthquake with a Mw 8.6. In this zone, the magnitude b-value (Table 2).
of the largest earthquake recorded is above 8.5, that is a value for
which an increase of 0.5 would lead to a Mmax that should be not
realistic. In this case, the catalogue has registered an event that is
Fig. 6. The zonation of the study area, taking into consideration the earthquakes distribution, the main active faults and the geological framework of the area.
7
Fig. 7. On the left, the magnitude-frequency

distributions and recurrence regressions for
the determination of the recurrence values
for the entire study area. White squares and
red dots are respectively the observed in
cremental and cumulative occurrence rates,
while the grey histogram and the red line
represent the incremental and cumulative
rates from the inverted Gutenberg–Richter
relation. It must be noted that the width of
noncumulative magnitude bins is not uni
form, as this is not a requirement of the
fitting method. On the right, the hypocentral
depth distribution is shown for the whole
study area with the filter for depths 10 and 33 km. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this
article.)
Table 1
In the table, the fault geometries are summarized (rectangular fault planes), the original length, the length of each fault after the segmentation procedure and the aspect
ratio for the segmented parts.
Fault Width Original Length Number of Length after the segments Aspect Slip rate from GPS data Slip rate used (mm/
(km) segments (km) ratio (mm/yr) yr)
Dauki 40 300 3 100 2.5 5.00 2.00

Dhubri 25 165 3 55 2.2 0.80 0.32
Kopili 50 200 2 100 2.0 2.50 1.00
Oldham 40 88 1 88 2.2 5.00 2.00
Tista 40 320 4 80 2.0 1.50 0.60
Yadong 25 125 2 62.5 2.5 0.50 0.20
HFT- 70 350 2 175 2.5 17.00 6.80
central
HFT-east 60 240 2 120 2.0 21.20 8.48
HFT-west 60 300 2 150 2.5 19.40 7.76
4.2. Fault sources Nandy, 2001; Stevens et al., 2020; Stevens and Avouac, 2015; Velasco
et al., 2007; Vernant et al., 2014; Yin, 2006). However, many times the
The main seismogenic faults of the area were considered and they value recorded by GPS is converted directly without considering that not
have been treated and modelled in a simplified way to perform the all the slip rates can generate future earthquakes, but much of this slip is
hazard calculation (Fig. 5). The information about faults focal mecha accommodated and converted in deformation (Murray et al., 2014;
nisms are extracted from the literature review (Baruah et al., 2016; Thatcher and Pollitz, 2008). After different sensitivity tests, the 40% of
Bilham and England, 2001; Islam et al., 2011; Kayal, 2010). In most the GPS recorded slip rate of each fault has been used (Table 1).
cases the dip and the depth information are not available, and the values The estimation of activity rates (N-values) involves the seismic
used refer to extrapolated data from figures or from a geological inter moment, M0 , the rigidity or shear modulus, μ, the total average
pretation. For a long fault or fault system, some segmentation is usually displacement (or slip) across the fault, D, the annual slip or slip rate, S,
needed when evaluating the potential seismicity, or the recurrence and the rupture area, A = LW, where L is fault length and W is fault
characteristics. Since it is not realistic that a fault ruptures along its width. Knowing the slip rate and the rupture length of the faults, we
entire length, faults and fault systems are segmented when evaluating extracted the a-value (keeping fixed the b-value at 1.05, like the area
their seismic potential in order to avoid unrealistically large magnitude sources; see Table 3) using an average model based on Anderson and
earthquakes. If the fault is modelled with the real length (without seg Luco (1983) (Eq. (1), Eq. (2), Eq. (3)) and Youngs and Coppersmith
mentation), the a-value and the maximum magnitude would be unre (1985) (Eq. (4)) relations (Bungum, 2007).
alistically high, corresponding to the fault length. For this reason, Anderson and Luco (1983) proposed the following relationships for
knowing the fault width factor, the aspect ratio was fixed to extract the the determination of the number of earthquakes N above the lower
segments fault length. The aspect ratio is given by the ratio between the bound magnitude (around 4–5) on a fault:
fault length and the fault width factor. The details of the segmented (( )/ )
model faults are provided in Table 1. In this work, the data about the slip N1 (M) = d − b d (S / β)eb(Mmax − M) e− ((d/2)Mmax ) Eq. 1
rates is derived from GPS measurements (Barman et al., 2017; Bilham
(( )/ ) [ ]
and England, 2001; De and Kayal, 2003; Diehl et al., 2017; England and N2 (M) = d− b b (S / β) eb(Mmax − M)
− 1 e− ((d/2)Mmax ) Eq. 2
Bilham, 2015; Goswami, 2005; Hauck et al., 1998; Hetényi et al., 2016;
Kayal et al., 2006; Le Roux-Mallouf et al., 2015; Marechal et al., 2016;
( ( )/ ) { / [ ] }
N3 (M) = d d − b b (S / β) 1 b eb(Mmax − M)
− 1 − (Mmax − M) e− ((d/2)Mmax ) Eq. 3
8
Table 2
All the parameters related to the area sources necessary for the hazard calculation. The column “Id” is relative to the identificative number of each area; the “Depth
range” put in relation the minimum and the maximum depth where the events can generated; the “Hypocentral depth” is expressed by a statistical analysis for a
specified depth; the “M range” is relative to the Mmin and Mmax used; the a and b-values of the Gutenberg-Richter recurrence law (obtained from the catalogue); and
finally the “Focal mechanisms” expressed in terms of probability, specifying the most representative strike, dip and rake of the area.
Id Name Depth range Aspect Hypocentral depth (km) ¼ M a- b- Focal mechanisms (probability -
(km) ratio probability range value value strike,dip,rake)
1 Burmese Arc 0–150 2.0 17.5 = 0.15 4.5–8.2 5.89 1.05 0.25–0,90,180
52.5 = 0.30 0.25–290,90,0
87.5 = 0.35 0.25–35,45,90
122.5 = 0.20 0.25–35,60,90
2 Bangal basin 0–80 2.0 10 = 0.18 4.5–7.7 5.06 1.05 0.25–90,45,90
30 = 0.44 0.25–270,45,90
50 = 0.28 0.25–45,70,45
70 = 0.10 0.25–45,90,0
3 Shillong 0–80 2.0 10 = 0.26 4.5–7.6 4.73 1.05 0.50–330,90,180
30 = 0.44 0.50–330,60,45
50 = 0.24
70 = 0.06
4 Mikir Massif 0–80 2.0 10 = 0.17 4.5–7.7 4.81 1.05 0.50–330,90,180
30 = 0.28 0.50–330,60,45
50 = 0.46
70 = 0.09
5 Eastern Himalaya 0–80 2.0 10 = 0.31 4.5–8.6 4.87 1.05 0.40–225,30,90
Syntaxis 30 = 0.52 0.40–315,30,90
50 = 0.14 0.20–50,50,90
70 = 0.03
6 India region 0–80 2.0 10 = 0.20 4.5–7.3 4.47 1.05 1.00–0, 90,180
30 = 0.48
50 = 0.16
70 = 0.16
7 Nepal (HFT-west) 0–80 2.0 10 = 0.32 4.5–8.5 5.25 1.05 0.20–320,90,180
30 = 0.31 0.25–270,30,90
50 = 0.21 0.25–270,10,90
70 = 0.16 0.30–45,45,-90
8 Bhutan (HFT-central) 0–80 2.0 10 = 0.43 4.5–7.0 4.85 1.05 0.50–270,30,90
30 = 0.30 0.50–270,10,90
50 = 0.18
70 = 0.09
9 Arunachal 0–80 2.0 10 = 0.33 4.5–7.8 4.99 1.05 0.50–225,30,90
Pradesh (HFT-east) 30 = 0.40 0.50–225,10,90
50 = 0.19
70 = 0.08
10 China 0–80 2.0 10 = 0.38 4.5–6.5 4.31 1.05 1.00–70,40,-90
30 = 0.34
50 = 0.18
70 = 0.10
existing ones) are periodically released. However, despite the presence

√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
where b = b(ln(10)), d = d(ln(10)), β = (αM0 (0)/(μW) and M0 (0) is the of large methodological similarities between GMPEs, the predicted
seismic moment for Ms = 0. The parameter d is the magnitude scaling ground motion proves to be highly variable. A careful evaluation should
coefficient in the log-linear relation between moment and magnitude, of be performed when selecting the most suitable GMPEs for a given study
the form logM0 = c – dM (Kanamori and Anderson, 1975). region. There are few relations available that have been developed
In addition to these three models, also Youngs and Coppersmith specifically for the Himalayan region and hardly for any region with
(1985) relation has been considered: reasonably similar tectonics. However, the approach that is used for
[ ] selecting GMPEs for the study area is to use empirical relations based on
( 0 ) μAf S(d − b) 1 − e
− β(mu − m0 )
observed data from other similar active regions. The proper calibration
N4 m = Eq. 4 of such relations is difficult, leading to considerable uncertainties in the
bM0u e− β(mu − m0 )
prediction of ground motions, and from the largest earthquakes. In a first
where m0 is some arbitrary reference magnitude, m = b(ln(10), Af = LW round, eleven GMPEs were selected as possible candidates, covering
is fault area and mu is an upper bound magnitude. three different tectonic contexts: active shallow crust (ASC), stable
The maximum magnitude of the faults is essentially assessed from continental crust (SCC), and subduction zone (SZ). These GMPEs have
the segmented fault length (Table 3). The Wells and Coppersmith (1994) been tested and compared with their hazard curves and response spectra
magnitude-scaling relationship was used, that is based on a global and only some of these have been considered suitable for the study area
database of historical earthquake ruptures and it works for shallow (Fig. 8), taking into consideration some information get from literature
earthquakes in active tectonic regions. and the magnitude range of operation.
While the aleatory component of the model uncertainty is generally
5. Ground motion model taken into account through the hazard integral, the epistemic compo
nent, which is related to the available level of knowledge and the
During the last three decades, more than 500 different Ground Mo adopted initial assumptions and simplifications, can be quantified by
tion Prediction Equations (GMPEs) have been produced around the using a logic-tree approach. The need of a logic-tree scheme is also due
world (Douglas, 2019), and new equations (or the update version of to the fact that there is no GMPE calibrated on local settings, since few
9
Table 3 to use a logic-tree based on Boore and Atkinson (2008) with weight 0.4
The necessary parameters to model the faults. The columns “Rake” and “Dip” are and Atkinson and Boore (2003) with weight 0.6. For the subduction
related to the faults geometry; the “Upper-Lower seismogenic depth” put in zone, it was decided to use Atkinson and Boore (2003) relation, specif
relation the minimum and the maximum depth where the events can generated; ically developed for this type of areas, in combination with Boore and
the “Hypocentral depth” is expressed by a statistical analysis for a specified Atkinson (2008) relation, since there is a zone component which can be
depth; the “M range” is relative to the Mmin and Mmax used; the a and b-values
modelled as active. All the area sources and the faults have been
of the Gutenberg-Richter recurrence law (obtained from the slip rates).
modelled like ASC, except for the Burmese Arc zone that is modelled like
Name Rake Dip Upper-Lower M range a- b- SZ.
seismogenic value value
depth (km)
6. Results
Dauki 90 50 0–40 4.5–7.75 4.69 1.05
(reverse)
Dhubri 0 90 0–35 4.5–7.29 3.64 1.05 Individual PSHA results from each seismicity model (area sources,
(strike- fault model and hybrid model) are presented and evaluated with
slip) OpenQuake (OQ) software (Version 2.7); it is an open source and
Kopili 180 60 0–50 4.5–7.91 4.47 1.05 community-driven seismic hazard and risk calculation software devel
(strike-
oped by the Global Earthquake Model (GEM) foundation. The
slip)
Oldham 90 57 0–40 4.5–7.69 4.67 1.05 magnitude-frequency distribution chosen is a double truncated
(reverse) Gutenberg-Richter distribution. This is described by means of Mmin and
Tista 180 90 0–40 4.5–7.69 4.14 1.05 Mmax and by the a and b-values of the Gutenberg-Richter relationship.
(strike-
The magnitude-scaling relationship adopted is the Wells and Copper
slip)
Yadong − 90 60 0–20 4.5–7.33 3.46 1.05
smith, 1944 (WC1944) for shallow earthquakes in active tectonic re
(normal) gions and it is based on a global database of historical earthquake
HFT- 90 / 0–48 4.5–8.24 5.49 1.05 ruptures. The seismic hazard calculations have been carried out by
central performing hazard computations at a grid interval of 15 km, covering
(reverse)
the entire study region delimited by latitude 24◦ -28.8◦ N and longitude
HFT-east 90 / 0–47 4.5–8.01 5.48 1.05
(reverse) 88◦ -94.5◦ E. Peak Ground Acceleration (PGA) and Spectral Acceleration
HFT-west 90 / 0–47 4.5–8.11 5.47 1.05 (Sa) values for six different periods (from 0.05 to 2.0 s) have been
(reverse) evaluated at bedrock level (Vs30 equal to 800 m/s) corresponding to a
probability of exceedance (POE) of 10% in 50 years. This exceedance
value corresponds to return periods of 475 year. For each grid point, all
data of the study region are available. The current logic-tree model is
the sources within a radius of 200 km were considered for the evaluation
using two separate branching levels for active shallow crust and sub
of PGA and Sa values. The hazard results are presented by hazard maps
duction zone regions (Fig. 9).
for the study area, and all ground-motion exceedance values are
For the active shallow crust region, two GMPEs (Boore et al., 2014;
computed through hazard curves and mean and quantile (0.15,0.5 and
Boore and Atkinson, 2008) were used with equal weights in the
0.85) Uniform Hazard Spectra (UHS) for three important cities:
logic-tree calculation (0.5 each). For the subduction zone, it was decided
Fig. 8. Comparison between Boore et al. (2014), Boore and Atkinson (2008) and Atkinson and Boore (2003) models used in the logic-tree scheme; the response
spectra are shown for different magnitudes (5.0, 6.0, 7.0, 8.0) and for different distances (5, 20, 50, 100 km).
10
Fig. 9. Logic-tree model using two separate branching levels for active shallow crust and subduction zone regions.
Fig. 10. The hazard map including only the area sources. The hazard map is shown in PGA using the g scale for the ground motion. The probability of exceedance is
10% in 50 years, corresponding to 475 years return periods.
Thimphu (27.28◦ N, 89.38◦ E), Guwahati (26.08◦ N, 91.44◦ E) and Shil maximum mean SA value is recorded at 0.1 s and it is equal to 0.45 g.
long (25.34◦ N, 91.53◦ E). For each model, an example is shown for
Shillong city and at the end in the hybrid model the results are plotted 6.2. Calculation with only faults
for all the three cities.
A second calculation has been carried out with only the faults with
6.1. Calculation with only area sources derivation of the recurrence model data from GPS data. In this model, all
the faults (except to the Himalayan Thrust) are modelled like simple
In a first calculation, only the area sources have been considered, fault: the simple adjective relates to the geometry description of the
using the recurrence model extracted from the seismic catalogue; all the source which is obtained by projecting the fault trace along a charac
areas have been modelled with the parameters listed in Table 2. teristic dip direction (by extending the fault up to the surface main
Fig. 10 shows the hazard map for the study area. The area sources are taining the fault dip). The Himalayan Frontal Thrust is modelled like a
used to describe the seismicity occurring over a wide area where the complex fault: the complex adjective refers to the fact that the fault can
identification and characterization of a single fault is difficult; in this be realized as a union of planes with different dips. Complex faults are
case, the seismicity is equally distributed in each point of each area. The generally used to model intraplate megathrust faults such as the big
maximum PGA calculated is 0.35 g and it is shown in the two areas subduction structures active in the Pacific, but also to create listric fault
called “Eastern Himalayan Syntaxis” and “Burmese arc”. In Fig. 11, the sources with a realistic geometry. The first part of the HFT, that reaches
hazard curve and the UHS (fixing the annual rate of exceedance like 10% the surface, has been built with 25◦ dip (40 km width), the second one
in 50 years) are shown for Shillong city). Looking at the UHS, the with 10◦ dip (60–70 km width) and the last part that arrives at 50 km
11
Fig. 11. Calculation with only the area sources: on the left, the mean hazard curves computed for a range of spectral periods, including PGA; on the right, mean and
quantile UHS for Shillong city for 10% POE in 50 years.
Fig. 12. The hazard map including only the faults. The hazard map is shown in PGA using the g scale for the ground motion. The probability of exceedance is 10% in
50 years, corresponding to 475 years return periods.
deep with 30◦ dip (40 km width) (Zhang et al., 2016). In the model with derived from GPS measurements. However, the problem of using a
only the faults, both simple and complex faults have been considered model which combines zones and faults is establishing the distribution
and the parameters needed for the calculation are listed in Table 3. of seismic potential between them, considering that they are derived
Fig. 12 shows the hazard map for the area, and, as it can be observed, from different data sources. The problem is that a part of the events
the Himalayan thrust affects most the results. The maximum PGA contained in the catalogue are linked to faults and they were already
calculated is 0.89 g and it is recorded in the northeast part of the Hi included in the seismic potential of the faults derived from slip rate
malayan thrust. This value is particularly high, and it could be attributed estimates. If all events are assigned to the zones, the events related to the
to the way the thrust is modelled. Looking at the UHS for Shillong in faults would be double-counted, and at the end the total seismic po
Fig. 13, the maximum mean SA recorded is correspondent to 1.45 g at tential can be overestimated. Finding a solution to distribute the seismic
0.1 s. The SA is very high due to the proximity of Shillong city near the potential appropriately is not easy. An approach to avoid the double-
Dauki and Oldham faults. counting could be to fix the magnitude at a certain value to separate
the seismic potential to the areas or to the faults. In this work, the
magnitude was fixed at 6.5 to distinguish the different sources for the
6.3. Calculation with hybrid model seismic potential (Fig. 14): for M < 6.5, the seismic potential is attrib
uted to the area sources and for M ≥ 6.5 the seismic potential is related
The purpose of this calculation is to combine the seismic potential of to the faults. In this way, the big events are related the faults and the low
both types of sources that is derived from different data: for the zones, grade of seismicity to the area sources like basic seismicity. Fig. 15
the recurrence model is estimated from the seismic catalogue; and for shows, as expected, how the Himalayan thrust dominates the seismic
the faults, it is deduced from fault geometries and slip rate estimates
12
Fig. 13. Calculation with only the faults: on the left, the mean hazard curves computed for a range of spectral periods, including PGA; on the right, mean and
quantile UHS for Shillong city for 10% POE in 50 years.
Fig. 14. The diagram shows how the seismic potential form the area and faults sources models are combined together into the hybrid model. To avoid the double-
counting, the magnitude is fixed at 6.5 to distinguish the different sources: for M < 6.5, the seismic potential is attributed to the area sources and for M ≥ 6.5 the
seismic potential is related to the faults.
Fig. 15. The seismic hazard results including both the two types of sources, zones and faults, in the study area. The hazard map is shown in PGA using the g scale for
the ground motion. The probability of exceedance is 10% in 50 years, corresponding to 475 years return periods.
13
Fig. 16. Calculation with the area sources and the faults (hybrid model): on the left, the mean hazard curves computed for a range of spectral periods, including PGA;
on the right, the mean and quantile UHS for different cities for 10% POE in 50 years.
14
hazard with the maximum PGA of 0.77 g in the northeastern part. influence of such parameter on the computed hazard. The a-value was
Looking at these results, it is very clear why is so important to compute a derived differently for the two sources: for what concerns the area
model considering both types of sources. Like in the previous cases, sources, the a-value was obtained from the analysis of past seismicity.
Fig. 16 shows the hazard curves and UHS for Shillong city where the Conversely, the a-value for the faults was calculated by means of the slip
PGA is 0.55 g and the maximum mean SA recorded is around 0.9 g at 0.2 rates. The uncertainties on the a-value for the faults are related to many
s. The PGA recorded is in the middle between the results for the calcu factors, in particular to GPS measurements errors. Moreover, it is
lations with only the area sources (0.28 g) and only the faults (0.90 g). important to consider that part of this slip could be accommodated and
The answer for these observations can be find in how the hybrid model is converted in plastic deformation; different values of the seismic
developed using different thresholds of magnitudes between the two coupling coefficient have been tested, ranging between about 10% and
sources. 80% of the total slip. A coefficient of 40% has finally been selected as
In addition, in Fig. 16 the hazard curves, for different spectral ac optimal value for the region, providing the most realistic hazard sce
celerations, and the UHS for other two cities are shown. For Guwahati, narios. However, future tests should better consider the influence of
the PGA is 0.35 g and the maximum mean SA is 0.55 g at 0.1 s; for lower slip rates values on the results. Several other aspects related to the
Thimphu the PGA is 0.78 g and the maximum mean SA is 1.35 g at 0.2 s. faults, such as geometry, focal mechanisms and Mmax, have been
analyzed, but their influence on the hazard results was found to be
7. Discussion and conclusions minor. This is particularly true and was actually expected for 475 years
return period, but such consideration might not be applicable to longer
The main issue in assessing the seismic hazard in Northeast India and return periods (e.g. 2% POE in 50 years). Regarding the area sources, the
Bhutan is the lack of base information necessary to calibrate seismicity Mmax for each zone has been taken from the maximum recorded value
parameters. In particular, the availability of a sufficiently complete adding 0.5 unit; for the faults, the procedure adopted is more complex
earthquake catalogue covering an appropriate magnitude range and and involves the use of specific conversion relations between fault
time span is critical. This is however conditional to the presence of a length and magnitude. The first tests were made by using the entire fault
proper local network, operational over many decades and sufficiently trace, but this has led to a very high Mmax values and unrealistic seismic
dense to detect down to very low-magnitude seismicity. A complete potential. For this reason, different alternative faults segmentation ge
catalogue is fundamental to determine the earthquake occurrence rates, ometries have been developed, trying to limit the ruptures length and
which is one of the most important parameters controlling the seismic consequentially the expected Mmax for each fault based on geological
hazard of a region. In this work we mostly relied on the information from field evidence from satellite images and seismicity distribution of past
the Mw homogenized ISC-GEM Extended catalogue, which is nonethe events. The segmentation model is unfortunately very subjective in most
less to be considered complete only for magnitudes of 5 and above. cases, but it has also a relevant impact on the hazard. A possible follow-
Therefore, a possible future upgrade could be the integration of data up of this study could be the analysis of different traces segmentation to
from local seismological networks, whenever available, to supplement compare how this aspect influence the hazard, which has been investi
the available catalogue, in order to increase the total number of recor gated only marginally in this study due to time constraints.
ded events and consequentially perform a better statistical analysis. A final critical element is represented by the selection of the Ground
Moreover, the effect of declustering is to be considered. A major Motion Prediction Equations. Since all the area sources and the faults
assumption for the calculation of the occurrence parameters is that the have been modelled like active shallow crust (except for the Burmese
earthquake generation follows a Poissonian process. For this reason, a Arc zone that has been modelled like subduction zone), a preliminary
declustering procedure based on the Gardner and Knopoff (1974) selection of the GMPEs has verted to the equations calibrated and suit
method was tested, by varying distance, magnitude and time windows. able for these tectonic frameworks. Then eleven GMPEs have been
However, the classical Gardner and Knopoff (1974) method removed selected and tested basing on the magnitude range of operation. A logic-
more than 40% of the total events, which leads to a drastic reduction of tree scheme was implemented in OpenQuake in order to account for
the activity rates of the source zones. A possible explanation is due to the epistemic uncertainty, using the three relations that have been consid
fact that the method has been originally calibrated on data from ered suitable for the study area. We found that the choice of the GMPEs
Southern California, which is expectedly different from present study heavily affect the hazard results: for these reasons could be useful a
area from a geological and seismotectonics perspective. To overcome future investigation where different equations and branching levels
this problem, a manual procedure was performed, based on the visual were adopted. It must be noted that the need of a logic-tree scheme is
selection of the cluster events, in order to have a good balance between also due to the fact that there are no GMPEs calibrated on local settings,
the number of aftershocks removed and the number of the events used since few data of the study region are available. An important future
for the statistical analysis. The procedure has removed about 25% of the development could therefore consist in creating a GMPE specific the
initial registered events. The choice of an optimal declustering approach Himalayan region, perhaps compensating the lack of earthquake ob
is nonetheless still open issue and must be better investigated in future servations with the results from numerical modelling.
analysis. Another aspect that we notice looking at Fig. 12 is the fact that the
Another important aspect related to the catalogue is the proper ground motion calculated doesn’t change along the thrust dip direction.
definition of the completeness periods for the different magnitudes. This problem could find an explanation in what type of distance metric is
Several tests have been performed assuming different magnitude of used in the GMPEs selected. For the faults, Boore et al. (2014) and Boore
completeness, both for the total study area and for each single source and Atkinson (2008) have been used and these relations are based on the
zone. From these tests, it was evident a strong sensitivity of this Joyner-Boore distance metric (Rjb). The Rjb is defined as the shortest
component on the final hazard results, particularly about the variability distance from a site to the surface projection of the rupture surface, and
of the occurrence rates, probably due to the spatially irregular distri in practice the faults dip values have not very great effects on the ground
bution of events. In this study, we assumed the earthquake occurrence motion results. As in the area source model, the recurrence values are
process to be fully represented by a double truncated Gutenberg-Richter the parameters that mostly drive the hazard results; however, in the
law. Due to the lack of local events, to obtain a more robust estimate of faults source model, also the dip and the geometry of fault are very
the activity rates, a unique b-value was first derived using all available important for the hazard calculation.
events for the whole investigated area. Such value (1.05) was then After all the previous considerations, the overall hazard distribution
imposed to the different sources, allowing only the rates (then the a- in the study area reflects the seismic zonation of the Indian building
value) to vary. For future development, however, different b-values code. The Bureau of Indian Standards (BIS, 2002) classifies the entire
should be tested, in order to better understand the sensitivity and the N-E India region into Zone V, that represents the highest hazard in India.
15
Table 4 3. Baro et al. (2018) carry out a Deterministic Seismic Hazard Assess
Comparison of computed PGA (in g) with other studies for 10% probability of ment (DSHA) of the Shillong Plateau; the hazard map underlines a
exceedance in 50 years (475 years return period) at selected cities. strong influence of the Oldham and Dauki faults on the results,
Shillong Guwahati Thimphu 4. Baro et al. (2020) present a PSHA of Shillong Plateau, using historical
Bhatia et al. (1999) 0.30 0.30 /
and instrumentally recorded regional earthquakes since 1411.
BIS (2002) 0.18 0.18 / 5. Stevens et al. (2020) is the first PSHA focusing on Bhutan and it is
Sharma and Malik (2006) 0.45 0.50 / based on faults locations, slip-rates and paleoseismic earthquake
NDMA (2011) 0.25 0.23 / data. The results show a non-uniform hazard level across the coun
Nath and Thingbaija (2012) 0.72 0.66 /
try; this outcome goes against the existing building code of Bhutan,
Das et al. (2016) 0.32 0.24 /
Baro et al. (2018) DSHA 0.36 / / adopted from the Indian Seismic Zonation of 2002 (BIS- 2002), that
Baro et al. (2020) 0.19 / / uses a PGA of 0.36 g uniformly applied across the entire country. The
Stevens et al. (2020) / / 0.77 PGA for Thimphu, computed for 475 years return period, is in
Present study (hybrid model) 0.55 0.35 0.78 accordance with the value calculated in the present study.
The analysis performed in this work allows obtaining a more accurate It is important to underline some aspects: the seismic hazard gap of
representation of the hazard level in this region. As it can be seen in Bhutan and the lower values obtained, compared to the ones calculated
Fig. 15, the state of Arunachal Pradesh has seismically higher hazard in this work. Both these facts can be attributed to the lack of faults
than in the Assam and Meghalaya states. The higher values in the north modelling, that leads to lower and maybe not realistic PGA values. In
part of the study area are due to the presence of the Himalayan particular, in this study, for Bhutan the hazard is driven by the Hima
décollement (and consequently to the high values of slip in this area). In layan thrust. In the calculation, however, the ground motion doesn’t
the Assam and Meghalaya states the seismic hazard increases signifi actually change along the thrust dip direction, as it would intuitively be.
cantly when active faults are considered. As already explain, the faults dip values have not very great effects on
Different studies have been carried out at regional scale and a the ground motion results due to how the considered GMPEs were built.
comparison of the PGA results for Shillong, Guwahati and Thimphu If a specific GMPE calibrated for this zone should be consider, probably a
between previous studies and the present investigation is shown in hypocentral distance metric could lead to better results with respect to
Table 4. the classical Rjb used in Boore et al. (2014) and Boore and Atkinson
(2008) equations.
1. Khattri et al. (1984) compute a maximum PGA of 0.8 g along the Furthermore, there is still a long way to go with respect to under
Brahmaputra valley and 0.4 g–0.6 g along the Himalayan mountain standing (and thereby predicting in more detail) the tectonic processes
belt, for the calculation at 10% probability of exceedance in 50 years, in this region, which is more complicated than in most other regions of
using 24 seismogenic sources for the entire India. the world. The conducted study represents a first step towards esti
2. Bhatia et al. (1999) and Giardini et al. (1999), under the Global mating seismic hazard for North-East India and Bhutan. However, the
Seismic Hazard Assessment Program (GSHAP), develop a seismic doubt that all the seismogenic faults have not been analyzed and taken
hazard map for the whole India, using 86 seismic source zones. The into consideration remains; in this area, which is very complex from the
highest values of PGA (0.35 g–0.4 g for 10% probability of exceed geological point of view, there are most certainly many hidden seis
ance in 50 years) are shown in the seismically active regions as the mogenic faults.
Burmese arc, Northeastern India and North-West Himalaya Finally, it is here important to note how the hybrid model is built.
(Hindukush). Since there are two different type of sources, the areas and the faults, it
3. Walling and Mohanty (2009) give an overview on the development was decided to assign the basic seismicity to the area sources, with an
of the seismic zonation map of India through time as well as a review equal distribution of the seismic potential determined through the
of seismic hazard and zoning studies; they point out that the seismic occurrence parameters calculated for each zone. The most relevant
zonation at a national level is not sufficient for a detailed seismic events are instead addressed to the faults. In order to consider only once
hazard assessment at the local level. the effects of each source, the influence of the areas is limited to a
4. NDMA (2011) report describes PGA values up to 0.35 g with 10% magnitude of 6.5, while the potential related to the faults is set from this
probability for exceedance in 50 years for the eastern Himalaya magnitude to the maximum magnitude calculated specifically for each
syntaxis and the Indo-Burmese arc. In the model, 32 source zones are fault. The seismic hazard values calculated (maximum PGA equal to
modelled. 0.77 g in the hybrid model) can be used like a basis for future applica
5. Nath and Thingbaijam (2012) show hazard distribution in the tions, like input data for a probabilistic risk computation and for a better
country significantly higher than GSHAP, and BIS (2002) studies. definition of scenario earthquakes for deterministic risk assessment.
The highest values are shown in northeastern India.
Data Availability
All these previous studies are performed for the whole India, using
only the area source model approach and often applying only a single The moment tensor from the Global Centroid Moment Tensor cata
attenuation relationship, disregarding the different geological prov logue (Ekström et al., 2012) was obtained from https://www.globalcmt.
inces. For these reasons, the detailed level of the hazard results is lower org/CMTsearch.html (last access January 2019).
with respect to the one analyzed in this study. In addition, it is important The USGS earthquake catalogue was obtained from https://earth
to mention some seismic hazard studies that they focus on the same quake.usgs.gov/earthquakes/search/(last access January 2019).
study area at local scale: The faults information was taken from the Geological Survey of India
https://www.gsi.gov.in/(last access January 2019).
1. Sharma and Malik (2006) study applies an area source model The Ground Motion Prediction Equations (GMPEs) (Douglas, 2019)
approach, and it shows PGA values between 0.05 g and 0.6 g for 475 were extracted from http://www.gmpe.org.uk (last update December
years return period, with the maximum values occurring around the 19, 2019).
Dauki fault in the Shillong massif. OpenQuake (OQ) software (Version 2.7) was used for the computa
2. Das et al. (2016) perform a PSHA for Northeast India region using the tion (GEM, 2017).
area sources model approach.
16
Funding Bungum, H., Lindholm, C.D., Mahajan, A.K., 2017. Earthquake recurrence in NW and
central Himalaya. J. Asian Earth Sci. https://doi.org/10.1016/j.jseaes.2017.01.034.
Carlton, B.D., Skurtveit, E., Bohloli, B., Atakan, K., Dondzila, E., Kaynia, A.M., 2018.
This work has been performed during an Erasmus Traneeship Pro Probabilistic seismic hazard analysis for offshore bangladesh including fault sources.
gram and the work has been partially funded by the University of Pavia https://doi.org/10.1061/9780784481462.015.
and NORSAR to cover mainly personal costs ( place to stay, food and Cornell, C.A., 1968. Engineering seismic risk analysis. Bull. Seismol. Soc. Am. 58,
1583–1606.
transportation) during that period. Curray, J.R., Emmel, F.J., Moore, D.G., Raitt, R.W., 1982. Structure, tectonics, and
geological history of the northeastern Indian ocean. In: The Ocean Basins and
Margins. Springer US, Boston, MA, pp. 399–450. https://doi.org/10.1007/978-1-
CRediT authorship contribution statement 4615-8038-6_9.
Curray, J.R., Moore, D.G., Lawver, L.A., Emmel, F.J., Raitt, R.W., Henry, M.,
Federica Ghione: Conceptualization, Methodology, Software, Kieckhefer, R., 1979. Tectonics of the andaman sea and Burma. In: Geological and
Geophysical Investigations of Continental Margins. American Association of
Formal analysis, Investigation, Writing – original draft, Writing – review
Petroleum Geologists, pp. 189–198.
& editing. Valerio Poggi: Validation, Resources, Data curation, Super Das, R., Sharma, M.L., Wason, H.R., 2016. Probabilistic seismic hazard assessment for
vision. Conrad Lindholm: Resources, Supervision. Northeast India region. Pure Appl. Geophys. 173, 2653–2670. https://doi.org/
10.1007/s00024-016-1333-9.
De, R., Kayal, J.R., 2003. Seismotectonic model of the Sikkim Himalaya: constraint from
microearthquake surveys. Bull. Seismol. Soc. Am. 93, 1395–1400. https://doi.org/
Declaration of competing interest
10.1785/0120020211.
Diehl, T., Singer, J., Hetényi, G., Grujic, D., Clinton, J., Giardini, D., Kissling, E., 2017.
The authors declare that they have no known competing financial Seismotectonics of Bhutan: evidence for segmentation of the Eastern Himalayas and
interests or personal relationships that could have appeared to influence link to foreland deformation. Earth Planet Sci. Lett. 471, 54–64. https://doi.org/
10.1016/j.epsl.2017.04.038.
the work reported in this paper. Douglas, J., 2019. Ground motion prediction equations 1964-2019 [WWW Document].
URL. http://www.gmpe.org.uk/gmpereport2014.pdf.
Ekström, G., Nettles, M., Dziewoński, A.M., 2012. The global CMT project 2004–2010:
Acknowledgments
Centroid-moment tensors for 13,017 earthquakes. Phys. Earth Planet. Inter. 200–
201, 1–9. https://doi.org/10.1016/j.pepi.2012.04.002.
This study is made possible by the generous support of NORSAR, that England, P., Bilham, R., 2015. The Shillong Plateau and the great 1897 Assam
earthquake. Tectonics 34, 1792–1812. https://doi.org/10.1002/2015TC003902.
provided technical assistant and scientific discussions.
Esteva, L., 1968. Bases para la formulacion de decisiones de diseño sísmico. Universidad
Autonoma Nacional de México.
References Esteva, L., 1967. Criterios para la construcción de espectros para diseño sísmico,
Proceedings of the XII Jornadas Sudamericanas de Ingeniería Estructural y III
Simposio Panamericano de Estructuras. Published later in Boletín del Instituto de
Akinci, A., Galadini, F., Pantosti, D., Petersen, M., Malagnini, L., Perkins, D., 2009. Effect
Materiales y Modelos Estructurales, 19. Universidad Central de Venezuela, 1967.,
of time dependence on probabilistic seismic-hazard maps and deaggregation for the
Caracas.
central Apennines, Italy. Bull. Seismol. Soc. Am. https://doi.org/10.1785/
Gahalaut, V.K., Rajput, S., Kundu, B., 2011. Low seismicity in the Bhutan Himalaya and
0120080053.
the stress shadow of the 1897 Shillong Plateau earthquake. Phys. Earth Planet. In.
Anderson, J.G., Luco, J.E., 1983. Consequences of slip rate constants on earthquake
186, 97–102. https://doi.org/10.1016/j.pepi.2011.04.009.
occurrence relations. Bull. Seismol. Soc. Am. 73, 471–496.
Gardner, J.K., Knopoff, L., 1974. Is the sequence of earthquakes in southern California,
Armijo, R., Tapponnier, P., Han, T., 1989. Late Cenozoic right-lateral strike-slip faulting
with aftershock removed, Poissonian? Bull. Seismol. Soc. Am. 64, 1363–1367.
in southern Tibet. J. Geophys. Res. Solid Earth 94, 2787–2838. https://doi.org/
GEM, 2017. The OpenQuake-engine user manual. Global Earthquake Model (GEM)
10.1029/JB094iB03p02787.
Technical Report 2017-11. https://doi.org/10.13117/GEM.OPENQUAKE.MAN.
Atkinson, G.M., Boore, D.M., 2003. Empirical ground-motion relations for subduction-
ENGINE.2.8/01.
zone earthquakes and their application to cascadia and other regions. Bull. Seismol.
Giardini, D., Grünthal, G., Shedlock, K.M., Zhang, P., 1999. The GSHAP global seismic
Soc. Am. 93, 1703–1729. https://doi.org/10.1785/0120020156.
hazard map. Ann. Geofisc. 42, 1225–1228.
Barman, P., Jade, S., Shrungeshwara, T.S., Kumar, A., Bhattacharyya, S., Ray, J.D.,
Goswami, S., 2005. Inverted metamorphism in the Sikkim-Darjeeling Himalaya:
Jagannathan, S., Jamir, W.M., 2017. Crustal deformation rates in Assam valley,
structural, metamorphic and numerical studies. University of Cambridge.
Shillong Plateau, eastern Himalaya, and Indo-Burmese region from 11 years
Greenwood, L.V., Argles, T.W., Parrish, R.R., Harris, N.B.W., Warren, C., 2016. The
(2002–2013) of GPS measurements. Int. J. Earth Sci. 106, 2025–2038. https://doi.
geology and tectonics of central Bhutan. J. Geol. Soc. London. 173, 352–369.
org/10.1007/s00531-016-1407-z.
https://doi.org/10.1144/jgs2015-031.
Baro, O., Kumar, A., 2017. Seismic source characterization for the Shillong Plateau in
Gulia, L., Wiemer, S., Wyss, M., 2012. Theme IV–understanding seismicity catalog
northeast India. J. Seismol. 21, 1229–1249. https://doi.org/10.1007/s10950-017-
artifacts and quality control. Community Online Resour. Stat. Seism. Anal. 1–26
9664-2.
https://doi.org/10.5078/corssa-93722864.
Baro, O., Kumar, A., Ismail-Zadeh, A., 2020. Seismic hazard assessment of the Shillong
Gutenberg, B., Richter, C.F., 1956. Magnitude and energy of earthquakes. Ann. Geofisc.
Plateau using a probabilistic approach. Geomatics, Nat. Hazards Risk 11,
9, 1–15.
2210–2238. https://doi.org/10.1080/19475705.2020.1833989.
Hainzl, S., Scherbaum, F., Beauval, C., 2006. Estimating background activity based on
Baro, O., Kumar, A., Ismail-Zadeh, A., 2018. Seismic hazard assessment of the Shillong
interevent-time distribution. Bull. Seismol. Soc. Am. https://doi.org/10.1785/
plateau, India. Geomatics. Nat. Hazards Risk. https://doi.org/10.1080/
0120050053.
19475705.2018.1494043.
Hauck, M.L., Nelson, K.D., Brown, L.D., Zhao, W., Ross, A.R., 1998. Crustal structure of
Baruah, Santanu, Baruah, Saurabh, Saikia, S., Shrivastava, M.N., Sharma, A., Reddy, C.
the Himalayan orogen at ~90◦ east longitude from Project INDEPTH deep reflection
D., Kayal, J.R., 2016. State of tectonic stress in Shillong Plateau of northeast India.
profiles. Tectonics 17, 481–500. https://doi.org/10.1029/98TC01314.
Phys. Chem. Earth, Parts A/B/C 95, 36–49. https://doi.org/10.1016/j.
Hetényi, G., Le Roux-Mallouf, R., Berthet, T., Cattin, R., Cauzzi, C., Phuntsho, K.,
pce.2015.11.009.
Grolimund, R., 2016. Joint approach combining damage and paleoseismology
Berthet, T., Ritz, J.-F., Ferry, M., Pelgay, P., Cattin, R., Drukpa, D., Braucher, R.,
observations constrains the 1714 A.D. Bhutan earthquake at magnitude 8 ± 0.5.
Hetenyi, G., 2014. Active tectonics of the eastern Himalaya: new constraints from
Geophys. Res. Lett. 43 https://doi.org/10.1002/2016GL071033, 10,695-10,702.
the first tectonic geomorphology study in southern Bhutan. Geology 42, 427–430.
Imsong, W., Choudhury, S., Phukan, S., 2016. Ascertaining the neotectonic activities in
https://doi.org/10.1130/G35162.1.
the southern part of Shillong Plateau through geomorphic parameters and remote
Bhatia, S.C., Kumar, M.R., Gupta, H.K., 1999. A probabilistic seismic hazard map of India
sensing data. Curr. Sci. 110, 91. https://doi.org/10.18520/cs/v110/i1/91-98.
and adjoining regions. Ann. Geofisc. 42, 1153–1166.
Islam, M.S., Shinjo, R., Kayal, J.R., 2011. Pop-up tectonics of the Shillong Plateau in
Bilham, R., England, P., 2001. Plateau ‘pop-up’ in the great 1897 Assam earthquake.
northeastern India: insight from numerical simulations. Gondwana Res. 20,
Nature 410, 806–809. https://doi.org/10.1038/35071057.
395–404. https://doi.org/10.1016/j.gr.2010.11.007.
BIS, 2002. IS 1893–2002 (Part 1): Indian Standard Criteria for earthquake resistant
Kanamori, H., Anderson, D.L., 1975. Theoretical basis of some empirical relations in
design of structures, Part 1—General Provisions and Buildings. Bureau of Indian
seismology. Bull. Seismol. Soc. Am. 65, 1073–1095.
Standards, New Delhi.
Kaverina, A.N., Lander, A.V., Prozorov, A.G., 1996. Global creepex distribution and its
Boore, D.M., Atkinson, G.M., 2008. Ground-motion prediction equations for the average
relation to earthquake-source geometry and tectonic origin. Geophys. J. Int. 125,
horizontal component of PGA, PGV, and 5%-damped PSA at spectral periods
249–265. https://doi.org/10.1111/j.1365-246X.1996.tb06549.x.
between 0.01 s and 10.0 s. Earthq. Spectra 24, 99–138. https://doi.org/10.1193/
Kayal, J.R., 2010. Himalayan tectonic model and the great earthquakes: an appraisal.
1.2830434.
Geomatics, Nat. Hazards Risk 1, 51–67. https://doi.org/10.1080/
Boore, D.M., Stewart, J.P., Seyhan, E., Atkinson, G.M., 2014. NGA-West2 equations for
19475701003625752.
predicting PGA, PGV, and 5% damped PSA for shallow crustal earthquakes. Earthq.
Kayal, J.R., 2008. Microearthquake Seismology and Seismotectonics of South Asia.
Spectra 30, 1057–1085. https://doi.org/10.1193/070113EQS184M.
Springer.
Bungum, H., 2007. Numerical modelling of fault activities. Comput. Geosci. 33,
808–820. https://doi.org/10.1016/j.cageo.2006.10.011.
17
Kayal, J.R., Arefiev, S.S., Barua, S., Hazarika, D., Gogoi, N., Kumar, A., Chowdhury, S.N., Tapponnier, P., Peltzer, G., Le Dain, A.Y., Armijo, R., Cobbold, P., 1982. Propagating
Kalita, S., 2006. Shillong plateau earthquakes in northeast India region: complex extrusion tectonics in Asia: new insights from simple experiments with plasticine.
tectonic model. Curr. Sci. 91, 109–114. Geology 10, 611. https://doi.org/10.1130/0091-7613(1982)10<611:PETIAN>2.0.
Khattri, K.N., Rogers, A.M., Perkins, D.M., Algermissen, S.T., 1984. A seismic hazard map CO;2.
of India and adjacent areas. Tectonophysics. https://doi.org/10.1016/0040-1951 Taroni, M., Akinci, A., 2021. Good practices in PSHA: declustering, b-value estimation,
(84)90156-2. foreshocks and aftershocks inclusion; A case study in Italy. Geophys. J. Int. https://
Lavé, J., Avouac, J.P., 2000. Active folding of fluvial terraces across the siwaliks hills, doi.org/10.1093/gji/ggaa462.
Himalayas of central Nepal. J. Geophys. Res. Solid Earth 105, 5735–5770. https:// Thatcher, W., Pollitz, F.F., 2008. Temporal evolution of continental lithospheric strength
doi.org/10.1029/1999JB900292. in actively deforming regions. GSA Today (Geol. Soc. Am.). https://doi.org/
Le Dain, A.Y., Tapponnier, P., Molnar, P., 1984. Active faulting and tectonics of Burma 10.1130/GSAT01804-5A.1.
and surrounding regions. J. Geophys. Res. 89, 453. https://doi.org/10.1029/ Uyeda, S., Kanamori, H., 1979. Back-arc opening and mode of subduction. J. Geophys.
JB089iB01p00453. Res. 84, 1049–1061.
Le Roux-Mallouf, R., Godard, V., Cattin, R., Ferry, M., Gyeltshen, J., Ritz, J.F., Valentini, A., Pace, B., Boncio, P., Visini, F., Pagliaroli, A., Pergalani, F., 2019. Definition
Drupka, D., Guillou, V., Arnold, M., Aumaître, G., Bourlès, D.L., Keddadouche, K., of seismic input from fault-based PSHA: remarks after the 2016 Central Italy
2015. Evidence for a wide and gently dipping Main Himalayan Thrust in western earthquake sequence. Tectonics. https://doi.org/10.1029/2018TC005086.
Bhutan. Geophys. Res. Lett. 42, 3257–3265. https://doi.org/10.1002/ Valentini, A., Visini, F., Pace, B., 2017. Integrating faults and past earthquakes into a
2015GL063767. probabilistic seismic hazard model for peninsular Italy. Nat. Hazards Earth Syst. Sci.
Marechal, A., Mazzotti, S., Cattin, R., Cazes, G., Vernant, P., Drukpa, D., Thinley, K., https://doi.org/10.5194/nhess-17-2017-2017.
Tarayoun, A., Le Roux-Mallouf, R., Thapa, B.B., Pelgay, P., Gyeltshen, J., Velasco, A.A., Gee, V.L., Rowe, C., Grujic, D., Hollister, L.S., Hernandez, D., Miller, K.C.,
Doerflinger, E., Gautier, S., 2016. Evidence of interseismic coupling variations along Tobgay, T., Fort, M., Harder, S., 2007. Using small, temporary seismic networks for
the Bhutan Himalayan arc from new GPS data. Geophys. Res. Lett. 43 https://doi. investigating tectonic deformation: brittle deformation and evidence for strike-slip
org/10.1002/2016GL071163, 12,399-12,406. faulting in Bhutan. Seismol Res. Lett. 78, 446–453. https://doi.org/10.1785/
Martin, S., Szeliga, W., 2010. A catalog of felt intensity data for 570 earthquakes in India gssrl.78.4.446.
from 1636 to 2009. Bull. Seismol. Soc. Am. 100, 562–569. https://doi.org/10.1785/ Vernant, P., Bilham, R., Szeliga, W., Drupka, D., Kalita, S., Bhattacharyya, A.K., Gaur, V.
0120080328. K., Pelgay, P., Cattin, R., Berthet, T., 2014. Clockwise rotation of the Brahmaputra
Mitchell, A.H.G., McKerrow, W.S., 1975. Analogous evolution of the Burma orogen and valley relative to India: tectonic convergence in the eastern Himalaya, naga hills, and
the scottish caledonides. Geol. Soc. Am. Bull. 86, 305. https://doi.org/10.1130/ Shillong Plateau. J. Geophys. Res. Solid Earth 119, 6558–6571. https://doi.org/
0016-7606(1975)86<305:AEOTBO>2.0.CO;2. 10.1002/2014JB011196.
Murray, J.R., Minson, S.E., Svarc, J.L., 2014. Slip rates and spatially variable creep on Vorobieva, I., Mandal, P., Gorshkov, A., 2017. Block-and-fault dynamics modelling of the
faults of the northern San Andreas system inferred through bayesian inversion of Himalayan frontal arc: implications for seismic cycle, slip deficit, and great
global positioning system data. J. Geophys. Res. Solid Earth. https://doi.org/ earthquakes. J. Asian Earth Sci. 148, 131–141. https://doi.org/10.1016/j.
10.1002/2014JB010966. jseaes.2017.08.033.
Nandy, D.R., 2001. Geodynamics of Northeastern India and the Adjoining Region. ACB Walling, M.Y., Mohanty, W.K., 2009. An overview on the seismic zonation and
Publication, Kolkata. microzonation studies in India. Earth Sci. Rev. https://doi.org/10.1016/j.
Nandy, D.R., 1983. The eastern Himalaya and the Indo-Burman Orogen in relation to the earscirev.2009.05.002.
Indian plate movement. Geol. Surv. India Misc. Pub. 43, 153–159. Weatherill, G.A., Pagani, M., Garcia, J., 2016. Exploring earthquake databases for the
Nath, S.K., Thingbaijam, K.K.S., 2012. Probabilistic seismic hazard assessment of India. creation of magnitude-homogeneous catalogues: tools for application on a regional
Seismol Res. Lett. https://doi.org/10.1785/gssrl.83.1.135. and global scale. Geophys. J. Int. 206, 1652–1676. https://doi.org/10.1093/gji/
NDMA, 2011. Development of probabilistic seismic hazard map of India technical report. ggw232.
Natl. Dis. Manag. Auth. Wells, D.L., Coppersmith, K.J., 1994. New empirical relationships among magnitude,
Oldham, R.D., 1899. Report on the great earthquake of the 12th June 1897, in: rupture length, rupture width, rupture area, and surface displacement. Bull. Seismol.
Geological Survey of India Mem, p. 379. Soc. Am. 84, 974–1002.
Panzera, F., Zechar, J.D., Vogfjörd, K.S., Eberhard, D.A.J., 2016. A revised earthquake Woessner, J., Laurentiu, D., Giardini, D., Crowley, H., Cotton, F., Grünthal, G.,
catalogue for south Iceland. Pure Appl. Geophys. 173, 97–116. https://doi.org/ Valensise, G., Arvidsson, R., Basili, R., Demircioglu, M.B., Hiemer, S., Meletti, C.,
10.1007/s00024-015-1115-9. Musson, R.W., Rovida, A.N., Sesetyan, K., Stucchi, M., Anastasiadis, A., Akkar, S.,
Paul Burgess, W., Yin, A., Dubey, C.S., Shen, Z.-K., Kelty, T.K., 2012. Holocene Engin Bal, I., Barba, S., Bard, P.Y., Beauval, C., Bolliger, M., Bosse, C., Bonjour, C.,
shortening across the main frontal thrust zone in the eastern Himalaya. Earth Planet Bungum, H., Carafa, M., Cameelbeeck, T., Carvalho, A., Campos-Costa, A.,
Sci. Lett. 357–358, 152–167. https://doi.org/10.1016/j.epsl.2012.09.040. Coelho, E., Colombi, M., D’amico, V., Devoti, R., Drouet, S., Douglas, J., Edwards, B.,
Poggi, V., Garcia-Peláez, J., Styron, R., Pagani, M., Gee, R., 2020. A probabilistic seismic Erdik, M., Fäh, D., Fonseca, J., Fotopoulou, S., Glavatovic, B., Gómez Capera, A.A.,
hazard model for North Africa, Bulletin of Earthquake Engineering. Springer Hauser, J., Husson, F., Kastelic, V., Kästli, P., Karatzetzou, A., Kaviris, G., Keller, N.,
Netherlands. https://doi.org/10.1007/s10518-020-00820-4. Kierulf, H.P., Kouskouna, V., Krishnamurty, R., Lang, D., Lemoine, A., Lindholm, C.,
Rajendran, C.P., Rajendran, K., Duarah, B.P., Baruah, S., Earnest, A., 2004. Interpreting Makropoulos, K., Manakou, M., Marmureanu, G., Martinelli, F., Garcia
the style of faulting and paleoseismicity associated with the 1897 Shillong, northeast Mayordomo, J., Mihaljevic, J., Monelli, D., Garcia-Moreno, D., Nemser, E.,
India, earthquake: implications for regional tectonism. Tectonics 23. https://doi. Pagani, M., Pinho, R., Pisani, A.R., Pitilakis, D., Pitilakis, K., Poggi, V., Radulian, M.,
org/10.1029/2003TC001605. Riga, E., Sandikkaya, M.A., Segou, M., Siegert, R., Silva, V., Stromeyer, D., Sousa, L.,
Rivas-Medina, A., Benito, B., Gaspar-Escribano, J.M., 2018. Approach for combining Sørensen, M.B., Tellez-Arenas, A., Vanneste, K., Wahlstöm, R., Weatherill, G.,
fault and area sources in seismic hazard assessment: application in south-eastern Viganò, D., Vilanova, S., Yenier, E., Zulfikar, C., Adams, J., Bommer, J.J., Bonilla, F.,
Spain. Nat. Hazards Earth Syst. Sci. 18, 2809–2823. https://doi.org/10.5194/nhess- Faccioli, E., Gülen, L., Koller, M., Pinto, A., Pinto, P., Papaioannou, C., Peruzza, L.,
18-2809-2018. Scherbaum, F., Scotti, O., Stirling, M., Theodoulidis, N., Wenk, T., Zschau, J., 2015.
Saikia, S., Chopra, S., Baruah, S., Singh, U.K., 2017. Shallow sedimentary structure of the The 2013 European Seismic Hazard Model: key components and results. Bull.
Brahmaputra valley constraint from receiver functions analysis. Pure Appl. Geophys. Earthq. Eng. https://doi.org/10.1007/s10518-015-9795-1.
174, 229–247. https://doi.org/10.1007/s00024-016-1371-3. Yadav, R.B.S., Tripathi, J.N., Rastogi, B.K., Das, M.C., Chopra, S., 2010. Probabilistic
Seeber, L., Armbruster, J.G., Quittmeyer, R., 1981. Seismicity and continental subduction assessment of earthquake recurrence in northeast India and adjoining regions. Pure
in the Himalayan arc. Himalaya Geodyn. Evol. 4, 215–242. Appl. Geophys. 167, 1331–1342. https://doi.org/10.1007/s00024-010-0105-1.
Sharma, A., Baruah, S., Piccinini, D., Saikia, S., Phukan, M.K., Chetia, M., Kayal, J.R., Yaghmaei-Sabegh, S., Shoaeifar, N., Shoaeifar, P., 2018. A bimodal hybrid model for
2017. Crustal seismic anisotropy beneath Shillong plateau - Assam valley in North time-dependent probabilistic seismic hazard analysis. Pure Appl. https://doi.org/
East India: shear-wave splitting analysis using local earthquakes. Tectonophysics 10.1007/s00024-018-1839-4. Geophys.
717, 425–432. https://doi.org/10.1016/j.tecto.2017.08.027. Yin, A., 2006. Cenozoic tectonic evolution of the Himalayan orogen as constrained by
Sharma, M.L., Malik, S., 2006. Probabilistic seismic hazard analysis and estimation of along-strike variation of structural geometry, exhumation history, and foreland
spectral strong ground motion on bed rock in North East India. 4th Int. Conf. Earthq. sedimentation. Earth Sci. Rev. 76, 1–131. https://doi.org/10.1016/j.
Eng. Taipei, Taiwan. earscirev.2005.05.004.
Shivamanth, A., Shama, P.S., K, D., M, S., C, H., 2016. A review on development of Youngs, R.R., Coppersmith, K.J., 1985. Implications of fault slip rates and earthquake
seismic hazard analysis of India. Int. J. Mod. Trends Eng. Res. 3. recurrence models to probabilistic seismic hazard estimates. Bull. Seismol. Soc. Am.
Stepp, J.C., 1971. An investigation of earthquake risk in the Puget Sound area by use of 75, 939–964.
the type I distribution of largest extremes. Pennsylvania State University. Zhang, L., Li, J., Liao, W., Wang, Q., 2016. Source rupture process of the 2015 Gorkha,
Stevens, V.L., Avouac, J.P., 2015. Interseismic coupling on the main Himalayan thrust. Nepal Mw7.9 earthquake and its tectonic implications. Geod. Geodyn. 7, 124–131.
Geophys. Res. Lett. 42, 5828–5837. https://doi.org/10.1002/2015GL064845. https://doi.org/10.1016/j.geog.2016.03.001.
Stevens, V.L., De Risi, R., Le Roux-Mallouf, R., Drukpa, D., Hetényi, G., 2020. Seismic Zhuang, J., Ogata, Y., Vere-Jones, D., 2002. Stochastic declustering of space-time
hazard and risk in Bhutan. Nat. Hazards. https://doi.org/10.1007/s11069-020- earthquake occurrences. J. Am. Stat. Assoc. https://doi.org/10.1198/
04275-3. 016214502760046925.
18

Kpseis 1

Uploaded by

Copyright:

Available Formats

Kpseis 1

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Kpseis 1

Uploaded by

Copyright:

Available Formats

Physics and Chemistry of the Earth 123 (2021) 103029

Contents lists available at ScienceDirect

Physics and Chemistry of the Earth

1. Introduction Historically, the study region is strongly affected by earthquakes

In order to completely represent source seismicity through a recur

Fig. 7. On the left, the magnitude-frequency

Dauki 40 300 3 100 2.5 5.00 2.00

existing ones) are periodically released. However, despite the presence

You might also like

Kpseis 1

Uploaded by

Copyright:

Available Formats

Kpseis 1

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Kpseis 1

Uploaded by

Copyright:

Available Formats

Physics and Chemistry of the Earth 123 (2021) 103029

Contents lists available at ScienceDirect

Physics and Chemistry of the Earth

1. Introduction Historically, the study region is strongly affected by earthquakes

In order to completely represent source seismicity through a recur­

Fig. 7. On the left, the magnitude-frequency

Dauki 40 300 3 100 2.5 5.00 2.00

existing ones) are periodically released. However, despite the presence

You might also like

In order to completely represent source seismicity through a recur