Earth and Planetary Physics

RESEARCH ARTICLE 2: 327–341, 2018

SOLID EARTH: SEISMOLOGY doi: 10.26464/epp2018030

Probabilistic seismic hazard assessment of Nepal using multiple

seismic source models
Md Moklesur Rahman1,2, and Ling Bai1,3*
1Key Laboratory of Continental Collision and Plateau Uplift, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing 100101, China;
2Department of Petroleum and Mining Engineering, Jessore University of Science and Technology, Jessore 7408, Bangladesh;
3Department of Geophysics, Stanford University, Stanford, California 94305-2215, USA

Abstract: The potential for devastating earthquakes in the Himalayan orogeny has long been recognized. The 2015 MW7.8 Gorkha, Nepal
earthquake has heightened the likelihood that major earthquakes will occur along this orogenic belt in the future. Reliable seismic hazard
assessment is a critical element in development of policy for seismic hazard mitigation and risk reduction. In this study, we conduct
probabilistic seismic hazard assessment using three different seismogenic source models (smoothed gridded, linear, and areal sources)
based on the complicated tectonics of the study area. Two sets of ground motion prediction equations are combined in a standard logic
tree by taking into account the epistemic uncertainties in hazard estimation. Long-term slip rates and paleoseismic records are also
incorporated in the linear source model. Peak ground acceleration and spectral acceleration at 0.2 s and 1.0 s for 2% and 10%
probabilities of exceedance in 50 years are estimated. The resulting maps show significant spatial variation in seismic hazard levels. The
region of the Lesser Himalaya is found to have high seismic hazard potential. Along the Main Himalayan Thrust from east to west beneath
the Main Central Thrust, large earthquakes have occurred regularly in history; hazard values in this region are found to be higher than
those shown on existing hazard maps. In essence, the combination of long span earthquake catalogs and multiple seismogenic source
models gives improved seismic hazard constraints in Nepal.
Keywords: Gorkha earthquake; probabilistic seismic hazard; peak ground acceleration; spectral acceleration; source models; logic tree
Citation: Rahman M. M., and Bai L. (2018). Probabilistic seismic hazard assessment of Nepal using multiple seismic source models. Earth
Planet. Phys., 2(4), 327–341. http://doi.org/10.26464/epp2018030

1. Introduction The Himalayan orogen has experienced many devastating earth-

quakes (Table 1), including the 2015 MW7.8 Gorkha, Nepal earth-
Due to the ongoing collision between the Indian and Eurasian
quake. Geodetic measurements (e.g., Stevens and Avouac, 2015)
plates since ~65 Ma (Ding L et al., 2005), the Himalayan mountain
have identified that the MHT is fully locked at the shallower por-
range has grown into the highest orogenic belt on Earth, experi-
tion in its interseismic period and the accumulation process of the
encing widespread seismicity. Situated in the central segment of
seismic strain continues until being released by large earthquakes
the Himalayan orogenic belt, Nepal is one of the most earth-
(Bollinger et al., 2016; Rajendran et al., 2015; Sapkota et al., 2013).
quake-prone countries in the world. The classical tectonic divi-
Previous studies have predicted that an earthquake of MW8.0 at
sions of this region from south to north are the Sub-Himalaya,
100 km east to the Kathmandu valley could potentially harm
Lesser Himalaya, Higher Himalaya, and Tethyan Himalaya (Yin A,
~1.3% of the population and damage 50% of the buildings (Wyss,
2006), which are separated by four major E-W trending faults: the
2005). The 2015 MW7.8 Gorkha earthquake caused about 9000
Main Frontal Thrust (MFT); the Main Boundary Thrust (MBT); the
casualties and destroyed over fifty thousands of buildings with
Main Central Thrust (MCT); and the South Tibetan Detachment
huge economic losses (Bilham, 2015). However, the Gorkha Earth-
System (STDS) (Figure 1). These surface faults have merged with
quake was considered to be smaller than the great earthquakes
the Main Himalayan Thrust (MHT), the detachment that separates anticipated in the central Nepal (Bilham, 2015; Hand and Pulla,
the underthrusting Indian plate from the overriding Himalayan 2015) and eventually heightened the likelihood of catastrophic fu-
orogeny (Nábělek et al., 2009). The convergence rate between the ture earthquakes along the Himalayan orogenic belt.
Indian and Eurasian plates is 30–40 mm/year towards the NNE
(Stevens and Avouac, 2015). About half of this rate is accumulat- Probabilistic seismic hazard assessment (PSHA) has long been em-
ing along the MHT (Bettinelli et al., 2006). ployed in determining appropriate building design code provi-
sions by many earthquake-resistant infrastructure projects (e.g.,
Correspondence to: L. Bai, [email protected]
IBC, 2006). The implementation of seismic design provisions for
Received 08 MAY 2018; Accepted 17 JUL 2018. buildings and other infrastructure is an effective way to reduce
Accepted article online 30 JUL 2018. seismic risk and consequently to avoid earthquake-related dis-
©2018 by Earth and Planetary Physics. asters. After the pioneer work of Cornell (1968), PSHA has been
328 Earth and Planetary Physics doi: 10.26464/epp2018030

80°0′0″E 81°0′0″E 82°0′0″E 83°0′0″E 84°0′0″E 85°0′0″E 86°0′0″E 87°0′0″E 88°0′0″E 89°0′0″E

31°0′0″N 31°0′0″N

30°0′0″N 30°0′0″N

TIBET

29°0′0″N 29°0′0″N

28°0′0″N 28°0′0″N
MW
4.0-5.0
5.1-6.0 Major Cities
27°0′0″N 6.1-7.0 Active faults 27°0′0″N
7.1-8.3
Historical earthquakes INDIA

BANGLADESH
80°0′0″E 81°0′0″E 82°0′0″E 83°0′0″E 84°0′0″E 85°0′0″E 86°0′0″E 87°0′0″E 88°0′0″E 89°0′0″E

Figure 1. Location map of the study area showing the spatial distribution of declustered earthquakes (red open circles) and some historical
earthquakes (MW>6.0) (black filled stars). The yellow star marks the location of the recent devastating 2015 MW7.8 Gorkha Earthquake. Brown
rectangles are seismic source zones numbered from Z1 to Z23. Purple solid lines are active faults modified from Styron et al. (2010). The
numbering of faults from f1–f8 is arbitrary. MFT: Main Frontal Thrust, MBT: Main Boundary Thrust, MCT: Main Central Thrust, STD: South Tibetan
Detachment

Table 1. List of major earthquakes (MW≥ 6.5) that have occurred in and around Nepal over the past 1,000 years
Date Latitude (°N) Longitude (°E) Location Magnitude (MW) Reference

June 7, 1255 27.70 85.30 Kathmandu 7.7 NOAA

June 6, 1505 29.50 83.00 Lo Mustang 8.2 A&D
Jul. 25, 1720 30.00 80.00 N Uttar-pradesh 7.5 A&D
Aug. 26, 1833 27.70 85.70 Nepal 7.6 A&D
Oct. 14, 1911 31.00 80.50 Tibet 6.8 ISC
Aug. 28, 1916 29.90 80.50 Uttaranchal 7.2 A&D
Jan. 15, 1934 27.55 87.09 Bihar-Nepal 8.2 S
May 27, 1936 28.50 83.50 Nepal 7.0 ISC
June 4, 1945 30.00 80.00 Nepal 6.5 ISC
Aug. 15, 1950 28.70 96.70 Assam 8.5 A&D
Sep. 4, 1954 28.30 83.80 Nepal 6.5 ISC
Sep. 20, 1988 26.72 86.63 Nepal 6.6 ISC
Sep. 18, 2011 27.81 88.20 Sikkim, India 6.9 ISC
April 25, 2015 28.28 84.79 Gorkha, Nepal 7.8 ISC
May 12, 2015 27.83 86.14 Nepal 7.1 ISC

Notes: NOAA: National Oceanic and Atmospheric Administration; A&D: Ambraseys and Douglas (2004); S: Sapkota et al. (2013); ISC:
International Seismological Centre.

widely studied on global, regional, and local scales (e.g., Bhatia et the complete part of the instrumental earthquake catalog is very
al., 1999; Chaulagain et al., 2015; Nath and Thingbaijam, 2012; Or- short (e.g., around 55–60 years for this study region) compared to
daz et al., 2014; Ram and Wang GX, 2013; Sawires et al., 2016; the time required for earthquake generation by tectonic pro-
Zhang PZ et al., 1999). Earthquake catalogs (both instrumental cesses (Bilham, 2013). This catalog limitation in PSHA causes low
and historical) are the primarily basis for PSHA studies. However, hazard value at some regions where great earhquakes have actu-

Rahman M M et al.: PSHA of Nepal

 Earth and Planetary Physics doi: 10.26464/epp2018030 329

ally occurred (Kijko et al., 2016). This limitation is evident from ical earthquake catalogs from the National Oceanic and Atmo-
some cases of big earthqukes, such as the 2008 Sichuan MW7.9, spheric Administration (NOAA), GEM Global Historical Earthquake
the 2010 Haiti MW7.0, and 2011 Japan MW9.0 earthquakes (Wang Archive (GEM-GHEA), and the National Seismological Centre of
ZM et al., 2012). In this study therefore we accumulate all avail- Nepal (NSC). Instrumental earthquake catalogs between 1906 and
able historical and instrumental earthquake catalogs to assess the 2016 are collected from the U.S. Geological Survey (USGS) and In-
probabilistic seismic hazard for this area. ternational Seismological Centre (ISC).
Seismic hazard maps for Nepal have been produced since the The catalogs from different sources are then compiled together
1990’s (e.g., Chaulagain et al., 2015; Rahman et al., 2018; Ram and and checked manually to remove duplicate events on the basis of
Wang GX, 2013; Zhang ZM et al., 1999). Most of these previous location, time, and magnitude. We convert the different types of
studies estimated the hazard value using only the complete part magnitude (e.g., body wave magnitude, surface wave magnitude,
of the catalogs (about 60 years of data), based on single seismic duration magnitude, and local magnitude) contained in the cata-
source models and old attenuation equations. A recent study of log into MW using the different empirical relationships shown in
the GEM Faulted Earth Project characterized the attribute of the the Table 2. Approximately 2100 events with MW≥4.0 are recor-
Himalayan Frontal Thrust (MFT) fault in the seismic hazard assess- ded in the comprehensive catalog. The seismicity model usually
ment for the frontal Himalaya (Berryman et al., 2014). In this study, assumes that the occurrence of an earthquake is independent. All
we use a methodology developed by Kijko and Smith (2012) to foreshocks and aftershocks are therefore excluded from the cata-
compute seismicity parameters (e.g., Gutenberg-Richter b value, log, deploying a declustering process following the algorithm of
rate of seismicity). In general, PSHA procedures assume that seis- Gardner and Knopoff (1974). A total of 900 events (including 13
micity parameters are constant over time. But study of earth- historical earthquakes with magnitude equal to or greater than
quake catalogs has shown that these parameters may change 6.0) are included in the final catalog (Figure 1).
over time (Kijko et al., 2016). Thus we calculate the b value and
seismicity rate by taking into account the incompleteness of the Assessment of different levels of completeness of various sub-
earthquake catalog since the 12th century. Moreover, three seis- catalogs for the instrumental earthquake catalog is very import-
mogenic source models (i.e. the smoothed gridded seismicity, the ant. The reason is that records of large earthquakes are generally
linear source model, and the areal source model) and two sets of complete for longer periods while the catalogs of smaller earth-
ground motion prediction equations (GMPEs) are combined us- quake are complete for shorter durations. The completeness of
ing a standard logic tree structure for capturing model-related each earthquake catalog is governed by such considerations of
epistemic uncertainties. The topographic elevation in the areal socioeconomic and historical circumstances, demographic vari-
source and smoothed gridded seismicity models are also taken in- ations, and seismic station coverage. Failure to acknowledge the
to account, while paleoseismic and geodetic records are incorpor- incompleteness of the catalogs would result in overestimate and
ated into the linear source model. A set of seismic hazard maps for underestimate of recurrence rates for small and large earth-
peak ground acceleration (PGA) and spectral acceleration (SA) are quakes, respectively. We first assume a threshold magnitude (M0)
produced in this analysis for a better understanding of the spatial of MW4.0 and then estimate completeness by following the visual
variation of seismic hazard. cumulative method (Tinti and Mulargia, 1985). The modern instru-
mental catalog is found to be complete since 1964 (Figure 2a). In
utilizing the incomplete instrumental earthquake catalog since
2. Data and Methods 1906, the entire catalog is divided into different sub-catalogs
2.1 Earthquake Catalog based on different levels of completeness. The complete time
Earthquake catalog provides the basic data for delineating seis- window for various magnitude intervals (e.g., 4.0–5.0; 5.0–6.0;
mogenic sources and computing the seismicity parameter, espe- 6.0–7.0) is assessed using the statistical method of Stepp (1972).
cially the mean seismic activity rate λ, the Gutenberg-Richter b The complete time windows are found for different levels of com-
value, and the maximum expected earthquake magnitude Mmax. pleteness of 4.0, 5.0 and 6.0 to be after 1964, 1925–1964, and
We consider the incompleteness of the earthquake catalog which 1906–1925, respectively (Figure 2b).
includes both historical earthquakes and the instrumental catalog,
with different levels of completeness (Kijko and Smit, 2012; Kijko 2.2 Earthquake Source Models
et al., 2016). Ambraseys and Douglas (2004) and Szeliga et al. Proper seismic source zoning is essential in PSHA. Seismogenic
(2010) provide information regarding historical major earth- sources with different characteristics are generally delineated ac-
quakes for Himalaya and adjacent regions. We also use the histor- cording to tectonic settings, seismological and geological attrib-

Table 2. Magnitude scaling relations used for conversion of different types of magnitude into MW
(Scordilis, 2006) (Grünthal and Wahlström, 2003) (Kaviris et al., 2008)

MW = 0.67 MS + 2.07 for 3.0≤MS≤6.1 MW = MD + 0.5 (MD<3.0)

MW = 0.99 MS + 0.08 for 6.2≤MS≤8.2 MW = 0.67(±0.11) + 0.56(±0.08) ML + 0.046(±0.013) ML2 MW = MD + 0.6 (MD≥3.0)

MW = 0.85 Mb + 1.03 for 3.5≤Mb≤6.2

Notes: MW: Moment magnitude; MS: Surface wave magnitude; ML: Local magnitude; Mb: Body wave magnitude; MD: Duration magnitude.

Rahman M M et al.: PSHA of Nepal

330 Earth and Planetary Physics doi: 10.26464/epp2018030

3500
(a)
3000

Accumulated number of events

2500

2000

1500

1000

500

0
1926 1936 1946 1956 1966 1976 1986 1996 2006 2016
Year
10
(b)
Standard deviation (σλ) of λ

0.1 Magnitude bins

4-4.5
4.6-5
5.1-5.5
5.6-6
>6
0.01
5 50 500
Time (year)

Figure 2. (a) Completeness verification for M0=4.0; and (b) Different complete time windows for corresponding magnitude intervals (solid
tangent lines show the constant standard deviation and vertical dashed lines show the length of the time interval)

utes, and fault characteristics. The Himalayan orogen is regarded Himalayan-Tibetan region, Styron et al. (2010) identified active
as a very complex earthquake producing environment. We thus faults using remote sensing imagery, aerial photographic inter-
use a model that combines three different seismic sources: areal, pretation, paleoseismic studies, and field investigations. For our
linear, and smoothed gridded sources. study region, 14 active faults are assigned as linear seismic
sources (Figure 1). Apart from the MFT, there are no well-con-
2.2.1 Areal Seismic Sources strained attributes for the other faults. Therefore, two magnitude-
In general, the area of a seismic source zone is delineated using area and magnitude-length scaling empirical relationships of Yen
the basic principle that seismicity within a single source zone is YT and Ma KF (2011) and Wells and Coppersmith (1994) are ap-
uniform and homogenous. It is therefore assumed that every plied to characterize these faults’ attributes. The compilation and
point within the source has an equal potential to generate a fu- evaluation of approximately 72 such kinds of relationships were
ture earthquake (Baker, 2015). Recent studies (Chaulagain et al., carried out in the GEM Faulted Earth and Regionalisation Com-
2015; Ram and Wang GX, 2013) have delineated 23 areal seismic ponents project (Stirling and Goded, 2012). They found that the
source zones (Figure 1) for the study region based on quantitat- relationship of Yen YT and Ma KF (2011) is better in quality be-
ive analysis of the earthquake distribution, fault information, and cause of its updated wide-spread datasets with greater mag-
seismotectonic settings. In this study, the same (23) seismic areal nitude coverage than that of Wells and Coppersmith (1994) who
source zones are utilized. High seismicity clusters are present in
used older datasets and relatively narrow magnitude coverage.
western (zones 16, 17 and 18) and east central (zones 9, 10, 11 and
They recommended the quality score of 1 (best available) and 2
12) Nepal. The frontal edge of the Himalaya along the boundary of
(good) for Yen YT and Ma KF (2011) and Wells and Coppersmith
southern Nepal shows sparse seismicity.
(1994), respectively. Accordingly, we assign the weight for Yen YT
and Ma KF (2011) and Wells and Coppersmith (1994) as two-thirds
2.2.2 Linear Seismic Sources
and one-third, respectively. For large magnitude earthquakes, the
Usually, major earthquakes are associated with active faults, and
occurrence rate along these faults is determined using long-term
in seismic hazard assessment these active faults are considered as
slip rates as:
linear seismic sources. Although, there are numerous methods for
earthquake source simulation (e.g., Zhang WB et al., 2006), for the Recurrence Interval = M0 /µu̇LW, (1)

Rahman M M et al.: PSHA of Nepal

 Earth and Planetary Physics doi: 10.26464/epp2018030 331

where μ is shear modulus 3.0 × 1011 dyne/cm2, L is rupture length, fault locations, we use a zone-free approach for the spatial
W is rupture width, u̇ is the fault slip rate, and M0 is the seismic smoothing of seismicity (Frankel, 1995; Woo, 1996) to avoid sub-
moment. The relationship between the seismic moment and mag- jectivity in delineating areal seismic sources. This approach is
nitude (M) is that of Hanks and Kanamori (1979): termed “kernel estimation”; it implements spatial smoothing of
log (M0 ) = 1.5M + 16.05. (2) seismicity and depicts the seismicity’s non-uniform spatial distri-
bution, by assuming that seismicity varies from place-to-place. For
Small scale fault attributes (e.g., dip, dip direction, slip, and slip smoothed gridded seismicity, thousands of 0.1º × 0.1º grid cells
rate) are not well documented for this study region. We therefore are made for the entire study area. At each grid cell, earthquakes
assign to small-scale faults the same values for the above men- of MW≥4.0 are counted for different magnitude intervals in estim-
tioned parameters as those of nearby known faults (MFT). The ating the seismicity rates for corresponding magnitude intervals.
Main Frontal Thrust (MFT) is one of the well-constrained active The final seismic activity rate is then obtained by smoothing the
faults in this study area (e.g., Ader et al., 2012). The GEM faulted calculated seismicity rates using the multivariate kernel probabil-
earth project (Berryman et al., 2014) characterized the different at- ity density function following the equation of Woo (1996) as:
tributes of the MFT for all of the three different segments in the ∑
n j e−∆i j /c
2 2

study area. In this study, we used the same attributes (e.g., rup-
n̂i = ∑
j
ture length, width, dip, dip direction, slip, slip rate, and expected
, (3)
e−∆i j /c
2 2

maximum magnitude) for the three segments of MFT (i.e. MFT-1, j

MFT-2, and MFT-3, respectively from west to east).

where nj is the rate of seismicity in the j-th grid cell, n̂i is the
We deployed two magnitude-frequency distribution models in smoothed rate of seismicity in the i-th cell, c is the correlation dis-
the linear source model: the Characteristic-earthquake model for tance accounting for location uncertainties, and Δij is the distance
the empirically estimated earthquakes with magnitude equal or between the i-th and j-th cells. The sum is taken over cells j within
higher than MW≥6.5, and the Gutenberg-Richter model for instru- a distance of 3c of cell i. It is important to note that the resulting
mentally recorded earthquakes. The computed linear source para- smoothed seismicity will be concentrated around the epicenters
meters (e.g., seismic moment, slip rates, rupture length, rupture) of the recorded earthquakes for very small c values, while larger c
for active faults are illustrated in Table 3 for the Himalayan region. values spread out the seismicity and eventually do not reflect the
true spatial variation of seismicity as assumed in the zone-free ap-
2.2.3 Smoothed Gridded Sources proach. Therefore, in this study, the c value of 0.5° is assigned, fol-
As earthquake hypocenters are widely scattered in relation to lowing Frankel (1995).

Table 3. Active faults parameters used in this study

Slip rate
Fault Name Length (km) M 0* Width (km) b σb λ
(mm/year)

MFT-1 1300 2.67888E+22 100 20.5 0.76 0.09 4.334

MFT-2 330 1.6133E+21 100 18.0 0.61 0.04 2.703
MFT-3 800 9.86865E+21 100 20.0 0.73 0.09 1.371
Tibrikot 62 5.50155E+19 39 20.5 0.69 0.14 0.055
Humla 76 8.27582E+19 47 20.5 0.61 0.13 0.370
Tingri 44 2.77124E+19 30 18.0 0.51 0.1 0.233
Kung Co 68 6.62049E+19 43 18.0 0.44 0.09 0.090
Karakoram 300 1.3287E+21 110 10.0 0.63 0.09 0.460
f1 70 7.01679E+19 44 20.5 0.54 0.12 0.154
f2 55 4.32837E+19 36 20.5 0.55 0.22 0.048
f3 65 6.04792E+19 41 20.5 1.2 0.12 0.164
f4 139 2.79178E+20 76 20.5 0.69 0.16 0.425
f5 92 1.21496E+20 54 18.0 0.71 0.1 0.223
f6 179 4.65654E+20 93 18.0 0.63 0.1 0.148
f7 103 1.52508E+20 60 18.0 0.48 0.1 0.066
f8 82 9.64029E+19 50 18.0 0.54 0.21 0.102

Notes: Fault numbering (f1–f8) is chosen arbitrarily since no naming appears on the source files. Rupture width for MFT-1–MFT-3 is taken from
Berryman et al. (2014); and for the rest of the faults is estimated using the relationship of Yen YT and Ma KF (2011). Slip rates are recorded from
Ader et al. (2012). λ, b value, and σb are used for the Gutenberg-Richter seismicity model. M0* is the weighted seismic moment.

Rahman M M et al.: PSHA of Nepal

332 Earth and Planetary Physics doi: 10.26464/epp2018030

2.3 Maximum Earthquake Magnitude and Focal Depth Priestley et al., 2008). In addition, hypocenter relocation results
The maximum expected earthquake magnitude (Mmax) for every from a nearby broadband seismic array found that the focal
seismic source zone is computed on the basis of historical (e.g., depths for the Gorkha earthquake sequence ranged primarily
1505 MW8.2 and 1934 MW8.2 earthquakes) and recent earthquake between 10 km and 20 km (Adhikari et al., 2015; Bai et al., 2016;
data, but also considering great earthquakes that occurred in Letort et al., 2016). Thus, in this analysis, the average focal depth
nearby seismic zones with similar tectonic settings. The estimate of 15 km is assigned as a constant depth for all the source models.
of Mmax is computed following the algorithm of Kijko and Singh It is essential to note that the topographic elevation of the study
(2011). This estimation takes into account the uncertainty in the area varies significantly from south to north. The elevation of the
earthquake magnitude determination. The estimated maximum source zones along the south frontal border is less than 1 km and
expected magnitude of an earthquake for each areal source zone, in the central and northern zones is about 3 km and 5 km, respect-
along with its standard deviation, is given in Table 4. The com- ively. Seismic intensity attenuates reasonably with respect to dis-
puted Mmax values vary between 6.5 and 8.6. For smoothed grid- tance from the source and thereby the topography has a consid-
ded seismicity, the same range of Mmax is used for the respective erable effect on hazard value. Thus, the topographic elevation is
grids. considered for each source zone along with the assigned average
Accurate determination of earthquake location along the moun- focal depth.
tain belt is usually difficult because of insufficient station cover-
age. The ISC catalog reveals that earthquakes are distributed 2.4 Seismicity Model and Parameters
throughout the thickness of the Tibetan and Indian crust ranging To estimate the earthquake magnitude exceedance rates, a Modi-
between 0 and 90 km. Moreover, some recent studies have found fied Gutenberg-Richter (MGR)-Poisson model is chosen as a seis-
that the hypocenters of most of the earthquakes in Nepal are loc- micity model for all seismogenetic sources except characteristic
ated at depths ranging between 0 and 30 km (Pandey et al., 1999; earthquakes. Herein, the seismicity is expressed as (Cornell and

Table 4. Computed seismicity parameters for areal source zones

Zones a-value b-value σb* λ* Mu* σ(Mu)*

Z1 4.48 0.85 0.20 0.086 6.5 0.86

Z2 4.16 0.77 0.18 0.115 6.5 0.57
Z3 3.40 0.70 0.33 0.013 6.5 0.49
Z4 4.22 0.82 0.12 0.060 6.5 0.29
Z5 4.10 0.82 0.12 0.033 6.5 0.47
Z6 4.84 0.82 0.12 0.020 7.2 0.86
Z7 3.22 0.47 0.20 0.033 7.5 0.72
Z8 4.84 0.82 0.12 0.019 7.7 1.00
Z9 4.04 0.70 0.07 0.157 8.1 0.30
Z10 3.75 0.56 0.05 0.137 8.5 0.46
Z11 4.23 0.60 0.06 0.240 8.5 0.45
Z12 4.16 0.53 0.04 0.240 8.5 0.70
Z13 4.46 0.80 0.17 0.187 7.0 0.34
Z14 4.31 0.66 0.11 0.198 8.0 0.93
Z15 4.10 0.66 0.17 0.138 8.5 0.43
Z16 4.40 0.68 0.06 0.493 8.5 0.56
Z17 4.67 0.64 0.05 0.482 8.6 0.51
Z18 4.63 0.69 0.08 0.243 8.5 0.31
Z19 3.98 0.71 0.14 0.068 6.5 0.36
Z20 4.48 0.77 0.15 0.081 7.0 0.44
Z21 4.96 0.82 0.12 0.080 6.5 0.36
Z22 4.51 0.63 0.09 0.362 6.5 0.30
Z23 4.38 0.64 0.09 0.201 6.6 0.30

Notes: σb* is the standard error of b value; λ* is the rate of seismicity per year; σ(Mu)* is the standard deviation of expected maximum magnitude
(Mu*).

Rahman M M et al.: PSHA of Nepal

 Earth and Planetary Physics doi: 10.26464/epp2018030 333

Van Marke, 1969): Once the β̂ -value is known, then the average annual rate of seis-
exp (−βM) − exp (−βMmax ) micity (λ0) at threshold magnitude for each seismic source zone is
λ (M) = λ0 , M0 ≤ M ≤ Mmax (4) computed using the joint maximum likelihood function (Kijko and
exp (−βM0 ) − exp (−βMmax )
Smit, 2012) as:
where λ0 is the rate of earthquake activity with threshold mag- n
λ̂ (mmin ) = ∑s [ ( )] . (7)
nitude M0 (here M0 = 4.0), β is a parameter equivalent to the b
ti exp −β̂ mimin − mmin
value (β = 2.303×b), and Mmax is the maximum expected mag- i=1

nitude for the sources. In computing the λ(M) value, the uncer-
tainty of both Mmax and β are taken into account. 2.5 Selection of Ground Motion Prediction Equations
The ground motion prediction equations (GMPEs), also known as
Conventionally, seismicity parameters are calculated based on the
attenuation relationships, are essential in estimating the ground
minimum magnitude of completeness (MC) of the catalog. The
motion parameters (e.g., PGA, SA). These equations usually give
Maximum Curvature (MAXC) method of Wiemer and Wyss (2000)
ground motion intensity as a function of many variables such as
is used to evaluate the MC value. The estimated MC value of MW4.0
earthquake magnitude, source-to-site distance, and soil condition.
is taken as the M0 because earthquakes smaller than MW4.0 are
In general, the attenuation relationships are obtained empirically
not likely to cause any damage to infrastructure, from an engin-
using a statistical regression method on a particular set of strong
eering point of view.
motion data.
In this study, the essential seismicity parameters are computed by
The seismotectonic setting of the Himalayan orogenic belt is very
taking into account the incompleteness of the instrumental cata-
complex. This study area is a tectonically active interplate region
log and the historical earthquake records. The Gutenberg-Richter
(Nath and Thingbaijam, 2012) and the earthquake occurrence pat-
b value and its standard error for every source zone are determ-
tern of this region is not simple and even rather compound in
ined following the joint maximum likelihood function of Kijko and combination. For example, Nath and Thingbaijam (2012) identi-
Smit (2012) based on the historical and instrumental earthquake fied that this study region belongs to the active shallow crustal re-
catalog. This new approach is the extension of the Aki-Utsu (Aki, gime (ACR) with many subduction zone interface (SZI) earth-
1965; Utsu, 1965) b value estimator which takes into account both quakes across the Himalaya. Earthquake focal depth and mechan-
incompleteness of the earthquake catalogs and temporal vari- ism study (Bai et al., 2017) stated that the occurrence of large
ation of seismicity. It provides straightforward approximations for earthquakes is fairly associated with subduction interface in this
the standard errors and confidence intervals (Kijko and Smit, region. Moreover, it was assumed by some previous seismic haz-
2012). In this method, the whole span of the earthquake catalog is ard studies that the Himalaya and its surrounding region belongs
divided into different sub-catalogs and the parameter, β (β = to either SIZ or ACR (e.g., Ram and Wang GX, 2013), or both (e.g.,
2.303×b) is calculated as: Chaulagain et al., 2015). In avoiding the controversy, we consider
∏
s ∏
ni ( ) ∏s ∏
ni [ ( )] that the study region behaves as both of these environments.
L (β) = f mij , β = β exp −β mij − mimin , (5) However, the strong ground motion data for the Tibetan-Himalay-
i=1 j=1 i=1 j=1 an region are insufficient and consequently no particular GMPEs
have been studied previously.
where mji is the sample of ni earthquake magnitudes recorded
during the time span of the i-th sub-catalog (i=1, 2, …, s). In addi- Usually, this problem is resolved by selecting the GMPEs de-
tion, maximization of the likelihood function in equation (5) veloped in other regions of similar seismotectonic settings, which
provides the generalized Aki–Utsu β̂ -value estimator as: could represent sufficiently the ground motion in this region.
( )−1 Therefore, a total of 8 GMPEs are selected in this analysis accord-
r1 r2 rs ing to the criteria (e.g., tectonic regime, functional form, regres-
β̂ = + +...+ , (6)
β̂1 β̂2 β̂ s sion coefficients, frequency range of the model, journal types,
ni ∑s datasets, etc.) of Cotton et al. (2006), and a number of trellis plots
where ri = ; n = i ni is the total number of events with mag- (i.e. predicted spectral acceleration (PSA) versus magnitude, PSA
n
nitudes equal to or exceeding the relevant level of completeness, versus distance, PSA versus structural periods) as Stewart et al.
and β̂i are the Aki–Utsu estimators of the β-values, calculated for (2013). Two Next Generation Attenuation (NGA-west2) models:
individual sub-catalogs i (i= 1, 2, …, s) according to the classic Abrahamson et al. (2014) and Chiou and Youngs (2014) from the
Aki–Utsu formula. global datasets; one model from Europe and Middle East datasets
of Ambraseys et al. (2005); and other one from mainly Japanese
However, in this analysis, the b value and its standard errors are datasets of Zhao JX et al. (2006) are selected for ACR. The weight
estimated for the whole region, as well as for each areal source assessment and thereby the ranking of different GMPEs are not
zone. For the data of the whole region, the estimated b value is carried out because of scarcity of strong motion records in the Hi-
0.82 and the standard error is 0.12. These values are assigned to malayan-Tibetan region. Due to insufficient observed data, many
zones 4, 5, 6, 8, and 21 because of low seismicity. On the other seismic hazard studies for Asian countries (e.g., Chaulagain et al.,
hand, for rest of the zones, the individual b value and its standard 2015; Kolathayar and Sitharam, 2012; Nath and Thingbaijam,
error are assigned for hazard computation (Table 4). In case of 2012; Ornthammarath et al., 2011) used equal weight for all mod-
smoothed gridded sources, the b value and its standard error of els. In this study, therefore, for each of these four models, equal
0.82 and 0.12 are employed to all of the grids. probabilistic weights (1/4) are assigned. Besides, three models

Rahman M M et al.: PSHA of Nepal

334 Earth and Planetary Physics doi: 10.26464/epp2018030

from global datasets covering the large distance, namely BC Hy- city parameters, the seismic hazard module CRISIS2015 (Ordaz et
dro 2016 known as Abrahamson et al. (2016), Atkinson and Boore al., 2015) is used. This module has high efficiency in hazard calcu-
(2003), and Youngs et al., (1997); and one for local dataset of Zhao lation and flexibility in model selection (Danciu et al., 2010). For
JX et al. (2006) are selected for SIZ. Equal probabilities (1/4) of areal and linear seismic sources, the principle of CRISIS in comput-
weight to each of the models are assigned, as for ACR. ing the seismic hazard is based on the standard Cornell approach
(Cornell, 1968). The assumption of the Module is that within a
Compatibility among these selected GMPEs is checked when ap-
source zone, seismicity is evenly distributed by unit area and
plied in the CRISIS2015 (Ordaz et al., 2015) program. Finally, these
thereby all the points might be a potential earthquake focus. A
two sets of GMPEs are combined together using a logic tree. In
spatial integration by subdividing the original sources is per-
the similar way, GMPEs are employed for all three source models.
formed by the CRISIS program in obtaining accurate hazard value.
Once a source is subdivided into sub-sources, the acceleration ex-
2.6 Logic Tree Structure ceedance rate due to the i-th single source can be computed us-
Uncertainties are commonly associated with different models ing the equation:
used in PSHA. The application of a logic tree allows the capturing
∑ ∫ ( ) ( )
Mmax
−dλ (M)
of epistemic uncertainties in different input models (Bommer et νi(a) = Wi j Pr A > a M, Ri j dM, (8)
al., 2005; Sabetta et al., 2005) by deploying alternative models in j M0 dM
the hazard estimation. In this study, three types of seismogenic
source models and two sets of GMPEs (each consisting of four dif- where M0 and M(max are the threshold
) and maximum magnitudes,
ferent GMPEs) are combined using the logic tree. Equal weights respectively; Pr A > a M, Ri j is the probability that the accelera-
(1/3) are assigned to each of the source models, as there is no sub- tion exceeds a certain value a at the site across distance Rij for an
stantial observed data for the study region in evaluating the rank- earthquake with magnitude M; and Wij is the weight of each sub-
ing of the weight. For the linear source model, equal weight (1/2) element. The above expression also assumes that ∑Wij = 1. The
is assigned for the Characteristic Earthquake and Gutenberg- total average acceleration exceedance rate at the site due to the
Richter seismicity models as there is no reason to prefer one mod- contributions of all the sources, N, within 300 km is calculated as:
el over the other. Figure 3 displays the standard logic tree frame- ∑N ∫ ( )
Mmax
−dλ (M)
work used for the linear source model in this analysis. ν(a) = Pr(A > a M, Ri j )dM. (9)
(n=1)
M0 dM

2.7 Seismic Hazard Estimation For smoothed gridded seismicity, once the seismicity parameters
To estimate the seismic hazards on the basis of computed seismi- and the GMPEs are known for each of the nodes of the grids, the

Seismicity Model Earthquakes Regime GMPEs

Abrahamson et al. (2014)

0.25
Ambraseys et al. (2005)
Active Shallow Crustal 0.25
0.50 Chiou & Youngs (2014)
0.25
Zhao et al. (2006)
MGR 0.25
0.50 Abrahamson et al. (2016)
0.25

0.50 Atkinson & Boore (2003)

0.25
Linear Source

Subduction Interface Youngs et al. (1997)

0.25
Zhao et al. (2006)
0.25

CE As above As above
0.50

MGR=Modified Gutenberg-Richter; CE=Characteristic Earthquake

Figure 3. The logic tree structure for the active linear seismogenic source model.

Rahman M M et al.: PSHA of Nepal

 Earth and Planetary Physics doi: 10.26464/epp2018030 335

hazard at a given node is calculated based on the effects of the to- ard value at each grid cell is computed using all of the eight differ-
tality of the nodes and their corresponding distances to the site of ent attenuation models for all three seismogenic sources separ-
interest as in equations (8) and (9). ately. Afterward, the final hazard value at each grid cell is ob-
tained by the combination of these individual values, using the lo-
gic tree structure. This analysis produces different seismic hazard
3. Results
maps for PGA value and SA value at short (0.2 s) and long (1.0 s)
For seismic hazard estimation, the entire study region is divided
natural periods at 5% critical damping factor for 10% and 2%
into small grids with the size of 0.1º×0.1º and the total number of
probabilities of exceedance in 50 years, corresponding to 475 and
grid cells is approximately 5000. At the center of each grid cell, the
2,475 year return periods, respectively. PGA hazard maps for the
PGA and SA values are computed for a referenced bedrock condi-
probabilities of exceedance of 10% and 2% in 50 years are shown
tion. The contribution of all the potential sources to hazard value
in Figure 4a, b. Figure 5a, b illustrates the spatial variations of SA
within the radial distance of 300 km from the center of each grid
value obtained at 0.2 s and 1.0 s natural periods for 2% probabil-
cell is taken into account. We estimate the hazard value by disag-
ity of exceedance.
gregating the hypocentral distance into small intervals of 1 km,
and the magnitude range (between the minimum and maximum The overall spatial distribution of hazard value regarding both
magnitude) into small incremental values of 0.1. Initially, the haz- PGA and SA is non-uniform. Because of heterogeneous character-

81°0′0″E 82°0′0″E 83°0′0″E 84°0′0″E 85°0′0″E 86°0′0″E 87°0′0″E 88°0′0″E

(a)

30°0′0″N 30°0′0″N

29°0′0″N 29°0′0″N

28°0′0″N 28°0′0″N

PGA (g)
27°0′0″N 27°0′0″N

0.21 0.29 0.38 0.47 0.55 0.64

81°0′0″E 82°0′0″E 83°0′0″E 84°0′0″E 85°0′0″E 86°0′0″E 87°0′0″E 88°0′0″E

(b)

30°0′0″N 30°0′0″N

29°0′0″N 29°0′0″N

28°0′0″N 28°0′0″N

PGA (g)
27°0′0″N 27°0′0″N
0.40 0.52 0.65 0.77 0.90 1.02

81°0′0″E 82°0′0″E 83°0′0″E 84°0′0″E 85°0′0″E 86°0′0″E 87°0′0″E 88°0′0″E

Figure 4. Spatial variation of PGA values for (a) 10% probability of exceedance; and (b) 2% probability of exceedance in 50 years.

Rahman M M et al.: PSHA of Nepal

336 Earth and Planetary Physics doi: 10.26464/epp2018030

81°0′0″E 82°0′0″E 83°0′0″E 84°0′0″E 85°0′0″E 86°0′0″E 87°0′0″E 88°0′0″E

(a)

30°0′0″N 30°0′0″N

29°0′0″N 29°0′0″N

28°0′0″N 28°0′0″N

SA (g)
27°0′0″N 27°0′0″N

0.87 1.17 1.47 1.78 2.08 2.38

81°0′0″E 82°0′0″E 83°0′0″E 84°0′0″E 85°0′0″E 86°0′0″E 87°0′0″E 88°0′0″E

(b)

30°0′0″N 30°0′0″N

29°0′0″N 29°0′0″N

28°0′0″N 28°0′0″N

SA (g)
27°0′0″N 27°0′0″N

0.29 0.47 0.65 0.83 1.01 1.19

81°0′0″E 82°0′0″E 83°0′0″E 84°0′0″E 85°0′0″E 86°0′0″E 87°0′0″E 88°0′0″E

Figure 5. Hazard maps of spectral acceleration at (a) 0.2 s natural period; and (b) 1.0 s natural period for 2% probability of exceedance in 50
years.

istics of the seismotectonic settings at the study region, there are ern-most part along the Nepal-India boundary is identified as a
clear lateral variations in the computed seismic hazard maps. The low seismic hazard zone. MHT has long been predicted as a po-
predicted hazard distribution pattern largely follows the back- tential source for producing great earthquakes in the Himalayan
ground seismicity distribution across the study area. The regions region (Avouac et al., 2015; Rajendran et al., 2015; Sapkota et al.,
of highest potential hazard are likely distributed along strike of 2013). Large earthquakes in the central portion of the Himalayan
the Indian-Eurasian plate collision zones. However, the hazard es- orogenic belt may greatly affect the densely-populated cities of
timate for 10% probability of exceedance shows a wide range of Nepal, such as Kathmandu and Pokhara (Bollinger et al., 2016).
values, and the minimum and maximum hazard values are found
as low as 0.21g, and as high as 0.64g, respectively (Figure 4a). The Nonetheless, the predicted seismic hazard value is increased sig-
zones of highest hazard are identified in the far western and cent- nificantly with decreasing probability of exceedance. For example,
ral to northeast central Nepal (Figure 4a) where the rate of seismi- at 2% probability of exceedance in 50 years, the computed haz-
city is very high compared to other parts of Nepal (Figure 1). For ard level for central and northeast central Nepal is about a factor
far western Nepal (i.e. south of Darchula city), we obtain peak haz- of 1.8 higher than that of 10% in 50 years; for far western Nepal,
ard values as high as 0.64g. In central Nepal around the Kath- the comparable factor is nearly 1.5 (Figure 4b). The maximum and
mandu valley, peak hazard values range up to 0.59g. The south- minimum PGA value for this case is obtained as low as 0.40g and

Rahman M M et al.: PSHA of Nepal

 Earth and Planetary Physics doi: 10.26464/epp2018030 337

as high as 1.02g, respectively. We produce the SA maps (Figure 5a, smoothed gridded and areal source models are 0.71g and 0.54g,
b) as per the requirement of the modern building design provi- respectively. We also find that the effect of sources on hazard
sions, which are analogous with the PGA maps in terms of hazard value varies spatially. In regions with high hazard, all the source
distribution pattern. SA value at 0.2 s and 1.0 s natural periods for models (areal, smoothed gridded, and linear) produce nearly sim-
2% probability of exceedance in 50 years is found in the order of ilar ground motion values (Figure 6). However, for northern-most,
0.87–2.38g, and 0.29–1.19g, respectively. southern-most, and the central west regions where background
seismicity is low, different levels of effect are observed among the
To understand how different source models contribute to the models. For example, in the southern and central west regions of
computed seismic hazard values, we estimate the hazard value for Nepal (e.g., around the cities of Bara, and Rukum), the linear
each model individually. The resulting hazard maps for these source model contributes comparatively higher seismic hazards
source models show that the levels of maximum and minimum than the other two source models, and consequently augments
hazard value are quite different from model to model. For in- the combined hazard value levels for these regions (Figure 6). On
stance, the linear source model produces maximum values as the other hand, in northern Nepal (near Gorkha city), the contribu-
high as 0.60g whereas the maximum hazard values for the tion from areal and smoothed gridded sources is comparatively

Combined Hazard
Combined Hazard Dhankuta Areal Source
1 Kathmandu Areal Source 1
Linear Source
Linear Source Smoothed Seismicity
Smoothed Seismicity

0.1 0.1
PE in 50 years

PE in 50 years

0.01 0.01

0.001 0.001
0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
1 Bara 1 Gorkha

0.1 0.1
PE in 50 years

PE in 50 years

0.01 0.01

0.001 0.001
0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
1 Kailali 1 Rukum

10% in 50 years 10% in 50 years

0.1 0.1
PE in 50 years

PE in 50 years

2% in 50 years 2% in 50 years

0.01 0.01

0.001 0.001
0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
PGA (g) PGA (g)
PE=Probability of Exceedance

Figure 6. Seismic hazard (PGA) curves for different cites of Nepal.

Rahman M M et al.: PSHA of Nepal

338 Earth and Planetary Physics doi: 10.26464/epp2018030

higher than that of the linear source model. In the southwest re- 4. Conclusions and Discussions
gion (near Kailali city), the highest contribution is from the To produce seismic hazard maps of Nepal, we incorporate the
smoothed seismicity source, and the areal source contributes the latest datasets, multiple seismogenic source considerations, up-
least hazard value, while the linear source model yields an inter- dated GMPEs, topographic elevation data, well constrained focal
mediate effect. Moreover, to infer the attenuation pattern of seis- depth data, and consideration of inherent uncertainties associ-
mic hazard with respect to various natural periods, the uniform ated with each of these inputs. An updated uniform declustered
earthquake catalog since 12th century is utilized. 23 areal sources,
seismic hazard spectra at 10% probability of exceedance in 50
14 linear sources and hundreds of smoothed gridded sources are
years for different cities is computed (Figure 7). The highest haz-
combined using a logic tree to account for the epistemic uncer-
ard value is obtained at the short natural period of 0.2 s for all cit- tainties. Essential seismicity parameters (e.g., b value and rate of
ies. With the increase of the natural periods, the hazard value falls seismicity, λ) are computed using the joint maximum likelihood
rapidly, and at the longer periods of about 1.5 s and above, the method (Kijko and Smit, 2012). The incompleteness of the catalog
values became nearly flattened and exhibit very gentle slopes of (i.e. different complete sub-catalogs) and the uncertainty in earth-
attenuation. quake magnitude determination are considered for seismicity

Combined Hazard Combined Hazard

1.5 Kathmandu Areal Source 1.5 Dhankuta Areal Source
Linear Source Linear Source
1.3 Smoothed Seismicity 1.3 Smoothed Seismicity

1.1 1.1

0.9 0.9
SA (g)

SA (g)
0.7 0.7

0.5 0.5

0.3 0.3

0.01 0.01
0.1 0.5 1.0 1.5 2.0 0.1 0.5 1.0 1.5 2.0
1.5 Bara 1.5 Gorkha

1.3 1.3

1.1 1.1

0.9 0.9
SA (g)

SA (g)

0.7 0.7

0.5 0.5

0.3 0.3

0.01 0.01
0.1 0.5 1.0 1.5 2.0 0.1 0.5 1.0 1.5 2.0
1.5 Kailali 1.5 Rukum

1.3 1.3

1.1 1.1

0.9 0.9
SA (g)

SA (g)

0.7 0.7

0.5 0.5

0.3 0.3

0.01 0.01
0.1 0.5 1.0 1.5 2.0 0.1 0.5 1.0 1.5 2.0
Natural Period (s) Natural Period (s)

Figure 7. Uniform hazard spectra for different cites at 475 years return period.

Rahman M M et al.: PSHA of Nepal

 Earth and Planetary Physics doi: 10.26464/epp2018030 339

parameters computation. The modified Gutenberg-Richter and ground motion relation for active crustal regions. Earthq. Spectra, 30(3),
Characteristic-earthquake magnitude-frequency models, two sets 1025–1055. https://doi.org/10.1193/070913EQS198M
Abrahamson, N., Gregor, N., and Addo, K. (2016). BC hydro ground motion
of GMPEs (four GMPEs for the active shallow crust and four GMPEs
prediction equations for subduction earthquakes. Earthq. Spectra, 32(1),
for the subduction interface) are utilized. In essence, the combina- 23–44. https://doi.org/10.1193/051712EQS188MR
tion of past seismicity data (both historical and instrumental), pa- Ader, T., Avouac, J. P., Liu-Zeng, J., Lyon-Caen, H., Bollinger, L., Galetzka, J.,
leoseismic and geodetic data, along with three sesimogenic Genrich, J., Thomas, M., Chanard, K., … Flouzat, M. (2012). Convergence rate
source models, thus provides more comprehensive and explain- across the Nepal Himalaya and interseismic coupling on the Main
able constraints to the computation of seismic hazard values. Himalayan Thrust: Implications for seismic hazard. J. Geophys. Res.: Solid
Earth, 117(B4), B04403. https://doi.org/10.1029/2011JB009071
The spatial variation pattern of the hazard value shows good Adhikari, L. B., Gautam, U. P., Koirala, B. P., Bhattarai, M., Kandel, T., Gupta, R. M.,
agreement with the maps of Zhang PZ et al. (1999), Ram and Timsina, C., Maharajan, N., Maharajan, K., … Bollinger, L. (2015). The
Wang GX (2013), and Rahman et al. (2018). In addition, the result- aftershock sequence of the 2015 April 25 Gorkha-Nepal earthquake.
ing hazard values for the Nepal area are comparable with results Geophys. J. Int., 203(3), 2119–2124. https://doi.org/10.1093/gji/ggv412
for the adjacent regions in India (e.g., Kolathayar and Sitharam, Aki, K. (1965). Maximum likelihood estimate of b in the formula logN=a−bM
and its confidence limits. Bull. Earthq. Res. Inst., 43(2), 237–239.
2012; Nath and Thingbaijam, 2012) and Tibet (Rahman et al.,
Ambraseys, N. N., Douglas, J., Sarma, S. K., and Smit, P. M. (2005). Equations for
2018). The rock-site PGA value around the epicenter of the 2015
the estimation of strong ground motions from shallow crustal earthquakes
Gorkha earthquake shown in previous studies (e.g., Chaulagain et using data from Europe and the middle east: horizontal peak ground
al., 2015; Ram and Wang GX, 2013; Zhang PZ et al., 1999) is below acceleration and spectral acceleration. Bull. Earthq. Eng., 3(1), 1–53.
0.4g. The hazard value for the source area of the Gorkha earth- https://doi.org/10.1007/s10518-005-0183-0
quake has been estimated to be low (Chaulagain et al., 2015; Ram Ambraseys, N. N., and Douglas, J. (2004). Magnitude calibration of north Indian
and Wang GX, 2013) based on the single source model and the earthquakes. Geophys. J. Int., 159(1), 165–206.
short span earthquake catalog. However, the PGA values ob- https://doi.org/10.1111/j.1365-246X.2004.02323.x
Atkinson, G. M., and Boore, D. M. (2003). Empirical ground-motion relations for
tained in this study are up to 0.45g at the same probability of ex-
subduction-zone earthquakes and their application to Cascadia and other
ceedance (i.e. 10%). We suggest that this difference is probably a
regions. Bull. Seismol. Soc. Am., 93(4), 1703–1729.
result of our consideration of multiple seismogenic sources along https://doi.org/10.1785/0120020156
with the wide span of our earthquake catalog including historic Avouac, J.-P., Meng, L. S., Wei, S. J., Wang, T., and Ampuero, J.-P. (2015). Lower
large earthquakes. edge of locked Main Himalayan Thrust unzipped by the 2015 Gorkha
earthquake. Nat. Geosci., 8(9), 708–711. https://doi.org/10.1038/ngeo2518
The hazard value is assessed in terms of PGA for 10% and 2%
Bai, L., Liu, H. B., Ritsema, J., Mori, J., Zhang, T. Z., Ishikawa, Y., and Li, G. H. (2016).
probabilities of exceedance on bedrock level. The spectral acceler- Faulting structure above the Main Himalayan Thrust as shown by relocated
ation distribution maps at 0.2 s and 1.0 s natural periods are pro- aftershocks of the 2015 Mw7.8 Gorkha, Nepal, earthquake. Geophys. Res.
duced for 2% probability of exceedance in 50 years. The region of Lett., 43(2), 637–642. https://doi.org/10.1002/2015GL066473
the Lesser Himalaya from east to west is found to be has high seis- Bai, L., Li, G. H., Khan, N. G., Zhao, J. M., and Ding, L. (2017). Focal depths and
mic hazard potential. The seismic hazards estimated in this study mechanisms of shallow earthquakes in the Himalayan–Tibetan region.
would help constrain the national building design provisions for Gondwana Res., 41, 390–399. https://doi.org/10.1016/j.gr.2015.07.009
Baker, J. W. (2015). Introduction to Probabilistic Seismic Hazard Analysis. White
the central segment of the Himalayan region. Better seismic haz-
Paper Version 2.1.
ard estimate in a tectonically active region requires detail pa-
Berryman, K., Ries, W., and Litchfield, N. (2014). The Himalayan Frontal Thrust:
leoseismic, geological, and geodetics studies. Strong ground mo- Attributes for seismic hazard Version 1.0. GEM Faulted Earth Project,
tion data is essential in order to develop the new attenuation rela- available from http://www.nexus.globalquakemodel.org/. Accessed 10 Apr
tionships for this tectonically active region. 2017
Bettinelli, P., Avouac, J.-P., Flouzat, M., Jouanne, F., Bollinger, L., Willis, P., and
Chitrakar, G. R. (2006). Plate Motion of India and Interseismic Strain in the
Acknowledgments Nepal Himalaya from GPS and DORIS Measurements. J. Geod., 80(8-11),
This research is supported by the grants of the National Nature 567–589. https://doi.org/10.1007/s00190-006-0030-3
Science Foundation of China (No. 41761144076, 41490611), the Bhatia, S. C., Kumar, M. R., and Gupta, H. K. (1999). A probabilistic seismic hazard
collaborative research program of the Disaster Prevention Re- map of India and adjoining regions. Ann. di Geofis., 42(6), 1153–1164.
search Institute of Kyoto University (No. 29W-03) and the COX vis- Bilham, R. (2015). Raising Kathmandu. Nat. Geosci., 8(8), 582–584.
iting professor fellowship of the Stanford University to L.B., and by https://doi.org/10.1038/ngeo2498
the Chinese Academy of Sciences (CAS) and The World Academy Bilham, R. (2013). Societal and observational problems in earthquake risk
assessments and their delivery to those most at risk. Tectonophys, 584,
of Sciences (TWAS) President’s PhD Fellowship to M.M.R.. We used
166–173. https://doi.org/10.1016/j.tecto.2012.03.023
earthquake catalogs from the ISC, USGS, GEM, NOAA and NSC. We
Bollinger, L., Tapponnier, P., Sapkota, S. N., and Klinger, Y. (2016). Slip deficit in
are grateful to Andrzej Kijko and his colleagues for providing the central Nepal: omen for a repeat of the 1344 AD earthquake?. Earth Planet.
“AUE” program, to Mario Ordaz for providing us the recently de- Space, 68, 12. https://doi.org/10.1186/s40623-016-0389-1
veloped version of the CRISIS software, and to Jack Baker and Bommer, J. J., Scherbaum, F., Bungum, H., Cotton, F., Sabetta, F., and
Fabrice Cotton for useful discussion. We thank the editor Abrahamson, N. A. (2005). On the use of logic trees for ground-motion
Xiaodong Song, Wenbo Zhang and an anonymous reviewer for prediction equations in seismic-hazard analysis. Bull. Seismol. Soc. Am., 95(2),
their constructive comments. 377–389. https://doi.org/10.1785/0120040073
Chaulagain, H., Rodrigues, H., Silva, V., Spacone, E., and Varum, H. (2015).
Seismic risk assessment and hazard mapping in Nepal. Nat. Hazards, 78(1),
References 583–602. https://doi.org/10.1007/s11069-015-1734-6
Abrahamson, N. A., Silva, W. J., and Kamai, R. (2014). Summary of the ASK14 Chiou, B. S. J., and Youngs, R. R. (2014). Update of the Chiou and Youngs NGA

Rahman M M et al.: PSHA of Nepal

340 Earth and Planetary Physics doi: 10.26464/epp2018030

model for the average horizontal component of peak ground motion and https://doi.org/10.1016/j.ijdrr.2014.05.004
response spectra. Earthq. Spectra, 30(3), 1117–1153. Ordaz, M., Faccioli, E., Martinelli, F., Aguilar, A., Arboleda, J., Meletti, C., and
https://doi.org/10.1193/072813EQS219M D’Amico, V. (2015). CRISIS2015 version 2.2: Computer program for
Cornell, C. A. (1968). Engineering seismic risk analysis. Bull. Seismol. Soc. Am., computing seismic hazard. Instituto de Ingenieria, UNAM, Mexico.
58(5), 1583–1606. Ornthammarath, T., Warnitchai, P., Worakanchana, K., Zaman, S., Sigbjörnsson,
Cornell, C. A, and Van Marke, E. H. (1969). The major influences on seismic risk. R., and Lai, C. G. (2011). Probabilistic seismic hazard assessment for Thailand.
In Proceedings of the Third World Conference on Earthquake Engineering (pp. Bull. Earthq. Eng., 9(2), 367–394. https://doi.org/10.1007/s10518-010-9197-3
69–73). Santiago, Chile. Pandey, M. R., Tandukar, R. P., Avouac, J. P., Vergne, J., and Héritier, T. (1999).
Cotton, F., Scherbaum, F., Bommer, J. J., and Bungum, H. (2006). Criteria for Seismotectonics of the Nepal Himalaya from a local seismic network. J.
selecting and adjusting ground-motion models for specific target regions: Asian Earth Sci., 17(5-6), 703–412. https://doi.org/10.1016/S1367-
Application to central Europe and rock sites. J. Seismol., 10(2), 137–156. 9120(99)00034-6
https://doi.org/10.1007/s10950-005-9006-7 Priestley, K., James, J., and Mckenzie, D. (2008). Lithospheric structure and deep
Danciu, L., Pagani, M., Monelli, D., and Wiemer, S. (2010). GEM1 Hazard : earthquakes beneath India, the Himalaya and southern Tibet. Geophys. J.
Overview of PSHA Software. GEM Technical Report 2010-2, Pavia, Italy: GEM Int., 172(1), 345–362. https://doi.org/10.1111/j.1365-246X.2007.03636.x
Foundation. Rahman, M. M., Bai, L., Khan, N. G., and Li, G. H. (2018). Probabilistic seismic
Ding, L., Kapp, P., and Wan, X. Q. (2005). Paleocene-Eocene record of ophiolite hazard assessment for Himalayan-Tibetan region from historical and
obduction and initial India-Asia collision, south central Tibet. Tectonics, instrumental earthquake Catalogs. Pure Appl. Geophys., 175(2), 685–705.
24(3), TC3001. https://doi.org/10.1029/2004TC001729 https://doi.org/10.1007/s00024-017-1659-y
Frankel, A. (1995). Mapping seismic hazard in the central and eastern United Rajendran, C. P., John, B., and Rajendran, K. (2015). Medieval pulse of great
States. Seismol. Res. Lett., 66(4), 8–21. https://doi.org/10.1785/gssrl.66.4.8 earthquakes in the central Himalaya: Viewing past activities on the frontal
Gardner, J., and Knopoff, L. (1974). Is the sequence of earthquakes in Southern thrust. J. Geophys. Res.: Solid Earth, 120(3), 1623–1641.
California, with aftershocks removed, Poissonian?. Bull. Seismol. Soc. Am., https://doi.org/10.1002/2014JB011015
64(5), 1363–1367. Ram, T. D., and Wang, G. X. (2013). Probabilistic seismic hazard analysis in Nepal.
Grünthal, G., and Wahlström, R. (2003). An MW based earthquake Catalogue for Earthq. Eng. Eng. Vibra., 12(4), 577–586. https://doi.org/10.1007/s11803-013-
central, northern and northwestern Europe using a hierarchy of magnitude 0191-z
conversions. J. Seismol., 7(4), 507–531. Sabetta, F., Lucantoni, A., Bungum, H., and Bommer, J. J. (2005). Sensitivity of
https://doi.org/10.1023/B:JOSE.0000005715.87363.13 PSHA results to ground motion prediction relations and logic-tree weights.
Hand, E., and Pulla, P. (2015). Nepal disaster presages a coming megaquake. Soil Dyn. Earthq. Eng., 25(4), 317–329.
Science, 348(6234), 484–485. https://doi.org/10.1126/science.348.6234.484 https://doi.org/10.1016/j.soildyn.2005.02.002
Hanks, T. C., and Kanamori, H. (1979). A moment magnitude scale. J. Geophys. Sapkota, S. N., Bollinger, L., Klinger, Y., Tapponnier, P., Gaudemer, Y., and Tiwari,
Res.: Solid Earth, 84(B5), 2348–2350. D. (2013). Primary surface ruptures of the great Himalayan earthquakes in
https://doi.org/10.1029/JB084iB05p02348 1934 and 1255. Nat. Geosci., 6(1), 71–76. https://doi.org/10.1038/ngeo1669
IBC. (2006). International Building Code. Washington DC: International Code Sawires, R., Peláez, J. A., Fat-Helbary, R. E., and Ibrahim, H. A. (2016). Updated
Council. probabilistic seismic-hazard values for Egypt. Bull. Seismol. Soc. Am., 106(4),
Kaviris, G., Papadimitriou, P., Chamilothoris, L., and Makropoulos, K. (2008). 1788–1801. https://doi.org/10.1785/0120150218
Moment magnitudes for small and intermediate earthquakes. In Scordilis, E. M. (2006). Empirical global relations converting MS and mb to
Proceedings of the 31st General Assembly of the European Seismological moment magnitude. J. Seismol., 10(2), 225–236.
Commission. ESC. https://doi.org/10.1007/s10950-006-9012-4
Kijko, A., and Singh, M. (2011). Statistical tools for maximum possible Stepp, J. C. (1972). Analysis of completeness of the earthquake sample in the
earthquake magnitude estimation. Acta Geophys., 59(4), 674–700. Puget Sound area and its effect on statistical estimates of earthquake
https://doi.org/10.2478/s11600-011-0012-6 hazard. In Proceedings of the 1st International Conference on
Kijko, A., and Smit, A. (2012). Extension of the aki-utsu b-value estimator for Microzonazion (vol. 2, pp. 897–910), Seattle.
incomplete catalogs. Bull. Seismol. Soc. Am., 102(3), 1283–1287. Stevens, V., and Avouac, J.-P. (2015). Interseismic coupling on the Main
https://doi.org/10.1785/0120110226 Himalayan Thrust. Geophys. Res. Lett., 42(14), 5828–5837.
Kijko, A., Smit, A., and Sellevoll, M. A. (2016). Estimation of earthquake hazard https://doi.org/10.1002/2015GL064845
parameters from incomplete data files. Part Ⅲ. Incorporation of uncertainty Stewart, J. P., Douglas, J., Javanbarg, M. B., Di Alessandro, C., Bozorgnia, Y.,
of earthquake-occurrence model. Bull. Seismol. Soc. Am., 106(3), 1210–1222. Abrahamson, N. A., Boore, D. M., Campbell, K. W., Delavaud, E., and Stafford,
https://doi.org/10.1785/0120150252 P. J. (2013). GEM-PEER Task 3 Project: Selection of a Global Set of Ground
Kolathayar, S., and Sitharam, T. G. (2012). Comprehensive probabilistic seismic Motion Prediction Equations. PEER Report 2013/22.
hazard analysis of the Andaman-Nicobar regions. Bull. Seismol. Soc. Am., Stirling, M., and Goded, T. (2012). Magnitude scaling relationships. Report
102(5), 2063–2076. https://doi.org/10.1785/0120110219 Produced for the GEM Faulted Earth & Regionalisation Global Componets,
Letort, J., Bollinger, L., Lyon-Caen, H., Guilhem, A., Cano, Y., Baillard, C., and GNS Science Miscellaneous Series, 42, 35.
Adhikari, L. B. (2016). Teleseismic depth estimation of the 2015 Gorkha- Styron, R., Taylor, M., and Okoronkwo, K. (2010). Database of active structures
Nepal aftershocks. Geophys. J. Int., 207(3), 1584–1595. from the Indo-Asian Collision. EOS, 91(20), 181–182.
https://doi.org/10.1093/gji/ggw364 https://doi.org/10.1029/2010EO200001
Nábělek, J., Hetényi, G., Vergne, J., Sapkota, S., Kafle, B., Jiang, M., Su, H. P., Chen, Szeliga, W., Hough, S., Martin, S., and Bilham, R. (2010). Intensity, magnitude,
J., Huang, B.-S., and the Hi-CLIMB Team. (2009). Underplating in the location, and attenuation in India for felt earthquakes since 1762. Bull.
Himalaya-Tibet collision zone revealed by the Hi-CLIMB experiment. Science, Seismol. Soc. Am., 100(2), 570–584. https://doi.org/10.1785/0120080329
325(5946), 1371–1374. https://doi.org/10.1126/science.1167719 Tinti, S., and Mulargia, F. (1985). An improved method for the analysis of the
Nath, S. K., and Thingbaijam, K. K. S. (2012). Probabilistic seismic hazard completeness of a seismic catalogue. Lett. Nuovo Cimento, 42(1), 21–27.
assessment of India. Seismol. Res. Lett., 83(1), 135–149. https://doi.org/10.1007/BF02739471
https://doi.org/10.1785/gssrl.83.1.135 Utsu, T. (1965). A method for determining the value of b in a formula
Ordaz, M. G., Cardona, O.-D., Salgado-Gálvez, M. A., Bernal-Granados, G. A., logN=a−bM showing the magnitude-frequency relation for earthquakes.
Singh, S. K., and Zuloaga-Romero, D. (2014). Probabilistic seismic hazard Geophys. Bull. Hokkaido Univ., 13, 99–103.
assessment at global level. Int. J. Disaster Risk Reduct., 10, 419–427. Wang, Z. M., Butler, D. T., Woolery, E. W., and Wang, L. M. (2012). Seismic hazard

Rahman M M et al.: PSHA of Nepal

 Earth and Planetary Physics doi: 10.26464/epp2018030 341

assessment for the Tianshui urban area, Gansu Province, China. Int. J. history, and foreland sedimentation. Earth-Sci. Rev., 76(1-2), 1–131.
Geophys., 2012, 461863. https://doi.org/10.1155/2012/461863 https://doi.org/10.1016/j.earscirev.2005.05.004
Wells, D. L., and Coppersmith, K. J. (1994). New empirical relationships among Youngs, R. R., Chiou, S.-J., Silva, W. J., and Humphrey, J. R. (1997). Strong ground
magnitude, rupture length, rupture width, rupture area, and surface motion attenuation relationships for subduction zone earthquakes. Seismol.
displacement. Bull. Seismol. Soc. Am., 84(4), 974–1002. Res. Lett., 68(1), 58–73. https://doi.org/10.1785/gssrl.68.1.58
Wiemer, S., and Wyss, M. (2000). Minimum magnitude of completeness in Zhang, P. Z., Yang, Z. X., Gupta, H. K., Bhatia, S. C., and Shedlock, K. M. (1999).
earthquake catalogs: examples from Alaska, the Western United States, and
Global seismic hazard assessment program (GSHAP) in continental Asia.
Japan. Bull. Seismol. Soc. Am., 90(4), 859–869.
Ann. Di Geof., 42(6), 1167–1190. https://doi.org/10.4401/ag-3778
https://doi.org/10.1785/0119990114
Zhang, W. B., Iwata, T., and Irikura, K. (2006). Dynamic simulation of a dipping
Woo, G. (1996). Kernel estimation methods for seismic hazard area source
fault using a three-dimensional finite difference method with nonuniform
modeling. Bull. Seismol. Soc. Am., 86(2), 353–362.
grid spacing. J. Geophys. Res., 111, B05301.
Wyss, M. (2005). Human losses expected in Himalayan earthquakes. Nat.
Hazards, 34(3), 305–314. https://doi.org/10.1007/s11069-004-2073-1 https://doi.org/10.1029/2005JB003725
Yen, Y. T., and Ma, K. F. (2011). Source-Scaling relationship for M 4.6-8.9 Zhao, J. X., Zhang, J., Asano, A., Ohno, Y., Oouchi, T., Takahashi, T., Ogawa H.,
earthquakes, specifically for earthquakes in the Collision Zone of Taiwan. Irikura K., Thio H. K., Fukushima, Y. (2006). Attenuation relations of strong
Bull. Seismol. Soc. Am., 101(2), 464–481. https://doi.org/10.1785/0120100046 ground motion in Japan using site classification based on predominant
Yin, A. (2006). Cenozoic tectonic evolution of the Himalayan orogen as period. Bull. Seismol. Soc. Am., 96(3), 898–913.
constrained by along-strike variation of structural geometry, exhumation https://doi.org/10.1785/0120050122

Rahman M M et al.: PSHA of Nepal