J Seismol (2019) 23:725–769

https://doi.org/10.1007/s10950-019-09832-3

ORIGINAL ARTICLE

Earthquake hazard potential of Indo-Gangetic Foredeep:

its seismotectonism, hazard, and damage modeling
for the cities of Patna, Lucknow, and Varanasi
Sankar Kumar Nath & Manik Das Adhikari &
Soumya Kanti Maiti & Chitralekha Ghatak

Received: 7 August 2017 / Accepted: 2 April 2019 / Published online: 4 May 2019 / Published online: 4 May 2019
# Springer Nature B.V. 2019

Abstract The Indo-Gangetic Foredeep region lies in and 2475 years have been estimated at firm rock site
close proximity to the Himalayan collision tectonics condition (site class B/C) with an average shear-wave
and the Peninsular Shield thereby subjecting it to repeat- velocity of about 760 m/s, of which, however, the results
ed strong ground shaking from large and great earth- of only 475 years of return period have been presented
quakes from these active tectonic regimes. An attempt here for urban development and earthquake engineering
is, therefore, made to understand the seismotectonic point of view. Surface-consistent probabilistic seismic
regime of the Indo-Gangetic Foredeep region while hazard is modeled using the International Building
performing probabilistic seismic hazard modeling of Code-compliant short and long period site factors cor-
its important cities of Patna and Lucknow and the reli- responding to topographic gradient-derived shear-wave
gious city of Varanasi based on consideration of velocity-based site classes. The estimated surface-
seismogenic source characteristics, smoothened gridded consistent PGA is seen to vary in the range of 0.222–
seismicity zoning, and generation of next generation 0.238 g in Patna City, while it varies in the range 0.257–
ground motion attenuation models appropriate for this 0.295 g in Lucknow and 0.146–0.172 g in the city of
region along with other existing region-specific ground Varanasi. The cumulative damage probabilities in terms
motion prediction equations in a logic tree framework. of ‘none,’ ‘slight,’ ‘moderate,’ ‘extensive,’ and ‘com-
In the hazard modeling, peak ground acceleration plete’ have been assessed using the capacity spectrum
(PGA) and 5% damped pseudo-spectral acceleration method in the Seismic Loss Estimation Approach
(PSA) at different time periods for 10 and 2% probabil- (SELENA) using both the fragility and capacity func-
ity of exceedance in 50 years with a return period of 475 tions for six model building types in these cities. The
discrete damage probability exhibits that the building
types ‘IGW-RCF2IL (PAGER/FEMA:C1L),’ ‘IGW-
S. K. Nath (*) : M. D. Adhikari : S. K. Maiti : C. Ghatak RCF21M (PAGER/FEMA:C1M),’ and ‘IGW-RCF11L
Department of Geology and Geophysics, Indian Institute of
Technology, Kharagpur 721302, India (PAGER/FEMA:C3L)’ will suffer minimum damage,
e-mail: [email protected] while ‘IGW-RCF21H (PAGER/FEMA:C1H),’ ‘IGW-
RCF11M (PAGER/FEMA:C3M),’ and ‘PAGER/
FEMA:C3H’ will suffer extensive damage in the event
M. D. Adhikari
e-mail: [email protected] of a maximum earthquake of Mw 7.2 impacting the
terrain as predicted from median nodal maximum mag-
S. K. Maiti nitudes in a heuristic search in the probabilistic protocol.
e-mail: [email protected] The estimated probabilistic seismic hazard and damage
C. Ghatak scenario are expected to play vital roles in the
e-mail: [email protected] earthquake-inflicted disaster mitigation and

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

726 J Seismol (2019) 23:725–769

management of these cities for better predisaster pre- scenario in the Indo-Gangetic Foredeep region with
vention, preparedness, and postdisaster rescue, relief, focus on probabilistic seismic hazard and damage as-
and rehabilitation. The results may also be incorporated sessment of the capital cities of Patna and Lucknow, and
in the building codal provisions of these smart cities as the Hindu religious city of Varanasi. The Bureau of
intended by the Federal Government of India. Indian Standards (BIS 2002) placed the entire IGF re-
gion in seismic zones II–V, corresponding to a peak
Keywords Probabilistic seismic hazard . GMPEs . Site ground acceleration (PGA) range of 0.1–0.36 g. The
condition . SELENA . Structural damage . Indo-Gangetic lowest hazard zone marked II pertains to the southwest-
Foredeep ern part of the IGF, while zone III covers the central part
of the Indo-Gangetic Foredeep region. The cities such as
Varanasi, Lucknow, Agra, Kanpur, and Bareilly are
1 Introduction classified under zone III. The northern part of the IGF
encompassing Patna, Delhi, Meerut, Ghaziabad, Farida-
The Indo-Gangetic Plain is a foredeep basin that follows bad, Gurgaon, and Noida is classified under zone IV.
the trend of the Himalayan collision zone. The basin is Zone V occupies predominantly the northernmost part
filled with thick alluvium deposits of varying degrees of of Bihar encompassing Madhubani, Supaul, Araria,
compaction overlying the basement faults, ridges, and Sitamarhi, Dharbhanga, Madhepura, and Saharsa region
other tectonic features, obliterating their surface expres- facing direct threat from the Nepal-Bihar earthquake
sion. The sediment thickness in the Indo-Gangetic regime. Stratigraphically, the IGF is characterized by
Foredeep (IGF) varies from 500 m to 3.9 km with Holocene alluvial deposits that are likely to be liquefied
maximum thickness observed along the foothills of the due to an impending large earthquake as reported in GSI
Himalaya (Srivatsava 2001; Srinivas et al. 2013; memoir (GSI 1939, 1993) during the onslaught of the
Chadha et al. 2016). The near-surface geological fea- 1833 Nepal earthquake of Mw 7.6, the 1934 Nepal-Bihar
tures, especially the overburden soils and soft sediments earthquake of Mw 8.1, and the 1988 Bihar-Nepal earth-
in the basin, can dramatically amplify seismic waves quake of Mw 6.8, respectively.
and cause severe damage and fatalities, even at sites In order to develop effective earthquake measures, it
relatively far away from the epicenter (e.g., Cassidy is, therefore, imperative that the seismic hazard is sys-
and Rogers 2004; Nath and Thingbaijam 2009, 2011b; tematically estimated for the terrain. Several attempts
Nath et al. 2013, 2014; Maiti et al. 2017). The occur- have been made by various researchers in the past for
rence of some devastating earthquakes viz. 1833 Nepal seismic hazard assessment in different parts of the coun-
earthquake of Mw 7.6, 1934 Nepal-Bihar earthquake of try (viz. Khattri et al. 1984; Bhatia et al. 1999; Parvez
Mw 8.1, 1988 Bihar-Nepal earthquake of Mw 6.8, 2011 et al. 2003; Das et al. 2006; Jaiswal and Sinha 2007;
Sikkim earthquake of Mw 6.9, and 2015 Nepal earth- Mahajan et al. 2010; NDMA 2010; Nath and
quake of Mw 7.8 in a region of 500 km radius around Thingbaijam 2012; Sitharam and Kolathayar 2013;
IGF signifies how vulnerable the region is toward seis- Nath and Adhikari 2013; Sitharam et al. 2015; Nath
mic devastations. et al. 2014; Adhikari and Nath 2016; Maiti et al.
In India, urban centers are more susceptible to seis- 2017). In assessing probabilistic seismic hazard, we
mic risk due to high population density, unplanned include two different earthquake source models viz.
growth, inadequate planning, poor land use, and sub- layered polygonal seismogenic sources and active tec-
standard construction practices, thus necessitating sound tonic seismogenic sources at three hypocentral depth
disaster mitigation and management plans through a ranges viz. 0–25, 25–70, and 70–180 km. The seismic-
judicious interplay of seismic hazard and structural vul- ity parameters, i.e., a value and b value, along with the
nerability. Therefore, the Ministry of Earth Sciences estimation of maximum earthquake (Mmax) have been
(MoES), Govt. of India in its XII FYP period planned performed. Smoothening seismicity has been performed
to take up seismic hazard microzonation (SHM) of to estimate the activity rate of earthquake occurrence for
urban agglomeration of 30 targeted cities with popula- both the layered polygonal and tectonic sources at dif-
tion more than one million and those located in seismic ferent hypocentral depth ranges. Due to nonavailability
zones III, IV, and V in the BIS zonation map of India of well-established attenuation relations for the region,
(BIS 2002). In the present study, we worked out seismic six new next generation attenuation models (NGA) have

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 727

been developed by nonlinear regression analysis for the Indian and the Eurasian plates. The major subsurface
three tectonic provinces, namely the Central Himalaya, ridges along the Indo-Gangetic plains are the Faiza-
the Central Indian Peninsular Shield, and the Indo- bad Ridge, the Munger-Saharsa Ridge, and the
Gangetic Alluvium Basin. The ground motion predic- Goalpara Ridge (Kayal 2008). The basement of the
tion equation (GMPE) models given by Atkinson and Indo-Gangetic Foredeep is formed by the extension
Boore (2006) and Campbell and Bozorgnia (2003) have of basement rocks and the depression. This basement
been used as fundamental equations for the generation configuration and the presence of thick sediments in
of NGA models appropriate for this region. Apart from the region render the Indo-Gangetic Foredeep prone
the new NGA models, we have altogether incorporated to seismic hazard. Some major faults identified in the
in this study 15 regional and global prediction equations Indo-Gange tic Forede ep as reported in the
based on suitability testing. Eventually, all the hazard Seismotectonic Atlas of India (Dasgupta et al. 2000)
contributing components viz. source attribution, seismic are the Hathusar Fault, the Moradabad Fault, the
activity rates, and GMPEs are judiciously integrated Mahendragarh Dehradun Fault, the Great Boundary
with appropriate ranks and weights in a logic tree frame- Fault, the Lucknow Fault, the Main Frontal Thrust,
work to generate probabilistic seismic hazard distribu- the West Patna Fault, the East Patna Fault, the
tion with 10 and 2% probability of exceedance in terms Munger-Saharsa Fault, and the Rajmahal Fault as
of PGA and 5% damped pseudo-spectral acceleration depicted in Fig. 1. It is believed that most of the
(PSA) for different time periods at firm rock site (site faults extend northwards transversely to the Himala-
class: B/C, Vs ~ 760 ms−1) for 475 and 2475 years of yan belt (Valdiya 1976).
return periods. Thereupon, surface consistent seismic The seismicity of the IGF is the result of the collision
hazard for 10% probability of exceedance in 50 years tectonics between the Indian plate and the Eurasian plate
has been estimated by using both the short and long (Chandra 1978; Seeber et al. 1981; Ni and Barazangi
period site amplification factors from the International 1984; Kayal 2008). Due to these collision tectonic ac-
Building Code (IBC 2009) corresponding to various tivities, major seismic events of Mw ≥ 6.0 from the near-
sites classified from Vs30 values derived from the topo- and far-field seismogenic sources like the Central
graphic gradient (Allen and Wald 2009). Subsequently, Himalaya, the Indo-Gangetic Alluvial Basin, and the
damage probability of different model building types Central Indian Peninsular Shield impacted the IGF and
has also been computed using the capacity spectrum its adjoining region as shown in Fig. 1. Some notably
method in Seismic Loss Estimation Approach disastrous earthquakes are the 1833 Nepal earthquake of
(SELENA) using a logic tree based on damage func- Mw 7.6, 1934 Nepal-Bihar earthquake of Mw 8.1, 1988
tions and relational analysis protocol and validated qual- Bihar-Nepal earthquake of Mw 6.8, 2015 Nepal earth-
itatively from damage reporting for historical earth- quake of Mw 7.8, 1999 Chamoli earthquake of Mw 6.8,
quakes in the region. and 1905 Kangra earthquake of Mw 7.8 that caused
widespread damage and loss of life and property in the
IGF and its adjoining region.
2 Seismotectonism of Indo-Gangetic Foredeep Past historical reporting exhibits that the damage
and its adjoining region potential in the IGF varies from MM intensity V to IX
from both the near- and far-field earthquakes. The 1833
The Indo-Gangetic Plain lies in the Himalayan Nepal earthquake of Mw 7.6 killed nearly 500 people
foredeep which constitutes vast alluvium plains of among which most of the fatalities were in the Kath-
the Ganges, Indus, and their tributaries. The basin is mandu valley and the northern part of the IGF as shown
formed as a consequence of flexing of the Indian in Fig. 2a (Dasgupta and Mukhopadhyay 2015). The
lithosphere due to the continued northward move- 1905 Kangra earthquake of Mw 7.8 occurred in the
ment of the Indian plate and the thrust fold loading West-Central Himalaya causing damage equivalent to
of the Himalayan Orogen. The structural limit be- MM intensity VI in the northern part of the IGF as
tween the Indo-Gangetic Foredeep and the Himala- depicted in Fig. 2b. The 1934 Bihar-Nepal earthquake
yan region in the north is defined by the Himalayan of Mw 8.1 occurred at the Indo-Nepal boundary causing
Frontal Thrust (HFT) which is a direct consequence widespread damage and 10,500 fatalities in both Nepal
of the compression resulting from collision of the and India (www.asc-india.org). Seismic intensity of

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

728 J Seismol (2019) 23:725–769

Fig. 1 a, b Seismotectonic setting of the Indo-Gangetic Foredeep and its adjoining region (modified after Dasgupta et al. 2000; Yin et al.
2009)

MM VII to IX has been reported in Patna district, while areas, nearly in 1608 cities of Nepal, India,
the same is reported to vary between MM intensity VI Bangladesh, and China. The MM intensity IV–VII was
and VII in Varanasi and MM intensity V and VI in felt in the IGF as shown in Fig. 2f. The structural
Lucknow City (Dasgupta et al. 2000; GSI 1939) as destruction and building collapse due to the near- and
depicted in Fig. 2c. The 1988 Bihar-Nepal earthquake far-source moderate to large earthquakes in the Indo-
of Mw 6.8 has killed 721 persons and injured 6553, and Gangetic Foredeep region is depicted in the photographs
64,470 buildings were damaged in eastern Nepal and embedded in Fig. 3.
India. The MM intensity IV to IX was felt in the IGF as
depicted in Fig. 2d. The 1999 Chamoli earthquake of
Mw 6.8 occurred in the West-Central Himalaya causing 3 Seismic hazard potential of the Indo-Gangetic
damage equivalent to MM intensity V–VII in the north- Foredeep region
ern part of the IGF region as depicted in Fig. 2e. The
2015 Nepal earthquake of Mw 7.8 was the worst natural Seismicity of the Himalaya and the IGF is not the sole
disaster to strike Nepal since the 1934 Nepal-Bihar factor accountable for the seismic hazard of the basin.
earthquake and destroyed 138,182 houses completely The interaction with sediment and subsequent amplifi-
and 122,694 houses partially across Nepal and its ad- cation of seismic waves enhances the vulnerability of
joining region. It took more than 9000 lives, uncount- the Himalayan Foredeep region (Bagchi and
able people were injured, more than 50,000 persons Raghukanth 2017). Additionally, thick deposition of
were rendered homeless, and the economic loss was alluvial or cratonic sediment over hard crust is the
estimated to exceed the GDP rate of the country reason behind the impedance contrast between the sed-
(www.usgs.gov). The earthquake was felt over large iment and the crust along the vertical direction. The

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 729

Fig. 2 Isoseismal maps in the IGF for the a 1833 Nepal earth- 2015 Nepal earthquake of Mw 7.8 (adopted from Dasgupta et al.
quake of Mw 7.6, b 1905 Kangra earthquake of Mw 7.8, c 1934 2000; Pandey and Molnar 1988; Dasgupta and Mukhopadhyay
Nepal-Bihar earthquake of Mw 8.1, d 1988 Bihar-Nepal earth- 2015; Ghosh and Mahajan 2013; www.usgs.gov)
quake of Mw 6.8, e 1999 Chamoli earthquake of Mw 6.8, and f

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

730 J Seismol (2019) 23:725–769

Fig. 3 Photographs of structural damage and building destruction Chamoli earthquake of Mw 6.8 in the Indo-Gangetic Foredeep
due to near- and far-source moderate to large magnitude earth- region (Nasu 1935; Duggal and Sato 1989; Rai et al. 2016;
quakes viz. a–c 1934 Nepal-Bihar earthquake of Mw 8.1, d–f 1988 Shrikhande et al. 2000; http://www.nicee.org/nepaleq/; http://123
Bihar-Nepal earthquake of Mw 6.8, g–i 2015 Nepal earthquake of himachal.com/ dharamsala/ links/1905.htm)
Mw 7.8, j–l 1905 Kangra earthquake of Mw 7.8, and m–o 1999

variation in the mechanical material properties of the the ‘stratigraphic effect.’ The combined effect of the
sediment also creates interfaces of vertical impedance basin and the strata alters the frequency content of the
contrast among various strata. This is the reason behind resultant ground motion at the surface (Nath 2011).

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 731

Repetitive reflection and refraction of the waves within 2011) as E (Vs30 < 180 ms−1), D1 (Vs30 180–240 ms−1),
the basin can cause either attenuation or amplification of D2 (Vs30 240–300 ms−1), D3 (Vs30 300–360 ms−1), C1
the incident seismic energy depending upon their fre- (Vs30 360–490 ms−1), C2 (Vs30 490–620 ms−1), C3 (Vs30
quency content. Therefore, an in-depth site characteri- 620–760 ms−1), and B (Vs30 > 760 ms−1) as depicted in
zation study is essential to understand the ‘basin/strati- Fig. 4.
graphic effect’ of the IGF. Generally, geotechnical and In order to establish a correlation between the
geophysical tests are carried out to characterize the soil topography-derived site class and geotechnical/
overburden. However, the topographic slope data is geophysical investigation-based site classification with
widely used for seismic site characterization at the mac- Vs30 as proxy, the city of Lucknow is vividly analyzed
roscale level (Nath et al. 2013). The site characterization for both the site classification maps as shown in Fig. 5
of the IGF has been carried out using topographic exhibiting around 95% similarity in site class D in both
gradient-based approach generated from the digital ele- maps, while in site class C, the correlation is found to be
vation model (DEM) data. The global data from post- of the order of about 60% on accuracy statistics, which
processing on the original Shuttle Radar Topography as a first approximation is considered satisfactory in the
Mission (SRTM) digital elevation data generated by present case. As seen on the map, the entire IGF is
the National Aeronautics & Space Administration mostly associated with site classes E and D, while site
(NASA) and the National Imagery & Mapping Agency classes C and B are located in the southern and northern
(NIMA) are used for this purpose. The processing in- parts of the region. From the NEHRP recommendation,
cludes application of hole-filling algorithm to provide soils with lower Vs30 values (i.e., toward site classes D
seamless and complete elevation surfaces for the entire and E) will experience more ground shaking due to the
region while eliminating areas of no data. Using the wave-amplifying properties of the soil. Generally, soft
spatial analysis tool available in ArcGIS, the soils increase the ground motion amplitude during an
topographic slope at each grid point has been earthquake and, therefore, are responsible for greater
evaluated. Based on the correlation studies conducted earthquake-induced disaster (Thitimakorn and
for active and stable continental regions, Allen and Wald Channoo 2012). Thus, most part of the IGF is expected
(2009) proposed slope ranges corresponding to each to exhibit higher ground motion amplification during
National Earthquake Hazard Reduction Program the onslaught of a moderate to large magnitude
(NEHRP) site class given in Table 1. The entire IGF is earthquake.
associated with the stable continental region (Nath and The surface and subsurface soil layers are very fertile
Thingbaijam 2011a). We, therefore, used correlation of in the IGF because the low-level floods in the Ganges
stable continental region for the assessment of site char- continually replenish the surface soil. The subsurface
acteristics of the entire IGF. Based on the average shear- lithological depth section of the cities of Patna, Luck-
wave velocity to a depth of 30 m (Vs30), the entire region now, and Varanasi is characterized by fine- to coarse-
has been classified into NEHRP site classes (Phillips grained sand overlaid by silt–clay horizon near the

Table 1 Summary of slope ranges for subdivided NEHRP Vs30 categories for active tectonic and stable continental region (Allen and Wald
2009)

Class Vs30 range (m/s) Slope range (active tectonic) Slope range (stable continental)

E < 180 < 1.0E-4 < 2.0E-5

D 180–240 1.0E-4–2.2E-3 1.0E-4–2.2E-3
240–300 2.2E-3–6.3E-3 2.2E-3–6.3E-3
300–360 6.3E-3–0.018 4.0E-3–7.2E-3
C 360–490 0.018–0.050 0.018–0.050
490–620 0.050–0.10 0.050–0.10
620–760 0.10–0.138 0.018–0.025
B > 760 > 0.138 > 0.025

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

732 J Seismol (2019) 23:725–769

Fig. 4 Seismic site condition map of the Indo-Gangetic Foredeep Gangetic Plain is associated with site class D. Subplots (a–c)
and its adjoining region derived from topographic gradient-based represent Vs30 distribution in Patna, Lucknow, and Varanasi cities
correlations of Allen and Wald (2009) exhibiting that the entire derived from topographic conversion

surface as depicted in Fig. 6. The ground water table varies from 0.3 to 3.0 m in the region. Hence, the thick
reported by the Central Ground Water Board (CGWB) alluvium with shallow ground water table possesses

Fig. 5 Site classification map of Lucknow City developed through the a topographic gradient-based approach and b geotechnical borehole
analysis and MASW surveys (Anbazhagan et al. 2013a)

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 733

Fig. 6 Subsurface lithological depth section of the cities of a aquiferindia.org/AboutAQUIM_Watershed_Patna.aspx for the
Patna, b Lucknow, and c Varanasi which exhibits loose to medium city of Patna, Anbazhagan et al. (2013a) for the city of Lucknow,
cohesionless sediment deposits (adopted from http://www. and Shukla and Raju (2008) for the city of Varanasi)

high risk of undergoing excessive liquefaction and sub- active Himalayan belt and the local site effects
sidence during an earthquake. On the other hand, the associated with thick alluvium deposits.
population density in this region is very high, with about
10,000 persons per square kilometer. The total popula-
tion in the IGF is about 300 million, which is 25% of the 4 Probabilistic seismic hazard and structural
total population of the country (Census 2011; damage modeling
http://censusindia.gov.in/). Therefore, the urban centers
located in different parts of the IGF are susceptible to The occurrence of devastating earthquakes in and around
earthquake damages due to its proximity to seismically the IGF region drew attention to the seismic hazard and

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

734 J Seismol (2019) 23:725–769

risk of the region and suggests that the Indo-Gangetic and categorized into five damage states viz. ‘none,’
Foredeep is neotectonically active and, therefore, pro- ‘slight,’ ‘moderate,’ ‘extensive,’ and ‘complete’ in
vides a possibility of triggering potential earthquakes in terms of total damaged area or the number of damaged
the near future. The contiguity to the active Himalayan buildings. The computational steps for seismic building
region and the Peninsular Shield region, the nature of damage modeling have also been illustrated in Fig. 7.
sediments, and the observed neotectonic activity render
this region vulnerable to seismic devastations, especially 4.1 Probabilistic seismic hazard assessment
due to amplification of ground motion in an alluvial filled
terrain due to local site effects and soil liquefaction. In the present study, rigorous formulations of hazard com-
Considering the seismogenic sources already defined, ponents have been adopted to deliver a case for a proba-
the subsurface lithology, the NEHRP-defined site condi- bilistic seismic hazard model (PSHM) of the cities of
tion, and the destructive earthquakes that triggered in and Patna, Lucknow, and Varanasi based on underlying
around the IGF, it is evident that the capital cities of Patna seismogenic source zones in the Indo-Gangetic Foredeep
and Lucknow and the famous Hindu religious city of and its adjoining region employing the earthquake catalog
Varanasi are vulnerable to earthquake-induced damage supplemented by records of historical earthquakes, focal
and destruction, thus necessitating an estimation of the mechanism data, published literatures, paleoseismicity
seismic hazard potential of the region with a focus on findings, and neotectonic database. In view of the site
these three important smart cities. characterization study of Nath et al. (2013) across the
The evaluation of probabilistic seismic hazard and country, Nath and Thingbaijam (2012) considered firm
structural damage involves the combination of three rock site condition (standard engineering bedrock) to be
major components: (a) probabilistic seismic hazard at most appropriate for regional hazard computation pur-
firm rock condition through a logic tree framework, (b) poses. Therefore, the standard engineering bedrock
surface level PSH model based on the International conforming to Vs30 ~ 760 ms−1 (defined as boundary site
Building Code (IBC 2009)-defined site effects par class B/C) is considered as the rock site in compliance to
NEHRP site classes, and (c) damage model defining which the hazard computation is performed (Nath and
discrete damage probability based on the capacity spec- Thingbaijam 2012). In the probabilistic seismic hazard
trum method dealt in SELENA. The key components analysis, the annual rate of ground motion exceeding a
and work frame of the probabilistic seismic hazard and specific value is computed to account for different return
damage assessment of the Indo-Gangetic Foredeep re- periods of the hazard. Contributions from all the relevant
gion are illustrated in Fig. 7. The hazard modeling sources and possible events are considered. Thereafter, a
involves hierarchical development of different hazard logic tree framework is developed toward hazard compu-
components such as seismogenic sources, ground mo- tation at each site to incorporate multiple models in source
tion prediction equations, and NEHRP site conditions, considerations, GMPEs, and seismicity parameters. The
appropriate hazard formulations considering both the hazard computation is performed on grid points covering
epistemic and aleatory uncertainties and integration of the entire study region in the cities of Patna, Lucknow, and
all in a logic tree framework. Varanasi at 0.0005° × 0.0005° interval. The hazard distri-
At the onslaught of a destructive earthquake in a butions are computed for the source zones at each depth
region, the prevailing georeferenced methodologies section separately and, thereafter, integrated. The prelimi-
such as HAZUS (Hazard-US, FEMA 1999), RADIUS nary model comprises of spatial distributions of seismic
(Risk Assessment Tools for Diagnosis of Urban areas hazard in terms of PGA and 5% damped PSA.
against Seismic Disasters, Okazaki et al. 2000), ELER
(Earthquake Loss Estimation Routine, Hancilar et al. 4.1.1 Preparation of earthquake catalog for the IGF
2010), EPEDAT (The Early Post-Earthquake Damage
Assessment Tool, Eguchi et al. 1997), and SELENA The preparation of a homogeneous and declustered
(Molina and Lindholm 2005; Molina et al. 2010) either earthquake catalog is the starting point of the steps to
individually or in unison are used for damage and loss be followed for PSHA of the region under study con-
modeling. In the present study, we implemented the sidering a period starting from the prehistoric era till the
capacity spectrum method in SELENA. Damage prob- present time. We, thus, prepared an earthquake catalog
ability of different model building types is computed of this region spanning for the period of 1900–2016 by

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 735

Fig. 7 Integrated computational framework for probabilistic seismic hazard and structural damage assessment

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

736 J Seismol (2019) 23:725–769

considering three major earthquake data sources, name- and Knopoff (1974), the length and duration of the
ly the International Seismological Centre (ISC, windows are given in Table 3. This method does not
h t t p : / / w w w. i s c . a c . u k ) , U . S . G e o l o g i c a l consider secondary and higher order aftershocks.
Survey/National Earthquake Information Center We adopted this technique since (a) there is higher
(USGS/NEIC, http://neic.usgs.gov.us), and Global likelihood of aftershocks of larger main shock events
Centroid Moment Tensor (GCMT, http://www. being recorded in the catalog compared to those for the
globalcmt.org), wherein the hypocentral depth entries smaller ones and (b) the spatial spans of aftershocks,
have been computed using the algorithm given by especially for those associated with larger earthquakes,
Engdahl et al. (1998). Other data sources used include are dynamic depending not only on the magnitude of the
the India Meteorological Department (IMD, event but also on the geological background.
http://www.imd.ernet.in) and Jaiswal and Sinha (2004 The parameters listed in Table 3 are adopted for
). For uniform magnitude scaling and establishing data magnitudes 3.0 ≤ Mw,GCMT ≤ 8.0, and the aftershock
homogeneity for meaningful statistical analysis, Mw is zone is identified by inspecting continuous spatial win-
preferred owing to its applicability for all ranges of dows of 0.25° × 0.25° for the presence of at least one
earthquakes: large or small, far or near, or shallow or event within the specified days’ limit corresponding to
deep-focused. To implement uniform magnitude scaling the main shock of a given magnitude. Once the zones
for the instrumental catalog, the Mw entries found in are demarcated, the events found within the zone from
GCMT are retained. The magnitude entry from the ISC the advent till the end of the catalog are examined with
catalog is selected maintaining a preference order of cumulative number of events against time. Nyffenegger
Ms,ISC, mb,ISC, ML,ISC, MD,ISC, mpv,ISC, and MLv,ISC; and Frohlich (2000) observed that the aftershock se-
mb,ISC, ML,ISC, MISC, MN,ISC MD,ISC, MLv,ISC mpv,ISC, quences for intermediate as well as deep earthquakes
mb,USGS, Ms,USGS, ukUSGS, and mw,USGS into Mw,GCMT. do not behave differently from those of the shallower
Therefore, the conversion relations derived by Nath ones. The algorithm, therefore, remains the same for the
et al. (2017) through the orthogonal standard regression deeper (hypocentral depth ≥ 70 km) earthquakes, and
(OSR) have been used to convert all the magnitude the termination of the aftershock sequences is decided
types into Mw,GCMT along with the existing reported accordingly. The analysis has uncertainties due to errors
equations as illustrated in Table 2. The uncertainties of associated with epicentral locations, time, and magni-
the unified moment magnitude due to the usage of the tudes. In the processing, the epicenters are grouped
conversion equations are incorporated during the within a distance bound, and consequently, the errors
compilation. associated are significantly reduced and so is with the
Eventually, we obtained a compilation with higher case of time bins, while the magnitude-wise correlation
data volume compared to the original sources. Thereaf- between the events is done with the assigned magni-
ter, the entire catalog has been declustered to remove tudes. We restricted to the identification of the most
foreshocks and aftershocks to derive a main shock cat- likely aftershocks, and henceforth, errors in the magni-
alog as elaborated in Nath et al. (2017). The space–time tudes are not given additional treatment. The same ap-
clustering of seismicity is mostly exhibited by fore- proach is used for the detection of likely foreshocks
shocks and aftershocks. Main shock catalogs are derived based on the increasing seismic activity. Finally, Fig. 8
by eliminating these clusters. Windowing algorithms are represents a seismicity map prepared using the derived
generally used for the purpose. The available algorithms main shock catalog.
(e.g., Gardner and Knopoff 1974; Reasenberg 1985; The hypocentral depth entries in the compiled catalog
Uhrhammer 1986; Zhuang et al. 2002; Hainzl et al. (only main shock events considered) are either fixed
2006) generally differ in terms of spatiotemporal win- (i.e., uncertainty not provided) or have standard devia-
dow parameters. On the other hand, deciding on optimal tion assigned. In order to reduce the associated uncer-
parameters is difficult in the light of diverse tainties, the Engdahl–van der Hilst–Buland (EHB) cat-
seismotectonic conditions (Gomberg et al. 2003). In alog from the International Seismological Centre (2009)
the present study, we used the window-based is consulted. The EHB catalog was prepared using data
declustering algorithm of Gardner and Knopoff (1974) from the ISC and preliminary determination of epicen-
to identify aftershocks and foreshocks depending on ters of USGS/NEIC; hypocentral depth was recomputed
interevent space–time distance. According to Gardner using the algorithm given by Engdahl et al. (1998). The

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 737

Table 2 Conversion relations used to convert all the magnitude types into Mw,GCMT (Nath et al. 2017)

Catalog accessed Magnitude scales Min. mag Max. mag. Converted to: Magnitude range Equations

ISC mb 3 9 Mw, GCMT 3.8–7.0 Mw, GCMT = 1.168 × mb, ISC − 0.663
ML 3 8 Mw, GCMT 3.5–7.3 Mw, GCMT = 0.499 × ML, ISC + 2.88
M 3 7.5 Mw, GCMT 4.7–7.2 Mw, GCMT = 0.978 × MISC + 0.1634
MN 3 5.7 ML, ISC 3.6–5.3 ML, ISC = 1.219 × MN, ISC − 0.972
MD 3 7.4 mb, ISC 4.0–6.2 mb, ISC = 1.428 × MD, ISC − 2.182
MLv 3 8.1 mb, ISC 2.0–4.5 mb, ISC = 0.962 × MLv, ISC − 0.0009
4.6–7.6 mb, ISC = 1.177 × MLv, ISC − 1.393
mpv 3 7.2 mb, ISC 3.4–6.6 mb, ISC = 1.337 × mpv, ISC − 1.625
USGS mb 3 7.5 Mw, GCMT 4.6–6.4 Mw, GCMT = 1.082 × mb, USGS − 0.413
MS 4 7.5 Mw, GCMT 4.5–5.6 Mw, GCMT = 1.15 × Ms, USGS − 0.628
5.7–7.0 Mw, GCMT = 1.21 × Ms, USGS − 1.45
7.1–7.5 Mw, GCMT = Ms, USGS
uk 5.6 7.1 Ms, ISC 6.5–6.8 Ms, ISC = ukUSGS + 0.2
mw 3.9 9.1 Mw, GCMT 5.1–7.0 Mw, GCMT = 1.017 × mw, USGS − 0.118
7.1–7.8 Mw, GCMT = mw, USGS

EHB catalog spans through 1960–2006 and does not is compiled on the Geographical Information System.
recalculate the magnitudes. The epicentral location and The sources include Seismotectonic Atlas of India and
the depth entries along with the uncertainty (standard NGLM published by the Geological Survey of India
deviation) in the present compilation are updated on the (Dasgupta et al. 2000; http://www.portal.gsi.gov.in)
basis of event-to-event comparison with the entries in and the one extracted from Landsat TM/MSS/ETM,
the EHB catalog. Additionally, records given by Stork LISS III/IV, Cartosat-DEM, and SRTM data. The
et al. (2008) are selectively used to update entries not seismogenic sources are defined by superimposing the
covered by the EHB catalog. homogeneous and declustered earthquake catalog for
Thus, the complete and homogeneous earthquake the period of 1900–2016 on the fault pattern in the
catalog prepared for Indo-Gangetic Foredeep and its region. In the present study, we classified seismogenic
adjoining region has been used for probabilistic seismic sources as (a) layered polygonal seismogenic sources
hazard assessment in terms of seismogenic source zo- and (b) active tectonic seismogenic sources as illustrated
nation, seismicity analysis, smoothened seismicity in the following subsections.
modeling, and seismic hazard assessment protocol.
Layered polygonal seismogenic source zones A popular
4.1.2 Seismogenic source definition in Indo-Gangetic approach in the seismogenic localization process is the
Foredeep and its adjoining region areal source zonation, wherein the objective is to capture
uniform seismicity. Source delineation is primarily
The source characterization includes both a homoge- based on tectonic trends and seismicity of the region.
neous earthquake catalog of the region just illustrated It has been observed that seismicity patterns and source
in the previous section and also the fault database which dynamics have significant variation with hypocentral

Table 3 Aftershock identification windows (Gardner and Knopoff 1974)

Magnitude 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0

Distance (km) 19.5 22.5 26 30 35 40 47 54 61 70 81 94.0

Time (days) 6 11.5 22 42 83 155 290 510 790 915 960 985

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

738 J Seismol (2019) 23:725–769

Fig. 8 Declustered seismicity covering the period 1900–2016 and comprising of 4597 main shock events. Subplots represent the histogram
of declustered earthquake distribution for the three hypocentral depth ranges of 0–25, 25–70, and 70–180 km

depth (Christova 1992; Tsapanos 2000; Allen et al. other researchers. In order to establish the
2004). Therefore, the consideration of a single set of seismotectonic description at each layer, we constructed
seismicity parameter over the entire hypocentral depth representative focal mechanism tensor (i.e., F̄ ) by cal-
range may lead to erroneous hazard estimation. Accord- culating the weighted average of the known moment
ingly, we considered three hypocentral depth ranges: tensors as follows:
upper crust (0–25 km), lower crust (25–70 km), and
N
lower crust (70–180 km). The methodology adopted in ∑ M n0 F nij
the present study can be outlined into three aspects: (1) n¼1
F ij ¼ N
ð1Þ
delineation of areal source zones on the basis of seis-
∑ M n0
micity distribution and fault patterns complemented by n¼1
available focal mechanism data, (2) formulation of seis-
micity model and associated uncertainty values for each Where N is the total number of the focal mechanisms,
source zone, and (3) application of seismicity smooth- M n0 is the scalar moment of the nth focal mechanism, and
ening algorithm to obtain activity rates for specific F nij is a function of the strike, dip, and rake of this focal
threshold magnitude/s. The source zonation at each mechanism (Aki and Richards 1980). Thus, following
depth layer is carried out by considering the seismicity the above steps and the findings of Nath and
patterns, fault networks, and similarity in the style of Thingbaijam (2012), 27 areal source zones have been
focal mechanisms (e.g., Cáceres et al. 2005). The lay- identified for the Indo-Gangetic Foredeep region as
ered seismogenic source model in the present study is depicted in Fig. 9a-c, while the GSHAP considered only
expected to facilitate resolving source characteristics 14 areal sources as shown in Fig. 9d.
more precisely than a single layer scheme that has been
considered hitherto in the study region under the Global Active tectonic seismogenic sources Additional
Seismic Hazard Assessment Program (GSHAP) and seismogenic sources considered here are the active

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 739

tectonic features such as the faults and lineaments identified and implicated earlier though they were also
(Azzaro et al. 1998; Slemmons and McKinney 1977). potential seismic sources in the region. The focal mech-
Many active faults and lineaments capable of producing anism data employed in the present study are derived
earthquakes of Mw 3.5 and above are expected to influ- from the Global Centroid Moment Tensor (GCMT,
ence seismic hazard of the Indo-Gangetic Foredeep available at www.globalcmt.org) database covering the
region and, hence, have been extracted from the period 1976–2016 and other available sources like
Seismotectonic Atlas of India (Dasgupta et al. 2000), Dasgupta et al. (2000), Chandra (1977), Singh and
the National Geomorphological & Lineament Mapping Gupta (1980), and Bilham and England (2001). We thus
(NGLM) on a 1:50,000 scale (available at http://www. identified around 527 active tectonic features shown in
Portal.Gsi.gov.in/portal/page; http://bhuvan.nrsc.gov. Fig. 10 in the hypocentral depth ranges of 0–25, 25–70,
in/gis/thematic/index.php), and additional features and 70–180 km.
picked by image processing of Landsat TM, SRTM,
ASTER, and LISS IV data as depicted in Fig. 10. 4.1.3 Smoothened gridded seismicity model
Emphasis has, however, been given to large-scale line-
aments having relevance to geomorphology, vegetation The contribution of background events in the haz-
patterns, tectonic contact zones, and aligned abrupt ard perspective is calculated using smoothened
drainage patterns/river which are generally related to gridded seismicity models wherein discrete earth-
faults. The seismicity of this region can be linked to all quake distributions are modeled into spatially con-
possible active faults/lineaments which were not tinuous probability distributions using the Frankel

Fig. 9 Layered polygonal seismogenic source framework for the Indo-Gangetic Foredeep region at the hypocentral depth ranges: a 0–
25 km, b 25–70 km, and c 70–180 km as adopted from Nath and Thingbaijam (2012), and d areal sources identified by GSHAP (1999)

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

740 J Seismol (2019) 23:725–769

Fig. 10 Active tectonic sources as identified to have seismic hazard contribution to the Indo-Gangetic Foredeep region

(1995) methodology. In this study, the IGF and its triggering the 1934 Nepal-Bihar earthquake of Mw 8.1,
adjoining region are gridded at 0.1° × 0.1° cells. 1988 Bihar-Nepal earthquake of Mw 6.8, and the recent
The smoothened function used is given as, 25th April 2015 Nepal earthquake of Mw 7.8.

∑ n j ðmr Þe−ðd ij =cÞ

2
4.1.4 Analysis of seismic activity rate on active tectonic
j sources
N ðm r Þ ¼ ð2Þ
∑ e−ðdij =cÞ
2

j In the present study, seismicity activity rates are

calculated for each active tectonic source inscribed
Where nj(mr) is the number of events with magnitude in each polygonal seismogenic source for the thresh-
≥mr, dij is the distance between the ith and jth cells, and c old magnitudes (mo) of Mw 3.5, 4.5, and 5.5 for the
denotes the correlation distance. The annual activity rate focal depth ranges < 25, 25–70, and 70–180 km. We
λmr is computed each time as N(mr)/T, where T is the employed the fault degradation technique of Iyengar
subcatalog period. The present analyses make use of
different subcatalogs with the threshold magnitudes of
Table 4 The subcatalogs for the three threshold magnitudes are
Mw 3.5, 4.5, and 5.5, respectively, as summarized in given here
Table 4 at the hypocentral depth levels of 0–25, 25–70,
and 70–180 km. Correlation distances of 55, 65, and Depth range (km) Subcatalog (threshold magnitude)
85 km are decided for the respective cases by calibrating Mw 3.5 Mw 4.5 Mw 5.5
the outputs from several runs of the algorithm with the
observed seismicity. 0–25 1997–2014 1964–2014 1901–2014
The smoothened seismicity models given in Fig. 11 25–70 1997–2014 1964–2014 1900–2014
depict possible stressed zones within the East-Central 70–180 1998–2014 1964–2014 1903–2014
Himalaya which incidentally had been the source for

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 741

Fig. 11 Smoothened seismicity models for different threshold magnitudes at three hypocentral depth levels of 0–25, 25–70, and 70–180 km

and Ghosh (2004) for this purpose. The number of

αs ¼ Ls =ΣLs and δs ¼ ns =N t ð3Þ
earthquake occurrence per year with m > mo in a
given seismogenic layered polygonal source Taking the mean of the above two weighting factors
consisting of n faults is denoted as N(mo). According which indicates the seismic activity of the sth fault in the
to the fault degradation technique, N(mo) should be zone, we can compute the annual activity rate of the sth
equal to the sum of the number of earthquakes Ns(mo) fault by,
along all the faults delineated in the seismogenic
source zone, i.e., N(mo) = ∑Ns(mo), where Ns(mo) N s ðmo Þ ¼ 0:5ðαs þ δs Þ  N ðmo Þ ð4Þ
represents the annual frequency of occurrence of an
event on sth subfault (s = 1, 2….n) with mo = 3.5, 4.5, The annual activity rate of each tectonic feature
and 5.5. The number of events Ns(mo) that occurs on inscribed in each of the 27 polygonal areal seismic
a given fault depends upon various factors like the sources has thus been computed wherein the regional
length of the fault (Ls) and the number of past earth- recurrence is degraded into individual faults/lineaments.
quakes (ns) of magnitude mo and above associated In Fig. 12, we presented representative plots of annual
with the sth fault having been used as weights for activity rate versus magnitude for a group of active
calculating Ns(mo). For example, if Nt is the total tectonic features inscribed in the polygonal areal
number of events occurring within a polygonal areal seismogenic source zones ‘2’ and ‘3’ for each of the
source, the weighting factor can be estimated as, three threshold magnitudes of Mw 3.5, 4.5, and 5.5. The

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

742 J Seismol (2019) 23:725–769

Fig. 12 Representative plots of annual activity rate versus magnitude for a group of active tectonic features inscribed in the polygonal areal
seismogenic sources ‘2’ and ‘3’ for the threshold magnitudes of Mw 3.5, 4.5, and 5.5

spatial distribution of fault activity rate for different

log10 ðeÞ
threshold magnitudes at three hypocentral depth ranges b¼
 ð6Þ
Δm
of 0–25, 25–70, and 70–180 km is also depicted in mmean − mt −
Fig. 13. 2

4.1.5 Seismicity analysis Where mmean is the average magnitude, mt is the

minimum magnitude of completeness, and Δm is the
Seismicity parameters: ‘a’ and ‘b’ value magnitude bin size (= 0.1 in the present study). The
assessment The evaluation of seismicity parameters is standard deviation of b value (δb) has been comput-
one of the most important steps in the seismic hazard ed by the bootstrapping method as suggested by
estimation. Earthquake occurrences across the globe Schorlemmer et al. (2003) which involves repeated
follow the Gutenberg and Richter (GR) (Gutenberg computations, each time employing redundant data
and Richter 1944) relationship, sample, allowing events drawn from the catalog to
be selected more than once. A minimum magnitude
log10 λðmÞ ¼ a−bm ð5Þ
constraint is generally applied on the GR relation
Where λ(m) is the cumulative number of events with given by Eq. (6) on the basis of the magnitude of
magnitude ≥m. The slope parameter, commonly termed completeness entailed by the linearity of the GR
the b value, is often employed as an indicator of stress relation on the lower magnitude range. An upper
regime in the tectonic reinforcements and to characterize magnitude has been suggested in accordance with
seismogenic zones (Schorlemmer et al. 2005). The maxi- the physical dissipation of energy and the con-
mum likelihood method for the estimation of b value given straints due to the tectonic framework (Kijko
by Aki (1965) and Utsu (1965) has been used here as, 2004). This is achieved by establishing the

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 743

Fig. 13 Fault activity rate for various threshold magnitudes at the three hypocentral depth ranges of 0–25, 25–70, and 70–180 km

maximum earthquake Mmax physically capable of seismic source zone is of appropriate size and the
occurring within a defined seismic regime in an temporal coverage of the catalog is also long
underlying tectonic setup. The magnitude distribu- enough, and the TGR model is reckoned to be more
tion is, therefore, truncated at Mmax such that Mmax ≫ appropriate considering the energy dissipations at
mmin. A modified version of Eq. (6) formulated by larger magnitudes. In several cases, zones with sim-
Page (1968) and Cornell and Vanmarcke (1969) is a ilar tectonics are merged to achieve sufficient num-
truncated exponential distribution (TGR) as follows, ber of events say ≥ 50 in the present case as well as
an acceptable uncertainty with the estimated seis-
10−bðm−mmin Þ −10−bðM max −mmin Þ micity parameters. This ultimately produced 20
λðmÞ ¼ λðmmin Þ ð7Þ zones out of a total of 27 zones initially considered.
1−10−bðM max −mmin Þ
Seismicity analysis has been performed in these
zones to estimate both the ‘a’ and ‘b’ values. The
Where mmin is the minimum magnitude and Mmax sample frequency magnitude distribution plots for
is the upper-bound magnitude. The maximum earth- main shock events in the seismogenic source zones
quake (Mmax) is the largest seismic event character- of ‘2,’ ‘3,’ and ‘18’ are depicted in Fig. 14, and the
istic of the terrain under the tectonostratigraphic seismicity parameters estimated for all the polygonal
consideration. The incomplete data (including the seismogenic sources are listed in Table 5.
historical data) is rendered return periods according
to the GR and TGR model. The linear GR relation Maximum earthquake prognosis The maximum earth-
can statistically accommodate large events if the q u a k e ( M m a x ) i s t h e l a rg e s t s e i s m i c e v e n t

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

744 J Seismol (2019) 23:725–769

Fig. 14 Representative frequency magnitude distribution plots at (GR) relation, while the circles and squares represent the instru-
some typical polygonal seismogenic source zones ‘2,’ ‘3,’ and mental events (the complete data coverage) and incomplete data
’18.’ The red line represents the truncated Gutenberg–Richter (including the historical data as extreme data coverage),
(TGR) relation, the blue line represents the Gutenberg–Richter respectively

characteristic of the terrain under the seismotectonic generally define the maximum extent of co-
consideration. The Mmax values are often calculated seismic fault rupture (Kanamori and Anderson
from fault dimensions and geodetic inferences 1975; Dietz and Ellsworth 1990; Wong et al.
(Wells and Coppersmith 1994; Anderson et al. 2000). Basically, an aftershock zone roughly corre-
1996), in addition to the frequency magnitude dis- sponds to the fault ruptured during the main shock.
tribution indicated by past earthquakes. Maximum Precise studies indicate that aftershocks are con-
earthquake prognosis has been performed for both centrated near the margin of the fault area where
the layered polygonal sources and the active tecton- large displacement occurred (e.g., Das and Henry
ic sources. For polygonal sources, a maximum like- 2003; Utsu 2002). The general assumption, based
lihood method for maximum earthquake estimation on worldwide data, is that one third to one half of
referred to as the Kijko–Sellevoll–Bayesian tech- the total length of the fault would rupture when it
nique (Kijko 2004; Kijko and Graham 1998) is generates the maximum earthquake (Mark 1977;
used. The technique is based on Bayesian equation Kayabalia and Akin 2003; Shukla and Choudhury
of frequency magnitude distribution. It has been 2012; Seyrek and Tosun 2011). In the present
observed that empirical magnitude distribution de- study, the fault rupture segmentation is identified
viates moderately from the Gutenberg–Richter rela- using the maximum length of the well-aligned main
tion following an exponential tail of a Gamma func- shock and aftershocks along the faults (e.g., Besana
tion at larger magnitudes. Hence, a readily available and Ando 2005; Utsu 2002; Wells and Coppersmith
Mmax estimator computer code proposed by Kijko 1994), which on digitization in GIS yields subsur-
(2004) has been employed in the present study. The face rupture length of each active tectonic feature
estimated maximum earthquake (Mmax) and the ob- whose maximum earthquake is estimated using the
served maximum earthquake of each polygonal co-seismic subsurface fault rupture dimension and
seismogenic source are listed in Table 5. magnitude of Wells and Coppersmith (1994). Ta-
The deterministic assessment of characteristic ble 6 enlists some major active tectonic sources,
earthquake viz. maximum earthquakes from a fault their total length (TFL), the associated observed
is generally achieved with a relationship between maximum earthquakes (M max,obs), the subsurface
earthquake magnitude and co-seismic subsurface rupture length (RLD), and the maximum predicted
fault rupture length. The primary method used to earthquake (Mmax) from RLD which is seen to fall
estimate subsurface rupture length and rupture area within the inner bounds of one third and one half
is the spatial pattern of early aftershocks (Wells and approximations of Mark (1977), Kayabalia and
Coppersmith 1994). Aftershocks that occur within Akin (2003), Shukla and Choudhury (2012), and
a few hours to a few days of the main shock Seyrek and Tosun (2011).

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 745

Table 5 Estimated seismicity parameters for all the polygonal seismogenic sources considered for PSHA in the Indo-Gangetic Foredeep
region

Zone b value a value Mmax (predicted) Observed (Mmax) No. of events in the zone

1 0.71 (± 0.19) 2.65 (± 0.55) 7.3 (± 0.4) 7.0 57

2 0.67 (± 0.04) 3.22 (± 0.64) 8.6 (± 0.2) 7.7 256
3 0.83 (± 0.10) 3.45 (± 0.10) 8.4 (± 0.2) 7.6 101
4 0.84 (± 0.09) 3.13 (± 0.38) 5.6 (± 0.4) 5.3 59
5 0.71 (± 0.11) 2.98 (± 0.26) 8.5 (± 0.3) 6.0 68
6+8 0.69 (± 0.06) 3.46 (± 0.58) 7.7 (± 0.4) 6.7 237
7+9 0.71 (± 0.08) 3.15 (± 0.41) 7.7 (± 0.4) 6.4 117
10 0.69 (± 0.05) 2.79 (± 0.33) 7.2 (± 0.4) 6.0 92
11 + 14 0.69 (± 0.14) 2.81 (± 0.14) 7.2 (± 0.3) 6.5 69
12 + 13 0.73 (± 0.10) 2.86 (± 0.16) 7.0 (± 0.4) 5.9 168
15 1.14 (± 0.23) 4.52 (± 0.36) 7.7 (± 0.2) 5.8 65
16 1.12 (± 0.07) 5.10 (± 0.43) 8.1 (± 0.4) 7.7 292
17 0.67 (± 0.10) 2.76 (± 0.37) 7.1 (± 0.3) 6.3 77
18 1.19 (± 0.13) 5.26 (± 0.34) 7.1 (± 0.4) 5.8 125
19 0.86 (± 0.04) 4.04 (± 0.34) 8.8 (± 0.4) 8.7 248
20 0.80 (± 0.09) 3.48 (± 0.29) 8.3 (± 0.2) 8.1 104
21 0.69 (± 0.12) 3.08 (± 0.33) 8.3 (± 0.2) 6.9 63
22 + 23 + 24 0.81 (± 0.12) 3.28 (± 0.29) 6.8 (± 0.3) 6.4 76
25 1.04 (± 0.26) 4.03 (± 0.38) 6.7 (± 0.2) 5.0 53
26 + 27 0.81 (± 0.23) 3.33 (± 0.60) 6.7 (± 0.3) 6.0 72

Table 6 Some major active tectonic features considered to be seismogenic in the region along with their total fault lengths (TFL), observed
maximum earthquake (mmax,obs), projected subsurface rupture length (RLD), and estimated maximum earthquake (Mmax)

Fault name TFL Mmax,obs Fault type RLD Mmax σ

Himalayan Frontal Thrust 1288.6 8.1 Reverse 577.3 8.6 0.26

Main Boundary Thrust 1328.6 7.7 Reverse 527.6 8.5 0.26
Jwala Mukhi Thrust 496.1 6.4 Reverse 123.5 7.6 0.26
Mahendragarh Dehradun Fault 315.3 5.1 Strike-slip 72.5 7.3 0.24
Son Narmada Fault 270.1 5.8 Strike-slip 42.6 6.9 0.24
Main Central Thrust 1916.5 8.7 Reverse 958.2 8.9 0.26
West Patna Fault 213.6 5.2 Strike-slip 35.8 6.8 0.24
East Patna Fault 177.1 6.8 Reverse 56.3 7.1 0.26
Munger-Saharsa Ridge Fault 220.7 6.5 Strike-slip 51.3 7.0 0.24
Lucknow Fault 130.9 4.2 Strike-slip 24.8 6.6 0.24
Great Boundary Fault 640.3 6 Strike-slip 85.9 7.4 0.24
Moradabad Fault 177.4 5.6 Strike-slip 54.2 7.1 0.24
Martoli Thrust 161.4 7.2 Reverse 98.6 7.5 0.26
Alaknanda Fault 58.6 5.5 Strike-slip 29.3 6.7 0.24
Ramgarh Thrust Fault 42.8 5.1 Reverse 21.4 6.5 0.26

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

746 J Seismol (2019) 23:725–769

4.1.6 Region-specific ground motion prediction Himalaya, the Central Indian Peninsular Shield, and
equations the Indo-Gangetic Alluvium Basin at 0.2 Mw intervals.
The source functions for earthquake simulation using
An evaluation of seismic hazard, whether deterministic EXSIM have been obtained from published literatures
or probabilistic, requires an estimate of expected ground and listed in Table 7. The amplification due to the
motion at the site of interest. The most common means shallow crustal effects, considered an important attribute
of estimating this ground motion in engineering prac- for ground motion simulation at the crustal level, is used
tices, including probabilistic seismic hazard analysis, is in the ground motion synthesis. The 1D crustal velocity
to use an existing attenuation relationship that relates a model of the Indo-Gangetic Foredeep adopted from
specific strong ground motion parameter of ground Monsalve et al. (2006) as shown in Fig. 15a is used to
shaking in terms of PGA, peak ground velocity incorporate crustal amplification. The crustal amplifica-
(PGV), PSA, pseudo-spectral velocity (PSV), or peak tion as a function of frequency is presented in Fig. 15b
ground displacement (PGD), and intensity to one or and is calculated from the shear-wave velocity profile
more attributes of an earthquake (e.g., Campbell and using the quarter wavelength approximation (Boore and
Bozorgnia 2003; Atkinson and Boore 2003; Nath and Joyner 1997).
Thingbaijam 2011a; Nath et al. 2012). There has been a Figure 16 exhibits the simulated acceleration time
large volume of work already available on the develop- history as well as the corresponding spectrum of (a)
ment and application of the GMPEs; recent reviews can the 2015 Nepal earthquake of Mw 7.8 simulated at Patna
be found in Douglas (2003), Campbell (2003), Power City, (b) the 1997 Jabalpur earthquake of Mw 5.8 simu-
et al. (2008), Nath and Thingbaijam (2011a), and lated at Lucknow City, and (c) the 1934 Nepal-Bihar
Anbazhagan et al. (2015a). earthquake of Mw 8.1 simulated at Varanasi City.
The ground motion parameters at a site of interest are Thereupon, nonlinear regression analyses have been
evaluated by using a ground motion prediction equation performed for different shaking parameters Y (i.e., PGA,
that relates a specific strong motion parameter of ground PSA, PGV, and PGD at different periods) following
shaking to one or more seismic attributes (Campbell and least square error minimization to estimate the coeffi-
Bozorgnia 2003; Nath et al. 2012). The appropriate cients of NGA models following Atkinson and Boore
ground motion prediction equations are not only useful (2006) and Campbell and Bozorgnia (2003) ground
in rapid hazard assessment but also important for seis- motion prediction formulations as given in Eqs. (8)
mic risk analysis. The selection of a model for the and (9), respectively, for the three major tectonic prov-
prediction equation is important as it should not only inces viz. the Central Himalaya, the Central Indian
be realistic but also a practical one and neither too Peninsular Shield, and the Indo-Gangetic Alluvium Ba-
complex nor too simple. Due to paucity of good mag- sin. The fundamental models adopted for nonlinear
nitude coverage of strong ground motion data, analytical regression analysis are individually given as,
or numerical approaches for a realistic prognosis of
possible seismic effects in terms of tectonic regime, (a) Atkinson and Boore (2006) (BA06):
earthquake size, local geology, and near fault conditions
necessitate systematic ground motion synthesis. There
are several algorithms available for ground motion syn-
thesis. However, the finite-fault stochastic method is LogPSA ¼ C 1 þ C 2 M þ C 3 M 2 þ ðC 4 þ C 5 M Þ f 1
considered to be best suited over a large fault rupture
distance and also the source characteristics for near-field þ ðC 6 þ C 7 M Þf 2 þ ðC 8 þ C 9 M Þf 0
approximation (Nath et al. 2009, 2012, 2014). Thus, in
þ C 10 Rcd ð8Þ
the present study, the stochastic finite-fault simulation is
performed using EXSIM of Motazedian and Atkinson
(2005) for strong ground motion synthesis. In order to
Where;
create a strong ground motion database, we simulated f 0 ¼ maxðlogðR0 =Rcd Þ; 0Þ; f 1 ¼ minðlogRcd ; logR1 Þ;
earthquakes of Mw 3.5 to the maximum earthquake in f 2 ¼ maxðlogðRcd =R2 ÞÞ; R0 ¼ 10; R1 ¼ 70; R2 ¼ 140;
the three tectonic provinces, namely the Central

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 747

Table 7 Parameters used for strong ground motion simulation in the Indo-Gangetic Foredeep region

Parameter 1934 Nepal-Bihar earthquake 2015 Nepal earthquake 1997 Jabalpur earthquake

Strike (°) 100 (e) 293 (f) 65 (a)

Dip (°) 30 (e) 7 (f) 70 (a)
Focal depth (km) 20 (e) 15 (f) 35 (a)
Source (location) 26.60° N, 86.80° E (e) 27.97° N, 84.48° E (f) 23.08° N, 80.09° E (a)
Observed magnitude (Mw) 8.1 (e) 7.8 (f) 5.8 (a)
Stress (bar) 275 (e) 275 (e) 270 (b)
Crustal density (g/cm3) 2.7 (e) 2.7 (e) 2.9 (d)
Shear-wave velocity (β) (km/s) 3.6 (e) 3.6 (e) 3.9 (c)
Pulsating area (%) 25 25 25
Quality factor 400f0.47 (e) 400f0.48 (e) 800f0.72 (d)
Kappa 0.03 (e) 0.03 (e) 0.035 (d)
Geometrical spreading 1/R (R < 100 km)
1/R0.5 (R > 100 km)
Windowing function Saragoni and Hart (1974)
Damping 5%

Source parameters have been adopted from (a) Saikia (2006), (b) Singh et al. (1999b), (c) Bhattacharya et al. (2009), (d) Sengupta (2012), (e)
Nath et al. (2009), and (f) GCMT

M is the magnitude in Mw, Rcd represents fault dis- the Central Himalaya, the Central Indian Peninsular
tance in kilometers, and C1…C10 are the regression Shield, and the Indo-Gangetic Alluvium Basin
coefficients. The obtained regression coefficients for seismogenic zones in the Indo-Gangetic Tectonic

Fig. 15 a Crustal velocity model for the Indo-Gangetic Foredeep region (adopted from Monsalve et al. 2006). b First-order approximated
crustal amplification is estimated using the crustal velocity model following Boore and Joyner (1997) for the Indo-Gangetic Foredeep region

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

748 J Seismol (2019) 23:725–769

Fig. 16 Simulated acceleration time history and the correspond- at Lucknow City, and c the 1934 Nepal-Bihar earthquake of Mw
ing spectrum of a the 2015 Nepal earthquake of Mw 7.8 simulated 8.1 simulated at Varanasi City
at Patna City, b the 1997 Jabalpur earthquake of Mw 5.8 simulated

Province using this next generation attenuation model significant increase in ground motion over the hanging
are given in Table 8. wall effect. Moreover, the hanging wall effect dies out
for rseis < 8 km, or sooner if rjb ≥ 5 km or δ ≥ 70°. Hence,
(b) Campbell and Bozorgnia (2003) (CB03): in the present scenario, the hanging wall effect is not
considered and the prediction equation has been
modified after Campbell and Bozorgnia (2003) to gen-
erate the next generation attenuation model suitable for
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi the entire Indo-Gangetic Foredeep Tectonic Province.
lnY ¼ c1 þ f 1 ðM w Þ þ c4 ln f 2 ðM w ; rseis ; S Þ The regression coefficients of the NGA models worked
þ f 3 ð F Þ þ f 4 ðS Þ ð9Þ out for the Central Himalaya, the Central Indian Penin-
sular Shield, and the Indo-Gangetic Alluvium Basin
seismogenic zones contributing to the seismic hazard
Where; of the Indo-Gangetic Foredeep Tectonic Province using
f 1 ðM w Þ ¼ c2 M w þ c3 ð8:5−M w Þ2 ; the fundamental Eq. (9) are given in Table 9.
 h i2
f 2 ðM w ; rseis ; S Þ ¼ r2 seis þ gðS Þ2 exp c8 M w þ c9 ð8:5−M w Þ2 ; For establishing the accuracy of these six NGA
g ðS Þ ¼ c5 þ c6 ðS V FS þ S SR Þ þ c7 S FR ; models worked out for the Indo-Gangetic Foredeep
f 3 ð F Þ ¼ c10 F RV þ c11 F T H ; Tectonic Province, we compared in Fig. 17 the PGA
f 4 ðS Þ ¼ c12 S V FS þ c13 S SR þ c14 S FR values of the predicted BA06 NGA models with the
SVFS = 1 (very firm soil), SSR = 1 (soft rock), SFR = 1 simulated ones in the Central Himalaya, the Indo-
(firm rock), SVFS = SSR = SFR = 0 (firm soil), FTH = 1 Gangetic Alluvium Basin, and the Central Indian Pen-
(thrust faulting), FRV = 1 (reverse faulting), and FRV = insular Shield seismogenic zones with a satisfactory
FTH = 0 (strike-slip and normal faulting). Mw represents agreement prevailing among all the three seismogenic
the moment magnitude and rseis represents the closest source zones.
distance to seismogenic rupture. According to Campbell The predicted NGA BA06 and CB03 models have
and Bozorgnia (2003), the nonlinear site effects inherent further been validated using both PGA and PSA residual
in large ground motion on firm soil do not permit a assessment following the formulations,

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 749

Table 8 Regression coefficients obtained by considering the Atkinson and Boore (2006) (BA06) NGA model for the Central Himalaya, the
Central Indian Peninsular Shield, and the Indo-Gangetic Alluvium Basin seismogenic source zones

PSA (s) C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 Std (δ)

Central Himalayan seismogenic source

0.05 − 0.7455 0.4406 − 0.0291 − 1.8857 0.1438 − 2.9953 0.5128 0.952 − 0.277 − 0.0049 0.3815
0.08 − 1.0128 0.5 − 0.0337 − 1.5842 0.1082 − 2.9009 0.4906 0.952 − 0.257 − 0.0052 0.4174
0.1 − 1.0336 0.5451 − 0.0377 − 1.7607 0.1367 − 2.1889 0.3985 0.952 − 0.247 − 0.0052 0.3512
0.2 − 1.2901 0.6466 − 0.0478 − 1.7489 0.1371 − 1.234 0.2473 0.952 − 0.257 − 0.0045 0.3478
0.3 − 2.6941 0.9476 − 0.0633 − 1.9235 0.1625 − 0.9253 0.1883 0.952 − 0.267 − 0.0041 0.216
0.5 − 4.061 1.2266 − 0.0811 − 1.941 0.168 − 0.5131 0.1551 0.952 − 0.257 − 0.0038 0.214
1 − 6.1944 1.7613 − 0.1164 − 2.1874 0.1937 − 0.5261 0.1191 0.952 − 0.237 − 0.0028 0.2507
2 − 9.3155 2.3031 − 0.1433 − 1.8582 0.1511 − 0.2993 0.0806 0.952 − 0.207 − 0.0024 0.323
5 − 14.1237 3.4073 − 0.208 − 2.4696 0.2293 − 0.5616 0.1029 0.952 − 0.207 − 0.0018 0.1989
PGA − 2.3422 0.7954 − 0.0556 − 2.0541 0.1833 − 2.5559 0.4648 0.952 − 0.277 − 0.0047 0.207
Central Indian Peninsular Shield seismogenic source
0.05 0.5289 0.2638 − 0.0365 − 3.072 0.3317 − 1.4051 0.1883 0.952 − 0.19 − 0.0038 0.2294
0.08 0.1602 0.2812 − 0.0305 − 2.5821 0.2674 − 2.3559 0.2723 0.952 − 0.217 − 0.0039 0.2283
0.1 0.4877 0.2284 − 0.0309 − 2.7991 0.3019 − 0.9959 0.2254 0.952 − 0.177 − 0.0039 0.2174
0.2 − 0.2576 0.4371 − 0.0411 − 2.6964 0.2732 − 1.926 0.1921 0.952 − 0.167 − 0.0034 0.2019
0.3 − 0.3213 0.5301 − 0.0527 − 3.1015 0.3232 − 1.6503 0.1273 0.952 − 0.177 − 0.0031 0.2007
0.5 − 2.1742 0.8671 − 0.068 − 2.725 0.2665 − 1.6422 0.1479 0.952 − 0.117 − 0.0027 0.2326
1 − 3.899 1.471 − 0.1206 − 3.7025 0.403 − 1.5473 0.1222 0.952 − 0.117 − 0.0023 0.2206
2 − 8.9452 2.3302 − 0.1505 − 2.5103 0.2325 − 1.7994 0.1368 0.952 − 0.127 − 0.0018 0.1505
5 − 12.5784 3.0259 − 0.1798 − 2.7426 0.2386 − 1.8738 0.1449 0.952 − 0.127 − 0.0009 0.1615
PGA − 0.1533 0.285 − 0.0316 − 2.6765 0.2854 − 2.2989 0.1954 0.952 − 0.177 − 0.0034 0.1442
Indo-Gangetic Alluvium Basin seismogenic source
0.05 − 4.5733 1.366 − 0.0883 − 1.18 0.0678 − 4.9977 0.8277 0.952 − 0.217 − 0.0057 0.2801
0.08 − 5.1964 1.5088 − 0.0962 − 0.9589 0.044 − 4.9463 0.8168 0.952 − 0.211 − 0.006 0.2783
0.1 − 5.1886 1.5206 − 0.0989 − 0.9899 0.052 − 4.2116 0.7179 0.952 − 0.217 − 0.006 0.272
0.2 − 5.3926 1.5517 − 0.099 − 1.065 0.0641 − 2.6592 0.4877 0.952 − 0.205 − 0.0052 0.2546
0.3 − 5.8496 1.6774 − 0.1096 − 1.2695 0.0931 − 2.2537 0.428 0.952 − 0.198 − 0.0048 0.2465
0.5 − 6.7985 1.8814 − 0.1224 − 1.3851 0.1073 − 1.9417 0.3777 0.952 − 0.215 − 0.0043 0.2223
1 − 8.6572 2.2887 − 0.1453 − 1.5744 0.1169 − 1.7556 0.2888 0.952 − 0.217 − 0.003 0.1877
2 − 10.7097 2.6633 − 0.164 − 1.8841 0.1558 − 0.4748 0.1095 0.952 − 0.187 − 0.0024 0.1921
5 − 12.2901 3 − 0.1865 − 3.1612 0.3156 − 0.0963 0.0411 0.952 − 0.175 − 0.0018 0.2965
PGA − 8.0915 2.4098 − 0.1758 − 1.8169 0.1917 − 3.1208 0.6408 0.952 − 0.207 − 0.0064 0.3552

 Alluvium Basin, and the Central Indian Peninsular

Y os
residual ¼ log10 ð10Þ Shield seismogenic sources are presented in Fig. 18. It
Yp is evident that the residuals have a zero mean and are
uncorrelated with respect to fault distance. Apparently,
Where Yos is the simulated PGA/PSA and Yp is the residual analysis of PGA and PSA of the NGA models
estimated PGA/PSA from the empirical attenuation re- predicted in the present investigation is found to be
lations (BA06 and CB03 in this case). Residual plots for unbiased in regard to both the magnitude and fault
PGA as a function of fault distance for NGA BA06 distance and, therefore, can be used along with other
models for the Central Himalaya, the Indo-Gangetic already available ground motion prediction equations

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 Table 9 Regression coefficients obtained by considering the Campbell and Bozorgnia (2003) (CB03) NGA model for the Central Himalaya, the Central Indian Peninsular Shield, and the
750

Indo-Gangetic Alluvium Basin seismogenic source zones

PSA (s) C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 C12 C13 C14 Std (δ)

Indo-Gangetic Alluvium Basin seismogenic source

0.05 0.623 − 0.148 − 0.2463 − 1.4358 275.03 − 0.005 274.9157 − 0.4579 − 0.2167 0.343 0.351 − 0.123 − 0.138 4.4469 0.2446
0.08 0.1761 0.0163 − 0.1907 − 1.5192 − 2.2962 − 0.005 − 2.3553 0.1493 0.0115 0.343 0.351 − 0.123 − 0.138 4.0501 0.2328
0.1 0.0174 − 0.1918 − 0.2405 − 1.1078 0.01 − 0.005 − 0.049 − 0.1003 − 4.2413 0.343 0.351 − 0.123 − 0.138 4.0615 0.3457
0.2 0.3174 − 0.2038 − 0.2405 − 1.1078 0.01 − 0.005 − 0.049 − 0.1003 − 8.2413 0.343 0.351 − 0.123 − 0.138 4.0615 0.2926
0.3 0.7449 − 0.1648 − 0.2843 − 1.3725 − 9.6722 − 0.005 − 9.7235 − 0.4487 − 0.2908 0.343 0.351 − 0.123 − 0.138 4.2989 0.2195
0.5 − 0.5902 − 0.001 − 0.3416 − 1.0608 1.1891 − 0.005 1.1299 − 3.1363 − 1.5552 0.343 0.351 − 0.123 − 0.138 2.8967 0.3208
1 − 0.9479 − 0.0035 − 0.3416 − 1.0608 1.1891 − 0.005 1.1299 − 3.1363 − 1.5552 0.343 0.351 − 0.123 − 0.138 2.8967 0.1884
2 − 3.2836 0.5089 − 0.2873 − 1.0918 0.0675 − 0.005 0.0084 0.5757 − 1.07 0.343 0.351 − 0.123 − 0.138 0.5334 0.1969
5 − 4.0052 0.5008 − 0.2873 − 1.0918 0.0675 − 0.005 0.0084 0.5757 − 1.07 0.343 0.351 − 0.123 − 0.138 0.5334 0.2833
PGA 1.2504 − 0.2297 − 0.3442 − 1.6043 0.3256 − 0.005 0.2727 0.5024 0.0364 0.343 0.351 − 0.123 − 0.138 5.0343 0.2395
Central Indian Peninsular Shield seismogenic source
0.05 − 3.6481 0.8304 − 0.0363 − 1.1591 0.2557 − 0.005 0.1953 0.5847 − 0.0902 0.343 0.351 − 0.123 − 0.138 0.2958 0.309
0.08 − 3.0219 0.7676 − 0.0458 − 1.2751 − 6.4483 − 0.005 − 6.5651 0.1666 − 0.163 0.343 0.351 − 0.123 − 0.138 0.9021 0.2679
0.1 − 3.1429 0.8078 − 0.0414 − 1.2993 − 2.337 − 0.005 − 2.4472 0.2989 − 0.1008 0.343 0.351 − 0.123 − 0.138 0.9011 0.2396
0.2 − 2.7474 0.5366 − 0.1041 − 0.9915 4.2461 − 0.005 4.1871 − 5.212 − 8.2057 0.343 0.351 − 0.123 − 0.138 1.1965 0.2789
0.3 − 2.5868 0.4743 − 0.1324 − 0.9823 1.2727 − 0.005 1.2137 − 1.6439 − 85.4231 0.343 0.351 − 0.123 − 0.138 1.1572 0.2894
0.5 − 3.486 0.7416 − 0.0893 − 1.0423 0.1701 − 0.005 0.0214 0.6771 − 0.1275 0.343 0.351 − 0.123 − 0.138 0.258 0.1292
1 − 3.2346 0.5264 − 0.1963 − 0.9271 0.0844 − 0.005 0.0254 − 0.237 − 24.5888 0.343 0.351 − 0.123 − 0.138 0.5094 0.1324
2 − 3.3042 0.4827 − 0.1963 − 0.9971 0.0844 − 0.005 0.0254 − 0.237 − 24.5888 0.343 0.351 − 0.123 − 0.138 0.5094 0.1324
5 − 6.1476 1.0171 − 0.3852 − 0.6758 5.3608 − 0.005 4.2939 − 2.2643 − 5.3532 0.343 0.351 − 0.123 − 0.138 − 2.4036 0.149
PGA − 4.2815 0.8857 0.0062 − 1.0574 0.0302 − 0.005 − 0.0296 1.3869 0.1152 0.343 0.351 − 0.123 − 0.138 − 0.5375 0.1092
Central Himalayan seismogenic source
0.05 − 4.6171 0.8249 − 0.0453 − 1.301 0.109 − 0.005 0.0476 0.5442 − 0.0344 0.343 0.351 − 0.123 − 0.138 1.6568 0.3999
0.08 − 4.5497 0.8247 − 0.0745 − 1.3805 1.6047 − 0.005 1.5263 0.1816 − 0.0562 0.343 0.351 − 0.123 − 0.138 2.3243 0.3843
0.1 − 4.5521 0.8502 − 0.0731 − 1.4046 0.864 − 0.005 0.7859 0.2772 − 0.0411 0.343 0.351 − 0.123 − 0.138 2.2918 0.3825
0.2 − 4.1649 0.7825 − 0.0996 − 1.3054 − 8.4724 − 0.005 − 8.5518 − 0.0142 − 0.2785 0.343 0.351 − 0.123 − 0.138 2.1076 0.3304
0.3 − 3.7394 0.5234 − 0.1899 − 1.1248 0.0296 − 0.005 − 0.0335 − 5.2922 0.8288 0.343 0.351 − 0.123 − 0.138 2.7046 0.3559
0.5 − 3.345 0.5771 − 0.1731 − 1.1603 2.698 − 0.005 2.6386 − 0.2742 − 0.6062 0.343 0.351 − 0.123 − 0.138 1.959 0.2043
1 − 3.5845 0.4921 − 0.2619 − 1.0097 4.962 − 0.005 4.9026 0.0289 − 0.8963 0.343 0.351 − 0.123 − 0.138 1.8235 0.2459
2 − 3.7552 0.4953 − 0.3382 − 0.9977 3.7662 − 0.005 3.7071 − 0.3544 − 1.7092 0.343 0.351 − 0.123 − 0.138 0.9988 0.2945

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

5 − 3.9236 0.5269 − 0.5548 − 0.8862 0.3253 − 0.005 0.2669 1.3351 − 5.3911 0.343 0.351 − 0.123 − 0.138 − 0.1496 0.2768
PGA − 2.4378 0.5256 − 0.0826 − 1.285 0.5294 − 0.005 0.4704 0.353 − 0.1526 0.343 0.351 − 0.123 − 0.138 1.3661 0.1318
J Seismol (2019) 23:725–769
J Seismol (2019) 23:725–769 751

Fig. 17 The blue dots represent the simulated PGA and the red dots represent the predicted PGA from the predicted NGA models BA06 for
a the Indo-Gangetic Alluvium Basin, b the Central Himalaya, and c the Central Indian Peninsular Shield seismogenic sources

for the IGF and its adjoining region and also those
1 N
available for similar tectonic setup in a logic tree frame- LLH ¼ − ∑ log ðgðxi ÞÞ ð11Þ
work for seismic hazard assessment. N i¼1 2
Apart from the NGA models worked out as a part of
this investigation, we also incorporated some regional and
global prediction models based on the suitability test per- Where xi represents the observed data for i = 1,... N.
formed on each such model for the estimation of seismic The parameter N is the total number of events and g(xi)
hazard of the region. Altogether, we adopted a total of 15 is the likelihood that model g has produced the obser-
GMPEs as given in Table 10. The coefficients of nine vation xi. In this case, g is the probability density func-
GMPEs already available for the region have been adopted tion given by a GMPE to predict the observation pro-
from their original publications. GMPEs are selected and duced by an earthquake with magnitude M at a site i that
ranked through the ‘efficacy test,’ proposed by Scherbaum is located at a distance R from the source.
et al. (2009) which makes use of average sample log- The smaller the value of LLH, the higher is the ranking
likelihood (LLH) computation for the purpose of ranking. index of the GMPE. The ranking analysis has been carried
The method has been tested successfully by Delavaud out using macroseismic intensity data (Martin and Szeliga
et al. (2009) and applied in the Indian context by Nath 2010; https://earthquake.usgs.gov/) and the PGA–
and Thingbaijam (2011a) and Anbazhagan et al. (2015a). European Macroseismic Scale (EMS, Grünthal 1998) re-
The LLH is computed as, lation at rock sites as given in Nath and Thingbaijam (2011

Fig. 18 Residuals of PGA with respect to fault distance for a the Indo-Gangetic Alluvium Basin, b the Central Himalaya, and c the Central
Indian Peninsular Shield seismogenic sources

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

752 J Seismol (2019) 23:725–769

Table 10 Selected ground motion prediction equations for PSHA and the Central Indian Peninsular Shield seismogenic
of any cities and urban centers located in the Indo-Gangetic
sources for the IGF as illustrated in Table 11.
Foredeep region

Selected ground motion prediction equations (GMPEs) 4.1.7 PSHA logic tree framework for the Indo-Gangetic
Foredeep region
Seismogenic sources Reference and code in brackets

Indo-Gangetic Present NGAs: Campbell and Bozorgnia The seismic hazard at a particular site is usually quanti-
Alluvium Basin (2003) (CB03), Atkinson and Boore fied in terms of level of ground shaking observed in the
(2006) (BA06) region. The methodology for probabilistic seismic haz-
Global/regional GMPEs: NDMA (2010)
ard analysis incorporates how often the annual rate of
(NDMA08), Abrahamson and Silva
(2008) (AS08), Raghukanth and ground motion exceeds a specific value for various
Kavitha (2014) (RK14) return periods of hazard at a particular site of interest.
Central Himalaya Present NGAs: Campbell and Bozorgnia In the hazard computation, all the relevant sources and
(2003) (CB03), Atkinson and Boore possible earthquake events are considered. A synoptic
(2006) (BA06)
probabilistic seismic hazard model is generated at engi-
Global/regional GMPEs: Anbazhagan
et al. (2013b) (ANBAZ13), Sharma neering bedrock based on the protocol given by Nath
et al. (2009) (SHAR09), Chiou and and Thingbaijam (2012), Nath et al. (2014), Adhikari
Youngs (2008) (CY08) and Nath (2016), and Maiti et al. (2017). The basic
Central Indian Present NGAs: Campbell and Bozorgnia methodology of the probabilistic seismic hazard analy-
Peninsular Shield (2003) (CB03), Atkinson and Boore sis involves computation of ground motion thresholds
(2006) (BA06)
Global/regional GMPEs: Raghukanth and
that are exceeded with a mean return period of say
Iyengar (2007) (RI07), Toro (2002) 475 years/2475 years at a particular site of interest.
(TORO02), NDMA (2010) (NDMA08) The effects of all the earthquakes of different sizes
occurring at various locations for all the seismogenic
sources at various probabilities of occurrences are inte-
a). Figure 19 presents the intensity as a function of distance grated into one curve that shows the probability of
for the indicated earthquakes derived from the ground exceeding different levels of a ground motion parameter
motion prediction equations. The individual normalized at the site during a specified time period. The computa-
weights of each GMPE have been derived by preparing a tional formulation as developed by Cornell (1968),
pairwise comparison matrix (Saaty 1980, 2000). The rank- Esteva (1970), and McGuire (1976) is given as,
ing analysis has been performed based on LLH values
ν ða > AÞ ¼ ∑ λi ∫m ∫r ∫δ Pða > Ajm; r; δÞ f m ðmÞf r ðrÞ f Δ ðδÞdmdrdδ
along with the weight assigned to each GMPE for the i
Central Himalaya, the Indo-Gangetic Alluvium Basin, ð12Þ

Fig. 19 The intensity as a function of distance for the indicated Shield, b the Central Himalaya, and c the Indo-Gangetic Alluvium
earthquakes derived from the ground motion prediction equations Basin seismogenic sources
for suitability testing of GMPEs for a the Central Indian Peninsular

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 753

Table 11 The weights and ranks assigned to respective GMPEs minimum magnitude mmin and a maximum magnitude
based on the average LLH ranking in the Central Indian Peninsular
Mmax. fΔ(δ) is the probability density function (in log-
Shield, the Central Himalaya, and the Indo-Gangetic Alluvium
Basin seismogenic zones normal distribution) associated with the standard devia-
tion of the residuals in GMPE. The GMPEs are de-
Model LLH Rank Weight scribed as relationships between a ground motion pa-
Indo-Gangetic Alluvium Basin seismogenic source
rameter ‘Y’ (i.e., PGA, PGV, or PSA at different pe-
riods), earthquake magnitude ‘M,’ source-to-site dis-
CB03 (present study) 2.144 1 0.33
tance ‘R,’ and uncertainty or residual (δ) as,
BA06 (present study) 2.346 2 0.27
NDMA08 2.386 3 0.20 lnðY Þ ¼ f ðM; RÞ þ δ ð14Þ
AS08 2.510 4 0.13
RK14 2.511 5 0.07
The ground motion uncertainty δ is modeled as a
normal distribution with a standard deviation, σln,y.
Central Himalaya seismogenic source
Hence, the above equation can be expressed as,
CB03 (present study) 2.482 1 0.33
BA06 (present study) 2.546 2 0.27 lnðY Þ ¼ f ðM; RÞ þ εσln;y ð15Þ
SHAR09 2.552 3 0.20
ANGB13 2.577 4 0.13
Where ε is the normalized residual, which is also a
CY08 2.892 5 0.07
normal distribution with a constant standard deviation,
and σln,y is the standard deviation associated with the
Central Indian Peninsular Shield seismogenic source
GMPE. In the PSHA formulation as given in Eq. (12),
CB03 (present study) 2.201 1 0.33
standard deviation denoted by δ is basically the residual
BA06 (present study) 2.219 2 0.27
associated with each GMPE. The probability density
Toro02 2.225 3 0.20
function fΔ(δ) follows a lognormal distribution that can
NDMA08 2.303 4 0.13
be expressed as,
RI07 2.389 5 0.07
" #
1 ðlny−lnymr Þ2
f Δ ðδÞ ¼ pffiffiffiffiffiffi exp − ð16Þ
2πσln;y 2σ2 ln;y
where ν (a > A) is the annual frequency of exceedance of
ground motion amplitude A, λ is the annual activity rate where lnymr = f(M, R) is the functional form of the
for the ith seismogenic source for a threshold magnitude, prediction model in terms of magnitude, distance.
and function P yields probability of the ground motion Ground motion variability constitutes aleatory uncer-
parameter a exceeding A for a given magnitude m at tainty intrinsic to the definition of GMPEs and, conse-
source-to-site distance r. The corresponding probability quently, to that of PSHA. Computations based only on
density functions are represented by fm(m), fr(r), and the median ground motions ignoring the associated
fΔ(δ). The probability density function for the magni- variability are known to underestimate the hazards, es-
tudes is generally derived from the GR relation pecially at low annual frequencies of exceedance
(Gutenberg and Richter 1944). In practice this relation- (Bommer and Abrahamson 2006). The value of εmax
ship is truncated at some lower and upper magnitude ranging from 2 to 4 has often been employed in proba-
values which are defined as the truncation parameters bilistic seismic hazard estimations (e.g., Marin et al.
related to the minimum (mmin) and maximum (Mmax) 2004). However, truncation at εmax < 3 has been sug-
values of magnitude, obtained by different methods. gested to be inappropriate (e.g., Bommer and
The present implementation makes use of the density Abrahamson 2006). In the present study, truncation at
function given by Bender (1983) as, εmax = 4 is considered to be pragmatic and implemented
uniformly for all the GMPEs.
βexp½−βðm−mmin Þ The distance probability function fr(r) represents the
f m ðm Þ ¼ ð13Þ
1−exp½−βðM max−mminÞ probability of occurrence of a given earthquake at a
distance in the range (r, r + dr). In the present analysis,
instead of considering probability function for the
Where β = b ln(10), and b refers to the b value of the source-to-site distance distinctively, we have imple-
GR relation. The distribution is bounded within a mented gridded point locations within the source zone,

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

754 J Seismol (2019) 23:725–769

where finite-fault ruptures are constructed based on the assigned weights equal to 0.20, 0.35, and 0.45, re-
rupture dimensions estimated for each magnitude. spectively. The seismicity model parameters, namely
The hazard computation is performed using a the annual rate of earthquakes λ(m) and β pair, are
Poisson occurrence model given by Eq. (17) below on assigned weights of 0.36, while the respective ± 1
grid points covering the entire study region at a spacing standard deviation gets weight equal to 0.32. Similar
of 0.0005° × 0.0005°. weight allotment is performed for Mmax. The weights
are allocated following the statistical rationale sug-
P ¼ 1−e−λt ð17Þ gested by Grünthal and Wahlström (2006). In order to
define appropriate weights, the percentage of proba-
Where λ is the rate of occurrence of the event (annual bility mass in a normal distribution for the mean
activity rate) and t is the time period of exceedance. value and ± 1 standard deviation are considered cor-
With this, the annual rate of exceedance for an event responding to the center of two equal halves.
with 10% probability in 50 years is given by,
4.2 Surface-consistent probabilistic seismic hazard
λ ¼ −½lnð1−0:1Þ=50 ¼ 0:0021=year ð18Þ modeling

In the present study, site response for both short and

A logic tree framework depicted in Fig. 20 is long periods as provided by IBC (2006) pertaining to
employed in the computation of probabilistic seismic NEHRP site classes in the IGF is shown as represen-
hazard for the capital cities of Patna and Lucknow tative samples for the city of Lucknow in Table 12
and the famous Hindu religious city of Varanasi at which in comparison with regional site amplification
0.0005° × 0.0005° grid resolutions to incorporate factors derived through geophysical and geotechnical
multiple models in the source considerations, investigation for the same city by Anbazhagan et al.
GMPEs, and seismicity parameters. In the present (2010) depict a satisfactory agreement. These site
study, the seismogenic sources, i.e., tectonic and lay- factors on convolution with firm rock level PGA
ered polygonal sources, are assigned weights equal to and PSA values generated surface-consistent proba-
0.60 and 0.40, respectively. The three derivatives for bilistic seismic hazard of the cities of Patna, Luck-
the threshold magnitude of Mw 3.5, 4.5, and 5.5 are now, and Varanasi for 475 years of return period.

Fig. 20 A logic tree formulation for probabilistic seismic hazard computation at each node of the region gridded at 0.0005° × 0.0005°
interval

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 755

Table 12 Summary of site amplification factor given by IBC (2006) for NEHRP site classes and site amplification factors derived through
geophysical and geotechnical investigation for Lucknow City by Anbazhagan et al. (2010)

Spectral acceleration Site class Amplification factor derived through

A B C D E geophysical and geotechnical analysis
Short period, SAS (g) Short period amplification factor, Fa (Anbazhagan et al. 2010)
≤ 0.25 0.8 1.0 1.2 1.6 2.5 1.06–2.5
(0.25, 0.50] 0.8 1.0 1.2 1.4 1.7
(0.50, 0.75] 0.8 1.0 1.1 1.2 1.2
(0.75, 1.0] 0.8 1.0 1.0 1.1 0.9
> 1.0 0.8 1.0 1.0 1.0 0.9
1-s period, Sal [g] 1-s period amplification factor, Fv
≤ 0.1 0.8 1.0 1.7 2.4 3.5
(0.1, 0.2] 0.8 1.0 1.6 2.0 3.2
(0.2, 0.3] 0.8 1.0 1.5 1.8 2.8
(0.3, 0.4] 0.8 1.0 1.4 1.6 2.4
> 0.4 0.8 1.0 1.3 1.5 2.4

4.3 Seismic damage modeling the person conducting the survey completes based on
visual observation of the building and using a set of
The damage probability in various socioeconomic questionnaires on the data collection form. Based on the
clusters of the cities of Patna, Lucknow, and Varanasi RVS and building characteristics, we have selected six
has been estimated in relationship with a given model building types in the cities of Patna, Lucknow,
ground motion parameter to evaluate the building and Varanasi and those have been described as ‘IGW-
performance for a particular seismic event in an RCF2IL (PAGER/FEMA:C1L),’ ‘IGW-RCF21M
open-source MATLAB-based seismic risk assess- (PAGER/FEMA:C1M),’ ‘IGW-RCF11L (PAGER/
ment package like SELENA developed by NORSAR FEMA:C3L),’ ‘IGW-RCF21H (PAGER/FEMA:C1H),’
(Norwegian Seismic Array)/ICG (International Cen- ‘IGW-RCF11M (PAGER/FEMA:C3M),’ and ‘PAGER/
ter for Geohazards, Norway) and the University of FEMA:C3H’ based on the construction, height, and
Alicante, Spain, for systematic seismic risk assess- number of stories as followed in PAGER/FEMA (
ment using the capacity spectrum method. The meth- 2000) and Pathak et al. (2015) nomenclature illustrated
odology consists of (i) classification of buildings in in Table 13.
different model building types as per FEMA nomen-
clature, (ii) development of uniform hazard response
spectra for each socioeconomic cluster, (iii) defini- 4.3.2 Structural damage assessment using the capacity
tion of capacity and fragility curve for each model spectrum method
building type, and (iv) assessment of discrete damage
probability according to different damage states. The The capacity spectrum method (CSM) is a nonlinear
detailed computational work flow is used as in static analysis, which compares the capacity curve of a
SELENA presented in Fig. 21. structure in terms of force and displacement with the
seismic response spectrum (Freeman 1978). It consists
4.3.1 Definitions of major building typologies of steps like generation of the capacity spectrum, com-
in the cities of Patna, Lucknow, and Varanasi putation of the design response spectrum, and determi-
nation of performance point. Structural capacity is rep-
The rapid visual screening (RVS) procedure has been resented by a force–displacement curve. A pushover
developed to identify and screen buildings that are po- analysis is performed for a structure with increasing
tentially seismically hazardous (FEMA 2000). The RVS lateral forces, representing the inertial forces of the
procedure uses a methodology based on a sidewalk structure under seismic demand. The process is contin-
survey of a building and a data collection form, which ued till the structure becomes unstable.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

756 J Seismol (2019) 23:725–769

Fig. 21 Computational framework used in SELENA for structural damage assessment (modified after Molina et al. 2010; Nath 2016;
Ghatak et al. 2017)

Capacity curve A building capacity curve is a plot of assumed to deform beyond the ultimate point without
lateral load resistance as a function of characteristic loss of stability, but the structural system provides no
lateral displacement (Yeh et al. 2000). It can be derived additional resistance to lateral load. Figure 22 depicts
from a plot of base shear versus roof displacement when the capacity curves for IGW-RCF2IL (PAGER/
the building is subjected to equivalent static forces. The FEMA:C1L), IGW-RCF21M (PAGER/FEMA:C1M),
building capacity curve has three control points: design, and IGW-RCF21H (PAGER/FEMA:C1H) model build-
yield, and ultimate capacity. A building is typically ing types as obtained from NIBS (2002).

Table 13 The model building types identified in the cities of Patna, Lucknow, and Varanasi as per FEMA (2000), WHE-PAGER (2008),
and Pathak et al. (2015)

Sl. Building classification Building material Building typology Number PAGER/

no. (Pathak et al. 2015) of story FEMA

1 Nonductile moment RC moment frame with unreinforced masonry IGW-RCF11L 1–3 C3L
resisting frame infills—made of rectangular fired bricks
RC moment frame with unreinforced masonry IGW-RCF11M 4–6 C3M
infills—made of rectangular fired bricks
RC moment frame with unreinforced masonry – 7+ C3H
infills—made of rectangular fired bricks
2 Ductile moment RC frame with unreinforced masonry infills—made of IGW-RCF2IL 1–3 C1L
resisting frame rectangular fired bricks IGW-RCF21M 3–6 C1M
IGW-RCF21H 7+ C1H

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 757

Fig. 22 Representative capacity curves for IGW-RCF2IL (PAGER/FEMA:C1L), IGW-RCF21M (PAGER/FEMA:C1M), and IGW-
RCF21H (PAGER/FEMA:C1H) model building types (adopted from NIBS 2002)

Seismic demand input The design response spectrum vulnerability curves or fragility curves for five damage
(DRS) is defined as the smoothened plot of maximum states are essential, which are developed as lognormal
acceleration as a function of frequency or time period of probability distribution of damage from the capacity
vibration for specific damping ratio for earthquake ex- curve. The fragility curve of a particular building can
citations at the base of a single degree of freedom system be constructed by (i) selecting earthquake ground mo-
(Nath 2016). The earthquake actions are represented in tion in terms of PGA and PSA, (ii) defining fine limiting
the form of a design response spectrum in terms of PGA states for discrete damage levels as per FEMA guide-
and PSA. The scheme given by IBC (2006, 2009) scales lines (FEMA 1999), (iii) analyzing building response
the design spectrum corresponding to the short and long using inelastic dynamic analysis, and (iv) conducting
periods, respectively, as presented in Table 12. The risk analysis to obtain the probability of exceeding
computational procedure for the design response spec- various limiting states. In the present study, fragility
trum is given in the Appendix. The spectral displace- curves for all model building types have been adopted
ment has been calculated for the assessment of ultimate from NIBS (2002) as listed in Table 14. For an expected
capacity of the building as, displacement, cumulative probabilities are defined to
obtain discrete damage probabilities of a structure in
S D ¼ 9:8  S A  T 2 ð19Þ terms of ‘none,’ ‘slight,’ ‘moderate,’ ‘extensive,’ and
‘complete.’ The representative fragility curves for IGW-
Where SD is the spectral displacement, SA is the RCF2IL (PAGER/FEMA:C1L), IGW-RCF21M (PAG-
spectral acceleration in g, and T is the time period. ER/FEMA:C1M), and IGW-RCF21H (PAGER/
FEMA:C1H) model building types are presented in
Fragility curve The fragility curves express the proba- Fig. 23.
bility of structural damage due to earthquakes as a
function of ground motion indices viz. PGA and PSA. Determination of performance point for the computa-
For the computation of damage probabilities, tion of discrete damage probability The peak building

Table 14 Fragility curve parameters for each model building type defined by NIBS (2002)

Model building types Slight Moderate Extensive Complete

Median Beta Median Beta Median Beta Median Beta

IGW-RCF2IL (PAGER/FEMA:C1L) 0.10 0.64 0.12 0.64 0.21 0.64 0.36 0.64
IGW-RCF21M (PAGER/FEMA:C1M) 0.09 0.64 0.13 0.64 0.26 0.64 0.43 0.64
IGW-RCF21H (PAGER/FEMA:C1H) 0.08 0.64 0.12 0.64 0.21 0.64 0.35 0.64
IGW-RCF11L (PAGER/FEMA:C3L) 0.10 0.64 0.14 0.64 0.21 0.64 0.35 0.64
IGW-RCF11M (PAGER/FEMA:C3M) 0.09 0.64 0.14 0.64 0.25 0.64 0.41 0.64
PAGER/FEMA:C3H 0.08 0.64 0.13 0.64 0.27 0.64 0.43 0.64

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

758 J Seismol (2019) 23:725–769

Fig. 23 Representative fragility curves for IGW-RCF2IL (PAGER/FEMA:C1L), IGW-RCF21M (PAGER/FEMA:C1M), and IGW-
RCF21H (PAGER/FEMA:C1H) model building types (adopted from NIBS 2002)

response at the point of interaction of the capacity curve 5 Results and discussion
and the design response spectrum is used with fragility
curve for the estimation of damage state probability. The The hazard distribution is estimated for the source zones
cumulative damage probabilities of all the model build- at all the hypocentral depth ranges of 0–25, 25–70, and
ing types in terms of none, slight, moderate, extensive, 70–180 km separately and thereupon integrated to ob-
and complete have been calculated by, tain the holistic hazard value. Hazard curves exhibit the
probability of exceeding different ground motion pa-
rameters at a particular site of interest. Figure 24 depicts
" !#
1 Sd the seismic hazard curves for the cities of Patna, Luck-
p½dsjS d  ¼ Φ In ð20Þ now, and Varanasi corresponding to PGA and PSA at
βds S d;ds 0.2 and 1.0 s, respectively, at engineering bedrock. Both
2 and 10% probability of exceedance in 50 years have
been demarcated by dotted lines in the diagram present-
Where p[ds| Sd] = probability of being in or exceed- ing both 475 and 2475 years of return period scenarios
ing a damage state, ds; Sd = given spectral displacement at firm rock condition.
(inches); S ds = median value of Sd at which the building The seismic hazard maps of Patna City corre-
reaches the threshold of the damage state ds; βds = sponding to the spatial distribution of PGA and
lognormal standard deviation of spectral displacement PSA at 0.2, 0.3, and 1.0 s for 10% probability of
of damage state, ds; and Φ = standard normal cumulative exceedance in 50 years are depicted in Fig. 25 that
distribution function. Both S d,ds and βds depend on a exhibits a PGA variation of 0.138 to 0.149 g. The
building type and its seismic design level. regions of Takiapar, Panapur Taufir, Sadikpur, and

Fig. 24 Annual frequency of exceedance versus ground acceler- 0.2 and 1.0 s for uniform firm rock site condition. Both 10 and 2%
ation plots usually termed as seismic hazard curves for the selected probabilities of exceedance in 50 years have been demarcated by
locations like a Danapur in Patna City, b Aliganj in Lucknow City, horizontal dotted lines in each plot
and c BHU in Varanasi City for peak and spectral accelerations at

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 759

Fig. 25 Seismic hazard distribution maps of Patna City in terms of PGA and PSA at 0.2, 0.3, and 1.0 s for 10% probability of exceedance in
50 years at firm rock site condition

Bahadurpur are placed in the higher hazard zone, 0.185 g. The regions of Janakipuram, Shivaji Puram,
while a moderate hazard level is associated with the and Kamta are seen with a higher hazard value, while a
regions of Danapur, Deedarganj, Gardanibagh, and moderate hazard level is associated with the regions of
Ramkrishna Nagar. A low hazard level of PGA Thakurganj, Vikas Khand, and Aliganj. A low hazard
0.138 g is observed in the southern part of the city level implicated with PGA 0.168 g is observed in the
e n co m p as s i n g a r e a s o f A ni s ab a d , R a n i p u r, southern part of the city encompassing areas of Eldeco
Murlichack, and Chhoti Badalpura. The PSA distri- II, Nilmatha, and Munnu Khera. The PSA distribution
bution for the short period of 0.2 s exhibits a varia- for the short period of 0.2 s exhibits a variation between
tion between 0.207 to 0.238 g, and at 0.3 s, it is seen 0.297 and 0.338 g, and at 0.3 s, it is seen to vary from
to vary from 0.204 to 0.232 g, while for a longer 0.258 to 0.289 g, while for a longer period, spectral
period of 1.0 s, spectral acceleration is seen to vary acceleration at 1.0 s ranges from 0.109 to 0.126 g.
from 0.068 to 0.090 g. The seismic hazard maps of Varanasi City corre-
Figure 26 depicts the seismic hazard maps of Luck- sponding to the spatial distribution of PGA and PSA
now City corresponding to the spatial distribution of at 0.2, 0.3, and 1.0 s for 10% probability of exceed-
PGA and PSA at 0.2, 0.3, and 1.0 s for 10% probability ance in 50 years with a return period of 475 years are
of exceedance in 50 years with a return period of depicted in Fig. 27 that shows a PGA variation of
475 years exhibiting a PGA variation of 0.168 to 0.091 to 0.109 g. The regions of Lamhi, Balirampur,

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

760 J Seismol (2019) 23:725–769

Fig. 26 Seismic hazard distribution maps of Lucknow City in terms of PGA and PSA at 0.2, 0.3, and 1.0 s for 10% probability of
exceedance in 50 years at firm rock site condition

BHU, Newada, and Balirampur are seen with a The results presented here indicate that the hazard
higher hazard level, while a moderate hazard level distributions are significantly higher than those speci-
is associated with the regions of Jaitpura, Barthara, fied in the earlier published works as listed in Table 15.
and Kurauti township areas. A low hazard level of The differences in the estimated hazard distribution
PGA 0.091 g is observed in the northeastern part of compared to the previously published maps can be
the city encompassing the area of Hiramanpur town- attributed to several factors such as (a) inclusion of
ship. The PSA distribution for the short period of new NGAs developed in this study and also the em-
0.2 s exhibits a variation between 0.174 and ployment of multiple GMPEs as appropriate for similar
0.210 g, and at 0.3 s, it is seen to vary from 0.153 seismotectonic regimes globally which were not includ-
to 0.182 g, while for a longer period of 1.0 s, spectral ed in the earlier studies, (b) layered seismogenic source
acceleration is seen to vary from 0.039 to 0.050 g. framework considerations and smoothened gridded

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 761

Fig. 27 Seismic hazard distribution maps of Varanasi City in terms of PGA and PSA at 0.2, 0.3, and 1.0 s for 10% probability of exceedance
in 50 years at firm rock site condition

Table 15 Comparison among all the computed PGAs reported seismicity models conforming to the variation of
here and estimated by other researchers for the cities of Patna, seismotectonic attributes with hypocentral depth, (c)
Lucknow, and Varanasi for 10% probability of exceedance in
50 years at firm rock condition depth-wise active tectonic specific source classification
apart from the already considered layered polygonal
References Peak ground acceleration (g) sources, and (d) multiple models of activity rates for
both the layered polygonal and tectonic sources based
Patna Lucknow Varanasi
on intensive seismicity analysis.
Present study 0.138–0.149 0.167–0.184 0.091–0.109 To understand the applicability of probabilistic seismic
Anbazhagan et al. 0.03–0.165 – – hazard on vulnerability aspect, we calculated damage
(2015b) probability in various socioeconomic clusters of the cities
Sitharam et al. (2015) 0.10–0.15 0.15–0.2 0.05–0.1 of Patna, Lucknow, and Varanasi in relationship with the
Sitharam and 0.1–015 0.06–0.12 0.05–0.09 given ground motion parameters to evaluate building per-
Kolathayar (2013)
formance for a particular seismic event. The design re-
Kumar et al. (2013) – 0.035–0.07 –
sponse spectrum of pseudo-spectral acceleration, the peak
Nath and Thingbaijam 0.12–0.16 0.16–0.20 0.08–0.12
(2012) building response, and the cumulative damage probabili-
NDMA (2010) 0.04–0.05 0.04 0.03 ties have been calculated for all the model building types
Bhatia et al. (1999) 0.10 0.08 0.06 based on surface-consistent ground motion and existing
Khattri et al. (1984) 0.10 0.05 0.05 capacity and fragility curves. The seismic hazard maps
presented in Fig. 28 correspond to the spatial distribution

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

762 J Seismol (2019) 23:725–769

Fig. 28 Seismic hazard distribution in the cities of a Patna, b Lucknow, and c Varanasi in terms of PGA spatial variation for 10% probability
of exceedance in 50 years at the surface for a return period of 475 years

Fig. 29 The 5% damped design response spectra for the cities of Patna, Lucknow, and Varanasi at surface-consistent level compliant with
NEHRP site class D (Vs30 ~ 180–360 ms−1)

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 J Seismol (2019) 23:725–769 763

Fig. 30 The discrete damage states ‘ds’ considering all the model building types in the cities of a Patna, b Lucknow, and c Varanasi

of PGA which have been generated at the surface level by Figure 29 presents 5% damping design response
convolving those generated at firm rock condition with the spectra for the cities of Patna, Lucknow, and Varanasi
site amplification factor given by IBC (2009) in compli- at NEHRP site class D (Vs30 ~ 180–360 ms−1). The
ance with the site classes in the cities of Patna, Lucknow, spectral displacement has been computed based on the
and Varanasi. The estimated surface-consistent PGA varies intersection between the design response spectra and
from 0.222 to 0.238 g for Patna City, while the same varies capacity curve of a model building type in order to
in the range of 0.257 to 0.295 g for Lucknow City and locate the performance point. This performance point
0.146 to 0.172 g for Varanasi City. Thereafter, the MM in conjunction with the fragility curve of the model
intensity has been estimated from the surface-consistent building type estimates the damage state probability of
probabilistic PGA using the relationship given by Wald each model building type in all the socioeconomic clus-
et al. (1999). The predicted MM intensity varies from VII ters of the cities of Patna, Lucknow, and Varanasi to
to VIII for Patna City, while the same is seen to vary from generate a composite damage scenario of each of these
MM intensity VII to VIII for Lucknow City and VI–VII three cities.
for Varanasi City. On the contrary, the maximum observed It has been estimated that out of 5000 buildings in
intensity till date due to the entire past moderate to large Patna City, about 48% of the buildings are expected to
earthquakes that visited the IGF region varies from MM suffer from ‘moderate’ damage followed by ~ 16%
intensity V–VIII. ‘complete,’ ~ 22% ‘extensive,’ and ~ 8% ‘slight’ dam-

Fig. 31 The discrete damage probability computed from the FEMA:C1H),’ ‘IGW-RCF11M (PAGER/FEMA:C3M),’ and
cumulative damage probability of ‘IGW-RCF2IL (PAGER/ ‘PAGER/FEMA:C3H’ model building types in terms of ‘slight,’
FEMA:C1L),’ ‘IGW-RCF21M (PAGER/FEMA:C1M),’ ‘IGW- ‘moderate,’ ‘extensive,’ and ‘complete’ damage states for the cities
RCF11L (PAGER/FEMA:C3L),’ ‘IGW-RCF21H (PAGER/ of a Patna, b Lucknow, and c Varanasi

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

764 J Seismol (2019) 23:725–769

age, and almost 6% of the buildings are seismic resistant components, namely the seismogenic source
as shown in Fig. 30a. In Lucknow City, approximately models, seismicity analysis, and ground motion
42% of the buildings are expected to suffer from ‘mod- prediction equations in a logic tree framework. On
erate’ damage followed by ~ 16% ‘complete,’ ~ 24% the other hand, structural damage has been estimat-
‘extensive,’ and ~ 13% ‘slight’ damage. However, ap- ed for the probable earthquake scenario of the cities
proximately 5% of the buildings are seismic resistant in of Patna, Lucknow, and Varanasi for a return period
the city as collectively shown in Fig. 30b. On the other of 475 years with a view to possible disaster miti-
hand, in Varanasi City, nearly 47% of the buildings are gation and management. The produced high-
expected to suffer from ‘moderate’ damage followed by resolution probabilistic seismic hazard maps and
~ 12% ‘complete,’ ~ 27% ‘extensive,’ and ~ 10% damage scenario will provide substantial informa-
‘slight’ damage, and about 4% of the buildings are tion for the development of these cities in terms of
seismic resistant in the city as shown in Fig. 30c. The land use planning and design of future infrastruc-
discrete damage probability exhibits that ‘IGW-RCF2IL tures. The emergency response capabilities can be
(PAGER/FEMA:C1L),’ ‘IGW-RCF21M (PAGER/ significantly improved to reduce casualties by rap-
FEMA:C1M),’ and ‘IGW-RCF11L (PAGER/ id, selective, and effective use of provided services.
FEMA:C3L) building typologies will suffer the mini- The architects and civil engineers may also use this
mum damage, while ‘IGW-RCF21H (PAGER/ information to assess the failure risk of the existing
F E M A : C 1 H ) , ’ ‘ I G W - R C F 1 1 M ( PA G E R / structures and, thus, design future earthquake-
FEMA:C3M),’ and ‘PAGER/FEMA:C3H’ building ty- resistant structures in these cities in the Indo-
pologies will experience severe destruction during a Gangetic Foredeep region.
strong impending earthquake of Mw 7.2 as predicted in
this study in the cities of Patna, Lucknow, and Varanasi Acknowledgments The critical review and constructive sugges-
and its adjoining region. The discrete damage probabil- tions of both the anonymous reviewers and the Journal Editorial
ities for different model building types in the cities of Board are thankfully acknowledged. Sincere thanks are also due to
the handling editor of this manuscript for apt handling of the same.
Patna, Lucknow, and Varanasi are depicted in Fig. 31,
which exhibits that the ‘complete’ and ‘extensive’ dam-
age probabilities are higher in the city of Patna followed Funding information This work has been supported by the
Geosciences/Seismology Division of the Ministry of Earth Sci-
by Lucknow and Varanasi. Incidentally, it is observed
ences, Government of India, vide sanction order no. CS/EHRA/5/
that the unreinforced masonry buildings in the region 2013 dated: 19/26 June 2014 and 23 June 2015.
are the most seismically vulnerable ones (GSI 1939;
Nath 2016) and, therefore, possess the probability of Appendix
‘complete’ damage. The present study also exhibits that
all the buildings in these cities are susceptible to damage The computation procedure followed in the design re-
and destruction at different levels from ‘slight’ to ‘com- sponse as given by IBC (2006, 2009) is as given below:
plete’ damage states.
(1) Compute the maximum considered earthquake
spectral response acceleration at 0.2 s and 1.0 s
6 Conclusion periods as,

The seismic hazard analysis has emerged as an

S MS ¼ F a S s ð21Þ
important issue in high-risk urban centers across
the globe and is considered an integral part of
earthquake-induced disaster mitigation practices.
Probabilistic seismic hazard analysis provides use-
ful solutions for end users, mainly as input to seis- S ML ¼ F v S l ð22Þ
mic design. This study delivers a next generation
probabilistic seismic hazard model of the cities of
Patna, Lucknow, and Varanasi in the IGF with the Fa and Fv correspond to the amplification fac-
incorporation of various seismic hazard tors for acceleration response spectra at 0.2 and

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 765

1.0 s periods as listed in Table 8. Ss and Sl denote geotechnical earthquake engineering and soil dynamics.
May 24-29, San Diego
the spectral accelerations at the respective periods.
Anbazhagan P, Kumar A, Sitharam TG (2013a) Seismic site
(2) Compute the design basis earthquake spectral re- classification and correlation between standard penetration
sponse acceleration at 0.2 and 1.0 s periods as, test N value and shear wave velocity for Lucknow City in
Indo-Gangetic Basin. Pure Appl Geophys 170(3):299–318
Anbazhagan P, Kumar A, Sitharam TG (2013b) Ground motion
S DS ¼ 2=3 S MS ð23Þ prediction equation considering combined dataset of record-
ed and simulated ground motions. Soil Dyn Earthq Eng 53:
92–108
Anbazhagan P, Sreenivas M, Ketan B, Moustafa SS, Al-Arifi NSN
(2015a) Selection of ground motion prediction equations for
S DL ¼ 2=3 S ML ð24Þ seismic hazard analysis of peninsular India. J Earthq Eng
20(5):699–737
Anbazhagan P, Bajaj K, Patel S (2015b) Seismic hazard maps and
spectrum for Patna considering region-specific
(3) Determine the characteristic time periods as, seismotectonic parameters. Nat Hazards 78(2):1163–1195
Anderson JG, Wesnousky SG, Stirling MW (1996) Earthquake
size as a function of fault slip rate. Bull Seismol Soc Am 86:
683–690
T 0 ¼ 0:2 S DL =S DS ð25Þ Atkinson GM, Boore DM (2003) Empirical ground-motion rela-
T S ¼ S DL =S DS ð26Þ tions for subduction-zone earthquakes and their applications
to Cascadia and other regions. Bull Seismol Soc Am 93:
(4) Construct the design response spectra as 1703–1717
8 9 Atkinson GM, Boore DM (2006) Earthquake ground-motion pre-
< 0:6ðS DS =T 0 ÞT þ 0:4S DS ; for T ≤T 0 = dictions for eastern North America. Bull Seismol Soc Am 96:
Sa ¼ S DS ; for T ≥ T 0 and T ≤T s 2181–2205
: ; Azzaro R, Ferreli L, Michetti AM, Serva L, Vittori E (1998)
S DL =T ; for T ≥Ts
Environmental hazard of capable faults: the case of the
ð27Þ Pernicana fault, Mt. Etna, Sicily. Nat Hazards 17(2):147–162
Bagchi S, Raghukanth STG (2017) Seismic response of the central
part of Indo-Gangetic Plain. Journal of Earthquake
Where Sa is the design spectral response acceleration Engineering. Available at https://doi.org/10.1080
/13632469.2017.1323044
and T is the fundamental time period of the structure. Bender B (1983) Maximum likelihood estimation of b-values for
magnitude grouped data. Bull Seismol Soc Am 73:831–851
Besana GM, Ando M (2005) The central Philippine fault zone:
location of great earthquakes, slow events, and creep activity.
References Earth Planets Space 57:987–994
Bhatia SC, Kumar MR, Gupta HK (1999) A probabilistic seismic
Abrahamson N, Silva W (2008) Summary of the Abrahamson & hazard map of India and adjoining regions. Ann Geofis
Silva NGA ground-motion relations. Earthquake Spectra 42(6):1153–1166
24(1):67–97 Bhattacharya SN, Suresh G, Mitra S (2009) Lithospheric S-wave
velocity structure of the Bastar craton, Indian peninsula, from
Adhikari MD, Nath SK (2016) Site-specific next generation
surface-wave phase-velocity measurements. Bull Seismol
ground motion prediction models for Darjeeling-Sikkim
Soc Am 99(4):2502–2508
Himalaya using strong motion seismometry. J Indian
Bilham R, England P (2001) Plateau ‘pop up’ in the great 1897
Geophys Union 20(2):151–170
Assam earthquake. Nature 410:806–809
Aki K (1965) Maximum likelihood estimate of b in the formula BIS (2002) IS1893-2002 (part 1): Indian standard criteria for
log N = a - bM and its confidence limits. Bull Earthq Res earthquake resistant design of structure part 1—resistant
Instit 43:237–239 provisions and buildings. Bureau of Indian Standards, New
Aki K, Richards PG (1980) Quantitative seismology: theory and Delhi
methods. W. H. Freeman, San Francisco Bommer JJ, Abrahamson NA (2006) Why do modern probabilis-
Allen TI, Wald DJ (2009) On the use of high-resolution topo- tic seismic-hazard analysis often lead to increase hazard
graphic data as a proxy for seismic site conditions (Vs30). estimates? Bull Seismol Soc Am 96:1967–1977
Bull Seismol Soc Am 99:935–943 Boore DM, Joyner WB (1997) Site amplifications for generic rock
Allen TI, Gibson G, Brown A, Cull JP (2004) Depth variation of sites. Bull Seismol Soc Am 87:327–341
seismic source scaling relations: implications for earthquake Cáceres D, Monterroso D, Tavakoli B (2005) Crustal deformation
hazard in southeastern Australia. Tectonophysics 390:5–24 in northern Central America. Tectonophysics 404:119–131
Anbazhagan P, Kumar A, Sitharam TG (2010) Site response study Campbell KW (2003) Prediction of strong ground motion using
of deep soil column in Lucknow, India. Recent advances in the hybrid empirical method and its use in the development of

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

766 J Seismol (2019) 23:725–769

ground-motion (attenuation) relations in eastern North emergency response decision support system: the early
America. Bull Seismol Soc Am 93:1012–1033 post-earthquake damage assessment tool (EPEDAT).
Campbell KW, Bozorgnia Y (2003) Updated near-source ground- Earthquake Spectra 13(4):815–832
motion (attenuation) relations for the horizontal and vertical Engdahl ER, van der Hilst R, Buland R (1998) Global teleseismic
components of peak ground acceleration and acceleration earthquake relocation with improved travel times and proce-
response spectra. Bull Seismol Soc Am 93:314–331 dures for depth determination. Bull Seismol Soc Am 88(3):
Cassidy JF, Rogers GC (2004) The Mw 7.9 Denali fault earthquake 722–743
of 3 November 2002: felt reports and unusual effects across Esteva L (1970) Seismic risk and seismic design decisions. In:
western Canada. Bull Seismol Soc Am 94(6B):S53–S57 Hansen RJ (ed) Seismic design for nuclear power plants.
Chadha RK, Srinagesh D, Srinivas D, Suresh G, Sateesh A, Singh Massachusetts Institute of Technology Press, Cambridge,
SK, Pérez-Campos X, Suresh G, Koketsu K, Masuda T, MA, pp 142–182
Domen K, Ito T (2016) CIGN, a strong-motion seismic Federal Emergency Management Agency (FEMA) (2000)
network in central Indo-Gangetic Plains, foothills of Prestandard and commentary for the seismic rehabilitation
Himalayas: first results. Seismol Res Lett 87(1):37–46 of buildings. FEMA, Washington, D.C., p 356
Chandra U (1977) Earthquakes of peninsular India—a FEMA N (1999) Earthquake loss estimation methodology—
seismotectonic study. Bull Seismol Soc Am 67:1387–1413 HAZUS 99. Federal Emergency Management Agency and
Chandra U (1978) Seismicity, earthquake mechanisms and tecton- National Institute of Buildings Sciences, Washington DC
ics along the Himalayan mountain range and vicinity. Phys Frankel A (1995) Mapping seismic hazard in the central and
Earth Planet Inter 16(2):109–131 eastern United States. Seismol Res Lett 66:8–21
Chiou B, Youngs RR (2008) An NGA model for the average Freeman SA (1978) Prediction of response of concrete buildings to
horizontal component of peak ground motion and response severe earthquake motion, publication SP-55. American
spectra. Earthquake Spectra 24:173–215 Concrete Institute, Detroit, pp 589–605
Christova C (1992) Seismicity depth pattern, seismic energy and b Gardner JK, Knopoff L (1974) Is the sequence of earthquakes in
value depth variation in the Hellenic Wadati-Benioff zone. Southern California, with aftershocks removed, Poissonian?
Phys Earth Planet Inter 72:38–48 Bull Seismol Soc Am 64:1363–1367
Cornell CA (1968) Engineering seismic risk analysis. Bull Ghatak C, Nath SK, Devaraj N (2017) Earthquake induced deter-
Seismol Soc Am 58:1583–1606 ministic damage and economic loss estimation for Kolkata,
Cornell CA, Vanmarcke EH (1969) The major influence on seis- India. J Rehabil Civil Eng 5(2):1–24. https://doi.
mic risk, the 4th world conference on earthquake engineering, org/10.22075/jrce.2017.3090.1166
Santiago, Chile, pp. 69–93
Ghosh GK, Mahajan SK (2013) Intensity attenuation relation at
Das S, Henry C (2003) Spatial relation between main earthquake
Chamba–Garhwal area in north-west Himalaya with epicen-
slip and its aftershock distribution. Rev Geophys 41(3)
tral distance and magnitude. J Earth Syst Sci 122(1):107–122
Das S, Gupta ID, Gupta VK (2006) A probabilistic seismic hazard
Gomberg J, Bodin PP, Reasenberg A (2003) Observing earth-
analysis of northeast India. Earthquake Spectra 22:1–27
quakes triggered in the near field by dynamic deformations.
Dasgupta S, Mukhopadhyay B (2015) Historiography and com-
Bull Seismol Soc Am 93:118–138
mentary from archives on the Kathmandu (Nepal)-India
Grünthal G (1998) European Macroseismic Scale 1998. Cahiers
earthquake of 26 August 1833. Indian Journal of History of
du Centre Européen de Géodynamique et de Séismologie 15,
Science (INSA), 50
Luxembourg
Dasgupta S, Pande P, Ganguly D, Iqbal Z, Sanyal K, Venaktraman
NV, Dasgupta S, Sural B, Harendranath L, Mazumdar K, Grünthal G, Wahlström R (2006) New generation of probabilistic
Sanyal S, Roy A, Das LK, Misra PS, Gupta H (2000) seismic hazard assessment for the area Cologne/Aachen con-
Seismotectonic atlas of India and its environs. Geological sidering the uncertainties of the input data. Nat Hazards 38:
Survey of India, Calcutta, India, Special Publication 59:–87 159–176
Delavaud E, Scherbaum F, Kuehn N, Riggelsen C (2009) GSI (1939) The Bihar-Nepal earthquake of 1934. Geol Surv India,
Information-theoretic selection of ground-motion prediction Memo 73:391
equations for seismic hazard analysis: an applicability study GSI (1993) Bihar–Nepal earthquake, August 20, 1988. Geol Surv
using Californian data. Bull Seismol Soc Am 99:3248–3263 India Spec Publ 31:104
Dietz LD, Ellsworth WL (1990) The October 17, 1989, Loma Gutenberg B, Richter CF (1944) Frequency of earthquakes in
Prieta, California, earthquake and its aftershocks: geometry California. Bull Seismol Soc Am 34:185–188
of the sequence from high-resolution locations. Geophys Res Hainzl S, Scherbaum F, Beauval C (2006) Estimating background
Lett 17(9):1417–1420 activity based on interevent-time distribution. Bull Seismol
Douglas J (2003) Earthquake ground motion estimation using Soc Am 96:313–320
strong-motion records: a review of equations for the estima- Hancilar U, Tuzun C, Yenidogan C, Erdik M (2010) ELER
tion of peak ground acceleration and response spectral ordi- software—a new tool for urban earthquake loss assessment.
nates. Earth Sci Rev 61:41–104 Nat Hazards Earth Syst Sci 10(12):2677–2696
Duggal R, Sato N (1989) Damage report of the Bihar-Nepal IBC (2006) International Code Council, Inc., Country Club Hills,
earthquake of August 21, 1988 (no. 89-2, p. 12), Available Illinois
at http://www.ers.iis.u-tokyo.ac.jp/PDF/ERSNo.22/1989-03- IBC (2009) International Code Council, Inc., Country Club Hills,
No.22-09.pdf Illinois
Eguchi RT, Goltz JD, Seligson HA, Flores PJ, Blais NC, Heaton Iyengar RN, Ghosh S (2004) Microzonation of earthquake hazard
TH, Bortugno E (1997) Real-time loss estimation as an in greater Delhi area. Curr Sci 87(9):1193–1202

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 767

Jaiswal K, Sinha R (2004) Webportal on earthquake disaster Nath SK, Adhikari MD (2013) Next generation attenuation
awareness in India. Available at www.earthquakeinfo.org models and time independent probabilistic seismic hazard
Jaiswal K, Sinha R (2007) Probabilistic seismic-hazard estimation of Darjeeling-Sikkim Himalaya. Int J Earthq Eng Hazard
for peninsular India. Bull Seismol Soc Am 97:318–330 Mitig 1(1):29–54
Kanamori H, Anderson DL (1975) Theoretical basis of some Nath SK, Thingbaijam KKS (2009) Seismic hazard assessment—
empirical relations in seismology. Bull Seismol Soc Am a holistic microzonation approach. Nat Hazards Earth Syst
65(5):1073–1095 Sci 9(4):1445–1459
Kayabalia K, Akin M (2003) Seismic hazard map of Turkey using Nath SK, Thingbaijam KKS (2011a) Peak ground motion predic-
the deterministic approach. Eng Geol 69:127–137 tions in India: an appraisal for rock sites. J Seismol 15:295–
K a ya l J R (2 00 8) Mi c r o e a r t hq u a k e s e i s m o l o gy a nd 315
seismotectonics of South Asia, Springer Science & Nath SK, Thingbaijam KKS (2011b) Assessment of seismic site
Business Media conditions: a case study from Guwahati city, Northeast India.
Khattri KN, Rogers AM, Perkins DM, Algermissen ST (1984) A Pure Appl Geophys 168(10):1645–1668
seismic hazard map of India and adjacent areas. Nath SK, Thingbaijam KKS (2012) Probabilistic seismic hazard
Tectonophysics 108:93–134 assessment of India. Seismol Res Lett 83:135–149
Kijko A (2004) Estimation of the maximum earthquake magni- Nath SK, Raj A, Thingbaijam KKS, Kumar A (2009) Ground
tude. Pure Appl Geophys 161:1655–1681 motion synthesis and seismic scenario in Guwahati city—a
Kijko A, Graham G (1998) Parametric-historic procedure for stochastic approach. Seismol Res Lett 80(2):233–242
probabilistic seismic hazard analysis: part I—estimation of Nath SK, Thingbaijam KKS, Maiti SK, Nayak A (2012) Ground-
maximum regional magnitude mmax. Pure Appl Geophys motion predictions in Shillong region, northeast India. J
152:413–442 Seismol 16:475–488
Kumar A, Anbazhagan P, Sitharam TG (2013) Seismic hazard Nath SK, Thingbaijam KKS, Adhikari MD, Nayak A, Devaraj N,
analysis of Lucknow considering local and active seismic Ghosh SK, Mahajan AK (2013) Topographic gradient based
gaps. Nat Hazards 69:327–350 site characterization in India complemented by strong
Mahajan AK, Thakur VC, Sharma ML, Chauhan M (2010) ground-motion spectral attributes. Soil Dyn Earthq Eng 55:
Probabilistic seismic hazard map of NW Himalaya and its 233–246
adjoining area, India. Nat Hazards 53(3):443–457
Nath SK, Adhikari MD, Maiti SK, Devaraj N, Srivastava N,
Maiti SK, Nath SK, Adhikari MD, Srivastava N, Sengupta P,
Mohapatra LD (2014) Earthquake scenario in West Bengal
Gupta AK (2017) Probabilistic seismic hazard model of
with emphasis on seismic hazard microzonation of the city of
West Bengal, India. J Earthq Eng 21:1113–1157
Kolkata, India. Nat Hazards Earth Syst Sci 14:2549–2575
Marin S, Avouac JP, Nicolas M, Schlupp A (2004) A probabilistic
Nath SK, Mandal S, Adhikari MD, Maiti SK (2017) A unified
approach to seismic hazard in metropolitan France. Bull
earthquake catalogue for South Asia covering the period
Seismol Soc Am 94:2137–2163
1900-2014. Nat Hazards 85(3):1787–1810
Mark RK (1977) Application of linear statistical model of earth-
quake magnitude versus fault length in estimating maximum NDMA (2010) Development of probabilistic seismic hazard map
expectable earthquakes. Geology 5:464–466 of India, technical report publicized by Govt. of India, New
Martin S, Szeliga W (2010) A catalog of felt intensity data for 570 Delhi, Working committee of experts (WCE), NDMA, avail-
earthquakes in India from 1636 to 2009. Bull Seismol Soc a b l e a t h t t p: / / w w w. h p sd ma .n ic .i n/ De v elo p m e nt
Am 100:562–569 % 2 0 o f % 2 0 P r o b a b l i s t i c
McGuire RK (1976) FORTRAN computer program for seismic %20Seismic%20Hazard%20Map%20of%20India.pdf
risk analysis. US Geological Survey:76–67 Ni J, Barazangi M (1984) Seismotectonics of the Himalayan
Molina S, Lindholm C (2005) A logic tree extension of the collision zone: geometry of the underthrusting Indian plate
capacity spectrum method developed to estimate seismic risk beneath the Himalaya. J Geophys Res Solid Earth 89(B2):
in Oslo, Norway. J Earthq Eng 9(6):877–897 1147–1163
Molina S, Lang DH, Lindholm CD (2010) SELENA—an open- NIBS (2002) Earthquake loss methodology, HAZUS 99, USA
source tool for seismic risk and loss assessment using a logic Nyffenegger P, Frohlich C (2000) Aftershock occurrence rate
tree computation procedure. Comput Geosci 36(3):257–269 decay properties for intermediate and deep earthquake se-
Monsalve G, Sheehan A, Schulte-Pelkum V, Rajaure S, Pandey quences. Geophys Res Lett 27:1215–1218
MR, Wu F (2006) Seismicity and one-dimensional velocity Okazaki K, Villacis C, Cardona C, Kaneko F, Shaw R, Sun J, ...,
structure of the Himalayan collision zone: earthquakes in the Tobin LT (2000) RADIUS: risk assessment tools for diagno-
crust and upper mantle. J Geophys Res Solid Earth 111(B10) sis of urban areas against seismic disasters. UN. International
Motazedian D, Atkinson GM (2005) Stochastic finite-fault model- Strategy for Disaster Reduction (ISDR). sd Geneva. CH.
ing based on a dynamic corner frequency. Bull Seismol Soc Page R (1968) Afetrshocks and micro aftershocks of the great
Am 95(3):995–1010 Alaska earthquake of 1964. Bull Seismol Soc Am 58(3):
Nasu N (1935) The great Indian earthquake of January 15, 1934. 1131–1168
Bull Earthq Res Inst 13(2):417–432 Pandey MR, Molnar P (1988) The distribution of intensity of the
Nath SK (2011) Seismic microzonation handbook. © Ministry of Bihar-Nepal earthquake of 15 January 1934 and bounds on
Earth Sciences, Government of India, New Delhi the extent of the rupture zone. J Nepal Geol Soc 5(1):22–44
Nath SK (2016) Seismic hazard, vulnerability and risk Parvez IA, Vaccari F, Panza GF (2003) A deterministic seismic
microzonation atlas of Kolkata, © Ministry of Earth hazard map of India and adjacent areas. Geophys J Int 155:
Sciences, Government of India, New Delhi, 530p 489–508

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

768 J Seismol (2019) 23:725–769

Pathak J, Bharali R, Deka B, Pathak S, Ahmed I, Dominik HL, Shukla J, Choudhury D (2012) Estimation of seismic ground
Meslem A (2015) Building classification scheme and vulner- motions using deterministic approach for major cities of
ability model for the city of Guwahati, Assam, EQRISK- Gujarat. Nat Hazards Earth Syst Sci 12:2019–2037
Technical Report, available at https://www.researchgate. Shukla UK, Raju NJ (2008) Migration of the Ganga river and its
net/publication/301198716 implication on hydro-geological potential of Varanasi area,
Phillips WM (2011) Redme documentation NEHRP site class map UP, India. J Earth Syst Sci 117(4):489–498
of Teton County, Idaho. Available at http://www. Singh DD, Gupta HK (1980) Source dynamics of two great
idahogeology.org/PDF/Digital_Data_(D)/Digital_ earthquakes of the Indian subcontinent: the Bihar-Nepal
D a t a b a s e s _ ( D D ) / D D 6 _ Te t o n _ N E H R P _ earthquake of January 15, 1934 and the Quetta earthquake
Liquefaction/NEHRP/README_Documentation_NEHRP_ of May 30, 1935. Bull Seismol Soc Am 70:757–773
Site_Class_Map_Teton_County_Idaho.PDF Singh SK, Ordaz M, Dattatrayam RS, Gupta HK (1999) A spectral
Power M, Chiou B, Abrahamson NA, Roblee C, Bozorgnia Y, analysis of the 21 May 1997, Jabalpur, India, earthquake (Mw
Shantz T (2008) An introduction to NGA. Earthquake 5.8) and estimation of ground motion from future earth-
Spectra 24:3–21 quakes in the Indian shield region. Bull Seismol Soc Am
Raghukanth STG, Iyengar RN (2007) Estimation of seismic spec- 89(6):1620–1630
tral acceleration in peninsular India. Journal of Earth System Sitharam TG, Kolathayar S (2013) Seismic hazard analysis of
Science 116:199–214 India using areal sources. J Asian Earth Sci 62:647–653
Raghukanth STG, Kavitha B (2014) Ground motion relations for Sitharam TG, Kolathayar S, James N (2015) Probabilistic assess-
active regions in India. Pure Appl Geophys 171(9):2241– ment of surface level seismic hazard in India using topo-
2275 graphic gradient as a proxy for site condition. Geosci Front
Rai DC, Singhal V, Raj SB, Sagar SL (2016) Reconnaissance of 6(6):847–859
the effects of the M7. 8 Gorkha (Nepal) earthquake of April Slemmons DB, McKinney R (1977) Definition of ‘active fault’,
25, 2015. Geomat, Nat Haz Risk 7(1):1–17 Miscellaneous Paper S-77-8, U.S. Army Engineer
Reasenberg PA (1985) Second-order moment of Central Waterways Experiment Station, Soils and Pavements
California seismicity. J Geophys Res 90:5479–5495 Laboratory, pp. 23, Vicksburg, Missouri
Saaty TL (1980) The analytic hierarchy process. McGraw-Hill Srinivas D, Srinagesh D, Chadha RK, Ravi Kumar M (2013)
International, New York Sedimentary thickness variations in the Indo-Gangetic
Saaty TL (2000) Models, methods, concepts and application of the foredeep from inversion of receiver functions. Bull Seismol
analytical hierarchy process. Kluwer Academic, Boston Soc Am 103:2257–2265
Saikia CK (2006) Modeling of the 21 May 1997 Jabalpur earth-
Srivatsava AB (2001) District resource maps of Geological Survey
quake in central India: source parameters and regional path
of India, Kanpur, Uttar Pradesh, India, available at www.
calibration. Bull Seismol Soc Am 96(4A):1396–1421
portal.gsi.gov.in
Saragoni GR, Hart GC (1974) Simulation of artificial earthquakes.
Stork AL, Selby ND, Heyburn R, Searle MP (2008) Accurate
Earthq Eng Struct Dyn 2:249–268
relative earthquake hypocenters reveal structure of the
Scherbaum F, Delavaud E, Riggelsen C (2009) Model selection in
Burma subduction zone. Bull Seismol Soc Am 98(6):2815–
seismic hazard analysis: an information-theoretic perspec-
2827
tive. Bull Seismol Soc Am 99:3234–3247
Schorlemmer D, Neri G, Wiemer S, Mostaccio A (2003) Stability Thitimakorn T, Channoo S (2012) Shear wave velocity of soils and
and significance tests for b-value anomalies: example from NEHRP site classification map of Chiang Rai City, northern
the Tyrrhenian Sea. Geophys Res Lett 30(16):1835 Thailand. Electron J Geotech Eng 17:2891–2904
Schorlemmer D, Wiemer S, Wyss M (2005) Variations in earth- Toro GR (2002) Modification of the Toro et al. (1997) Attenuation
quake size distribution across different stress regimes. Nature equations for large magnitudes and short distances, Risk Eng.
437:539–542 Inc.
Seeber L, Armbruster JG, Quittmeyer RC (1981) Seismicity and Tsapanos TM (2000) The depth distribution of seismicity param-
continental subduction in the Himalayan arc, Zagros Hindu eters estimated for the South American area. Earth Planet Sci
Kush Himalaya Geodynamic Evolution. American Lett 180(1):103–115
Geophysics Union 215–242 Uhrhammer RA (1986) Characteristics of northern and central
Sengupta P (2012) Stochastic finite-fault modelling of strong California seismicity. Earthq Notes 57:21
earthquakes in Narmada south fault, Indian shield. J Earth Utsu T (1965) A method for determining the value of b in the
Syst Sci 121(3):837–846 formula log N=a-bM showing the magnitude–frequency re-
Seyrek E, Tosun H (2011) Deterministic approach to the seismic lation for the earthquakes. Geophys Bull Hokkaido Univ 13:
hazard of dam sites in Kizilirmak basin. Turkey Nat Hazards 99–103
59:787–800 Utsu T (2002) Statistical features of seismicity. Int Geophys Ser
Sharma ML, Douglas J, Bungum H, Kotadia J (2009) Ground- 81(A):719–732
motion prediction equations based on data from the Valdiya KS (1976) Himalayan transverse faults and folds and their
Himalayan and Zagros regions. J Earthq Eng 13(8):1191– parallelism with subsurface structures of north Indian plains.
1210 Tectonophysics 32(3–4):379–386
Shrikhande M, Basu S, Mathur BC, Mathur AK, Das JD, Bansal Wald DJ, Quitoriano V, Heaton TH, Kanamori H (1999)
MK (2000) Strong motion data. Report on Chamoli Relationships between peak ground acceleration, peak
Earthquake of March 29, 1999. Department of Earthquake ground velocity, and modified Mercalli intensity in
Engineering, University of Roorkee, Roorkee, pp 27–43 California. Earthquake Spectra 15(3):557–564

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Seismol (2019) 23:725–769 769

Wells DL, Coppersmith KJ (1994) New empirical relationships Yin An, Dubey CS, Kelty TK, Webb, AAG, Harrison TM, Chou
among magnitude, rupture length, rupture width, rupture CY, Julien C (2009) Geologic correlation of the Himalayan
area, and surface displacement. Bull Seismol Soc Am 84: orogen and Indian craton: Part 2. Structural geology, geo-
974–1002 chronology, and tectonic evolution of the Eastern Himalaya.
WHE-PAGER (2008). WHE-PAGER phase 2, development of Geol Soc Am Bull : B26461-1.
analytical seismic vulnerability functions. EERI-WHE-US Zhuang J, Ogata Y, Vere-Jones D (2002) Stochastic declustering of
Geological Survey (Available at http://pager.world-housing. space-time earthquake occurrences. J Am Stat Assoc 97:369–
net/) 380.
Wong I, Silva W, Bott J, Wright D, Thomas P, Gregor N, Li S,
Mabey M, Sojourner A, Wang Y (2000) Earthquake scenario Publisher’s note Springer Nature remains neutral with regard to
and probabilistic ground shaking maps for the Portland, jurisdictional claims in published maps and institutional
Oregon, metropolitan area. Portland Hills fault M, 6 affiliations.
Yeh CH, Jean W, Loh CH (2000) Building damage assessment for
earthquake loss estimation in Taiwan. Twelfth World
Conference on Earthquake Engineering Vol 30

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”),
for small-scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are
maintained. By accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use
(“Terms”). For these purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or
a personal subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or
a personal subscription (to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the
Creative Commons license used will apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data
internally within ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking,
analysis and reporting. We will not otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of
companies unless we have your permission as detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that
Users may not:

1. use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to
circumvent access control;
2. use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil
liability, or is otherwise unlawful;
3. falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by
Springer Nature in writing;
4. use bots or other automated methods to access the content or redirect messages
5. override any security feature or exclusionary protocol; or
6. share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer
Nature journal content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates
revenue, royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain.
Springer Nature journal content cannot be used for inter-library loans and librarians may not upload Springer Nature journal
content on a large scale into their, or any other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any
information or content on this website and may remove it or features or functionality at our sole discretion, at any time with or
without notice. Springer Nature may revoke this licence to you at any time and remove access to any copies of the Springer Nature
journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express
or implied with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or
warranties imposed by law, including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be
licensed from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other
manner not expressly permitted by these Terms, please contact Springer Nature at

[email protected]