Nat Hazards (2013) 68:1089–1108

DOI 10.1007/s11069-013-0678-y

ORIGINAL PAPER

Probabilistic seismic hazard analysis of Tripura

and Mizoram states

Arjun Sil • T. G. Sitharam • Sreevalsa Kolathayar

Received: 25 November 2011 / Accepted: 1 April 2013 / Published online: 23 April 2013
Ó Springer Science+Business Media Dordrecht 2013

Abstract A probabilistic seismic hazard analysis for the states of Tripura and Mizoram in
North East India is presented in this paper to evaluate the ground motion at bedrock level.
Analyses were performed considering the available earthquake catalogs collected from
different sources since 1731–2010 within a distance of 500 km from the political
boundaries of the states. Earthquake data were declustered to remove the foreshocks and
aftershocks in time and space window and then statistical analysis was carried out for data
completeness. Based on seismicity, tectonic features and fault rupture mechanism, this
region was divided into six major seismogenic zones and subsequently seismicity
parameters (a and b) were calculated using Gutenberg–Richter (G–R) relationship. Faults
data were extracted from SEISAT (Seismotectonic atlas of India, Geological Survey of
India, New Delhi, 2000) published by Geological Survey of India and also from satellite
images. The study area was divided into small grids of size 0.05° 9 0.05° (approximately
5 km 9 5 km), and the hazard parameters (rock level peak horizontal acceleration and
spectral accelerations) were calculated at the center of each of these grid cells considering
all the seismic sources within a radius of 500 km. Probabilistic seismic hazard analyses
were carried out for Tripura and Mizoram states using the predictive ground motion
equations given by Atkinson and Boore (Bull Seismol Soc Am 93:1703–1729, 2003) and
Gupta (Soil Dyn Earthq Eng 30:368–377, 2010) for subduction belt. Attenuation relations
were validated with the observed PGA values. Results are presented in the form of hazard
curve, peak ground acceleration (PGA) and uniform hazard spectra for Agartala and Ai-
zawl city (respective capital cities of Tripura and Mizoram states). Spatial variation of
PGA at bedrock level with 2 and 10 % probability of exceedance in 50 years has been
presented in the paper.

Keywords Seismicity  Faults  Ground motion  PSHA  Hazard curves 

Spectral acceleration

A. Sil (&)  T. G. Sitharam  S. Kolathayar

Department of Civil Engineering, IISc, Bangalore 560012, India
e-mail: [email protected]; [email protected]

123
1090 Nat Hazards (2013) 68:1089–1108

1 Introduction

Tripura and Mizoram are the North Eastern states in India located between the latitude
21.91°–24.50°N, longitude 91.15°–93.50°E (Fig. 1). The seismic zone of North East India
is considered as one of the six most active plate tectonic regions in the world (Kayal 1998;
Tiwari 2000). It is well established that Indian plate is continuously dipping down under
Eurasian plate, which caused to the formation of the Himalayas (Oldham 1882, 1899;
Kayal 1987; Kumar et al. 2005; Mukhopadhyay and Dasgupta 1988; Mukhopadhyay 1992;
Nandy 2001; Tandon and Srivastava 1975; Mitra et al. 2005). The rate of movement
between the Indian plate and the Eurasian plate converge about 45 mm/year (Kumar et al.
2007). The latest version of the seismic zoning map of India given in the earthquake
resistant design code of India [BIS 1893 (Part 1) 2002] assigns four levels of seismicity for
the entire India in terms of different zone factors. The study region has been placed in
seismic zone V as per BIS-1893-(2002), which has a zone factor of 0.36 g. The main
drawback of the seismic zonation code of India (BIS-1893 2002) is that, it is based on the
past seismic activity, not based on a comprehensive scientific seismic hazard analysis. In
the study area, five historical earthquakes of Mw [8 have occurred and 15 events of
magnitude Mw [7 have occurred in the last (1910–2010) 100 years. The region is sub-
jected to both intraplate and interplate (including subduction) seismic activities, because of
its complex seismotectonic setting. The differential movements of the plate are capable of
producing future ground motion that could damage the area. For example, in Shillong
plateau (SP), the depth ranges from 40 to 80 km, whereas in Indo-Burmese region (IBR),
the depth ranges from 90 to 200 km (Kayal 1987, 1998; Nandy 2001). Many researchers
have suggested that the Himalayan arc is involved in a collision tectonics between the
plate-to-plate boundaries, whereas the Burmese arc is involved in a subduction process
(Verma et al. 1976; Mukhopadhayay and Dasgupta 1988; Kayal 2008). Researchers have
found a big seismic gap known as Assam gap (Khattri and Weiss 1978). Since 1950, no
major event has occurred in between the Eastern Himalaya (EH), SP and in the IBR in
comparison to the past history. Khattri and Weiss (1978) have suggested and expected a
big earthquake in this region at any time in the future. In 2011, the study region experi-
enced three earthquakes of magnitude M 6.4 (Feb 2011), M4.2 (June 23, 2011) and M4.3
(April 19, 2011) (IMD, India). The epicenters of these quakes were located at the border of
IBR and another was in the Bengal basin (BB) in SW direction with respect to the study
area.
Parvez and Ram (1997), Sharma and Malik (2006), Das et al. (2006) and Yadav et al.
(2010) have made attempts to study the seismicity of North East India. Earlier works have
been carried out by various researchers to assess the seismic hazard of specific regions of
North East India like Guwahati (Raghukanth and Dash 2010; Nath et al. 2008, 2009),
Sikkim (Nath et al. 2008, 2009), Manipur (Kumar Pallav 2010), etc. Although many
researchers have reported about the status of seismicity on North Eastern Region, only very
few studies have been carried out for Tripura and Mizoram region. From a review of
available literature, no specific study exists for Tripura and Mizoram till date, whereas
almost the whole of the study area is highly vulnerable for severe shaking, amplification,
liquefaction, and landslides. Because of the increasing population and rapid growth of
unplanned urban infrastructures that inhabits this region; it becomes prudent to carry out
the seismic hazard analysis using the latest data and methodology available, which will
assist the designer/planner and the governments to mitigate the hazard.
The widely used methods for seismic hazard assessment are deterministic seismic
hazard (DSHA) and probabilistic seismic hazard (PSHA). In DSHA, hazard is evaluated

123
Nat Hazards (2013) 68:1089–1108 1091

Fig. 1 Location map of the study area (Scale 1:1 million)

considering the close distance between the source and site of interest and past maximum
magnitude occurred within the fault. In hazard analysis, epistemic and aleatory uncer-
tainties can be accounted for within a probabilistic framework. The PSHA procedure
assumes that the distances and magnitudes probability are distributed uniformly throughout
the fault rupture. The hazard assessment, mainly involves collection of events and iden-
tification of seismic sources, developing new seismotectonic models and selection of
appropriate ground motion models. Two suitable attenuation relations were selected, which
have been validated with the actual instrumental recorded data available for IBR sub-
duction zone and North East India. The hazard has been calculated using the updated
PSHA methodologies for 2 and 10 % probabilities of exceedance in 50 years.

2 Seismicity of the study area

The study area covers the latitude 18.40°–29°N and longitude 86.20°–97.30°E within
500 km radial distance from the political boundary of both the states (Fig. 2). It covers all

123
1092 Nat Hazards (2013) 68:1089–1108

Fig. 2 Seismotectonic map of the study area with identified sources and events (Scale 1:1 million)

the North Eastern states of India, whole of Bangladesh and west part of Burma in the east,
south part of China in the north, some part of the Bay of Bengal in the south, and part of
West Bengal in the west direction (Fig. 1).
The region was subdivided into six sub major zones like IBR, SP, Eastern Himalayan
arc (EH), Bengal Basin (BB), Mishmi Thrust (MT) and Naga Thrust (NT). Identification
and characterization of these zones have been done based on seismicity, tectonic charac-
teristics, and fault rupture mechanism considering the similar approach suggested by Kayal
(1998). The seismotectonic map was prepared after superimposing all the declustered
events (until 2010) on the digitized tectonic map published by GSI. The events were
segregated and separate subcatalogs were created for each source zone (IBR, EH, SP, BB,
NT, MT). The earthquake magnitude frequency distribution plots for the zones are shown
in Fig. 3. The seismicity parameters ‘a’ and ‘b’ were calculated for each of these zones
separately using G–R relationship and these values have been reported in Table 1. IBR and
EH are found to have comparatively higher a, b values, which indicates a higher level of
seismicity (a values) with a greater proportion of smaller size events (b values).

2.1 Eastern Himalayas (EH)

The seismicity of this zone is due to collision tectonics between the Indian plate and the
Eurasian plate (Verma and Mukhopadhyay 1977; Khattri and Tyagi 1993). Seismic events

123
Nat Hazards (2013) 68:1089–1108 1093

No of Earthquake per year

Mw

Fig. 3 Magnitude-frequency plots for the identified seismic source zones

Table 1 Seismicity parameters calculated for each zone

Sl. No. Seismic regions/zones Parameter- Parameter- R2 (coefficient of
(a) value (b) value correlation)

1 Indo-Burma range (IBR) 4.94 0.79 0.96

2 Eastern Himalaya (EH) 4.88 0.86 0.92
3 Shillong Plateau (SP) 3.28 0.61 0.98
4 Bengal Basin (BB) 3.03 0.61 0.93
5 Naga Thrust (NT) 2.57 0.54 0.94
6 Mishmi Thrust (MT) 2.54 0.54 0.90

are mostly located in between main boundary thrust (MBT) and the main central thrust
(MCT). The MCT is mainly a ductile shear zone (Kayal 2001). In the entire NE Himalayan
belt a transverse tectonic regime is observed (Kayal 2001). Depth of focus in this zone
ranges from 0 to 40 km with few having depth range 41–70 km. Two earthquakes of
magnitude Mw 7.1 in 1941 and another Mw 7.8 in 1947 have so far been reported. However,
recently Mw 6.8 on September 17, 2011 has occurred and caused a devastating effect in
Sikkim state.

2.2 Indo-Burma ranges (IBR)

The Burmese arc is convex in NE–SW direction having approximately 1,100 km long plate
boundary. The IBR ranges are Arakan-Yoma, Chin and Naga hills which pass through
northeast direction to join the Himalayas. In the southern direction, the ranges continue up
to Andaman-Nicobar islands and having a width around 150 km. It is well known that this
zone is subduction (inslab) earthquake zone characterized by high seismicity (Satyabala
2003). In this area, about 10 large earthquakes (M [ 7) have occurred during last

123
1094 Nat Hazards (2013) 68:1089–1108

100 years. The depth of focus goes up to 200 km in the south and in the north side, the
depth becomes lesser.

2.3 Shillong plateau (SP)

The activity in this zone is considered as the plate boundary activity. The vicinity of
‘SP’ is bounded by Brahmaputra river fault in the north, Dauki fault in the south which
is oriented in E–W direction. Dhubri fault in the west, oriented in N–S and Disang
thrust toward the east side. Dauki fault behaves as normal faults. The 3rd highest
seismic activity was observed in this zone after IBR and EH. This region has experi-
enced a great earthquake of 1897 (M [ 8.7) and three other major earthquakes of
M [ 7 occurred in this zone. The focal depth varies well within 60 km. The structural
discontinuities are Dudhnai, Kulsi, Samin, Dauki and Dhubri and also Barapani thrusts
(Kharshing et al. 1986).

2.4 Bengal basin (BB)

This zone is characterized by low seismicity. However, Tripura folded belt shows mod-
erate activity. It is prudent to highlight that this zone contains intraplate as well as plate
boundary activity. The Srimangal earthquake occurred in 1918 (Mw = 7.6) along the
Sylhet fault while Cacher earthquake of 1984 (M = 5.6) occurred within the Tripura fold
belt. The main tectonic domain is Dauki fault, which separates the boundary between the
SP and BB in E–W direction. Islam et al. (2011) suggested that tectonic features of BB are
the Eocene Hinge zone. Curray et al. (1982) suggested that this hinge zone marks the
boundary between continental crust and the young oceanic crust that extends toward the
Bay of Bengal.

2.5 Naga Thrust (NT)

The seismic activity in this zone is much lesser than SP. In this zone, since 1548 no
medium or large earthquakes were reported (Bapat et al. 1983). A survey performed on
micro earthquakes concluded that this zone is aseismic. Kayal (1987) and Khattri and
Weiss (1978) named it as an Assam gap or aseismic corridor. This gap extends parallel to
the Naga Thrust in the east and ends at the Dauki fault to the south.

2.6 Mishmi Thrust (MT)

This zone is located between the Himalayan arc to the north and Burmese arc to the east. In
this zone, the seismicity is higher compared to Eastern Himalaya due to the collision of
Indian plate and the Eurasian plate (Molnar and Tapponnier 1977; Verma and Mukho-
padhyay 1977). This zone is considered to have high stress concentration and accepted as a
special zone with block tectonics (Gansser 1974). The Great Assam earthquake of 1950
(M = 8.7) occurred in this zone which might be due to strike slip movement along the Po
Chu fault. Further, the Himalayan arc takes a 90° sharp bend and meets the IBR in this
region.

123
Nat Hazards (2013) 68:1089–1108 1095

3 Data collection

3.1 Earthquake catalog

The earthquake catalogs have been compiled for the study area covering from 18.40° to
29°N latitude and 86.20°–97.30°E longitude having 500 km distance from the political
boundary of Tripura and Mizoram. The event details were collected from various national
and international seismological agencies like, Indian Meteorological Department (IMD),
Geological Survey of India (GSI), United States Geological Survey (USGS) and Interna-
tional Seismological Center (ISC), etc. Earthquake magnitude scale is one of the most
fundamental parameters to quantify the earthquake size. For quantification, a uniform
magnitude scale is desirable, but practically it is not possible to maintain single scale due to
the type of instrumentation and its recording, station distribution, magnitude formulas, etc.
(Kanamori 1977, 1983). As a result, various magnitude scales such as ML, mb and MS are
being used. The problem comes when they do not record uniformly for all magnitude
ranges because of their saturation nature to detect higher magnitudes. This limitation
brings the underestimation or overestimation of magnitudes (Scordilis 2006). To overcome
this problem, moment magnitude (MW) scale has been selected, which has no saturation
effect for higher magnitude and it is based on seismic moment. Since the moment mag-
nitude characterizes more accurately the earthquake size, it is considered as a uniform
magnitude scale for seismic hazard evaluation. For PSHA, a uniform scale is necessary to
have an accurate assessment of the hazard. In this study, all the events were converted into
moment magnitude scale (MW) using the relation given by Kolathayar et al. (2011)
developed for the Indian subcontinent. The equations are given below:
MW ¼ 1:08ð0:0152Þmb  0:325ð0:081Þ mb  4  7:2 R2 ¼ 0:732 ð1Þ
MW ¼ 0:815ð0:04ÞML þ 0:767ð0:174Þ 3:3  ML  7 R2 ¼ 0:884 ð2Þ
MW ¼ 0:693ð0:006ÞMS þ 1:922ð0:035Þ 3:7  MS  8:8 R2 ¼ 0:90 ð3Þ
Before seismic hazard analysis, earthquake catalog was declustered using the meth-
odology proposed by Gardner and Knopoff (1974) and modified by Uhrhammer (1986).
Recent study by Stiphout et al. (2010) states that the seismicity background derived from
the methods of Gardner and Knopoff (1974) follows a Poisson distribution. The space and
time window used to remove foreshocks and aftershocks is given in below equations. This
algorithm assumes that time and spatial distribution of foreshocks and aftershocks is
dependent on the magnitude of the main event.
Distance = e1:024þ0:804Mw Time = e2:87þ1:235Mw ð4Þ
After declustering, 3,246 events (main shocks) were identified in the study area (within
500 km from political boundary) among which, 2,422 events were of MW greater than 4.
Hence, in total 3,246, declustered seismic events recorded in the study area since
1731–2010 (280 years) were used for further analysis. The data set contains 825 events
which were less than MW 4, 1,278 events from 4 to 4.9, 995 events from 5 to 5.9, 128
events from 6 to 6.9, 15 events from 7 to 7.9, and 5 events of MW greater than 8 which is
shown in Table 2. The earthquake frequency of occurrence in each decade is shown in
Fig. 4. Then, statistical analyses were carried out for data completeness using Stepp’s
method (Steep 1972) as shown in Fig. 5. From the figure, it is clear that the magnitude
range 4–5 was complete for 39 years, 5–6 was complete for 49 years, 6–7 was complete

123
1096 Nat Hazards (2013) 68:1089–1108

for 89 years, 7–8 was complete for 129 years and greater than 8 was complete for
199 years.

3.2 Seismic sources

Identification of seismic sources is a prerequisite for the seismic hazard analysis. The linear
seismic sources were identified from the seismotectonic atlas (SEISAT 2000), published by
the Geological Survey of India, from the available literature and also from remote sensing
images. The SEISAT (2000) contains 43 maps presented in 42 sheets covering the entire
India and adjacent countries with 1:1 million scale. The sheets of SEISAT (2000) repre-
senting the features of the study area were scanned, digitized and georeferenced using
MapInfo 10.0 version. After this, tectonic features and seismic events were superimposed
on the digitized map of the seismic sources to prepare the seismotectonic map as shown in

Table 2 Number of earthquakes reported in each decade of the study area

From To Mw \ 4 4 \ Mw \ 5 5 \ Mw \ 6 6 \ Mw \ 7 7 \ Mw \ 8 Mw [ 8 Total

1731 1740 – – – – 1 – 1
1741 1750 – – – – – – 0
1751 1760 – – – – – – 0
1761 1770 – 1 – 2 – – 3
1771 1780 – – – – – – 0
1781 1790 – – – – – – 0
1791 1800 – – – – – – 0
1801 1810 1 2 – – – – 3
1811 1820 – – 2 – – – 2
1821 1830 – 6 4 1 – – 11
1831 1840 – – 1 1 1 1 4
1841 1850 2 3 8 1 1 – 15
1851 1860 1 1 2 2 – – 6
1861 1870 – 2 5 2 1 – 10
1871 1880 – – 1 – – – 1
1881 1890 – 1 1 1 – – 3
1891 1900 – 1 – – – 1 2
1901 1910 – – 2 – 1 1 4
1911 1920 – – 1 1 1 – 3
1921 1930 – – 9 14 2 – 25
1931 1940 – – 7 17 2 1 27
1941 1950 – – 12 7 1 1 21
1951 1960 – – 17 8 1 – 26
1961 1970 – 23 93 20 – – 136
1971 1980 7 129 146 8 – – 290
1981 1990 40 268 293 16 2 – 619
1991 2000 321 581 287 22 1 – 1,212
2001 2010 453 260 104 5 – – 822
Total 825 1,278 995 128 15 5 3,246

123
Nat Hazards (2013) 68:1089–1108 1097

Fig. 4 Histogram of earthquake data in the study area

Completeness Ranges

(1971-2010)=39years
1 (1961-2010)=49years
4<Mw<5
Sigma(Standard Deviation)

5<Mw<6
6<Mw<7
7<Mw<8
-1/2 Mw>8
Slope-T
0.1

(1921-2010)=89years
(1881-2010)=129years
0.01
(1811-2010)=199years

10 100
Time (Years)

Fig. 5 Plot showing data completeness analysis (Steep 1972)

Fig. 2. The maximum magnitude (Mmax) has been estimated for each fault after super-
imposing all the events on the fault map. The evaluation of seismicity parameters were
done based on the data from an entire source zone. For differentiating the fault level
activity rate, it is necessary to find the fault level recurrence rate. This was estimated by
considering the weighting factors for length and number of events associated with a fault
(relative to the total length of all faults in the zone and total number of earthquakes in the
zone, respectively).

123
1098 Nat Hazards (2013) 68:1089–1108

4 Selection of ground motion models

Different tectonic regimes give rise to different ground motion models. The ground motion
models are described by the probability density function (PDF) of the ground motion given
by the properties of the earthquake source to given magnitude, type-of-faulting, distance
and the site response. The ground motion is assumed to follow a lognormal distribution and
the ground motion equation gives the median ground motion and the standard deviation.
Three main categories of tectonic regimes are typically used for ground motion estimation
for seismic hazard assessments: shallow crustal earthquakes in active tectonic regions,
shallow crustal earthquakes in stable continental regions and subduction zone earthquakes.
Crustal interplate earthquakes are mostly observed near plate margins where the crust
stiffness is lesser; temperature is relatively high and the resulting continuous thrusting
cause frequent earthquakes. Intraplate earthquakes are generally observed within the plate
where crust stiffness is higher with higher stress drop and low temperature. The earth-
quakes generated in this portion cause higher ground motions compared to crustal inter-
plate earthquakes. The subduction earthquakes produce lower ground motion compared to
crustal earthquakes with low attenuation.
There is scarcity of strong motion data and region specific attenuation relations in India.
Some of the important ground motion equations available for India include one developed
for the Himalayan region (Sharma 1998), for the Delhi region (Iyengar and Ghosh 2004);
for Peninsular India (Raghu Kanth and Iyengar 2007); for Sikkim Himalaya (Nath et al.
2005); for the Himalayan Region (Sharma et al. 2009); for the North East India (Das et al.
2006) and for Indo-Myanmar subduction zone (Gupta 2010). The strong motion data have
been collected from ‘‘COSMOS virtual data center’’ (http://db.cosmos-eq.org/) to validate
the selected attenuation relationship for this region. Based on seven earthquakes during
1987 to1997, the data recorded in different stations (see Table 3; Fig. 6) using digital
network (SMART-1) for Indo-Burma Range (IBR) and North East India (NEI) have been
used to validate the prediction equations. Only rock level recorded PGAs have been used as
shown in Fig. 7.
The PGA versus distance plots (for Magnitude 5.9 earthquake) obtained using the
GMPES given by Sharma (1998), Atkinson and Boore (2003), Raghu Kanth and Iyengar
(2007), Sharma et al. (2009) and Gupta (2010) are shown in Fig. 7. It can be seen that
Sharma (1998) and Sharma et al. (2009) give low PGA value at short distance and
attenuate very less even at larger distance. Raghu Kanth and Iyengar (2007) give lower
PGA value compared to observed PGA both at smaller and large distances. It is observed
that the attenuation model (Gupta 2010) overestimates the PGA at shorter distances but
underestimates beyond 300 km distance. Atkinson and Boore’s (2003) inslab attenuation
relation predict median values with the observed PGAs at short to medium distances.
Iyengar et al. (2010) developed a model for North East India and this has also been used for
validation. This model also underestimates the PGA for all distances. Therefore, for hazard
analysis the attenuation relations of Gupta (2010) and Atkinson and Boore (2003) have
been used. The prediction equation is given below and their coefficients are shown in
Tables 4 and 5 up to the bedrock level.
Log Y¼C1 þ C2 Mw þC3 hþC4 R  g log R ð5Þ
where Y represents mean PGA at bedrock level in cm/sec2, Mw is the moment magnitude,
h is the focal depth in kilometer, R is the hypocentral distance in kilometer and g is the
geometric attenuation parameter. It is observed that Atkinson and Boore (2003) and Gupta

123
Nat Hazards (2013) 68:1089–1108 1099

Table 3 Some strong motion recording stations located at rock sites in North East India that were used for
validation of ground motion models (http//db.cosmos-eq.org)
Earthquake Date Stations Hypo central Mw Recorded
distance (km) PGAmax
(g)

1 IBR 10/9/1986 Ummulong 44.9 5.0 0.113

Latitude 25.385 N Pynursla 48.2 5.0 0.093
Longitude 92.077E Khliehriat 52.5 5.0 0.046
Depth (km) 43 Umsning 58.3 5.0 0.101
Nongkhlaw 68.1 5.0 0.093
Nongpoh 71.4 5.0 0.056
Nongstoin 92.4 5.0 0.019
2 IBB 6/2/1988 Pynursla 84.6 5.9 0.050
Latitude 24.688 N Mawphlang 94.7 5.9 0.081
Longitude 91.57E Shillong 110.2 5.9 0.048
Depth (km) 15 Ummulong 117.2 5.9 0.056
Nongkhlaw 117.3 5.9 0.114
Khliehriat 117.5 5.9 0.080
Umsning, 127.7 5.9 0.061
Nongpoh, 145.5 5.9 0.086
3 IBR 9/1/1990–97 Ummulong 342.9 6.0 0.051
Latitude 24.753 N Pynursla 67.1 6.0 0.042
Longitude 95.241E Ummulong, 78.4 6.0 0.155
Depth (km) 119 Shillong 90.1 6.0 0.072
Umsning 106.8 6.0 0.077
Nongpoh, 124.1 6.0 0.072
Nongstoin 125 6.0 0.048
Khliehriat 323.2 6.0 0.025
Site conditions all are rock sites

(2010) underestimate the PGA at larger distance, but give reasonable value that match with
observed PGA for short to medium distance. Hence, those GMPEs were selected for the
present study. In addition, both of these GMPEs are region specific relations developed for
regions with subduction earthquakes, which are prominent in the study area. The relation
by Gupta (2010) was developed specifically for Indo-Myanmar subduction zone of North
East India and well-accepted ground motion model (Atkinson and Boore 2003) was
developed for subduction zone earthquakes (Cascadia). It is shown in Fig. 7 that for how
far distance those equations (Gupta 2010; Atkinson and Boore 2003) are valid or efficient
to predict the future ground motion correctly. Keeping this limitation in view, it is rec-
ommended to develop new GMPEs for the region for large distances. The lack of strong
motion data at different distances makes the attenuation models inefficient to predict the
future ground motion especially at large distance. As more strong motion records become
available, attempts should be made by researchers to develop site specific ground motion
prediction equation for the states of Tripura and Mizoram, suitable for large distance as
well.

123
1100 Nat Hazards (2013) 68:1089–1108

Fig. 6 The distribution of data recording stations in North East India (source: http//db.cosmos-eq.org/)

5 Hazard analysis

Gutenberg and Richter’s (1944) parameters a and b were estimated for each zone which
has a negative correlation between earthquake magnitude and rate of occurrences. The
relationship assumes an exponential distribution of magnitude and is expressed as:
Log N¼a  bM ð6Þ
where a represents the level of seismicity of an area for a particular period of interest and
b is the slope of the regression equation that describes the relative size distribution of
events. The higher a value indicates the higher level of seismicity. The higher b value
means smaller size events are associated with larger proportion. This implies a condi-
tioning expression for a given magnitude size and the probable mean annual rate of
occurrences. The seismicity parameters for each source zone were estimated using G–R
relationship. For differentiating the fault level activity rate, it is necessary to find the fault

123
Nat Hazards (2013) 68:1089–1108 1101

100
Observed PGA
10
Gupta(2010) (inslab)
1
Atkinson and Boore(2003)
(inslab)
0.1
PGA (g)

Atkinson and Boore(2003)

0.01 (Interface))
Sharma(1998)
0.001
Sharma et al. (2009)
0.0001
Raghukant and Iyengar
0.00001 (2007)
0 100 200 300 400 500 600
Distance (km)

Fig. 7 Validation of ground motion models using available strong motion data (Mw 5.9) for subduction
earthquakes in North East India

Table 4 Attenuation coefficients given by Gupta (2010) for North East India subduction zone
Period (s) C1 C2 C3 C4 r (standard deviation, cm/s2)

0.02 0.4598 0.6909 0.01130 -0.00202 0.347

0.04 0.7382 0.6327 0.01275 -0.00234 0.343
0.10 1.0081 0.6667 0.01080 -0.00219 0.341
0.20 1.2227 0.6918 0.00572 -0.00192 0.340
0.40 0.8798 0.7727 0.00173 -0.00178 0.341
1.00 -0.3339 0.8789 0.00130 -0.00173 0.344
2.00 -2.0677 0.9964 0.00364 -0.00118 0.347
3.00 -3.4227 1.1169 0.00615 -0.00045 0.351

Table 5 Attenuation coefficients given by Atkinson and Boore (2003) for subduction zone
Period (sec) C1 C2 C3 C4 r (standard deviation, cm/s2)

0.02 -0.04710 0.69090 0.01130 -0.0020 0.27

0.04 0.50697 0.63273 0.01275 -0.0023 0.25
0.10 0.43928 0.66675 0.01080 -0.0022 0.28
0.20 0.51589 0.69186 0.00572 -0.0019 0.28
0.40 0.00545 0.77270 0.00173 -0.0018 0.28
1.00 -1.0213 0.87890 0.00130 -0.0017 0.29
2.00 -2.3923 0.99640 0.00364 -0.0012 0.30
3.00 -3.7001 1.11690 0.00615 -0.0005 0.30

level recurrence rate. This was estimated by considering the weighting factors for length
and number of events associated with a fault (relative to the total length of all faults in the
zone and total number of earthquakes in the zone, respectively).

123
1102 Nat Hazards (2013) 68:1089–1108

The seismic hazard evaluation of the study area based on a state-of-the-art PSHA study
has been performed using the classical Cornell–McGuire approach (Cornell 1968). The
hazard curves obtained from PSHA show the variation of peak horizontal acceleration (PHA)
or spectral acceleration (Sa) against the mean annual rate of exceedance (Kramer 1996). For
calculating the seismic hazard values, the entire study area was divided into grids of size
0.05° 9 0.05° (approximately 5 km 9 5 km) and the hazard values at the center of each grid
cell were calculated by considering all the seismic sources within a radius of 500 km. The
occurrence of an earthquake in a seismic source is assumed to follow a Poisson’s distribution.
The probability of ground motion parameter, Z, at a given site, will exceed a specified level, z,
during a specified time; T is represented by the expression:
PðZ [ zÞ ¼ 1  emðzÞT  mðzÞT ð7Þ
where m(z) is the mean annual rate of exceedance of ground motion parameter, Z, with respect
to z. The function m(z) incorporates the uncertainty in time, size and location of future
earthquakes and uncertainty in the ground motion they produce at the site. It is given by:
2 1 3
XN Z
mmax Z
mðzÞ ¼ Nn ðm0 Þ fn ðmÞ4 fn ðrjmÞPðZ [ zjm; rÞdr5dm ð8Þ
n¼1
m¼m0 r¼0

where Nn (m0) is the frequency of earthquakes on a seismic source n, having a magnitude

equal to or greater than a minimum magnitude m0 (in this study, it is taken as 4.0); fn(m) is
the probability density function for a minimum magnitude of m0 and a maximum mag-
nitude of mmax; fn(r|m) is the conditional probability density function (probability of
occurrence of an earthquake of magnitude m at a distance r from the site for a seismic
source n); P(Z [ z|m, r) is the probability at which the ground motion parameter Z exceeds
a predefined value of z, when an earthquake of magnitude m occurring at a distance of r
from the site. The integral in Eq. 8 can be replaced by summation and the density functions
fn(m) and fn(r|m) can be replaced by discrete mass functions. The resulting expression for
m(z) is given by:
"r ¼r #
N miX
X ¼mmax jX max

mðzÞ ¼ kn ðmi Þ Pn ðR ¼ rj jmi ÞPðZ [ zjmi ; rj Þ ð9Þ

n¼1 mi ¼m0 rj ¼rmin

where kn(mi) is the frequency of occurrence of magnitude mi at the source n obtained by

discretizing the earthquake recurrence relationship for the source n.
Many uncertainties are associated with the ground motion models and in order to get a
rational value of hazard, hazard analysis with the logic tree approach was employed in this
work assigning equal weightages to both the GMPEs. The peak ground accelerations
(PGA) at the rock level were evaluated for the entire area. The contour maps showing the
spatial variation of PGA (2 and 10 %) values were evaluated. The uniform hazard spec-
trum for the same return periods was developed for Agartala (capital city of Tripura) and
Aizawl (capital city of Mizoram state) cities.

6 Results and discussion

The primary output from PSHA is a hazard curve showing the peak ground acceleration
against the mean annual rate of exceedance. Seismicity parameters a and b were estimated

123
Nat Hazards (2013) 68:1089–1108 1103

Fig. 8 Seismic Hazard Curve at rock level for a Agartala and b Aizawl City

for each source zone as shown in Table 1. The hazard curves for all six major sources were
prepared and compiled to get the total hazard curve which represents the cumulative hazard
of all sources. Figure 8a, b shows the hazard curves for the Agartala and Aizawl city at
bedrock level. These results highlight that the ‘IBR’ is the most active zone followed by
‘SP,’ and ‘BB’ to produce more annual frequency of occurrence and higher ground
acceleration due to earthquakes compared to others zones within the study area. The
corresponding uniform hazard spectra (UHS) curve for Agartala and Aizawl cities is shown
in Fig. 9a, b. PGA at each grid point has been estimated to define seismic hazard at rock
level. The PGA distribution maps for 2 and 10 % probability of exceedance in 50 years are
shown in Fig. 10a, b. The PGA value for Tripura varies from 0.06 to 0.5 g (2 % probability
exceedance) and 0.03–0.26 g (10 % probability exceedance). Similarly, the PGA value for

123
1104 Nat Hazards (2013) 68:1089–1108

(a)

Spectral Acceleration(g)

Period(sec)
(b)

Fig. 9 Uniform hazard spectrum at rock level for 2,475 and 475 years of return periods of a Agartala and
b Aizawl City

Mizoram varies from 0.11 to 0.4 g (2 % probability of exceedance) and 0.08–0.23 g (10 %
probability of exceedance). The highest PGA values were observed in the northern district
of Tripura for both the return periods which can go beyond 0.46 and 0.26 g. The developed
UHS for Agartala city shows that for 475 years return period, the zero period spectral
acceleration is 0.14 g and similarly 0.20 g for 2,475 years. For Aizawl city, zero period
spectral acceleration for 2,475 and 475 years return period shows 0.17 and 0.1 g,
respectively. The comparison of present values is shown in Table 6 as reported by other
researchers in this region, along with the hazard specified by BIS 1893 (2002). The present
PHA value of Agartala city matches well with that of Das et al. (2006) and Iyengar et al.
(2010), but shows a large deviation from the value reported by Sharma and Malik (2006).
The present PGA value of Aizawl city shows deviation from all the other studies. The
reason for this deviation may be due to use of different seismogenic source zones and
different attenuation relationships in these works. Das et al. (2006) and Iyengar et al.
(2010) used attenuation relations developed by them to estimate the hazard. Sharma and
Malik (2006) employed the relation suggested by Youngs et al. (1997). We have employed
two well-recognized and latest relations in a logic tree framework, after validating the
same with actual strong motion data from the study area. The zone factor specified by BIS

123
Nat Hazards (2013) 68:1089–1108 1105

Fig. 10 PGA (g) contours at bed rock level for a 2 and b 10 % probability of exceedance (PE) in 50 years

123
1106 Nat Hazards (2013) 68:1089–1108

Table 6 Comparison of present PGA value with previous studies

Location Present study Das et al. Sharma and Iyengar et al. (2010)
(using PSHA) (2006) Malik (2006)

Agartala (capital of Tripura) 0.11–0.20 g 0.18–0.22 g 0.33 g 0.12–0.18 g

Aizawl (capital of Mizoram) 0.1–0.17 g 0.22–0.28 g 0.3–0.4 g 0.22–0.35 g

1893 (2002) is uniform for the entire North East India (as entire NE India falls in Seismic
zone V). The present hazard value shows spatial variation even within small areas of
Tripura and Mizoram, as it takes into account the effect of sources and events at each grid
point separately.

7 Conclusions

In this paper, the results of probabilistic seismic hazard analysis (PSHA) of Tripura and
Mizoram states in North East India have been presented. Analysis has been carried out
using the data collected over a distance of 500 km from the political boundary of the state.
The hazard curve has been developed for Agartala and Aizawl cities. Further, UHS has
been generated at the rock level considering 2 and 10 % probability of exceedance in
50 years. In Fig. 3, their frequency distribution shows that the IBR and SP zone has greater
potential to have greater magnitude earthquakes in the future. EH and MT has a chance to
produce magnitudes greater than Mw 7.5. Northern part of Tripura has been found to have
high hazard compared to other parts. This part is very close to Sylhet (Bangladesh) fault
having two major earthquakes Mw [ 7 in the past history. Lowest hazard has been eval-
uated for the southern part of the Tripura (0.03–0.06 g). Both capital cities show moderate
hazard values at bedrock level. Attenuation relations were validated with recorded PGA
values before they were used in the hazard analysis. The attenuation equations of Gupta
(2010) and Atkinson and Boore (2003) matches well, but both the models underestimates
at large distances. However, it is necessary to modify/develop an attenuation model with
large distances for [300 km for this region, to define the hazard in a better way.

Acknowledgments The authors would like to acknowledge IMD, USGS, ISC, BARC and NGRI for
providing the details of earthquake events in Tripura, Mizoram and adjoining areas. Comments and valuable
suggestions by Thomas Glade, Editor-in-Chief (NHAZ journal) for improving the quality of paper are
greatly appreciated. The authors acknowledge the anonymous reviewers for going through the manuscript
thoroughly and giving their valuable suggestions to enhance the quality of the paper.

References

Atkinson GM, Boore DM (2003) Empirical ground-motion relations for subduction-zone earthquakes and
their application to Cascadia and other regions. Bull Seismol Soc Am 93:1703–1729
Bapat A, Kulkarni RC, Guha SK (1983) Catalogue of earthquakes in India and neighborhood–from his-
torical period up to 1979. Indian Society of Earthquake Technology, Roorkee
BIS-1893 (2002) Indian standard criteria for earthquake resistant design of structures, part 1—general
provisions and buildings. Bureau of Indian Standards, New Delhi
Cornell CA (1968) Engineering seismic risk analysis. Bull Seismol Soc Am 58:1583–1606
Curray JR, Emmel FJ, Moore DG, Raitt RW (1982) Structure tectonics and geological history of the
northeastern Indian Ocean. In: Nairn AEM, Stehli FG (eds), The ocean basins and margins, The Indian
Ocean, vol 6. Plenum, New York, pp 399–450

123
Nat Hazards (2013) 68:1089–1108 1107

Das S, Gupta ID, Gupta VK (2006) A probabilistic seismic hazard analysis of Northeast India. Earthq
Spectra 22(1):1–27
Gansser A (1974) The ophiolitic melange, a world-wide problem on Tethyan examples. Eclogae Geol Helv
67:479–507
Gardner JK, Knopoff L (1974) Is the sequence of earthquakes in Southern California with aftershocks
removed, Poissonian. Bull Seismol Soc Am 64(5):1363–1367
Gupta ID (2010) Response spectral attenuation relations for in-slab earthquakes in Indo-Burmese subduction
zone. Soil Dyn Earthq Eng 30:368–377
Gutenberg B, Richter CF (1944) Frequency of earthquakes in California. Bull Seismol Soc Am 34:185–188
Islam MS, Shinjo R, Kayal J (2011) The tectonic stress field and deformation pattern of northeast India,
the Bengal basin and the Indo-Burma Ranges: a numerical approach. J Asian Earth Sci 40(1):
121–131
Iyengar RN (2010) Development of probabilistic seismic hazard map of India (A technical report of the
working committee of experts (WCE) constituted by the National disaster management authority govt.
of India, New Delhi
Iyengar RN, Ghosh S (2004) Microzonation of earthquake hazard in greater Delhi area. Curr Sci
87:1193–1202
Iyengar RN, Chadha RK, Balaji Rao K, Raghukanth STG (2010) Development of probabilistic seismic
hazard map of India. A technical report of the working committee of experts (WCE) constituted by the
National disaster management authority govt. of India, New Delhi
Kanamori H (1977) The energy release in great earthquakes. J Geophys Res 82:2981–2987
Kanamori H (1983) Magnitude scale and quantification of earthquakes. In: Duda SJ, Aki K (eds), Quan-
tification of earthquakes. Tectonophysics, 93:185–199
Kayal JR (1987) Microseismicity and source mechanism study: Shillong Plateau, Northeast India. Bull
Seismol Soc Am 77:184–194
Kayal JR (1998) Seismicity of Northeast India and surroundings–development over the past 100 years.
J Geophys 19(1):9–34
Kayal JR (2001) Microearthquake activity in some parts of the Himalaya and the tectonic model. Tec-
tonophysics 339:331–351
Kayal JR (2008) Microearthquake seismology and seismotectonics of south Asia, 1st edn. Springer Verlag
and Capital Publishing Company, India, ISBN: 978–1–4020–8179–8
Kharshing AD, Khattri KN, Moharir PS, Chander R (1986). Neotectonics in Shillong massif and neigh-
boring regions. In: Proceedings of the international symposium neotectonics in South Asia, India. Feb.
18–21, 407–416
Khattri KN, Tyagi AK (1993) Seismicity patterns in the Himalayan plate boundary and identification of the
areas of high seismic potential. Tectonophysics 96:281–297
Khattri KN, Weiss M (1978) Precursory variation of seismic rate in Assam Area, India. Geology 6:685–688
Kolathayar S, Sitharam TG, Vipin KS (2011) Spatial variation of seismicity parameters across India and
adjoining areas, Nat Hazards doi:10.1007/s11069-011-9898-1
Kramer SL (1996) Geotechnical earthquake engineering; Prentice-Hall international series in civil engi-
neering and engineering mechanics. Prentice-Hall, New Jersey
Kumar N, Parvez IA, Virk HS (2005) Estimation of coda waves attenuation for NW Himalayan region using
local earthquakes. Phys Earth Planet Interiors 3:243–258
Kumar P, Yuan X, Ravi Kumar M, Kind R, Li X, Chadha RK (2007) The rapid drift of Indian tectonic plate.
Nature 449:894–897
Mitra S, Priestley K, Bhattacharyya AK, Gaur VK (2005) Crustal structure and earthquake focal depths
beneath northeastern India and south Tibet. Geophys J Int 160:227–248
Molnar P, Tapponnier P (1977) Relation of the tectonics of Eastern China to the India-Eurasia collision:
application of slip-line field theory to large-scale control tectonics. Geology 5:212–216
Mukhopadhyay M (1992) On earthquake focal mechanism studies for the Burmese arc. Curr Sci 1:72–88
Mukhopadhyay M, Dasgupta S (1988) Deep structure and tectonics of the Burmese arc: constraints from
earthquake and gravity data. Tectonophysics 149:299–322
Nandy DR (2001) Geodynamics of Northeastern India and the adjoining region. ABC Publications, Calcutta,
p 209
Nath SK, Vyas M, Pal I, Sengupta P (2005) A hazard scenario in the Sikkim Himalaya from seismotectonics
spectral amplification source parameterization and spectral attenuation laws using strong motion
seismometry. J Geophys Res 110:B01301. doi:10.1029/2004/2004JB003199
Nath SK, Thingbaijam KKS, Raj A (2008) Earthquake hazard in the Northeast India. A seismic microzo-
nation approach with typical case studies from Sikkim Himalaya and Guwahati city. J Earth Syst Sci
117(S2):809–831

123
1108 Nat Hazards (2013) 68:1089–1108

Nath SK, Raj A, Thingbaijam KKS, Kumar A (2009) Ground motion synthesis and seismic scenario in
Guwahati city-a stochastic approach. Seismol Res Lett 80:233–242
Oldham RD (1882). A Catalogue of Indian earthquakes from the earliest time to the end of A.D. 1869, by
the late Thomas Oldham R. D. Oldham (Editor), Geol Soc India Memoir 163–215
Oldham RD (1899) Report on the great earthquake of 12 June 1897. Geol Soc India Memoir 29:379
Pallav K (2010) Seismic microzonation of Imphal City and probabilistic seismic hazard assessment of
Manipur state; A PhD thesis of IIT Guwahati, Assam India
Parvez IA, Ram A (1997) Probabilistic assessment of earthquake hazards in the North-East Indian Peninsula
and Hindukush regions. Pure Appl Geophys 149(4):731–746
Raghu Kanth STG, Iyengar RN (2007) Estimation of seismic spectral acceleration in Peninsular India.
J Earth Syst Sci 116(3):199–214
Raghukanth STG, Dash SK (2010) Deterministic seismic scenarios for northeast India. J Seismol
14:143–167
Satyabala SP (2003) Oblique plate convergence in the Indo-Burma (Myanmar) subduction region. Pure Appl
Geophys 160:1611–1650
Scordilis EM (2006) Empirical global relations converting Ms and mb to moment magnitude. J Seismol
10:225–236
SEISAT (2000) Seismotectonic atlas of India. Geological Survey of India, New Delhi
Sharma ML (1998) Attenuation relationship for estimation of peak ground acceleration using data from
strong-motion arrays in India. Bull Seismol Soc Am 88:1063–1069
Sharma ML, Malik SHIPRA (2006) Probabilistic seismic hazard analysis and estimation of spectral strong
ground motion on bed rock in North East India, 4th international conference on earthquake engi-
neering, Taipei, Taiwan, pp 12–13
Sharma ML, Douglas J, Bungum H, Kotadia J (2009) Ground-motion prediction equations based on data
from the Himalayan and Zagros Regions. J Earthq Eng 13:1191–1210
Steep JC (1972) Analysis of completeness of the earthquake sample in the Puget sound area and its effects
on statistical estimates of earthquake hazard. Proc Inter Conf Microzonat for safer construct Res Appl
64(4):1189–1207
Stiphout VT, Zhuang J, Marsan D (2010) Seismicity declustering, community online resource for statistical
seismicity analysis. doi:10.3929/ethz
Tandon AN, Srivastava HN (1975) Focal mechanism of some recent Himalayan earthquakes and regional
plate tectonics, Bull Seismol Soc Am 65:963–970
Tiwari RP (2000) Earthquake hazards and mitigation in India with special reference to Northeastern India.
ENVIS Bull 8(2):15–22
Uhrhammer RA (1986) Characteristic of northern and central California seismicity abstract. Earthq Notes
1:21
Verma RK, Mukhopadhyay M (1977) An analysis of the gravity field in Northeast India. Tectonophysics
42:283–317
Verma RK, Mukhopadhyay M, Ahluwalia MS (1976) Seismicity, gravity and tectonics of Northeast India
and Northern Burma. Bull Seismol Soc Am 66:1683–1694
Yadav RBS, Tripathi JN, Rastogi BK, Das MC, Chopra S (2010) Probabilistic assessment of earthquake
recurrence in Northeast India and adjoining regions. Pure Appl Geophys ISSN 0033–4553
Youngs RR, Chiou SJ, Silva WJ, Humphrey JR (1997) Strong ground motion relationships for subduction
earthquakes. Seismol Res Lett 68:58–73

123