You are on page 1of 15

Nat Hazards

DOI 10.1007/s11069-010-9618-2

ORIGINAL PAPER

Potential source zones for Himalayan earthquakes:


constraints from spatial–temporal clusters

Basab Mukhopadhyay • Anshuman Acharyya • Sujit Dasgupta

Received: 15 October 2008 / Accepted: 6 September 2010


Ó Springer Science+Business Media B.V. 2010

Abstract The Himalayan fold-thrust belt has been visited by many disastrous earth-
quakes (magnitude [ 6) time and again. This active collisional orogen bordering Indian
subcontinent in the north remains a potential seismic threat of similar magnitude in the
adjoining countries like India, Pakistan, Nepal, Bhutan and China. Though earthquake
forecasting is riddled with all conjectures and still not a proven presumption, identifying
likely source zones of such disastrous earthquakes would be an important contribution to
seismic hazard assessment. In this study, we have worked out spatio-temporal clustering of
earthquakes (Mb C 4.5; 1964–2006) in the Himalayas. ‘Point density’ spatial statistics has
helped in detecting 22 spatial seismicity clusters. Earthquake catalog is then treated with a
moving time-distance window technique (inter-event time 35 days and distance
100 ± 20 km) to bring out temporal clusters by recognizing several foreshock-main
shock-aftershock (FMA) sequences. A total of 53 such temporal sequences identified in the
process are confined within the 22 spatial clusters. Though each of these spatio-temporal
clusters deserves in-depth analysis, we short-listed only eight such clusters that are dis-
sected by active tectonic discontinuities like MBT/MCT for detail study. Spatio-temporal
clusters have been used to constrain the potential source zones. These eight well-defined
spatio-temporal clusters demonstrate recurrent moderate to large earthquakes. We assumed
that the length of these clusters are indicating the possible maximum rupture lengths and
thus empirically estimated the maximum possible magnitudes of eight clusters that can be
generated from them (from west to east) as 8.0, 8.3, 8.2, 8.3, 8.2, 8.4, 8.0 and 7.7. Based on
comparative study of the eight cluster zones contemplating with their temporal recurrences,
historical seismic records, presence of intersecting faults and estimated magnitudes, we
have guessed the possibility that Kangra, East Nepal, Garhwal and Kumaun–West Nepal
clusters, in decreasing order of earthquake threat, are potential source zones for large
earthquakes (C7.7 M) in future.

Keywords Himalayas  Earthquake  Spatio-temporal clusters 


Characteristic earthquake  Seismic quiescence  Fault interaction  Rupture

B. Mukhopadhyay (&)  A. Acharyya  S. Dasgupta


Geological Survey of India, 27, Jawaharlal Nehru Road, Kolkata 700016, India
e-mail: basabmukhopadhyay@yahoo.com

123
Nat Hazards

1 Introduction

The active deformation of the Himalayan fold-thrust belt has been witnessed by the
occurrence of repeated earthquakes of different size. Manifestation of ongoing active
tectonics in this collisional convergent margin includes disastrous earthquakes at Shillong
(1897, M 8.7), Kangra (1905, M 7.8), Assam (1950, M 8.5), West Nepal–India (1980,
M 6.5), Assam (1988, M 7.2), Bihar–Nepal (1988, M 6.6), Uttarkashi (1991, M 6.6),
Chamoli (1999, M 6.8) and Kashmir (2005, M 7.7). In addition, several large-sized and
medium-sized earthquakes have spawned along the 2,500-km-long Himalayan arc
inflicting major damage to life and property in the adjoining densely populated foothills
and in the Ganges—Brahmaputra alluvial plains.
Forecasting earthquakes in the Himalayan tectonic domain is not uncommon. Four large
earthquakes (M 8.6) are forecasted in the Himalayas (Bilham and Ambraseys 2005; Bilham
and Wallace 2005) within the inferred ‘seismic gap areas’. At least two large earthquakes
(M [ 8.0 or even larger) might also take place in the areas west of Kathmandu in Nepal
(Bettinelli et al. 2006). Large earthquakes (Mw [ 8.6 with rupture length *400 km) are
also speculated in the seismic gaps at western and central Himalayas and eastern Nepal
(Feldl and Bilham 2006). Lines of evidence such as estimation of plate convergence rate
(Bettinelli et al. 2006) and differential shortening rate (Banerjee and Burgmann 2002;
Chen et al. 2004; Bettinelli et al. 2006; Feldl and Bilham 2006; Jade et al. 2007) from GPS
observations and average slip deficit (Bilham and Ambraseys 2005; Bilham and Wallace
2005) also support the continuing active tectonism and recurrence seismogenesis in the
Himalayas.
Earthquake forecasting based on direct scientific observation is replete with conjec-
tures and has not yielded success so far. Delineation of ‘seismic gap’ between ruptured
fault segments found some success in the oceanic subduction margins, though the data
from the Himalayas yet to attest such hypothesis. Predictions are rather ineffective if not
specified in the recognizable size time–space window. Even the available statistical
models do not yield unique solutions because of both random and non-random nature of
earthquake distribution in space and time (Kagan 1997; Wang and Kuo 1998; Ebel and
Kafka 2002). Moreover, detecting specific locales of strain accumulation in the litho-
sphere is still beyond the reach of the earth scientists. Accumulation and release of strain
never follow any uniform rule with the ‘Gutenberg–Richter (G–R) relation’ and ‘Char-
acteristic Earthquake’ slip distribution model both observed in real situations. While the
G–R relation appears to be valid in many seismotectonic ‘spatial domains’, ‘individual
fault segments’ have a tendency to produce repeated large earthquakes characteristic to
the particular fault without producing transient smaller earthquakes (see Schwartz and
Coppersmith 1984).
In the present study, we have examined the earthquake distribution in the Himalayas
from the spatio-temporal perspective. The objective of this study is to define spatio-
temporal clusters of earthquakes through ‘point density’ and ‘moving time window’
analysis. Principal outcome of our finding is the disposition and characteristics of spatio-
temporal clusters, with a few corollaries as a spin-off. We have described seismotectonic
characteristics of clusters and its possible association with inferred rupture areas. Behavior
of these clusters has also been investigated symptomatic of ‘characteristic’ model. We
have derived maximum rupture lengths of individual clusters and estimated the maximum
possible size of earthquakes for each cluster. Eventually, the threat potential of possible
source zones in the Himalayas has been prognosticated.

123
Nat Hazards

2 Methodology for cluster analysis

Similar elements occurring closely together in space usually produce ‘clusters’. Earth-
quakes may show spatial, preferred distribution that simulates clusters. In a region, an
earthquake cluster may be suspected if the assemblage consists of multiple events with
magnitude greater than a threshold value occurring within a specific period of time
(McGuire 2004). The ISC catalog has been used for the period from 1964 to 2002, and the
NEIC catalog is employed for the time span 2003 to June 2006. There has been no overlap
of data between the two catalogs. The data within the catalog may have minor locational
and depth errors consistent with the teleseismic data and, to our knowledge, will not affect
substantially the analysis. The earthquakes cover a range of magnitude from Mb 3.0 to 7.0
with focal depths ranging between *5 and 100 km. The earthquakes covered a broad
extent of the Himalayas (25°–37°N, 73°–95°E) comprising tectonic domains of main
frontal thrust (MFT) in the south, main boundary thrust (MBT), main central thrust (MCT)
to beyond Indus–Tsangpo Suture (ITS) in the north (Fig. 1). The cumulative earthquake
frequency curves are constructed using the 2,345 earthquake events with Mb C 4.0. The
b-value of earthquakes with Mb C 4.0 calculated by maximum-likelihood method (Aki
1965) is 1.09. However, we have selected 1,266 events of a magnitude domain of
Mb C 4.5 for the time period 1964–2006 because the cumulative curve (Fig. 2a) over
the magnitude (Mb C 4.5) is smooth and the catalog is by and large complete above

Fig. 1 Spatio-temporal clusters outlined in black (A, B, C, D, E, F, G and H) in Lesser Himalaya. The name
of the clusters are A—Kashmir, B—Kangra, C—Garhwal, D—Kumaun–West Nepal, E—East Nepal,
F—Sikkim, G—Bomdila and H—Eastern Syntaxis. Note the interaction between Himalayan Thrust planes
(MFT-MBT-MCT) and Peninsular crosscutting faults (RF—Ropar fault, MDF—Mahendragarh–Dehradun
Fault, GBF—Great Boundary Fault, WPF—West Patna Fault, EPF—East Patna Fault, MSRMF—Munger
Saharsha Ridge Marginal Fault, MKF—Malda-Kishanganj Fault and BF—Bomdila Fault). MFT—main
frontal thrust, MBT—main boundary thrust, MCT—main central thrust, ITS—Indus-Tsangpo Suture,
Jam—Jammu, Si—Simla, Le—Leh, Dd—Dehra Dun, Nd—New Delhi, Jai—Jaipur, All—Allahabad,
Sh—Shillong

123
Nat Hazards

Fig. 2 a Frequency magnitude


relationships for 2345 events of
mb C 4.0. See the catalog is
complete above magnitude 4.5
which was the cutoff magnitude
taken for further analysis.
b Histogram showing total
number of earthquakes in the
period 1964–2006 against
clusters

magnitude 4.5 within the time frame. The events (Mb C 4.5) are plotted on a generalized
tectonic map (Dasgupta et al. 2000) of the area (Fig. 1).
A simple visual examination of the map clearly brings out significant numbers of spatial
clusters of earthquake epicenters (Fig. 1). To constrain the spatial extents of such visible
clusters, we have utilized the spatial statistical function ‘point density’. Point density is a
classical spatial statistical tool to identify areas where data points are concentrated more or
vice versa. To calculate the point density, the distance between the adjacent earthquakes is
measured and a mean distance (*20 km) is calculated. Half of the mean distance (i.e.,
10 km) is taken as the radius of the neighborhood. Point density is then calculated as the
total number of earthquake epicentral points that fall within a circular neighborhood with a
specific radius (in this case 10 km) divided by the area of the neighborhood. This process
scans each time a total area of *314 sq. Km. A factor resulting from the size of the
earthquake is also introduced for deriving the point density value, e.g., five points are
counted instead one for an earthquake of magnitude five in the selected neighborhood. This
is done to offer more weight to larger earthquakes in the calculation. The measurement is
then taken in an overlapping grid pattern where the center of the circle has been moved
across the map (both along latitude and along longitude) by a sliding distance of 10 km.
The calculated point density value is stored in a grid point at the center of the circle. The
resulting values obtained by this sliding grid process have a mean (M) 22.78 and standard
deviation (SD) 78.58. The areas with anomalous point density (value [ (m ? 1
SD) = 101.36) have been marked as zones of spatial clusters and shown as closed poly-
gons with black outline (Fig. 1). This process identifies 22 spatial clusters within the study
area. It is to be mentioned here that other than size, any other source parameter like data
pertaining to energy release or seismic moment of the earthquakes can also be used
alternately to calculate point density provided the energy release data/seismic moment for
all earthquakes is available in the selected catalog, which are indeed rare.

123
Nat Hazards

The foreshock-main shock–aftershock/main shock–aftershock sequences in the catalog


are identified by a moving time-distance window technique (after Kafka and Walcott
1998), the ‘time’ taken as 35 days whereas ‘distance’ as 100 ± 20 km. The ‘time’ refers to
‘inter-event time’ between successive earthquakes within a circular area; ‘distance’ indi-
cates the radius of the circle. By this method of moving time-distance window, 53 numbers
of foreshock-mainshock–aftershock (FMA)/mainshock–aftershock (MA)/foreshock-main-
shock (FM) temporal sequences have been identified in the catalog. Interestingly, these 53
numbers of temporal sequences are clustered within the 22 spatial clusters already defined
by point density analysis. We have attempted detailed analysis on 8 clusters (designated as
A, B, C, D, E, F, G and H, from west to east for the eight clusters viz., Kashmir, Kangra,
Garhwal, Kumaun–West Nepal, East Nepal, Sikkim, Bomdila and Eastern Syntaxis,
respectively) confined between the MFT and north of MCT in the Lesser Himalayas
(Fig. 1).

3 Characteristics of the eight cluster zones

The cluster pattern in the eight zones reveals temporal–spatial association of earthquakes
(Table 1). Between 1964 and 2006, Kashmir cluster (A) accommodates maximum number
of earthquakes (210) followed by Kumaun–West Nepal (cluster D, 94) and Bomdila
(cluster G, 82). While Eastern Syntaxis (cluster H) contains only 12 events, the other four
clusters, Kangra (B, 74), Garhwal (C, 47), East Nepal (E, 28) and Sikkim (F, 45), have a
moderate number of events (Fig. 2b). Within individual clusters, occurrence of events
through time depicts temporal variance (Fig. 3). The period 1986–1990 registered prolific
events for clusters E, F and H, whereas cluster C had an acme during 1991–2000 and B
reached its zenith during 1970–1980. Cluster D experienced a maximum number of
earthquakes during 1964–1969, while G registered a maximum number during 1981–1985.
Cluster A is absolutely different from any other clusters and earthquake incidence
abounded only during 2001–2006.
Analyzing the pattern of earthquakes within each cluster, it is apparent that the clusters
are constituted of independent events as well as FMA sequence and its combinations
(Table 1). In the clusters, earthquakes that form complete FMA sequence or any combi-
nation of FMA are placed in temporal sequence, whereas the events that do not register to
any foreshock (FS) and/or aftershock (AS) are treated as independent events. The char-
acteristics of temporal sequence and independent events in the cluster zones show inter-
esting variations. It is noteworthy that in all the clusters, there are more independent events
than the earthquakes classified under temporal (FMA) sequences. The only exception is the
‘cluster A’ that has more events in FMA sequence due to Kashmir earthquake of
08.10.2005 (Mb 7.7) with 1 FS and 195 AS. The magnitudes of ‘individual earthquakes’
vary from 4.5 to 5.8 in all the clusters, apart from a higher magnitude event of Mb 6.1
(01.06.2005) in cluster H.
It is worthwhile to investigate the seismotectonic implications of these clusters.
According to the characteristic earthquake model (Schwartz and Coppersmith 1984),
faults rupture in a series of characteristic earthquakes, each identical, with the same slip
distribution and length. Especially in the case of a fault of finite length, the characteristic
earthquake model fits fine with attributes such as ‘constant displacement per event’ at a
point, variable slip rate along the fault and persistent large events with infrequent
moderate earthquakes (Scholz 2002). In the light of above, we propose that the identified
cluster zones are strong candidates for earthquakes of repetitive nature.

123
Table 1 Description and statistics of temporal and independent events in the eight cluster zones (A–H of Fig. 1) in Lesser Himalaya
Cluster No of events Temporal sequence in cluster Independent events in cluster
(Mb C 4.5)
No of Main shock date No of recorded Name of Years No of Range

123
Mmax with date
Fore-shock and magnitude in Mb Aftershock earthquake events in Mb

A (Kashmir) 210 – 28.12.1974 (5.0) 3 1966 1 5.1 (06.04.1966) 5.1


– 14.02.2004 (5.5) 2 1971 2 5.2 (27.12.1971) 4.8–5.2
1 08.10.2005 (7.7) 195 Kashmir earthquake 1978 1 4.9 (27.04.1978) 4.9
1984 2 4.6 (04.06.1984) 4.5–4.6
B (Kangra) 74 2 20.02.1967 (5.5) 2 1966 1 4.7 (16.03.1966) 4.7
1 24.10.1973 (5.3) 2 1968–1970 4 4.9 (05.11.1968) 4.5–4.9
2 05.12.1975 (5.3) 8 1974–1975 2 4.8 (16.11.1974) 4.6–4.8
1977–2006 47 5.5 (26.04.1986) 4.5–5.5
C (Garhwal) 47 1 19.10.1991 (6.4) 3 Uttarkashi earthquake 1967–1990 23 5.6 (10.07.1986) 4.5–5.6
– 28.03.1999 (6.3) 8 Chamoli earthquake 1992–1996 6 4.8 (08.12.1994) 4.5–4.8
2002–2005 4 5.0 (14.12.2005) 4.5–5.0
D (Kumaon– 94 – 26.09.1964 (5.9) 1 1964–1965 4 5.3 (20.12.1964) 4.8–5.3
West Nepal) – 27.06.1966 (6.0) 12 1967–1979 16 5.7 (20.05.1979) 4.5–5.7
1 29.07.1980 (6.1) 4 1981–1984 10 5.1 (19.02.1984) 4.5–5.1
– 18.05.1984 (5.6) 5 1986–2000 20 5.6 (09.12.1991) 4.5–5.6
– 27.11.2001 (5.5) 4 2002–2006 12 5.0 (05.05.2006) 4.5–5.0
E (East Nepal) 28 – 26.02.1970 (5.0) 1 1966–1968 2 4.9 (28.10.1968) 4.5–4.9
– 24.03.1974 (5.4) 1 1971 1 4.9 (06.06.1971) 4.9
1 20.04.1988 (5.4) 1 1975–1987 7 5.2 (04.10.1978) 4.5–5.2
– 20.08.1988 (6.4) 2 Bihar–Nepal earthquake 1989–2006 8 5.0 (03.02.2006) 4.5–5.0
Nat Hazards
Table 1 continued

Cluster No of events Temporal sequence in cluster Independent events in cluster


(Mb C 4.5)
Nat Hazards

No of Main shock date No of recorded Name of Years No of Mmax with date Range
Fore-shock and magnitude in Mb Aftershock earthquake events in Mb

F (Sikkim) 45 – 12.01.1965 (5.9) 1 1964 3 5.1 (30.08.1964) 4.8–5.1


1 21.08.1972 (5.1) – 1971 1 5.2 (04.12.1971) 5.2
1 19.11.1980 (6.0) – Gangtok earthquake 1975–1978 5 5.0 (26.11.1975) 4.5–5.0
– 27.11.1997 (5.0) 1 1982–1996 17 5.7 (09.01.1990) 4.5–5.7
1998–2006 11 5.6 (03.09.1998) 4.5–5.6
G (Bomdila) 82 1 09.12.1965 (5.3) – 1964 2 5.6 (01.09.1964) 5.6
– 07.01.1985 (5.4) 2 1966–1984 27 5.8 (15.09.1967) 4.5–5.8
– 12.10.1985 (5.3) 4 1987–1997 32 5.1 (08.03.1989) 4.5–5.1
– 26.09.1985 (5.5) 1 1999–2006 6 5.3 (25.01.2006) 4.7–5.3
– 23.02.2006 (5.8) 2
H (Eastern Syntaxis) 12 – 09.05.1988 (5.1) 2 1977–1981 4 4.7 (14.10.1981) 4.5–4.7
1991–2005 5 6.1 (01.06.2005) 4.5–6.1

123
Nat Hazards

Fig. 3 Temporal distribution of earthquakes in 5-year bin in different clusters

The cluster zones in the Himalayas contain tectonic surfaces like main central thrust
(MCT) and/or main boundary thrust (MBT)/main frontal thrust (MFT), segmented by
discontinuities of transverse faults; many of which are transgressing from Peninsular shield
area. Each of the clusters displays occurrence of more or less uniform magnitude earth-
quakes [e.g., Uttarkashi (6.4) and Chamoli (6.3) in Garhwal cluster C; Mainshock 5.5–6.1
in Kumaun–West Nepal cluster E; Mainshock 5.3–5.5 in Kangra cluster B] suggesting
more or less ‘similar displacement per event’. The principal tectonic surfaces (viz., MCT,
MBT, MFT) have been deformed with variable slip rate along strikes (increasing differ-
ential shortening) from west to east (e.g., 14 ± 1 mm/year in western Himalaya, Banerjee
and Burgmann 2002; 19±1 mm/year in eastern Himalaya, Chen et al. 2004). If we look
into the historical records, it is observed that the size of large earthquakes that recurred in
each cluster is more or less similar in size (Table 2). While Kangra cluster (B) experienced
two large earthquakes of Mw 7.6 (1555) and 7.8 (1905), East Nepal (cluster E) was visited
by two large earthquakes of Mw 7.7 (1833) and 8.1 (1934) and Kumaun–West Nepal
(cluster D) had a record of Mw 7.5 (1720) and 7.3 (1916). Integrating the above, the
characteristic earthquake recurrence model (Schwartz and Coppersmith 1984) defined by
constant displacement of faults per event at a point with variable slip rate along length and

123
Nat Hazards

Table 2 Historical earthquakes (1500 to 1963 AD) (after Ambraseys and Douglus, 2003) that ruptured the
cluster zones in Lesser Himalaya
Type of zone Name of earthquake Year and Moment Expected magnitude as per
magnitude estimation in column_3
of Table 3
(1) (2) (3) (4)

Cluster zone
B (Kangra) Srinagar earthquake September 1555, 7.6 8.29
Kangra earthquake April 1905, 7.8
Chamba earthquake June 1945, 6.3
C (Garhwal) Kumaon Earthquake September 1803, 8.1 8.25
D (Kumaon–West Nepal) Uttarpradesh earthquake July 1720, 7.5 8.28
Kumaon earthquake August 1916, 7.3
Nepal earthquake July 1926, 6.5
E (East Nepal) Bihar Nepal earthquake January 1934, 8.1 8.19
Nepal earthquake August 1833, 7.7
G (Bomdila) East Bhutan earthquake January 1941, 6.8 7.99

constant size large earthquakes coupled with infrequent moderate earthquakes seems to be
best suited for explaining these cluster events. Further, for a fault of a finite length where
slip rate varies strongly along strike and is segmented by discontinuities (as by the
transverse faults in the Himalayas), the characteristic earthquake model is fulfilling all
required criteria (see Scholz 2002).

4 Estimation of size of possible earthquakes in cluster zones

Within the clusters, we have employed the empirical relation between rupture length
(RL) and size of earthquake (M) as a function of type of faulting (reverse) and region
(inter-plate). The strike lengths of clusters have been measured in kilometers on map
(Fig. 1). These strike lengths may be indicative of the maximum rupture length (RL) that
can be generated by an earthquake. On measuring RL, the expected size of earthquake is
empirically calculated using the equation log RL = -2.86 ? 0.63 M for reverse faults
(Wells and Coppersmith, 1994). We have excluded strike slip motions from our calcu-
lations since the tectonic surfaces of MCT/MBT/MFT show predominant thrust move-
ment. The estimates show magnitude range between *7.7 and 8.4 (Table 3) in the eight
clusters. The estimated magnitude is obviously indicative of a maximum possible
earthquake size for respective clusters when the entire strike length ruptures. Smaller
rupture length would of course yield smaller magnitude earthquake. Using M, the
expected average displacement (AD) has also been calculated following the relation Log
AD = -4.8 ? 0.69 M (Wells and Coppersmith 1994). The range of slip displacement
would vary from 3.04 to 9.61 m. The expected moment magnitude calculated empirically
and magnitude of historical earthquakes occurring in particular clusters is more or less
comparable (Table 2).

123
Nat Hazards

5 Seismotectonic characteristics of cluster zones

After demonstrating the existence and characteristics of eight cluster zones, the behavior of
those spatio-temporal clusters in relation with regional tectonics may be examined at this
point. There may be number of factors responsible for earthquakes to occur as clusters. The
suggested factors include fault interactions, type of tectonics and its relation to far field
stress, characteristics of rupture, activation of basement and cover rocks etc.
The geometric disposition of the fault system in relation to interaction of conjugate/en
echelon fault/lineament etc. plays a pivotal role in the kinematics and dynamics of
earthquake triggering process. Domains of intersecting major discontinuity surfaces are
favorable locales for stress buildup and are considered to be seismically potential sites
(Andrew 1989; Talwani and Gangopadhyay 2003). Localized stress buildup has been
shown to be a function of preexisting zones of weakness in response to plate tectonic force.
The existing model of stress builtup near interacting/intersecting faults depends on relative
length and orientation of faults. The optimum condition of strain accumulation is found
when a suitably oriented fault is intersected and offset by transverse, shorter faults. This
model may not be applied as a general principle valid for all types of tectonic setting may
hold good in many tectonic domains including the present study area in the Himalayas
particularly in cluster C, D, E, F and G as discussed below.
Probing into the 42 years of seismic record (1964–2006) in the Himalayas, it is per-
ceived that clusters B, E, C and D successively may be propounded for potential source
according to seismic vulnerability. However, due to inherent uncertainties in the complex
earth system, a telltale suggestion is difficult to make. The pattern observed in the past may
or may not indicate the most vulnerable locales in the future. The concentration of
earthquakes in cluster A supersedes all other clusters, solely because of the Kashmir
earthquake of 2005. Threat potential of cluster A might have been possibly minimized due
to the recent rupture.
The Kangra cluster (B) had experienced last independent moderate earthquake (Mb 5.5)
20 years ago (26.4.1986) suggesting a temporal quiescence. On the other hand, this cluster
is characterized by the history of two large earthquakes (Fig. 4) in the past (1555—7.6;

Table 3 The expected magnitude (M) of the earthquake derived from rupture length (RL) and its average
displacement (AD) in the cluster zones of Lesser Himalaya. A—Kashmir, B—Kangra, C—Garhwal,
D—Kumaun–West Nepal, E—East Nepal, F—Sikkim, G—Bomdila and H—Eastern Syntaxis
Type of zone Expected (maximum) rupture Expected magnitude (M) Expected average displacement
length (RL) (in km) (for reverse fault) (AD)/slip (in meter) calculated
taking M of reverse fault
(1) (2) (3) (4)

Cluster
A 151 8.00 5.236
B 230 8.29 8.298
C 217 8.25 7.780
D 226 8.28 8.147
E 200 8.19 7.128
F 263 8.38 9.616
G 149 7.99 5.164
H 92 7.66 3.040

123
Nat Hazards

1905—7.8). Empirical estimation also shows that in case of total rupture of the cluster
length, an earthquake of M 8.2 is expected in cluster B.
Garhwal cluster (C) is dotted with two recent catastrophic earthquakes, viz., Uttarkashi
(1991, 6.4) and Chamoli (1999, 6.3). MCT passes through cluster C and shows a cross-
cutting relationship with Ropar fault (RF) and Mahendragarh-Dehradun Fault (MDF). The
Garhwal cluster has one significant great historical event (1803—8.1).
In the Kumaun–West Nepal cluster (D), there is no record of large earthquakes in the
recent past; only moderate tremors visited the area. However, it contains Main Central
Thrust (MCT) as predominant seismogenic surface (also MBT in the southern fringe of the
cluster) intersected by transverse Great Boundary fault (GBF). Two large historical events
(1720—7.5; 1916—7.3) had dotted the area. Such interaction of crosscutting faults both
from the Himalayas (E–W trending MFT, MBT, MCT) and from N–S to NE–SW trending
deep-seated faults traversing from Peninsular India [viz. GBF, East Patna Fault (EPF) etc.]
may likely to have a role in earthquake incidence.
Intersecting domains of thrusts/faults seems very relevant for East Nepal cluster (E).
MFT, MBT and MCT pass through cluster E and are intersected by three dominant
transverse faults (Dasgupta et al. 1987) traversing from the plains of India (Fig. 1). These
faults are East Patna Fault (EPF), West Patna Fault (WPF) and Munger Saharsha Ridge
Marginal Fault (MSRMF). The epicenter of the disastrous Bihar–Nepal earthquake (1988,
6.1) is located near such a fault (see Dasgupta 1993). Further, two major historical events
(1833—7.7; 1934—8.1) of cluster E had shaken northern India awfully (Fig. 4).
We have indicated clusters B, E, C and D above in view of vulnerable source zones
taking into consideration the temporal events, historical records, presence of intersecting
faults and estimated expected M. Besides, intersections of fault systems are also prominent
in other clusters. Sikkim Cluster F illustrates a folded MCT intersected by Malda
Kishanganj Fault (MKF). Bomdila Cluster G, located in East Bhutan, is characterized by

Fig. 4 The plot of historical earthquakes (refer Table 2) and four identified clusters as most vulnerable
source zones (gray zones—B, C, D, E) for future Lesser Himalayan earthquakes. Annotation is similar to
Fig. 1

123
Nat Hazards

the presence of all the three tectonic surfaces of MFT, MBT and MCT with the Bomdila
fault as the major cross fault interaction (see also Nandy and Dasgupta 1991). Since these
seismic clusters are seemingly controlled by the faults from Peninsular India, the far field
Indian plate driving forces reactivate the whole system and as a result moderate to large
earthquakes may continuously recur. Occurrence of earthquakes in similar conditions has
also been inferred elsewhere (see Kenner 2003).

6 Discussion

In a collision orogen like the Himalayas, thrust tectonics dominates in the foreland and
hinterland flanks of the orogenic belt. In the Lesser Himalayan foreland, deformation either
may be restricted to the cover sequence (‘‘thin skinned’’ tectonics) or may actively involve
both the cover sequence and crystalline basement underlying it (‘‘thick-skinned’’ tectonics;
Sibson 2002). Thick-skinned tectonics thus involves compressional reactivation of base-
ment and its penetrating fault systems inherited from previous crustal extension (e.g., fault
systems from Peninsular India, Fig. 1). With a view to infer the depth distribution of
earthquakes in eight clusters (Table 4), depth sections for seven representative clusters
(A, B, C. D, E, F, G) (Fig. 5a–g) are constructed. Orientation of depth section lines is
approximately perpendicular to the thrust zones (Fig. 5). In cluster A, though the epicenter
of recent Kashmir earthquake (focal depth 26 km) is located in the upper crust, there is
number of earthquakes including Mb 5.0 that have been produced in deep crustal condi-
tions (60–70 km, Fig. 5a). In clusters B, C and G, earthquakes [5 Mb are mostly confined
in the upper crustal level (Fig. 5b, c, g). The events in clusters D, E and F have been
originated from upper as well as from lower crust, including few located at the interface
between lower crust and mantle (Fig. 5d–f). The conspicuous absence of deep-seated faults
from Peninsular India within inter-cluster zones might have made the inter-cluster zones
seismically less active. Thus it may be surmised that cluster zones of earthquakes do exist
in the Himalayan arc where ‘thick-skinned’ tectonics coupled with reactivation of base-
ment penetrating faults from Peninsular India could be the main source of earthquakes.
Earthquakes may strike in the same locale replicating previous record by re-rupturing.
The incidence of earthquakes within each cluster may also follow characteristic earthquake
model as already discussed. Inference may be made with caution from the comparative
study of the eight cluster zones. A speculation is insinuated that the clusters B (Kangra),
E (East Nepal), C (Garhwal) and D (Kumaon–West Nepal) are potential source zones
[marked gray in Fig. 4, in decreasing order of earthquake threat to bounce back to produce

Table 4 Depth (km) distribution


Cluster Number of earthquake Number of earthquake
of the earthquake epicenter in
(Mb C 4.5) [45 km depth
eight cluster zones in Lesser
Himalaya
A 210 9
B 73 13
C 46 8
D 94 14
E 28 3
A—Kashmir, B—Kangra,
C—Garhwal, D—Kumaun–West F 45 12
Nepal, E—East Nepal, G 82 18
F—Sikkim, G—Bomdila and H 12 3
H—Eastern Syntaxis

123
Nat Hazards

Fig. 5 Enlarged view of the


seven clusters along with
corresponding seismic depth
sections; (a) Cluster A,
(b) Cluster B, (c) Cluster C,
(d) Cluster D, (e) Cluster E
(f) Cluster F and (g) Cluster G.
‘Star’ in depth section indicates
the main shock events in the
clusters (see Table 1). Note: BF
bomdila fault, MBT main
boundary thrust, MFT main
frontal thrust, MCT main central
Thrust

123
Nat Hazards

large earthquakes (C7.8 M)]. If these earthquakes in shallow depth show reverse fault
mechanism, the expected size may range from M 7.7 to 8.4 with a comparable displace-
ment (3.04–9.61 m) along rupture plane possibly accommodating the slip deficit in the
Himalayas.
Historically observed pattern may portray occurrence scenario but may not be a strong
enough tool for forecasting. The estimated size of the earthquakes should not be accepted
as a ‘forecast’ since our main objective was to identify source zones based on historical
seismic records. Nevertheless, the earthquake threat potential envisaged in the source
zones cannot be ignored altogether.

Acknowledgments We express our thankfulness to Dr. Thomas Glade, Editor in Chief, for his thoughtful
comments on the earlier version of the manuscript. We convey our gratefulness to three erudite anonymous
reviewers whose suggestions have helped immensely to improve the scientific contents of the paper.

References

Aki K (1965) Maximum likelihood estimate of b in the formula log N = a-bM and its confidence limits.
Bull Earthquake Res Inst Tokyo Univ 43:237–239
Ambraseys N, Douglas J (2003) Magnitude calibration of north Indian earthquakes. J Geophy Int 158:1–42
Andrew DJ (1989) Mechanics of fault junction. J Geophy Res 94:9389–9397
Banerjee P, Burgmann R (2002) Convergence across the northwest Himalaya from GPS measurements.
Geophys Res Lett 29(13):31–34
Bettinelli P, Avouac JP, Flouzat M, Jouanne F, Bollinger L, Willis P, Chitrakar GR (2006) Plate motion of
India and interseismic strain in the Nepal Himalaya from GPS and DORIS measurements. J Geodesy
80:567–589
Bilham R, Ambraseys NN (2005) Apparent Himalayan slip deficit from the summation of seismic moments
for Himalayan earthquakes, 1500–2000. Current Science 881:658–1663
Bilham R, Wallace K (2005) Future Mw [ 8 earthquakes in the Himalaya: Implication from the 26 Dec
2004 MW = 9.0 earthquake on India’s Eastern Plate Margin. Geol Surv India 85:1–14 special
publication
Chen QZ, Freymueller JT, Wang Q, Yang ZQ, Xu CJ, Liu JN (2004) A deforming block model for the
present-day tectonics of Tibet. J Geophys Res 109(B1):B01403
Dasgupta S (1993) Seismotectonics of the Eastern Himalaya and its foredeep. Geol Surv India 31:73–81
Special Publication
Dasgupta S, Mukhopadhyay M, Nandy DR (1987) Active transverse features in the central portion of the
Himalaya. Tectonophysics 136:255–264
Dasgupta S, Pande P, Ganguly D, Iqbal Z, Sanyal K, Venkatraman NV, Dasgupta S, Sural B, Harendranath
L, Mazumdar S, Sanyal S, Roy A, Das LK, Misra PS, Gupta H (2000) Seismotectonic Atlas of India
and Its Environs. Geol Surv, India, Kolkata
Ebel JE, Kafka AL (2002) A non-poissinian element in the seismicity of the Northeastern United States. Bull
Seismol Soc Am 92:2040–2046
Feldl N, Bilham R (2006) Great Himalayan earthquakes and the Tibetan plateau. Nature 444:165–170
Jade S, Mukul M, Bhattacharyya AK, Vijayan MSM, Jaganathan S, Kumar A, Tiwari RP, Kumar A, Kalita
S, Sahu SC, Krishna AP, Gupta SS, Murthy MVRL, Gaur VK (2007) Estimates of interseismic
deformation in Northeast India from GPS measurements. Earth Planet Sci Lett 263:221–234
Kafka AL, Walcott JR (1998) How well does the spatial distribution of smaller earthquakes forecast the
locations of larger earthquakes in the Northeastern United States? Seismol Res Lett 69:428–440
Kagan YY (1997) Are earthquakes predictable? Geophys J Int 131:505–525
Kenner SJ (2003) Understanding the sources of stress which load faults in low strain rate tectonic regimes.
Abst., Indo-US workshop on Seismicity & Geodynamics, October 6–10, 2003, NGRI, Hyderabad,:
21–24
McGuire RK (2004) Seismic hazard and risk analysis. Earthquake engineering research institute (EERI)
Okland California USA Monograph series 10
Nandy DR, Dasgupta S (1991) Seismotectonic domains of Northeastern India and adjacent areas. In geology
and geodynamic evolution of the Himalayan collision zone. Phys Chem Earth 18:371–384
Scholz CH (2002) The mechanics of earthquakes and faulting, 2nd edn. Cambridge University Press, UK

123
Nat Hazards

Schwartz D, Coppersmith K (1984) Fault behaviour and characteristic earthquakes: example from the
Wasatch fault and San Andreas fault zones. J Geophys Res 89(B7):5681–5698
Sibson RH (2002) Geology of the crustal earthquake source. In: Lee WHK, Kanamori H, Jennings PC,
Kisslinger C (eds) International handbook of earthquake and engineering seismology, 81st edn.
Academic Press, Amsterdam, pp 455–473
Talwani P, Gangopadhyay A (2003) Seismogenesis of intraplate earthquakes. Abst., Indo-US workshop on
Seismicity & Geodynamics, October 6–10, 2003, NGRI, Hyderabad,: 16–17
Wang JH, Kuo CH (1998) On the frequency distribution of interoccurrence times of earthquake. J Seismol
2:351–358
Wells DL, Coppersmith KJ (1994) New empirical relationships among magnitude, rupture length, rupture
width, rupture area and surface displacement. Bull Seismol Soc Am 84(4):974–1002

123

You might also like