Acta Geophysica

vol. 62, no. 4, Aug. 2014, pp. 737-761

DOI: 10.2478/s11600-013-0186-1

Probabilistic Seismic Hazard Analysis for

the City of Quetta, Pakistan
Shafiq Ur REHMAN1, Conrad LINDHOLM2,
Najeeb AHMED3, and Zahid RAFI3
1
National Seismic Monitoring Center,
Pakistan Meteorological Department, Islamabad, Pakistan
e-mail: srltp@yahoo.com (corresponding author)
2
Norwegian Seismic Array, NORSAR, Kjeller, Norway
3
Pakistan Meteorological Department, Islamabad, Pakistan

Abstract
Seismic hazard assessment for Quetta is carried out using probabil-
istic seismic hazard analysis technique based on area sources and aug-
mented by line source used for the first time in Pakistan. Seismic data
has been collected and analyzed in spatial and temporal domains. Five
Seismic Zones have been modeled in line with tectonics of the region
with b-value of 1.14 using regression. The b-value is slightly higher,
which is attributed to the fact that aftershocks were not removed as it dis-
torted the dataset. Five fault sources are modeled, with three as reverse
and two as strike-slip with 7.8 as maximum magnitude. Mach Structure
is included in the tectonics for the first time. The attenuation relation
used in the present study is recommended by various researchers. The
expected Peak Ground Acceleration for 500-year return period is
4.79 m/s2 for rock outcrop and characterized as very high. Furthermore,
variation in spectral acceleration within Quetta city is observed, for
which spectral curves are developed for four different places.

Key words: seismic hazard, tectonics and faults, recurrence, Peak

Ground Acceleration, attenuation relation.

© 2014 Institute of Geophysics, Polish Academy of Sciences

Unauthenticated
Download Date | 9/24/15 11:30
PM
738 S.U. REHMAN et al.

1. INTRODUCTION
Quetta City is the provincial capital of Balochistan Province in Pakistan
(Fig. 1) and has a population of 1.2 million. The city is located in a valley
and has experienced severe earthquakes in the past. The most famous was
the 31 May 1935, magnitude 7.7 earthquake that killed 40000 people with
almost entire city infrastructure destroyed. The world famous Chaman Fault
passes close to Quetta. Over the last decade, the city has grown with many
buildings three to four stories high. This stipulates the need for accurate
seismic hazard map utilizing state-of-the-art technique for hazard analysis.
The method is based on a technique defined by Cornell (1968) for line
sources and later modified by McGuire (1976, 1978) for area sources; the
method consists of the four steps: (i) definition of seismic sources, (ii) defi-
nition of seismicity recurrence parameters, (iii) ground motion prediction,
and (iv) evaluation of return period and probability of exceedance of ground
motions. The technique is well developed and well described and imple-
mented by many researchers, e.g., Rafi et al. (2012), Bhatti et al. (2011), and
Das (2011) for use of seismic hazard evaluation in Pakistan. A significant
enhancement in this technique is made in the present study by inclusion of
geological structures as line sources modeling using Wells and Copper-
smith’s (1994) results for rupture length and magnitude relations. Further-
more, a new structure, named the Mach Structure, is added after findings of

Fig. 1. Location of Quetta City. Colour version of this figure available in electronic
edition only.
 PROBABILISTIC SEISMIC HAZARD ANALYSIS 739

Rafi et al. (2013) to further update the fault system in the area. The program
CRISES 2007 (Ordaz et al. 2007) with area and line sources has been used
in the present study.
Over the past few years, different studies have been conducted for seis-
mic hazard analysis of Quetta. The study conducted by the Pakistan Meteor-
ological Department and NORSAR (PMD-NORSAR 2007) for probabilistic
hazard of Pakistan is utilizing a very coarse gird of 1° × 1° which cannot be
recommended for hazard assessment at local level. For a return period of 500
years, the Peak Ground Acceleration is suggested to be 3.85 m/s 2. Building
codes of the Pakistan Seismic Provisions 2007 (BCP SP 2007), the Quetta
district and its Tehsil Panjpai (S/T) are located in Zone 3 with the highest
design values of 0.32 g for this zone. Areas in the south, which include
Mastung and Mach, are also in Zone 3. The areas north of Quetta, including
Pashin, and Muslim Bagh and Ziarat in the north east are in Zone 4. Das
(2011) has estimated that Quetta city at 2% probability in 50 years is
expected to have PGA of 0.67 g (corresponding to 6.6 m/s 2), with highest
values estimated for Sheikh Manda and Chiltan Housing Society. For the
present study, the tectonic setting and fault system around Quetta have been
studied in detail. The seismic catalogue is updated with data from various
contributing agencies and the existing seismotectonic and fault maps have
been updated. Probabilistic Seismic Hazard analysis technique is adopted
with area source and line source modeling.

2. TECTONICS AND LOCAL FAULTS

Over the geological time scale, the Indian plate has moved northward and
rotated in counterclockwise direction. The collision between the Indian and
Eurasian plates began some 30 to 40 myBP (Aitchison et al. 2007) and the
Tethys Ocean has entirely been consumed between the Eurasian, the Arabi-
an, and the Indian plates (Powell 1979). The present-day tectonics is marked
by collision and thrusting along the Main Boundary Thrust (MBT), Pamir
Himalaya, and the Hindu Kush region, forming the northern plate boundary.
On the western side, the tectonics of colliding Indian and Eurasian plates is
governed by transform plate boundary, consisting of the Chaman and Ornach
Nal Fault Zone with left lateral strike slip motion (Fig. 2). In the southern-
most areas, i.e., west of Ornach Nal Fault Zone, the oceanic lithoshpere is
subducting below the continental crust. In terms of structural trends, the
Main Boundary Thrust (MBT) and allied thrust faults form an elongated
zone in NW-SE direction, whereas the western margin along Chaman and
Ornach Nal Zone exhibits a dominent N-S trend. Due to compressional re-
gime of the Arabian plate, the Chagai Arc and the Markan Zone have struc-
tural trends in the E-W direction. The structural trends observed on local
740 S.U. REHMAN et al.

scale around Quetta have two distinctive natures, i.e., thrust and strike slip.
The Chaman Fault Zone is located on the western side of Quetta city and Su-
leiman Arc and Sibbi Trough on eastern side of the Quetta city. To the south,
the Kirthar ranges stretches due south from Quetta.
The Chaman Fault was first discovered in 1893 after the 1892 earth-
quake which offset the Quetta-Chaman railroad by 75 cm in left lateral
movement (Griesbach 1893). The fault is a left lateral transform fault. It is
considered responsible for famous 1935 (Ms = 7.7) earthquake. Lawrence
and Yeats (1979) divided it into four segments. First is the active fault which
is a linear feature and extends without interruption over the entire length of
the segment. Alluvial fans are cut by this segment of the fault with visible
marks in aerial photos over areas of bedrock. The second part is the Chaman
Fault Zone which is one kilometer wide and with a fault gouge, and an ac-
tive fault zone is present in this zone. The third segment is the Chaman
Fault system which is a series of four or five branches that curve away from
the main trace towards south west and west. The fourth is the Chaman trans-
form boundary which includes the Chaman, Ornach Nal, and other faults
and constitutes the boundary between the Indian and the Eurasian plates.
The Sulaiman Lobe or Sulaiman Arc is a broad (>300 km) and gentle
(<1° sloping hills) fold and thrust belt that is tectonically active. It has de-
veloped by transpression as a result of the left-lateral strike slip motion along
the Chaman Fault and southward thrusting along the western terminus of the
Indian subcontinent (Sarwar and DeJong 1979, Lawrence et al. 1981). The
present-day structural style and tectonics of the Ziarat area was developed
due to the interaction between the Suleiman Lobe, the Chaman Fault, and the
Sibbi Trough. Two major thrusts, the Gogai Nappe and the Babai Nappe, are
also present along with associated strike slip faults (Niamatullah et al. 1989)
The geological survey of Pakistan has worked in the field of identifica-
tion of tectonic regimes, fault systems, and geological features in Pakistan
over the years; however, the existing fault maps for Pakistan were not re-
vised. Khan et al. (2012), working to develop the fault map for Pakistan,
provided the most reliable and updated information of faults around Quetta
city (Fig. 3). It is evident that the fault closest to Quetta city is the Kalat
Fault, which starts north of Anjira town and passes immediately west of the
city, as denoted by number 7 in Fig. 3. The length of Kalat Fault is 107 km
with a dip in the west-northwest direction. The fault continues with a change
in orientation to east, making a bend northwest of Kach and beyond this
point is known as the Chinjan-Zakriazi Thrust (No. 19 in Fig. 3) with report-
ed length of 239 km. The Chaman Fault was first discovered in 1892
(Griesbach 1893) and originates as Ornach Nal Fault (Fig. 2) south of Quet-
ta, stretching in the north-south direction until it meets the Herat Fault in the
extreme north (Fig. 2). The Chaman Fault is a left-lateral strike slip feature
 PROBABILISTIC SEISMIC HAZARD ANALYSIS 741

Fig. 2. Regional tectonic map of the area: B – Bannu, HKS – Hazara Kashmir
Syntaxis, J-K – Jacobabad Khairpur high, K – Karachi, KB – Kalabagh, KK –
Khuzdar Knot, KP – Kalat Plateau, MBH – Mari Bugti Hills, NH – Nanga Parbat
Haramosh Massif, P – Peshawar, TS – Trans Indus Salt Range, Q – Quetta, Z –
Ziarat (Sarwar and DeJong 1979).

that in recent history has not released a regular high earthquake activity.
Ambraseys and Bilham (2003) have considered this fault as zone of signifi-
cant accumulation of stress that has made this structure ready for one or
more major earthquakes. The Chaman Fault streaches over 600 km in length
(Khan et al. 2012, Kazmi 1979).
The Harnai-Tatra Fault (Fig. 3, No. 17) is located north east of Quetta
and is traced in the area between the Kach and Khost towns; it has a length
of 140 km with a northwest-southeast orientation and it passes close to
Harnai. It further continues in the SE direction till it bends and aligns in
almost east-west direction. The reported dip of the fault is in northeast direc-
tion. The Harnai-Tatra Fault is located in area that has a high seismic activity
north of Quetta (Figs. 11 and 12). Another important fault is the Ghazaband
Fault (Fig. 3, No. 6) with reported length of 425 km. It has the same orienta-
742 S.U. REHMAN et al.

Fig. 3. Fault system around Quetta city. Explanations: 1 – Chaman Fault, 2 –

Dalbadin Fault, 3 – Ahmadwal Fault, 4 – Ban Fault, 5 – Usman Fault, 6 – Ghaza-
band Fault, 7 – Kalat Fault, 8 – Pandran Fault, 9 – Gokurt Fault, 10 – Bibai Fault,
11 – Anjira Fault, 12 – Nagau Fault, 13 – Bolan Zardak Fault, 14 – Kotra Fault,
15 – Gandava Fault, 16 – Pirkoh Thrust, 17 – Harnai Tatra Fault, 18 – Karahi Fault,
19 – Chinjan-Zakriazi Fault, 20 – Zhob Valley Thrust, 21 – Dasht Murga Fault,
22 – Katawaz Fault, 23 – Chaman Fault splay, 24 – Mokur Fault. Source: Khan
et al. (2012). Colour version of this figure available in electronic edition only.

tion as the Kalat Fault and runs parallel to it, passing west of Quetta. This
lies between the Chaman Fault and the Kalat Fault. Other important faults
include Anjira, Bibai, and Gandava Faults. The local faults have been dis-
cussed in detail later in the text relating to in source zones and recurrence pa-
rameters.
 PROBABILISTIC SEISMIC HAZARD ANALYSIS 743

3. SEISMIC DATA
The catalogues used in this study are established from reports of various
international agencies that have been and are in operation over a fairly long
time. This is important since the length of the catalogue is essential, given
the fairly long recurrence times for the largest earthquakes. The catalogues
used in the present study are International Seismological Center (ISC), Unit-
ed States Geological Survey (USGSPDE), and Pakistan Meteorological
Department (PMD) earthquake data catalogues, whereas the Harvard Cen-
troid Moment Tensor (CMT) catalogue provided data for focal mechanism
solutions. A total of 1935 events were analyzed in the study in a larger re-
gion bounded by latitudes 26°N to 33°N and 63°E to 73°E. Using the rela-
tion of Scordilis (2006) catalogues have been homogenized to obtain
moment magnitude (Mw). During the conversion, originally reported
moment magnitudes have been preserved with preference to ISC-reported
Mw. The second and third preference for conversion is mb and Ms to Mw,
respectively. The conversion of both catalogues (Fig. 4) using Scordilis

(a)

(b)

Fig. 4. Conversion of magnitude mb and Ms to Mw from: (a) the ISC catalogue, and
(b) USGS-PDE catalogue.
744 S.U. REHMAN et al.

(a)

(b)

(c)

Fig. 5. Time distribution of magnitudes in the ISC catalogue.

 PROBABILISTIC SEISMIC HAZARD ANALYSIS 745

(2006) gives a linear relation for both mb and Ms, showing the consistency in
conversion at a wide range of magnitudes.
The data has been analyzed in time domain to identify the duration of da-
ta completeness and corresponding magnitude. The technique has been
adopted as used by Rafi et al. (2012) for analysis of data catalogue in time
domain. In a glance it is evident that the ISC catalogue is considered to be
complete for magnitude 7.2 and above for the period 1909 onwards (Fig. 5a).
The data in lower magnitude range is available from 1961 with first reported
earthquake of magnitude below 5.0 in 1961; however, the bulk of input of
data starts from 1964 and the catalogue is considered to be complete for
magnitude 4.5 and above for the duration 1965 onwards (Fig. 5b, c). There is
a pattern of reduced seismic activity from 1967 to 1978, the reason for which
may be the old, decaying instrumentation. For magnitude 4.0 and above, the
ISC catalogue is considered to be complete from 1995.
The USGS PDE provided data from 1973 onwards for the area (Fig. 6).
For the initial period, up to 1978, the earthquake frequency is lower com-
pared to the rest of the period. However, the catalogue can be considered
complete for magnitude 5.0 and above from 1973, whereas it is complete for
magnitude 4.2 and above from 1992 onwards. The findings from the com-
parison of the results (Figs. 5 and 6) for both the catalogues suggest that the
results of the analysis are similar to findings of Rafi et al. (2012). It is con-
cluded that ISC catalogue is more reliable for use in seismic hazard evalua-
tion. During the analysis of catalogue three earthquakes were assigned
magnitudes as reported by Ambraseys and Bilham (2003) with details given
in Table 1, instead of ISC derived magnitudes.

Fig. 6. Time distribution of magnitudes in the USGS catalogue.

746 S.U. REHMAN et al.

Table 1
Detail of earthquakes with manual magnitude assessment
from Ambraseys and Bilham (2003)
Day Month Year Latitude Longitude Depth Mw used
30 5 1935 29.5 66.75 35 7.7
27 8 1931 29.75 67.25 35 7.3
24 8 1931 30.25 67.75 35 6.8

3.1 Historical earthquakes

In the study of greater region around Quetta city, records of several signifi-
cant historical earthquakes are available. The most significant is the Febru-
ary 1893 earthquake with intensity IX on MMI scale (Fig. 7), with the
epicenter located virtually in the city with no other information available
with PMD. An earthquake with the highest intensity (X on MMI scale) in the
area occurred in 1872, and is known as Lehri earthquake. Two significant
earthquakes, with intensities VIII and VII, are located northwest of Quetta
near town of Pishin.

Fig. 7. Historical earthquakes with high intensities (marked with light stars) around
Quetta city (dark star). Major seismic zone north of alignment of seismicity.
 PROBABILISTIC SEISMIC HAZARD ANALYSIS 747

(a)

(b)

Fig. 8. Instrumental seismicity from: (a) ISC catalogue, and (b) USGS catalogue.
Colour version of this figure available in electronic edition only.
748 S.U. REHMAN et al.

In the analysis of historical earthquake data for Quetta region, a remark-

able alignment of the historical earthquakes is observed, as shown in Fig. 7.
These earthquakes clearly demarcate a NW-SE trend that runs nearly parallel
from Sibbi and Mach to Quetta and further north to Chaman, with the
instrumentally recorded earthquakes (Fig. 8) with a clear separation towards
the north. It is believed that this distinction between instrumental catalogues
and the historical seismicity is real and locations are not biased. The histori-
cal earthquakes that form this line are given in Table 2. It is further interest-
ing to note that all of these earthquakes making an alignment occurred in a
short span of 35 years (1867-1904). This high seismic activity in short span
may indicate the activity along a hidden lineament. It is interpreted from an
alignment of historical epicentres and the relative time-clustering that there
exists a deeper lineament with large historical earthquakes occurring along
this lineament. This interpretation is later used for modelling a fault line.
Table 2
Details of earthquakes from 1867 to1903 forming an alignment
Max I
Day Month Year Latitude Longitude (roman)
– – 1867 29.1 68.3 VII
15 12 1872 29.2 68.2 X
20 12 1892 30.9 66.4 IX
13 2 1893 30.2 67.0 IX
– – 1900 30.4 67.0 VIII
– – 1902 30.6 66.8 VII
23 12 1903 29.5 67.6 VII

3.2 Instrumental seismicity

Data of instrumental seismicity have been plotted in Fig. 8a for ISC cata-
logue and Fig. 8b for USGS-PDE catalogue. Both the catalogues develop
a cluster of epicenters north east of Quetta. In a higher magnitude range, i.e.,
for magnitude 6.1 and above, there are ten earthquakes in the ISC catalogue
(Fig. 8a) and four earthquakes in the USGSPDE catalogue (Fig. 8b). Even in
the lower magnitude range, the seismicity reported in ISC database is more
reliable. The areas north and northeast of the Quetta city have a very high
seismic activity rate and most of large earthquakes are located in this zone,
whereas the rest of the area exhibits a scattered seismicity in spatial domain.
This spatial distribution is later utilized for characterization of area sources
for probabilistic analysis. Some important earthquakes in the instrumental
 PROBABILISTIC SEISMIC HAZARD ANALYSIS 749

era include the 1931 Sharigh earthquake, 1931 Mach earthquake, 1935 Quet-
ta earthquake, and 2008 Quetta-Ziarat earthquake.
The Sharigh earthquake originated on 24 August 1931 with a surface
wave magnitude Ms = 6.8 and depth of 33 km. It was followed by the
Ms = 7.3 Mach earthquake on 27 August (Ambraseys and Bilham 2003).
The information for intensity distribution of Sharigh earthquake is incom-
plete (West 1934). In the Mach earthquake, approximately 120 people died.
West (1934, 1937) has reported a maximum intensity of VIII. Most of the
damage was reported in Mach. All the buildings in New Mach were dam-
aged to some extent, while many houses built of dried mud bricks (adobe) or
stone with mud mortar were completely ruined along with two government
rest houses. In a more recent research, Szeliga et al. (2009) have concluded
that the earthquake occurred on a 42 km wide east-west and 72 km long
north-south Dezghat/Bannh fault system west of Sibbi. The fault slipped in a
reverse sense to the east with a maximum slip of 3.2 m and a mean slip of
1.2 m. They have also promoted (argued for) a mechanism that stored elastic
strain from past events that may account for a 3.2 m of local slip observed
up-dip from the inferred locking line of the 1931 earthquake.
The famous Quetta earthquake of 1935 originated on 30 May with an es-
timated surface wave magnitude of 7.7 (Ambraseys and Bilham 2003) four
years after the Mach earthquake, due to unclamping of the fault by the earlier
earthquake – Mach 1931. A total of 35000 people died in the earthquake.
About 26 000 people were killed in total in Quetta. In Kalat tribal areas,
some 8410 people died. This was the deadliest earthquake in the subconti-
nent of Pakistan before the occurrence of 8 October 2005 Kashmir earth-
quake. Quetta was divided into two major parts at that time, the cantonment
(military compound) and the civil areas. 15000 people died in the civil area
of Quetta city (Pinhey 1938). In the state of Kalat, 2900 people out of its
population of 10000 died and 5000 were injured. All villages between Quet-
ta and Kalat were destroyed with 70% of population either dead or injured.
Ambraseys and Bilham (2003) have placed the epicenter 160 km south of
Quetta about 40 km west of Kalat with an estimated seismic moment of
17.0 × 1027 dyne cm.
In the more recent activity, an earthquake of magnitude 6.5 Mb occurred
on 29 October 2008 in Quetta-Ziarat region with foreshock of 5.0 mb. Rafi
et al. (2013) has reported that the intensity of the main shock was VIII in and
around the Ziarat–Pashin areas, while the Peak Ground Acceleration record-
ed at Quetta area (60 km from the epicenter) was 0.17 m/s 2 (horizontal com-
ponent) and 0.06 m/s2 (vertical component) with source mechanism to be
strike slip for both events. The findings of Rafi et al. (2013) have been uti-
lized to model the line source in the region.
750 S.U. REHMAN et al.

4. SEISMIC HAZARD ANALYSIS

4.1 Area sources and recurrence parameters
Instrumental seismicity from catalogues of ISC and USGS and its correlation
with tectonic structures is used to identify the seismic source zones for haz-
ard evaluation of Quetta city (Figs. 2, 3, 8). The area has been divided into
four main zones (Main Zone, North Zone, South Zone, and West Zone) and
two sub zones, i.e., Remaining East and Remaining West (Fig. 9). Most of
seismicity is concentrated in the Main Zone (39%) shown by thick cluster in
Fig. 9. Most of earthquakes above magnitude 6.0 in instrumental era are lo-
cated in this zone with a shallow focus. Many of the earthquakes are in depth
range of 33 km. This zone contains a northwest-southeast trend in Sulaiman
Arc due to intrusion of Sibbi Trough and Jacobabad Khairpour high (Fig. 2).
The Harnai-Tatra Fault is located in this zone. The North Zone defines the
area of high seismicity largely related to Suleiman ranges that form an arc
known as Sulaiman Arc with Mari Bugti Hills on foreland side of the
Sulaiman Arc. Both Sulaiman ranges and Mari Bugti Hills terminate against
the intrusion of Sibbi Trough. The bend in the pattern of seismicity corre-
sponds to these tectonic features in the North Zone. All the seismicity in this
zone is below magnitude 6.0 with most of events in a magnitude range 4.1 to

Fig. 9. Source zones for seismic hazard analysis of Quetta city. Larger study area
around Quetta (Circum Zone) is shown by darker area. Colour version of this figure
available in electronic edition only.
 PROBABILISTIC SEISMIC HAZARD ANALYSIS 751

5.1 and shallow focus. This zone contributes with 9% of the total seismic
activity. In the South Zone, seismic activity is related to Kirthar Ranges
south of Quetta in north-south direction. Two large earthquakes (M > 6.1)
have occurred in this zone in the area where the seismic activity is low. In
extreme lower part of this zone, the seismic activity is high but the magni-
tude range of 4.0 to 6.0 is dominating. It is inferred that the low seismic
activity in northern part of this zone corresponds to high magnitudes in tem-
poral domain. This zone has a contribution of 15% in seismic activity. The
West Zone is unique with thick cluster of earthquakes just west of Quetta
city. This part contains the left-lateral Chaman transform fault. In larger
study area (Circum Zone) around Quetta city (Fig. 9), the seismicity is not
uniformly distributed and this has been marked as Remaining East and
Remaining West Zones. However, it is important to note that both of these
source zones contain one significant earthquake (M > 6.1) and cannot be
ruled out in seismic hazard analysis. Relative and absolute distribution with
the source zones have been summarized in Table 3.
Table 3
Definition of areas (absolute and relative) within the zones defined
Area Area Average activity
[km2] [%] [%]
Circum Zone 292672
Main Zone 16408 5.60627597 39
South Zone 42624 14.5637437 15
West Zone 6168 2.10747868 13
North Zone 40616 13.8776514 09
Remaining
(East and West) 186856 63.8448502 25

The regressions for recurrence parameters (Fig. 10) were conducted in-
dependently on two catalogues (ISC and USGS-PDE) on a larger zone,
comprising the area around Quetta. For these catalogues, the completeness
analysis results were used for constraining time and magnitude cutoffs as
follows:
1. For the ISC catalogue (instrumental starting from 1909) the following
periods of completeness were determined:
 from 1909 and onwards for M ≥ 7.2
 from 1965 and onwards for M ≥ 4.5
 from 1995 and onwards for M ≥ 4.0
2. For the USGS catalogue (instrumental starting from 1974) the fol-
lowing periods of completeness were determined:
 from 1974 and onward for M ≥ 5.0
 from 1992 and onward for M ≥ 4.2
752 S.U. REHMAN et al.

For these five completeness periods and magnitude windows of the cata-
logues, regression analyses were conducted in order to obtain the Guten-
berg–Richter recurrence relation: log(N) = a – b × Mw. The results for the

Fig. 10. Gutenberg–Richter regressions conducted on: (a) three independent time
and magnitude windows of the ISC catalogue, and (b) two independent time and
magnitude windows of the USGS catalogue. Thick line represents the average. All
data were normalized to 1 year. Colour version of this figure available in electronic
edition only.
 PROBABILISTIC SEISMIC HAZARD ANALYSIS 753

ISC catalogue are shown in Fig. 10a and for the USGS catalogue in Fig. 10b.
The results from the two catalogues yielded log(N) = 6.506 0 – 1.1422 × Mw
for ISC data and log(N) = 6.500 – 1.1327 × Mw for USGS data. It is ob-
served that the b-value of 1.14 is well established through the instrumental
catalogues, however at the same time it is recognized to be in the higher end
of expectations compared with some earlier investigations (e.g., Ambraseys
and Bilham 2003). This high b-value is attributed to the possible inclusion of
aftershocks that were not removed due to the risk of removing too many oth-
er earthquakes but accounts for any temporal uncertainty.
Using the above-mentioned method and analysis, the activity for the
source zones has been computed and tabulated in Table 4.
Table 4
Recurrence values used in the computational model as established
through regressions on the data
b N a T=6 T=5
Main Zone 1.14 2.43166156 6.08590313 5.67 0.41
South Zone 1.14 0.92001915 5.66379687 15.00 1.09
West Zone 1.14 0.84183517 5.62522707 16.40 1.19
North Zone 1.14 0.56981026 5.45573027 24.23 1.75
Remaining East 1.14 0.61216549 5.48686884 22.55 1.63
Remaining West 1.14 0.93408181 5.67038492 14.78 1.07
Explanations: N-value corresponds to an M ≥ 5.0 earthquake. The return period , T (in years),
given for each zone is for M = 6.0.

4.2 Line sources and recurrence parameters

It is very clear that the empirical data for PSHA investigation is extremely
short compared to the probabilities of interest in a standard hazard investiga-
tion, which are normally around 500 years recurrence, corresponding to
0.002 × 10–3 annual exceedance rates. For this reason, it was decided to also
include some local faults as contributing to the hazard of Quetta. While there
exists evidence that these faults and structural zones are indeed seismically
active, the actual recurrence or quantification of the earthquake activity of
these faults remains uncertain. Furthermore, in our study area, the seismicity
cannot be related to particular faults, however the technique of modeling a
characteristic earthquake for faults is utilized which has previously never
considered for areas in Pakistan territory in studies, e.g., Rafi et al. (2012),
Das (2011) and BCP-SP2007. In terms of rupture length as a function of
fault type, the Wells and Coppersmith (1994) relations were applied. Faults
modeled in this analysis contribute only ten percent of area sources to
account for any uncertainties (Table 5). These line sources are shown in
Figs. 11 and 12.
754 S.U. REHMAN et al.

Table 5
Parameters used in fault modeling. The total seismic moment release
from the faults is less than 10%
Modeled activity Min. and max. Mode
Fault name b-value
at M = 7.3 magnitude expected of faulting
Chaman Fault Λ = 0.00057 0.80 7.3-7.8 Strike slip
Kalat–Chinjin Fault Λ = 0.00057 0.80 7.3-7.8 Reverse
Harnai–Tatra Fault Λ = 0.00057 0.80 7.3-7.8 Reverse
Ghazaband–Zhob Fault Λ = 0.00057 0.80 7.3-7.8 Reverse
Mach Structure Λ = 0.00057 0.80 7.3-7.8 Strike slip

Chaman Fault. The Chaman Fault (or fault group) stretches sub N-S
until it meets the Herat Fault in the north. The Chaman Fault is a left-lateral
strike slip feature that in recent history has not released a regular high earth-
quake activity. Ambraseys and Bilham (2003) speculate whether there is a
significant accumulation of stress that has made this structure ready for one
or more major earthquakes.
Kalat–Chinjin Structure. The Kalat Fault starts north of Anjira town and
continues in a northward direction, passing through Kalat town and then
passes west of Quetta city. It further continues with a change in orientation
in a northeast direction. The reported length of the Kalat Fault is 107 km
with a dip in a west-northwest direction. The structure continues under the
tectonics of the Suleiman Arc in an easterly direction where it is called the
Chinjin-Zakriazi Thrust. This portion has a strike of 234 degrees and also in
northwest direction. These faults are considered as two different faults but
for the present study the Kalat Fault and the Chinjin-Zakriazi Thrust have
been merged and modeled as a single structure termed the Kalat-Chinjin
Structure.
Harnai–Tatra Structure. The fault originates in the area between
the Kach and Khost towns and has a length of 140 km with a northwest-
southeast orientation and passes close to Harnai. It further continues in SE
direction till it bends and aligns in almost east-west direction. The dip of the
fault is estimated to be in a northeast direction.
Ghazaband–Zhob Structure. The Ghazaband–Zhob Structure is also
modeled as a single fault similar to the Kalat–Chinjin Structure. It runs par-
allel to the Chaman Fault and the Kalat–Chinjin Structure in the south but
bends in the north and is coupling the Kalat–Chinjin Structure to the Sulei-
man lobe. The dip direction is southeast (opposite to the Kalat–Chinjin
Fault).
Mach Structure. This structure has not been previously described in lit-
erature and geological/fault maps. It has been included in the study keeping
 PROBABILISTIC SEISMIC HAZARD ANALYSIS 755

Fig. 11. Structural map of the region surrounding Quetta, Balochistan. Base map
courtesy of Khan et al. (2012). Stars denote historical earthquakes with intensities,
circles – seismicity above magnitude 5.0 from ISC, thick black lines – faults from
Khan et al. (2012), dashed line – Mach Structure after Rafi et al. 2013. Colour ver-
sion of this figure available in electronic edition only.

Fig. 12. Local faults modeled in the present PSHA analysis.

756 S.U. REHMAN et al.

in view the results of Rafi et al. (2013). This study indicated a northwest-
southeast structure and is contrary to previous descriptions of the area as
thrust. This structure is poorly recognized by surface manifestations; howev-
er, the large historical earthquakes seem to align along this structure. It may
be responsible for the recent Quetta-Ziarat earthquake of magnitude 6.5 mb
with a foreshock of magnitude 5.0 mb. The structure runs tentatively from the
Harnai–Tatra Fault in a southeasterly direction and passes close to Sibbi.

4.3 Attenuation relation and input parameters

Rafi et al. (2012), Das (2011), and Douglas (2004) have compared and ana-
lyzed different ground motion relations. For the present study, the attenua-
tion relation by Ambraseys et al. (2005) has been utilized which is in line
with Rafi et al. (2012) and Das (2011). This relation utilizes data of 595
triaxial strong motion records from seismically active regions of Europe and
Middle East (including Iran) with moment magnitudes as defined by
Kanamori (1977) and selection of events based on original estimates of
seismic moment. In terms of source to site distance relation, the distance
proposed by Joyner and Boore (1981) is used for earthquake with Mw great-
er than 6.0 and epicentral distance for lower magnitudes with a cutoff value
at Mw = 5.0. Ambraseys et al. (2005) utilized data from stations with known
velocity profile only to overcome the problem of incorrect coefficients as in
Ambraseys et al. (1996). In terms of spectral ordinates the period is truncat-
ed at 2.5 s due to paucity of data, particularly beyond 4 s.
Computer program CRISES2007 (Ordaz et al. 2007) is used for calcula-
tion of seismic hazard on a fine grid of 0.02 degrees. The capabilities of the
program were fully utilized for polygon, line sources, and characteristic
earthquake model. Seismic hazard has been computed in four different parts
of the city namely North, South, East, and West Quetta in terms of spectral
ordinates. For the area sources, the Gutenberg–Richter relations were used as
a basis for modeling the seismic activity. The fault sources were modeled as
an added contribution. This attempts to cover the possibly higher seismicity
than observed in the short observation period. The cumulative moment re-
lease of the modeled faults lines amounts to less than 10% of the cumulative
moment release of the area sources. The faults (structures) are modeled as
Gutenberg–Richter distributions with M = 7.3 as the minimum magnitude
and 7.9 as the maximum magnitude. Both area and line sources have been
discussed in detail earlier. The maximum contribution distance was set to
256 km in the software. Hazard was computed for 8 periods and for 20
ground motion levels ranging from 0.01 to 20 m/s2.
Annual rates used in the computational model together with the b = 1.14
as derived from the overall data (Fig. 10 and Table 4). Furthermore, a Mmax
 PROBABILISTIC SEISMIC HAZARD ANALYSIS 757

of 7.9 was used. This is regarded as the maximum credible earthquake that
may take place in the region albeit tapered off against M = 8.1 (Table 6).
Table 6
Annual recurrence for M = 5 for 6 area zones used
in the computational model together with mean Mmax value (7.9)
Annual N
Zone (M = 5) Mmax
Main Zone
(Suleiman Range) 2.43 7.9
West Zone
(west of Quetta) 0.84 7.9
North Zone
(northern Suleiman) 0.57 7.9
South Zone
(stretching south from Quetta) 0.92 7.9
Remaining East
(southeastern part) 0.612 7.9
Remaining West
(northwestern part) 0.934 7.9

5. RESULTS AND DISCUSSION

The hazard results obtained for the central part of Quetta city are given in
Table 7 and Fig. 13 for hard rock sites. There is a variation within Quetta
city in terms of spectral acceleration (Fig. 14), however the PGA for
500 years return period averages to 4.79 m/s 2. This corresponds to rock out-
crop values. The values obtained are characterized as very high, but may
indeed confirm the damages from previous earthquakes as, e.g., the 1935
earthquake.
Table 7
Ground motion results for 3 return periods and 8 periods
for central Quetta at rock site
Period Horizontal ground motion [m/s2] for recurrence period
[s] 100 years 500 years 1000 years
PGA 2.35 4.79 6.15
0.10 5.07 10.33 13.19
0.20 5.79 11.56 14.78
0.50 3.36 7.73 10.81
1.00 1.47 4.07 6.28
1.50 0.81 2.42 3.86
2.00 0.69 2.11 3.39
2.50 0.49 1.64 2.83
758 S.U. REHMAN et al.

Fig. 13. Seismic hazard curve for Quetta city center for hard rock site.

Fig. 14. Seismic hazard curve for different areas of Quetta city in terms of spectral
acceleration for hard rock site.

It is evident from the review and analysis in the present study that sever-
al large earthquakes have rattled the area in historical and instrumental eras.
The cause of these destructive earthquakes lies in the tectonic and fault sys-
tems in the region. During the present study the usual practice of using area
sources and recurrence parameters have been augmented with use of line
 PROBABILISTIC SEISMIC HAZARD ANALYSIS 759

sources or fault modeling. Seismic source zone are based on the tectonics
and seismicity of the region. In area sources, regression calculation was car-
ried out on catalogues of ISC and USGS to acquire the recurrence parame-
ters. In terms of line sources, four faults reported by Khan et al. (2012) and
a new significant feature after Rafi et al. (2013) are included in the study.
This is the first time such a technique is adopted in probabilistic analysis in
Pakistan. Quetta city was divided into four parts. It is observed that there is
a considerable difference in ground acceleration at spectral period of about
0.3 s (Fig. 14) for four zones of the city. This result can be used to facilitate
engineers for designing of buildings. This behavior of variation of spectral
acceleration is analogous to that described in Building Codes of Pakistan.
The value of ground acceleration in the present study is 4.79 m/s 2 for return
period of 500 years at spectral period of PGA, which is higher compared to
study of Rafi et al. (2012) with ground acceleration of 3.94 m/s2. It is due to
inclusion of line sources or fault system around Quetta.

Acknowledgment. Authors are thankful to Norwegian Ministry of

Foreign Affairs for providing financial support for research and Planning
Commission of Pakistan for facilitation in bilateral collaboration with Nor-
wegian Ministry of Foreign Affairs. Authors are also thankful to Norwegian
Seismic Array (NORSAR), Norway, for provision of technical and computa-
tional support during this work.

References

Aitchison, J.C., J.R. Ali, and A.M. Davis (2007), When and where did India
and Asia collide?, J. Geophys. Res. 112, B5, B05423, DOI: 10.1029/
2006JB004706.
Ambraseys, N., and R. Bilham (2003), Earthquakes and associated deformation in
northern Baluchistan 1892-2001, Bull. Seismol. Soc. Am. 93, 4, 1573-1605,
DOI: 10.1785/0120020038.
Ambraseys, N.N., K.A. Simpson, and J.J. Bommer (1996), Prediction of horizontal
re- sponse spectra in Europe, Earthq. Eng. Struct. Dyn. 25, 4, 371-400, DOI:
10.1002/(SICI)1096-9845(199604)25:4<371::AID-EQE550>3.0.CO;2-A.
Ambraseys, N.N., J. Douglas, S.K. Sarma, and P.M. Smit (2005), Equations for the
estimation of strong ground motions from shallow crustal earthquakes
using data from Europe and the Middle East: Horizontal peak ground
acceleration and spectral acceleration, Bull. Earthq. Eng. 3, 1, 1-53, DOI:
10.1007/s10518-005-0183-0.
BCP SP (2007), Building Codes of Pakistan Seismic Provisions Government of
Islamic republic of Pakistan Ministry of Housing and Works, Islamabad.
760 S.U. REHMAN et al.

Bhatti, A.Q., S.Z. Ul Hassan, Z. Rafi, Z. Khatoon, and Q. Ali (2011), Probabilistic
seismic hazard analysis of Islamabad, Pakistan, J. Asian Earth Sci. 42, 3,
468-478, DOI: 10.1016/j.jseaes.2011.05.006.
Cornell, C.A. (1968), Engineering seismic risk analysis, Bull. Seismol. Soc. Am. 58,
5, 1583-1606.
Das, A.J. (2011), Vulnerability assessment of structures subjected to earthquakes
(VASE) in Pakistan, Ph.D. Thesis, National Institute of Transportation,
School of Civil and Environmental Engineering, NUST, Islamabad, Paki-
stan.
Douglas, J. (2004), Ground estimation equations 1964-2003; a comprehensive
worldwide summary of strong motion attenuation relationships for peak
ground acceleration and spectral ordinates (1969 to 2000) with corrections
and additions, Imperial College Research Report No: 04-001.
Griesbach, C.L. (1893), Notes on the earthquake in Baluchistan on the 20th Decem-
ber 1892, Rec. Geol. Surv. India, 26, 2, 57-64.
Joyner, W.B., and D.M. Boore (1981), Peak horizontal acceleration and velocity
from strong-motion records including records from the 1979 Imperial Val-
ley, California, earthquake, Bull. Seismol. Soc. Am. 71, 6, 2011-2038.
Kanamori, H. (1977), The energy release in great earthquakes, J. Geophys. Res. 82,
20, 2981-2987, DOI: 10.1029/JB082i020p02981.
Kazmi, A.H. (1979), Active fault systems in Pakistan. In: A. Farah, and K.A.
DeJong (eds.), Geodynamics of Pakistan, Geological Survey of Pakistan,
Quetta, 285-294.
Khan, M.A., M. Javed, W. Ahmed, and M. Sayab (2012), Fault map and associated
database of fault parameters for Pakistan, International Project – WP1 Re-
port Earthquake Model of Middle East (EMME).
Lawrence, R.D., and R.S. Yeats (1979), Geological reconnaissance of the Chaman
Fault in Pakistan. In: A. Farah and K.A. DeJong (eds.), Geodynamics of
Pakistan, Geological Survey of Pakistan, Quetta, 351-357.
Lawrence, R.D., R.S. Yeats, S.H. Khan, A. Farah, and K.A. DeJong (1981), Thrust
and strike slip fault in interaction along the Chaman transform zone,
Pakistan, Geol. Soc. Spec. Publ. London 9, 1, 363-370, DOI: 10.1144/
GSL.SP.1981.009.01.33.
McGuire, R.K. (1976), FORTRAN computer program for seismic risk analysis, U.S.
Geol. Survey Open-File Report No. 76-67.
McGuire, R.K. (1978), FRISK: Computer program for seismic risk analysis using
faults as earthquake sources, U.S. Geol. Survey Open File-Report No. 78-
1007.
Niamatullah, M., K.H. Durrani, A.R. Qureshi, Z. Khan, D.M. Kakar, M.R. Jan, and
A. Ghaffar (1989), Emplacement of Bibai and Gogai Nappes, northeast of
Quetta, Geol. Bull. Univ. Peshawar 22, 153-158.
 PROBABILISTIC SEISMIC HAZARD ANALYSIS 761

Ordaz, M., A. Aguilar, and J. Arboleda (2007), Crisis 2007: Program for computing
seismic hazard, Ver. 3.01.
Pinhey, L.A. (1938), The Quetta earthquake of 31 May 1935, Government of India,
New Delhi.
PMD-NORSAR (2007), Seismic hazard analysis and zonation of Pakistan, Azad
Jammu and Kashmir, Pakistan Meteorological Department and NORSAR,
Technical Report July 2007.
Powell, C.McA. (1979), A speculative tectonic history of Pakistan and surroundings:
some constraints from the Indian Ocean. In: A. Farah, and K.A. DeJong
(eds.), Geodynamics of Pakistan, Geological Survey of Pakistan, Quetta, 5-
24.
Rafi, C., Lindholm, H. Bungum, A, Laghari, and N. Ahmed (2012), Probabilistic
seismic hazard of Pakistan, Azad-Jammu and Kashmir, Nat. Hazards 61, 3,
1317-1354, DOI: 10.1007/s11069-011-9984-4.
Rafi, Z., N. Ahmed, S. Ur Rehman, T. Azeem, and K.A. Abd el-Aziz (2013), Analy-
sis of the Quetta–Ziarat earthquake of 29 October 2008 in Pakistan, Arab.
J. Geosci. 6, 6, 1731-1737, DOI: 10.1007/s12517-011-0485-2.
Sarwar, G., and K.A. DeJong (1979), Arcs, oroclines, syntaxes: the curvature of
mountain belts in Pakistan. In: A. Farah and K.A. DeJong (eds.), Geody-
namics of Pakistan, Geological Survey of Pakistan, Quetta, 341-350.
Scordilis, E.M. (2006), Empirical global relations converting Ms and m b to moment
magnitude, J. Seismol. 10, 2, 225-236, DOI: 10.1007/s10950-006-9012-4.
Szeliga, W., R. Bilham, D. Schelling, D.M. Kakar, and S. Lodi (2009), Fold
and thrust partitioning in a contracting fold belt: Insights from the 1931
Mach earthquake in Baluchistan, Tectonics 28, 5, TC5019, DOI: 10.1029/
2008TC002265.
Wells, D.L., and K.J. Coppersmith (1994), New empirical relationships among mag-
nitude, rupture length, rupture width, rapture area and surface displacement,
Bull. Seismol. Soc. Am. 84, 4, 974-1002.
West, W.D. (1934), The Baluchistan earthquakes of August 25th and 27th, 1931,
Mem. Geol. Surv. India 67, 01-82.
West, W.D. (1937), Earthquakes in India. In: Proc. Indian Sci. Congress, Delhi,
Vol. 24, 189-224.

Received 2 March 2012

Received in revised form 16 May 2013
Accepted 20 May 2013