Directory UMM :Data Elmu:jurnal:J-a:Journal of Asian Earth Science:Vol18.Issue5.2000:
Journal of Asian Earth Sciences 18 (2000) 547±559
Fault systems and Paleo-stress tensors in the Indus Suture Zone
(NW Pakistan)
G. Zeilinger a,*, J.-P. Burg a, N. Chaudhry b, H. Dawood c, S. Hussain c
a
Geologisches Institut, ETH-Zentrum, Sonneggstr. 5, CH-8092 ZuÈrich, Switzerland
Institute of Geology, Punjab University, Quaid-e-Azam Campus, Lahore 54590, Pakistan
c
Museum of Natural History, Garden Avenue, Shakaraparian, Islamabad 44000, Pakistan
b
Accepted 12 October 1999
Abstract
Analysis of fault-striations measured in the Kohistan part of the Indus Suture Zone (NW Himalaya, Pakistan) has been
carried out to document dynamic evolution during the brittle stage of the collision of India and Asia. Processing of the data
with a direct inversion method identi®ed four stress ®elds which were chronologically ordered from ®eld evidence as SSE±NNW
compression, E±W compression, radial extension and SSW±NNE compression. The last corresponds to the present-day stress
®eld de®ned from seismic activity. The earlier stress ®elds are related to times during the Miocene, when convergence-related
stresses were disturbed by the formation of the nearby Nanga Parbat and Indus syntaxes. 7 2000 Elsevier Science Ltd. All
rights reserved.
1. Introduction
In NW Pakistan, the Kohistan Complex developed
as an island arc above the northward subducted Tethyan lithosphere during the Mesozoic (Bard, 1983;
Coward et al., 1986). The southern boundary of the
Kohistan Complex is the Indus Suture (the Main Mantle Thrust=MMT of Tahirkheli et al., 1979), which is
a crustal-scale, 35±508 northward dipping fault contact
(Malinconico, 1989), along which the Kohistan Complex has been thrust over India (Fig. 1). Collision
between India and Kohistan began at about 65 Ma
(Beck et al., 1996), and continued with subsequent
obduction onto India (Coward et al., 1987). The synand late-collisional history of the Indus Suture includes
southward directed thrusting and extension (Treloar et
al., 1991; Burg et al., 1996). Fission track ages (Zeitler
* Corresponding author. Tel.: +41-1-632-3690; fax: +41-1-6321030.
E-mail address: gerold@erdw.ethz.ch (G. Zeilinger).
et al., 1982) record the termination of signi®cant dierential movement across the Indus Suture at 15 Ma.
Ductile structures have been investigated by several
authors (e.g. Coward and Butler, 1985; Coward et al.,
1986, 1987; Treloar et al., 1990; Arbaret et al., 2000)
in order to understand early collisional processes.
However, faulting is an important part of the longlived deformation history that has produced the present day structure of the suture zone [Fig. 2(a)]. Some
areas are seismically active (Seeber and Armbruster,
1979) and it is evident in the ®eld that, despite the lack
of surface rupturing during particular earthquakes
(e.g. Pennington, 1979), rocks have been fractured.
This study concentrates on the analysis of fault-striations [Fig. 2(b) and 2(c)] in order to document the
dynamics of this part of the Indus Suture Zone during
hypercollision. The investigated area straddles the
Indus Suture and comprises three main units that are
from S to N, i.e. from the structurally lowest (Treloar
et al., 1996; Kazmi and Jan, 1997):
Ð The Indian unit that includes granodiorite and
intensely foliated and folded gneisses with a marked
1367-9120/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved.
PII: S 1 3 6 7 - 9 1 2 0 ( 9 9 ) 0 0 0 8 4 - X
548
G. Zeilinger et al. / Journal of Asian Earth Sciences 18 (2000) 547±559
hornblende-gabbros, diorites, norites, amphibolites and
small bodies of hornblendite (the Kamila Amphibolite
Belt of Jan, 1988) towards the top.
Ð The Chilas Complex which is composed of gabbro-norite and subordinate diorites (Jan et al., 1984).
2. Method
Fig. 1. Studied area (squared) located in the structural framework of
Northern Pakistan, with the Kohistan Complex, main sutures and
syntaxes.
SW trending stretching lineation (Coward et al.,
1988; Treloar et al., 1989).
Ð The lower Kohistan unit comprising ultrama®c
and ma®c rocks at the bottom, and mainly gabbros,
Approximately 650 measurements of fault/striation
pairs from 30 sites form our data set. Outcrops de®ned
as sites are sections not longer than 100±200 m. The
size of faults, quality of criteria for sense of movement
and faulting style (e.g. discrete plane vs gouge zone)
were also noted in order to characterize homogeneous
fault populations. In particular, we distinguished faults
with new crystallization (chlorite, apatite etc.) from
``dry'' faults with striated slip planes and without new
crystallization, indicating that ``dry'' faults were
formed at lower temperature, closer to the surface, and
therefore later than the crystal coated planes. Conju-
Fig. 2. (a) Fault zone in meta-gabbro, 4 km NE of Duber Kale (locality in Fig. 4) (circled cow as scale). Slip plane: 305/36. This fault shows
sinistral movement and is attributed to population 4. (b) Normal fault (slip plane 129/88) with serpentine ®bres (046/58) indicating the downward
relative movement of the hangingwall block (4 km N of Duber Bazaar located in Fig. 4). (c) Same fault plane as 1b with several ®bre orientations: 046/58 and 050/21; ®bre continuity indicates a relatively short change in movement direction. For fault population distinction only the
principal ®bre direction was used, ®tting population 3. (d) Superposed striations on one fault plane (slip plane 290/18), the older one (lower
arrow, 274/17, population 2) is cut by the younger one (upper arrow, 011/03, population 4).
G. Zeilinger et al. / Journal of Asian Earth Sciences 18 (2000) 547±559
549
Fig. 3. Data processing illustrated with data set 29/08 (6 km NE of Patan along the Karakorum Highway, length of section 150 m). This site is
part of the ``regional data set'', a combined set of measurements belonging to an area where faulting can be assumed to be homogeneous. Once
the successive, preliminary regional tensors have been calculated from the ``regional data set'', each site was ®ltered through these stress tensors
(mis®t angle R 408). In addition, data sets from each site were separated without employing the regional stress tensors. Both resulting subsets
were compared and rearranged to populations with coherent s1 directions. These separated faults were processed again with a random tensor
search to obtain local deviations in terms of shape and orientation of the local stress tensor. Populations of less than 5 faults and unexplained
data were considered to be not signi®cant. Therefore they are not shown in Figs. 5±8.
550
G. Zeilinger et al. / Journal of Asian Earth Sciences 18 (2000) 547±559
Fig. 4. Unseparated data from 30 sites. Sites used for calculation of the regional stress tensors are not separated by major faults and they are
clustered in one, reasonably small geographic area (marked by dashed line) in a single rock unit (a massive metagabbro). Fault/striation plotted
in stereo-net, lower hemisphere (fault plane: great circle, striation: arrow, pointing towards movement direction of the hanging block).
G. Zeilinger et al. / Journal of Asian Earth Sciences 18 (2000) 547±559
gate fault sets were used in the ®eld for a ®rst estimate
of the bulk shortening (compression) direction giving
evidence of separate faulting events. Superposed striations on single fault planes further supported the
chronological classi®cation [Fig. 2(d)]. These ®eld observations, which yielded homogeneous data sets, were
the key for computer aided fault distinction and stress
tensor calculations. A ®rst computer analysis of singlesite data consistently required more than one stress
tensor to explain the ®eld measurements. Consequently, fault striation data have been separated for
every site to determine successive stress tensors (Fig. 3).
Data processing was carried out with Software FSA
written by B. CeÂleÂrier, which is based on a direct inversion algorithm (Bott, 1959; Compton, 1966; Etchecopar et al., 1981). Assumptions of the method are: (1)
direction of the resolved shear stress on the fault is
parallel to the measured striation and records the slip
direction on the fault; (2) slip on each fault is independent from other new or reactivated faults; and (3) the
stress directions are homogenous on the scale of a site.
To ascertain regional stress ®elds, 248 measurements
within a single rock unit, a metagabbro of Lower
Kohistan, near Patan, were processed as one data set
(Fig. 4). The aim was to eliminate small, local deviations and to integrate over a large area the bulk
brittle tectonics. In the ®rst processing step, tensors of
maximum (s1), intermediate (s2) and minimum (s3)
principal stresses (positive in compression) best ®tting
30% of the measurements were randomly calculated.
This random tensor search is a Monte Carlo approach
to search a stress tensor that best explain the slip
directions. This mainly follows the method proposed
in Etchecopar et al. (1981) and Etchecopar (1984). In
a ®rst step of the random tensor search, the stress tensor data are generated by using a random variable, so
that the orientations are uniformly distributed in
space. In a second step, for each tensor, the angular
mis®t between the predicted and observed slip direction for each fault slip data is computed. In a third
step, for each tensor, an average angular mis®t, is
computed. An adjustable percentage of the fault slip
data must be explained. The angular mis®t is the average of the angular mis®t for the chosen percentage of
the data that have the lowest angular mis®t. In a
fourth step the tensors are ranked by increasing value
of the angular mis®t and only the ®rst tensors are
retained. These stress tensors are the most compatible
with the fault data set.
The best ®tting tensor integrated the highest number
of mis®t angles
Fault systems and Paleo-stress tensors in the Indus Suture Zone
(NW Pakistan)
G. Zeilinger a,*, J.-P. Burg a, N. Chaudhry b, H. Dawood c, S. Hussain c
a
Geologisches Institut, ETH-Zentrum, Sonneggstr. 5, CH-8092 ZuÈrich, Switzerland
Institute of Geology, Punjab University, Quaid-e-Azam Campus, Lahore 54590, Pakistan
c
Museum of Natural History, Garden Avenue, Shakaraparian, Islamabad 44000, Pakistan
b
Accepted 12 October 1999
Abstract
Analysis of fault-striations measured in the Kohistan part of the Indus Suture Zone (NW Himalaya, Pakistan) has been
carried out to document dynamic evolution during the brittle stage of the collision of India and Asia. Processing of the data
with a direct inversion method identi®ed four stress ®elds which were chronologically ordered from ®eld evidence as SSE±NNW
compression, E±W compression, radial extension and SSW±NNE compression. The last corresponds to the present-day stress
®eld de®ned from seismic activity. The earlier stress ®elds are related to times during the Miocene, when convergence-related
stresses were disturbed by the formation of the nearby Nanga Parbat and Indus syntaxes. 7 2000 Elsevier Science Ltd. All
rights reserved.
1. Introduction
In NW Pakistan, the Kohistan Complex developed
as an island arc above the northward subducted Tethyan lithosphere during the Mesozoic (Bard, 1983;
Coward et al., 1986). The southern boundary of the
Kohistan Complex is the Indus Suture (the Main Mantle Thrust=MMT of Tahirkheli et al., 1979), which is
a crustal-scale, 35±508 northward dipping fault contact
(Malinconico, 1989), along which the Kohistan Complex has been thrust over India (Fig. 1). Collision
between India and Kohistan began at about 65 Ma
(Beck et al., 1996), and continued with subsequent
obduction onto India (Coward et al., 1987). The synand late-collisional history of the Indus Suture includes
southward directed thrusting and extension (Treloar et
al., 1991; Burg et al., 1996). Fission track ages (Zeitler
* Corresponding author. Tel.: +41-1-632-3690; fax: +41-1-6321030.
E-mail address: gerold@erdw.ethz.ch (G. Zeilinger).
et al., 1982) record the termination of signi®cant dierential movement across the Indus Suture at 15 Ma.
Ductile structures have been investigated by several
authors (e.g. Coward and Butler, 1985; Coward et al.,
1986, 1987; Treloar et al., 1990; Arbaret et al., 2000)
in order to understand early collisional processes.
However, faulting is an important part of the longlived deformation history that has produced the present day structure of the suture zone [Fig. 2(a)]. Some
areas are seismically active (Seeber and Armbruster,
1979) and it is evident in the ®eld that, despite the lack
of surface rupturing during particular earthquakes
(e.g. Pennington, 1979), rocks have been fractured.
This study concentrates on the analysis of fault-striations [Fig. 2(b) and 2(c)] in order to document the
dynamics of this part of the Indus Suture Zone during
hypercollision. The investigated area straddles the
Indus Suture and comprises three main units that are
from S to N, i.e. from the structurally lowest (Treloar
et al., 1996; Kazmi and Jan, 1997):
Ð The Indian unit that includes granodiorite and
intensely foliated and folded gneisses with a marked
1367-9120/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved.
PII: S 1 3 6 7 - 9 1 2 0 ( 9 9 ) 0 0 0 8 4 - X
548
G. Zeilinger et al. / Journal of Asian Earth Sciences 18 (2000) 547±559
hornblende-gabbros, diorites, norites, amphibolites and
small bodies of hornblendite (the Kamila Amphibolite
Belt of Jan, 1988) towards the top.
Ð The Chilas Complex which is composed of gabbro-norite and subordinate diorites (Jan et al., 1984).
2. Method
Fig. 1. Studied area (squared) located in the structural framework of
Northern Pakistan, with the Kohistan Complex, main sutures and
syntaxes.
SW trending stretching lineation (Coward et al.,
1988; Treloar et al., 1989).
Ð The lower Kohistan unit comprising ultrama®c
and ma®c rocks at the bottom, and mainly gabbros,
Approximately 650 measurements of fault/striation
pairs from 30 sites form our data set. Outcrops de®ned
as sites are sections not longer than 100±200 m. The
size of faults, quality of criteria for sense of movement
and faulting style (e.g. discrete plane vs gouge zone)
were also noted in order to characterize homogeneous
fault populations. In particular, we distinguished faults
with new crystallization (chlorite, apatite etc.) from
``dry'' faults with striated slip planes and without new
crystallization, indicating that ``dry'' faults were
formed at lower temperature, closer to the surface, and
therefore later than the crystal coated planes. Conju-
Fig. 2. (a) Fault zone in meta-gabbro, 4 km NE of Duber Kale (locality in Fig. 4) (circled cow as scale). Slip plane: 305/36. This fault shows
sinistral movement and is attributed to population 4. (b) Normal fault (slip plane 129/88) with serpentine ®bres (046/58) indicating the downward
relative movement of the hangingwall block (4 km N of Duber Bazaar located in Fig. 4). (c) Same fault plane as 1b with several ®bre orientations: 046/58 and 050/21; ®bre continuity indicates a relatively short change in movement direction. For fault population distinction only the
principal ®bre direction was used, ®tting population 3. (d) Superposed striations on one fault plane (slip plane 290/18), the older one (lower
arrow, 274/17, population 2) is cut by the younger one (upper arrow, 011/03, population 4).
G. Zeilinger et al. / Journal of Asian Earth Sciences 18 (2000) 547±559
549
Fig. 3. Data processing illustrated with data set 29/08 (6 km NE of Patan along the Karakorum Highway, length of section 150 m). This site is
part of the ``regional data set'', a combined set of measurements belonging to an area where faulting can be assumed to be homogeneous. Once
the successive, preliminary regional tensors have been calculated from the ``regional data set'', each site was ®ltered through these stress tensors
(mis®t angle R 408). In addition, data sets from each site were separated without employing the regional stress tensors. Both resulting subsets
were compared and rearranged to populations with coherent s1 directions. These separated faults were processed again with a random tensor
search to obtain local deviations in terms of shape and orientation of the local stress tensor. Populations of less than 5 faults and unexplained
data were considered to be not signi®cant. Therefore they are not shown in Figs. 5±8.
550
G. Zeilinger et al. / Journal of Asian Earth Sciences 18 (2000) 547±559
Fig. 4. Unseparated data from 30 sites. Sites used for calculation of the regional stress tensors are not separated by major faults and they are
clustered in one, reasonably small geographic area (marked by dashed line) in a single rock unit (a massive metagabbro). Fault/striation plotted
in stereo-net, lower hemisphere (fault plane: great circle, striation: arrow, pointing towards movement direction of the hanging block).
G. Zeilinger et al. / Journal of Asian Earth Sciences 18 (2000) 547±559
gate fault sets were used in the ®eld for a ®rst estimate
of the bulk shortening (compression) direction giving
evidence of separate faulting events. Superposed striations on single fault planes further supported the
chronological classi®cation [Fig. 2(d)]. These ®eld observations, which yielded homogeneous data sets, were
the key for computer aided fault distinction and stress
tensor calculations. A ®rst computer analysis of singlesite data consistently required more than one stress
tensor to explain the ®eld measurements. Consequently, fault striation data have been separated for
every site to determine successive stress tensors (Fig. 3).
Data processing was carried out with Software FSA
written by B. CeÂleÂrier, which is based on a direct inversion algorithm (Bott, 1959; Compton, 1966; Etchecopar et al., 1981). Assumptions of the method are: (1)
direction of the resolved shear stress on the fault is
parallel to the measured striation and records the slip
direction on the fault; (2) slip on each fault is independent from other new or reactivated faults; and (3) the
stress directions are homogenous on the scale of a site.
To ascertain regional stress ®elds, 248 measurements
within a single rock unit, a metagabbro of Lower
Kohistan, near Patan, were processed as one data set
(Fig. 4). The aim was to eliminate small, local deviations and to integrate over a large area the bulk
brittle tectonics. In the ®rst processing step, tensors of
maximum (s1), intermediate (s2) and minimum (s3)
principal stresses (positive in compression) best ®tting
30% of the measurements were randomly calculated.
This random tensor search is a Monte Carlo approach
to search a stress tensor that best explain the slip
directions. This mainly follows the method proposed
in Etchecopar et al. (1981) and Etchecopar (1984). In
a ®rst step of the random tensor search, the stress tensor data are generated by using a random variable, so
that the orientations are uniformly distributed in
space. In a second step, for each tensor, the angular
mis®t between the predicted and observed slip direction for each fault slip data is computed. In a third
step, for each tensor, an average angular mis®t, is
computed. An adjustable percentage of the fault slip
data must be explained. The angular mis®t is the average of the angular mis®t for the chosen percentage of
the data that have the lowest angular mis®t. In a
fourth step the tensors are ranked by increasing value
of the angular mis®t and only the ®rst tensors are
retained. These stress tensors are the most compatible
with the fault data set.
The best ®tting tensor integrated the highest number
of mis®t angles