Journal of Asian Earth Sciences 91 (2014) 218–226

Contents lists available at ScienceDirect

Journal of Asian Earth Sciences

journal homepage: www.elsevier.com/locate/jseaes

A paleo-seismological study of the Dauki fault at Jaflong, Sylhet,

Bangladesh: Historical seismic events and an attempted rupture
segmentation model
Michio Morino a,⇑, A.S.M. Maksud Kamal b, S. Humayun Akhter b, Md. Zillur Rahman b,
Reshad Md. Ekram Ali c, Animesh Talukder c, Md. Mahmood Hossain Khan c, Jun Matsuo d, Fumio Kaneko d
a
International Short-Term Consultant of United Nations Development Programme (UNDP), 2-9-6 Hiromachida, Kure-shi, Hiroshima 737-0104, Japan
b
Faculty of Earth and Environmental Sciences, University of Dhaka, Dhaka 1000, Bangladesh
c
Geological Survey of Bangladesh, 153 Pioneer Road, Segunbagicha, Dhaka 1000, Bangladesh
d
OYO International Corporation, Tomin Nissay Kasugacho Bldg. 3F, 1-17 Koishikawa 1-Chome, Bunkyo-ku, Tokyo 112-0002, Japan

a r t i c l e i n f o a b s t r a c t

Article history: A paleo-seismological study was conducted at Jaflong, Sylhet, Bangladesh, which is on the eastern part of
Received 20 December 2013 the Dauki fault. The geomorphology around Jaflong is divided into the Shillong Plateau, the foothills, the
Received in revised form 24 May 2014 lower terraces, and the alluvial plain from north to south. Because the foothills and lower terraces are
Accepted 2 June 2014
considered to be uplifted tectonically, an active fault is inferred to the south of the lower terraces. This
Available online 12 June 2014
fault, which branches from the Dauki fault as a foreland migration, is known as the Jaflong fault in this
paper. The trench investigation was conducted at the southern edge of the lower terrace. The angular
Keywords:
unconformity accompanied by folding, which is thought to be the top of the growth strata, was identified
Dauki fault
Shillong Plateau
in the trench. An asymmetric anticline with a steep southern limb and gentle northern limb is inferred
Jaflong from the back-tilted lower terrace and the folding of the gravel layer parallel to the lower terrace surface.
Active fault The timing of the seismic event which formed the folding and unconformity is dated to between AD 840
Trench investigation and 920.
Radiocarbon dating The trench investigation at Gabrakhari, on the western part of the Dauki fault, revealed that the Dauki
fault ruptured in AD 1548 (Morino et al., 2011). Because the 1897 great Indian earthquake (M P 8.0;
Yeats et al., 1997) was caused by the rupture of the Dauki fault (Oldham, 1899), it is clear that the Dauki
fault has ruptured three times in the past one thousand years. The timing of these seismic events coin-
cides with that of the paleo-liquefactions confirmed on the Shillong Plateau. It is essential for the paleo-
seismological study of the Dauki fault to determine the surface ruptures of the 1897 earthquake. The
Dauki fault might be divided into four rupture segments, the western, central, eastern, and easternmost
segments. The eastern and western segments ruptured in AD 840–920 and in 1548, respectively. The
1897 earthquake might have been caused by the rupture of the central segment.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction et al., 1997). However, there is no evidence of surface ruptures

The Dauki fault which passes along the southern margin of the
Shillong Plateau is an E–W trending reverse fault inclined to the
north (Fig. 1a). The 1897 great Indian earthquake (M P 8.0; Yeats
et al., 1997) is one of the four great earthquakes that have occurred
in the Himalayan front over the past one hundred years (Yeats
et al., 1997; Ambraseys and Bilham, 2000). This high-intensity dis-
tribution suggests that the 1897 earthquake was caused by the
rupture of the Dauki fault (Oldham, 1899; Molnar, 1987; Yeats
⇑ Corresponding author. Tel.: +81 823 36 4610.
E-mail address: morino@oyointer.com (M. Morino).
on the Dauki fault (Yeats et al., 1997). The Sylhet Trough on the
south of the Shillong Plateau exhibits a remarkable negative Bou-
guer gravity anomaly of over 70 mgal, whereas the Shillong Pla-
teau shows a positive Bouguer gravity anomaly of over +20 mgal
(Rahman et al., 1990; Narula et al., 2000; Rajendran et al., 2004).
This finding indicates that the Sylhet Trough is subsiding through
the repeated ruptures of the Dauki fault, and the thick Quaternary
sediments that are presently accumulating. The Shillong Plateau
has a differential topographic profile with a southern steep slope
and a northern gentle slope (Fig. 1b). This finding is indicative of
the tectonic geomorphology of the Dauki fault (Morino et al.,
2011). This evidence suggests that the Dauki fault is active during

http://dx.doi.org/10.1016/j.jseaes.2014.06.002
1367-9120/Ó 2014 Elsevier Ltd. All rights reserved.
 M. Morino et al. / Journal of Asian Earth Sciences 91 (2014) 218–226 219

Fig. 1. (a) An active fault map of Bangladesh. The Dauki fault passes along the southern margin of the Shillong Plateau. The plate boundary mega-thrust between the Indian
and the Burman plate around Bangladesh is divided into the Tripura and the Arakan segments. The broken line represents its deformation front (DF). The fault traces of the
Arakan segment and the deformation front are after Maurin and Rangin (2009). The rectangle to the northwest of the Shillong Plateau represents the area of the paleo-
liquefaction study after Sukhija et al. (1999). The shaded background map is made using the elevation data of the Shuttle Radar Topography Mission (SRTM). Gb: trench site at
Gabrakhari (Morino et al., 2011), Jf: trench site at Jaflong, Ch: Chhatak, (b) A topographic profile of the Shillong Plateau across the trench site at Jaflong in the N–S direction.
The line A–B is shown in Fig. 1a.

the Late Quaternary time period. However, previous studies
suggested that the Dauki fault has not ruptured in the historical
past (Chen and Molnar, 1990; Kayal, 2001; Rajendran et al.,
2004), although no tectonic geomorphological or paleo-seismolog-
ical surveys were performed.
Morino et al. (2009, 2011) first identified the active Dauki fault
in the trench at Gabrakhari Village and suggested that the Dauki
fault ruptured in AD 1548, though it was a small secondary fault
that had branched off from the main fault. The 1548 earthquake,
which is the first large earthquake recorded in Bangladesh, shook
both Sylhet and Chittagong (http://www.banglapedia.org/HT/
E_0002.HTM, last accessed on 18 December 2013). Another opin-
ion is that the mega-thrust between the Indian and Burman plate,
the Tripura segment or deformation front (DF) shown in Fig. 1a,
generated the 1548 earthquake (Steckler et al., 2008). However,
if the mega-thrust was the seismic source of this earthquake,
Dhaka or nearby Sonargaon, the capital at the time, would have
received severe damage. However, there are no historical records
of the damages to Dhaka in 1548, although the records might be
unreliable. The timing of the seismic event at the Gabrakhari
trench site by Morino et al. (2011) is credible because the radiocar-
bon ages are based on samples of charcoal, peat, and wood
fragments that are proper for dating. However, Morino et al.
(2011) is also unable to explain the damages to Chittagong in
1548. Synthetic paleo-seismological studies of the Dauki fault
and the plate boundary mega-thrust are necessary to reveal the
seismic source of the 1548 earthquake.
It is essential to the paleo-seismological study of the Dauki fault
to find the surface ruptures of the 1897 earthquake. Although the
paleo-liquefactions in 1897 are evident around Gabrakhari, on
the western part of the Dauki fault (Morino et al., 2011), there is
no evidence of any surface ruptures in 1897. It is unclear whether
the Dauki fault around Gabrakhari did not ruptured in 1897, or
whether a subsurface rupture of the Dauki fault did not reach the
220 M. Morino et al. / Journal of Asian Earth Sciences 91 (2014) 218–226

Fig. 2. (a) Geomorphic classification and active faults around Jaflong. H: foothill, T1: higher terrace, T2: lower terrace. The arrows represent back-tilted surfaces. The solid line
and the broken lines represent an active fault and blind active faults, respectively. Two active faults, the Dauki fault and the Jaflong fault, are inferred around Jaflong. The
Jaflong fault is interpreted as a branched fault of the Dauki fault. This is indicative of the foreland migration of the Dauki fault. The cross on the X–Y line represents the
exposure of the gravel (Fig. 3c), (b) A topographic profile across the lower terrace that was measured by a simple water level and staff method. The shape is similar to that of
the Shillong Plateau (Fig. 1b). The X–Y line is shown in Fig. 2a.

surface. The former suggests that the Dauki fault is divided into
several rupture segments (Ferguson et al., 2012) with the rupture
segment of 1897 being different from that of 1548. The latter sug-
gests that the Dauki fault is basically blind (Steckler et al., 2012),
and the subsurface rupture does not always reach the surface. Near
Gabrakhari, the secondary fault strands in 1548 might have inci-
dentally appeared at the surface, while the subsurface rupture in
1897 may not have reached the surface. Is one or both of these
characterizations correct for the Dauki fault?
We conducted a paleo-seismological study of the Dauki fault at
Jaflong, Sylhet, Bangladesh, which is on the eastern part of the Dau-
ki fault. In this paper, we report the outcome of the trench investi-
gation at Jaflong, and discuss the timing of the historical seismic
events and the rupture segmentation model of the Dauki fault.
2. Active fault traces along the Dauki fault
We examined the stereo-paired satellite images from CORONA
(resolution: 7.5 m) and ALOS (resolution: 2.5 m) to detect the
active fault traces along the Dauki fault. The tectonic geomorphol-
ogy is not clear because Bangladesh is located in a tropical mon-
soon region with abundant rainfall and the fault topographies,
such as fault scarp and folding scarp, tend to be eroded by floods.
Although the fault scarps are occasionally inferred on the southern
edges of the uplifted terraces, for example, on the westernmost
part of the Dauki fault, it seems that the fault scarps are eroded
by floods and result in a continuous E–W running scarp, whereas
a reverse fault originally exhibits complicated fault scarps. The
downthrown sides on the footwall are the depositional area. The
sedimentation rate is fast and the surface ruptures are easily
buried. In the trench at Gabrakhari (Morino et al., 2011), the layer
deposited in about the 16th century on the footwall of the Dauki
fault lies at the bottom of a 2 m-deep trench. The sedimentation
rate is more than 4 mm/year. Additionally, Bangladesh is a densely
populated country and most of the land is cultivated for rice fields.
Because the geomorphology is severely modified by cultivation,
there are few places where tectonic geomorphologies are left in
their original condition. In addition, because the Dauki fault runs
along the border between India and Bangladesh, many places have
restricted entry.
Foothills and terraces have developed at the foot of the Shillong
Plateau. These geomorphologies are considered to be uplifted by
the activity of the Dauki fault, while the Sylhet Trough to the south
is subsiding (Fig. 1a). Therefore, the active Dauki fault is inferred
just to the south of the foothills and terraces. The satellite image
interpretation focused on the terraces. The terrace surfaces are
covered with dense bamboo and trees; therefore, it is difficult to
distinguish the fault scarps and the deformation of terrace
surfaces. The terraces were identified on the satellite images and
then the field survey targeting the terraces was performed. The
back-tilted terrace surfaces which are indicative of pressure ridges
were found at Gabrakhari (Morino et al., 2011) and Jaflong (Fig. 1a).
Some lower terraces were preserved in natural condition for resi-
dential space in villages, though most of the land was modified
by cultivation. The tectonic-geomorphic investigation along the
Dauki fault is to determine natural geomorphology such as those
of Gabrakhari and Jaflong. The active fault traces along the Dauki
fault shown in Fig. 1a are not based on distinct fault topographies.
The traces are inferred, although not clearly, immediately to the
south of the tectonically uplifted terraces and foothills.
 M. Morino et al. / Journal of Asian Earth Sciences 91 (2014) 218–226 221

3. Geomorphology in and around Jaflong

The geomorphology in and around Jaflong, which was inter-

preted by the stereo-paired ALOS images, is composed of the Shil-
long Plateau, the foothills, the terraces, and an alluvial plain from
north to south (Figs. 2a, 3a and b). The terrace is divided into the
lower terrace (T2) and the higher terrace (T1). The flat surface,
which is recognized at the top of the foothills, is the higher terrace
(T1). It might be a strath terrace because the deposits of T1 were
not confirmed in the field. The lower terrace (T2), which is 6- to 7-
m higher than the active alluvial plain, developed at the base of
the foothills (Fig. 2). The surface of T2 is overlain with a gravel layer
containing pebbles and cobbles of Precambrian metamorphic rocks
transported from the Shillong Plateau. The gravel layer was depos-
ited by the fluvial process of the Dauki River and uplifted to the pres-
ent height tectonically. The main stream of the Dauki River flows
southward and the tributary flows westward along the southern
edge of the Shillong Plateau. Judging from the distribution of T2 at
the base of the foothills, the old Dauki River flowed east- to south-
east-ward at that time. The terrace surfaces generally tilt down-
stream. The surface of T2 is tilted to the north around the trench
site (Fig. 2a and b) and is estimated to be a tectonic-geomorphology
called ‘‘back-tilted’’ or ‘‘pressure ridged’’ in technical terms.
The lower terraces on the eastern side of the trench site do not
show back-tilted surfaces. Most of the lower terraces were
removed to widen the rice fields or by the erosion from repeated
floods, and the natural geomorphologies no longer remain there.
Only the lower terrace, where the trench investigation was carried
out, was preserved in natural condition for the residential space of
Muslimnagar Village (Fig. 2b).
Two active faults are inferred around Jaflong (Fig. 2a). The
northern fault is inferred from the E–W trending straight cliff with
a topographic drop of over 500 m on the southern margin of the
Shillong Plateau. The southern fault is inferred to the south of
the lower terraces from the tectonic-geomorphology of the back-
tilted lower terrace and the monoclinal flexure confirmed by the
trench investigation as mentioned below. According to Alam
et al. (1990), the northern fault is considered to be the Dauki fault.
The southern fault, though it is blind, is a newly proposed active
fault. In this paper we temporarily call this new fault the ‘‘Jaflong
fault’’. The active fault traces of the Dauki fault become obscure
to the east, while the Jaflong fault continues eastward stepping
to the south (Fig. 1a).

4. Trench investigation at Jaflong

Two trenches at sites Nos. 1 and 2 were excavated on the south-
ern edge of the lower terrace at Jaflong (Figs. 2a and 3b). The size of
these trenches is 7–10 m long, 5 m wide, and 2 m deep. The log and
mosaic photos of the western wall at trench site No. 1 are shown in
Fig. 4. Trench No. 2 was excavated close to site No. 1 for a deeper
trench. However, we could not achieve a deeper trench because
the trench was underwater due to springs from the gravel layer
and occasionally heavy showers. Both trenches at sites Nos. 1
and 2 show the same stratigraphy and structure. In this chapter,
only the results from the trench at site No. 1 will be discussed.
Samples for radiocarbon dating were collected and analyzed in
both trenches at sites Nos. 1 and 2. The log of the eastern wall at
trench site No. 2 is shown in Fig. 6. The results of radiocarbon
dating are shown in Table 1.
4.1. Stratigraphy in the trench and its radiocarbon age
The stratigraphy in the trench is divided into units A to G which
are composed of fine-grained deposits such as clay, silt, and fine- to
Fig. 3. (a) A photograph showing the geomorphology around Jaflong. The view is
toward the north. The geomorphology is divided into three surfaces: the Shillong
Plateau, the foothills, and the lower terrace. The higher terrace is inferred at the top
of the foothills. (b) A close-up view of the lower terrace. The trench was excavated
at the place where the village people are standing. (c) The gravel layer overlying the
lower terrace with fresh rounded to sub-rounded pebbles and cobbles which are
composed of Precambrian metamorphic rocks carried in from the Shillong Plateau.
The location of the exposure is shown in Fig. 2a as a cross. The gravel layer is
thought to be a main stream deposit of the Dauki River.
medium-sand (Fig. 4). Units A to F are channel, back marsh, or
pond deposits left behind from the Dauki River. The Dauki River
is still transporting gravel from the Shillong Plateau, and the gravel
is currently collected and used as a building material. Unit G which
is the same as the gravel layer overlying the lower terrace was
deposited in the main stream of the old Dauki River.
Unit A: This unit contains blackish organic clay. The absence of
bedding suggests that this unit is a marsh deposit. At site No. 2 this
unit interfingers with unit B. The organic sediment (Jaf-5) has a
radiocarbon age of AD 1265–1385.
Unit B: Yellowish brown, medium-sized sand suggests small
channel deposits in unit B and unit C. These units cover units D
222 M. Morino et al. / Journal of Asian Earth Sciences 91 (2014) 218–226

Fig. 4. (a) A log of the western wall at Jaflong trench site No. 1. The stratigraphy in the trench is divided into units A to G. Unit G is the gravel layer overlying the lower terrace.
Units D to G are gently warped and covered by units A to C with an angular unconformity. The broken lines represent bedding. The solid triangles represent the locations of
radiocarbon dating samples. (b) A mosaic photograph of the western wall.

Table 1
14
The results of radiocarbon dating. All the samples were analyzed by beta analytic radiocarbon dating laboratory. conventional C ages were corrected by d13C using the Libby
half-life of 5568 years.

Trench Unit Sample Laboratory Method Material Measured d13C Conventional Calibrated Calendar
number number radiocarbon (permil) radiocarbon age (yBP) year (2 sigma)
age (yBP)
Jaflong A Jaf-5 Beta-319595 AMS Organic sediment 540 ± 30 16.3 680 ± 30 AD 1265 to 1385
B Jaf 2-5 Beta-319587 AMS Charred material organic sediment 1120 ± 30 20.9 1190 ± 30 AD 925 to 975
C Jaf 2-6 Beta-319588 AMS Charred material organic sediment 1130 ± 30 22.6 1170 ± 30 AD 915 to 965
Jaf-4 Beta-319594 AMS Charred material organic sediment 1030 ± 30 22.8 1070 ± 30 AD 902 to 953
Jaf 2-8 Beta-319590 AMS Charred material organic sediment 1050 ± 30 24.5 1060 ± 30 AD 895 to 932
Jaf 2-7 Beta-319589 AMS Charred material 1110 ± 30 26.7 1080 ± 30 AD 887 to 922
E Jaf-2 Beta-319592 AMS Charred material organic sediment 1160 ± 30 23.3 1190 ± 30 AD 817 to 894
Jaf 2-2 Beta-319584 AMS Organic sediment 1110 ± 30 19.7 1200 ± 30 AD 799 to 880
Jaf 2-4 Beta-319586 AMS Organic sediment 1180 ± 30 23.7 1200 ± 30 AD 784 to 863
Jaf 2-1 Beta-319583 AMS Wood 1090 ± 30 24.8 1090 ± 30 AD 777 to 843
F Jaf-3 Beta-319593 AMS Charred material 1150 ± 30 27.1 1120 ± 30 AD 771 to 828
Jaf-1 Beta-319591 AMS Charred material organic sediment 1330 ± 30 22.7 1370 ± 30 AD 608 to 689
Jaf 2-3 Beta-319585 AMS Organic sediment 1540 ± 30 19.8 1630 ± 30 AD 387 to 543

to G with an angular unconformity. The organic sediments includ-
ing the charcoals (Jaf2-5) have a radiocarbon age of AD 925–975.
Unit C: This unit contains alternation of dark gray silt and fine
gray sand. This unit buries the footwall of the folding scarp of units
D to G. The bedding plane is almost horizontal. The charcoal
(Jaf2-7) and the organic sediments including charcoals (Jaf-4,
Jaf2-6, and Jaf2-8) have a radiocarbon age of AD 880–930 and AD
895–965, respectively.
Unit D: This unit contains whitish-yellow silt partly alternating
with gray clay. This unit, along with units E and F, is a marsh or
pond deposit. The top of the layer is 5–10 cm thick, characterized
by a white color, and distinctly demarcated from units B and C
(Fig. 5a and b). There are no samples for radiocarbon dating.
Unit E: This unit contains an alternation of yellowish-brown silt
and gray clay finer than unit D. The wood fragments (Jaf2-1) in this
layer have a radiocarbon age of AD 770–850. The organic sedi-
ments (Jaf-2, Jaf2-2, and Jaf2-4) have a radiocarbon age of AD
780–900.
Unit F: This unit contains gray clay overlying the gravel of unit
G. The charcoal (Jaf-3) has a radiocarbon age of AD 770–830. The
organic sediments including charcoals (Jaf-1) and the organic
sediment (Jaf2-3) have radiocarbon ages of AD 600–690 and AD
 M. Morino et al. / Journal of Asian Earth Sciences 91 (2014) 218–226
Fig. 5. (a) A photograph of the western wall at Jaflong site No. 1. Unit D, which is
characterized by a whitish color, is warped. (b) A close-view of unit D. The red
threads represent a grid spacing of 1 m.
380–550, respectively. These organic sediments are older than the
charcoal.
Unit G: This unit contains gravel with rounded to sub-rounded
pebbles less than 5 cm in diameter. The pebbles are composed of
Precambrian metamorphic rocks. The matrix is filled with sandy
clay. This unit is the main deposit of the Dauki River because the
Dauki River is now transporting gravel from the Shillong Plateau.
This unit is inferred to be the Holocene deposit because the gravels
are fresh and not weathered. There is a path running southward
from the top of T2 along the X–Y line in Fig. 2a. The gravel layer
is exposed on the cutting of the path. The gravel layer at the top
of T2 continues southward beside the trench site. Unit G in the
trench is the same as the gravel layer at the top of T2. There are
no samples for radiocarbon dating.
4.2. Folding scarp identified in the trench
Units D to G are inclined to the south with a dip of 10°
between the W0 and W2 + 50 cm horizontal markers. The dip grad-
ually becomes steeper up to a maximum dip of 20° between
Fig. 6. A log of the eastern wall at Jaflong trench site No. 2.
223
W2 + 50 cm and W4 + 20 cm. To the south of W4 + 20 cm, the dip
is almost horizontal (Figs. 4 and 5). Units D, E, and F are composed
of fine deposits of silt and clay. Because these units were deposited
under the calm environment of a marsh or pond, the original sed-
imentary structure is almost horizontal. The structure of units D to
G dipping southwards indicates a tectonic deformation. The folding
scarp, the base of which is located at W4 + 20 cm, is inferred and
units A to C cover units D to G with an angular unconformity. In
the trench site at No. 2, units D to G are inclined to the south with
a dip of 10° between the E0 and E5 horizontal markers (Fig. 6). As
mentioned above, units D to G are warped and covered by the
unconformity of units A to C.
The southern slope of the lower terrace is inclined to the south
at an angle of 8° (Fig. 2b) which is almost the same as the dip of
units D to G in trench No. 1 between W0 and W2 + 50 cm horizon-
tal markers. The gravel layer overlying the surface of T2 (Fig. 3c)
continues up to unit G at the bottom of the trench. Therefore, it
is presumed that the geological structure near the ground surface
is parallel to the topographic profile of the lower terrace
(Fig. 2b). This finding is indicative of an asymmetric anticline with
a steep southern limb and a gentle northern limb.
The 1999 Chi–Chi, Taiwan, earthquake (Mw 7.6) was caused by
the rupture of the Chelungpu thrust fault. At the place where the
tip of the fault is buried in thick sediments, the rupture did not
reach the surface. Instead coseismic, wide surface deformations
were formed. Chen et al. (2007) called these surface deformations
fold scarps, distinguishing them from fault scarps. At Jaflong an
active fault was not identified in the trench where the folding
was identified. The southern slope of T2 is estimated to be a fold
scarp by Chen et al. (2007). If the buried tip of the Jaflong fault
was projected to the surface, it could reach the base of the folding
scarp.
4.3. Timing of the seismic event
The seismic event is inferred from the angular unconformity
between units C and D. Units B and C overlying the angular uncon-
formity show almost the same ages as units E and F. To limit the
timing of the seismic event, the analysis of OxCal v.4.2.3 (Bronk
Ramsey and Lee, 2013) using IntCal 13 (Reimer et al., 2013) was
practiced (Fig. 7). The calibrated calendar year of the seismic event
is dated to AD 840–920 with a probability of 2r.
4.4. Recurrence interval of the Jaflong fault
If the lower terrace was uplifted once to the present height by
the rupture of the Jaflong fault in AD 840–920, and the tip of the
fault reached near the surface at that time, the amount of vertical
displacement for a single event is inferred to be 7–8 m from a dif-
ference of altitude between the top of the lower terrace and the
bottom of the trench. According to GPS data across the southern
224 M. Morino et al. / Journal of Asian Earth Sciences 91 (2014) 218–226

Fig. 7. Calibrated age distributions of radiocarbon dating samples and the timing of the seismic event, calculated by OxCal v.4.2.3 (Bronk Ramsey and Lee, 2013) considering
deposit models. IntCal 13 (Reimer et al., 2013) is used for atmospheric curves. The light gray and dark gray areas show the prior and posterior probability distributions of
radiocarbon age. The lines show the range of 2r (95.4-percentile) confidence. The timing of the seismic event is dated to AD 840–920. The samples in bold italics represent
wood/charcoal for materials. Other samples are organic sediment/organic sediment including charcoals.

margin of the Shillong Plateau, the shortening is 7 mm/year
(Steckler et al., 2012). If the Jalong fault has a dip of 45° (Bilham
and England, 2001), the vertical velocity is estimated to be
7 mm/year. The recurrence interval of the Jaflong fault is inferred
to be about 1000 years from the vertical displacement and velocity.
If the lower terrace was uplifted twice by the ruptures of the
Jaflong fault, the recurrence interval of the Jaflong fault is about
500 years.
5. Discussion
5.1. The historical seismic events of the Dauki fault
Although the trench investigation across the Dauki fault has
been performed only at the two sites of Gabrakhari and Jaflong,
two seismic events were identified over the past one thousand
years. At the Gabrakhari trench site, on the western part of the
Dauki fault, the AD 1500–1630 seismic event was inferred, consis-
tent with the 1548 earthquake that is the first large earthquake
recorded in Bangladesh (Morino et al., 2011). The Dauki fault is
thought to have ruptured in 9–10th century because the AD
840–920 seismic event is inferred at Jaflong trench site, on the
eastern part of the Dauki fault. Although the surface ruptures dur-
ing the 1897 great India earthquake are not confirmed, the seismic
source of this earthquake is most likely the Dauki fault, judging by
the hat-shaped high-intensity distribution, the base of which is
located on the Dauki fault (Oldham, 1899; Molnar, 1987).
According to the studies of paleo-liquefaction along the Krish-
nai and Dudnai Rivers running to the north on the Shillong Plateau
(Fig. 1a), two paleo-liquefactions are reported during AD 1450–
1650 and AD 700–1050 except AD 1897 (Sukhija et al., 1999;
Rajendran et al., 2004). The timing of these paleo-liquefactions
coincides with that of the paleo-earthquakes identified by the
trench investigations across the Dauki fault at Gabrakhari and Jaf-
long. These paleo-liquefactions may have been caused by the rup-
tures of the Himalayan Front fault and nearby active faults on the
Shillong Plateau other than the Dauki fault. Bilham and England
(2001) suggested that the south-dipping ‘‘Oldham fault’’ was the
seismic source of the 1897 great Indian earthquake. However,
there is no remarkable tectonic geomorphology on the Shillong
Plateau (Morino et al., 2011). The 1934 Bihar–Nepal earthquake
(Mw 8.1) and the 1950 Assam earthquake (Mw 8.4) occurred
along the Himalayan front to the north of the Shillong Plateau
(inset in Fig. 1a).The paleo-seismological studies between the
rupture areas of the 1934 and 1950 earthquakes revealed that
the great earthquake occurred in AD 1100 (Kumar et al., 2010;
inset in Fig. 1a). However, the timing of these earthquakes does
not coincide with the timing of the paleo-liquefactions on the
Shillong Plateau. The north-dipping Dauki fault extends beneath
the Shillong Plateau. It is very likely that the Shillong Plateau has
been widely shaken by the ruptures of the Dauki fault.
 M. Morino et al. / Journal of Asian Earth Sciences 91 (2014) 218–226
If these earthquakes in AD 840–920, 1548, and 1897 were pro-
duced by the ruptures of the entire Dauki fault, the recurrence
interval is estimated to be 350–700 years. GPS data across the
Dauki fault suggests its shortening by 7 mm/year (Steckler
et al., 2012). Because the Dauki fault dips northward at approxi-
mately 45° (Bilham and England, 2001), the slip rate along the fault
is estimated to be 10 mm/year. If the recurrence interval of the
Dauki fault is 350–700 years, an average slip of 3.5–7.0 m of
the fault would be expected. Although the fault length is uncertain
because the eastern end is unclear, if the fault length and depth are
260 km and 35 km, respectively, the fault dip is 45°, the average
displacement is 3.5 m, and a shear modulus is 3  1011 dyne/cm2,
a moment magnitude of Mw 8.1 is extracted from the seismic
moment M0, based on the methods of Hanks and Kanamori
(1979). According to the empirical relationship between the mag-
nitude (Mw) and the surface rupture length (SRL), based on the
methods of Wells and Coppersmith (1994), a moment magnitude
of Mw 7.9 is expected.
5.2. Surface ruptures of the 1897 earthquake
A paleo-seismological study of the Dauki fault is essential to
find the surface ruptures of the 1897 great Indian earthquake.
Two reasons are proposed as briefly mentioned in the introduction.
The first reason is that the Dauki fault is divided into several
rupture segments, and each segment would have ruptured at a dif-
ferent time. The second reason is that the Dauki fault is basically
blind and the subsurface ruptures and deformations might not
always reach the surface.
In the first case, a segmentation model is suggested as follows.
The fault trace oversteps northward around Gabrakhari and Chha-
tak at 3 km wide (Gb and CH in Fig. 8). Additionally, at northeast
of Sylhet (Sy in Fig. 8), the foothills migrate southward and the
fault trace is assumed to overstep southward at 10 km wide
(Fig. 8). If these oversteps (based on our satellite photo interpreta-
tion) indicate segment boundaries, the Dauki fault could be divided
into four segments: west (W), central (C), east (E), and easternmost
(EM) (Fig. 8). Our trench investigations reveal that the west and
east segments ruptured in AD 1548 and AD 840–920, respectively.
The 1897 earthquake might be attributed to the rupture of the
central segment, though no paleo-seismological studies have been
carried out until the current study. In this case, the length of the
central segment is approximately 130 km, half of the length of
the entire Dauki fault. If the magnitude is estimated from the
empirical relationship between Mw and SRL, according to the
methodology of Wells and Coppersmith (1994), a Mw of 7.6 is
expected.
Fig. 8. A segmentation model of the Dauki fault. W: western segment, C: central
segment, E: eastern segment, EM: easternmost segment. The thin solid lines
represent Bouguer gravity anomaly contours at intervals of 10 mgal which are
copied from Fig. 4a after Rajendran et al. (2004). The Dapsi fault is after Ferguson
et al. (2012).
225
In the second case, the characteristics of the Dauki fault might be
typically exhibited at Jaflong where the Jaflong fault is blind and an
asymmetric anticline has developed on the hanging wall of the
fault. At Jaflong the seismic event horizons might be shown as
angular unconformities, which represent growth strata. The small
fault strands in the trench at Gabrakhari might have unexpectedly
appeared on the surface. The surface rupture or deformation in
1897 may have appeared farther south at Gabrakhari and Jaflong.
But there is no tectonic geomorphology in the southern area.
Ultimately we propose that the Dauki fault is divided into four
rupture segments, as shown in Fig. 8 because the timing of the lat-
est rupture event at Jaflong is different from that at Gabrakhari. To
prove this rupture segmentation model, the paleo-seismological
study of the central segment is indispensable. Also, the seismic
reflection survey and drilling survey, e.g. at Jaflong, is necessary
to reveal the subsurface geological structure across the Jaflong fault.
6. Summary
(1) The back-tilted lower terrace, which is estimated to be a tec-
tonic-geomorphology, was found at Jaflong on the eastern
part of the Dauki fault.
(2) A folding scarp was confirmed in the trench at Jaflong. The
timing of the seismic event that formed the folding scarp
is dated to AD 840–920.
(3) The asymmetric anticline near the ground surface is inferred
from the topographic profile across the lower terrace and the
distribution of the gravel layer. The folding scarp covered by
the horizontal layers with an angular unconformity in the
trench is interpreted to be the top of the growth strata.
(4) The timing of paleo-liquefactions which was confirmed on
the Shillong Plateau coincides with that of the seismic
events of the Dauki fault. The paleo-liquefactions on the
Shillong Plateau are most likely to have been caused by
the ruptures of the Dauki fault.
(5) Three seismic events in AD 840–920, 1548, and 1897
occurred on the Dauki fault. If these paleo-earthquakes are
attributed to ruptures of the entire Dauki fault, the recur-
rence interval is estimated to be 350–700 years.
(6) The Dauki fault might be divided into four rupture segments
of western, central, eastern, and easternmost portions based
on the fault geometry of overstepping. The seismic events
from AD 840–920 and in 1548 are inferred on the eastern
and the western segments, respectively. The 1897 great
Indian earthquake might have been caused by the rupture
of the central segment, although there is no paleo-seismo-
logical investigation of the central segment. If the lower ter-
race at Jaflong was uplifted once to the present height by the
rupture of the Jaflong fault in AD 840–920, the recurrence
interval of the Jaflong fault (the eastern segment) might be
about 1000 years.
(7) At Jaflong, the northern Dauki fault might be exposed on the
surface. But, the fault is not very active because it does not
continue eastward. The trench investigation at Jaflong sug-
gests that the southern Jaflong fault is more active than
the northern Dauki fault. However, it is basically blind, and
an asymmetric anticline and growth strata have developed
near the ground surface.
Acknowledgments
The authors are appreciative of the Comprehensive Disaster
Management Programme phase II (CDMP II) of the Ministry for
Food and Disaster Management (MoFDM) of the Government of
Bangladesh for the permission to publish this paper. CDMP II is
226 M. Morino et al. / Journal of Asian Earth Sciences 91 (2014) 218–226
supported by UNDP, EC, and DFID. The authors appreciate Dr. Kerry
Sieh, of the Earth Observatory of Singapore (EOS) for valuable sug-
gestions and discussions. The authors appreciate Dr. Koji Okumura,
of Hiroshima University, for helping with the Oxcal analysis.
Finally, thanks are offered to the anonymous reviewers for valuable
comments and suggestions.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jse-
aes.2014.06.002. These data include Google maps of the most
important areas described in this article.
References
Alam, M.K., Hasan, A.K.M.S., Khan, M.R. Whitney, J.M., 1990. 1: 1000,000 Geological
map of Bangladesh.
Ambraseys, N., Bilham, R., 2000. A note on the Kangra Ms = 7.8 earthquake of 4 April
1905. Curr. Sci. 79, 101–106.
Bilham, R., England, P., 2001. Plateau pop-up during the great 1897 Assam
earthquake. Nature 410, 806–809.
Bronk Ramsey, C., Lee, S., 2013. Recent and planned developments of the program
OxCal. Radiocarbon 55 (2–3), 720–730.
Chen, W.-P., Molnar, P., 1990. Source parameters of earthquakes and intraplate
deformation beneath the Shillong Plateau and the northern Indoburman ranges.
J. Geophys. Res. 95, 12,527–12,551.
Chen, Y-G., Lai, K-Y., Lee, Y-H., Suppe, J., Chen, W-S., Lin, Y-N.N., Wang, Y., Hung, J-H.,
Kuo, Y.-T., 2007. Coseismic fold scarps and their kinematic behavior in the 1999
Chi–Chi earthquake Taiwan. J. Geophys. Res. 112, B03S02. http://dx.doi.org/
10.1029/2006JB004388.
Ferguson, E.K., Seeber, L., Steckler, M.S., Akhter, S.H., Mondal, D., Lenhart, A., 2012.
The Dauki Thrust Fault and the Shillong Anticline: an Incipient plate boundary
in NE India. 2012 American Geophysical Union Fall Meeting, T51F-2665.
Hanks, T.C., Kanamori, H., 1979. A moment-magnitude scale. J. Geophys. Res. 84,
2348–2350.
Kayal, J.R., 2001. Microearthquake activity in some parts of the Himalaya and the
tectonic model. Tectonophysics 339, 331–351.
Kumar, S., Wesnousky, S.G., Jayangondaperumal, R., Nakata, T., Kumahara, Y., Singh,
V., 2010. Paleoseismological evidence of surface faulting along the northeastern
Himalayan front, India: timing, size, and spatial extent of great earthquakes. J.
Geophys. Res. 115, B12422. http://dx.doi.org/10.1029/2009JB006789.
Maurin, T., Rangin, C., 2009. Structure and kinematics of the Indo-Burmese Wedge:
recent and fast growth of the outer wedge. Tectonics 28, TC2010. http://
dx.doi.org/10.1029/2008TC002276.
Molnar, P., 1987. The distribution of intensity associated with the great 1897 Assam
earthquake and bounds on the extent of the rupture zone. J. Geol. Soc. India 30,
13–27.
Morino, M., Kamal, A.S.M.M., Muslim, D., Ali, R.Md.E., Kamal, M.A., Kaneko, F., 2009.
Activity of Dauki Fault during the 1897 Ms 8.0 Great Assam earthquake
confirmed by Trench Investigation at Gabrakahri Village, Haluaghat,
Mymensingh, Bangladesh. J. Geol. Soc. Jpn 115, Pictorial 11–12.
Morino, M., Kamal, A.S.M.M., Muslim, D., Ali, R.Md.E., Kamal, M.A., Rahman, Md.Z.,
Kaneko, F., 2011. Seismic event of the Dauki Fault in 16th Century Confirmed by
Trench Investigation at Gabrakhari Village, Haluaghat, Mymensingh,
Bangladesh. J. Asian Earth Sci. 42, 492–498. http://dx.doi.org/10.1016/j.jseaes.
2011.05.002.
Narula, R.L., Acharrya, S.K., Banerjee, J. (Eds.), 2000. Seismotectonic atlas of India
and its environs. Geological Survey of India, Kolkats, 43p.
Oldham, R.D., 1899. Report of the great earthquake of 12th June 1897. Memoirs of
the Geological Survey of India 29, 379p (Reprinted by Geological Survey of
India, Calcutta, 1981).
Rahman, M.A., Mannan, M.A., Blank, H.R., Kleinkopf, M.D., Kucks, R.P., 1990. Bouguer
gravity anomaly map of Bangladesh. Geological survey of Bangladesh.
Rajendran, C.P., Rajendran, K., Duarah, B.P., Baruah, S., Earnest, A., 2004. Interpreting
the style of faulting and paleoseismicity associated with the 1897 Shillong,
northeast India, earthquake: implications for regional tectonism. Tectonics 23,
TC4009. http://dx.doi.org/10.1029/2003TC001605.
Reimer, P.J., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Bronk Ramsey, C.,
Grootes, P.M., Guilderson, T.P., Haflidason, H., Hajdas, I., Hatt, C., Heaton, T.J.,
Hoffmann, D.L., Hogg, A.G., Hughen, K.A., Kaiser, K.F., Kromer, B., Manning, S.W.,
Niu, M., Reimer, R.W., Richards, D.A., Scott, E.M., Southon, J.R., Staff, R.A., Turney,
C.S.M., van der Plicht, J., 2013. IntCal13 and Marine13 radiocarbon age
calibration curves 0–50,000 years cal BP. Radiocarbon 55 (4), 1869–1887.
Steckler, M.S., Akhter, S.H., Seeber, L., 2008. Collision of the Ganges-Brahmaputra
Delta with the Burma Arc: implications for earthquake hazard. Earth Planet. Sci.
Lett. 273, 367–378.
Steckler, M.S., Akhter, S.H., Seeber, L., Bilham, R.G., Kogan, M.G., Masson, F., Maurin,
T., Mondal, D., Piana Agostinetti, N., Rangin, C., Saha, P., 2012. GPS Velocities and
Structure across the Burma Accretionary Prism and Shillong Anticline in
Bangladesh. 2012 American Geophysical Union Fall Meeting, T51F-2667.
Sukhija, B.S., Rao, M.N., Reddy, D.V., Nagabhushanam, P., Hussain, S., Chadha, R.K.,
Gupta, H.K., 1999. Timing and return period of major paleoseismic events in the
Shillong Plateau, India. Tectonophysics 308, 53–65.
Wells, D.L., Coppersmith, K.J., 1994. New empirical relationships among magnitude,
rupture length, rupture width, rupture area, and surface displacement. Bull.
Seismol. Soc. Am. 84 (4), 974–1002.
Yeats, R.S., Sieh, K., Allen, C.R., 1997. The geology of earthquakes. Oxford University
Press, 568p.