Acta Geophysica

vol. 61, No. 4, Aug. 2013, pp. 923-934

DOI: 10.2478/s11600-013-0125-1

Seismicity and Reservoir

Induced Crustal Motion Study
around the Tehri Dam, India
Swapnamita CHOUDHURY, Param K. GAUTAM,
and Ajay PAUL

Wadia Institute of Himalayan Geology, Dehradun, India

e-mail: swapnamita.choudhury@gmail.com

Abstract
The Tehri Dam is located in a seismotectonically active region in
the Indian Himalayan belt. This 260.5 m high dam has a live water stor-
age of 2615 × 106 m3 and is capable of generating crustal deformation
corresponding to water fluctuation. Filling of the reservoir started in
October 2005. Seismic data around the dam between 2000 and 2010
shows that seismicity is corresponding to drawdown levels of the reser-
voir rather than to higher water levels. GPS data at twelve local bench-
marks were collected from 2006 to 2008 during filling and drawdown
reservoir levels. The velocity vectors show ground motion to be between
~0.69-1.50 mm in the different filling-drawdown cycles with reference to
the permanent station at Ghuttu. The motion appears to be inwards into
the reservoir when the reservoir is filled and outwards when the reservoir
is drained. This ground motion corresponds to elastic deformation and
rebound due to effect of the oscillating water levels.
Key words: space geodesy, seismicity, reservoir induced ground motion,
Tehri Dam (India).

1. INTRODUCTION
The Himalaya has given birth to several perennial rivers flowing from its
glaciers into the plains. The rivers have a high potential for hydropower gen-
eration apart from providing drinking and irrigation water. The Tehri project
is an irrigation and hydropower project, completed in the year 2005 and fill-
________________________________________________
© 2013 Institute of Geophysics, Polish Academy of Sciences
924 S. CHOUDHURY et al.

ing started immediately thereafter. The construction of the dam was associ-
ated with several concerns, especially following the magnitude M6.8 Uttar-
kashi earthquake of 20 October 1991 (Fig. 1). The region experiences
frequent seismicity. Subsequent to the Uttarkashi earthquake, the Chamoli
earthquake, M6.6, occurred on 29 March 1999 (Fig. 1). The concern for the
safety of the dam in case of a major earthquake was mitigated with measures
adopted for a safe dam structure. However, the concern for reservoir trig-
gered seismicity (RTS) still instils speculations. The earliest example of RTS
was reported from studies of induced earthquakes in the Lake Mead created
by the 220 m high Hoover Arch Dam (Carder 1945). In India, the Shivajisa-
gar Reservoir due to the Koyna Dam has been a significant example of per-
sistent triggered seismicity since the last 50 years (Gupta 1992, Talwani
1997). The incremental stress under a two dimensional artificial lake was
calculated by Gough (1969) and the implications of loading and unloading of
a reservoir on seismic activity were discussed by Gough and Gough
(1970a, b).

Fig. 1. Regional seismotectonics around the Tehri Dam, India. Seismicity is from the
WIHG Garhwal Seismic Network. The area shown in this figure is beyond the area
chosen for selection of seismic data for analysis. Figure also shows local faults
around the Tehri region: (1) Dewal Tear, (2) Gadolia Tear, and (3) Marh Tear. Ma-
jor tectonics features are shown after Valdiya (1980).
 SEISMICITY AND RESERVOIR INDUCED CRUSTAL MOTION 925

Fig. 2. Locations of the VSAT linked broadband seismic network established in

2007 by WIHG especially focusing around the Tehri Dam, India. Twelve GPS
benchmarks (inset 1) are the locations of campaign mode surveys. Inset 2 shows the
location of the study area in India.

The Wadia Institute of Himalayan Geology (WIHG), Dehradun (India),

established an eleven-station broadband seismic network around the dam in
the states of Uttarakhand and Himachal Pradesh (Fig. 2). In this study, appli-
cation of space geodesy like Global Positioning System (GPS) and
Interferometric Synthetic Aperture Radar (InSAR) (Fig. 2) has been initiated
to monitor crustal deformation due to the loading of the dam. InSAR did not
yield good results. We have also investigated probable influence of the load-
ing and unloading of the reservoir on the local seismicity of the region. Re-
sults of this investigation are discussed in this paper.

2. STUDY AREA
The dam site (~640 m a.s.l.) lies downstream of the confluence of the Bhagi-
rathi and Bhilangana Rivers. The reservoir with a total lake area of 42 km2
extends 44 km in the Bhagirathi Valley and 25 km in the Bhilangana Valley.
The Tehri Dam is located about 5 km south of the moderate to steep south
926 S. CHOUDHURY et al.

dipping North Almora Thrust (NAT) (Fig. 1). The NAT is seen to be dis-
placed for about 500 m by the NE-SW trending Dewal Tear which is a trav-
erse fault in the Bhilangana River Valley. The NW-SE trending Gadolia
Tear is a strike slip fault running parallel to the NAT in the same valley and
finally merging with the NAT. The NW-SE trending Tehri Tear cuts across
the Jalkur River (a tributary of the Bhagirathi River) and runs about 10 km
upstream of the Bhilangana River (Nawani et al. 1990). The E-W trending
Marh Tear is exposed about 4 km downstream from the dam. The Tehri-
Koteshwar Fault attributes a straight flow of the Bhagirathi River down-
stream from the dam to Koteshwar, where a 400 MW dam is soon nearing
completion.
The contribution to runoff in the Bhagirathi and Bhilangana Rivers in the
non-monsoon season is from snowmelt. The discharge in both rivers has
been monitored in gauzing stations at Uttarkashi (for Bhagirathi River) and
at Ghansali (for the Bhilangana River). The inflows of both rivers are seen to
vary between 30 and 1600 × 106 m3, the highest being observed on 20 Sep-
tember 2010. Figure 3 shows the complete per day water level history of the
Tehri reservoir from October 2005 to 2010. The minimum drawdown level
(MDDL) of the reservoir has been fixed at 740 m a.s.l. The water level histo-
ry has been used to correlate the seismic data around the dam.
The entire Himalayan belt is one of the most active seismotectonic belts
of the world, marked by numerous large/great earthquakes. The location of
the dam falls in an identified 700 km long seismic gap between the 1905
Kangra earthquake in the west and the 1934 Bihar-Nepal earthquake in the
east. Recent GPS studies report convergence at 15-20 mm/yr in the frontal
thrust region in a 100-150 km zone within the seismic gap region (Banerjee
and Burgmann 2002). Geomorphic evidences in the folded and thrusted Ter-
tiary Siwalik molasse southwest of the dam also show ~14 ± 2 mm/yr short-
ening rates across the Kangra re-entrant and 11 ± 5 mm/yr across the
Dehradun re-entrant (Powers et al. 1998). Microseismicity studies show that
the MCT is seismically inactive (Khattri et al. 1989). In the Garhwal Hima-
laya, earthquakes are of thrust faulting mechanism under a dominant com-
pressive regime (Khattri et al. 1989, Kayal 2001, Yadav 2009) with a few
strike-slip faulting corresponding to prominent transverse faults/lineaments
that cut across the Himalaya. Figure 1 also shows the seismicity (2007 to
2010 from the WIHG network) in the study area.

3. RESULTS AND DISCUSSION

In the region surrounding the Tehri Dam, seismicity is generated due to ac-
tive tectonics in consequence to the Himalayan orogeny. The loading of the
reservoir, however, may play a triggering role. Generally, it is perceived that
the load induces activation of crustal structures corresponding to a normal
 SEISMICITY AND RESERVOIR INDUCED CRUSTAL MOTION 927

fault system. The Himalayan orogen belongs to a compressive stress regime

along its linear belts. In such a scenario, induced loading should rather stabi-
lize any tendency for upthrusting of any tectonic block. If this holds true, the
stability of this region should be enhanced rather than inducing seismicity.
Other older dams in the Himalayan rivers, at Tarbela, Pong, Bhakra,
Ramganga, show no increase or induced seismicity (Gupta 1992). However,
diffusion of water may increase pore pressure in any environment. A general
tendency therefore will be an elastic response, rather with a coupled effect of
compaction of pore spaces due to loading and diffusion of water to depths
(Simpson et al. 1988). The increase of pore pressure at hypocentral depths
may lead to the activation/triggering of seismicity. The response of the effect
of loading on any fragile tectonics due to a reservoir may take place immedi-
ately following the filling or may be delayed until several filling-drawdown
cycles are completed. In the Tehri Dam the water level oscillates about 80-
100 m every filling-drawdown cycle from MDDL.
The seismic network established in 2007 by WIHG is located 50 km
from each other at Dehradun, Tapovan, Adibadri, Gaurikund, Ghuttu,
Deoban, Kotdwar, Chakrata, and Kharsali in the state of Uttarakhand, and
Nahan and Kotkhai in Himachal Pradesh, India (Fig. 2). Threshold limit of
1.5 has been attained with the data with a complete coverage around the
study area and the catalogue is complete for the region. Seismic data (M > 3)
from the ISC catalogue (www.isc.ac.uk) for the years 2000 to 2010 shows
that the initial period (2005-2007) of the reservoir is associated with lower
magnitude earthquakes and also reduced number of seismic events (Fig. 3).
During this period the reservoir completed its first cycle of filling from base
level at 638 to 784 m a.s.l. (maximum for that rainy season) and oscillated
back to MDDL in the lean months of 2007. Increased frequency of earth-
quake events in later periods in the oscillating reservoir, especially in 2009 is

Fig. 3. Earthquakes from 2000 to 2010 (data from ISC catalogue, M > 3) and from
2007 to 2010 (data from WIHG network, M > 1.5) in the region around the dam
along with water level history of the Tehri Dam from October 2005 (since dam fill-
ing) to 2010.
928 S. CHOUDHURY et al.

Fig. 4. Number of earthquake events per filling period (filling and drawdown cycles)
around the Tehri Reservoir, India. Seismicity appears to be more pronounced in the
drawdown periods.

observed (Fig. 3). WIHG network data shows that for all filling cycles be-
tween 2007 and 2010 seismicity appears to be more pronounced during low
reservoir water levels (December to May) rather than higher levels (Fig. 4).
The GPS data has been acquired in twelve benchmark stations (HNBG,
BHAL, MOTN, GWAD, OKHL, KANG, NEGI, RAZA, TIPR, KUTH,
JAAK, and JUDA – Fig. 2) on campaign mode between the years 2006 and
2008 twice a year, corresponding to the filling and drawdown cycles of the
Tehri reservoir (Fig. 5A-C). We have also established a permanent station at
Ghuttu located ~34 km (aerially) northeast from the dam site (Fig. 2). Top-
con GPS’s HiPer GGD receivers are used in the data acquisition.
The GPS raw data is acquired in TPS format. Conversion to RINEX
format was performed for further analysis through TPS2RIN and TEQC tool
(unavco, http://facility.unavco.org). The software GAMIT Version 10.32
(King and Bock 2002) has been used for next phase of processing and esti-
mation of three dimensional relative positions of ground stations and satellite
orbits. The primary output of GAMIT is a loosely constrained solution. In
 SEISMICITY AND RESERVOIR INDUCED CRUSTAL MOTION 929

Fig. 5. Water level data of the Tehri Reservoir from 2005 to 2010. GPS data was
collected in campaign mode in 2006, 2007, and 2008 corresponding to: (A) draining
period (October 2006 to July 2007), (B) while the reservoir was filled (July to No-
vember 2007), and again (C) when the reservoir was drained (November 2007 to
May 2008). Figure also shows GPS velocity vectors with respect to Ghuttu perma-
nent station. The motion of the ground was observed to be away from the reservoir
area (a and c) during drawdown period (A and C) and inwards (b) during filling pe-
riod (B).

addition to our campaign mode stations we used data from the International
Global Navigation Satellite System (GNSS) Service (IGS) stations, namely,
BAHR, TEHN, POL2, KIT3, WUHN, KUNM, IISC, HYDE, BAN2, and
DGAR stations. We obtained the time series data for our twelve sites to
check the component variations. Figure 6 shows the horizontal component
variations of the station JUDA in the different campaign surveys. The proc-
930 S. CHOUDHURY et al.

Fig. 6. Horizontal component variations of the station JUDA corresponding to dif-

ferent campaign surveys between 2006 and 2008.
 SEISMICITY AND RESERVOIR INDUCED CRUSTAL MOTION 931

Fig. 7. Horizontal component variations of the station GHUT (Ghuttu) correspond-

ing to different campaign surveys between 2007 and 2008. The permanent station at
Ghuttu though established in the year 2006, acquisition of data happened only in
year 2007.

essing of the entire campaign mode GPS data was done with reference to
ITRF05. Since we wanted to analyse the local deformation of the region
around the reservoir due to the loading and unloading, we obtained velocity
vectors with respect to our permanent station at Ghuttu, expecting that
Ghuttu will not be influenced by the oscillating reservoir. Simpson (1976)
pointed out that in most cases of RTS, earthquakes are seen to occur within
25 km of the reservoir area. Horizontal components of time series of the sta-
tion Ghuttu (GHUT) are shown in Fig. 7.
932 S. CHOUDHURY et al.

At twelve benchmarks, the velocity vectors for the three campaign sur-
veys (Fig. 5A-C) show a motion of 12-51, 28-86, and 40-114 mm/yr with re-
spect to ITRF05. The velocity of the permanent station at Ghuttu is
38.06 mm/yr with reference to ITRF05. The error of position coordinates is
± 1.55 mm for East offset and ± 1.36 mm for North offset. We have estimat-
ed ground motion exhibited by our 12 benchmark locations with respect to
the station at Ghuttu. In the filling cycle between October 2006 and July
2007 (Fig. 5A) the velocity vectors show a slow crustal movement of 0.69-
1.5 mm away from the reservoir (Fig. 5a). This can be correlated with low
initial seismicity in the region during this period (2005-2007). Ground mo-
tion increases in the subsequent period in the reservoir filling cycle. In the
filling cycle from July 2007 to November 2007 (Figs. 5B, 6b) the ground
motion was seen to be towards the reservoir in the order of 0.85-1.39 mm. In
the cycle from November 2007 to May 2008 (Fig. 5C), where the reservoir
water level moved down to MDDL, crustal motion was estimated to be 0.73-
1.27 mm. In our GPS data processing, error ellipses are at 95% confidence
level. During the cycles where the reservoir is filling (Fig. 5B) with simulta-
neous increase in loading, the velocity vectors appear to move inwards into
the Tehri Lake (Fig. 5b). In the cycles where the reservoir is draining
(Fig. 5A, C) the crust appears to move outwards (Fig. 5a, c). Under the load
of the reservoir water, the crust, though heterogeneous, should behave like
an elastic sheet, sagging with increase in load and rebound when the load is
released. With elasticity all GPS benchmark points should show ground mo-
tion with the loading and unloading of the dam, which is reflected by our
campaign survey results.

4. CONCLUSIONS
The large hydropower dam at Tehri, with a load of 2615 × 106 m3 water, has
been speculated at large to trigger earthquakes in the already tectonically ac-
tive region. The earth-and-rock-filled dam at Tehri is advantageous for the
safety of the huge dam considering the strain accumulation in this region.
Such dams show high inertia, flexibility and high damping absorbing earth-
quake energy enabling the dam to undergo strains without cracking (Moreira
and Poole 1993). Our analysis shows a correlation of the water level fluctua-
tions in the reservoir with the seismicity observed in and around the dam.
Our seismic data shows that the oscillating reservoir has been working to-
wards inducing stability during peak reservoir periods with higher seismicity
observed during unloading periods. A unique observation in seismicity is in
the occurrence of higher magnitude earthquakes in conditions of either fill-
ing or decreasing reservoir levels rather than at peak reservoir levels. GPS is
a precise space geodetic technique and has proved to be capable of estimat-
ing crustal motion due to loading and unloading of the Tehri Reservoir. The
 SEISMICITY AND RESERVOIR INDUCED CRUSTAL MOTION 933

present study has brought out preliminary observations for induced effect of
the reservoir. Long term study has been initiated for correlation of the oscil-
lating reservoir to the local seismotectonics and associated crustal deforma-
tion.

A c k n o w l e d g e m e n t s . We thank Director, WIHG for his support during

this work. We thank Ministry of Earth Sciences, Government of India, for providing
funds to Wadia Institute of Himalayan Geology, Dehradun for establishing the
Multiparametric Geophysical Observatory (MPGO) and the Seismic Network. We
also thank ESA for the ENVISAT (SAR) data and Mr. A.L. Sah, Ex-Chief General
Manager, Tehri Hydro Development Corporation, for the hydrological data of the
Tehri Dam. We also thank the two anonymous reviewers for their constructive sug-
gestions and efforts to modify this paper.

References

Banerjee, P., and R. Bürgmann (2002), Convergence across the northwest Himalaya
from GPS measurements, Geophys. Res. Lett. 29, 13, 30-1-30-4, DOI:
10.1029/2002GL015184.
Carder, D.S. (1945), Sesimic investigations in the Boulder Dam area, 1940-1944,
and the influence of reservoir loading on local earthquake activity, Bull.
Seismol. Soc. Am. 35, 4, 175-192.
Gough, D.I. (1969), Incremental stress under a two-dimensional artificial lake, Can.
J. Earth Sci. 6, 5, 1067-1075, DOI: 10.1139/e69-109.
Gough, D.I., and W.I. Gough (1970a), Stress and deflection in the lithosphere near
Lake Kariba – I, Geophys. J. Roy. Astron. Soc. 21, 1, 65-78, DOI: 10.1111/
j.1365-246X.1970.tb01767.x.
Gough, D.I., and W.I. Gough (1970b), Load-induced earthquakes at Lake Kariba –
II, Geophys. J. Roy. Astron. Soc. 21, 1, 79-101, DOI: 10.1111/j.1365-
246X.1970.tb01768.x.
Gupta, H.K. (1992), Reservoir-Induced Earthquakes, Elsevier, Amsterdam.
Kayal, J.R. (2001), Microearthquake activity in some parts of the Himalaya and the
tectonic model, Tectonophysics 339, 3-4, 331-351, DOI: 10.1016/S0040-
1951(01)00129-9.
Khattri, K.N., R. Chander, V.K. Gaur, I. Sarkar, and S. Kumar (1989), New seis-
mological results on the tectonics of the Garhwal Himalaya, Proc. Indian
Acad. Sci. – Earth Planet. Sci. 98, 1, 91-109, DOI: 10.1007/ BF02880378.
King, R.W., and Y. Bock (2002), Documentation for the GAMIT analysis software,
release 10.32, Massachusetts Institute of Technology, Cambridge, USA.
Moreira, J.R., and A.D. Poole (1993), Hydropower and its constraints. In:
T.B. Johansson, H. Kelly, A.K.N. Reddy, R.H. Williams (eds.), and
934 S. CHOUDHURY et al.

L. Burnham (ex. ed.), Renewable Energy: Sources for Fuels and Electricity,
Island Press, Washington D.C., 73-119.
Nawani, P.C., T.S. Pangtey, R. Sanwal, and J.S. Rawat (1990), On the continuation
of construction stage geological investigations of Tehri Dam project on
river Bhagirathi, Tehri (Garhwal) District, Uttar Pradesh, FS: 1988-1989,
Geol. Surv. India, Progress Rep. No. 27, Accession No. NRO-S6484.
Powers, P.M., R.J. Lillie, and R.S. Yeats (1998), Structure and shortening of the
Kangra and Dehra Dun reentrants, Sub-Himalaya, India, Geol. Soc. Am.
Bull. 110, 8, 1010-1027, DOI: 10.1130/0016-7606(1998)110<1010:
SASOTK>2.3.CO;2.
Simpson, D.W. (1976), Seismicity changes associated with reservoir loading, Eng.
Geol. 10, 2-4, 123-150, DOI: 10.1016/0013-7952(76)90016-8.
Simpson, D.W., W.S. Leith, and C.H. Scholz (1988), Two types of reservoir-
induced seismicity, Bull. Seismol. Soc. Am. 78, 6, 2025-2040.
Talwani, P. (1997), Seismotectonics of the Koyna–Warna area, India, Pure Appl.
Geophys. 150, 3-4, 511-550, DOI: 10.1007/s000240050091.
Valdiya, K.S. (1980), Geology of Kumaun Lesser Himalaya, Wadia Institute of Hi-
malya Geology, Dehradun,
Yadav, D.K., N. Kumar, and A. Paul (2009), Recent seismicity and stress pattern in
NW Himalaya, J. Himalayan Geol. 30, 2, 139-145.

Received 31 July 2012

Received in revised form 31 January 2013
Accepted 25 February 2013