Harsh K. Gupta

A review of recent studies of triggered earthquakes by artificial water reservoirs with special emphasis on earthquakes in Koyna, India

Abstract Triggering of earthquakes by filling of artificial water reservoirs is known for over six decades. As of today, over 90 sites have been globally identified where earthquakes have been triggered by filling of water reservoirs. The question of earthquakes triggered by artificial reservoirs has been addressed and reviewed in a number of papers and books. In the present review, the book “Reservoir-Induced Earthquakes” by Gupta [Gupta, H.K., 1992. Reservoir-Induced Earthquakes. Elsevier, Amsterdam, 364 pp.], which contains all the necessary information on this topic till 1990, has been taken as the base. An effort has been made to add information on this important topic gathered over the last 10 years. Koyna, India continues to be the most significant site of artificial-water-reservoir-triggered earthquakes. During 1990s, two events exceeding M 5 and several smaller events occurred in the vicinity of Koyna, and recently impounded Warna Reservoir. Detailed studies have addressed the relocation of earthquakes, stress drop, nucleation, migration and other important aspects of these earthquakes. In a unique experiment, twenty-one 90- to 250-m deep borewells have been drilled in the seismically active Koyna–Warna region and the water levels are continuously monitored. Step-like coseismic changes of several centimeters have been observed in some wells associated with a few M≥4 events. Detailed tomographic studies conducted on one of the best recorded triggered earthquake sequence at Lake Oroville in California revealed that this sequence was associated with a southwest dipping structure characterized by low velocity, while in adjacent areas, seismic activity occurs in regions of higher velocity. Similar investigations in Aswan showed that shallow activity is associated with low P-wave velocity. Several new reservoir sites that have triggered earthquakes have been reported during 1990s. The most important being Srinagarind Dam in Thailand, which had an M 5.9 earthquake and the sequence had all the characteristics of triggered earthquake sequences. New theoretical work, particularly the effect of pore fluid pressure in anisotropic rocks and its implication in triggered seismicity is an important development. However, much more needs to be done to fully comprehend the role of artificial water reservoirs in triggering earthquakes.

Fluids below the hypocentral region of Latur Earthquake, India: Geophysical indicators

A set of geophysical measurements and observations was undertaken to investigate the nature of the crust beneath the epicentral region of the deadly Mw 6.1 Latur earthquake of September 30, 1993. With an estimated focal depth of 2.6 km and the associated well defined but subtle surface ruptures, it is a rare stable continental region (SCR) earthquake with surface rupture. The focal depth of 69 out of 73 well located aftershocks is less than 5.5 km. Broad band (10³–10−3 Hz) magnetotelluric (MT) soundings reveal the presence of an anomalously high conductivity zone at a shallow depth range of 6–10 km. Consistent with this result is the observation of a Pc phase, lagging behind the Pg phase by about 0.6 to 0.8 sec in the aftershock seismograms indicating a low velocity layer (LVL) at 7 to 10 km depth. A Bouguer gravity low of 5 m.gal, nearly coincident with this feature, is also observed. Above evidences indicate that the focal zone of the Latur earthquake sequence is limited to depths of about 5 to 6 km in the upper crust by an underlying low-velocity and high conductivity layer. We interpret this high conductive, low velocity layer as a fluid filled fractured rock matrix. The inferred stress regime, including due to uplift of the Deccan Plateau, triggered by erosion of basalt cover is likely to be confined mostly in the upper part of the crust. Existence of a low velocity, high conductivity fluid filled layer will enhance stress concentration in the uppermost brittle part of the crust causing mechanical failure.

Anatomy of surface rupture zones of two stable continental region earthquakes, 1967 Koyna and 1993 Latur, India

Soil-helium surveys in the surface rupture zones of the 1993 Latur (Mw 6.2) and the 1967 Koyna (Mw 6.3) stable continental region (SCR) earthquake sites in Peninsular India showed anomalies defining surface traces of the causative faults. Propagating from the Archaean crystalline basement through the Deccan basalt cover, the seismic fault produced a scarp by uplift along a thrust in the Killari area of the Latur earthquake, whereas the Koyna earthquake was associated with a strike-slip fault which expressed itself as an en echelon fissure zone. Core drilling has confirmed that the fault at Killari extends downward from the surface rupture zone with an approximate dip of 50° towards SSW. The level differences of flow contacts obtained by drilling in the hanging wall and foot wall sides of the fault, do not unequivocally establish the amount of displacement, but suggest that it might be anything from 1 m to 6 m. If the higher figure of 6 m is accepted, it would indicate reactivation of an old fault. Drilling established a WNW dip of the Koyna fault, resolving a long-standing debate.

The deadliest stable continental region earthquake occurred near Bhuj on 26 January 2001

A large destructive earthquake occurred on 26 January 2001 in the region of Kutch, Gujarat, in Western India, with magnitude Mw 7.7. The earthquake caused very heavy damage and a large number of casualties with more than 20,000 deaths. A preliminary study of ground deformation, damage pattern and aftershock distribution is presented.

Paleoliquefaction evidence and periodicity of large prehistoric earthquakes in Shillong Plateau, India

Abstract The tectonic setting and the occurrence of the great Assam earthquake (M=8.7) of 1897 in the Shillong Plateau succeeded by three great earthquakes (1905, 1934 and 1950) in the adjoining Himalayan frontal arc, indicates the vulnerability of the Shillong Plateau to large earthquakes. The lack of seismicity records of the region earlier than 100 years and data on the recurrence of damaging earthquakes led us to investigate the paleoseismicity of the Shillong Plateau. Our paleoseismic investigations in the meizoseismal area of the 1897 earthquake revealed well-preserved liquefaction and deformed syndepositional features at 10 selected sites in the alluvial deposits along two north flowing tributaries of the Brahmaputra river. These features are 14C dated using associated organic samples. As the liquefaction of sediments was an important feature of the 1897 earthquake, we identified this seismic event, and other large/major prehistoric earthquakes through paleoliquefaction and other coseismic structural deformation at the investigated sites. In addition to the 1897 event, we provide geological evidence for at least three large seismic events. Two of them occurred during 1450–1650 and 700–1050 AD, the third predates 600 AD. The analysis of the 14C data suggests a return period of about 400–600 yr for the large earthquakes in the Shillong Plateau. This finding is the first of its kind from the Himalaya and adjoining region.

Timing and return period of major palaeoseismic events in the Shillong Plateau, India

Abstract The close temporal occurrence of four great earthquakes in the past century, including the great Assam earthquake of 1897 in the Shillong Plateau, necessitated examination of the palaeoseismicity of the region. The results from such investigation would definitely aid in addressing the problem of the earthquake hazard evaluation more realistically. Our recent palaeoseismological study in the Shillong Plateau has led us to identify and provide geological evidence for large/major earthquakes and estimate the probable recurrence period of such violent earthquakes in parts of the Shillong Plateau and the adjoining Brahmaputra valley. Trenching along the Krishnai River, a tributary of the River Brahmaputra, has unravelled very conspicuous and significant earthquake-induced signatures in the alluvial deposits of the valley. The geological evidence includes: (1) palaeoliquefaction features, like sand dykes and sand blows; (2) deformational features, like tilted beds; (3) fractures and syndepositional deformational features, like flame structures caused by coeval seismic events. Chronological constraints of the past large/major earthquakes are provided from upper and lower radiocarbon age bounds in the case of the palaeoliquefaction features, and the coeval timing of the palaeoseismic events is obtained from the radiocarbon dating of the organic material associated with the deformed horizon as well as buried tree trunks observed wide distances apart. Our palaeoseismic measurements, which are the first from the area, indicate that the Shillong Plateau has been struck by large/major earthquakes around 500±150, 1100±150 and >1500±150 yr BP, in addition to the well-known great seismic event of 1897, thereby the 14 C dates indicate a recurrence period of the order of 500 yr for large earthquakes in the Shillong Plateau.

Enhanced reservoir-induced earthquakes in Koyna region, India, during 1993-95

Reservoir induced earthquakes began to occur in the vicinity of Shivajisagar Lake formed by Koyna Dam in Maharashtra state, western India, soon after its filling started in 1962. Induced earthquakes have continued to occur for the past 34 years in the vicinity of this reservoir, and so far a total of 10 earthquakes of M ≥ 5.0, over 100 of M ≥ 4 and about 100,000 of M ≥ 0.0 have occurred. Every year, following the rainy season, the water level in the reservoir rises and induced earthquakes occur. Seismic activity during 1967–68 was most intense when globally, the largest reservoir induced earthquake occurred on 10 December, 1967. Other years of intense seismic activity are 1973 and 1980. During 1986 another reservoir, Warna, some 20 km south of Koyna, began to be filled. The recent burst of seismic activity in Koyna-Warna region began in August, 1993, and was monitored with a close network of digital and analog seismographs. During August, 1993–December, 1995, 1,272 shocks of magnitude ≥ 2 were located, including two earthquakes of M 5.0 and M 5.4 on 8 December, 1993 and 1 February, 1994, respectively. Two parallel epicentral trends in NNE-SSW direction, one passing through Koyna and the other through Warna reservoir are delineated. The 1993 increase in seismicity has followed a loading of 44.15 m in Warna reservoir during 11 June 11, 1993 through August 4, 1993, with a maximum rate of filling being 16 m/week. The larger shocks have been found to be preceded by a precursory nucleation process.

An investigation into the Latur earthquake of September 29, 1993 in southern India

Abstract The devastating ( M w 6.2) Latur earthquake of September 29, 1993 in South India has claimed an estimated 11.000 human lives. With an I max of VIII, the earthquake was felt to an average distance of 750 km. More than 125 shocks were reported to have been felt during August 1992–March 1993. Out of these, during October–November 1992, several shocks of M ≥ 2.0 were recorded at the NGRI seismic station at Hyderabad which is the closest (220 km) to the epicentre. No such shocks occurred for at least 8 months before the Latur earthquake. The aftershocks were monitored by a network of up to 21 stations between October 8, 1993 and January 31, 1994. A majority of the aftershocks occurred within a 10-km radius from the main shock. On the basis of the location of aftershocks in the first few days, a plane dipping at an angle of 45° towards the southwest and striking at 135° is inferred to be the fault plane which extends to a depth of 4.5 km and on projection meets the surface in the vicinity of the observed surface rupture. Assuming the aftershock zone to be the rupture zone, the stress drop is estimated to be 7 MPa with a maximum displacement of 1.7 m for the main earthquake. A unique discovery is the high concentration of helium in the soil in the immediate vicinity of the surface rupture indicating that the rupture extends to the surface from a depth of a few km. A detailed broadband magnetotelluric (MT) investigation revealed the presence of an anomalously high conductive zone at a depth of 6–10 km. Observation of a PC phase, lagging behind the Pg phase by about 0.6 to 0.8 s in the seismograms of aftershocks, indicating a low-velocity zone at 7 to 10 km depth, is consistent with the MT results. This highly conductive low-velocity layer is inferred to be fluid-filled. The main stress regime in Peninsular India is NE compressive stress due to plate tectonic movement. Erosion of the basalt cover in the Deccan Plateau may be adding additional compressive stress in the region. The existence of a low-velocity highly conductive fluid-filled layer will enhance stress concentrations in the uppermost brittle part of the crust causing the earthquake.

Gas-hydrates in Krishna-Godavari and Mahanadi basins: New data

KALACHAND SAIN, MAHESWAR OJHA, NITTALA SATYAVANI, G.A. RAMADASS, T. RAMPRASAD, S. K. DAS and HARSH GUPTA CSIR National Geophysical Research Institute, Uppal Road, Hyderabad 500 007 National Institute of Ocean Technology, Velachery-Tambaram Main Road, Chennai 600 100 CSIR-National Institute of Oceanography, Dona Paula, Goa 403 004 Ministry of Earth Sciences, Prithvi Bhavan, Lodhi Road, New Delhi 110 003 Email: kalachandsain@yahoo.com

Crustal seismic velocity structure in the epicentral region of the Latur earthquake (September 29, 1993), southern India: inferences from modelling of the aftershock seismograms

Abstract We present 1-D models of crustal P and S wave velocity structure in the epicentral region of the devastating 1993 Latur earthquake in southern India. Travel time and relative amplitude modelling of a seismogram section of well located aftershocks out to 80 km offset, by reflectivity synthetic seismograms, reveal alternating low-velocity layers (LVLs) for P and S waves in the upper crust at depths of 6.5–9.0 km and 12.3–14.5 km with about 7% velocity reduction. A lower crustal LVL at 24–26 km depth is also inferred by modelling. The seismic activity in the Latur region is essentially confined to shallow upper crustal depths of less than 6 km, above the LVLs. The upper crustal velocity models further reveal a relatively low Vp/Vs ratio of 1.65, while synthetic seismogram models consistent with the upper crustal P wave reflection phases reveal a Qp/Qs ratio certainly less than 1. A high-conductivity zone at a shallow depth of 6–10 km brought out by broad band MT soundings in this region, is found to be coincident with the upper crustal LVLs. These coincident anomalous properties of the upper crust suggest the presence of fluids (probably in a partially- or slightly under-saturated state) below the hypocentral region of the Latur earthquake and its aftershocks.