The Himalayan region is seismically very active and has experienced several large earthquakes in the past. The intense folding, faulting and thrusting that took place during the Himalayan Orogeny have given rise to several tectonic provinces characterised by changes in seismicity. The deformation of the Indian Plate has resulted in several stress regimes such as the Western and Eastern Syntaxes, the Frontal Arc and others. The Himalayan Frontal Arc, flanked by the Arakan-Yoma in the east and the Chaman fault in the west, forms one of the most seismically active segments of the Alpide belt [1] [2] . Focal mechanism solutions of earthquakes in the Himalayan region have revealed that over 607 events are of thrust fault type [3] .

Seismological investigations in the Himalayan region have been undertaken by many researchers [4] - [15] . [16] observed that the high seismicity of northeastern Kumaon and adjoining Nepal is primarily related to strike-slip movement along some transverse faults. The seismicity of the Himalayan region has been attributed to movement along either the Main Boundary Thrust [17] [18] or with the Main Central Thrust [19] . Detailed studies of the Himalayan seismicity and surrounding regions have been carried out by [19] [20] based on the contouring of a-values ranging from 3.0 to 7.0 and identified various seismically active areas. These studies have been undertaken in a large area, indicating that the Himalayan region’s different parts are highly active. Such studies must be confined to a particular area to better understand the seismicity associated with a single tectonic unit. Such studies are pre-requisite for pattern recognition.

The study area ( Figure 1 ) is situated at the boundary of the Lower and Higher Himalayas and is probably traversed by the Main Boundary Fault [21] . The central part of the area has experienced intensity VIII (Rossi-Forel scale) isoseismal of the great Kangra earthquake of 1905 [22] .

A proper understanding of the seismicity of a region and its meaningful interpretation requires a comprehensive and reliable database. Such a database is helpful to study the spatial distribution of events to a region’s tectonic features and to undertake various investigations for pattern recognition. Given this, the present paper aims to prepare a PC-based seismicity database for Tehri and its vicinity to study the spatiotemporal variations of seismicity and associated phenomena. An attempt has also been made to identify precursory swarm and anomalous seismicity rate changes based on the cumulative number of earthquake occurrences as a function of time (CNET).

A seismicity database for the region bounded by 25˚N - 40˚N and 65˚E - 85˚E has been prepared with the help of a compatible personal computer system using a standard database management package [23] [24] . The whole region (25˚N - 40˚N and 65˚E - 85˚E) has been divided into three zones based on seismic activity (Singh and Raghavan, 1991a, b), namely Zone I, Zone II and Zone III. The seismicity of Zone II and Zone III (a part of the Himalayan region) are comparable to each other. In contrast, Zone I is extremely active since it includes the Hindukush and Tien Shan regions [25] . The pertinent literature and available earthquake catalogues (Catalogues of International Seismological Summary (ISS); Preliminary determinations of Epicentres (POE) of USGS); [26] - [35] have been critically examined to prepare a comprehensive database. A total of 6105 events from November 1, 1853, to March 31, 1989, have been incorporated in the database.

Before 1853, the reporting of epicentral location and magnitude were highly inconsistent [36] . Since the location of the events is of critical importance for seismological investigations, only those events are listed for which epicentral locations are available. The database has been critically scrutinised, and duplicates (the same events reported in different sources) have been eliminated. For the first time, a seismicity database of part of the Indian region has been compiled in computer format using a PC system that is supposed to be the most reliable and accurate since it is free from apparent errors often observed in the existing catalogues. The existing catalogues have minimal application in their present form since they are not prepared in computer format or arranged systematically [23] [24] .

Tehri area is approximately centrally located in Zone III, where 543 events have been reported from November 1, 1853, to March 31, 1989 ( Figure 1 ). During the entire period, the Himalayan Frontal Arc (Zone experienced 5 large earthquakes of M ≥ 7 till 1962 and medium-sized earthquakes of mb ≥ 6 ( Table 1 ) from 1963 to March 31, 1989 ( Figure 2(a) ). The great Kangra earthquake of April 4, 1905, of M = 8.6, occurred in this region about 400 km northwest of Tehri. The recent seismic activity due to medium-sized earthquakes in the area seems to be clustered in the north and east of Tehri. The database record shows no major earthquake has been reported in the Tehri area during the last 17.5 years [37] . The only recent destructive earthquake of October 20, 1901 (IST) occurred at Uttarkashi, about 40 km northeast of Tehri ( Figure 2(a) ).

The database is subjected to identify precursory swarm and seismicity rate changes associated with medium size earthquakes. Such investigations on a regional basis in the Himalayas have been carried out by [50] - [52] . [24] reported that medium to large-magnitude earthquakes are preceded by precursory swarm and seismicity rate changes. The basis of these investigations is the Cumulative Number of Earthquake occurrences as a Function of Time (CNET) curves, which, in the present case, have been obtained for two schemes: (1) Total database since 1902 and (2) database for mb ≥ 4.6 since 1967. Since 1963, the database has been considered for detailed studies only since the same is not complete before 1963.

Three medium size earthquakes (mb = 5.4, 5.4 & 5.6) have occurred in the Tehri area since 1963 ( Table 2 ). Swarms are events that occur quickly in an area spread over a period and followed by a quiescence before the occurrence of mainshocks. Also, in a particular area if there are a series of small earthquake in a short duration, but none of the events is a principal event, then the series of earthquakes are considered as a swarm. On analysis of the database for the two schemes mentioned above, it was found that none of these medium-sized earthquakes in the Tehri area were preceded by swarm-like activity ( Figure 3(b) ).

The changes in the seismicity rate of a particular strange period may show an increasing or decreasing trend as compared to rates of preceding periods. Properly assessing dependent and independent events in the database is necessary to differentiate accurate rates caused by earthquake processes in the earth’s interior from those associated with detection changes. In the present case, total events and events that complete the database (mb ≥ 4.6) have been considered. It is presumed that the dependent events (swarm. foreshocks, clusters, aftershocks, etc.) are not present, especially on considering the database for events with mb ≥ 4.6 since

The region is moderately active. The CNET curve shown in Figure 3(b) cannot be interpreted accurately before 1963 since the database is incomplete due to poor location and detection capabilities that prevailed during the period. The CNET curves pertaining to total events and events with mb ≥ 4.6 show a similar pattern of continuous increasing trend. Three anomalous periods can be identified on careful examination of the CNET curve of total events from 1963 to 1988: 1963 to 1973, 1974 to 1985 and 1986 to 1988, during which the seismicity rates deviated in the ratio 1:2.2:4.5 respectively. This indicates that the continuous increase of the seismicity rates in successive anomalous periods may largely be attributed to the detection changes ( Figure 3(b) ).

On the other hand, three anomalous periods can also be identified by carefully examining the CNET curve obtained for the events with mb ≥ 4.6 ( Figure 3(b) ): 1963-1973, 1974-1985 & 1986-1988. The seismicity rates in these periods are 0.5, 0.9 & 1.7 events/year, respectively. The CNET curve ( Figure 3(b) ) shows a uniform gradient till 1985, which may be largely due to detection changes. However, there is a two-fold increase in the seismicity rate from 1986 compared to the preceding anomalous period due to a sharp change in the gradient of the CNET curve ( Figure 3(b) ). This change is conspicuous and may be considered a real one caused by the earthquake processes in the earth’s interior. When the present database is analysed in addition to data reported by USGS for the period from April 1, 1989, to October 1, 1991, it was observed that the Uttarkashi earthquake of October 20, 1991, was not preceded by a precursory swarm. However, the two-fold increase in seismicity rate change from 1986 may be considered as precursory to the Uttarkashi earthquake. The validity of this change may be obtained only if the present data have been available on a wide range of magnitude.

We acknowledge the constraints of the database, which are restricted to events with a magnitude of mb ≥ 4.6 only, and recognize the inherent error associated with estimating focal depths. This has limited the mapping of the extent of four main tectonic features, even considering total events. Despite these limitations and the number of observation points, the focal depth contour map ( Figure 2(c) ) seems very reasonable, separating normal focal depth areas (less than 36 km) from the regions where considerably deeper events have been reported.

The identification of seismic precursors is constrained due to the lack of lower-magnitude events. The area has not had much seismic activity for the past 135 years since November 1, 1853. It was observed that none of the medium-sized earthquakes of the Tehri area were found to be associated with the precursory swarm. However, seismicity rate changes obtained from the CNET curve for the total events indicated a continuously increasing trend, which is interpreted mainly due to detection changes ( Figure 3(b) ). At the same time, an increase in real rate change has been observed from 1986 on considering events for mb ≥ 4.6 ( Figure 3(b) ). The changes in the seismicity rate indicate only the probability of occurrence of future major earthquakes in a region under consideration. Hence, it isn’t easy to estimate the occurrence time and magnitude and to pinpoint the location with the help of this precursor. The Tehri area has been moderately active, and no event of M ≥ 6 occurred from 1902 to March 31, 1989 ( Table 2 ). Hence, it can be said that if this rate change is real, then the associated mainshock magnitude would be of moderate size. This has been inferred based on past seismicity reported in the area. The spatial distribution of events ( Figure 2(b) ) indicates that the more significant events in the area are located northeast of Tehri.

The main conclusions of the present investigations are:

The seismicity database of the Tehri area from 1963 to 1988 is complete for the events with mb = 4.6 only.

Based on the spatial distribution of events, four main moderately/active tectonic features responsible for the Tehri area’s seismicity have been delineated. General seismicity in the Tehri area is considerably low, particularly near it. Seismic activity is mainly confined in the areas beyond 60 km east and northwest of Tehri. These two tectonic features are separated by a well-defined north-south trending gap of about 40 km wide (east-west), where no event has been reported ( Figure 2(b) ). However, the spatial distribution of earthquakes may not correlate directly with the number of events or encompass all earthquakes in the database.

Earthquakes in the Tehri area are of shallow focus, with a maximum concentration of 25 - 36 km. The magnitude of the majority of events ranges from 4.1 to 5.0. The events with focal depths (less than 36 km) are confined in a stretch oriented along the northeast-southwest where Tehri is centrally located.

None of the medium size earthquakes in the Tehri area are preceded by precursory swarms. However, a signature of a 2-fold increase in real seismicity rate change has been observed from 1986 on considering events for mb ≥ 4.6 from 1963 to 1988. The earthquake associated with this rate change would be of moderate size, which has been envisaged based on the pattern of earthquake occurrences in the past. The seismicity rate change observed from 1986 may be considered a possible precursor to the recent Uttarkashi earthquake of October 20, 1991 (IST).

The authors sincerely thank the Department of Earthquake Engineering and Civil Engineering, Indian Institute of Technology Roorkee, India, for providing computational facilities for carrying out this work.