Regional seismic event location with a sparse network: Application to eastern Kazakhstan, USSR

Abstract

Three‐component data from a sparse three‐station seismic network in eastern Kazakhstan, surrounding the Soviet nuclear test site, have been analyzed to determine location estimates for regional events recorded by two or three stations. Included among these events are the September 1987 chemical explosions whose locations are known. Locations are calculated using arrival times of P and S phases and arrival azimuths from first P. Location uncertainties are estimated using a combination of a priori and a posteriori data uncertainties. A layered P wave velocity model adapted from Soviet Deep Seismic Sounding surveys is employed for calculating travel times, and two S models are tried. Location results for the chemical explosions are excellent, even if only two stations are used: absolute location errors are less than 10 km, and estimated 90% confidence uncertainties are only a few kilometers. The data are also adequate to determine correctly their depth (i.e., focus at the surface). The other regional events include numerous suspected mine blasts and two earthquakes from the Tien Shan. The calculated locations of the latter events agree well with a teleseismic location for one of them, falling within a belt of regular seismic activity. Nearly all of the presumed blasts can be associated with mapped mines, and we have been able to identify the source areas for two sets of blasts in high‐resolution satellite images. “Before and after” photographs allow us to identify specific active mines. Our location estimates agree quite well with the observed active mines.

Location performance of a sparse regional network

“…First, we used one velocity model for all of North America east of the Rocky Mountain front. Considering this, we feel that the results of our study are generally consistent with those of Thurber et al (1989), and we support their conclusion that sparse regional networks can accurately locate small events recorded at few stations. Station corrections for the azimuth estimates may also offer some improvement.…”

“…Our final model M2L fits the complete travel time data set very well (Figure 2), but we have not investigated whether different models for smaller subregions would do even better. Thurber et al (1989) and Ruud et al {1988) also claimed that depth can be constrained if multiple secondary arrivals are observed at some stations. We plan to use our database to determine optimal time windows and frequency bands for measuring azimuths.…”

Seismic Discrimination of Nuclear Explosions

“…In practice, the main use of in-country high-frequency signals (i.e. above 10 Hz) in improving identification capability may lie in interpretation of the improved locations, including depth estimates (Ruud et al 1988, Thurber et al 1989.…”

Infrasound Detection of Large Mining Blasts in Kazakstan

“…The location and origin time of each event were determined from the seismic cross-array and are listed in Table 1. Ekibastuz comprises a number of coal mines centered about (51.67°N, 75.40°E) as determined by satellite photographs (THURBER et al, 1989).…”