Assessing Event Location Capability With Ground Truth Events at the Degelen Mountain Test Site, Kazakhstan

Abstract : We are investigating seismic event location capability in Kazakhstan using first-P arrivals from nuclear explosions with relatively precise or exact ground truth information. We use a waveform cross-correlation (WCC) method to determine high-precision relative arrival times. We then use the improved arrival times to examine several issues regarding location accuracy and precision. Since the true explosion locations are known precisely, we can evaluate both relative (joint hypocenter determination, JHD) and absolute (single-event) location capability. Aspects we are currently investigating include relative and absolute location accuracy, in particular using sparse regional-distance observations, the spatial variability of derived source-specific station corrections (that is, path corrections), and the influence of global heterogeneity on absolute location accuracy. We have combined information from several sources in order to establish adequate ground truth information for Degelen Mountain. Locations and origin times for 68 Degelen explosions were published by Bocharov et al. (1989). Tunnel portal coordinates for all Degelen explosions were released as part of a tunnel closure project (Leith, 1998). In addition, we obtained sections of a map showing all of the Degelen tunnels, albeit with some apparent distortions and no coordinates (W. Leith, pers. comm.). By combining the true portal coordinates with the tunnel maps, we were able to determine approximately the complete location and geometry of all the tunnels. Using the Bocharov et al. (1989) ground truth information, we then verified our estimated tunnel geometry for those events, and determined that the explosions all took place at essentially the tunnel ends. We then adopted the other tunnel ends as our best estimates of the locations for the remaining Degelen explosions.