Revision of earthquake hypocenter locations in GEOFON bulletin data using global source-specific station terms technique

Abstract:

The use of a 1D velocity model for seismic event location is often associated with significant travel-time residuals. Particularly for regional stations in subduction zones, where the velocity structure strongly deviates from the assumed 1D model, residuals of up to <span" serif"="Serif"">±10 seconds are observed even for clear arrivals, which leads to strongly biased locations. In fact, due to mostly regional travel-time anomalies, arrival times at regional stations do not match the location obtained with teleseismic picks, and vice versa. If the earthquake is weak and only recorded regionally, or if fast locations based on regional stations are needed, the location may be far off the corresponding teleseismic location. In this case, implementation of travel-time corrections may leads to a reduction of the travel-time residuals at regional stations and, in consequence, significantly improve the relative location accuracy. Here, we have extended the source-specific station terms (SSST) technique to regional and teleseismic distances and adopted the algorithm for probabilistic, non-linear, global-search earthquake location. The method has been applied to specific test regions using P and pP phases from the GEOFON bulletin data for all available station networks. By using this method, a set of timing corrections has been calculated for each station varying as a function of source position. In this way, an attempt is made to correct for the systematic errors, introduced by limitations and inaccuracies in the assumed velocity structure, without solving for a new earth model itself. In this presentation, we draw on examples of the application of this global SSST technique to relocate earthquakes from the Tonga-Fiji subduction zone and from the Chilean margin. Our results have been showing a considerable decrease of the root-mean-square (RMS) residual in earthquake location final catalogs, a major reduction of the median absolute deviation (MAD) of the travel-time residuals at regional stations and sharper images of the seismicity compared to the initial locations.