Using beamforming for the global network location

Abstract

The aim of this study is the design and trial of a novel sparse network beamforming (NB) technique to improve earthquake location. For this purpose bulletin phase arrival time data were processed via the use of complex exponents form and then used in an optimization via grid-search to search for the maximum semblance in hypocenter space. The use of the robust semblance statistic, provides reliable location results. The NB location results for a set of test events were compared to standard iterative ISC location procedure “iscloc” and its prototype the Jeffreys maximum likelihood estimator. For this purpose a data-base of 139 reference ground-truth events was extracted from the catalog of the International Seismological Centre (ISC) (97 GT5 events, assumed earthquakes, and 42 GT0 nuclear tests). The tuned NB procedure has shown excellent location results for events with a “good” ISC location and demonstrated the large epicenter deviations have been decreased in “bad” cases. Further developments of the algorithm would include allowances for 3D earth structure and a priori site-specific information.