Improving seismic event location: An alternative to three-dimensional structural models

We devise and apply a method to account for the effect of the aspherical st ructure of the Earth in locating earthquakes. This technique relies upon th e ability to detect the average structural signal present in the residuals between source and receiver and correct for this signal during location, us ing a phenomenological description that we call Empirical Heterogeneity Cor rections (EHC). EHC are employed in the relocation of a large set of well-c onstrained teleseismic earthquakes selected among the events reported by th e Bulletins of the International Seismological Centre 1964-1995. The rms le ngth of EHC relocation vectors for these events is about 10 km. The method is also tested against a selected set of ground-truth events, both earthqua kes and explosions, whose locations are independently known by nonseismic m eans. The rms length of the mislocation vectors for the test events, compar ed to their original mislocation in the reference 1-D model SP6, is reduced in the EHC relocation by 17% for explosions and 12% for earthquakes. Out t echnique provides a successful alternative to the use of 3-D structural mod els, approximately reaching the same value of effectiveness in improving ev ent location.