Automatic phase pick refinement and similar event association in large seismic datasets

Large digital seismic data sets typically have approximate and inconsistent P and S wave arrival estimates. If not corrected, this situation frequentl y blurs the true hypocenter distribution so as to mask important fault, vol canic, or other seismogenic structure. Recent applications which address th is problem for increasingly large data sets have resulted in dramatic illum ination of seismogenic features in hydrocarbon and geothermal reservoirs, a s well as in tectonic and volcanic earthquake regions. Existing algorithms frequently require substantial analyst interaction, which can become prohib itive once more than a few tens of events are considered. We summarize prog ress towards developing a fully automatic technique for removing pick incon sistencies and subsequently re-estimating hypocenters, and discuss our deve lopment of an adaptive, correlation-based algorithm which points the way to wards techniques that appear suitable for the largest data sets presently e xisting.