Automatic Phase-Detection and Identification by Full Use of a Single Three-Component Broadband Seismogram

Abstract

Automatic-time picking continues to be a significant issue in seismo- gram analysis. Any individual method is based on separation of signal and noise in a particular domain and cannot ensure consistent onset time picking under all source, receiver, path, and noise conditions. However, a combination of three methods based on differing aspects of seismograms can be used to provide automated and consistent phase detection and interpretation using a single three-component record. For arrival detection, a combination of energy analysis, differences in instantaneous frequency, and an auto-regressive representation of the seismic record are used to develop an algorithm for robust phase extraction, suitable for use in real time. The character of the phases is extracted using polarization analysis using complex traces to separate S- and P-wave phases. The combination of four different methods offers a very effective technique for automatic arrival detection and association from a single three-component broadband seismogram and from thence an approximate event lo- cation procedure.