Matched Field Processing Framework for Coherent Detection Over Local and Regional Networks

摘要

In this final report it is demonstrate that it is possible to generalize the single-phase matched field processing template developed under a previous contract to a template constructed from an entire seismogram. This template can be used as a basis for a detection algorithm implemented in a subspace detector framework. Furthermore, the framework can represent correlation detectors and the incoherent (narrowband) matched field detectors and gradation between the two. It is shown that detectors in the framework can be made to adapt, as new event waveforms become available, through a subspace update mechanism. A detector that begins as a matched field detector will adapt to resemble a correlation detector, if the source produces highly repeatable waveforms. This form of adaptation is demonstrated with a highly repeatable source in Fennoscandia, and somewhat different adaptive behavior that improves detection performance with a mining explosion source in the Korean peninsula. It is demonstrated that the performance of the matched field detector can improve dramatically as the number of available sensors increases, and that the matched field and subspace detectors are applicable to very large apertures over which conventional notions of waveforms coherence do not apply.