Seismic Space-Time Processing

Abstract

Frequency filtering and spatial filtering have been usefully applied for many years to the problem of extraction of signals from noise in active seismic exploration and in passive earthquake surveillance. Because broadband (octave or more) transients are of interest in both areas, frequency dependent spatial filtering has been frequently used. In some cases, interfering noise is spatially nonrandom, and “optimum” space-time filters are used to advantage. Essentially similar spatial response is usually obtained independent of the “optimization” criteria (e.g., least squares, maximum likelihood). Similar techniques may be used to employ arrays with complex inter-element signal relationships (such as combinations of horizontal and vertical motion seismometers). Although useful coherent properties of signal and noise can be exploited with an assumption of space stationarity, time stationarity, or space-time stationarity, in many cases use of space-time varying optimization is worthwhile and is being increasingly employed. Some of the current approaches to tracking down useful statistical space-time properties of signal and noise and their exploitation in seismic information extraction will be discussed.