The Concept of Virtual Arrays in Seismic Data Acquisition

Abstract

We are presenting a new way of improving seismic-array responses. By analyzing the relationship between the covariance matrix forms from the real sensors of seismic arrays and the fourth-order crosscumulants from the same sensors, we find that artificial sensors can be constructed from the real sensors. We have called these artificial sensors virtual sensors and the combination of real and virtual sensors a virtual seismic array. For example, we can construct from an equally weighted linear array of five sensors, a weighted virtual array of nine sensors. Basically, the virtual sensors allow us to introduce new sensors in the seismic arrays as well as new weightings of the existing real sensors.

The key assumption behind this approach is that seismic data are considered nonGaussian; hence the fourth-order crosscumulants of the real sensor responses are nonzero.