When applying statistical deconvolution techniques to seismic data, it is almost universal to assume that the spectrum of the earth's reflectivity is white over some particular bandwidth. In situations where seismic data are acquired in reasonable proximity to an existing well with one or more reflectivity logs, however, the nonstationary deconvolution of the seismic traces can be improved by replacing the conventional white reflectivity spectral assumption with a simple 2D colour function (of frequency and time), whose shape has been derived from spectral analysis of the actual well log. Least-squares fitting of log reflectivity spectra to low-order polynomial curves has been shown to provide a convenient representation of the `colour' of well log spectra. To facilitate the testing and application of `colour-correction' in the nonstationary Gabor deconvolution algorithm to large data sets, the colour-correction scheme has been added to our existing ProMAX Gabor deconvolution and analysis modules. We describe here the specifics of those modules, including parameter selection and format requirements for the ASCII file which conveys into Gabor deconvolution the polynomial coefficient tables from MATLAB well log analysis; and we demonstrate the updated algorithms on field data.
View full article as PDF (0.68 Mb)