Surface-consistent minimum-phase deconvolution

Abstract

A method is provided for adaptive minimum-phase deconvolution. The deconvolution method removes source and receiver wavelet variations in a surface-consistent manner from traces of a seismic survey which has the traces indexed by source and receiver locations. A plurality of source-pairs and a plurality of a receiver-pairs are selected from the traces. Each source-pair has two designated sources and each receiver-pair has two designated receivers. Cross-correlations are performed on traces in each source-pair which have common receivers to provide a source-pair cross-correlation signal for each common receiver. Cross-correlation are performed on traces in each receiver-pair which have common sources to provide a receiver-pair cross-correlation signal for each common source. The cross-correlation signals of each receiver-pair and each source-pair are stacked to provide stacked cross-correlation signals. The stacked signals are then used in a surface-consistent convolutional model to obtain within a selected bandpass an amplitude frequency spectrum for each source and receiver wavelet. In one embodiment, an edited amplitude spectrum is obtained. A minimum-phase wavelet spectrum is then determined from each edited amplitude spectrum using a Hilbert transform. The minimum-phase wavelet spectra are then used to determine a two-sided finite-length time domain deconvolution filter for each source and receiver location. The respective filters for each location are then convolved with the trace to deconvolve the trace.