The wavelets measured in VSP downgoing waves are extracted at arrival time of the first break in the VSP data, and then convolved with a time-frequency domain constant-Q filter. The filtered wavelets obtained from VSP are compared to the propagating wavelet estimated from surface data by Wiener, frequency domain spiking and Gabor deconvolutions. To verify the accuracy of the results from surface data, we use normalized cross-correlation to carry out the comparison. This study shows that wavelet estimates from VSP surveys can be used to evaluate the accuracy of the wavelets estimated from surface data. Our results are consistent with the hypothesis that Gabor deconvolution can accurately estimate the nonstationary wavelets embedded in real seismic records. These results also suggest that Gabor deconvolution is superior to a multi-window Wiener or frequency domain spiking deconvolution.
View full article as PDF (0.46 Mb)