Seismic data decomposition using singular points with a matching pursuit scheme

摘要

Using a simple thin-bed tuning concept, spectral decomposition is routinely used to map subtle stratigraphic features in both clastic and carbonate terrains, thin bed tuning effects. However, spectral decomposition also forms the mathematic basis for many Q-estimation, pore-pressure prediction, chronostratigraphic mapping, unconformity mapping, and bandwidth extension algorithms. We extend earlier introduced concepts of maximum modulus lines to better decompose a seismic trace into its constituent wavelet components, resulting in an algorithm that is a generalization of the popular matching technique. In this paper, we reconstruct the most important information of the data and remove steeply dipping noise using multiple levels of ridges. We also note that these ridges are sensitive to waveform dispersion, suggesting that we can extend the bandwidth of the data, compensate for geometric and intrinsic Q.