AUTOMATIC SELECTION OF PARAMETERS FOR SPECTRAL DECOMPOSITION OF SEISMIC DATA

Spectral decomposition is a widely used method of qualitative seismic interpretation, a key limitation of which lies in the subjective selection of frequency components for visualization. Existing approaches based on the analysis of the global Fourier amplitude spectrum do not account for the temporal localization of the signal’s spectral content. We introduce a new method for automatic frequency selection based on approximating local spectra obtained via the short-time Fourier transform (STFT) as a linear combination of three Morlet wavelet spectra. This approach enables the extraction of time-dependent frequency trajectories that capture the evolution of the signal’s spectral content, rather than relying on a fixed frequency triplet over the entire interval. We have solved the optimization problem using the differential evolution algorithm and have tested the method on field data from the West Siberian oil-and-gas province. Quantitative evaluation using Shannon and Rényi entropy metrics, as well as a colorfulness metric, has shown increased visual separability and color contrast in RGB composites. The proposed method reduces the subjectivity of interpretation and ensures reproducible results.