A high-wavenumber full waveform inversion method based on sparse decomposition

Complex reservoir imaging and characterization impose higher requirements on the accuracy of velocity modeling. High-wavenumber full waveform inversion (FWI) is an effective way to realize high-precision velocity modeling. To address the instability and local minima issues inherent in high-wavenumber FWI, appropriate multi-scale inversion strategies are required. This study implements high-resolution time-frequency decomposition of time-domain seismic data through a local peak-constrained sparse decomposition method, and constructs a sparse decomposition-based FWI algorithm. This approach enables multi-scale inversion from low to high wavenumbers to reconstruct the velocity model, thereby providing a high-accuracy velocity field for reservoir imaging and characterization. Numerical model tests show that sparse decomposition-based high-wavenumber FWI enhances inversion stability and yields high-precision velocity models. The application to practical data shows that reservoir description can be directly carried out using the velocity model derived from high-wavenumber FWI to a certain extent.