Full Seismic Waveform Inversion Algorithm for Gas Cloud

Abstract

Gas cloud is a critical structure of subsurface and wave propagation through subsurface anomalies such as ‘gas cloud’ is a complex phenomenon. Due to the complexity of wavefield propagation through gas cloud, the conventional inversion algorithms do not provide satisfactory results. Wave-equation based seismic waveform inversion has provided significant improvement in the spatial and temporal domain. High-resolution inversion results in geologically complex areas derive the need to develop techniques that exploit the special structure of full-waveform inversion to regularize the inverse problem. We have developed a full seismic waveform inversion algorithm to minimize the difference between observed and modeled data via optimization. In order to make the inversion efficient, a reduced Hessian matrix was used during the inversion process, given by the combined action of the Jacobian operator and its adjoint. We inverted a gas cloud model in 5 frequency bands, selected from 1 to 12 Hz, where each band contains 3 frequencies. To validate our proposed algorithm, we introduce Gaussian anomalies in the synthetic data of gas cloud model, which shows that our hierarchical strategy for acoustic full-waveform inversion yields better results than the conventional methods at reduced computational costs.