Multi-scale seismic envelope inversion method based on sparse representation theory

Abstract

Combining seismic data with a constrained model can obtain a high-resolution inversion result. However, it is very difficult to obtain an accurate low-frequency model because of the lack of wells in the offshore exploration area. We propose a multi-scale seismic envelope inversion method based on sparse representation to solve the problem of seismic inversion in offshore well-less areas. First, we demodulate the envelope of the seismic signal based on signal modulation theory and take it as the low-frequency component for inversion. Then, decompose the seismic signal based on MPSTFT (Short-Time Fourier Transform based on Matching Pursuit algorithm) method to provide high-resolution components of intermediate-frequency and high-frequency. Next, we take the linear impedance model as the initial model and update the inversion results by multi-scale inversion based on Bayesian inversion theory. Finally, we use the synthetic data to demonstrate the effectiveness of the proposed method and apply it to real seismic data. Comparing inversion results, the result of the proposed is more accurate and possesses a higher resolution. The results show that the proposed method has good adaptability to solve the problem of offshore well-less area inversion with a high-resolution inversion result.