Fast stochastic inversion based on simulated annealing algorithm

Abstract

Seismic resolution is the key factor to the extraction of stratigraphic information from seismic data and it is very important for the thin layer description. Due to the band limited nature of seismic data, conventional inversion methods are unable to meet the needs of current exploration and exploitation demands. Compared to the deterministic inversion, stochastic inversion can effectively integrate the high frequency information of well-logging data and have a higher resolution. This paper starts from the original framework of stochastic inversion, combined with FFT-Moving Average (FFT-MA) simulation and Simulated Annealing (SA) algorithm, proposes a new fast stochastic inversion method. According to the Bayesian theory, using well-logging data as conditional data and seismic data as constraints, we integrate geostatistical priori information into the posteriori probability density function of the model. Finally, through the numerical calculations, we can conclude that FFT-MA simulation after conditional processing, combined with SA algorithm, can reduce the time consumption. The inversion result can be converged to the seismic data, and the final results match the model very well, which demonstrates the effectiveness of the proposed inverdion method.