AVO Inversion in Orthotropic Media Based on SA-PSO

The seismic inversion of anisotropic media is a popular research direction. Orthotropic media closely approximate real strata, but their inversion remains difficult because of their many parameters and large variations. In this article, particle swarm optimization (PSO) and simulated annealing (SA) are combined to invert prestack amplitude variation with offset (AVO) data in orthotropic media. PSO is a stochastic algorithm that easily falls into local minima and has a poor global search ability. Integrating SA into PSO adds the probabilistic jump-out mechanism of SA, modifying the global optimization of PSO by adjusting the temperature. The hybrid SA-PSO algorithm can overcome premature phenomena, avoids easily falling into local minima, and fissures stronger global optimization. With the P-wave reflection coefficient formula, we establish an objective function for the anisotropic parameters of orthotropic media and use SA-PSO to invert prestack AVO data in orthotropic media. The inversion process is stable, and the inversion error is small. The proposed method is applied to the partial Marmousi2 model, and seven anisotropic parameters of orthotropic media are inverted. According to the relationships between anisotropic and fracture weakness parameters, profiles of the weakness parameters are obtained, and the fracture infill materials are predicted. The prediction results are consistent with the actual model, verifying the correctness of the method. The difference between the reservoir and surrounding rock is significant in the inversion profiles, benefiting reservoir identification. These results will provide a basis for reservoir identification and prediction and provide theoretical support for AVO inversion in anisotropic media.

Automated summary

The study successfully inverted AVO data in orthotropic media through the combination of PSO and SA, introducing a hybrid SA-PSO algorithm that overcomes local minima and improves global optimization. The proposed method demonstrates stable inversion results with small errors, showing potential for reservoir identification and prediction in anisotropic media.