Pre-stack Seismic Probabilistic Inversion Method for Lithofacies and Elastic Parameters of Volcanic Reservoir

Abstract

Seismic inversion is the primary way to obtain subsurface models, lithologic and stratigraphic information. However, seismic elastic parameters inversion and ‘discrete lithofacies’ identification for complex volcanic reservoirs are usually independent during the whole inversion process. Also, the influence of reservoir lithology on elastic parameters is not always considered directly before lithofacies prediction. This paper proposes a probabilistic pre-stack seismic inversion method for lithofacies and elastic parameters of volcanic reservoirs. Under the framework of Bayesian inversion, considering that the prior probability distribution of elastic parameters of volcanic reservoirs is affected by volcanic lithofacies, a posteriori probability distribution characterized by a mixed probability model is first derived. Then, a single-point-direct sequential simulation stochastic algorithm with simultaneous optimization of multiple solutions is used to simulate the posterior probability distribution of elastic parameters and lithofacies of volcanic reservoirs, which improves the resolution of lithofacies prediction results of volcanic reservoirs. The feasibility and stability of our method are ensured through synthetic and field applications. The prediction results highly agree with logging curves and lithology logging interpretation data. We have improved the resolution of volcanic rock reservoir lithofacies prediction results. In one-dimensional tests, we achieved the prediction of lithofacies and elastic parameters for three types of volcanic lithofacies. The error compared to prior information is no higher than 15%, thereby verifying the method’s good noise resistance.