Estimation of P- and S-wave impedances using Bayesian inversion and adaptive neuro-fuzzy inference system from a carbonate reservoir in Iran

Abstract

P- and S-wave impedances are accounted as two significant parameters conventionally inverted from seismic amplitudes for evaluation of gas and oil reservoirs. They may not be the final goal of interpretation studies; however, they play an important role in many methods such as reservoir characterization, rock physical modeling, geostatistical simulation, fluid detection. Bayesian inversion is a conventional method used by many researchers and even by industry to invert these parameters. To compare this method with intelligent methods, the adaptive network-based fuzzy inference system (ANFIS) was utilized to construct a model for the prediction of P- and S-wave impedances. Two ANFIS models were implemented, subtractive clustering method (SCM) and fuzzy c-means clustering method. The prediction capabilities offered by ANFIS models were shown by using field data obtained from a carbonate reservoir in Iran. Unlike other studies, input parameters, in this study, are pre-stack seismic data and attributes, while the P- and S-wave impedances are the output parameters in all methods. Mean square error was used for comparison of the performance of those models. The obtained results show that the ANFIS-SCM model generates the best indirect estimation of P- and S-wave impedances with high degree of accuracy and robustness.