Identification of challenging gas-bearing reservoir based on machine learning (ML) and computed conversion-based AVO analysis: a study from Jaisalmer Sub-basin, India

Abstract

AbstractAmplitude variation with offset (AVO) analysis is an important tool for identifying natural gas-bearing reservoirs. The changes in seismic amplitudes based on the variation of density and velocity of the rock matrix are captured through the AVO analysis. The current work was performed in the Ghotaru region of the Jaisalmer Sub-basin area, where limited and discrete hydrocarbon discoveries were observed from the Lower Goru Formation during the earlier various exploration campaigns. The discrete nature of the reservoir lithofacies developed challenging scenarios for the successful exploratory campaign. The campaign encountered more difficulties because of limited advanced datasets, which affected the study to capture the extension of hydrocarbon-bearing reservoir lithofacies and its characterization towards a successful exploration campaign. This study shows the way to overcome these challenges using an existing conventional dataset. The study shows the possibility of AVO analysis using a post-stack seismic dataset. A unique conversion method from post-stack to pre-stack seismic is introduced in this study based on the uses of the integrated velocity model. An integrated, robust velocity model was developed with consideration of anisotropy factors. Introducing a machine learning-based algorithm in the petrophysical study, including the conventional approach, provides a robust validation of this work. Intercept (A) and gradient (B) are the basic outcome of AVO analysis. The well-based study and AVO analysis based on intercept (A) and gradient (B) complement each other for finding hydrocarbon-bearing reservoir rock. Cross-plots and AVO analysis show the reservoir's lithofacies extension and fluids. The study reveals the potential of natural gas retained in the Lower Goru Formation, which is composed of patchy sandstone. Two AVO classes (Class I and Class III) of gas-bearing sandstone have been identified in this study. This study presents a unique method for identifying natural gas reservoirs with limited old conventional data.