Geophysical prediction technology for sweet spots of continental shale oil: A case study of the Lianggaoshan Formation, Sichuan Basin, China

•High-precision data volume of parameters was derived from 3D seismic data using the nonlinear pre-stack AVO inversion methodology.•It correlated elastic parameters with sweet spot evaluation metrics through petrophysical analyses, predicting the planar distributions of critical evaluation parameters, such as total organic carbon (TOC) content, reservoir porosity, high-quality reservoirs’ thickness, brittleness of strata, and fractures.•This study built a comprehensive evaluation model using the random forest regression algorithm, achieving the seismic prediction and comprehensive evaluation of the sweet spots of continental shale oil. China possesses significant potential shale oil resources, primarily of continental shale oil. Compared with marine shale oil, continental shale oil reservoirs have more complex lithology and stronger heterogeneity, which leads to unclear geophysical response characteristics of shale oil sweet spots, and it is difficult to optimize and evaluate sweet spots. Therefore, it is urgent to study effective seismic prediction technology of continental shale oil sweet spots to support the scale and efficiency development of shale oil. In this study, based on logging interpretation and seismic rock physics analysis, the geophysical response characteristics of reservoir sweet spots are determined, and the sensitive elastic parameters are optimized. High-precision data volume of these parameters was derived from 3D seismic data using the nonlinear pre-stack AVO (Amplitude variation with offset) inversion methodology. With the data volume, this study correlated the elastic parameters with the key evaluation parameters of shale oil sweet spots, predicting the spatial distributions of key evaluation parameters, such as total organic carbon (TOC) content, reservoir porosity, high-quality reservoirs’ thickness, formation brittleness, and fractures. Finally, the comprehensive evaluation model is established using the random forest regression algorithm, and the quantitative prediction and classification evaluation of continental shale oil sweet spots are realized. The nonlinear inverse weighted AVO pre-stack inversion method and the random forest shale oil dessert comprehensive evaluation method proposed in this study have realized the high-precision quantitative prediction and classification evaluation of continental shale oil sweet spots. Our methodology effectively categorized three classes of shale oil sweet spots in the Lianggaoshan Forma