Advanced digital techniques applied to outcrop models: Integrating Local Binary Pattern (LBP) and Convolutional Neural Network (CNN) to support stratigraphic and sedimentological interpretation of reservoir analogs in the Salta Basin, Argentina

Digital Outcrop Models (DOMs), empowered by advanced digital techniques, have revolutionized the study of outcrop analogs for petroleum reservoir characterization by enabling the extraction of key quantitative parameters for modeling. The limited availability of subsurface data often constrains reservoir characterization, making outcrop analogs essential tools for improving geological models. The analogs bridge the gap between borehole-derived information and regional-scale seismic data, providing crucial mesoscale insights. In this context, this study proposes an integrative workflow combining high-resolution sequence stratigraphy (HRSS) with digital techniques to enhance the understanding of depositional settings and extract data from the Balbuena III Sequence of the Salta Basin, Argentina, a well-established stratigraphic basin analog for Brazilian pre-salt carbonate reservoirs. The workflow combines traditional field-based methods with advanced digital techniques applied to photogrammetric data, including Local Binary Pattern (LBP) analysis and Convolutional Neural Networks (CNNs). LBP analysis correlated with stratigraphic interpretation demonstrated promising potential for characterizing the high-frequency cyclicity observed in the study area. CNN-based segmentation classified and delineated eleven lithofacies, including carbonate, siliciclastic, mixed, and volcanic facies. This segmentation allows for the generation of lithofacies-classified 3D point clouds and a detailed spatial representation of facies distribution across the outcrop. Digital approaches enable more in-depth analysis by increasing efficiency, accuracy, and the capacity to analyze large datasets. By combining digital and traditional methods, this work improves the analysis of outcrop analogs, which contributes to more accurate geological modeling and enhances the predictive capability of petroleum fields and hydrocarbon recovery.