Understanding regional structural and horizon deformation using machine learning: A case study Kwanza Basin, Angola

This study presents the successful application of machine learning (ML) using convolutional neural networks (CNNs) to predict extensive faults and consistent high-resolution salt and stratigraphic horizons on a massive ∼40,000 km2 complex Kwanza offshore dataset. ML assisted in predicting small-to-large scale fault surfaces that helped in the identification of various structural features, improved classification of different deformational styles, enhanced the differentiation of structural provinces, and better distinguished the prevailing stress and strain fields within the regional extent of Kwanza Basin. In addition, a separate CNN expedited the prediction of regional horizons that consistently tracked the proper events for the water bottom (WB), top of salt (TOS), base of salt (BOS), and stratigraphic horizons. These horizons were essential to characterize the horizon deformation, sediment deposition, and salt and sediment thickness distribution. Integrating the insights gained from the ML assisted faults and horizons advances a better understanding of the structural evolution of the Kwanza Basin, leading to the framework for a holistic basin-wide play and prospectivity evaluation.