Machine Learning in Triassic Paleogeography for Fossil Fuel Extraction Site Discovery

This study utilizes machine learning algorithms to determine the coordinates of potential fossil fuel reserves by associating existing fossil fuel fields with the geological structure of the Triassic period. A method has been developed to convert the current coordinates of the fields to those of the Triassic period, and the characteristics of the fields during the Triassic period have been extracted from two-dimensional maps using computer vision techniques. K-nearest neighbors, random forest, and XGBoost models were trained, and the best model predicted the presence of fossil fuels in the fields with 88% accuracy. The interdisciplinary approach employed demonstrates the effectiveness of machine learning in geo-spatial analysis and opens new avenues for improving the management of energy resources.