Advances in geothermal energy prospectivity mapping research based on machine learning in the age of big data

Abstract

Many researchers and engineers are working diligently to develop new methods, materials, and technologies to solve various issues that arise in geothermal energy exploitation and utilization, including geothermal energy prospectivity mapping, exploration and drilling, geothermal well logging, and geothermal power generation. To address these challenges, big data and machine learning (ML) approaches are rapidly advancing in the field of geothermal energy exploitation and utilization. This investigation presents aspects of the progress of research into geothermal energy prospectivity mapping in the age of big data and ML. A detailed summary of geothermal energy distribution factors, spatial data analysis issues and techniques, and modeling methods' pros and cons is presented, and the distinction between conventional and ML-enhanced play fairway analysis (ePFA) in geothermal energy prospectivity mapping is highlighted. The case study results indicate that, compared with conventional play fairway analysis, ePFA can more effectively and accurately analyze site data, extracting hidden features, particularly when applied to geothermal energy discovery, characterization, and production in the Texas region, USA. It concludes that the advancement and successful integration of ML-enhanced methods in geothermal energy prospectivity mapping should be encouraged by policymakers to improve the accuracy of geothermal energy prospectivity mapping for stakeholder decision-making, potentially leading to reduced time, cost, and risk in geothermal energy exploitation and utilization.