Machine Learning Techniques in Predicting Bottom Hole Temperature and Remote Sensing for Assessment of Geothermal Potential in the Kingdom of Saudi Arabia

The global demand for energy is increasing rapidly due to population growth, urbanization, and industrialization, as well as to meet the desire for a higher standard of living. However, environmental concerns, such as air pollution from fossil fuels, are becoming limiting factors for energy sources. Therefore, the appropriate and sustainable solution is to transition towards renewable energy sources to meet global energy demands by using environmentally friendly sources, such as geothermal. The Harrat Rahat volcanic field, located in the western region of the Kingdom of Saudi Arabia (KSA), gets more attention due to its geothermal potential as a viable site for geothermal energy exploration due to its high enthalpy. The prime objective of this study is to present up-to-date and comprehensive information on the utilization of borehole temperature and remote sensing data to identify the most prospective zones with significant geothermal activity favorable for exploration and drilling. A brief description of the selected wells and the methodology used to determine the petrophysical parameters relevant to the geothermal potential assessment are presented. Special emphasis is given to gamma-ray ray and temperature logs for calculating heat production and the geothermal gradient. The effectiveness of various machine learning techniques are assessed throughout this study for predicting the temperature-at-depth to evaluate the suitability of employing machine learning models for temperature prediction, and it is found that XG Boost provided excellent results. It can be observed that some linear anomalies can be traced in the NW, trending on the west side of the Harrat volcanic field based on magnetic data interpretation. The land surface temperature in 2021 exhibited higher temperatures compared to 2000, suggesting potential volcanic activity in the subsurface. It is concluded that the integration of remote sensing data with subsurface data provides the most reliable results.