Evaluating the potential distribution of deep geothermal resources in the North Jiangsu Basin, East China, using machine learning

Abstract

Substantial reserves of deep geothermal resources (DGRs) in the North Jiangsu Basin (NJB) offer considerable potential for energy supply in eastern China. However, the complicated geological structure and miscellaneous influencing factors associated with DGRs present challenges for ascertaining their potential distribution. In this paper, we introduce an enhanced evaluation index system, wherein nine indices are organized into three categories: geophysical presentations, tectonic and magmatic activities, and geothermal indicators, to highlight the distinctive characteristics of deep geothermal energy. Subsequently, we applied a machine learning approach, specifically the MaxEnt model, to quantify the probability distribution of DGRs within the NJB. The results demonstrate that the Jianhu Uplift, situated in the central region of the NJB, is the most favorable area for DGR development. In addition, the southwestern region of Huai'an, the northern area of Taizhou, and the eastern coastal zone of the basin were identified as primary potential areas for DGRs. The distribution of these promising areas was predominantly influenced by the distance from deep-large faults. The depth of high-conductivity and low-velocity bodies emerged as the second most significant factor, followed by the P-wave velocity distribution. Collectively, these three factors account for over 60 % of the impact on DGRs' distribution. These findings provide robust quantitative evidence for the optimization of favorable areas for DGR development. They also suggested that our methodology is effective in maximizing spatial distribution inference with limited data, offering considerable merits and promising prospects for geoscientific research in data-scarce environments.