Machine learning for fault-controlled geothermal systems exploration in Chenzhou and Huizhou region, Southeast China

This study investigates fault-controlled geothermal systems in southeastern China, focusing on representative regions in Hunan, Guangdong, and Jiangxi provinces. A machine learning approach—non-negative matrix factorization with k-means clustering (NMFk)—was applied to classify geothermal water types and delineate favorable exploration zones based on hydrochemical composition, flow rate, heat flow, and fault proximity. Geothermal waters were classified into three types with distinct geochemical and geological attributes: Type A, Type B, and Type C. Representative geothermal fields—Nuanshui (Type A), Longmen and Reshui (Type B), and Chengkou (Type C)—were selected to validate the classification and analyze reservoir characteristics and genetic mechanisms. Type B geothermal water exhibits the highest exploration potential, characterized by deep circulation (1900–5300 m), high reservoir temperatures (66–143 °C), strong confinement, and enrichment in Na⁺ and Li⁺. Its formation is primarily controlled by NW-trending faults and high heat-producing granites. Type C geothermal water shows moderate potential, with the highest heat flow (83 mW/m²), deep circulation (3500–5400 m), and elevated temperatures (109–127 °C), despite lower flow rates. It is hosted in granitic reservoirs associated with NE–N-trending faults. In contrast, Type A demonstrates the lowest geothermal potential, featuring shallow circulation (900–2100 m), lower temperatures (42–75 °C), high flow rates, and enrichment in Mg²⁺, Ca²⁺, and Sr²⁺, reflecting strong meteoric recharge and limited geochemical evolution. A conceptual model is proposed in which meteoric water infiltrates through fault zones, absorbs heat during deep circulation within granitic or carbonate rocks, and ascends to form geothermal reservoirs or surface springs. The classification results align well with spatial patterns of geothermal favorability, offering a robust framework for geothermal resource assessment and supporting sustainable development strategies in southeastern China.