Potential Zoning and Target Optimization of Geothermal Resources in the Eastern Margin of the Qinghai–Xizang Plateau: Multi-Source Data GIS Modeling Based on Fusion Machine Learning

The eastern margin of the Qinghai–Xizang Plateau (QXP) represents one of the regions with the most significant and concentrated high-temperature geothermal potential in China. Previous studies have primarily focused on localized geothermal belts, such as the Xianshui River and Litang zones, with limited comparative analysis of distinct geothermal systems distributed across the eastern margin of the QXP. The prediction results of a single GIS model are discrete and inefficient, while machine learning approaches rely on data and ignore the inherent logical relationship. It remains a key issue to be urgently solved to realize the prediction of regional geothermal resource potential by combining the advantages of the two models and build a more reliable spatial intelligent analysis framework. Therefore, based on systematic investigations of geothermal geology and genesis mechanisms along the eastern margin of the QXP, indicators such as faults, terrestrial heat flow, and hot spring heat flux were selected. By integrating GIS-based spatial analysis with intelligent ensemble learning algorithms, a comprehensive geothermal resource potential prediction method was developed, enhancing the ability to model nonlinear relationships and significantly improves prediction accuracy under complex geological conditions. This GIS-integrated learning model has identified four zones of extremely high geothermal potential (EHp): Xianshui River EHp, Litang EHp, Sanjiang EHp, and the Eastern Himalayan Syntaxis EHp. Geothermal exploration target areas were delineated by identifying the dominant controlling factors of different EHp, providing both a scientific basis and strategic guidance for the exploration of geothermal resources along the eastern margin of the QXP.