Physics-Guided Machine Learning with Deterministic-Stochastic Signal Decomposition and XGBoost Residual Correction for Deep Reservoir Thermal Stratification Forecasting: A Case Study of Miyun Reservoir, the Largest Drinking Water Source Reservoir in North China

Background (from the associated paper): Predicting thermal stratification in deep reservoirs is challenging: 3D hydrodynamic models demand intensive computation, while pure deep learning often systematically overestimates hypolimnetic temperatures. To address this, we developed a physics-guided surrogate framework for Miyun Reservoir. The data provided here (CE-QUAL-R1 simulations from 2014–2022) were used to train and validate that framework.