As a core essential climate variable (ECV) in the Global Climate Observing System, land surface temperature (LST) plays a pivotal role in climate change monitoring, urban thermal environment assessment, agricultural management, and ecosystem surveillance. To obtain high-precision LST data, a novel machine learning-based LST downscaling framework integrating feature interaction optimization and Shapley additive explanations (SHAP) scoring was proposed. SHAP scoring was employed for feature importance analysis to identify optimal predictors, while 10 distinct models were comparatively evaluated to establish a high-resolution downscaling framework adaptable to homogeneous surface characteristics. The results show that the SHAP-based feature selection significantly enhanced prediction accuracy by prioritizing nonlinear determinants. The red-blue band interaction feature demonstrated consistent dominance across all algorithms (XGBoost, LightGBM, GradientBoost), exhibiting both the broadest SHAP value range (−2.0 to 2.0) and the highest relative contribution weight. By explicitly addressing spatial heterogeneity, the spatial random forest (SRF) model achieved superior downscaling performance, particularly in vegetated regions. It generated reliable 10 m-resolution LST estimates (R² = 0.74, RMSE = 6.28℃), demonstrating robust generalization capabilities in complex terrain conditions. The SHAP-based land surface temperature downscaling method can effectively capture the nonlinear interactions among spectral, topographic, and other features, demonstrating high accuracy and strong physical interpretability in high-resolution temperature retrieval over areas dominated by homogeneous vegetation.

Changes in precipitation extremes and associated risks under the 1.5 and 2.0°C global warming targets in the Yangtze River basin (YRB) were assessed. The projections from 10 global climate models (GCMs) of the Coupled Model Intercomparison Project phase 6 (CMIP6) were bias-corrected and averaged with Bayesian and arithmetic mean methods, respectively. The results show that the Bayesian weights can reflect the performance of each GCM in capturing seasonal precipitation extremes. Thus, its multimodel ensemble projections noticeably improve the performance of the mean, interannual variability, and trends of precipitation extremes. The areal-mean risks of Rx5day (maximum consecutive 5-day precipitation) are projected to increase by ratios of 3.3 in summer, 2.9 in autumn, 2.2 in spring, and 1.9 in winter under the 1.5°C target. Spatially, the northwestern part of the YRB may experience the highest risk of increments in Rx5day extreme in summer and autumn. In response to an additional 0.5°C warming from 1.5 to 2.0°C, the risks of seasonal Rx5day extreme for all 20-, 50-, and 100-yr return periods are projected to increase respectively. The higher probabilities of extreme precipitation events under the warming targets may cause more hazardous flooding; therefore, new strategies and infrastructures for climate change and hydrological risk mitigation are imperative in the YRB.