Associations of Urbanization Intensity, Ecosystem Services, and Climate with Land Surface Temperature and PM2.5 in the Yellow River Basin (China): A Spatial Regression and XGBoost–SHAP Analysis

Understanding how urbanization, ecosystem-service conditions, and climate are associated with environmental responses is important for regional land management. Using three discrete county-level snapshots from the Yellow River Basin in 2013, 2018, and 2023, this study examined two environmental response variables: land surface temperature (LST) and the concentration of fine particulate matter with an aerodynamic diameter of 2.5 μm or less (PM2.5), in relation to urbanization intensity (UII), carbon storage (CS), water yield (WY), soil retention (SR), air temperature (T), and wind speed (WS). Pearson correlation analysis, multiple linear regression, and extreme gradient boosting (XGBoost) combined with SHapley Additive exPlanations (SHAP) were used to characterize linear and nonlinear associations. To reduce spatial leakage, model performance was evaluated using nested five-fold buffered spatial block validation, in which contiguous spatial blocks were held out, and counties within 50 km of each test block were excluded from model training. The pooled out-of-block R2 values ranged from 0.765 to 0.777 for LST and from 0.728 to 0.854 for PM2.5. T was the dominant predictor of both responses, accounting for 79.26–86.62% of the total SHAP importance for LST and 53.12–65.99% for PM2.5. UII showed positive model-based associations with both responses that generally appeared to level off at higher observed values, whereas the associations of ecosystem-service variables differed among the three independently analyzed years. Because the three years were modelled separately, these cross-year contrasts are interpreted descriptively rather than as continuous temporal trends. These findings characterize contrasting explanatory structures for surface thermal conditions and particulate pollution and may inform the formulation of hypotheses for spatially differentiated environmental management in the Yellow River Basin.