Accurate projections of surface downward shortwave radiation (RSDS) are essential for understanding regional energy balance and assessing solar energy potential. However, substantial systematic biases remain in CMIP6 models, particularly over regions with complex topography and large climatic contrasts. Using ERA5 reanalysis and 16 CMIP6 models, we constrain future RSDS projections over the Silk Road Economic Belt (SREB) via the Ensemble Pattern Regression (EPR) method. EOF analysis identifies the dominant spatial modes of historical model bias, revealing widespread overestimation across much of the study region and persistent underestimation in Western Europe and Southeast Asia. By linking historical model performance to future change biases, the EPR constraint reduces projection spread and restores spatial gradients that are often weakened in conventional multi-model means. It reduces the inter-model projection spread by 30%–50% across all scenarios and periods, with local reductions exceeding 75% in the mid-to-low latitudes, and enhances spatial gradients by about 45% in most scenarios relative to the conventional multi-model mean. The constrained projections show that regions with high solar energy potential remain concentrated in the Arabian Peninsula, the Iranian Plateau, and northwestern China, indicating strong latitudinal and topographic controls. Solar energy resources generally increase under future climate change, although the magnitude of change does not vary monotonically across emission scenarios. Overall, the results indicate that the pattern-based EPR constraint enhances the reliability of regional RSDS projections over complex terrains. This improvement strengthens the robustness of regional solar energy assessments and provides a more reliable climatic basis for renewable energy planning.
LIU et al. (Wed,) studied this question.