Accurate simulation of surface solar radiation (SSR) remains challenging in regions with sparse observations and complex terrain. This study proposed a meteorologically driven semi-empirical model (Semi-Model) for daily SSR estimation in temperate regions, fully implemented on the Google Earth Engine (GEE) platform. The model integrated potential clear-sky radiation with terrain effects and cloud macro-/micro-physical properties to capture key drivers of SSR variability. We further developed GEE.SVF tool that quantifies terrain shielding effects, thereby improving the estimation of sky diffuse radiation and terrain-reflected radiation. Evaluation at Baseline Surface Radiation Network (BSRN) sites showed high accuracy: the mean error of −0.03 ×10 3 KJ/m 2 (10.21%) with a RMSE of 3.31 ×10 3 KJ/m 2 . In complex terrain, the RMSE remained within 1.10–2.49 ×10 3 KJ/m 2 . Compared with the widely used Black model, the Semi-Model reduced RMSE by up to 1.12 ×10 3 KJ/m 2 . We also compared Semi-Model with a machine-learning model (XGBoost). The Semi-Model matched or outperformed XGBoost in accuracy across BSRN and China Meteorological Administration stations. The model effectively quantified the influences of clouds and terrain and exhibits strong transferability across temperate climate zones. These findings highlighted a practical, scalable approach for solar resource assessment and planning in data-scarce, topographically heterogeneous regions. • GEE.SVF, a terrain obstruction quantification tool, was developed on the Google Earth Engine (GEE) Platform. • An empirical model accounting for both cloud and terrain effects was developed for estimating all-sky solar radiation in temperate regions. • This model exhibits high accuracy and robustness across complex terrains and diverse weather conditions, while also offering superior usability.
Zhang et al. (Wed,) studied this question.