Generating Station-Like Downward Shortwave Radiation Data by Using Sky Condition-Guided Model Based on ERA5-Land Data

Accurate downward shortwave radiation (DSR) data is essential for the utilization of solar energy resources. However, existing various DSR data have certain limitations in applications. Among these, ERA5-Land DSR data is considered a relatively high-quality dataset. Nevertheless, it is reported notable deviations under different sky conditions, resulted from deficiencies in cloud and aerosol simulations. Therefore, this study aims to generate high-quality DSR data based on ERA5-Land data. We propose a sky condition-guided model using deviation characteristics of ERA5-Land DSR across different sky conditions as guidance, incorporating reliable cloud and aerosol parameters to provide atmosphere information. Consequently, station-like DSR data with accuracy close to station data is generated. Our model achieves high accuracy under different sky conditions, with RMSE below 32.19 W/m 2 . At the individual-station scale, 86.04 % of validation sites exhibit R values exceeding 0.9. At the seasonal scale, R values consistently surpass 0.87. Moreover, our data exhibits better accuracy compared to mainstream DSR datasets. Furthermore, it depicts the spatiotemporal variation of DSR and further reveals the photovoltaic potential in China, with relatively high values in the Tibetan Plateau and northwest China. The station-like DSR data is promising to advance the utilization of solar energy and accelerate the energy transition.