Abstract
Surface incident solar radiation (R(s)) of reanalysis products is widely used in ecological conservation, agricultural production, civil engineering and various solar energy applications. It is of great importance to have a good knowledge of the uncertainty of reanalysis R(s) products. In this study, we evaluated the R(s) estimates from two representative global reanalysis (ERA-Interim and MERRA-2) using quality- controlled surface measurements from China Meteorological Administration (CMA) and Multi-layer Simulation and Data Assimilation Center of the Tibetan Plateau (DAM) from 2000 to 2009. Error causes are further analyzed in combination radiation products from the Earth's Radiant Energy System (CERES) EBAF through time series estimation, hotspot selection and Geodetector methods. Both the ERA-Interim and MERRA-2 products overestimate the R(s) in China, and the MERRA-2 overestimation is more pronounced. The errors of the ERA-Interim are greater in spring and winter, while that of the MERRA-2 are almost the same in all seasons. As more quality-controlled measurements were used for validation, the conclusions seem more reliable, thereby providing scientific reference for rational use of these datasets. It was also found that the main causes of errors are the cloud coverage in the southeast coastal area, aerosol optical depth (AOD) and water vapor content in the Sichuan Basin, and cloud coverage and AOD in the northeast and middle east of China.