Evaluation of satellite and reanalysis products of downward surface solar radiation over East Asia: Spatial and seasonal variations

[1] Surface solar radiation plays a crucial role in surface energy and water budgets, and it is also an important forcing for land hydrological models. In this study, the downward surface solar radiation (DSSR) from two satellite products, the Fengyun-2C satellite (FY-2C) and the Fast Longwave and Shortwave Radiative Fluxes project (FLASHFlux), and two reanalysis datasets, NCEP-DOE and ERA-Interim, was evaluated against ground-based observations (OBS) from 94 stations over mainland China during July 2006 to June 2009. It is found that the mean DSSR derived from FY-2C is comparable to OBS, with small positive biases of 3.0 Wm for daily data and 3.5 Wm for monthly data and moderate RMSEs of 49.3 Wm (daily) and 31.9 Wm (monthly). These results are comparable to those for FLASHFlux, which has the lowest RMSEs (43.2 Wm and 30.5 Wm for daily and monthly data, respectively) and the strongest correlations with OBS (r = 0.90 and 0.93 for daily and monthly data, respectively) among the four products. The DSSR from the reanalyses has much larger RMSEs and generally lower correlations with OBS than the satellite products, especially for the NCEP-DOE products. Results also show that daily DSSR values are sensitive to the averaging grid size, while monthly mean DSSR is largely insensitive to the averaging scale. The DSSR from the four datasets over East Asia shows similar spatial patterns with large seasonal variations but differs in magnitude. In summer, high DSSR is observed over western China, while low DSSR is seen primarily over South Asia and the Sichuan Basin associated with extensive cloud cover (CC) and large precipitable water (PW). In winter, the high DSSR center shifts to South Asia due to decreased CC and PW, and the DSSR decreases from the South to the North. Deficiencies in the parameterizations of clouds, aerosols, and water vapor, as well as errors in atmospheric and surface properties for the retrieval algorithms contribute to the lower correlation of the DSSR derived from FY-2C (r = 0.82 and 0.90 for daily and monthly data) with OBS than those from FLASHFlux product. Further improvements to the representation of clouds and aerosols in the FY-2C retrieval algorithm are needed.