Analysis of surface incident shortwave radiation from four satellite products

Abstract

Incident solar radiation (R<inf>s</inf>) over the Earth's surface is important for studying our climate and environment. Global observation networks have been established, but many land surfaces are under-represented. Satellite remote sensing is the only way to estimate R<inf>s</inf> at both global and regional scales. Many efforts have been made to evaluate the accuracy of current R<inf>s</inf> products generated from satellite observations, but only a limited amount of ground measurements was generally used and the individual satellite products were used for analyzing R<inf>s</inf> variability. In this study, four satellite estimates of R<inf>s</inf>, including the Global Energy and Water Cycle Experiment - Surface Radiation Budget (GEWEX-SRB V3.0), the International Satellite Cloud Climatology Project - Flux Data (ISCCP-FD), the University of Maryland (UMD)/Shortwave Radiation Budget (SRB) (UMD-SRB V3.3.3) product, and the Earth's Radiant Energy System (CERES) EBAF, were evaluated using comprehensive ground measurements at 1151 sites around the world from the Global Energy Balance Archive (GEBA) and the China Meteorological Administration (CMA). It was found that these satellite estimates of R<inf>s</inf> agree better with surface measurements at monthly than at daily time scale and can capture the seasonal variation of R<inf>s</inf> very well, but these satellite products overestimated R<inf>s</inf> by approximately 10wm^-2. The mean bias and the root mean square error (RMSE) of the monthly mean estimates from these four data sets were 10.2wm^-2 and 24.8wm^-2 respectively. The global annual mean values of R<inf>s</inf> were 186.7wm^-2, 185.4wm^-2, and 188.6wm^-2 for CERES-EBAF, ISCCP-FD, and GEWEX-SRB V3.0 respectively. The averaged global annual mean R<inf>s</inf> value from ground-measured-calibrated three satellite derived R<inf>s</inf> products was 180.6wm^-2, which is smaller than that estimated from individual satellite-derived products. The CERES-EBAF product shows the best accuracy among these four data sets, which indicates that including more accurate cloud information from active instruments can improve the accuracy of R<inf>s</inf>. These satellite products show different temporal trends. Both GEWEX-SRB V3.0 and ISCCP-FD showed similar trends at the global scale but with different magnitudes. A significant dimming was found between 1984 and 1991, followed by brightening from 1992 to 2000, and then by a significant dimming over 2001-2007. The CERES-EBAF product showed a brightening trend, but not significantly since 2000. The variability from satellite estimates at pixel level was also analyzed. The results are comparable with previous studies based on observed R<inf>s</inf> at the surface for specific regions, although some inconsistencies still exist and the magnitudes of the variations should be further quantified. We also found that clouds contribute more to the long-term variations of R<inf>s</inf> derived from satellite observations than aerosols.