A fusion method to accurately derive all-sky surface shortwave net radiation based on remotely sensed data

Accurate surface shortwave net radiation (SWNR) estimation is crucial for understanding surface energy balance and climate change. However, existing algorithms derive SWNR either indirectly from downward radiation and surface albedo or directly using the top-of-atmosphere (TOA) albedo, often leading to error propagation and insufficient representation of atmospheric effects, particularly under snow-covered and thick cloud conditions. Moreover, daily surface albedo products struggle to capture rapid intraday variations during snowfall and snowmelt. This study corrects MODIS surface albedo using two independent TOA albedo estimation schemes. Based on the corrected albedo, SWNR is estimated using a parameterization based on TOA albedo and a lookup table approach, and a combined estimate is obtained by averaging the two results. Validation against SURFRAD measurements shows that the corrected surface albedo substantially improves, with bias and RMSE decreasing from –0.217 and 0.397 to –0.036 and 0.177, respectively. This improvement extends to SWNR estimation, reducing bias by 5~30 W/m² and RMSE by 40~70 W/m². Evaluations using BSRN and SURFRAD demonstrate that the combined estimate consistently outperforms individual methods under diverse conditions, due to complementary bias characteristics. In addition, TOA albedo and SWNR exhibit enhanced spatial details at 5 km resolution and closely match MODIS true-color imagery.