Net shortwave radiation (NSR) plays an essential role in surface energy budget and has been reported to exhibit low estimation accuracy when ignoring topographic effects. Several algorithms have been developed to estimate clear-sky NSR in mountains, but none have estimated NSR in mountainous terrain under all-sky (i.e., both clear and cloudy) conditions. Moreover, surface heterogeneity is considerable in mountains, and coarse-spatial-resolution satellites may not provide sufficient details for fine-scale applications. In this study, to address these challenges, we developed a method for fine-spatial-resolution (i.e., from 20 m to 30 m resolution) all-sky NSR estimation in mountains directly from Landsat 8 and Sentinel-2 top-of-atmosphere (TOA) observations with mountainous radiative transfer scheme (herein denoted ‘TOPO’). For comparison, we also created the model neglecting topography (herein called ‘FLAT’). The validation of the TOPO model in complex terrain against Chengde ground-based pyrometers in China was good, with an R2 of 0.91 and a root-mean-square error (RMSE) of 80.9 W/m2, which was an improvement of 46.1% (reduction of RMSE) against the FLAT model. Meanwhile, the consistency of NSR generated from Landsat 8 and Sentinel-2 using TOPO model was satisfactory. Time-series comparisons of NSR derived from TOPO and FLAT models showed that topographic effects were substantial both in clear sky (maximum deviation of 643.4 W/m2) and cloudy sky (270.2 W/m2) when NSR estimates from the two models were compared. The evaluation of the FLAT model with a dataset simulated from the mountainous radiative transfer (created for the TOPO model) showed that neglecting topographic effects over mountainous terrain could lead to RMSEs of 541.7 W/m2 (relative RMSE = 159.5%) and 201.4 W/m2 (relative RMSE = 217.1%) for NSR estimation under clear and cloudy skies, respectively. This study provides a straightforward method of deriving fine-spatial-resolution all-sky NSR with topographic consideration, which can be further extended to the global mapping of NSR over mountainous terrain.
Read full abstract