Assimilating remote sensing data into GIS-based all sky solar radiation modeling for mountain terrain

Abstract

Solar radiation is the ultimate energy resource of earth surface energy balance and the main driving force of atmospheric, ecological and hydrological processes. Solar radiation over complex terrain has large spatial and temporal variation because of terrain shading and high cloud heterogeneity. While most existing GIS-based solar radiation models only work under clear sky condition, this research presents a solar radiation model which considers both terrain shading and anisotropic cloud attenuation and diffuse radiation using MODIS atmospheric products. Specifically, we use skyshed map, sky cloud maps, and sky weight map to represent angular distribution of sky obstruction, anisotropic cloud properties, and diffuse radiance over hemispherical sky, respectively. Combining skyshed map, sky weight map and sky cloud maps, we develop a solar radiation model where 3D geometrical relationships among sun, cloud, and terrain are considered and anisotropic diffuse radiance and cloud attenuation are modeled. Model results are evaluated using field observations in the Kunlun Mountains of western China. At Terra and Aqua overpass time, our model performs well with a mean relative bias (MRB) of −0.2%. It underestimates in clear and partly cloudy sky with a MRB of −5.86% and −4.79% and has a mean absolute relative bias (MARB) of 8.11% and 21.59% respectively. It overestimates under overcast sky with a MRB of 1.68% and has a MARB of 31.71%. Our model performs better when compared with existing instantaneous solar radiation products. For daily solar radiation, our model shows good performance with a MRB of 1.43% and MARB of 17.02%. Our model shows significant spatial variation of solar radiation within the study area with the influences from terrain and cloud. Our research provides a novel and improved approach to assimilating remote sensing data into GIS-based solar radiation modeling in mountainous terrain where observations are sparse and difficult to obtain.