Calibration and sensitivity analysis of a spatially-distributed solar radiation model

Abstract

SRAD, a spatially-distributed solar radiation model, was applied to a Canadian boreal forest environment in north-western Ontario. SRAD is grid based, and factors in both topo- and meso-scaled processes using a digital elevation model (DEM) and local monthly atmospheric parameters as inputs. SRAD generates estimates of incident, outgoing, and net irradiance, as well as surface and air temperatures for each point in the DEM, over any time period ranging from one day to one year. Cloudiness and other atmospheric conditions are factored into the shortwave irradiance estimates. From the DEM, the terrain effects of slope angle, aspect, and topographic shading are calculated and used to modify the estimates of shortwave irradiance. The SRAD-generated irradiance estimates for the study region were found to be consistent with irradiance data from other sources. Estimates of irradiance were most sensitive to the parameters of sunshine fraction and cloudiness. Radiation estimates were generated and compared using both a 20 m and a 100 m resolution DEM. Extremely low irradiance estimates generated at the fine scale were absent at the coarser scale. However the mean value of irradiance at both scales was estimated to be the same at 12.4 MJm-2 day-1 annually. Radiation estimates were also determined at the two scales for a series of forest research plots. For 91% of the plots, the irradiance estimates during the growing season differed by less than 0.5 MJm-2 day-1 between the two scales. Results presented here suggest that SRAD can provide useful input for predictive spatial models in boreal forest ecosystems.