CR prepress abstract  -  doi: 10.3354/cr00750

ABSTRACT: Lack of site-specific daily solar radiation (Rg) data is a significant impediment for most crop modeling applications. For this reason, five methods for estimating Rg were tested: the Ångström-Prescott equation (AP), the Supit - van Kappel formula (SK) and three temperature based methods (the Campbell-Donatelli model, the Hargreaves equation and piecewise multiple linear regressions with a breakpoint (PLR)). To overcome the lack of long and continuous time series of Rg measurements at multiple sites, satellite-derived Rg from HelioClim-1 database, after having been tested against observations from two stations, was interpolated at 12 additional stations. When sunshine duration data were available, the AP equation was preferred because it (1) produces the closest to zero and unity intercepts and slopes, respectively and (2) shows the lowest relative RMSE (varied from 9% to 18.6%). When cloud cover observations and maximum and minimum temperatures exist, SK equation is proposed, as it is equally efficient as AP in most assessment metrics. If only daily maximum and minimum temperatures exist on-site, the PLR approach with a breakpoint, which reflects the value at which the behavior of solar radiation changes as a function of the extraterrestrial Rg and the diurnal temperature range, is suggested. Compared with AP, PLR approach increased the mean relative RMSE by less than 3.7% and decreased d from 0.98 to 0.97. SK equation was the most suitable substitute for measured Rg for the CERES-Wheat crop model, while among the temperature-based methods PLR produced the smallest yield errors.