CR 29:91-102 (2005)  -  doi:10.3354/cr029091

ABSTRACT: The availability of daily solar radiation (SR) data has been limited until recent years, restricting modeling applications when long-term data are required. The main goal of this study was to improve the WGENR (Weather Generator) SR generator and to evaluate its performance with recorded SR data for the southeastern USA. WGENR includes adjustment factors (AF) for obtaining proper average annual daily SR during dry (RMD) and wet (RMW) days and for amplification of its amplitude during dry and wet days. Thus, 3 sets of AFs, related to the occurrence of dry and wet days and precipitation greater than zero, were evaluated. Recorded SR data from 17 weather stations located in South Georgia, with records ranging from 1992 to 2003, were used for evaluation. The chi-square (χ²) goodness-of-fit and the Kolmogorov-Smirnov (KS) tests were applied to analyze the frequency distributions and to determine significant differences between the cumulative distribution functions (CDFs) of generated and recorded SR. For most of the locations, generated SR differed from recorded data but improvements for RMD and RMW were obtained. Based on the χ² and the smallest difference between CDFs for most of the locations, a final set of AFs was selected that provided an overall improvement of the WGENR solar radiation generator. Further work will focus on the impact of these changes on decision support systems, specifically yield prediction with crop models using generated SR.

KEY WORDS: Weather generator · Automated weather stations · Decision Support System · Computer simulation model · Natural resource management