An Evaluation of Satellite-based Insolation Estimates for Ohio

Abstract

The National Environmental Satellite, Data and Information Service (NESDIS) has developed an operational procedure to estimate hourly and daily totals of global solar radiation (insolation) from geostationary operational environmental satellite (GOES) observations. Production of the insolation estimates began for the eastern two-thirds of the United States in the summer of 1980 and is expected to continue through 1986. The NESDIS estimates comprise a unique set of insolation data, but their accuracy needs to be assessed for atmospheric and surface conditions which differ from those for the Great Plains. Accuracy of 1982 NFSDIS daily estimates for Ohio was assessed through comparison with insolation measurements from eight high-quality stations. The NESDIS estimates are for target areas of approximately 50 ? 50 km centered on the intersection of 1° lines of latitude and longitude. In this study, only estimates for the target centered closest to each station were compared with measurements from that station. For the snow-free season, men errors (bias) of the estimates for all days are positive but generally less than 0.75 MJ m?2 day?1; both standard deviations of the errors and root-mean-square (rms) errors are generally less than 2.25 MJ m?2 day?1; and correlation coefficients generally exceed 0.95. Expressed as a percentage of mean-measured insolation, bias is generally less than +3.5% and both standard deviations and rms errors are generally less than 12%. For clear days, bias is approximately ?3% of mean insolation and the rms and standard deviations of the errors are less than 6% of mean insolation. Bias is unaffected by distance between the station and the center of the target area for which the estimates were made, but rms and standard deviations of the errors increase and correlation coefficients decrease as the separation distance increases. Estimates are much leer accurate when a snow cover existed. Generally, the bias is negative and exceeds ?1.25 MJ m?2 day?1, standard deviation of the errors exceed 2.70 MJ m?2 day?1, rms errors exceed 3.00 MJ m?2 day?1, and correlation coefficients are less than 0.90. Expressed as a percentage of mean insolation, bias generally exceeds ?15%, standard deviations of the error exceed 30%, and rms errors exceed 35%.