Predicting Daily Insolation with Hourly Cloud Height and Coverage

Solar radiation information is used in crop growth, boundary layer, entomological and plant pathological models, and in determining the potential use of active and passive solar energy systems. Yet solar radiation is among the least measured meteorological variables. A semi-physical model based on standard meteorological data was developed to estimate solar radiation received at the earth's surface. The radiation model includes the effects of Rayleigh scattering, absorption by water vapor and permanent gases, and absorption and scattering by aerosols and clouds. Cloud attenuation is accounted for by assigning transmission coefficients based on cloud height and amount. The cloud transmission coefficients for various heights and coverages were derived empirically from hourly observations of solar radiation in conjunction with corresponding cloud observations at West Lafayette, Indiana. The model was tested with independent data from West Lafayette and Indianapolis, Madison, WI, Omaha, NE, Columbia, MO, Nashville, TN, Seattle, WA, Los Angeles, CA, Phoenix, AZ, Lake Charles, LA, Miami, FL, and Sterling, VA. For each of these locations a 16% random sample of days was drawn within each of the 12 months in a year for testing the model. Excellent agreement between predicted and observed radiation values was obtained for all stations tested. Mean absolute errors ranged from 1.05 to 1.80 MJ m−2 day−1 and root-mean-square errors ranged from 1.31 to 2.32 MJ m−2 day−1. The model's performance judged by relative error was found to be independent of season and cloud amount for all locations tested.