Articles
PREDICTION OF SOLAR RADIATION FROM AIR TEMPERATURE
Article number
593_27
Pages
209 – 217
Language
English
Abstract
Radiation algorithms, with the aid of crop simulation models, will be useful for crop yield predictions, particularly in locations where measured radiation data are not available.
Several empirical functions were incorporated into radiation algorithms to predict solar radiation from measured maximum and minimum air temperatures (Tmax and Tmin). Each function was tested with historical weather data from five locations: Stoneville, Mississippi; Meridianville, Alabama; Shafter, California; Corpus Christi, Texas; and Florence, South Carolina.
Monthly mean values of each function’s coefficients were derived, from which daily values were obtained via cubic spline interpolation.
Calibrated with Mississippi weather data, each function predicted solar radiation from measured Tmax and Tmin data for each location.
Predictions were compared to measured radiation for validation, and the best functions were identified.
Agreement was obtained between predicted and measured radiation (r2 ranged from 0.49 to 0.89), depending on function choice and location.
The functions performed best in California, and the 2nd order polynomial showed superior performance (r2 = 0.89 in California). Performance deteriorated as polynomial order increased above N = 2. Mean-squared-deviation-based statistical analysis showed that the lack of correlation weighted by standard deviations accounted for most of the variation between predicted and measured radiation.
Results show the usefulness of radiation algorithms in predicting solar radiation from measured air temperature.
Several empirical functions were incorporated into radiation algorithms to predict solar radiation from measured maximum and minimum air temperatures (Tmax and Tmin). Each function was tested with historical weather data from five locations: Stoneville, Mississippi; Meridianville, Alabama; Shafter, California; Corpus Christi, Texas; and Florence, South Carolina.
Monthly mean values of each function’s coefficients were derived, from which daily values were obtained via cubic spline interpolation.
Calibrated with Mississippi weather data, each function predicted solar radiation from measured Tmax and Tmin data for each location.
Predictions were compared to measured radiation for validation, and the best functions were identified.
Agreement was obtained between predicted and measured radiation (r2 ranged from 0.49 to 0.89), depending on function choice and location.
The functions performed best in California, and the 2nd order polynomial showed superior performance (r2 = 0.89 in California). Performance deteriorated as polynomial order increased above N = 2. Mean-squared-deviation-based statistical analysis showed that the lack of correlation weighted by standard deviations accounted for most of the variation between predicted and measured radiation.
Results show the usefulness of radiation algorithms in predicting solar radiation from measured air temperature.
Authors
A..G. richardson, K.R. Reddy, M.L. Boone
Keywords
Radiation algorithm, polynomial, calibration, validation, crop simulation model
Online Articles (31)