Evaluation of Solar Radiation Prediction Models in North America

摘要: Solar radiation data at the earth's surface (R s , MJ m -2 d -1 ) are not typically recorded weather stations, but they may be predicted from other meteorological measurements. For one location, Keiser, AR, we developed an empirical equation for predicting R The mechanistic models of Hargreaves-Samani (HS) and two forms Bristow-Campbell model, described by Thornton Running (TR) Weiss et al. (WS), were also evaluated 13 sites, covering a 23° range in latitude 42° longitude. HS, TR, WS models, used coefficients as originally published, HS site-specific coefficient (HS-SS) was derived each site. Regression vs. observed values using Keiser gave r 2 (0.77) similar to best models. HS-SS model had lowest root mean square error 3.50 followed TR (3.56), (3.86), (4.33) Predicted ranging 0.72 (TR model) 0.56 (WS model). There slight superiority over Similar fits (r > 0.87) errors found among HS-SS, when averaged 7-d period, it concluded that these three provided accurate precise R, estimations our sites without further modification.