DEVELOPING A MODEL TO IMPROVE THE PREDICTION OF DAILY SOLAR RADIATION FOR THE PREDICTION OF REFERENCE EVAPOTRANSPIRATION

Three modification of the Hargreaves and Samani (HS) model were proposed to account for the effect of the atmospheric vapor on the prediction of the global solar radiation. The predicted values of the solar radiation using the three models and the values of the original HS model were compared to the measured solar radiation using root mean square error (RMSE), mean absolute bios error (MABE), mean bios error (MBE) and the T-test in pairs. The results showed that the first model M1, and the second model M2 performed better than the original HS. The T- test results showed that there are no significant differences between the means values of the two models prediction and the mean value of the measured solar radiation. The MBE for both M1 and M2 indicates a tendency to slightly over estimate the solar values for M1, and a slight under estimation of the M2 prediction. The model M1 increased the values of the ETo predicted using the FPM than the original HS model and thus solved the problem of ETo under estimation by FPM. It is recommended to use M1 for application where over predictions is safer such as evapotranspiration prediction and use M2 where under prediction is safer such as solar radiation prediction for thermal solar application and solar collectors.