Abstract
Estimates of solar radiation (Ri) by 15-day periods were calculated from other available meteorological data by multiple regression analysis of Eo = f(Ri, u, Ta) ed), and then solving for Ri from the most significant equation of the model given above. Only solar radiation wind speed (u) and air temperature (Ta) were found to correlate significantly with Eo. The solution in terms of Ri (equation /4/) then was used to obtain estimates of solar radiation for 1- to 5-day periods. The mean ratio of estimated Ri to measured Ri approached unity (0.95), while the coefficient of variation was 8.9 percent, as compared to 5.1 percent for the original 15-day period data. It was found when these were compared to measured values that neglect of air temperature reduced precision of the estimates. Equation /4/ (the indirect solution) next was evaluated on the basis of foreign data, representing extreme meteorological conditions as those in Bet Dagan, Israel and Cristobal, Panamá. This equation also was compared at the same time to a direct solution of Ri by multiple regression analysis (equation /6/), a solution which directly minimizes the deviations about Ri. Statistical data are presented which compare the precision of the estimates by either equation (/4/ and /6/).