Empirical modelling of solar radiation for selected cities in Nigeria using multivariate regression technique

In this paper, the data for solar radiation, minimum and maximum temperatures, wind speed and evaporation for the years 1970-1995 and for fourteen stations taken over Nigeria were obtained from the archives of the Nigerian Meteorological Agency (NIMET), Oshodi, Lagos. The distributions of solar radiation with each of the four meteorological variables were observed. It was found that solar radiation was adequately monitored by each of the variables indicating that the linear combination of the variables could be used to developed model from where solar radiation can be evaluated. Consequently, a multivariate linear regression model was developed for each of the stations. The statistical indicators such as R2, MBE, RMSE and MPE were calculated to estimate the efficiency of the developed models. For instance, the values of the R2, MBE, RMSE and MPE calculated for Sokoto, a Sahelian station are 0.5349, 0.0273, 0.4154 and 0.0811, respectively, which show that the model for Sokoto is significant. Those for other stations have also been calculated with all showing high level of signicance at 0.05 alpha level.   Key words: Solar radiation, meteorological variables, correlation coefficient, multivariate regression, evaporation, significance.