Rainfall modelling using backward generalized estimating equations: a case study for Fasa Plain, Iran

In recent years, plenty of researches have accomplished to make the relationship between the climatic variables for daily, monthly, and seasonal rainfall occurrence and magnitude around the world. In this study, monthly rainfall modeling was performed using backward generalized estimating equation (GEE). In this regard, monthly average maximum and minimum temperature, sunshine hours, wind speed, and relative humidity data from 1967–2014 for the Fasa Plain at Fars province, Iran were selected as predictors to investigate their effects on response variable of rainfall. Results indicated that in February, March, April, June, August, and October the term of humidity has positive effect (B > 0 and P < 0.05) and the terms of maximum and minimum temperature, sunshine hours, and wind speed have negative effects (B < 0 and P < 0.05) on rainfall. Then, to assess the verification and accuracy of the final equation of GEE model, the monthly rainfall was forecasted and compared with the observed rainfall values. The determination coefficients of more than 96.0% between the observed data and the forecasted values illustrates the goodness of this model in prediction. The goodness of fit indices showed that the GEE nicely modeled the rainfall.