Monthly Reference Evapotranspiration Modeling Using Gene Expression Programming From Minimum Climatic Data

Abstract

Evapotranspiration is a key factor for water balance, irrigation scheduling, and crop yield. Even though, Penman-Monteith FAO-56 (PMF-56) equation had estimated the highest accuracy reference evapotranspiration (ETo), it required complete climatic records, which may not be easily available. The present study is to develop and evaluate a gene expression programming (GEP) model for estimating mean monthly ETo using minimal number of climatic data. Climatic variables used to estimate ETo are maximum and minimum air temperature (Tmax and Tmin), mean relative humidity (RH), solar radiation (Rs), and wind speed at 2-m height (u2). The data used in the analysis refer to 32 weather stations available at different locations in Egypt through the CLIMWAT database. The PMF-56 method was used as the reference standard for evaluating the developed GEP models based on statistical criteria such as: index of agreement (IA) and the root mean square error (RMSE). The results showed that the accuracy of the GEP model significantly improved when either RH or u2 was used as additional input variables. The GEP model with the inputs: Tmax, Tmin, RH, and u2 showed the highest IA (0.991 and 0.990) and the lowest RMSE (0.426 mm d-1 and 0.430 mm d-1) for training and testing sets, respectively. Comparing the results of GEP models with other empirical models showed that ETo values estimated by using the GEP models are more accurate. Accordingly, the GEP technique can be employed successfully in modeling ETo from the available climatic data and allowed for providing simple algebraic formulas.