Estimation of Daily Reference Evapotranspiration (ET0) in the North of Algeria Using Adaptive Neuro-Fuzzy Inference System (ANFIS) and Multiple Linear Regression (MLR) Models: A Comparative Study

Abstract

The aim of this study is to model the daily reference evapotranspiration (ET0) in the Mediterranean region of Algiers, Algeria country, using Adaptive Neuro-fuzzy inference system (ANFIS) and multiple linear regression (MLR) models. Various daily climatic data, i.e. daily mean relative humidity, sunshine duration, maximum, minimum and mean air temperature and wind speed obtained from Dar El Beida weather station, are used as inputs to the ANFIS and MLR models so as to estimate ET0. In order to find the optimal topology of the ANFIS, different architectures were trained and examined and the network with minimum Root Mean Squared Error (RMSE), Mean Absolute Error (MAE) and high Coefficient of Correlation (CC) has been selected as an optimal configuration. A comparison was conducted between the estimates provided by the ANFIS and by MLR. The results showed that ANFIS using the climatic data successfully estimated ET0 and the ANFIS simulated ET0 better than the MLR. Totally 2,193 daily samples were used for training the model, and 730 daily samples for testing and validation of the model. The developed ANFIS model for the ET0 modelling shows good performance with an MAE index in the range of 0.32–0.75, RMSE between 0.41 and 0.75 and the CC in the range of 0.80–0.96, which endows with high performance of predictive ANFIS system to make use for modelling daily reference evapotranspiration (ET0).