Feasibility of artificial intelligence and CROPWAT models in the estimation of uncertain combined variable using nonlinear sensitivity analysis

Reference evapotranspiration (ET0) estimation with reliable accuracy is critical for the management of water resources and irrigation practices. The aim of this study is to estimate ET0 using CROPWAT model in Kano and Katsina meteorological stations of northwestern Nigeria. Artificial neural network (ANN) and multiple linear regression (MLR) were also developed for comparison. Monthly mean data for 34 years (1983–2016) including maximum, minimum and mean temperatures (Tmax, Tmin and Tmean), relative humidity (RH) and wind speed (U2) were used as inputs. Penman Monteith (FAO-56-PM) regarded as the standard method for computing ET0 was used as the benchmark. Initially, nonlinear correlation analysis was carried out to determine the best input variable. Thereafter, 7 models were developed based on different combinations to ascertain the most reliable for comparison to Crop Water and Irrigation Requirements Program of FAO (CROPWAT) model. The normalized Determination coefficient (R2) and root mean square error (RMSE) were used as the criteria for checking the performance of the models. The results showed that RH was the most dominant input, model 6 that has a combination of Tmax, RH and U2 provided the most reliable performance. The results also demonstrated that CROPWAT model is comparable in performance to ANN and MLR and can be efficiently used to estimate ET0 in the study stations with R² of 0.923, 0.962 and RMSE of 0.065, 0.046 in the validation phase for Kano and Katsina stations, respectively.