Applicability of support vector machines and adaptive neurofuzzy inference system for modeling potato crop evapotranspiration

Abstract

Estimation of crop evapotranspiration (ETC) for certain crops such as potato is very important for irri- gation planning, irrigation scheduling and irrigation sys- tems management. The primary focus of this study was to investigate the accuracy of the adaptive neurofuzzy infer- ence system (ANFIS) and support vector machines (SVM) for potato ETC estimation when lysimeter measurements or the complete weather data for applying the FAO method are not available. The estimates of the ANFIS and SVM models were compared with the empirical equations of Blaney-Criddle, Makkink, Turc, Priestley-Taylor, Har- greaves and Ritchie. The performances of the different SVM and ANFIS models were evaluated by comparing the corresponding values of root mean square error (RMSE), mean absolute error (MAE) and correlation coefficient (r). The drawn conclusions confirmed that the SVM and AN- FIS models could provide more accurate ETC estimates than the empirical equations. Overall, the minimum RMSE (0.042 mm/day) and MAE (0.031 mm/day) values and the maximum r value (0.98) were obtained using the SVM model with mean air temperature, relative humidity, solar radiation, sunshine hours and wind speed as inputs.