Evapotranspiration estimation by two different neuro-fuzzy inference systems

SummaryThe potential of two different adaptive network-based fuzzy inference systems (ANFIS) based neuro-fuzzy systems in modeling of reference evapotranspiration (ET0) are investigated in this paper. The two neuro-fuzzy systems are: (1) grid partition based fuzzy inference system, named G-ANFIS, and (2) subtractive clustering based fuzzy inference system, named S-ANFIS. In the first part of the study, the performance of resultant FIS was compared and the effect of parameters was investigated. Various daily climatic data, that is, solar radiation, air temperature, relative humidity and wind speed from Santa Monica, in Los Angeles, USA, are used as inputs to the FIS models so as to estimate ET0 obtained using the FAO-56 Penman–Monteith equation. In the second part of the study, the estimates of the FIS models are compared with those of artificial neural network (ANN) approach, namely, multi-layer perceptron (MLP), and three empirical models, namely, CIMIS Penman, Hargreaves and Ritchie methods. Root mean square error, mean absolute error and determination coefficient statistics are used as comparing criteria for the evaluation of the models’ performances. Based on the comparisons, it is found that the S-ANFIS model yields plausible accuracy with fewer amounts of computations as compared to the G-ANFIS and MLP models in modeling the ET0 process.

Research highlights
► The neuro-fuzzy model with temperature-based data could be used in irrigation scheduling problem.
► The temperature based NF model results are better than the four-parameter empirical model.
► The RH and U2 significantly increase the estimation accuracy of temperature-based models.
► The Hargreaves and Ritchie models should be used after calibration with FAO-56 PM ET0 values.
► Calibrated CIMIS Penman model gives more accurate results than those of the other empirical models.