Forecasting Weekly Reference Evapotranspiration Using Wavelet-ANFIS Hybrid Model

Evapotranspiration is one of the most important components in the optimization of water use in agriculture and water resources management. In recent years, artificial intelligence methods and wavelet based hybrid model have been used for forecasting of hydrological parameters. In the present study, applications of the adaptive network-based fuzzy inference system (ANFIS) and Wavelet-ANFIS models to forecast weekly reference evapotranspiration at the synoptic stations of Tabriz, Ahvaz, Bandar Abbas, and Ramsar were investigated. For this purpose, a 20-year statistical period (1990-2009) was considered: 15 years (1990-2004) for training and the final five years (2005-2009) for testing the various models. Various combinations of input data (various lag times) and different kinds of mother wavelets were evaluated. Results showed that, compared to the ANFIS model, the hybrid model Wavelet-ANFIS had greater ability and accuracy in forecasting weekly evapotranspiration at all of the four synoptic stations. Moreover, the use of yearly lag times in the ANFIS model increased its accuracy. However, in the Wavelet-ANFIS, yearly lags not only did not increase the accuracy of the Wavelet-ANFIS model, but also reduced its accuracy in some cases. Investigation of various kinds of mother wavelets also indicated that the Meyer wavelet was the most suitable mother wavelet for forecasting weekly reference evapotranspiration. Results of this study can also be used for irrigation scheduling in the studied regions.