M5 Model Trees and Neural Networks Based Prediction of Daily ET0 (Case Study: Bonab Station)

Evapotranspiration as an important element of hydrologic cycle plays a crucial role in the assessment of watershed balance. To calculate plant water requirements، first reference evapotranspiration (ET0) must be computed، then on this basis this calculation can be applied for any other plants. In the present study the precise assessment of daily reference evapotranspiration was carried out in Bonab station by standard FAO-Penman-Monteith then the combination of daily climatic parameters such as average، minimum and maximum of air temperature، average، minimum and maximum of relative humidity، rainfall، wind speed and sunlight hours were all considered as an input of Multi-Layered-Perceptrons Feed Forward Back Propagation Neural Networks and M5 model tree were both applied to achieve better results. It can be inferred that though the Neural Network approach may render more exact result than M5 model tree، however، M5 model tree provides a more understandable، applicable and simple linear relation in predicting evapotranspiration.