Estimation of Daily Reference Evapotranspiration with Fuzzy Inference Systems

Evapotranspiration is one of the major components of hydrologic cycle and estimation of irrigation needs. In recent years the use of intelligent systems for estimating hydrological phenomena has increased significantly. In this study the possibility of using fuzzy inference system efficiency، creating a bridge between meteorological parameters and evapotranspiration، and comparing the accuracy of reference evapotranspiration using these systems were investigated. After analyzing the different models and different combinations of daily meteorological data، five models for estimating daily reference evapotranspiration were presented. For these models، the calculated evapotranspirationfrom Penman-Monteith-FAO equation was considered as a baseand the efficiency of other models was evluated using statistical methods such as root mean squared error، error of the mean deviation، coefficient of determination،Jacovides (t) and Sabbaghet al. (R2/t) criteria. The used data were collected from Mashhad’s meteorological synoptic station for a period of 50-years (from 1339 to 1389). From the available data، 75 percentwas used for training the model and the rest of 25 percent was utilized for the testing purposes. The results derived from the fuzzy models with different input parameters as compared with Penman-Monteith-FAO and Hargreaves-Samani methods showed that fuzzy systems were very well able to estimate the daily reference evapotranspiration. Fuzzy model so that the highest correlation with the four input variables (r=0. 99) had in mind and evaluate other parameters، the model with two parameters، temperature and relative humidity (RMSE=0. 96، MBE =0. 18، R2=0. 95، t=22، = and R2 / t=0. 04) match very well with the model Penman - Monteith - FAO had stage training. In the test phase، training phase was very similar results and the model with the second phase of temperature and relative humidity will get the best match. According to the results of this study it can be concluded that fuzzy model approach is an appropriate method to estimatethe daily reference evapotranspiration. In addition، the fuzzy models do not require complex calculations which are required forcombination methods.