Application of Fuzzy Logic-Based Modeling to Estimate Hourly Reference Evapotranspiration Using the Least input Data

Evapotranspiration as one of the most important components of the hydrologic cycle, plays a key role in water resources management, crop yield simulation and irrigation scheduling. Therefore, presenting a low cost and precision model is very essential for calculations of hourly ETo. Although, there are empirical formulas, their performances are not all satisfactory due to the complicated nature of the hourly evapotranspiration process, the data availability, and high cost and error for gathering data. This paper develops hourly ETo estimation model based on fuzzy inference system (FIS) technique. We follow the idea of using the least input parameters, so the net radiation (Rn) selected, as the only input parameter. The used data has been picked on UC-Andrade station for training model, that have the most variation on Rn and climatically conditions, and another thirteen stations, that selected randomly, among 114 automated stations in US California. There is not proper hourly data in Iran. FIS model estimates hourly ETo as crisp number using of 230 rules with 48 level, centeroid defuzzification method and inference Mamdani method. FIS results compared with Penman-Monteith-FA056 and CIMIS-Penman combined model. It has been found that FIS technique has high accuracy and good performance (for the train data set, R2 = 0.97, RMSE= 0.07, MBE=-0.004 and R2/t (t: Jacovides criteria)=0.21). Comparing FIS with CIMIS and FAO56 results shows that FIS has better correlation with CIMIS than FAO56 for test data set, with R2 = 0.94, RMSE= 0.0693, MBE=-0.0384 and R2/t=0.018. Among FIS, CIMIS and FAO56, FIS model is economical, because of the parsimony principal; in conclusion, it raises model accuracy.