Evaporation Estimation Using Adaptive Neuro-Fuzzy Inference System and Linear Regression

Abstract

Evaporation is important for water planning, management and hydrological practices, and it plays an influential role in the management and development of water resources. This study demonstrates the application of two different models, adaptive neuro-fuzzy inference system (ANFIS), and linear regression (LR) models for estimating monthly pan evaporation in Basrah City, south of Iraq. In the first part of this study, the ANFIS model is used twice, in the first one, the temperature is used as input data only, and in the second one, the temperature and relative humidity are used as input data for predicting the evaporation. A verification test is added to check the model correctness by matching the calculated evaporation with the once observed in Basrah city for the period (1980-2009). In the second part of the study, the results obtained by ANFIS models are compared with results of linear regression model. The comparison reveals that the ANFIS models give better accuracy in estimating monthly pan evaporation than the linear regression model. The accuracy is improved about 5% in correlation coefficient (R) and determination coefficient (R2). The results proved that monthly pan evaporation could be successfully estimated through the use of ANFIS models.