A Comparative Study of Daily Pan Evaporation Estimation Using ANN, Regression and Climate Based Models

Evaporation estimates are needed for efficient management of water resources at a farm scale as well as at a regional or catchment scale. This paper presents application of artificial neural networks (ANN), statistical regression and climate based models viz.: Penman, Priestley–Taylor and Stephens and Stewart, for estimation of daily pan evaporation. Six different measured weather variables comprising various combinations of maximum and minimum air temperature, sun shine hours, wind speed, relative humidity I and II were used. Randomly selected 1,096 daily records were used to develop the models of ANN and regression, and 365 daily records were used as independent data set for performance evaluation, which was not used previously in any of the model development process. The results of the developed ANN and multiple linear regression (MLR) models along with Penman, Priestley-Taylor and Stephens and Stewart models were compared statistically with observed pan evaporation values. Comparison showed that there is slightly better agreement between the ANN estimations and measurements of daily pan evaporation than other models. Copyright Springer Science+Business Media B.V. 2010