COMPARATIVE EVALUATION OF DIFFERENT POTENTIAL EVAPOTRANSPIRATION ESTIMATION APPROACHES

Abstract

Accurate estimation of potential evapotranspiration is necessary step in water resources management. Recently, the FAO-56 version of Penman-Monteith equation has been established as a standard for calculating reference evapotranspiration (ET0). Still there are different approaches (requiring less data) which estimate ET0 closely to Penman-Monteith method for different climatological conditions. Performance Evaluation of all the approaches on the same basis is prerequisite for selecting an alternative approach in accordance with available data. Therefore, two most popular temperature-based approaches (Hargreaves and Thornthwaite) and two radiation based approaches (Priestley-Taylor and Turc) were used to estimate monthly potential evapotranspiration (ET0) at Pantnagar (Uttarakhand), India. Further, the performance of all these methods were evaluated by regression and error analysis between standard ET0 derived using FAO-56 Penman-Monteith method and ET0 values estimated using all the four methods, on monthly and seasonal basis. On monthly basis Turc method performed best with lowest RMSE (0.562), ARE (0.137), AAD (0.448) and high coefficient of determination (0.792). On seasonal basis, the Priestley-Taylor methodwas found to be the best for Rabiseasonwith lowest error values and minimum seasonal over/under ET0 predication rate with respect to standard ET0.Turc method holds second rank in Rabi season. However in Kharif season Truc method performed better than any other method with lowest error terms and lowest seasonal over/under predication rate. In summer season all the methods performed poorly compared to other two seasons but Hargreaves method performed better than other methods. Though the performance and accuracy of FAO-56 Penman-Monteith method can never be debated in theoretical or practical applications yet the comparative evaluation performed in this paper can be used as guideline for selection of alternative or less data dependent methods in case of non-availability of data.