Reference evapotranspiration (ETo) is an important and one of the most difficult components of the hydrologic cycle to quantify accurately. Estimation/measurement of ETo is not simple as there are number of climatic parameters that can affect the process. There exists copious conventional (direct and indirect) and non conventional/soft computing (artificial neural networks, ANNs) methods for estimating ETo. Direct methods have the limitations of measurement errors, expensive, impracticality of acquiring point measurements for spatially variable locations, whereas the indirect methods have the limitations of unavailability of all necessary climate data and lack of generalizability (needs local calibration). In contrast to conventional methods, soft computing models can estimate ETo accurately with minimum climate data which have advantages over limitations of conventional ETo methods. This chapter reviews the application of ANN methods in estimating ETo accurately for 15 locations in India using six climatic variables as input. The performance of ANN models were compared with the multiple linear regression (MLR) models in terms of root mean squared error, coefficient of determination and ratio of average output and target ETo values. The results suggested that the ANN models performed better as compared to MLR for all locations.
Part of the book: Advanced Evapotranspiration Methods and Applications