A Comparative Study of Potential Evapotranspiration Estimation by Three Methods with FAO Penman–Monteith Method across Sri Lanka

Abstract

Among numerous methods that have been developed to estimate potential evapotranspiration (PET), the Food and Agricultural Organization Penman–Monteith model (FAO P–M) is often recognized as a standard method to estimate PET. This study was conducted to evaluate the applicability of three other PET estimation methods, i.e., Shuttleworth–Wallace (S–W) model, Thornthwaite (TW) and pan methods, to estimate PET across Sri Lanka with respect to the FAO P–M model. The meteorological data, i.e., temperature, relative humidity, wind speed, net solar radiation, and pan evaporation, recorded at 14 meteorologic stations, representing all climate and topographic zones of Sri Lanka, were obtained from 2009 to 2019. The models’ performances were assessed based on three statistical indicators: root mean squared error (RMSE), bias, and Pearson correlation coefficient (R). In comparison with the FAO P–M model estimates, the seasonal and annual estimates of all three models show great differences. The results suggested that pan and S–W methods perform better in the dry zone of the country. Both S–W and pan methods underestimated PET over the entire county in all seasons. TW does not show consistent results over the country, thus being found as the least reliable alternative. Although S–W is highly correlated with the FAO P–M model, the application of the model in a data-scarce region is more constrained, as it requires more parameters than the FAO P–M model. Thus, the study suggests employing alternative methods based on the region of the country instead of one single method across the entire country.