Modeling Regional Evapotranspiration for Forested Watersheds across the Southern United States

Abstract

Evapotranspiration (ET) is the process that returns water to the atmosphere and therefore completes the hydrologic cycle. ET is a major component in the hydrological balance, and therefore is important to understanding forest water yield, sediment and nutrient movement. However, direct measurement of forest ET for a large region is not possible. The objectives of this study were to develop a model to estimate long-term annual actual evapotranspiration (AET) for forested watersheds across the southern United States (U.S.) and to compare the differences among six potential evapotranspiration (PET) methods. The developed AET model will be used to study hydrologic effects of climate and landuse changes. The six compared PET methods include three temperature-based methods (the Thornthwaite, Hamon and Hargreaves-Samani method) and three radiation-based methods (the Turc, Makkink and Priestley-Taylor method).

A GIS database including land cover, hydrology and climate was developed for thirty-nine forested watersheds across the southern U.S.. Based on these data, a long-term annual AET model was developed. The independent variables included in the model are rainfall, latitude, elevation and percentage of land cover of conifer forests and water body in the watersheds. The model has a R2 of 0.85 and is sufficient to predict long-term annual AET for forested watersheds across the southern U.S.. Six PET methods were highly correlated but significantly different from each other. Greater differences were found among the temperature-based PET methods than radiation-based PET methods. In comparisons of the six PET methods, the Priestley-Taylor, Hamon and Turc methods performed better than the Thornthwaite, Makkink and Hargreaves-Samani methods.