Evapotranspiration in the Okefenokee Swamp watershed: a comparison of temperature-based and water balance methods

Abstract

In the Okefenokee Swamp, actual evapotranspiration (AET) should approximate potential evapotranspiration (PET) as there is rarely a moisture deficit. Even during droughts, a high evapotranspiration rate is maintained owing to a large area of wetlands and a shallow groundwater table in the watershed. Therefore, if AET can be estimated in the Okefenokee Swamp watershed, it can be used to evaluate the accuracy of PET estimates for the region. In this study, temperature-based PET estimates were compared with AET obtained by the water balance method for the Okefenokee Swamp watershed. Swamp water level data were used to identify subperiods with identical storage at the beginning and end of the period, so that a steady-state water balance model could be used to estimate AET as the difference between precipitation and outflow runoff. When the PET estimates were regressed upon AET, Thornthwaite PET had the highest R 2 value (0.817), followed by Blaney-Criddle PET (0.781), and Holdridge PET proportioned by biotemperature (0.768). The Thornthwaite method also rendered a long-term average very close to that of AET. Seasonal errors in Thornthwaite PET were reduced by using pan evaporation and temperature data to partition annual values into monthly values. Holdridge PET overestimated evapotranspiration from the Okefenokee Swamp watershed. Using standard crop coefficients, the Blaney-Criddle method would have overestimated evapotranspiration in the humid Okefenokee Swamp watershed. However, by substituting a very low crop coefficient ( K = 1.5), the Blaney-Criddle method gave results for the Okefenokee region similar to those obtained by the Thornthwaite method with the lowest error among all the methods examined. Pan evaporation correlated well with AET ( R 2 = 0.628) once an appropriate pan coefficient (0.71) had been determined.