Long-term potential and actual evapotranspiration of two different forests on the Atlantic Coastal Plain
A reliable estimate of potential evapotranspiration (PET) for a forest ecosystem is critical in ecohydrologic modeling related with water supply, vegetation dynamics, and climate change and yet is a challenging task due to its complexity. Based on long-term on-site measured hydro-climatic data and predictions from earlier validated hydrologic modeling studies, this study compared different methods for estimating monthly and annual potential evapotranspiration (PET) for two Atlantic coastal plain forests. One study site is a naturally drained mature pine mixed hardwood forest in South Carolina (SC), and the other site is a drained pine plantation forest in North Carolina (NC). The Hargreaves-Samani (HS) method was used for estimating grass PET for the SC site, while the Penman-Monteith (PM) method was used for calculating a standard grass reference (REF-ET) and simulating forest PET for the NC site. Daily HS-grass PET was used as an input in the Thornthwaite water balance (WBA) model, which was validated with long-term monthly streamflow to obtain simulated ET for the 1946-2008 period at the SC site. A process-based field-scale model, DRAINMOD-FOREST, was used for quantifying ET for the 1988-2008 period for the NC site by using REF-ET and forest PET as inputs separately. The monthly ET/PET ratios were further calculated for both sites. The long-term mean annual HS-grass PET was 1137 (Â±69) mm at the SC site. HS-grass PET for a recent four-year (2011-2014) period was 11% higher than the forest PET obtained by the PM method using above-canopy microclimate data and canopy resistance parameters. The long-term annual ET/HS PET varied from 0.76 to 1.0, with an average of 0.92. Annual PM-forest PET simulated using the validated DRAINMOD-FOREST model at the NC site varied from 1014 to 1335 mm with a long-term mean of 1146 Â±87Â mm, which is about 13% higher than the REF-ET (1010 Â±123 mm) at the NC site but very similar to the values obtained for the HS-grass PET for the SC site. The estimated annual ET/PM-forest PET ratios varied from 0.81 to 0.97, with an average of 0.90. We also found similar seasonal values and variations of ET/HS PET at the SC site and ET/PM PET at the NC site, both of which were largely different from the ET/REF-ET values and their seasonal distribution reported for another pine forest site (Parker Tract) in coastal NC with eddy flux-based measurements of ET. Results from this study showed large difference of PET estimations by using different methods, particularly for the grass and forest reference. This study also highlighted the importance of properly defining and estimating forest PET, rather than using the standard REF-ET, and related mean monthly ET/PET ratios for estimating ET for a forest reference in forest hydrologic models and water balance studies.