Fully-automated estimation of actual to potential evapotranspiration in the Everglades using Landsat and air temperature data as inputs to the Vegetation Index-Temperature Trapezoid method

Thursday, 18 December 2014
Ali Levent Yagci1,2 and John W. Jones2, (1)George Mason University Fairfax, Fairfax, VA, United States, (2)USGS Eastern Geographic Science Center, Reston, VA, United States
While the greater Everglades contains a vast wetland, evapotranspiration (ET) is a major source of water “loss” from the system. Like other ecosystems, the Everglades is vulnerable to drought. Everglades restoration science and resource management requires information on the spatial and temporal distribution of ET. We developed a fully-automated ET model using the Vegetation Index-Temperature Trapezoid concept. The model was tested and evaluated against in-situ ET observations collected at the Shark River Slough Mangrove Forest eddy-covariance tower in Everglades National Park (Sitename / FLUXNET ID: Florida Everglades Shark River Slough Mangrove Forest / US-Skr). It uses Landsat Surface Reflectance Climate Data from Landsat 5, and Landsat 5 thermal and air temperature data from the Daily Gridded Surface Dataset to output the ratio of actual evapotranspiration (AET) and potential evapotranspiration (PET). When multiplied with a PET estimate, this output can be used to estimate ET at high spatial resolution. Furthermore, it can be used to downscale coarse resolution ET and PET products. Two example outputs covering the agricultural lands north of the major Everglades wetlands extracted from two different dates are shown below along with a National Land Cover Database image from 2011. The irrigated and non-irrigated farms are easily distinguishable from the background (i.e., natural land covers). Open water retained the highest AET/PET ratio. Wetlands had a higher AET/PET ratio than farmlands. The main challenge in this study area is prolonged cloudiness during the growing season.