An evaluation of ECOSTRESS products of a temperate montane humid forest in a complex terrain environment

Abstract

Plant water use is difficult to monitor and predict in complex terrain. NASA's ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) provides new data for understanding hydrologic cycling in these forested mountain areas. Evaluation of ECOSTRESS evapotranspiration (ET) has relied primarily on eddy flux towers, but most flux towers are located in flat terrain, sampling relatively homogeneous vegetation. To address this knowledge gap, the accuracy of the ECOSTRESS land surface temperature (L2_LST) and ET (L3_ET_PT-JPL) data were evaluated against data from the U.S. Department of Agriculture (USDA) Forest Service's Coweeta Hydrologic Laboratory: a humid temperate forest ecosystem in the southern Appalachian Mountains of the southeastern U.S. with extensive data from one eddy covariance tower and five climate stations varying in elevation. ECOSTRESS LST showed a strong linear relationship with the weather station near-surface air temperature (R2 = 0.85; offset = −2.5 °C), showing no difference across slope, aspect and elevation, but a higher correlation during the night than the day. ECOSTRESS tended to overestimate ET compared with our site eddy covariance measurements (R2 = 0.43; RMSE = 146 W m−2). To evaluate potential sources of error associated with remotely-sensed inputs to the ECOSTRESS Priestly-Taylor Jet Propulsion Laboratory (PT-JPL) ET model, we ran PT-JPL using ground-based data from our site. ET estimated with locally-collected data showed much better performance in capturing variability (R2 ~ 0.70), illustrating the uncertainty contributed to ECOSTRESS by the coarse scale meteorological inputs in areas of complex topography. Notably, both the ECOSTRESS LST and ET captured important topographic gradients and spatially-explicit diurnal variability. The valley floor of the Coweeta Basin was warmer than the higher elevations by 8 °C in the day, but cooler at night. On the south-facing aspects, LST and ET were higher, consistent with observations. This also highlights the ability of ECOSTRESS and PT-JPL to decouple ET from LST when and where appropriate, as they are normally inverse to one another. Our study provides the first detailed analysis of ECOSTRESS for forested ecosystems in complex terrain and offers insight for future evaluation.