Improving ECOSTRESS-based Evapotranspiration Estimates Using Harmonized Landsat Sentinel-2 Imagery

Evapotranspiration (ET) is a major component of the hydrologic cycle and plays a fundamental role in water and land management. However, previous studies have shown that estimating ET is quite challenging, particularly at fine temporal and spatial scales. Dense time series of harmonized Landsat 8 and Sentinel-2 imagery (HLS) combined with ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) data provide a unique opportunity to enhance monitoring, mapping, and characterizing ET at unprecedented spatial and temporal resolutions. In this study, we develop and evaluate an improved ET estimation method based on Priestley-Taylor Jet Propulsion Laboratory algorithm (PT-JPL) with HLS optical reflectance imagery and ECOSTRESS land surface temperature (LST) and surface emissivity, in addition to MODIS surface albedo and ERA5-Land climate reanalysis data. The new approach creates denser times series of ET from HLS imagery, which can also be used to map ET at 30 m spatial resolution. The new ET estimates are evaluated against ground-based observations from the AmeriFlux network and compared with the performance of the original ECOSTRESS PT-JPL ET estimates across different ecosystems and landcover settings over the continental United States. We present results for evaluating ET estimates, the remote sensing and climate reanalysis inputs, and illustrating the sensitivity and uncertainty of the improved ET method and its performance related to land cover and terrestrial ecosystem properties.