Evaluation of ECOSTRESS Collection 2 Evapotranspiration Products: strengths and uncertainties for evapotranspiration modeling

Authors: PIERRAT, Z - Jet Propulsion Laboratory; PURDY, A - California State University; MALLICK, K - Luxembourg Institute Of Science & Technology; FISHER, J - Chapman University; HALVERSON, G - Jet Propulsion Laboratory; PASCOLINI-CAMPBELL, M - Jet Propulsion Laboratory; RYU, Y - Seoul National University; VILLANUEVA-WEEKS, C - California State University; JOHNSON, M - Jet Propulsion Laboratory; Anderson, Martha; HATCH, B - Jet Propulsion Laboratory; CAWSE-NICHOLSON, K - Jet Propulsion Laboratory

Submitted to: Water Resources Research
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/25/2025
Publication Date: 6/14/2025
Citation: Pierrat, Z., Purdy, A., Mallick, K., Fisher, J., Halverson, G., Pascolini-Campbell, M., Ryu, Y., Villanueva-Weeks, C., Johnson, M., Anderson, M.C., Hatch, B., Cawse-Nicholson, K. 2025. Evaluation of ECOSTRESS Collection 2 Evapotranspiration Products: strengths and uncertainties for evapotranspiration modeling. Water Resources Research. 61(6). Article e2024WR039404. https://doi.org/10.1029/2024WR039404.
DOI: https://doi.org/10.1029/2024WR039404

Interpretive Summary: The ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) is a satellite instrument that helps support water and ecosystem research. ECOSTRESS does this by collecting land surface temperature measurements and modeling evapotranspiration (ET). Initial ET estimates from ECOSTRESS identified key areas for improvement in ET modeling. This study provides an overview of updates to the ECOSTRESS ET product and evaluates its performance against site-level data across North America. We observe an overall improvement from the initial ET product but variable performance based on time of day, climate and vegetation type. We also assess auxiliary data used to drive the ET models and find key areas for model improvement. Overall, this work helps support data users by providing accuracy metrics and key considerations for data use and helps support future work improving satellite-driven ET models.

Technical Abstract: The ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) collects thermal observations from the International Space Station to support evapotranspiration (ET) research at fine spatial resolutions (70 m x 70 m). Initial ET estimates from ECOSTRESS Collection 1 have been used in a wide range of scientific studies and applications, though subsequent analyses identified areas for improvement. This study provides an overview of updates to ECOSTRESS Collection 2 ET and presents an accuracy assessment of ET and auxiliary variables against in situ data from AmeriFlux. Key updates in Collection 2 include: four independent model estimates of ET and improved auxiliary forcing data. We find the multi-model ensemble ET estimate achieves a root mean square error (RMSE) of 109 Wm-2 for instantaneous observations and 1.5 mm/day for daily retrievals. When considering uncertainty in energy balance closure approaches for site-level data, the RMSE improves to 48 Wm-2 for instantaneous ET. We observe variable performance based on time of day, climate and vegetation type. Evaluation of auxiliary data highlight limitations in down-scaled net radiation and relative humidity, contributing to a diurnal hysteresis in ET estimates. We provide accuracy metrics and model sensitivity to auxiliary data to help the user community with data adoption, interpretation, confidence, and applications. ECOSTRESS is the only instrument capable of providing ET at different times of day at high spatial scales, thus, this work is an important step toward enhancing the capabilities of current satellite-driven ET models in resolving diurnal ET variations and guiding directions for future improvements.