Interoperability of ECOSTRESS and Landsat for mapping evapotranspiration time series at sub-field scales

Authors: Anderson, Martha; YANG, YANG - US Department Of Agriculture (USDA); XUE, JIE - US Department Of Agriculture (USDA); Knipper, Kyle; YANG, YUN - US Department Of Agriculture (USDA); Gao, Feng; HAIN, C. - Nasa Marshall Space Flight Center; Kustas, William - Bill; CAWSE-NICHOLSON, KERRY - Jet Propulsion Laboratory; HULLEY, GLYNN - Jet Propulsion Laboratory; FISHER, JOSHUA - Jet Propulsion Laboratory; Alfieri, Joseph; MEYERS, TILDEN - National Oceanic & Atmospheric Administration (NOAA); Prueger, John; BALDOCCHI, DENNIS - University Of California; REY-SANCHEZ, CAMILO - University Of California

Submitted to: Remote Sensing of Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/9/2020
Publication Date: 11/20/2020
Citation: Anderson, M.C., Yang, Y., Xue, J., Knipper, K.R., Yang, Y., Gao, F.N., Hain, C., Kustas, W.P., Cawse-Nicholson, K., Hulley, G., Fisher, J., Alfieri, J.G., Meyers, T., Prueger, J.H., Baldocchi, D., Sanchez, C. 2020. Interoperability of ECOSTRESS and Landsat for mapping evapotranspiration time series at sub-field scales. Remote Sensing of Environment. 252. Article 112189. https://doi.org/10.1016/j.rse.2020.112189.
DOI: https://doi.org/10.1016/j.rse.2020.112189

Interpretive Summary: Thermal infrared satellite imagery, collected from air- or spaceborne platforms, has demonstrated value for mapping vegetation water use (evapotranspiration, or ET) and stress over agricultural landscapes. To date, Landsat has been the only routine, global source of thermal imaging at spatial resolutions fine enough to discriminate between individual farm fields. The 8-16 day imaging interval of Landsat, however, is not frequent enough to support operational water management applications, particularly in cloudy regions where clear views of the land-surface are infrequent. The recent ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) mission, with an average 4-day revisit and variable overpass time, allows the unique opportunity to investigate requirements for future operational water use mapping missions. Comparing ECOSTRESS and Landsat-derived ET maps, results suggest that a 4-day revisit can significantly improve seasonal ET estimates; however, sensor disaggregation may add unacceptable uncertainty for operational uses.

Technical Abstract: Land-surface temperature retrieved from thermal infrared (TIR) remote sensing has proven to be a valuable constraint in surface energy balance models for estimating evapotranspiration (ET). For optimal utility in agricultural water management applications, frequent thermal imaging (<4-day revisit) at sub-field (100 m or less) spatial resolution is desired. While, the current suite of Landsat satellites (7 and 8) provides the required spatial resolution, the 8-day combined revisit can be inadequate to capture rapid changes in surface moisture status or crop phenology. The new ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) mission, with an average 4-day revisit interval and nominal 70-m resolution, provides a valuable research platform for augmenting Landsat TIR sampling and for investigating TIR-based ET mapping mission requirements more broadly. This study investigates the interoperability of Landsat and ECOSTRESS imaging for developing ET image timeseries with high spatial (30-m) and temporal (daily) resolution. A data fusion algorithm is used to fuse Landsat and ECOSTRESS ET retrievals at 30 m with daily 500-m retrievals using TIR data from the Moderate Resolution Imaging Spectroradiometer (MODIS) over target agricultural sites spanning the United States.The added value of the combined multi-source dataset is quantified in comparison with daily flux tower observations collected within these target domains. In addition, we investigate ET model performance as a function of ECOSTRESS view angle, overpass time, and time separation between TIR and visible to shortwave infrared (VSWIR) data acquisitions used to generated land-surface temperature, leaf area index, and albedo inputs to the surface energy balance model. The results demonstrate the value of the higher temporal sampling provided by ECOSTRESS, especially in areas that are frequently impacted by cloud cover. Limiting usage to ECOSTRESS scenes collected between 9:00 a.m. to 5:00 p.m. and nadir viewing angles <20O yielded daily (24-hr) ET retrievals of comparable quality to the well-tested Landsat baseline. We also discuss challenges in using LST from a thermal free-flyer system for ET retrieval, which will have ramifications for future TIR water-use mapping missions.