Liu et al. (2026) Integrating Landsat and ECOSTRESS to explore the value of additional thermal infrared satellite observations for field-scale evapotranspiration mapping

Identification

• Journal: Remote Sensing of Environment
• Year: 2026
• Date: 2026-10-14
• Authors: Hui Liu, Yun Yang, Martha C. Anderson, Feng Gao, Christopher Hain, John Volk, Zoe Pierrat, Kerry A. Cawse-Nicholson, Forrest Melton, Christine Lee, A. J. Purdy, Weina Duan, Tian Li, Jiaxin Wang
• DOI: 10.1016/j.rse.2026.115738

Research Groups

• Soil and Crop Sciences Section, School of Integrative Plant Science, Cornell University, Ithaca, NY, USA
• USDA, Agricultural Research Service, Hydrology and Remote Sensing Laboratory, Beltsville, MD, USA
• NASA, Marshall Space Flight Center, Earth Science Branch, Huntsville, AL, USA
• Desert Research Institute, Reno, NV, USA
• Department of Geography, University of California, Santa Barbara, CA, USA
• NASA Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA
• NASA Ames Research Center, Moffett Field, CA, USA
• California State University, Monterey Bay, Seaside, CA, USA

Short Summary

This study evaluates the value of integrating ECOSTRESS thermal observations with Landsat data to improve field-scale evapotranspiration (ET) time series, finding that ECOSTRESS significantly reduces interpolation errors and enhances temporal consistency, especially when it effectively shortens Landsat sampling gaps.

Objective

• To quantify the value of more frequent thermal observations from ECOSTRESS for improving field-scale evapotranspiration (ET) time series when integrated with Landsat.
• To determine the conditions under which ECOSTRESS provides significant added value for daily ET estimation.
• To evaluate how ECOSTRESS overpass time influences ET accuracy.
• To provide practical insights into the effective use of ECOSTRESS for improving field-scale ET time-series construction.

Study Configuration

• Spatial Scale: Field-scale (30 meters), covering the contiguous United States (CONUS).
• Temporal Scale: Daily evapotranspiration (ET) time series, with observations spanning from 2019 onwards, utilizing Landsat's approximately 8-day effective revisit and ECOSTRESS's 1-5 day revisit.

Methodology and Data

• Models used: Disaggregated Atmosphere-Land Exchange Inverse (DisALEXI) model, Data Mining Sharpener (DMS) framework, Two-Source Energy Balance (TSEB) model, Atmosphere-Land Exchange Inverse (ALEXI) model. Daily ET interpolation uses linear interpolation of the actual-to-reference ET ratio.
• Data sources:
  • Satellite: Landsat Collection 2 (7, 8, 9) for visible, shortwave infrared (VSWIR), and thermal infrared (TIR) data; ECOSTRESS Collection 2 Level 2 Land Surface Temperature and Emissivity (L2_LSTE) for approximately 70-meter Land Surface Temperature (LST); Harmonized Landsat Sentinel-2 (HLS) Version 2.0 for 30-meter surface reflectance.
  • Observation: Daily evapotranspiration (ET) data from 25 AmeriFlux eddy covariance flux tower sites across the contiguous United States.
  • Reanalysis/Ancillary: North American Land Data Assimilation System (NLDAS) for initial air temperature; Climate Forecast System Reanalysis (CFSR) for hourly solar radiation and 3-hourly meteorological variables; 4-kilometer GOES-based ALEXI daily ET product; 30-meter National Land Cover Database (NLCD); OpenET 4-kilometer grass reference ET (derived from gridMET dataset).

Main Results

• ECOSTRESS-based daily evapotranspiration (ET) on satellite overpass dates demonstrated comparable accuracy to Landsat-based ET, with Root Mean Squared Error (RMSE) values of 1.14 millimeters per day (mm/day) and 1.17 mm/day, respectively.
• Integrating ECOSTRESS observations consistently improved the accuracy and reduced the bias of interpolated ET estimates across daily, weekly, monthly, and annual temporal scales. Mean Bias Error (MBE) reduction ranged from 6.97% to 8.00%, while RMSE and Mean Absolute Error (MAE) also decreased.
• The largest improvements in ET time series agreement occurred when ECOSTRESS observations reduced effective Landsat sampling gaps by approximately 50% (Gap Reduction Ratio of 0.4–0.5).
• ECOSTRESS observations acquired near midday (10:00–14:00 local time) yielded the most reliable ET estimates, showing the smallest median absolute error and highest consistency with flux tower measurements. Early morning (<10:00) and late afternoon (>17:00) overpasses exhibited larger deviations.
• Spatial comparisons of same-day ET maps revealed strong agreement between Landsat and ECOSTRESS products, with Mean Absolute Differences (MAD) ranging from 0.183 to 0.662 mm/day for representative sites. The highest agreement (R² > 0.85, lowest RMSD/MAD) was observed for ECOSTRESS overpasses between 10:00 and 13:00 local time.

Contributions

• Systematically quantified the conditions under which ECOSTRESS provides significant added value for daily ET estimation, moving beyond overall accuracy gains to a mechanistic understanding.
• Evaluated how ECOSTRESS's variable overpass time impacts daily ET reconstruction and accuracy, identifying optimal acquisition windows (10:00–14:00 local time) for reliable estimates.
• Provided a decision-oriented evaluation framework clarifying when ECOSTRESS meaningfully enhances Landsat-based ET time series, considering factors like gap reduction and overpass-day quality.
• Offered empirical guidance for integrating ECOSTRESS into operational ET workflows and informed the temporal sampling requirements for next-generation thermal satellite missions.

Funding

• USGS contract (140G0126C0007)
• NASA ECOSTRESS program (Grant No. 80NSSC26K0294 and 80NSSC26K0331)
• USGS OpenET Project Cooperative Agreement (G23AC00676)
• Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration (for Zoe Amie Pierrat, Kerry Cawse-Nicholson, and Christine M. Lee)
• USGS Grant G24AC00499 (for Adam J. Purdy)
• NASA's Western Water Action Office (for Adam J. Purdy)

Citation

@article{Liu2026Integrating,
  author = {Liu, Hui and Yang, Yun and Anderson, Martha C. and Gao, Feng and Hain, Christopher and Volk, John and Pierrat, Zoe and Cawse-Nicholson, Kerry A. and Melton, Forrest and Lee, Christine and Purdy, A. J. and Duan, Weina and Li, Tian and Wang, Jiaxin},
  title = {Integrating Landsat and ECOSTRESS to explore the value of additional thermal infrared satellite observations for field-scale evapotranspiration mapping},
  journal = {Remote Sensing of Environment},
  year = {2026},
  doi = {10.1016/j.rse.2026.115738},
  url = {https://doi.org/10.1016/j.rse.2026.115738}
}