Laluet et al. (2026) Assessing the suitability of global evapotranspiration products over irrigated areas

Identification

• Journal: Hydrology and earth system sciences
• Year: 2026
• Date: 2026-04-01
• Authors: Pierre Laluet, Chiara Corbari, Oscar M. Baez-Villanueva, Sophia Walther, Yongqiang Zhang, Joaquín Muñoz-Sabater, Clement Albergel, Wouter Dorigo
• DOI: 10.5194/hess-30-1779-2026

Research Groups

• Department of Geodesy and Geoinformation, TU Wien, Vienna, Austria
• Department of Civil and Environmental Engineering, Politecnico di Milano, Milan, Italy
• Hydro-Climate Extremes Lab (H-CEL), Ghent University, Ghent, Belgium
• Max Planck Institute for Biogeochemistry, Jena, Germany
• Key Laboratory of Water Cycle and Related Land Surface Processes, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, China
• European Centre for Medium-Range Weather Forecasts (ECMWF), Reading, UK
• European Space Agency Climate Office, ECSAT, Didcot, UK

Short Summary

This study evaluates the suitability of six global evapotranspiration (ET) products over irrigated croplands by comparing their spatial patterns, seasonal dynamics, and magnitudes against irrigation maps, an independent ET ensemble, and eddy covariance measurements across diverse agro-climatic regions. The assessment reveals significant differences, with PMLv2, SSEBop v6.1, and FLUXCOM RS generally showing the strongest and most consistent agreement with reference datasets, while ERA5-Land exhibits the weakest correspondence.

Objective

• To assess the ability of six widely used global evapotranspiration (ET) products (FLUXCOM RS, GLEAM4.3a, PMLv2, ERA5-Land, MOD16A2, and SSEBop v6.1) to capture irrigation-related ET signals over irrigated croplands.

Study Configuration

• Spatial Scale: Global ET products (native resolutions from 500 m to 0.1°, resampled to a common 0.1° grid for intercomparison). Evaluation conducted across seven major irrigated regions in the Contiguous United States (CONUS), Europe, and South Asia. Local-scale assessment at two eddy covariance sites in the Po Valley, Italy (tower footprints 100-300 m).
• Temporal Scale: ET products cover periods from 2000/2001 to 2020/2025. Evaluation periods vary: Global Map of Irrigated Areas (GMIAv5) reference period around 2005; OpenET comparison from 2016 to 2023; Eddy covariance data from 2008 to 2014. Monthly mean ET values were computed, focusing on peak irrigation seasons.

Methodology and Data

• Models used: FLUXCOM RS, GLEAM4.3a, PMLv2, ERA5-Land, MOD16A2, SSEBop v6.1.
• Data sources:
  • Reference Datasets:
    • Global Map of Irrigated Areas version 5 (GMIAv5) for spatial evaluation of irrigation infrastructure.
    • OpenET ensemble (median of six models: METRIC, DisALEXI, geeSEBAL, SSEBop, PT-JPL, SIMS) for seasonal dynamics and magnitude assessment in CONUS.
    • Eddy covariance measurements from two irrigated maize sites (Landriano, Livraga) in the Po Valley, Italy, for local-scale validation.
  • Input Data for ET Products (examples): MODIS-derived variables (Land Surface Temperature (LST), fraction of absorbed photosynthetically active radiation (fAPAR), Enhanced Vegetation Index (EVI), Leaf Area Index (LAI), albedo, emissivity), CERES (net radiation), ESA CCI (surface soil moisture), ERA5 (atmospheric fields, radiative fluxes), NASA GMAO (GEOS-5 reanalysis), VIIRS-based LST, TerraClimate (reference evapotranspiration), GLDAS-2.1 (meteorological forcings, VPD), MSWEP v2.8 (precipitation).
  • Ancillary Data: WorldCereal database for land cover classification.

Main Results

• Overall Performance: PMLv2, SSEBop v6.1, and FLUXCOM RS show the strongest and most consistent agreement with reference datasets in capturing irrigation-related spatial patterns and seasonal dynamics. MOD16A2 and GLEAM4.3a capture temporal variability but show more variable spatial agreement and generally lower ET magnitudes. ERA5-Land exhibits the weakest correspondence.
• Spatial Evaluation (vs. GMIAv5):
  • FLUXCOM RS shows the strongest spatial correspondence with irrigated areas (mean Pearson correlation coefficient R = 0.65 in arid/semi-arid regions, 0.43 in humid regions).
  • SSEBop v6.1 (R = 0.53 arid/semi-arid, 0.31 humid) and PMLv2 (R = 0.41 arid/semi-arid, 0.30 humid) also show good agreement.
  • ERA5-Land shows weak correspondence (R = -0.19 arid/semi-arid, 0.13 humid).
  • Correlations are generally higher in arid and semi-arid regions where irrigation creates stronger ET contrasts.
• Seasonal Dynamics and Magnitude (vs. OpenET):
  • SSEBop v6.1 shows the highest average ET during the irrigation season (106.6 mm per month), closely aligning with OpenET (mean bias of +10.9 mm per month) and high correlation (R = 0.94).
  • MOD16A2 yields the lowest mean ET (52.7 mm per month) and largest negative bias (-43.0 mm per month) relative to OpenET.
  • FLUXCOM RS, PMLv2, SSEBop v6.1, and GLEAM4.3a generally show high correlations with OpenET (often R > 0.9).
• Seasonal Dynamics and Magnitude (vs. Eddy Covariance):
  • FLUXCOM RS, PMLv2, MOD16A2, and GLEAM4.3a show strong temporal agreement with flux tower data (mean R values of 0.88, 0.90, 0.91, and 0.89, respectively).
  • PMLv2 and SSEBop v6.1 exhibit the lowest mean Root Mean Square Deviation (RMSD) values (25.4 and 23.3 mm per month, respectively).
  • All products tend to underestimate ET relative to tower measurements, with MOD16A2 and GLEAM4.3a showing the strongest negative biases.
• Interannual Variability and Drought Sensitivity: All products capture marked negative ET anomalies during the 2014 California Valley drought, with FLUXCOM RS showing the strongest standardized response (z-scores of -2.29 in June and -1.96 in July).

Contributions

• Presents the first comprehensive evaluation of six global ET products over irrigated croplands using three independent reference sources: irrigation infrastructure maps, the OpenET ensemble, and eddy-covariance measurements.
• Provides a multi-scale, multi-source assessment across diverse agro-climatic regions, offering a robust comparison of how global ET products represent irrigation-related ET variability.
• Identifies key factors influencing ET product behavior over irrigated areas, such as the representation of vegetation water stress (e.g., integration of LST or vegetation indices) and the inclusion of irrigated flux tower data in model training/calibration.
• Offers guidance for selecting appropriate ET products for applications in irrigated regions and highlights priorities for improving the representation of irrigation in global ET models.

Funding

• European Space Agency (Climate Change Initiative – Anthropogenic Water Use (CCI-AWU) precursor project, contract no. 4000142449/23/I-NB)
• European Space Agency (Climate Change Initiative – Land Evaporation (CCI-E) project, contract no. AO/1-12490/24/I-LR)

Citation

@article{Laluet2026Assessing,
  author = {Laluet, Pierre and Corbari, Chiara and Baez-Villanueva, Oscar M. and Walther, Sophia and Zhang, Yongqiang and Muñoz-Sabater, Joaquín and Albergel, Clement and Dorigo, Wouter},
  title = {Assessing the suitability of global evapotranspiration products over irrigated areas},
  journal = {Hydrology and earth system sciences},
  year = {2026},
  doi = {10.5194/hess-30-1779-2026},
  url = {https://doi.org/10.5194/hess-30-1779-2026}
}