Lunel et al. (2026) Systematic overestimation of evapotranspiration over irrigated areas by an offline land surface model

Identification

• Journal: Hydrology and earth system sciences
• Year: 2026
• Date: 2026-08-13
• Authors: Tanguy Lunel, Belén Martí, Aaron Anthony Boone, Patrick Le Moigne
• DOI: 10.5194/hess-30-5117-2026

Research Groups

• Centre National de Recherches Météorologiques, Université de Toulouse, Météo-France, CNRS, Toulouse, France
• ISPA, INRAE, Bordeaux Sciences Agro, Villenave d’Ornon, France

Short Summary

This study quantifies the systematic overestimation of evapotranspiration by offline Land Surface Models (LSMs) over irrigated areas due to their neglect of atmospheric feedback from irrigation, finding an average overestimation of 25%. It details the underlying physical processes and the dependence of this bias on LSM configuration and crop water stress.

Objective

• To quantify the atmospheric effects of irrigation on near-surface meteorological conditions (temperature, humidity, wind speed) using coupled model simulations and observational data.
• To systematically investigate and quantify the atmospheric feedback effect on evapotranspiration in the Interactions Soil Biosphere Atmosphere (ISBA) Land Surface Model (LSM) for various configurations.
• To refine the understanding of LSM response to atmospheric feedback, particularly concerning modelled ecophysiological processes and the resulting overestimation of evapotranspiration in offline LSMs over irrigated areas.

Study Configuration

• Spatial Scale: Point-scale for offline ISBA simulations; 2 km horizontal resolution for coupled Meso-NH simulations over the LIAISE campaign domain (Ebro basin, northeastern Spain).
• Temporal Scale: Two weeks (15–29 July 2021) for the LIAISE Special Observation Period (SOP); assessed representativeness over the full LIAISE Long Observation Period (mid-April to mid-October 2021).

Methodology and Data

• Models used:
  • Interactions Soil Biosphere Atmosphere (ISBA) Land Surface Model (LSM) (offline point-scale simulations).
  • SURFace EXternalisée (SURFEX) platform (version 8.1).
  • Meso-NH atmospheric model (coupled with SURFEX for atmospheric forcing generation).
  • FAO-56 Penman-Monteith formulation (for reference evapotranspiration comparison).
• Data sources:
  • Land surface Interactions with the Atmosphere over the Iberian Semi-arid Environment (LIAISE) project field campaign data (15–29 July 2021).
  • In situ observations from La Cendrosa alfalfa field (Canut, 2022) and Institut de Recerca i Tecnologia Agralimentàries (IRTA) corn field (Martínez-Villagrasa et al., 2022).
  • ERA5 reanalysis (Hersbach et al., 2020) for LOP representativeness assessment.
  • ECOCLIMAP-II land cover database for albedo values.

Main Results

• Irrigation significantly cools the 2 m air temperature by 2.6 °C, increases 2 m specific humidity by 1.8 g kg⁻¹, and reduces 2 m wind speed by 0.7 m s⁻¹ on average during the LIAISE SOP over the La Cendrosa alfalfa field.
• Offline ISBA simulations, when forced with atmospheric conditions that neglect irrigation effects (i.e., warmer, drier, windier), systematically overestimate evapotranspiration over irrigated areas.
• For well-irrigated crops, this overestimation of evapotranspiration averages about 25%, ranging from 4% to 35% depending on the ISBA configuration.
• For water-stressed crops, the atmospheric feedback (cooler, more humid, less windy conditions) delays stomatal closure and advances stomatal reopening, which can compensate for or even lead to an increase in transpiration, resulting in a negligible or even negative (underestimation) bias in total evapotranspiration.
• The choice of stomatal conductance scheme (Jarvis vs. ISBA-A-gs) and canopy representation (composite vs. Multi Energy Balance (MEB)) significantly influences the magnitude and nature of the evapotranspiration bias. ISBA-A-gs allows for more stomatal compensation, reducing overestimation, while the composite approach exaggerates the compensatory effect of bare soil evaporation.
• The FAO-56 reference evapotranspiration formula, when fed with non-irrigation-affected atmospheric data, shows a similar overestimation (20% to 24%) to the ISBA model for irrigated conditions.

Contributions

• First detailed quantification of the atmospheric feedback effect on evapotranspiration in the ISBA LSM across a diverse range of model configurations, demonstrating its general applicability to other LSMs.
• Elucidation of the specific LSM processes (stomatal closure timing, interaction between transpiration and bare soil evaporation, influence of canopy representation) that modulate the atmospheric feedback effect under different water stress conditions.
• Robust quantification of the systematic overestimation of evapotranspiration (mean 25%) in offline LSMs when irrigation-induced atmospheric feedback is ignored, highlighting a critical bias for hydrological and agricultural applications.
• Utilizes high-resolution coupled model simulations and unique observational data from the LIAISE campaign to provide a strong basis for attributing atmospheric changes to irrigation.

Funding

• Agence Nationale de la Recherche, HILIAISE project (grant no. ANR-19-CE01-0017)

Citation

@article{Lunel2026Systematic,
  author = {Lunel, Tanguy and Martí, Belén and Boone, Aaron Anthony and Moigne, Patrick Le},
  title = {Systematic overestimation of evapotranspiration over irrigated areas by an offline land surface model},
  journal = {Hydrology and earth system sciences},
  year = {2026},
  doi = {10.5194/hess-30-5117-2026},
  url = {https://doi.org/10.5194/hess-30-5117-2026}
}