Kukal et al. (2026) Artificial intelligence/machine learning estimation of FAO and ASCE standardized reference ET erodes the physical basis and intended purpose of ET standardization

Identification

• Journal: Agricultural Water Management
• Year: 2026
• Date: 2026-07-22
• Authors: Meetpal Singh Kukal, Richard G. Allen, Ayşe Kiliç, Philip A. Blankenau, Kendall C. DeJonge, Kelly Robert Thorp, Justin Huntington, Luis Santos Pereira, R. López-Urrea, Maher Salman, Paula Paredes, Clarence W. Robison
• DOI: 10.1016/j.agwat.2026.110657

Research Groups

• Department of Soil and Water Systems, University of Idaho, Boise, ID, USA
• Department of Soil and Water Systems, University of Idaho, Kimberly, ID, USA
• University of Nebraska, Lincoln, USA
• Idaho Department of Water Resources, ID, USA
• USDA-ARS Water Management and Systems Research Unit, Fort Collins, CO, USA
• USDA-ARS Grassland Soil and Water Research Laboratory, Temple, TX, USA
• Desert Research Institute, Reno, NV, USA
• LEAF—Linking Landscape, Environment, Agriculture and Food Research Center, Associated Laboratory TERRA, School of Agriculture, University of Lisbon, Portugal
• Desertification Research Centre (CIDE), CSIC-UV-GVA, Moncada, Valencia, Spain
• FAO, Rome, Italy

Short Summary

The paper argues that using artificial intelligence/machine learning to directly estimate FAO and ASCE Standardized Reference Evapotranspiration (ETref) undermines the physical basis and intended purpose of ET standardization. Instead, AI/ML can be used to improve the quality, continuity, and representativeness of meteorological inputs for the FAO/ASCE ETref equation.

Objective

• Investigate the limitations and drawbacks of using artificial intelligence/machine learning (AI/ML) to directly estimate FAO and ASCE Standardized Reference Evapotranspiration (ETref)
• Explore alternative ways AI/ML can be leveraged to advance evapotranspiration (ET) science

Study Configuration

• Spatial Scale: Global, with a focus on agricultural regions
• Temporal Scale: Long-term, with a focus on climate variability and change

Methodology and Data

• Models used: FAO/ASCE ETref equation, AI/ML models for gap-filling and downscaling meteorological data
• Data sources: Satellite, observation, reanalysis datasets, including ERA5 and pyfao56

Main Results

• Direct estimation of ETref using AI/ML can lead to loss of physical basis and intended purpose of ET standardization
• AI/ML can be used to improve the quality, continuity, and representativeness of meteorological inputs for the FAO/ASCE ETref equation
• Reduced-input estimates of ETref can create a conceptual mismatch and undermine the importance of weather monitoring infrastructure

Contributions

• The paper contributes to the ongoing debate on the use of AI/ML in hydrology and evapotranspiration science, highlighting the limitations and drawbacks of direct estimation of ETref using AI/ML
• The paper provides an alternative perspective on the use of AI/ML to improve the quality, continuity, and representativeness of meteorological inputs for the FAO/ASCE ETref equation

Funding

• This research was funded by the University of Idaho, the USDA-ARS Water Management and Systems Research Unit, and the Desert Research Institute.

Citation

@article{Kukal2026Artificial,
  author = {Kukal, Meetpal Singh and Allen, Richard G. and Kiliç, Ayşe and Blankenau, Philip A. and DeJonge, Kendall C. and Thorp, Kelly Robert and Huntington, Justin and Pereira, Luis Santos and López-Urrea, R. and Salman, Maher and Paredes, Paula and Robison, Clarence W.},
  title = {Artificial intelligence/machine learning estimation of FAO and ASCE standardized reference ET erodes the physical basis and intended purpose of ET standardization},
  journal = {Agricultural Water Management},
  year = {2026},
  doi = {10.1016/j.agwat.2026.110657},
  url = {https://doi.org/10.1016/j.agwat.2026.110657}
}