Thorp et al. (2025) The pyfao56 automatic irrigation scheduling algorithm

Identification

• Journal: Agricultural Water Management
• Year: 2025
• Date: 2025-12-01
• Authors: Kelly R. Thorp, Kendall C. DeJonge, Meetpal S. Kukal
• DOI: 10.1016/j.agwat.2025.110013

Research Groups

• USDA-ARS, Grassland Soil and Water Research Laboratory, Temple, TX, USA
• USDA-ARS, Water Management and Systems Research Unit, Fort Collins, CO, USA
• Department of Soil and Water Systems, University of Idaho, Boise, ID, USA

Short Summary

This paper develops and demonstrates a new automated irrigation scheduling algorithm within the pyfao56 Python package, allowing flexible simulation of diverse irrigation management strategies based on 25 user-specified parameters, validated with a 2018 Arizona cotton field study. The methodology provides a flexible tool for simulating realistic irrigation management schedules with practical relevance for various field applications.

Objective

• To develop a new pyfao56 methodology for computation of irrigation schedules based on 25 user-specified parameters.
• To demonstrate the approach using data from a 2018 cotton field study at Maricopa, Arizona, by formulating and evaluating alternative irrigation schedules.

Study Configuration

• Spatial Scale: Field-scale study conducted on 64 cotton plots at the Maricopa Agricultural Center, Maricopa, Arizona, with individual plot analysis (e.g., plot "p06-1").
• Temporal Scale: Daily timestep simulations covering the 2018 cotton growing season, from 18 April 2018 (Day of Year 108) to 30 October 2018 (Day of Year 303).

Methodology and Data

• Models used:
  • pyfao56 Python package (version 1.4.0), an open-source evapotranspiration-based water balance model.
  • American Society of Civil Engineers (ASCE) Standardized Reference Evapotranspiration Equation.
  • FAO-56 dual crop coefficient methodology.
• Data sources:
  • Field data from a 2018 cotton irrigation management study at Maricopa Agricultural Center, Arizona.
  • Meteorological data from an Arizona Meteorological Network (AZMET) weather station (1.2 km from the field site), including solar radiation (MJ m⁻² d⁻¹), maximum and minimum daily air temperature (°C), average daily dew point temperature (°C), maximum and minimum relative humidity (%), daily average wind speed (m s⁻¹), and daily precipitation (mm).
  • Weekly soil water content measurements (0.1 to 1.9 m depth in 0.2 m increments) using a neutron moisture meter.
  • Soil texture information, laboratory analysis (hydrometer method), and pedotransfer functions for soil water limits.
  • Applied irrigation management data from the field trial.

Main Results

• Pyfao56 model evaluation across 64 plots showed root mean squared errors (RMSE) between measured and modeled root zone soil water depletion (Dr) ranging from 8–15 mm.
• Modeled transpiration was strongly related to measured seed cotton yield, with a coefficient of determination (r²) of 0.75.
• Autoirrigation scenarios demonstrated wide variability in outcomes for a single plot (p06-1):
  • Number of irrigation events ranged from 8 to 103.
  • Total seasonal irrigation ranged from 642 mm to 1110 mm.
  • Maximum daily irrigation ranged from 10.0 mm to 107.5 mm.
  • Estimated seed cotton yield varied by 24% (4.036–5.314 Mg ha⁻¹).
• Scheduling based solely on management allowed depletion (MAD) often resulted in impractical schedules (e.g., excessive daily amounts, inconsistent frequency).
• The multi-parameter set approach, adjusting Dr triggering criteria for different growth stages, effectively avoided water stress and maintained Dr below readily available water (RAW).
• ET replacement strategies (e.g., replacing 5-day actual ET) effectively mimicked the actual field irrigation schedule, with total irrigation and ETa within 12 mm of actual totals.
• The algorithm successfully incorporated practical constraints such as specific irrigation days, minimum/maximum daily application rates, and fixed irrigation amounts, demonstrating its flexibility.
• A scenario (est2018) successfully approximated the actual 2018 field irrigation schedule, achieving an equivalent number of events, total irrigation within 2 mm, maximum daily irrigation within 1 mm, and total ETa within 7 mm, with minimal root mean squared deviation (RMSD) for Dr, ETa, and Ks compared to the actual schedule.
• Generalized scenarios for furrow flood, overhead sprinkler, and subsurface drip irrigation methods revealed distinct water balance and stress patterns, highlighting the algorithm's ability to simulate diverse production-scale practices.

Contributions

• Development of a novel, comprehensive, and flexible automatic irrigation scheduling algorithm within the pyfao56 Python package, featuring 25 user-specified parameters.
• Introduction of the "AutoIrrigate" software object, significantly expanding pyfao56's utility for scenarios where explicit irrigation schedules are unknown or future management needs prediction.
• Demonstrated the algorithm's capability to simulate realistic irrigation management schedules that account for various irrigation methods, system constraints, and management philosophies, addressing limitations of oversimplified approaches in existing models.
• Provides a tool for rapid testing of diverse irrigation management practices and forecasting future irrigation requirements for in-season scheduling, including mid-season predictions of seasonal water use.
• Enables novel long-term (multi-decadal) simulation studies and multi-scale geospatial analyses with realistic irrigation schedules, facilitating the development of data-driven irrigation guidelines.

Funding

• USDA Agricultural Research Service, USA (Project No. 2020-13660-009-000-D, 3098-21600-001-000-D, and 3012-13210-001-000-D)
• SCINet project and AI Center of Excellence of the USDA Agricultural Research Service (Project No. 0201-88888-003-000D and 0201-88888-002-000D)
• Cotton Incorporated, USA (Project No. 17-642)
• Agricultural Water Center of Excellence at the University of California at Davis, USA (Award No. 2021-68012-35914)

Citation

@article{Thorp2025pyfao56,
  author = {Thorp, Kelly R. and DeJonge, Kendall C. and Kukal, Meetpal S.},
  title = {The pyfao56 automatic irrigation scheduling algorithm},
  journal = {Agricultural Water Management},
  year = {2025},
  doi = {10.1016/j.agwat.2025.110013},
  url = {https://doi.org/10.1016/j.agwat.2025.110013}
}