Independent site about precision irrigation and the FIGARO research project (2012–2016). Not affiliated with the FIGARO consortium, its partners or the European Commission. About this site EN | FR
FIGARO Irrigation Notes
Wheel-line sprinkler irrigating a meadow

AquaCrop: how FAO's crop model turns water into yield, and what it needs to be trusted

Published on 19 November 2026 · FIGARO Irrigation Notes

Photo: rawpixel (CC0, credits)

FAO-56 tells a farmer how much water a crop uses. It does not say what happens to the harvest when the crop gets less. That second question is the one that matters when water is short, and it is the one AquaCrop, the crop model published by FAO, was built to answer. Part of the FIGARO project's research on deficit irrigation relied on it. This article explains how the model reasons, what it needs and where its answers stop.

What AquaCrop is for

AquaCrop was developed by FAO's Land and Water Division to simulate the yield response of herbaceous crops to water: cereals, cotton, potato, tomato, sugar beet and similar field and vegetable crops. It is meant for conditions where water is the main factor limiting production, from rainfed farming to full irrigation. The model was introduced in 2009 in a series of papers in Agronomy Journal, starting with the description of its concepts by Pasquale Steduto, Theodore Hsiao, Dirk Raes and Elías Fereres, and it became the core of FAO Irrigation and Drainage Paper 66, Crop yield response to water (2012).

FAO lists the uses it has in mind: comparing attainable and actual yields, identifying constraints on water productivity, developing irrigation schedules, designing strategies under water shortage such as deficit irrigation, and studying the effect of climate change on crop production. The intended users are practitioners in extension services, government agencies, NGOs and farmer associations, as well as researchers and economists who need a simple model for scenarios. The current release, version 7.3, is distributed free of charge with a graphical interface, as stand-alone programs for Windows, Linux and macOS, and as Python source code.

The calculation in four steps

The model's authors describe its growth engine as water-driven: transpiration is calculated first, and everything else follows from it. Each simulated day goes through four steps.

  1. Green canopy cover. Instead of the leaf area index used by many crop models, AquaCrop follows the fraction of the ground covered by green leaves, from zero at sowing to a maximum at mid-season, then declining as the crop senesces. Canopy cover can be estimated by eye, from a photograph or from satellite images, which makes the model easier to check against a real field.
  2. Transpiration. With enough water, crop transpiration is the reference evapotranspiration (ETo) multiplied by a crop coefficient proportional to canopy cover. The rest of the water lost from the field is soil evaporation, which the model calculates separately. This split matters: water that evaporates from bare soil between rows produces no biomass.
  3. Biomass. Above-ground biomass is proportional to cumulative transpiration. The factor of proportionality is the biomass water productivity, and AquaCrop normalises it for the evaporative demand of the climate and for the CO2 concentration of the air. The normalised value is treated as a conservative, crop-specific parameter that holds across sites and seasons. It differs mainly between plant types: C4 crops such as maize produce more biomass per unit of normalised transpiration than C3 crops such as wheat or potato.
  4. Yield. The harvestable yield is the biomass multiplied by a harvest index. The model starts from a reference harvest index for the crop and adjusts it during the season according to the stresses the crop has met, and when they occurred.

Written as a chain, the logic is: water available in the root zone, then canopy and stomata, then transpiration, then biomass, then yield. In the FAO-56 method described in our article on crop coefficients and evapotranspiration, the calculation stops at the second link; AquaCrop carries it through to the harvest.

How water stress is represented

The soil water balance is kept day by day for the root zone, as in FAO-56: rain, irrigation and capillary rise come in, evaporation, transpiration and deep percolation go out. When the root zone dries, AquaCrop does not apply a single penalty. It uses four separate stress responses, because plants do not react to drought all at once:

Each response starts at its own threshold of soil water depletion, and the thresholds shift with the evaporative demand of the day. This structure is what makes the model useful for deficit irrigation: it can show that the same seasonal water cut costs little when it falls during vegetative growth and a great deal when it falls at flowering. Temperature stress on pollination, soil fertility and salinity are also represented in simplified form, so the effect of a salty irrigation water can be explored alongside the water deficit.

What it needs

FAO groups the inputs into four families:

InputContentWhere it usually comes from
ClimateDaily minimum and maximum temperature, ETo, rainfall, CO2 concentrationField or agro-meteorological station; ETo computed with the FAO Penman-Monteith method
CropDevelopment stages, canopy growth and decline, rooting depth, water productivity, harvest index, stress thresholdsCalibrated crop files supplied with the model, with cultivar-specific parameters adjusted locally
SoilLayers, texture, water content at saturation, field capacity and wilting point, permeability; depth and salinity of any shallow water tableField observations, default values or pedotransfer functions
ManagementFertility level, mulches and practices affecting the water balance, weed control; irrigation method, dates and depths, irrigation water salinityThe farmer's records or the schedule being tested

The crop files carry the parameters that the developers consider conservative, such as normalised water productivity, and those that depend on the variety and the sowing date, such as the length of each growth stage. A serious application starts by checking the second group against local observations: emergence date, time to full canopy, flowering and maturity.

How FIGARO used it

The FIGARO partners used AquaCrop as a research tool for irrigation scheduling. Raphael Linker and colleagues at the Technion and partner institutes used it to compute optimal deficit irrigation schedules for cotton, potato and tomato in a simulation study published in Agricultural Water Management in 2016, and further papers applied it to cotton in northern Greece and to irrigation scheduling with imperfect weather forecasts. The full references are on our scientific publications page. The approach was to let the model explore many possible schedules over past seasons and find the one that gave the best yield for a given amount of water, something no field trial could do at that scale.

Limits to keep in mind

Used within these limits, AquaCrop gives what an evapotranspiration table cannot: an estimate of what a given volume of water is worth in yield, and of when each millimetre counts most.

Sources