A new algorithm simplifies and improves calculations for the prediction of crops' water requirements

Evapotranspiration is defined as the amount of water lost by crops through evaporation from the soil and transpiration from plant cover. Its estimation is fundamental to calculate the water needs of plants, allowing for the establishment of an efficient irrigation program to minimize water consumption. Traditionally, the reference values of this parameter are calculated usinga complex algorithm, a mathematical function based on energy balance and requiring a large amount of climatic data. But, would it be possible to simplify this process to get the same result? What if a shorter path could be found that leads to the same place?

This is precisely what a research group in the Engineering Projects area at the University of Cordoba has achieved in a new study whose results are published in the journal Agronomy. The team has developed new models based on Artificial Intelligence that produce, a week in advance, estimates clarifying the water needs of crops through a new algorithm that, unlike others developed to date, needs fewer meteorological variables.

Specifically, it is fed up to 9 variables, all related to thermal parameters, such as, for example, the times of high and low temperatures, and thermal energy. "The great advantage of this algorithm," explains researcher Javier Estévez, is that "with a thermometer, practically, you can accurately predict the reference evapotranspiration and, subsequently, the crop's water demand."

To calculate exactly the same thing, traditional systems need to have a whole series of meteorological variables,like wind speed, humidity and solar radiation; "parameters that are more expensive to measure and, unlike temperature, not available at all weather stations," concludes Professor Estévez.

These models developed make it possible to estimate, with great reliability, the amount of water that a certain crop needs, a week in advance, although it performs better in inland areas, where the temperature is not affected by large bodies of water. In addition, during the first three days, the predictions are more accurate,achieving better results than other models published in the scientific literature," says Juan Antonio Bellido, another of the authors who participated in the study.

The algorithm, developed under the auspices of the SMARITY research project, has been validated in five locations in Andalusia subject to different climatic aridity conditions, from wetter areas, near Cadiz and Huelva, to desert areas, in Almeria. According to researcher Amanda García, both the model and the source code are available in open format, "which means that anyone who wants to use it can do so, improve it, and adapt it to any type of crop."

In this way, the new work helps to improve the management of water resources and to establish irrigation schedules that optimize water use,which is vital in areas suffering from scarce rainfall and contributes to the fight against drought, one of the red-letter objectives on the United Nations' road map.