Artificial intelligence saves from drought

Using climate data and powerful neural networks, Spanish researchers have developed a breakthrough tool for the olive industry. Scientists from the Faculty of Agronomy at the University of Cordoba have developed an artificial intelligence tool that will help farmers predict how much water they will need for irrigation for the week ahead.

The researchers added that this latest tool, LSTMHybrid, is part of a broader effort to digitize irrigation, which they say will help farmers reduce production costs by saving water and energy.

The latest tool is based on the Cangenfis model, developed in 2021 and trained using four years of climate data from Zuhar in the Andalusian province of Granada. When implemented, it can predict long-term irrigation water needs with an accuracy of up to 80%, and for various crops, including rice, corn and tomatoes.

LSTMHybrid allows farmers to plan their water needs more precisely and compare expected irrigation needs with different tariff periods. The researchers hope that these more accurate data will help farmers make the most economically sound decisions to optimize the use of water and energy.

The need to modernize Spain's irrigation system, which, according to researchers, has traditionally been based on historical experience rather than forecast data, has become even more urgent due to the ongoing drought and dangerously low water levels in reservoirs.

While CANGENFIS used hundreds of neural networks that take into account half a million different factors, LSTMHybrid makes its predictions based on average temperature, reference total evaporation, humidity and previous watering records.

The new model can also save previously entered data, which helps improve its ability to predict year after year.

This simplification allows farmers and irrigation managers to manually enter weekly data into the system using a conventional computer, predicting how much water will be needed for irrigation next week. “The knowledge is there, and the technology has been tested and is working,” said third lead researcher Emilio Camacho. “Now we have to develop a tool that will allow communities to use this technology in a simple way so that companies that are going to provide a technological solution to the irrigation community implement these advances.”