Israeli scientists use AI to improve irrigation and detect plant stress earlier

Israeli researchers have shown that artificial intelligence tools can significantly improve irrigation management and help detect plant stress at an earlier stage. This is according to a new study by the Hebrew University of Jerusalem, which demonstrates how machine-learning models can be used to accurately predict daily crop water use.

The study focuses on plant transpiration — the process by which water evaporates through leaves and a key indicator of a plant’s actual water consumption. Unlike conventional approaches that rely on indirect indicators such as weather data or soil moisture, the researchers analyzed the plants’ own physiological behavior. The work is based on seven years of continuous, high-resolution measurements of tomatoes, wheat and barley grown under semi-commercial greenhouse conditions.

Using a high-precision load-cell lysimeter system, the research team recorded subtle changes in plant weight in real time, enabling exceptionally accurate measurement of daily transpiration. These data were then used to train machine-learning models, including Random Forest and XGBoost, which successfully predicted daily transpiration across different crops and environmental conditions. In independent tests, the XGBoost model achieved an R² value of 0.82.

A key innovation of the study is that the model predicts how a healthy, well-irrigated plant should behave. Plant biomass and daily temperature emerged as the most influential variables affecting water use. When actual plant behavior deviates from the model’s prediction, such differences may serve as early warning signals of stress caused by drought, salinity, disease or root damage — often before visible symptoms appear.

Although the current approach is mainly suited to research and controlled growing environments, the authors see it as an important step toward plant-driven decision tools. In the longer term, combining such AI models with field-ready sensors could underpin new precision-agriculture systems that improve irrigation scheduling, save water and provide early alerts of emerging plant stress.

Tags: Israel, AI, irrigation