Electronic Sentinel Networks for Intelligent Environmental Management: Prediction of Rainfall, Water Need, and Adjusted Irrigation

Modern agriculture faces major challenges in the sustainable management of water, particularly in regions affected by water stress. To address these issues, we propose an intelligent irrigation system based on a network of electronic sentinels and machine learning models. Agro-meteorological data from the Algiers station, covering the period 2016–2025 with 3439 records, were used to train three predictive modules: rainfall, water need, and adjusted irrigation. The originality of this work lies in integrating rainfall and water need predictions to anticipate adjusted irrigation, thereby optimizing water management. The approach, built on the Random Forest algorithm, achieved an accuracy of 84.1% for rainfall, a coefficient of determination R2 = 0.926 for water need (RMSE = 3.40 mm, MAE = 2.36 mm) and R2 = 0.939 for adjusted irrigation (RMSE = 3.14 mm, MAE = 2.44 mm). Combining these predictions enables more precise water management, reducing unnecessary consumption while maintaining soil moisture at levels favorable to plant growth. This system represents a significant step toward sustainable agriculture, contributing to climate change resilience and long-term water resource management.