Smart Irrigation Decision Support Using Soil and Weather Data for Sustainable Agriculture

Efficient irrigation management is essential for sustainable agriculture, particularly under increasing water scarcity, climate variability, and growing pressure on freshwater resources. Conventional irrigation methods often rely on fixed schedules or instantaneous soil-moisture thresholds and therefore may not adequately consider short-term changes in soil and weather conditions. This study proposes a lightweight smart irrigation decision-support framework that integrates real-time soil measurements, environmental observations, short-term weather information, machine-learning-based soil-moisture prediction, and embedded edge computing. At the algorithmic level, soil moisture, recent moisture variation, air temperature, relative humidity, soil temperature, rainfall, and forecast precipitation are used to estimate near-future soil-moisture conditions. A weather-aware decision mechanism then determines whether irrigation is required and estimates the corresponding irrigation depth, water volume, and pump or valve operating duration. At the hardware level, a low-cost embedded controller acquires sensor data, receives weather information when connectivity is available, executes the selected lightweight prediction model locally, and generates either irrigation recommendations or direct actuator commands. Several machine-learning models, including Linear Regression, Decision Tree, Random Forest, Gradient Boosting, and a compact Multilayer Perceptron, are compared in terms of predictive accuracy and embedded computational requirements. The proposed approach is evaluated against fixed-schedule and soil-moisture-threshold irrigation using prediction error, water consumption, irrigation frequency, pump operating time, soil-moisture regulation, and inference performance. Ablation experiments further compare soil-only, weather-only, and combined soil-and-weather configurations. The framework aims to provide an affordable, interpretable, and resource-efficient solution for adaptive irrigation management in sustainable agriculture.

کلیدواژه ها:

Smart irrigation ، sustainable agriculture ، machine learning ، Internet of Things (IoT) ، edge computing ، precision agriculture