A Machine Learning-Driven System for Football Field Irrigation using Satellite Analysis and Low Cost IoT Electronics

This project presents an intelligent, data-driven irrigation framework designed to regulate severe water
scarcity in football field management, with a specific focus on the extreme water-stressed climate of
Cyprus. Our system transitions traditional experience-based watering habits into a precise, automated
decision-support system. The architecture integrates multi-level technology: ground-based capacitive
soil moisture and AHT10 temperature and humidity sensor managed by a Kypruino UNO+ microcontroller,
remote sensing data from the ESA Sentinel-2 mission, and cloud-based machine learning analytics. While
initial deployment between January to March 2026 required manual data logging due to localized ESP8266
Wi-Fi connectivity limitations, an app was fully coded and optimized between April to June 2026. Our
model using Ridge Regression and XGBoost models established a high-accuracy correlation between
environmental variables and grass vitality, achieving a remarkably low Mean Absolute Error 0.0356. Our
app splits the football pitch into eight separate irrigation zones showing real-time soil, meteorological, and
satellite inputs against predefined system data. The app delivers zone-specific recommendations and
features a sustainability dashboard demonstrating reduction in water consumption, reduction in required
maintenance labor, and sprinkles operated duration.