IoT-Enabled Precision Irrigation Management Using Soil Moisture Sensor Networks and Machine Learning-Based Evapotranspiration Prediction

Irrigated agriculture accounts for approximately 70% of global freshwater withdrawals, yet irrigation water use efficiency in traditional flood irrigation seldom exceeds 40–50%. The convergence of IoT sensor technology, wireless communication, cloud computing, and machine learning presents a transformative opportunity to improve agricultural water use efficiency through real-time precision irrigation management. This study presents the design, implementation, and two-season field validation of an IoT-enabled precision irrigation management system (IoT-PIMS) for rain-fed rice cultivation in semi-arid agro-climatic zones of Telangana, India. The system integrates a wireless sensor network of 144 soil moisture sensors, automated weather stations, LoRaWAN communication, and a cloud-hosted Random Forest ET₀ prediction model trained on 12 years of IMD weather data, with a fuzzy logic irrigation decision engine delivering recommendations via SMS and Android app. Two-season field trials (Kharif 2023 and Rabi 2023–24) across three sites demonstrated: 31.4% water use reduction relative to conventional flood irrigation; paddy yield improvement from 5.21 to 5.84 t/ha; water use efficiency improvement from 0.41 to 0.67 kg grain/m³ (+63.4%); ET₀ prediction RMSE of 0.31 mm/day (NSE=0.89); and system uptime of 97.4%. Farmer perception surveys indicate 84.6% willingness to continue using IoT-PIMS.