Abstract
In Mediterranean agriculture, olive orchards represent a crucial cultural and economic asset, yet they face severe water scarcity challenges aggravated by climate change. Traditional irrigation practices often lead to either over-watering or water stress, both of which negatively impact yield and resource efficiency. This study presents a novel, deep learning-based smart irrigation framework designed to optimize water conservation in Mediterranean olive orchards using Internet of Things (IoT) soil moisture sensors. We deployed a wireless sensor network across a representative olive grove in southern Spain to collect high-resolution temporal data on soil volumetric water content, ambient temperature, and relative humidity. These multi-sensor streams were integrated with local meteorological forecasts and processed using a Long Short-Term Memory (LSTM) recurrent neural network to predict soil moisture dynamics up to 72 hours in advance. Based on these predictive insights, an automated irrigation controller dynamically schedules watering events to maintain optimal soil moisture levels while minimizing water waste. Over a six-month trial period, the proposed system demonstrated a 28.4% reduction in water consumption compared to conventional evapotranspiration-based scheduling, while maintaining crop health and expected olive yields. Our findings demonstrate that integrating deep learning with IoT networks offers a highly scalable and resource-efficient solution for sustainable orchard management in water-stressed regions.
