The SWAMP project recently published a paper in a special issue of the IEEE Internet of Things Magazine focused on IoT-based smart agriculture. Advancing IoT-Based Smart Irrigation is the first SWAMP paper to specifically address the combination of IoT and artificial intelligence, together with physical agronomical models, to provide better irrigation prescription plans that at the same time save water and energy and improve productivity.
We developed an IoT-based platform for smart irrigation, with a flexible architecture to easily connect IoT and Machine Learning (ML) components to build application solutions. Our architecture enables multiple and customizable analytical approaches to precision irrigation, giving room for the improvement of ML approaches. Impacts to different stakeholders can be anticipated, as IoT professionals, by facilitating system deployment, and farmers, by providing cost reduction and safer crop yields.
The water need estimation process is divided up into two key activities. Soil water content and dynamics estimation consists in estimating soil water content and dynamics through the direct measurement of soil water content, rainfall, and irrigation, as well as physical models of soil water dynamics applied over the collected data (weather data mainly) and soil and crop characteristics. Soil water need forecast consists in calculating soil water content forecasts and water need forecasts for each moment of a planning horizon, using techniques such as simulation and machine learning algorithms.
We developed a smartphone Farmer App where farmers are informed of immediate water needs, water balance time series, and the current soil moisture information for a 3-depth sensor probe. With this real-time status of the farm at hand, and equipped with the optimized irrigation plans computed by the system, the farmer can achieve better use of the water resources without harming productivity.
SWAMP project pilots have just been deployed, they are operating properly, and data is being collected. The next step, expected by the end of 2020, is to analyze the data and disseminate quantitative impact results.