Machine learning to predict nitrogen leaching at field level

Nitrate leaching is a major pollutant in freshwater systems, and farmers are encouraged to adopt management practices that can help reduce this leaching. However, current nitrate-leaching models lack precision, do not incorporate actionable management variables, and fail to deliver predictions at the field level. This deficit makes it difficult for farmers to implement effective practices to reduce nitrate leaching.

This PhD project aims to provide more accurate predictions of nitrate leaching at the field level by employing machine-learning methods and integrating various data sources. These sources will include monitoring networks, soil data, weather data, crop information, satellite images, and historical data. The project will involve collaboration with partners, including the departments of Ecoscience and Agroecology, as well as SEGES.