AISAAC – AI-based sustainability assessment of sub-area-specific managed cropland

The AISAAC project is developing a monitoring approach for the systematic, small-scale assessment of the sustainability of agricultural land, with the aim of optimising yields and identifying potential restoration areas in accordance with the EU Nature Restoration Regulation. This assessment will take into account resource use and crop yields. The project therefore focuses on researching AI-based methods for predicting crop yields from agricultural land managed on a plot-by-plot basis. An Explainable AI (XAI) model links historical data from partners Weinland Agrar and aGRAR-ZT with satellite, weather, soil and management data to map relationships between resource use, soil moisture and yields. In addition, the model incorporates soil moisture data from permanently installed sensors, thereby supporting both irrigation planning and yield modelling. High-resolution yield maps are generated using PlanetScope data and compared with Sentinel-2-based results. The methods are validated on a reference field. AISAAC contributes to sustainable agriculture, improves decision-making, reduces resource consumption and emissions, and promotes biodiversity by identifying suitable areas for restoration.