Digital twins have become a major technology trend and a critical component in the implementation of smart environments. With their ability to mimic the behaviour of real-world entities in virtual environments, they provide advanced monitoring, diagnostics, prognostics, and optimization capacities. However, as their implementation requires convergence of many technologies and non-technical aspects, ranging from Internet of-things to artificial Intelligence with integrated domain-specific knowledge, their usage today is limited to highly controlled environments, such as smart factories and smart homes. Their immense potentials to provide environmental intelligence, thus, remain unutilised, specially, when considering protection of Earth from degradation through sustainable management of its natural resources and urgent actions against climate changes.

Today, food production is amongst the main producers of greenhouse gases (GHG), while being under immense pressure due to the rapid urbanisation. It is, therefore, critical to address the trade-off between safeguarding food production, while lowering GHG emissions. This can only be achieved by deepening understanding of our interactions with agricultural ecosystems. The proposed project addresses contemporary challenges of digital twins for modelling such socio-environmental interactions by providing significant advances beyond state-of-the-art in the following aspects:

• A new in-situ , capable of simultaneously capturing CO₂, N₂O and CH₄ emissions, together with temperature and moisture of surroundings as well as levels of plant photosynthesis using quantum sensor with location data provided by Galileo,
• A data harvesting system, intended for gathering and aligning IDEAL’s in-situ data with open Earth Observation data sources (e.g. Copernicus satellite images, GEOSS thematic maps, and LiDAR data from Slovenian environmental agency) for common representation of spatiotemporal entities,
• An advanced data fusion framework designed for mining IDEAL’s data sources by the principles of deep and feature learning for spatiotemporal extrapolations and crop-growth simulations,
• Process optimization and visual analytics services for providing for prescriptive analytics capacities of socio-environmental interactions with the support of explainable artificial intelligence.

As a result, IDEAL digital twin shall enable:

• of farmers’ interaction with agricultural ecosystems,
• of green-house-gas emissions, soil health, and crop development parameters,
• of their changes during the time, and
• Optimization of farming processes, accordingly.

In accordance with user-centric design, project development shall be governed by three complementary pilots, each addressing the specifics of a particular agricultural ecosystem that all together cover 98% of Slovenian farmland, namely, grasslands, arable lands, and permanent crops. Within each of the pilots, systematic data collections shall be conducted periodically during crop and grass growth, before and after all major farming activities, including tillage, fertilization, planting, and harvesting in order to ensure accurate profiling of the following parameters:

• High-resolution GHG emission that includes CO₂, N₂O and CH₄.
• Soil health parameters and derived nutrition levels, as for example fertility indices, pH, and manganese, and
• Crop and grass development parameters based on their physical features like levels of photosynthesis productions and growth.

In order to maximize the project potentials, IDEAL digital twin shall be plugged-in into existing precision farming infrastructure provided by industrial partner (namely Igea d.o.o.), turning natural ecosystem into a smart environment. IDEAL shall, thus, provide the necessary social innovation infrastructure to the researchers and practitioners that are currently struggling with low level of general digitalization in agricultural sector.

This project has received funding from the Slovenian Research Agency, under "Public Call for co-financing of research projects in 2020", research grand L7-2633.