Periodic Reporting for period 1 - AI4SoilHealth (AI4SoilHealth: Accelerating collection and use of soil health information using AI technology to support the Soil Deal for Europe and EU Soil Observatory)
Reporting period: 2023-01-01 to 2024-06-30
Healthy soils are crucial for ensuring safe food and water supplies worldwide and play a key role in reducing greenhouse gas emissions. However, 60 to 70% of soils in Europe are in poor condition and can't fully support these vital functions.
Project Objective: The AI4SoilHealth project (Jan 2023 – Dec 2026) is short for “Accelerating collection and use of soil health information using AI technology to support the Soil Deal for Europe and EU Soil Observatory”. The project aims at co-designing, creating and maintaining an open access European-wide digital infrastructure for the continuous monitoring and forecasting of Soil Health metrics, and supporting the improvement of the health of European soils in the future. To do so, AI4SoilHealth is working to:
- Identify robust & realistic Soil Health proxies for detecting state & change in soil health relative to the desired soil ecosystem functions/services, and
- Develop a sensor-fusion-based soil health assessment toolbox (spectroscopy, genomics and in-situ measurements).
- Map soil health indicators across Europe
The project will create a “Soil Digital Twin” infrastructure (mimicking the physical environment in a big-data virtual replica) to map, collect and integrate soil data. This open access environment will be based on AI methods (e.g. machine learning & artificial neural networks), acting as an interface for stakeholders. In turn, this aggregated data can help to effectively share soil health information to inform decision making and empower soil managers at all levels across Europe. Develop a flexible progressive web app-based cyberinfrastructure termed “AI4SoilHealth” (including Soil Health Data Cube) to serve users with current data at farm scale (30m spatial resolution or finer) with a secure API (Application Programming Interface).
The Consortium: The AI4SoilHealth consortium is led by Aarhus University (AU), a recognized leading center within soil physics and soil mapping and integrates 28 leading entities – covering 14 countries - with expertise in soil health research. The partners represent research institutions, companies, and non-profit organisations. More information on this project is available at: https://ai4soilhealth.eu/(opens in new window)
Project Objective: The AI4SoilHealth project (Jan 2023 – Dec 2026) is short for “Accelerating collection and use of soil health information using AI technology to support the Soil Deal for Europe and EU Soil Observatory”. The project aims at co-designing, creating and maintaining an open access European-wide digital infrastructure for the continuous monitoring and forecasting of Soil Health metrics, and supporting the improvement of the health of European soils in the future. To do so, AI4SoilHealth is working to:
- Identify robust & realistic Soil Health proxies for detecting state & change in soil health relative to the desired soil ecosystem functions/services, and
- Develop a sensor-fusion-based soil health assessment toolbox (spectroscopy, genomics and in-situ measurements).
- Map soil health indicators across Europe
The project will create a “Soil Digital Twin” infrastructure (mimicking the physical environment in a big-data virtual replica) to map, collect and integrate soil data. This open access environment will be based on AI methods (e.g. machine learning & artificial neural networks), acting as an interface for stakeholders. In turn, this aggregated data can help to effectively share soil health information to inform decision making and empower soil managers at all levels across Europe. Develop a flexible progressive web app-based cyberinfrastructure termed “AI4SoilHealth” (including Soil Health Data Cube) to serve users with current data at farm scale (30m spatial resolution or finer) with a secure API (Application Programming Interface).
The Consortium: The AI4SoilHealth consortium is led by Aarhus University (AU), a recognized leading center within soil physics and soil mapping and integrates 28 leading entities – covering 14 countries - with expertise in soil health research. The partners represent research institutions, companies, and non-profit organisations. More information on this project is available at: https://ai4soilhealth.eu/(opens in new window)
In the first 18 months, the project has made significant progress in developing the infrastructure to evaluate soil health across the EU. We have concentrated on five key activities. First, we have developed a robust indicator selection framework, and we are exploring new soil health indicators. Second, we established pilot sites and created field protocols and database standards, all contributing to the development of a sensor-fusion-based soil health assessment toolbox of the future, incorporating spectroscopy, genomics, and in-situ measurements. Third, we compiled and harmonized soil health data across various regions of Europe. Fourth, we worked on mapping soil health indicators at the pan-EU level. Lastly, we carried out activities to support the implementation of the Soil Monitoring Law.
Main achievements
The indicator selection framework has been formulated from previous policy and stakeholder needs, including consultation with the Joint Research Centre’s (JRC) Land Use and Coverage Area Frame Survey (LUCAS) soils team – and from synthesizing previous soil health indicators based on literature, policy and known databases across the EU, UK and elsewhere. A short synthesis of this work has been published and can be found here: https://www.openaccessgovernment.org/article/developing-a-robust-soil-health-indicator-selection-framework/178004/(opens in new window).
Multi actor engagement pilot sites have been set up across 11 different countries within different pedoclimatic conditions in the EU. These sites have identified various stakeholders and engagement strategies, as well as existing and new datasets to validate AI4SoilHealth predictions. We have also secured access to national datasets, data collection resources, and laboratory facilities for analysis. To ensure the production of harmonized data and storage, access and usability, we developed protocols for sampling methodology and an integrated data access system. We also selected the sensors (e.g. rapid enzymatic activity reader, and a soil temperature, moisture and salinity penetrometer) and methods (e.g. eDNA analysis, Beerkan water infiltration method, Moulder app for aggregate stability) to be tested within the pilot sites.
We have produced maps that represent or support the development of soil health indicators. All these maps are combined in a data cube, here called Soil Health Data Cube for Europe (SHDC4EU), and it is presented at 30-m spatial resolution and covering the period from 2000—2023+(https://zenodo.org/communities/ai4soilhealth/(opens in new window)). We also developed the core infrastructure for the soil health phone-app, which serves as a foundation for implementing the specified features and functionalities.
To support the Soil Mission and the Soil Monitoring Law, we are in contact with JRC, and we have published a short paper on the Soil Mission and an approach to soil health assessment. We have also written a paper on the UK soil monitoring experience with 5 decades of monitoring with design and development principles in support of the mission. We published a paper exemplifying the use of soil monitoring units for benchmarking soil health in Danish agricultural areas using SHIs and site-specific benchmarks, and we are working to implement it at Pan-EU scale with our new soil maps. We met with JRC colleagues in June to reprofile and refine the work priorities which are now being implemented. The major impact has been to reprofile work on vegetation cover, woodlands, and landscape heterogeneity to other soil descriptors, mostly carbon, salinity and biodiversity. The project is currently organising around internal clusters that will collaborate with a JRC representative with the cluster helping to set the agenda for the work plan.
Main achievements
The indicator selection framework has been formulated from previous policy and stakeholder needs, including consultation with the Joint Research Centre’s (JRC) Land Use and Coverage Area Frame Survey (LUCAS) soils team – and from synthesizing previous soil health indicators based on literature, policy and known databases across the EU, UK and elsewhere. A short synthesis of this work has been published and can be found here: https://www.openaccessgovernment.org/article/developing-a-robust-soil-health-indicator-selection-framework/178004/(opens in new window).
Multi actor engagement pilot sites have been set up across 11 different countries within different pedoclimatic conditions in the EU. These sites have identified various stakeholders and engagement strategies, as well as existing and new datasets to validate AI4SoilHealth predictions. We have also secured access to national datasets, data collection resources, and laboratory facilities for analysis. To ensure the production of harmonized data and storage, access and usability, we developed protocols for sampling methodology and an integrated data access system. We also selected the sensors (e.g. rapid enzymatic activity reader, and a soil temperature, moisture and salinity penetrometer) and methods (e.g. eDNA analysis, Beerkan water infiltration method, Moulder app for aggregate stability) to be tested within the pilot sites.
We have produced maps that represent or support the development of soil health indicators. All these maps are combined in a data cube, here called Soil Health Data Cube for Europe (SHDC4EU), and it is presented at 30-m spatial resolution and covering the period from 2000—2023+(https://zenodo.org/communities/ai4soilhealth/(opens in new window)). We also developed the core infrastructure for the soil health phone-app, which serves as a foundation for implementing the specified features and functionalities.
To support the Soil Mission and the Soil Monitoring Law, we are in contact with JRC, and we have published a short paper on the Soil Mission and an approach to soil health assessment. We have also written a paper on the UK soil monitoring experience with 5 decades of monitoring with design and development principles in support of the mission. We published a paper exemplifying the use of soil monitoring units for benchmarking soil health in Danish agricultural areas using SHIs and site-specific benchmarks, and we are working to implement it at Pan-EU scale with our new soil maps. We met with JRC colleagues in June to reprofile and refine the work priorities which are now being implemented. The major impact has been to reprofile work on vegetation cover, woodlands, and landscape heterogeneity to other soil descriptors, mostly carbon, salinity and biodiversity. The project is currently organising around internal clusters that will collaborate with a JRC representative with the cluster helping to set the agenda for the work plan.
Expected impact: By developing a set of soil health indicators and the pan-european maps the AI4SoilHealth project will help key actors from politicians to farmers to reverse the decline in soil health in Europe and bring healthier soils for all, from food producers to recreational users. Also, in the end, thanks to AI4SoilHealth, land users will be able to take out their phones, scan their soils and get real-time advice about the health of their soils. They will have access to information that will allow them to make data-driven decisions regarding land management and ultimately will help in improving the health of European soils.