AgroTech project as an example of how CREODIAS can be used for food and environmental research

CloudFerro together with the QED Software team is working on a Research & Development project AgroTech to test and review its cloud capabilities for agriculture-related applications. After 1 year of collaboration on the project, we would like to share our main thoughts about implementing solutions for agriculture on CREODIAS.

EXTRACTING LAND INFORMATION FROM EARTH OBSERVATION DATA

One of the land cover aspects is the possibility of extracting its types, density, or frequency from Earth observation. Conscious land management requires a trustworthy source of current data. Land cover and land use classification is the most common basic source of information for the Earth's surface. Satellite imagery creates an opportunity to see the land surface in plenty of various bands and with wide radiometric resolution. Therefore, thematic aggregation is needed. Classification is used for simplifying the content as well as reducing the weight of data.

There are plenty of available methods and solutions for providing such information. CREODIAS users can conduct their own classification or choose from widely available ready thematic layers e.g., S2GLC or Copernicus High-Resolution Layers available on the CREODIAS EO data cloud repository. However, these methods have their limitations starting from chosen methodology and timeframe and ending with imperfect accuracy. They are great for general analysis on a wide scale of Europe, but when it comes to details, they seem insufficient.

AGROTECH AS A GOOD EXAMPLE

Looking at our users’ demands and needs, CloudFerro has started a research and development project AgroTech to test and review our cloud capabilities in order to provide the best possible environment for:

• creating new methods
• convenient gaining from existing thematic layers
• generating on-demand dedicated products
• efficient use of cloud infrastructure
• testing various settings and properties of Virtual machines suitable for user algorithms

It is a great challenge to provide satisfactory land cover classification on demand. CloudFerro team is facing it in the AgroTech project. It is our research goal to facilitate the usage of CREODIAS resources by generating thematic layers at any time and range of user needs. CloudFerro team is focused on implementing machine learning techniques in land cover and crop type classification, as well as the recognition of anomalies. Created methods will be ingested fully in the PGaaS (Product-Generation-as-a-Service) workflow and available for our clients as an example of the demonstration of CREODIAS capabilities.

AgroTech not only aims at creating innovative tools and information for CREODIAS users to incorporate in their value chains – ready-to-use classification models ranging from land cover to crop type and status classification and tools to train and retrain new networks. It also allows CloudFerro to evaluate and optimize its services. New capabilities will be offered both as containerized applications executed by PGaaS workflows – which require less configuration on the users’ side - and as new VM (Virtual Machine) images for the users to choose from.

Combining existing CREODIAS PGaaS and VM-based processing capabilities with Copernicus data repository and AgroTech results will further simplify generating added value from satellite imagery for the agriculture sector and beyond. Thus, it enables more stakeholders to participate in the market and promotes the uptake of Copernicus-born data throughout European industries.

Below you can find examples of conducted land cover classification as part of the AgroTech project for northern Belgium and the southern Netherlands.