A new cloud-based processing system
The improved processing workflows will be implemented and deployed as a flexible and open processing system to further boost user-friendliness. This activity will include the necessary actions to deploy the system as a cloud-based processing system, to offer WorldCereal as a service through the ESA Network of Resources (NoR) program and to develop a new set of user interfaces allowing easy access to dedicated tools for downloading products, submitting and monitoring customized processing jobs both for model training and inference and computing map-based statistics based on generated products.
To accomplish these goals, the WorldCereal processing system (see figure) will be integrated into openEO by means of dedicated openEO process graphs. A cloud-based openEO backend offers access to the necessary EO data archives and computational power to satisfy both the demands of (a) generating global high resolution products and (b) running user-defined model training and inference over a custom area and time period. The WorldCereal system will be deployed on the new Copernicus Data Space Ecosystem backend.
Communication between the different system components (RDM, VDM and openEO processing system) will be managed through STAC catalogs. New in-situ reference datasets being added to RDM will trigger automated training data extraction workflows in openEO, enabling smooth usage of these data for model training. Generated WorldCereal products will be added to a product catalog, from where they can be easily visualized in the VDM.
The transformation of the WorldCereal system into a cloud-based openEO service allows for easy access, a smooth user experience and seamless integration with ESA Network of Resources.
A local version
By default, the new WorldCereal system will be offered as a cloud-based processing system. Interested users will however have the possibility to implement and run the system locally (on-premise) in stand-alone mode. Dedicated instructions will be prepared on how to start the OpenEO backend using a Docker container to guarantee a consistent environment. The local workflows can be preconfigured with remote STAC catalogs for EO data access, or the user can configure own collections that are exposed via STAC metadata. This should allow a user with profound IT skills to set up the system locally with limited effort.