openEO Platform is a new cloud processing and analytics environment built on top of openEO. It brings openEO to production and offers data access and data processing services to the EO community. The project is funded by ESA and starts in September 2020. The consortium working on this project consists of EODC (lead), EGI, EURAC, GEO, Sinergise, VITO, and WWU Münster.
openEO Platform builds upon the EO cloud processing platforms managed by VITO, Sinergise, and EODC, and the platform management and software development experience of all partners. openEO platform unites:
integrating all of these with the openEO API, a community project that started with H2020 funding and that comes with an ecosystem of user-friendly, graphical, and command-line-based clients. The openEO API’s original goal of fostering the current heterogeneity in interfaces of cloud-based EO processing back-ends matches the ESA requirement of realizing a federated infrastructure and matches the current situation that no single processing back-end serves (or will any time soon serve) all datasets of interest.
An idea that is unique to OpenEO is the combined facility of predefined and user-defined functions. This ‘best of both worlds’ approach allows optimization of everyday tasks by the backend provider, while still giving the user access to the set of libraries and tools that reflect years of work by the earth observation and open source community such as TensorFlow, Keras, GDAL, SNAP, and OpenCV. Using such ‘user-defined functions’ will not require a complex process but will be as simple as writing local code.
To scale up the processing function and workflow library available in OpenEO, as well as available examples and documentation, we want to engage the users to share their work and participate in the community. This can be done in the ‘classical’ way, by providing open source code, documentation or simply being an active forum member, or by providing their work on a marketplace, where a researcher or company can actually be rewarded for making a workflow, product or service available.
Machine learning (ML) and artificial intelligence (AI) are key technologies enabled by the platform. These data-hungry algorithms rely on combining earth observation raster data together with other datasets in various formats, which will be supported by flexible data access mechanisms. The platform will support integrated ML algorithms for ease-of-use, external libraries such as Tensorflow for flexibility, and export for use in external AI services to build a truly federated ecosystem. Data Cube service will be upgraded to serve as a distribution mean for various derived products, not just for visualization but also for ad-hoc analysis of the results.