The management of long EO data time series of continuing or historical missions, with up to 40 years of data available already today, requires dedicated technical solutions and technologies which might differ from the ones exploited by existing systems.
The ESA project LOOSE aims at investigating and analysing available and mature new technologies for such long EO data time series processing.
LOOSE developed a Blueprint infrastructure connecting mature open source components from different heritages into a comprehensive framework. This includes ingestion, discovery, exploitation-optimized access, processing and optimized analysis of EO data timeseries.
Throughout the project, several thematic real-world pilot-applications were implemented to underline the capabilities of the LOOSE platform. This demo will refer to such an application: The reprocessing of the 40 years AVHRR instrument data time series conducted at German Aerospace Center Oberpfaffenhofen. The demo illustrates several technical features and trade-offs explored in LOOSE, specifically:
• the comparison of openEO (aka “white-box-processing”) vs. OGC EO Application Packages (aka “black-box-processing”) for EO timeseries processing – particularly from the developer’s perspective,
• the usage of STAC metadata for management of EO timeseries processing,
• the scalability issues involved in massive parallel processing using DASK,
• the “bridges” implemented between Kubernetes and HPC paradigms for the openEO > Actinia > GRASS GIS implementation,
• the asynchronous archive bulk-reload and ingestion capabilities for EO timeseries production.