The volume of satellite EO data is rising rapidly, never before have so many EO missions been providing observation of our planet. These vast EO collections require new and innovative approaches to data management, processing and analysis. The conventional approach of local data storage and in-house processing is no longer feasible. More viable solutions take advantage of distributed storage and computing resources that are co-located with the EO data archives. Cloud computational infrastructures combined with scalable analytical platform capabilities represent a modern viable solution for operational EO data processing and analysis.
To respond to these challenges we have completely redesigned and upgraded our EO data processing workflows. We have developed our internal production framework that relies on open-source based, cloud agnostic and scalable solution. It fully supports distributed computing and server clustering enabling high availability, load balancing and parallel processing. Standalone processing tools dedicated to particular processing and analytical tasks are developed using the microservice architecture. Large scale production is possible thanks to the big data concept that is inherited in the system.
Robust, proven and operational processing chains are implemented covering diverse approaches for integrated analysis of optical & SAR satellite imagery. They are based on the use of state-of-the-art analytical methods, such as artificial intelligence (machine & deep learning, neural networks), decision trees or object-based image analysis. The EO data pre-processing and integration are centralized to generate Analysis Ready Data (ARD). The EO data analyses are supported by additional expertise, such as geospatial statistics, feature extraction, data mining or area frame sampling.