The Austrian Semantic EO Data Cube Infrastructure

Abstract

Big optical Earth observation (EO) data analytics usually start from numerical, sub-symbolic reflectance values that lack inherent semantic information (meaning) and require interpretation. However, interpretation is an ill-posed problem that is difficult for many users to solve. Our semantic EO data cube architecture aims to implement computer vision in EO data cubes as an explainable artificial intelligence approach. Automatic semantic enrichment provides semi-symbolic spectral categories for all observations as an initial interpretation of color information. Users graphically create knowledge-based semantic models in a convergence-of-evidence approach, where color information is modelled a-priori as one property of semantic concepts, such as land cover entities. This differs from other approaches that do not use a-priori knowledge and assume a direct 1:1 relationship between reflectance values and land cover. The semantic models are explainable, transferable, reusable, and users can share them in a knowledgebase. We provide insights into our web-based architecture, called Sen2Cube.at, including semantic enrichment, data models, knowledge engineering, semantic querying, and the graphical user interface. Our implemented prototype uses all Sentinel-2 MSI images covering Austria; however, the approach is transferable to other geographical regions and sensors. We demonstrate that explainable, knowledge-based big EO data analysis is possible via graphical semantic querying in EO data cubes.

Semantic Querying in Earth Observation Data Cubes

“…We see it as an additional benefit of our approach that the implemented workflow of chaining together different building blocks into a query recipe can easily be supported by a visual programming interface, taking away the need for an application expert to write code. This is demonstrated already in an operational setting by Sen2Cube.at, a nation-wide semantic data cube infrastructure for Austria, which uses the semantique package in the background (Sudmanns et al, 2021a).…”

“…Translating these values into information about the real world therefore requires interpretation, bridging the so-called sensory and semantic gaps (Arvor et al, 2019). This is not a trivial task, because it involves reconstructing the four-dimensional physical world from two-dimensional imagery containing data that can only describe a limited set of properties of a real-world entity or event, such as its color (Sudmanns et al, 2021a). Dodging these barriers requires advanced technical expertise in the field of EO analytics.…”

Data Cubes for Earth System Research: Challenges Ahead

“…Cloud computing environments offer a promising solution for effectively managing vast amounts of Earth system data. Platforms such as GEE, the European Open Science Cloud (EOSC) 29 , Google Colaboratory 30 , Amazon SageMaker 31 , DeepESDL 32 , and Kaggle 33 provide opportunities for efficient data storage, processing, and collaboration in scientific research. However, it is important to note that these platforms often have certain limitations imposed on users.…”

“…Going beyond research applications, data cubes have also played a pivotal role in informing governmental actions and policies, as evidenced by their integration into national data cube frameworks [23,24]. Examples of these initiatives are, among others, the data cubes produced under the Committee on Earth Observation Satellites (CEOS) Open Data Cube [ODC, 25] 1 initiative (e.g., Digital Earth Australia, 26, 27, the Colombian Data Cube, 28, the Swiss Data Cube, 29), the Semantic Austrian EO Data Cube Infrastructure [30], the Euro Data Cube 2 system (providing a platform for efficient use of ARD), as well as tailored regional [e.g., the Regional Earth System Data Lab, 20] and global data cubes [e.g., the Earth System Data Cube, 8].…”

Collection, Standardization and Attribution of Robust Disaster Event Information—A Demonstrator of a National Event-Based Loss and Damage Database in Austria

“…Since event data are mostly recorded based on point features and usually only at the time of the event, we investigated how additional evidence from open and freely available Earth Observation data could be derived to complement the event information by (1) spatial explicit delineation of events with larger impacts (e.g., storm damages) and (2) to monitor areas before and after an event for exact time identification and also potential recovery monitoring. In our prototypical experiments, we used the Sen2Cube.at [23] (www.sen2cube.at, accessed on 26 April 2022) system, a semantic data cube of all available Sentinel-2 data for Austria (>13,000 images until end of 2021). The system allows spatiotemporal ad hoc queries based on semantic concepts for local damage events from 2015 (launch of the first Sentinel-2 satellite) to present.…”