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.
Highlights
Our concept and implementation of a semantic Earth observation (EO) data cube architecture aims to facilitate semantic queries by including computer vision (CV) directly at the EO data cube level
Users can apply them as the basic building blocks to define general semantic entities; Convergence-of-evidence to increase the semantic granularity: semantic entities are defined using multiple sources of evidence, including color information, the temporal dimension, and potential auxiliary data and information, e.g., a digital elevation model (DEM); Extraction of land use/land cover units using the semantic entities in specific applications for better-posed big EO data queries and analyses (e.g., semantic content-based image retrieval (SCBIR, a method to retrieve images based on a semantic description of their content), cloud-free compositing, automatic change detection)
We developed Sen2Cube.at, a scalable semantic EO data cube architecture, as a step towards the production of an explainable AI-based CV system for big EO data, implementing it prototypically for Sentinel-2 images covering Austria
Summary
Our concept and implementation of a semantic Earth observation (EO) data cube architecture aims to facilitate semantic queries by including computer vision (CV) directly at the EO data cube level. Systems and interfaces with high applicability and many features are more use and require programming skills, limiting the target users (green polygon), e.g., web-based code editors. Images and presents an opportunity to implement semantic approaches in big EO data analyses These analyses allow users to formulate queries using semantic concepts instead of writing low-level procedural or declarative code, which starts with reflectance values. In automated big EO data analyses involving thousands of images (or even more), the meta quality indicators of products and processes becomes increasingly important Some of these indicators are not easy to achieve and obtain, e.g., the timeliness and frequency of products, or the transferability and reproducibility of methods. Shares our conclusions and outlook as well as a variety of other potential use-cases beyond using Sentinel-2 data covering Austria
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
