ACube4Floods

Starting point / motivation

Flooding is a recurrent phenomenon in Austria. Every year sees at least several smaller-scale flood events occurring in different parts of the territory. While the extent and duration of these events are limited, their combined impact on society and economy is significant. Unfortunately, field work to assess the flood status and its impact are time demanding and costly.

Current state-of-the-art approaches for mapping floods with satellite images such as provided by Sentinel-1 and Sentinel-2 often fail to capture these smaller-scale events. While some of these flood events may indeed be impossible to capture due to sampling limitations, an important part of the problem is that current approaches for flood mapping - such as e.g. implemented by the Copernicus Emergency Mapping Services - just work with single images and image-based classification techniques. This approach does not allow to exploit the full information content of the Copernicus data as is contained in the full collocated Sentinel-1 and Sentinel-2 time series.

Contents and goals

To overcome this limitation, the ACube4Floods project will strive to extract the maximum information content from Sentinel-1 and Sentinel-2 time series by applying change detection- and machine learning algorithms to analysis-ready Sentinel data cubes.

Methods

This will be realised on a cloud platform, thereby leveraging and further improving the capabilities of the Austrian Data Cube (ACube) which is envisioned to become an open government data service.

Scientifically, ACube4Floods will bring advances in the fusion of SAR, optical and ancillary data (terrain model, land cover, etc.) through machine learning techniques, and in our capability to predict the limitations and uncertainties of the satellite-derived information. The latter point is deemed essential for the acceptance of ACube4Floods by the users, as users can deal with known unknowns but not with unknown unknowns.

Expected results

By capturing also smaller-scale flood events, ACube4Floods will help the project's pilot users, most importantly the Austrian Ministry for Sustainability and Tourism and the Federal Ministry of Defense, to enhance disaster resilience, through better preparedness, response, inventory and recovery. This involves e.g. the documentation and analysis of flood events, by which a comparison of the actual observed flood events with the simulated scenarios for certain occurrence probabilities from the hazard zone mapping can take place.

Furthermore, ACube4Floods will combine Copernicus data with dynamic weather and mobile movement data in order to support near-real-time risk assessment.

Project Coordinator

Vienna University of Technology

Project partner

• Central Institute for Meteorology and Geodynamics (ZAMG)
• Federal Ministry of Defence
• Federal Ministry for Sustainability and Tourism (BMNT)
• University of Natural Resources and Applied Life Sciences Vienna
• EODC Earth Observation Data Centre for Water Resources Monitoring GmbH
• Luxembourg Institute of Science and Technology, En Abrege List
• Invenium Data Insights GmbH

Vienna University of Technology
Karlsplatz 13
A-1040 Vienna
Tel.: +43 (1) 58801 0
Fax: +43 (1) 58801 41275
Web: www.tuwien.at