In the context of anticipated climate change, meteorologists expect an increase in extreme weather events in Austria, which may serve as precursors of gravitational mass movements such as landslides. As such, landslides represent a safety risk for people and infrastructure, and often cause great damage. Consequently, their detection is crucial in order to be able to act promptly and to avert potential damage at an early stage.
A high-quality and complete data inventory is an essential prerequisite for a better understanding of landslides and for the creation of maps, risk analyses or the development of an early warning system. So far, this consists of data from historical archives, results from field mapping, data derived from remote sensing, as well as combined inventories. We propose to supplement these inventories by exploiting information from new digital elevation models and earth observation data from the Sentinel missions. These are large amounts of valuable data, which have only been used to a very limited extent up to now. This is mainly attributable to the lack of methods to analyze these data in a timely manner, which could be used to create an improved data inventory as a basis for determining the occurrence probability of mass movements.
The project focus is thus on creating high-quality data inventories for landslides. The gAIa approach is to enhance existing, federal data sets of gravitative mass movements (in particular landslides) with the help of modern artificial intelligence (AI) methods. For this purpose, the newly generated information from the airborne-laser-scanning (ALS) elevation models, and multi-spectral, optical satellite data (Sentinel-2) will thus be fused and harmonized with the current landslide land-registry data.
More specifically, Machine learning (ML) techniques will be employed to support the data-fusion. The resulting geo-data inventory will consider data standards which are currently under development. Additionally, aspects of geo-data management and long-term archiving will be examined.
Furthermore, deep learning (DL) architectures will be employed to develop a prediction model for the probability of landslide occurrences. This way, circumstances, which may lead to such mass movements, can be detected automatically.
The main goal of gAia is to generate a hazard-warning map for landslides which will serve as a basis for policy makers and civil protection authorities in order to facilitate their decision making. Therefore, they will be equipped with the generated information which will serve to recommended actions in governmental crisis management in order to avert potential catastrophes.
ProjektleiterIn / Name und Institut/Unternehmen:
SBA Research gGmbH
Mag. DI Rudolf Mayer
Floragasse 7/5, 1040 Wien
+43 (1) 505 36 88
Auflistung der weiteren Projekt- bzw. KooperationspartnerInnen
AIT Austrian Institute of Technology GmbH
Dr. Jasmin Lampert
Center for Digital Safety & Security, Data Science
Giefinggasse 4, 1210 Wien
Bundesministerium für Landesverteidigung
Mjr OR Dipl.-Ing. Christian Meurers
Roßauer Lände 1, 1090 Wien
Disaster Competence Network Austria - Kompetenznetzwerk für
Dipl.Ing. Susanna Wernhart
Peter-Jordan-Straße 82, 1190 Wien
Dr. Marc Ostermann
Programmkoordinator für Geomonitoring and Katastrophenschutz
Neulinggasse 38, 1030 Wien
GeoVille Informationssysteme und Datenverarbeitung GmbH
MSc. Michaela Seewald
Sparkassenplatz 2, 6020 Innsbruck
Zentralanstalt für Meteorologie und Geodynamik (ZAMG) – Teilrechtsfähige
Einrichtung des Bundes
Dr. Michael Avian
Hohe Warte 38, 1190 Wien
Homepage im WWW: