ALaDyn

Millions of people worldwide live in areas with high landslide risk, annually causing numerous fatalities and significant economic losses. Rainfall-induced landslides cause the majority of those losses. Particularly Europe is heavily affected, and within Europe, Austria is one of the most vulnerable countries. In Austria slopes are often prone to gravitational mass movements due to the geomorphological and geological conditions. Landslide susceptibility modelling, typically providing landslide hazard maps as end-products, usually focuses on static factors such as geology and morphology. Dynamic influences like extreme weather, deforestation, or land-use changes are often overlooked. However, they are becoming increasingly important due to climate change.

This study addresses this gap by developing a digital twin model for landslide susceptibility that considers both, static conditions like geomorphology and geology as well as dynamic triggers such as rainfall, land-use changes, and land cover changes. The model area is located in southern Styria, where well-documented landslide-inventories have been established in the last two decades. Thousands of landslides were documented during rain-fall triggered landslide events in 2009 and 2023. The August 2023 event, triggered by the low-pressure system “Zacharias,” highlights the significance of more frequent and intense rainfall events. In this single event over 3,000 landslides were documented, and damages in southeastern Styria amounted to more than 30 million Euros.

A major challenge for accurate modelling is the lack of detailed time series data on extreme weather events and their associated landslides. However, extensive data is available in the project region, which can be used for both calibration and validation of the models. Additionally, weather data from ECMWF, developed under the EU/ESA initiative DestinE ("Destination Earth"), as well as weather data from the Wegener Network and INCA (Geosphere Austria), are utilized. Furthermore, land-use data from Copernicus is incorporated into the models, along with data from Austrian databases such as GTIF-AT. Remote sensing technologies like LiDAR are also employed to better capture the protective function of forests.

The goal is to develop an automated prototype that generates daily hazard maps based on various scenarios, such as sudden extreme weather events, land-use changes, or preventive measures. Calibration and validation will be based on the heavy rainfall event, associated with rain-fall triggered landslide events in 2009 and 2023.