Landslide-induced river course changes and lake formation (RiCoLa)

Funding Agency: Austrian Academy of Sciences (ÖAW, 2017-2020)

Principle Investigators: Dr. Günther Prasicek, Dr. Daniel Hölbling

Project Team: Ann-Lore Argentin (PhD Student)

Landslides are among the most serious threats to human lives and infrastructure in mountain ranges worldwide. Beyond the direct hazard posed by the moving mass, landslides can initiate natural hazard cascades by damming rivers and initiating catastrophic flash floods and debris flows. Such long-range effects render even unwitnessed landslides occurring in remote areas significant. Critically, insufficient information exists on landslide occurrence and recurrence intervals, and thus the hazard potential of landslide hazard cascades, as well as possible prediction and prevention measures. This lack of information is mostly due to the remoteness of many mountain regions as well as the complex dynamics of natural hazard cascades, even so the hazard posed by landslide dam failures is often orders of magnitudes greater than that of the initial landslide event. Better understanding of landslide-river interaction is hence crucial to assess and predict the resulting natural hazard.

Here we propose utilizing an interdisciplinary approach to build a solid statistical base for evaluating and understanding landslide-induced river course changes and lake formation, as well as flood risk resulting from landslide dam breaches. The project is a) tailored for long-term (~ the past 30 years) monitoring to better understand the role of extreme events in the interaction of the hillslope and channel systems, b) directly promotes the scientific advance of two early career researchers, and c) is well integrated into the international research agenda of ICGP and Future Earth.

Major outcomes of the project will be:

- A semi-automated and transferable technique for detection of landslide-induced river course changes and lake formation.

- An inventory of automatically detected landslide-induced river course changes and lakes.

- Statistical analyses evaluating the occurrence of landslide-induced river course changes and lakes dependent on (a) triggers such as earthquakes and large storm events, (b) causes such as structural weaknesses of valley flanks, and (c) consequential natural hazards.

 

For this, we will combine expert knowledge in high-mountain geomorphology and advanced remote sensing techniques to (a) extract river courses and lakes on optical satellite imagery using object-based image analysis and (b) detect and quantify river course changes by performing object-based change detection and time-series analysis over the past 30 years.

Our method will:

- Combine information on the drainage system derived from LANDSAT, SENTINEL-2 and ASTER (available at no charge since April 1, 2016) satellite imagery, as wells as SRTM and ASTER GDEM terrain data with (nearly) global coverage and available at no cost.

- Be characterized by a high level of automation.

- Avoid costly photogrammetric construction of digital elevation models from image data for each time slice.

- Be easily transferrable to other regions, different spectral domains or spatial scales (e.g. VHR satellite imagery, aerial photographs).

- Be transferable to other fields of application such as volcanic eruptions, flooding, etc.

RiColA