PhD-Project I8 by Lisa Luna (UP): Learning landslide triggers from inventory data

Timescale: July 2019 – Sept. 2021


Prof. Dr. Oliver Korup, University of Potsdam;

Prof. Dr. Jürgen Kurths, PIK Potsdam


Regional disturbances such as rainstorms and strong earthquakes trigger a large number of all landslides worldwide. These triggers can mobilize between tens and hundreds of thousands of slope failures in minutes to days, excluding reactivated slope failures. The success of predicting such regional landsliding episodes therefore hinges on how well we can detect the timing and magnitude of their triggers. While earthquake prediction (or forecasting) remains challenging, more progress has been made in the field of forecasting rainstorms. Novel methods such as event synchronisation based on spatially embedded complex networks (at the core of NatRiskChange’s Project P1), for example, have helped to objectively map out extreme rainfall trajectories from tropical cyclones.

The next step is to learn how these extreme rainfall patterns map to landslide distributions, which can also inform our understanding of how climate change will affect future landslide susceptibility.  To do so, we propose to first develop probabilistic methods to identify rainfall-triggered landslides from existing major landslide inventories (for example from Central America, USA, Canada, and Japan), and thus separate them from earthquake-triggered landslides. This approach will include the extreme rainfall trajectories from Project P1 and new insights into earthquake-triggered landslides investigated in the NatRiskChange Kumamoto Task Force. With inventories of rainfall-triggered landslides in hand, we will then investigate what drives the spatial distribution of landslides triggered during rainstorms.  We will integrate insights gained from historical inventories with modeling exercises to examine how climate change (for example, higher rainfall intensities or more frequent storms) could affect future landslide susceptibility.  In the context of a changing climate, this project seeks to improve our ability to detect and predict rainfall-triggered landslides and thus to advance practical landslide forecasting to aid robust vulnerability and risk assessments.

Objectives and Methods

  • to separate rainfall- from earthquake-triggered landslides using meteorological and topographic predictors such as extreme rainfall trajectories and seismic directivity effects (developed and derived in NatRiskChange Phase I)
  • to understand what drives the spatial distribution of rainfall triggered landslides and how, in this context, climate change might affect future landslide susceptibility
  • Methods: probabilistic classification methods using Bayesian Reasoning and Machine Learning, physical landscape modeling, numerical landscape modeling, spatially embedded complex networks