To run operational landslide early warning systems (LEWS) the definition of the precursory conditions that caused landslides in the past is essential. Usually, only precipitation (rain) is considered as the main precursor and is used to derive rainfall thresholds based on two characteristics: rainfall event duration (D) and (mean) intensity (I) or cumulated depth. However other precursors can be relevant for landslide triggering like snowmelt and degree of soil saturation.
Landslide thresholds are the most used tools to predict the occurrence of landslides over large areas. Thresholds are obtained by analyzing landslide occurrence and precursory factors that have triggered them in the past. Empirical (statistical or historical) and physically based methods can be used to define landslide thresholds.
The landslide early warning system that is operational in Norway uses only a landslide threshold valid for the entire country. In the next years, there will be a revision of the existing thresholds, therefore there is interest to investigate:
- A new threshold using the hydrometeorological parameters from the new Open Distributed model. We can evaluate later if the new threshold can be proposed to predict landslides that occur under a specific weather condition or that occur in a specific region.
- A new threshold that can predict landslides triggered by local convective rainfall during summer.
- Create a common platform for operational purposes that can include different available landslide thresholds and a comparison of warning performance.
Observed interpolated hydrometeorological parameters from xgeo.no and from meteorological stations, as well as landslide events from the landslide database will be used.
The expected results will be new landslide thresholds and a new warning platform that will be used by landslide forecasters in the daily hazard evaluation.
The thesis will use GIS tools, Python, R or tools for statistical analyses.