LAMPRE has developed a Landslide-Road Impact Model (LRIM) to explore different potential scenarios of regional road network disruption by different numbers of landslides. Triggers such as heavy rainfall or earthquakes may cause many landslides across a region within minutes to weeks of the event. Some of these landslides may block the transportation network, making it difficult to move about a region.
The LRIM developed by LAMPRE is applicable at local to regional scales. We have applied the model to medium to high-relief topographic regions susceptible to low-mobility landslides (i.e., not rock falls and debris flows). This LRIM can be applied to any region where triggered landslide events occur and road network, susceptibility and elevation data is available.
The LRIM may be run before, during or after a triggeredlandslide event to simulate the most likely impact upon the road network in terms of number of roads blocked by landslides or number of landslides nearby the roads, and potential resultant road network disruptions.
Civil Protection authorities use LRIM to model scenarios of network impact of different sized triggered landslide events and potential resources that might be needed.
Planning & development authorities use LRIM identify potentially vulnerable road network scenarios and plan appropriate redundancies in the road networks.
Transportation authorities & utility managers use LRIM to model potential overall road distance unavailable in the road network, and resultant disruption, as a result of landslides.
Agricultural & forest agencies use LRIM to model potentially vulnerable road network scenarios in forests.
Scientists use LRIM to simulate the potential impact of landslides, and other types of hazards, on different kinds of infrastructure networks.
LAMPRE prepares LRIM at scales ranging from 1:100,000 (smaller scale) to 1:10,000 (larger scale). LAMPRE needs a landslide susceptibility map (LSMM), a road network map, a digital elevation model (DEM) and knowledge of the landslide statistical distribution (e.g., via LStats). The model takes minutes to days to run, depending on the extent and complexity of the study area.
Malamud et al. (2004) doi: 10.1002/esp.1064
Guzzetti et al. (2003) doi: 10.5194/nhess-3-469-2003