![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
A Tool for Modelling the Probability of Landslides Impacting Road Networks |
VerfasserIn |
Faith E. Taylor, Michele Santangelo, Ivan Marchesini, Bruce D. Malamud, Fausto Guzzetti |
Konferenz |
EGU General Assembly 2014
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 16 (2014) |
Datensatznummer |
250092328
|
Publikation (Nr.) |
EGU/EGU2014-6663.pdf |
|
|
|
Zusammenfassung |
Triggers such as earthquakes or heavy rainfall can result in hundreds to thousands of
landslides occurring across a region within a short space of time. These landslides can in turn
result in blockages across the road network, impacting how people move about a region.
Here, we show the development and application of a semi-stochastic model to simulate
how landslides intersect with road networks during a triggered landslide event.
This was performed by creating “synthetic” triggered landslide inventory maps
and overlaying these with a road network map to identify where road blockages
occur.
Our landslide-road model has been applied to two regions: (i) the Collazzone basin (79
km2) in Central Italy where 422 landslides were triggered by rapid snowmelt in January
1997, (ii) the Oat Mountain quadrangle (155 km2) in California, USA, where 1,350
landslides were triggered by the Northridge Earthquake (M = 6.7) in January 1994. For both
regions, detailed landslide inventory maps for the triggered events were available, in addition
to maps of landslide susceptibility and road networks of primary, secondary and tertiary
roads.
To create “synthetic” landslide inventory maps, landslide areas (AL) were randomly
selected from a three-parameter inverse gamma probability density function, consisting of a
power law decay of about -2.4 for medium and large values of AL and an exponential
rollover for small values of AL. The number of landslide areas selected was based on the
observed density of landslides (number of landslides km-2) in the triggered event
inventories. Landslide shapes were approximated as ellipses, where the ratio of the major and
minor axes varies with AL. Landslides were then dropped over the region semi-stochastically,
conditioned by a landslide susceptibility map, resulting in a synthetic landslide inventory
map. The originally available landslide susceptibility maps did not take into account
susceptibility changes in the immediate vicinity of roads, therefore our landslide
susceptibility map was adjusted to further reduce the susceptibility near each road based on
the road level (primary, secondary, tertiary). For each model run, we superimposed the
spatial location of landslide drops with the road network, and recorded the number,
size and location of road blockages recorded, along with landslides within 50 and
100 m of the different road levels. Network analysis tools available in GRASS
GIS were also applied to measure the impact upon the road network in terms of
connectivity. The model was performed 100 times in a Monte-Carlo simulation for each
region.
Initial results show reasonable agreement between model output and the observed
landslide inventories in terms of the number of road blockages. In Collazzone (length of road
network = 153 km, landslide density = 5.2 landslides km-2), the median number of modelled
road blockages over 100 model runs was 5 (±2.5 standard deviation) compared to the
mapped inventory observed number of 5 road blockages. In Northridge (length of road
network = 780 km, landslide density = 8.7 landslides km-2), the median number of modelled
road blockages over 100 model runs was 108 (±17.2 standard deviation) compared to the
mapped inventory observed number of 48 road blockages. As we progress with model
development, we believe this semi-stochastic modelling approach will potentially aid
civil protection agencies to explore different scenarios of road network potential
damage as the result of different magnitude landslide triggering event scenarios. |
|
|
|
|
|