This is a continuation of the discussion of landslide susceptibility in Saint Lucia, and extends into a preliminary analysis of landslide risk to human habitation, in a qualitative sense.
We have corrected the locations of the inventoried landslides as best we could, given the accuracy of the source data, which as I mentioned before is derived from a landslide inventory map included in the Saint Lucia Landslide Response Plan dated 31 January 2008, and is easily located on-line.
The landslide susceptibility map has been updated by re-analysing landslide incidence in comparison with a variety of spatial data, including:
DEM interpretations (using 2009 topography provided by Government of Saint Lucia):
– aspect (orientation of the slope, i.e. north, south, SW, etc)
– plan curvature
– profile curvature
– slope angle
Soils data (using shapefile provided by Government of Saint Lucia):
– soil erosion class (interpretation methodology unknown)
– gradient class (i.e. gentle, steep, very steep etc)
The analysis uses a modified “weights of evidence” approach. I obtained weight factors for each of these spatial themes, and tried several different combinations to obtain the best possible susceptibility map. In the end, the best model uses elevation, aspect, plan curvature and slope angle. The model does not use the soils data, nor overall curvature or profile curvature.
The inferred landslide susceptibility map looks like this:
The way I have broken out the landslide susceptibility is by relative expected frequency. Given a major landslide triggering event (i.e. significant tropical storm with unusually heavy rainfall, major earthquake, etc), we would expect widespread landslide occurrence. Part of the island (green area) would have significantly lower landslide frequency (~ 25 % average spatial frequency), and selected area would have significantly higher frequency (orange and red with nearly twice and about thrice average frequency).
We might focus our attention to the areas of the island with elevated (i.e. above “average”) landslid susceptibility, and this map provides a somewhat modified interpretation:
Now if we are interested to know where landslides might affect residences, we can look at our inferred population density map:
We can then overlay the elevated landslide susceptibility layer with population density to obtain an initial sense of qualitative landslide risk to human habitation:
In this last map, the red areas are places on the island where there is both a relatively high population density AND an elevated landslide susceptibility. Hence, in the case of widespread landslide occurrence, these areas would be at greater risk to property and human health.
It MUST be noted that the preceding analysis is limited to the potential interaction between local landslides and stationary property or people. There also exists the separate (and SIGNIFICANT) hazard and risk associated with channelized landslide phenomena such as debris flows/floods, hyperconcentrated flows, and other events where landslide debris flows along an existing channel, with impacts at some distance downstream. Some of the most significant landslide events in Saint Lucia have involved such events. For example, during Hurricane Tomas in 2010, approximately 10 people died when a debris flow impacted residences at Fond St. Jacques (see following photo, courtesy DFL Consult, Castries, St Lucia, photographer unknown):
Further, some form of channelized landslide at Ravine Poisson in 1938 caused > 60 fatalities. This map shows the current layout.
The bridge at Ravine Poisson is understood to be regularly impacted by debris during major storm events. This type of hazard requires separate consideration beyond the landslide susceptibility and risk mapping presented above in this post.