This should be the final post in a series of three to develop some ideas about regional scale risk associated with large landslides in sensitive clay in eastern Canada. The first two parts are here: landslide susceptibility and risk in eastern Canada part 1 and here: landslide susceptibility and risk in eastern Canada part 2.
Now before I discuss prioritization of further study, I’d like to explore the validity of the model a little bit. The susceptibility model discussed previously should do a decent job of isolating future landslide occurrence to a relatively small proportion of the overall study area. However, a lot of the study area is either unpopulated or infrequently used/occupied, so many landslides can occur with minimal chance of impact to people, buildings or other infrastructure. The risk map is intended to isolate areas that are more or less likely to have future DAMAGE or other impact, such as LOSS OF LIFE.
One way to test the validity of the model is to look at past damaging landslides to see whether they occurred within higher risk areas of the map. I have good location data for a small number of past damaging landslides, including the following:
2010 Saint-Jude, Quebec – house destroyed, 4 fatalities
1972 Saint-Jean-Vianney, Quebec – subdivision destroyed, 31 fatalities
1924 Kenogami, Quebec – mill facilities damaged, complete mill shut down
1955 Nicolet, Quebec – school destroyed, church property damaged, 4 fatalities
1918 Saint-Thuribe, Quebec – 1 fatality
1908 Notre-dame-de-la-Salette, Quebec – numerous homes destroyed, 33 fatalities
I find these results to be very encouraging. All of the damaging landslides that I have good location data for fall within the elevated risk areas delineated by the risk map, suggesting the map has some validity.
Now the primary purpose in developing this regional scale landslide risk map is to aid in the prioritization of efforts for additional, more detailed, study in selected areas. Such further work might include air photo study, site reconnaissance, or detailed, site-specific landslide risk mapping, such as that being completed for selected communities in Quebec by the MTQ (Quebec transport ministry).
As a first exploration, we can try to prioritize on the basis of political boundaries. This map shows county boundaries superimposed over the risk model:
There are 881 full or partial counties in this map, thus yielding what might be an unwieldy number of geographic divisions, unless they can be ranked with a fairly small number of higher priority counties. I’ve looked at two different ways of trying to prioritize the counties by risk: first by determining the total relative risk within the county, and then by “risk density,” where this risk density is the total risk divided by the area of the county. Here are those two representations:
The first map, showing total risk by county, has some unexpected results. In particular, most of eastern Ontario has relatively high values. This is inconsistent with the fact that most damaging landslides have happened in Quebec, and much of eastern Ontario has limited susceptibility, particularly toward the southwest end of the area. This result is due to the fact that county size in eastern Ontario tends to be very large, particularly in comparison with central Quebec.
The second map, which shows risk density by county, gives a result that is a little closer to intuition, with elevated values in parts of the study area where landslides are more common. In this case (and similarly to the previous map), there is an unintuitively high risk in the urban areas. While much of the area of the major urban centres (e.g. Ottawa, Montreal, Quebec) have relatively low (or nil) susceptibility to large landslides, the other side of the risk equation drives the risk higher in these areas. Since risk is combination of possibility of hazard and possibility of presence of receptors, the risk model is right, but it defies intuition, which for most people would probably put an intuitive bias on the probability of the hazard in an instinctive assessment of risk, whereas population density is equally important.
Closer examination of the county-based risk maps shows that neither does a particularly good job of showing significant risk levels near the past damaging landslides. Therefore, perhaps neither of these county-based risk maps can be used on their own to aid prioritization of further work, but rather serve simply as useful additional contextual information.
I think the better approach may be to generate something similar to the county-based maps, but on some form of equal-area subdivision of the study area, not relying on political boundaries.
I guess that means I’ll need another post to wrap this up. 🙂