Landslide Risk Assessment

Data were collected from the following sources: 

1. UBC Department of Geography

• Popultation Data
• British Columbia DEM
• Vancouver Island shapefile
• Vancouver Island roads
• British Columbia Rivers

2. Hectares BC

• Vegetation
• Soil

3. ClimateBC

• Precipitation

Generating climate data                   

The DEM was resampled so that the cell size was increased from 77m² to 770m². Converting from raster to point in the same manner as previously now outputted 54 000 points, a much more reasonable amount of data to work with. X and Y coordinates were calculated from the point-DEM using the “calculate geometry” tool, and the attribute table was organized in the format required for input into ClimateBC (ID1 – point number, ID2 – second location option, Latitude – y coordinate, Longitude – x coordinate, Elevation). The attribute table was then exported as a .dbf file, converted to a .csv file using Microsoft Access, and inputted in ClimateBC. ClimateBC calculated climate data (we chose the decade of 1991-2000) for each of the given points and outputted a .csv file which was edited to isolate Mean Annual Precipitation (other climate variables were removed) then imported back into ArcMap.

Figure 1: The Climate BC program; multi-location input was used and the period selected was 1991-2000.

       Within the attribute table of the point-DEM layer, a new field was added: “Precipitation”. The climate data were joined with the point-DEM via the Point ID columns; then Mean Annual Precipitation data was added in the Precipitation column using the field calculator (Precipitation = MAP [mean annual precipitation]). The join was then removed, and the point-DEM layer was converted to a raster using the “Feature to Raster” tool.

Multi-Criteria Analysis

Each of the layers used in the multi-criteria evaluation (MCE) was normalized to 1-0 using the fuzzy membership operation. Vegetation, soil, roads and rivers are all categorized data, and the category values needed to be re-classified based on the different areas’ landslide risk level. The new classification values were created by adding a new field to the attribute table, and using the field calculator to add new values individually (see Tables 1, 2, and 3). The new values attempted to represent the level of risk associated with the category (based on research by Hungr et al, 2005, Dai et al, 2002, and Ozdemir, A., 2009). Any areas that were listed as having no data were reclassified to having a value of zero.

Table 1 – Vegetation Re-classification

Areas with strong root systems, or systems that are only found on flat areas, were given lower values, while unvegetated surfaces were given the most likely value.

Figure 2: Vegetation type map reclassified based on landslide risk.

 Table 2 – Soil Re-classification

Hard rock was the least likely soil type to experience landslides (although rockslides are possible, they were not considered in greater depth in this assessment), organic soil is able to absorb more water in high precipitation events, so it was less important than mineral soils, the most likely to result in debris flows (Hungr et al, 2005).

Figure 3: Soil type map reclassified based on landslide risk.

Table 3 – Road Re-classification

Well constructed roads (expressways and highways) usually take preventative measures and make slope stabilizing efforts, thus they were given a mid-range value; less well constructed roads are more likely to destabilize slopes and thus were given the highest value. 

          Finally, with re-classified, normalized data in raster format, the MCE was set up. Using the weighted sum tool, the factors were weighted according to their influence on landslides: slope (35%), precipitation (25%), distance to roads (10%), soil type (12.5%), vegetation type (12.5%), and distance to rivers (5%). The resulting hazard map was divided into high, medium, and low-risk areas for landslides. A second MCE was run with the variables weighted equally. Finally, a map overlaying population density by dissemination area with landslide risk was created.