| Summary: | This research demonstrates the use of airborne synthetic aperture radar (SAR) and Geographic
Information Systems (GIS) for mapping and predicting large rock landslide occurrences in southwestern
British Columbia.
Lineaments in the Chilliwack area are mapped using geocoded 10 m resolution 1:20,000,
1:50,000, and 1:80,000 scale C-band SAR and compared to lineaments mapped with 1:50,000 scale
black and white stereo airphotos. SAR lineament length and trend correlate well with airphoto
lineaments and provide additional structural information about the Chilliwack area. Testing the ability of
SAR to locate specific airphoto lineaments reveals that SAR tends to find the longer airphoto lineaments.
SAR is useful for detecting geomorphological landslide features such as lineaments, antislope
scarps, deformation zones, and especially hummocky debris lobes, cone and fans. In areas covered by
cultivation, logging activity, or mature forest, SAR proves most useful for revealing landslide debris and
lineaments especially where tree canopy height corresponds subtle changes in ground morphology.
Depending on the viewing geometry, SAR is able to detect landslide source and deposition zones not
apparent in stereo airphotos.
Spatial relationships between regional faults, lineaments, plutonic contacts, bedrock geology,
slope angle, and landslide hazard occurrence are explored using the IDRISI GIS and an objective
multivariate statistical methodology called weights of evidence. Weights of evidence modelling
determines landslide potential for regions where representative landslide occurrences are known,
estimates uncertainty, ranks predictive power of input maps, and accounts for missing or incomplete
data. The method is particularly well-suited for modelling multiclass maps and proximity to linear
features.
The top five predictor maps are: 1) Custer Gneiss; 2) within 1300 m of a fault trace; 3) within
1200 m of a lineament; 4) Cultus Formation; 5) Chilliwack Formation, and; 5) Spuzzum Pluton. The
proximity to a plutonic contact and slopes less than 20° are not positive predictors of landslide
occurrence. Predictor map weights are combined and map of posterior probability for predicting landslide
occurrence is created. Areas of relative high landslide occurrence include the Fraser Canyon near Yale,
the slopes above Chilliwack and the upper Chilliwack Valley, and adjacent to the Ross Lake Fault south
of Hope. Adding this hazard zonation map with socio-economic criteria and landslide runout models
within a GIS builds risk maps essential for development planning in mountainous terrain. === Science, Faculty of === Earth, Ocean and Atmospheric Sciences, Department of === Graduate
|
|---|
