Probabilistic approach for quantifying weather conditions for landslide occurrences at a 60 m-high weathered rock slope in western Canada

A 500 m long, 60 m high rock slope adjacent to Highway 837 in Alberta, Canada (C018 site) has a long history of landslides, with multiple occurrences of fallen material reaching and blocking the highway. Since 2000, 11 landslides have been reported. Previous studies have shown that a relationship exists between weather conditions and three distinct failure modes at this site: slides of frozen slabs of heavily weathered material, earthflows, and rock falls. Weather data have been analyzed for each recorded landslide, and the corresponding weather signature that led to each landslide has been investigated. Results show that antecedent precipitation, seasonal thawing, and short-term temperature fluctuations all play a significant role in landslide occurrences, and their effect can be quantified from a probabilistic approach. A clear weather signature has been identified for landslides occurring in the winter months. These landslides have occurred after episodes of precipitation that were followed by a decrease in temperature into freezing values with a duration of several days to weeks and subsequently followed by thawing with temperatures fluctuating around 0 °C. Moreover, statistical analyses on landslides occurring in spring and summer showed that if there is more than 20 mm of rainfall in a 14 day period, there is a 6% probability of a landslide, with a 0.1% probability of a landslide if there is less than 20 mm of rainfall in the preceding 14 days. This probabilistic approach provides a means to identify periods when landslide hazard is 60 times higher than the other time periods in spring and summer. Furthermore, the paper illustrates how this knowledge can be used to enhance our understanding of the potential effects of climate change on landslide risk and quantify the increase in landslide hazard based on climatic predictions.