Do Apparent "Precursory" Rockfalls Coincide with Large Rockfalls by Chance?

ABSTRACT: Rockfalls present a hazard to infrastructure, particularly for slopes along major roadways. Rock slope monitoring studies have linked large failures with the presence of previous, smaller spatially co-located precursory rockfalls. Most of these studies focus on large-scale failures of thousands of cubic meters in volume, and few studies focus on precursory rockfalls associated with small-scale rockfalls of less than 100 m3. If precursory rockfalls can be identified prior to small-scale block failure, there is potential to forecast and prevent these failures. However, it has not been definitively established that these apparent precursory rockfalls are not simply a product of coincidental co-location on a slope with high background rockfall activity. This study seeks to evaluate this possibility through comparison of a simple simulation model with observations from field data. First, terrestrial lidar and photogrammetry data were collected over time across two slopes to create point clouds, and the rockfalls present at each slope were identified. Across the slopes, 77 rockfalls were identified above a volume threshold of 0.1 m3, and for each of these rockfalls, it was determined whether there were any associated precursory rockfalls. A random rockfall simulation was then created to represent the "null" condition, where all rockfalls were assigned uniform random locations, and it was determined how many rockfalls above the volume threshold would have had spatially co-located precursory rockfalls for many different realizations. Results confirm that the number of precursory rockfalls is significantly higher relative to the null simulation, consistent with the idea that precursory rockfall occurrence is driven by mechanical interaction between adjacent rockfall blocks rather than being random in nature.