Identifying and Monitoring Rockfall Precursors Using Terrestrial Laser Scanning for Improved Rockfall Hazard Management

Abstract

Rockfalls threaten communities and infrastructure in mountainous regions worldwide and have been a particularly problematic hazard along the transportation corridors in western Canada. These types of failures are traditionally managed using rockfall hazard management frameworks that are based on a historical analysis of rockfall activity. These frameworks do not fully consider the complexity of the natural rock slopes, nor indicate where and when a potential failure may occur. Terrestrial Laser Scanning (TLS) has been beneficial for monitoring and assessing rock slope hazards and has the potential to identify incipient signs of rock slope failure. The aim of this thesis is to use TLS to better manage rockfall hazard along transportation corridors by including quantitative measures of precursor activity into hazard analysis strategies and through the development of near-real time automated TLS systems for early warning purposes. In the first part of the thesis, monitoring at 2-3-month intervals was conducted with TLS at a variety of rock slopes located along the CN Rail line in the Thompson River valley over the four-year duration of this research. Over one hundred rockfalls in the range of 0.1 to 4200 m3, exhibiting precursor activity in the form of pre-failure deformation, smaller precursor rockfalls and tension crack openings were studied and compiled into a database. The approach included: (i) identifying potential rockfall source zones based on incipient signs of failure; (ii) tracking kinematics in three dimensions to better understand the mechanisms of failure; (iii) estimating potential failure volumes based on bounding joint structure; and (iv) transmitting this information to the railway operator for an assessment of risk. The second part of the thesis focused on monitoring at near continuous levels, which is required to identify the accelerating phase prior to rockfall failure which is necessary to forecast failure time in early warning systems. A complete series of analysis tools was developed to process and analyze TLS data autonomously in near real-time. This included tools to filter outliers, register point clouds, conduct change detection using spatial and temporal neighbourhood averaging and to display time series of deformation. The system was tested for a 6-week period at a rockslide in the French alps. The contributions in this thesis (i) enhance current rockfall hazard analysis methods by including quantitative measurements of precursor activity, (ii) improve our understanding of the pre-failure stage of rock slope failures and (iii) opened the door for future studies of rock slope failure at a high temporal density using TLS.