A Quantitative Rockfall Risk Analysis System for Highway Rock Slopes

This paper presents a quantitative methodology for rockfall risk analysis focusing on highway crystalline rock slopes. The proposed method resulted in a risk classification given by a matrix, which relates the rockfall likelihood, with its consequence. The likelihood considers the characteristics of the rock masses and the consequence takes into account the chance of the block reaching the highway and causing accidents. The rockfall likelihood and the consequences are obtained by equations, whose values allow classifying the slopes as high, medium (transition zone), or low rockfall risk. These equations and the matrix of the risk analysis system were generated using linear discriminant analysis and confidence ellipses, both multivariate statistical techniques. Principal component analysis was applied to the dataset before the linear discriminant analysis in order to change qualitative variables into quantitative ones. To establish the classes of failure likelihood and its consequences, cluster analysis was applied. Afterwards, discriminant analysis was applied using the scores of the principal components as independent variables and the resulting clusters as dependent variables. The dataset used in this research has 220 rock slopes and the main variables considered are the parameters traditionally used in hazard and risk classifications for highway slopes.