Rockfall Occurrence and Precipitation Modeling to Simulate Correlation Patterns

Abstract

ABSTRACT: Rockfalls are naturally occurring geohazards that are frequently associated with steep or vertical slopes. Falling rocks have the potential to disrupt transportation systems, including highway and railway passages. A common trigger known to influence the frequency and severity of rockfall events is precipitation. For many rock slopes, rainfall is considered the primary rockfall trigger. In some cases, the relationship between rainfall amount and rockfall frequency presents as a linear correlation when studied using data averaged over an extended period (i.e. due to monitoring data resolution limitations). However, when considered over smaller timescales, the actual underlying relationship is likely to be non-linear, and may involve step changes in behavior at different levels of rainfall (i.e. threshold-based behavior). The aim of this study is to evaluate the ways in which the apparent relationship between precipitation and rockfall depends on the time period over which data are being compared. We achieve this by developing a simulation that includes several components that overall model rockfall outcomes on a day-to-day basis. Daily precipitation totals are simulated according to a lognormal distribution. Lognormal distributions were also used to represent different ranges of rockfall volumes corresponding to different precipitation thresholds. The model can simulate daily rockfall volume distributions using varying probabilities of occurrence to reproduce the differences in characteristics of various hypothetical rock slopes. Overall, the simulation results are used to interpret how the appearance of correlations between variables averaged over longer time scales depends on the nature of the underlying precipitation-rockfall relationships on a day-to-day basis. 1. INTRODUCTION Rockfalls represent a significant geohazard, particularly for vertical slopes along transportation corridors. Among the various factors influencing rockfall occurrences, liquid precipitation (rainfall) is a primary trigger (Birien and Gauthier, 2023, Pratt et al., 2018). Studies with a larger monitoring interval often reveal an apparent linear correlation between rainfall and rockfall frequency due to data resolution constraints (Banji et al., 2021; Malsam, 2022; Weidner & Walton, 2021); however, examination on a smaller time scale suggests a non-linear relationship characterized by threshold-based behavior (Schneider et al., 2023). This research aims to explore how the emergent linear correlation between precipitation and rockfall events is influenced by the temporal scale of data analysis through the development of a simulation model incorporating daily precipitation values and corresponding rockfall volumes based on designated thresholds. By simulating scenarios across various hypothetical rock slopes, insights can be derived regarding the interplay between precipitation and rockfall occurrences on day-to-day basis and what would be observed over longer time scales.