Laboratory simulations of rainfall-induced landslide reactivation mechanisms following the 2016 Kaikōura Earthquake

2021

Abstract

Increases in rainfall-induced landsliding following large earthquake are well documented but the time frames over which this heightened hazard persists in the land scape remains poorly understood. Whilst it is well known that the presence of failed and partially slopes after earthquakes significantly reduces the rainfall thresholds required to activate slope movement, their failure susceptibility during specific storms and how this changes through time remains poorly studied. To improve knowledge in this field requires well documented slope failures following earthquakes and a detailed understanding of their potential failure mechanisms when pore pressures are elevated in the slope. The 2016 Mw 7.8 Kaikōura earthquake provides a unique opportunity to study how rainfall events following the earthquake may impact the timing and mechanisms of landslide reactivation. This study conducted a suite of specialist triaxial cell experiments, designed to replicate varying rainfall scenarios on remoulded samples collected from two sites where numerous earthquake-induced landslides were recorded in similar Late Cretaceous to Neogene sediments with similar physical properties (the Leader Dam Landslides (LDL) and the Limestone Hill landslide (LHL)). In each experiment rainfall events were simulated using a series of different pore pressure scenarios (increases and decreases in mean effective stress) at representative field stress conditions whilst monitoring material deformation behaviour. The results demonstrate that both the deformation behaviour and pore pressure required to generate failure were influenced by the previous changes in pore pressure. Samples subjected to stepped increases in pore pressure were subject to greater pre-failure deformation (dilation) and subsequently failed at lower pore pressures (higher mean effective stress) when compared to samples subjected to linear increases in pore pressure. In addition, increases in the rate of pore pressure also increased the amount of pre-failure deformation allowing failure to occur when pore pressures were lower. In contrast a sample subjected to both increases and decreases in pore pressure underwent pre-failure densification and subsequently required a larger increase in pore pressure to fail. The results demonstrate that landslide reactivation is influenced by a number of factors including the amount and rate of previous changes in pore pressure and the slope drainage history. The results provide new insights into why landslide susceptibility may remain elevated for prolonged periods of time (e.g. decades) in the landscape as well as why the rainfall thresholds for site specific failures during storms may be difficult to predict.

Abstract

Related Papers