The impact of atmospheric pressure change and rainfall for triggering landslides during weather events

Landslides are a complex phenomenon which triggering depends on both intrinsic properties of soils and rocks and external influences such as the action of weather conditions, or earthquakes. Around 6,000 landslides failed the 6^th of September 2018 during the Mw 6.6 Hokkaido Eastern Iburi earthquake (Japan), one day after the typhoon Jebi hit the region. If the ground acceleration induced by the seismic waves likely played a major role in the triggering of these landslides, it is unclear how it compares to the respective role of rainfall and atmospheric pressure drop induced by the typhoon. The aim of this work is therefore to investigate the influence of weather conditions on landslide triggering, and more specifically to characterize the relative contributions of rainfall and atmospheric pressure changes on slope stability.

For this purpose, a simple model is developed to describe the two mechanisms and to compare their respective impact on slope stability. The model considers a homogeneous isotropic tilted infinite half-space in one dimension. Slope stability is estimated using a safety factor and a Mohr-Coulomb criterion. In the static case, groundwater is accounted for by adding an unconfined aquifer into the model. Analytical models based on diffusion equations have been used to describe the impact of rainfall and atmospheric pressure changes on slope stability (Iverson, 2000; Schulz, 2009). Extracting a response function from these models allows us to compute the stability change due to any rainfall or pressure time series. The model parameters are taken for a typical slope in Taiwan tilted with a 25° angle and with characteristics of a fully saturated loamy soil at 4 m depth and put under conditions similar to the Morakot typhoon, with more than 240 mm of rain on a 24 h period and an associate atmospheric pressure drop of 4 kPa.

Atmospheric pressure change and rainfall impacts the media in a very different way despite being associated to the same physical phenomenon, pressure diffusion. The atmospheric effect is instantaneous and directly affects the effective stress with a maximum of 4 kPa. This effect decreases over time while the pore pressure is adjusted to the atmosphere. The rainfall effect is delayed in time but has a greater impact on the effective stress, reaching 11.7 kPa almost 2 days after the end of the rainfall event. While atmospheric pressure does not change significantly the safety factor, it can exacerbate the effect of rainfall and advance the failure in time because of the respective temporal lag between the 2 processes.  Therefore, this study may lead to a better understanding of the effect of weather events such as typhoons on landslide triggering and slope stability. Our results call for revisiting in a more systematic approach the role of atmospheric pressure change on landslide triggering during extreme weather events. Because a 1D model may hide some effects associated to hillslope geometry, we then consider 2D numerical models which allow us to offer some first insights on slope stability during weather events, accounting for topography.