Abstract
Understanding what controls the travelling distance of large landslides has been the topic of considerable debate. By combining observation and experimental data with depth-averaged continuum modelling of landslides and generated seismic waves, it was empirically observed that lower effective friction had to be taken into account in the models to reproduce the dynamics and runout distance of larger volume landslides. Moreover, such simulation and observation results are compatible with a friction weakening with velocity as observed in earthquake mechanics. We investigate here as to whether similar empirical reduced friction should be put into discrete element models (DEM) to reproduce observed runout of large landslides on Earth and on Mars. First we show that, in the investigated parameter range and for a given volume, the runout distance simulated by 3D DEM is not much affected by the number (i.e. size) of grains once this number attains ~ 8000. We then calibrate the model on laboratory experiments and simulate other experiments of granular flows on inclined planes, making it possible for the first time to reproduce the observed effect of initial volume and aspect ratio on runout distances. In particular, the normalised runout distance starts to depend on the volume involved only above a critical slope angle > 16–19°, as observed experimentally. Finally, based on field data (volume, topography, deposit), we simulate a series of landslides on simplified inclined topography. The empirical friction coefficient, calibrated to reproduce the observed runout for each landslide, is shown to decrease with increasing landslide volume (or velocity), going down to values as low as 0.1–0.2. No distinguishable difference is observed between the behaviour of terrestrial and Martian landslides.
Highlights
Rock and debris avalanches of large volumes can travel very long distances along almost flat topographies and represent a high risk for populations (Legros 2002)
Using the simplified shape of the released mass and of the topography, we identify the value of the friction coefficient that makes it possible to best fit the observed runout distance of each landslide
Our results show that lower grain–grain friction is necessary to reproduce the runout distance of larger landslides, in agreement with what was found previously using depth-averaged continuum models
Summary
Rock and debris avalanches of large volumes can travel very long distances along almost flat topographies and represent a high risk for populations (Legros 2002). Numerical modelling of these complex granular flows helps in understanding and predicting such events. Legros 2002; Lucas et al 2014; Delannay et al 2017) These hypotheses include bulk fluidisation or lubrication by air, gas, water, ice, heating or acoustic waves (see, e.g. references in Shreve 1987; Goren et al 2010; Ferri et al 2011; Bulmer 2012; Liu et al 2015; Mitchell et al 2015; Charrière et al.2016; Johnson et al 2016; Wang et al 2018), or the presence of an erodible bed No consensus exists on the relative impact of these processes on landslide dynamics and deposit because of the lack their quantification
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
