The mobility of long-runout landslides

Abstract

Several issues relevant to the mobility of long-runout landslides are examined. A central idea developed in this paper is that the apparent coefficient of friction (ratio of the fall height to the runout distance) commonly used to describe landslide mobility is physically meaningless. It is proposed that the runout distance depends primarily on the volume and not on the fall height, which just adds scatter to the correlation. The negative correlation observed between the apparent coefficient of friction and the volume is just due to the fact that, on the gentle slopes on which landslides travel and come to rest, a large increase in runout distance due to a large volume corresponds to a small increase in the total fall height, hence to a decrease in the apparent coefficient of friction. It is shown that the spreading of a fluid-absent, granular flow is not able to explain the large runout distances of landslides, and in particular does not allow the centre of mass to travel further than expected for a sliding block. This contrasts with the behaviour of natural landslides, for which the centre of mass is shown to travel much further than expected from a simple Coulomb model. The presence of an interstitial fluid which can partly or entirely support the load of particles allows the effective coefficient of solid friction to be reduced or even suppressed. Air is not efficient for fluidising large landslides and a loose debris cannot slide over a basal layer of entrapped and compressed air, as air would rapidly pass through the debris in the form of bubbles during batch sedimentation. Water is much more efficient as a fluidising medium due to its higher density and viscosity, and its incompressibility. As water is known to enhance the mobility of the saturated debris flows, it is proposed that water is also responsible for the long runout of landslides. This is consistent with the fact that the increase in runout with volume is similar for debris flows and landslides. Field evidence suggests that most landslides are unsaturated with water but not dry, even on Mars. Comparison of the velocity of well-documented landslides with that predicted by fluid-absent, granular models shows that these models predict landslides that are much faster and less responsive to topography than natural ones. The relatively low velocities of landslides suggest that energy dissipation is dominated by a velocity-dependent stress and that the coefficient of solid friction is very low. This is consistent with the physics of fluidised or partly fluidised debris and suggests that landslide velocity may be controlled by local slope and flow thickness rather than by the initial fall height. In the absence of a supply of fluid at the base, fluidisation requires a net downward flux of sediment, implying some deposition at the base of landslides, which may thus progressively run out of material. In such a model, the spreading of the portion of a landslide beyond a certain distance would primarily depend on the volume passing this distance and not on the total volume of the landslide. Landslide deposits may therefore have self-similar shapes, in which the area covered beyond a certain distance is a constant function of the volume beyond that distance. It is shown that the shape of some well-documented landslide deposits is in reasonable agreement with this prediction. One consequence is that, as recently proposed for debris flows, assessment of hazards related to landslides should be based on the correlation between the volume and the area covered by the deposit, rather than on the apparent coefficient of friction.