EP51D-01
Physical Limits on the Predictability of Erosion and Sediment Transport by Landslides and Debris Flows
Friday, 18 December 2015: 08:00
2005 (Moscone West)
Richard M Iverson, USGS, Vancouver, WA, United States
Abstract:
Episodic landslides and debris flows play a key role in sculpting many steep landscapes, and they also pose significant natural hazards. Field evidence, laboratory experiments, and theoretical analyses show that variations in the quantity, speed, and distance of sediment transport by landslides and debris flows can depend strongly on nuanced differences in initial conditions. Moreover, initial conditions themselves can be strongly dependent on the geological legacy of prior events. The scope of these dependencies is revealed by the results of landslide dynamics experiments [Iverson et al., Science, 2000], debris-flow erosion experiments [Iverson et al., Nature Geosci., 2011], and numerical simulations of the highly destructive 2014 Oso, Washington, landslide [Iverson et al., Earth Planet. Sci. Let., 2015]. In each of these cases, feedbacks between basal sediment deformation and pore-pressure generation cause the speed and distance of sediment transport to be very sensitive to subtle differences in the ambient sediment porosity and water content. On the other hand, the onset of most landslides and debris flows depends largely on pore-water pressure distributions and only indirectly on sediment porosity and water content. Thus, even if perfect predictions of the locations and timing of landslides and debris flows were available, the dynamics of the events – and their consequent hazards and sediment transport – would be difficult to predict. This difficulty is a manifestation of the nonlinear physics involved, rather than of poor understanding of those physics. Consequently, physically based models for assessing the hazards and sediment transport due to landslides and debris flows must take into account both evolving nonlinear dynamics and inherent uncertainties about initial conditions. By contrast, landscape evolution models that use prescribed algebraic formulas to represent sediment transport by landslides and debris flows lack a sound physical basis.