Numerical Modeling of the 2014 Oso, Washington, Landslide.

Friday, 19 December 2014: 2:00 PM
David L George and Richard M Iverson, USGS, Vancouver, WA, United States
Numerical simulations of alternative scenarios that could have transpired
during the Oso, Washington, landslide of 22 March 2014 provide insight
into factors responsible for the landslide's devastating high-speed runout.
We performed these simulations using D-Claw, a numerical model we
recently developed to simulate landslide and debris-flow motion from
initiation to deposition. D-Claw solves a hyperbolic system of fi ve partial
diff erential equations that describe simultaneous evolution of the thickness,
solid volume fraction, basal pore-fluid pressure, and two components of
momentum of the moving mass. D-Claw embodies the concept of
state-dependent dilatancy, which causes the solid volume fraction m to
evolve toward a value that is equilibrated to the ambient stress state and
shear rate. As the value of m evolves, basal pore-fluid pressure coevolves,
and thereby causes an evolution in frictional resistance to motion. Our Oso
simulations considered alternative scenarios in which values of all model
parameters except the initial solid volume fraction m0 were held constant.
These values were: basal friction angle = 36; static critical-state solid
volume fraction = 0.64; initial sediment permeability = 10-8 m2; pore-fluid
density = 1100 kg/m3; sediment grain density = 2700 kg/m3; pore-fluid
viscosity = 0.005 Pa-s; and dimensionless sediment compressibility
coefficient = 0.03. Simulations performed using these values and m0 = 0.62
produced widespread landslide liquefaction, runaway acceleration, and
landslide runout distances, patterns and speeds similar to those observed or
inferred for the devastating Oso event. Alternative simulations that used
m0 = 0.64 produced a much slower landslide that did not liquefy and that
traveled only about 100 m before stopping. This relatively benign behavior
is similar to that of several landslides at the Oso site prior to 2014. Our
findings illustrate a behavioral bifurcation that is highly sensitive to the
initial solid volume fraction. They suggest that the destructiveness of the
2014 Oso event may have resulted in part from prior slope deformation that
produced a dilated sediment state that made the sediment susceptible to
contraction and liquefaction as it began to fail on March 22.