We have used a raster GIS (Idrisi for Windows) to calculate
the static Factor of Safety (assuming infinite slab failure) for all the ca.
550,000 30-m cells in the urban Seattle area (covers the Shilshole Bay, Seattle
North, Duwamish Head, and Seattle South 7.5' quadrangles).
From
the Static Factor of Safety maps, we calculated the critical acceleration
needed to induce downslope movement, after the method of Newmark (1965). Next,
several ground motion scenarios were generated: 1) a probabilistic set, in
which the Peak Ground Accelerations (PGAs) with 2%, 5%, and 10% probability of
exceedance (PE) in 50 years were assigned to all pixels, and 2) a deterministic
set, representing a M=6.7 earthquake on the Seattle fault, a M=7.5 earthquake
on the South Whidbey Island fault, and a M=8.5 earthquake on the Cascadia
megathrust. In each of these models PGA was converted to Arias Intensity using
empirical relationships.
Finally,
the ground motion scenarios were applied to the maps of Critical Acceleration
to predict Newmark displacement in each scenario. For each of the 6 ground
motion scenarios listed above, we assumed one of four different groundwater
conditions at the time of the earthquake: 1) dry, i.e. the 5-ft thick potential
failure slabs on hillslopes are completely dry, 2) intermediate, i.e. the lower
parts of the slabs near the Vashon outwash/Lawton Clay contact were saturated,
3) wet, in which the zone of saturation rises in slabs near the outwash/clay
contact, spanning a range of 200 m from the contact, and 4) saturated, i.e. all
cells in the area are saturated right to the surface (=worst-case scenario).
This
combination of the 6 ground motion scenarios, and the 4 possible groundwater
conditions, yield 24 scenario maps of landslides. To simplify interpretation,
the Newmark displacements have been reclassified into categories of: 1) no
slide (green), meaning predicted Newmark displacement is less than the
threshold for catastrophic failure in that unit (ranges from 5 cm in sands such
as the Vashon advance outwash, to 20 cm in cohesive sediments such as the
Lawton Clay), 2) threshold slides (yellow), in which the predicted Newmark
displacement exceeds the threshold by 5 cm or less, and 3) slides (red), in
which the predicted Newmark displacement is more than 5 cm larger than the
assumed threshold for catastrophic failure. An example is the small number of
landslides predicted in the intermediate ground water condition, if shaking has
a 10% PE in 50 years (Fig. 1)
Each of
the 24 scenarios has a probability of occurrence in the next 50 years,
calculated as the product of: 1) the probability of the ground motion occurring
(or being exceeded) in 50 years, and 2) the probability that a given
groundwater condition will occur on the day of the earthquake. Probabilities
for the probabilistic scenarios are self-defined. Probabilities for the
deterministic scenarios are determined as the inverse of the following
estimated recurrence times for the three fault sources: 1) Seattle fault, 4000
years; 2) South Whidbey Island Fault, 4000 years, and 3) Cascadia Megathrust,
500 years. Probabilities of the four groundwater scenarios occuring were
crudely estimated from local consulting companies. The dry condition is
estimated to prevail over 50% of a given (typical) year, and the saturated
condition to occur about 2 weeks per year (about 5% of the year). Probabilities
of the other groundwater scenarios are unknown, but were arbitrarily assumed to
be about 30% for the intermediate and 15% for the wet scenario.
The number of predicted landslide cells (does not include threshold landslides) and the probability of the scenarios occurring in the next 50 years are given below for the full 4-quadrangle map area. Note that the scenarios with the higher probabilities (such as (Fig. 1), which have net probabilities of 0.03-0.05) predict the smallest number of landslides. Conversely, the "worst-case" scenarios, which assume both high ground motions (which have low probabilities) and wet or saturated groundwater (also low probability) predict very large numbers of landslides (roughly 15,000-25,000 cells predicted to Fig. 2 ). However, note that these scenarios also have very low associated probabilities (e.g., 0.000625, or 1 chance in 1600 of occurring in a 50 year period). Mitigation of these "worst-case" scenarios would be very expensive, and engineers and planners may decide that it is uneconomical to try to protect against landslide scenarios with only a 1 in 1600 chance of occurring during a structure's lifetime. However, they may judge that landslide areas predicted in higher-probability scenarios (e.g., 0.01-0.05 in 50 years) would be the best beneficiaries of the limited funds for structural or nonstructural mitigation.
Return to GEO-HAZ Home Page.
