What makes some fault systems form complex patterns, and
what controls the timing, location, and size of earthquakes?
How rigid are the Earth's tectonic plates at depth?
Which forces move tectonic plates, and drive deformation of the Earth's
crust over regional scales?
I address these and other questions by numerically modeling earthquake-related
deformation over time scales from seconds to thousands of
years. The same lithosphere in a particular region must explain deformation
over various time scales while being consistent with geological and geophysical
data. These days, we usually model short and long term deformation separately,
and then make sure these models are at least consistent with each other.
In the future, realistic deformation models will be able to incorporate a
wide range of time scales (e.g., CIG).
Instantaneous deformation of the Earth’s lithosphere due to an
earthquake.
That is, permanent elastic deformation
after shaking has stopped, but before significant inelastic deformation
(creep) has occurred. This modeling is useful for estimating
how faults slip kinematically (how deep? patchy slip distribution
[high stress drop]?). Static deformation modeling also shows how elastic
stresses change on local faults in response to a large earthquake,
which is of use in estimating changes in the likelihood that these
faults will fail in the near future. So far, I have modeled stress
changes due to the 1999 Izmit, Turkey earthquake, and have looked at strike-slip
earthquakes in general, to see how heterogeneous elasticity influences
deformation. Since faults are heterogeneous, coseismic fault slip and
postseismic fault creep (next topic) are interrelated: often it's hard
to tell where one ends and the other begins.
Transient deformation.
I also model accelerated deformation of
the Earth's surface after large earthquakes in an effort
to characterize the structure and rheology of fault zones (or the
lithosphere as a whole) below the brittle-ductile transition.
Right now, the $64,000 question for most plate boundary regions is:
is the upper mantle gooey and incapable of supporting large deviatoric
stresses, or is it essentially rigid over “long” time periods? That
is, how far can we push the classic ideas of plate tectonics, in which
the relative motion of rigid plates (lithosphere) is permitted only
by slip along fault surfaces? This must be known before we can
understand how stresses build up along active fault systems between
earthquakes. Modeling postseismic deformation is a good way to get at this
information, since we are modeling the Earth's response to a known step
in stress (known from modeling the coseismic displacements). The answer
could vary for different plate boundaries.
Relating coseismic and postseismic deformtion to "interseismic"
deformation.
Example. Suppose that the early postseismic deformation following a large
earthquake is attributed to relaxation of low-viscosity (gooey) lower crust.
However, before the earthquake, strain associated with relative plate motion
was was very strongly concentrated within about 50 km of the fault zone.
If the lower crust viscosity is low compared to the recurrence interval then
this is impossible. An explanation is needed: perhaps a strongly nonlinear
or transient rheology, perhaps the early postseismic deformation model was
nonunique (and afterslip also explains the GPS postseismic deformation data)...
Long-term dynamics of fault systems
I am interested in the development of plate boundary fault systems
and the related seismicity. Why are some plate boundaries simple (a single
fault) while others comprise a complex network of faults? Do certain geological
settings lead to the formation of geometrically simple faults? In tectonically
complex regions like southern California, does the timing of earthquakes
(or even high slip rates over many earthquake cycles) on individual faults
depend on changes in loading? Or does it depend on changes in effective normal
stress on faults (pore pressure) during and between earthquakes? Water is
ubiquitous in the upper crust and it is known to play a role in (effectively)
weakening faults, but the interactions between pore pressure change and faulting
must be investigated,
We model the development of fault zones over time assuming a brittle
upper crust whose properties depend on total strain (damage rheology).
The upper crust may be loaded in several different ways, by assuming different
boundary conditions and rheologies at depth. This modeling is in some ways
analogous to modeling postseismic deformation because we are attempting
to learn about aseismic deformation below the brittle crust by modeling
and comparing fault zone complexity and seismicity statistics to observations.
Since the damage model itself contains many parameters, (even excluding
pore pressure effects), this research is fairly complicated.
Some links to like-minded people and groups:
Southern California Earthquake Center
Earthscope
GSC-Pacific
Geodynamics Program
Part of the FE mesh for modeling the Izmit, Turkey earthquake.