New Research Opportunities in the Earth Sciences - New Research Opportunities in the Earth Sciences - page 36

includes realistic 3D simulations of fault roughness; spatially variable frictional
properties; and other effects, such as basin reverberation and focusing, soil
nonlinearity, and soil-structure interactions. Quantitative modeling now provides a
prospect of eliminating dependence on poorly constrained empirical models, thus
linking seismic hazard analysis for the first time to physics-based concepts such as
stress-time evolution.
Much of this work is currently coordinated by the SCEC, where there is a
community effort to develop 3D rupture models with full 3D implementations, from
finite-element codes to high-resolution 3D community crustal models (Olsen et al.,
2008). These models are currently being used to predict shaking in Los Angeles, San
Francisco, and other cities from ruptures on the San Andreas Fault or other regional
faults. This has been an interdisciplinary effort bridging rock physics, seismology,
soil mechanics, structural geology, and earthquake engineering and requires the use
of today's most powerful supercomputers because representations of faults must span
spatial scales covering many orders of magnitude and because physical quantities
must be calculated at all causally connected points to properly account for stress and
slip evolution.
Advancing earthquake source studies to a full physics-based model of
initiation, propagation, and arrest requires knowledge of the stresses on faults, how
those stresses change with time, and the influence of pore fluid pressure. Resolving
these questions requires improvements in computational resources and support for
theoretical developments so that 3D wavefields can be computed for realistic crustal
environments. Furthermore, additional ground displacement records recorded near
faults during large earthquakes are needed to test the results of dynamic rupture
models. While much of the San Andreas Fault has been instrumented (by other
government agencies) with strong motion sensors, accelerometers do not directly
record ground displacements and cannot distinguish rotations from accelerations.
Combining strong motion records with GPS position estimates (in the same way that
GPS is often combined with more precise gyroscopes in navigation systems) would
address the limitations of strong motion data. It is desirable to support collocation of
strong motion sensors when GPS receivers are installed in fault zones.
The recent earthquake drills, or ShakeOuts, conducted in California (Perry et
al., 2008) and since expanded 1 to Nevada, Utah, Oregon, Idaho, the Central U.S.,
British Columbia, and Guam have used realistic shaking simulations to guide the
responses of millions of people to scenario events. The effort has just begun, and as
computers, 3D methods, and the interfaces between the scientists and engineers,
scientists and first responders, and societal engagement improve, this area will greatly
expand (see Box 2.4). NSF's role includes interagency engagement with the U.S.
Geological Survey (USGS) in SCEC, along with direct funding of many related basic
research efforts in each component (theory, computational support, new observations)
that feed into this hazard area. This work has the potential to transform probabilistic
hazard analysis and to greatly enhance public preparedness for earthquake disasters.
With the recent demonstration that physics-based approaches to probabilistic
seismic hazard analysis are both viable and important, the research opportunity is
1 http://www.shakeout.org/regions