The challenge facing the ground motion group is validation of codes against available data so that broadband ground motions relevant to building damage can be predicted with confidence from future earthquakes. While considerable progress has been made at frequencies below 1 Hz, large-scale high-frequency modeling is beyond both computational resources and our detailed knowledge of source and path. Various empirical schemes have been used to add high frequencies to computed seismograms, but without a physical basis their reliability is in question. Even at low frequencies, inadequate knowledge of the path limits how much of the coda can be predicted. Both high frequency strong shaking and long-term coda are important for engineering considerations. The ground motion group has made significant progress in attacking these problems with a series of numerical calculations, validations and experiments coordinated with other groups across SCEC (e.g., ESP, IIG, CVM, CFM, USR,CME).

Numerical Simulations

Larger imageTerashake movie frame. The rupture travels SE along the San Andreas fault. Note the directivity to the southeast, trapping of waves in the Los Angeles and Ventura basins and irregular shaking pattern from fault segmentation.

A number of different methods have been implemented to model the high frequency part of seismograms. Yehua Zeng uses a combination of rough source, randomly distributed scatterers and reverberations in near surface layering. Arben Pitarka and Rob Graves add a stochastic component to low frequency deterministic calculations. Tom Heaton adds high frequency data from nearby strong motion instruments to wavefields calculated using Jeroen Tromp’s spectral element method. Kim Olsen uses finite differences to model low frequencies and ray synthetics to deterministically model high frequencies with sources constrained by pseudo dynamics. The method claims smoother phase transitions across the spectral band. The same source model is used and a weighted superposition is used to combine the low and high frequency bands. The optimal method continues to be an area of active enquiry.

The ground motion group were active participants in the PEER-USGS-SCEC sponsored NGA-E (Next Generation Attenuation) project that has now been funded directly by NSF for a 3-year project. The SCEC research involves comparison of broadband simulations against data from large earthquakes (See IIG report). The NGA modeling effort presents an essential conduit through which SCEC research flows to practical use.

Modeling and Observations

Larger imageFoam rubber simulations. The model is shown upper left. The data (red lines) is compared with Steve Day’s dynamic rupture code, which matches low frequencies but not the high frequencies generated on the fault plane.

Work continues on finding geological constraints on historic strong ground motion by analyzing survivability of precariously balanced rocks. After the remarkable observation of a line of rocks mid-way between the Elsinore and San Jacinto faults, presumed to be located just far enough from each to survive historic shaking, a new reconnaissance study of the region between the San Andreas and San Jacinto is being undertaken. Further theoretical and observational work continues to quantify the observations.

Path and Scattering Effects

Kim Olsen introduced an empirical Q model into the CVM, where Q was taken to be proportional to S wave speed. This resulted in significant variance reduction between theory and data. Peter Shearer and Egill Haukkson are inferring Q from local earthquake data. They use stacked spectra from hundreds of thousands of double difference-relocated events to isolate source, path, and site spectral response. At the SCEC meeting they presented a new Q model for the upper crust. Jamie Steidl and Pengcheng Liu have inverted for Q(z,f) at various depths in the SCEC borehole seismic array, and show that at high frequencies the last few hundred meters can be as attenuative as the remaining path. Ralph Archuleta is examining source, path and site effects on the 150 station Yokahama seismic array with the objective of explaining long durations and the various ground motion factors relative to the geology.

Larger imageShear wave velocity cross sections through the SCEC and Harvard 3D velocity models (click for details...)

Larger imageMap of peak simulated ground velocity for the Northridge earthquake.

Engineering Applications

Tom Heaton leads an effort involving active collaboration between building and ground modelers to model the hazard presented by the newly discovered Puente Hills blind thrust. Ground motions up to 1.5 Hz are calculated using Tromp’s spectral elements program, coupled with historic recordings of strong ground motion for higher frequencies. Programs written by (structural engineer) Hall’s group have been used to model building response for (up to) 40- floor buildings. The simulations are used to identify times and locations of structural damage. The structure codes can take the buildings all the way to collapse. Various movies of building excitation were presented at the SCEC annual meeting with assessment of maximum damage presented in graphical form on the structural drawings.

Future Directions

Future directions include comparing and validating broadband ground motion with observations and improving modeling schemes; Testing of the CVM and inversion of observed seismograms to improve the CVM; Adding to the CVM the SCEC scattering and attenuation model by identifying and modeling sources of scattering; Developing methods for incorporating nonlinear site response for large amplitude ground motion events in Southern California including site and structural response; Developing collaborations with engineers (with IIG) to add building response to synthetic seismograms and identify seismogram characteristics important for damage.