Validation of 3D Ray-Theoretical and Finite-Frequency Tomographic Images of the Western U.S. Lithosphere: Comparison of Synthetic Seismograms Generated via the Spectral Element Method

Central to the needs of both natural earthquake monitoring and nuclear treaty verification is the abililty to accurately predict seismic travel times through models used for seismic event location. We report on progress in validation of 3D tomographic models for the Western USA, using synthetic seismograms generated via both ray-theoretical and finite-frequency methods. Allen et al. (2008) constructed three-dimensional tomographic seismic velocity images of mantle structure beneath the western USA using finite-frequency kernels. Their Dynamic North America (DNA) Models of P- and S-velocity structure (DNA098-P and DNA098-S) used teleseismic body-wave traveltime residuals recorded at over 1000 seismic stations: half provided by the Earthscope Transportable Array Flexible Arrays, and the other half from regional seismic networks. Using the same data, models have been constructed using both ray-theory and finite-frequency sensitivity kernels. Capitalizing on the high-speed supercomputing capacity available at Los Alamos National Laboratory, we present model validation for both the ray-theoretical and finite-frequency DNA models using the Spectral Element Method (SEM) to generate synthetic seismograms through the 3D models. These synthetics are compared to the actual data to assess the success of travel time prediction for both ray-theoretical and finite-frequency inversion methods. Implementation of the method for a densely instrumented region such as that covered by the DNA model provides a useful testbed for validation methods we will subsequently apply to other, more challenging study areas.