Initially the CUS model was used. This proved to be inadequate for waveform modeling since it could not model the observed slower surface wave train. Next Green;s functions for the Utah-Pechmann model were used. This model is adequate for wavefrom analysis of all earthquakes in the western United States. Direct waveform inversion succeeded.
We noticed though that the average dispersion on all paths to European stations from this earthquakes showed much lower group velocities than predicted by this western US model as in the dispersion comparison.
The side-by-side comparison of waveforms in different frequency bands shown in http://www.eas.slu.edu/Earthquake_Center/ MECH.EU/20060529022004/20060529022004.gm shows some interesting features. First in term os peak amplitude, the Utah model generally does quite well up to frequencies of 0.2Hz. The intial part of the signal is matched quite often, although the duration is not. In the slightly lower frequency band, up to 0.1 Hz, the waveform agreement for the vertical component is consistently good, although the scattered wavefield is significant for some stations.
The ability to model to 0.1 Hz is important since waveform modeling of smaller earthquake will only be possible at higher frequencies.
It is also interesting that some paths, e.g., MCRV and VULT ahve simple waveforms.