Initial processing is done with the CUS model. This stage is important for quality control which identifies stations that have significant outliers. The model is not appropriate for time-domain modeling at frequencies greater than 0.05 Hz. A Pacific Northwest model is preferred. This model will eventually be constructed using waveform data from earthquakes such as this. It is only by attempting to fit the data with the simple CUS model that we can understand the imperfections in using that model.
The preferred solution is that obtained using the modified Yellowstone model, which at least does better in fitting the shorter period dispersion seen in the data. In comparing the solutions, note that the EW trending tension axis is defined, but that the nodal places differ.
The following link used the modified Yellowstone model to search for the surface-wave spectral amplitude focal mechanism estimate and also compares the observed and predicted waveforms. The source depth and moment magnitude are essentially the same as those estimated using the CUS model.
Interestingly, the depth sensitivity is much better using the modified this model. The presentation shows the waveform fit in two frequency bands 0.02 - 0.05 Hz for comparison with the similar figure with the CUS model and 0.02 - 0.40 Hz to indicate how well peak amplitudes are modeled at high frequency.