Tanner: took an event from Gary's catalog, tried to identify the Rayleigh part of it, about 30sec from a local blast. Filtered the data, estimating the frequency of the wave by eye. Then do principal component analysis: compute the covariance matrix, then its eigenvalues, then project the motion onto the two main eigenvectors. Real data looks messy, tried with simulations and it worked fine. Probably have to improve on the narrowband filter. Will have some plots for the next meeting. Daniel provided code and information too, will try to go through it and include it in the coming days.
Daniel: all sounds great. A more common approach people use is to fix the source direction, and then force the azimuth. But Tanner's approach is probably better.
Daniel: showed plots that can be used to convince ourselves that we are looking at the Rayleigh wave, by looking at the velocity vs frequency or period (i.e. dispersion curve). Velocity increases at lower frequencies (larger wavelength) for Rayleigh waves and this is what you would look for.
Gary: could provide additional tools used in seismology to help Tanner's measurement.
Tanner: want to do the analysis at all frequencies (eg 10-20 steps between 0.1 and 10 Hz) and at all stations, to study the amplitude vs depth dependence. And fit a power law to extract the fit parameters.
Daniel: looking at the eigenfunctions as a function of depth would also be very interesting.
Gary: making progress on understanding the data from the surface shots. Have a table for when data is good in a given station. Then cross-correlate stations to get time-lags for each shot to within 1ms. Difficult problem, but seems doable.
Ross: processed the seismic line data at 2000 level, seeing reflectors in the data. Not sure if from open cut or the free surface. Should have a presentation next meeting.
Vuk: Victor gave Levi some pointers about how to look for the missing information on minerals. Levi will follow up on it.