Re: [MSNoise] compute mwcs error

Hi Martha et al., By 1 day, I meant 1 week I guess. Sorry about that. Here are my notes from Aurelien's original message about the FTAN parameters. The bmin and bmax parameters are the bandwidth min and max of the gaussian filters used in the FTAN algorithm. The FTAN applies a set of Gaussian filters to the seismograms with different central frequencies. The width of the filter (in Hz) for the minimum central frequency is bmin and vice versa for bmax. The width for the intermediate filters increases linearly between bmin and bmax. So you can determine the beta for each central frequency (f) as, *beta = linspace(bmin, bmax, nfreqs)* These parameters are equivalent to the "alpha" parameters in this paper defining the FTAN: Herrmann, R. B. (1973). Some aspects of band-pass filtering of surface waves. *Bulletin of the Seismological Society of America*, *63*(2), 663-671. If you look into the cpp source code in libfta.cpp (starting at line 31) you see the Gaussian filter definition. *double filtre_Gaussien(double f, double fh, double beta){//printf("freq : %f filtrage : %f beta : %f\n",f, fh, beta);return 1.0/beta*exp(-0.5*(f-fh)*(f-fh)/(beta*beta));}* Keep in mind that in this msnoise-tomo case, alpha/beta changes with each frequency. If I recall correctly, Aurelien's recommendation was to have the bmin/bmax vary by orders of magnitude (e.g. bmin=0.03, bmax=3.0). I can't find his email stating this or the reason, but this is what I recollect. If you play with the values you will see how the bandwidth changes the structure in the FTAN image. In the end, this changes how the autopicking picks. Please let me know if you have any other questions. Dylan On Wed, Dec 8, 2021 at 8:15 AM Dylan Mikesell <DylanMikesell(a)boisestate.edu> wrote: