From: Randall Peters PETERS_RD@..........

Date: Tue, 22 Dec 2009 16:20:39 -0500

Chuck, First do a high pass filter on your data with a corner frequency of say= 0.005 Hz, to remove the secular trend in the record and eliminate the prob= lem of Gibb effect. Then take the FFT of the complete one-day record and l= ook to see if there is a microseismic hump in the spectrum. Note the frequ= ency of the maximum in that hump and use its value to again do filtering, b= ut this time both a high and a low pass at the same value, corresponding to= a bandpass filter at the hump frequency. Then open the filtered signal wi= th an x-scale of say 2000 s. After doing this you will see that the micro'= s 'come and go' according to what is called their correlation time, typical= ly no more than 10 or twenty cycles. For the hump having a period of 3 s (= 0.33 Hz), this means a given micro-signal persists for no more than about 3= 0 s before dying out, only to be followed by another short lived signal. T= hese approximately 30 s pieces comprise individually a set of very sharp sp= ectral lines, separated from one another (closest neighbors) by what appear= s to be values of the eigenmode frequencies, in mHz. To see these individu= al lines use the WinQuake FFT that is the rightmost one of the pair in Larr= y's code. They will be most vivid if displayed in linear rather than log s= cale. If you place the cursor on a given line you can read its frequency a= nd then move on to an adjacent line, left or right; read it and then write = down the difference value between them. I am especially interested in seei= ng how many of these difference frequency values comprise a distinct set of= values, as opposed to being randomly distributed group of numbers. If so,= then the free modes must be modulating the microseisms. The only way to k= now for sure is to collect enough delta-f values in the manner I've just de= scribed to generate a statistical base having reasonable confidence. I did= a cursory inspection of some of my records this way, and the results look = quite promising. Those of us familiar with heterodyne processes in radio s= hould not be surprised. The process of modulation involves a carrier that = is influenced by lower frequency signals by means of nonlinear processes. = In my thinking, the mircroseisms would be the carrier, and the eigenmodes b= e "what our planet is trying to tell us if we have the demodulation means i= n a receiver to hear what she is saying".=20 Randall= __________________________________________________________ Public Seismic Network Mailing List (PSN-L)