Page 1 of 1
Posted: Sun Jan 09, 2011 8:47 pm
I was trying to invert dispersion curves from an oceanic region. However, dinver gets stuck when the S-velocity of the first layer is set to zero. Is there any way that dinver can handle a water layer at the top?
Posted: Fri Jan 21, 2011 9:45 am
This is one improvement of the forward computation code that can be foreseen. I had discussions some years ago about that and I also tried to set a null Vs but it does not work so easily. I generates divisions by zero in the current formulation of the problem. If you take Aki & Richard matrices for Rayleigh waves with a null Vs it does not work either. Surface waves measured with a fluid cap are called Stoneley waves and they requires a slightly different formalism. Codes from Robert Herrmann
have or had an option for an ocean/sea layer. A similar computation should be introduced in gpdc/dinver codes. However, if you have some literature references about that topic there are pretty welcome. Personally, I have had not time yet to concentrate on that implementation but if you feed me with some references, things can be sped up.
Posted: Wed Feb 09, 2011 9:27 pm
Sorry for my latish reply. I just saw that you were discussing to include Saito's code into dinver in a separate thread. As this code can also handle liquid layers you could kill two birds with one stone. I further found some remarks on this problem in Aki and Richards (2nd edition; Chapter 7.2; Box 7.6). They refer to Gilbert, F., 'Elastic waves in a transitional solid with arbitrarily small rigidity", GJI, 133, 1998 for a more detailed discussion.
Not sure if this helps.