geopsy.org

Dear Marc,

I just cannot find info about the meaning of colours in the slowness/freq curve on the low-left side, when presenting results of spac2disp. There is yellow, green, blue etc and there seem to be more pronounced when the number of cells is reduced. Why? Also, in the documentation for MSPAC, the green dots are chosen (out of a multitude of yellow) to deduce the correct dispersion curve. Why is that?

MY last question is: how one can chose the most representative dispersion if curves look very different on each ring?

I send you an attachment of my results. I draw the kmin, kmax and upper and lower slowness band in a way similar to your example. These results are from 10 min spac data from 2 triangles, inner and outer.

many thanks

Theodora

Everything is explained in Wathelet et al. 2005 (BSSA)

Wathelet, M. , D. Jongmans, and M. Ohrnberger (2005). Direct Inversion of Spatial Autocorrelation Curves with the Neighborhood Algorithm. Bulletin of the Seismological Society of America, 95, 1787--1800

If you have further questions about it, we can continue the discussion in this forum

Hi Marc,

Thanks for the suggestion-paper. It was very helpful. Question: The auto-correlation curves must be inverted all together, or individually? In my case I noticed a misfit of ~0.8 if all rings are processed together and a value of 0.2 if individually. It is very difficult to drop the misfit when all together. I was wondering if it makes sense to invert each ring individually and then compare the models...

No, it is required to have at least several rings. With one single ring there are risks of "aliasing" the correct dispersion curve. Better to have at least 2 or 3 rings together.

Usually, it is difficult to get a misfit better than 0.4 with SPAC in a real case. And results "look" fine with such misfit. With a misfit of 0.8, I would seek for the errors. Try to identify which rings or which frequency ranges are supporting the largest deviation. Eventually revise your selection in spac2disp, compare with FK processing if available at the same frequency,... is there any reason to discard some parts of SPAC curves?