joint inversion Rayleigh and Love

This forum is dedicated to discuss all problems and suggestions related to the inversion software
Post Reply
fabioca
Posts: 9
Joined: Tue Jun 07, 2022 4:55 pm

joint inversion Rayleigh and Love

Post by fabioca »

Dear Marc,
is there a way to compute a joint inversion (isotropic) for both Rayleigh and Love fundamental modes?

Thank you,
admin
Site Admin
Posts: 811
Joined: Mon Aug 13, 2007 11:48 am
Location: ISTerre
Contact:

Re: joint inversion Rayleigh and Love

Post by admin »

Hi,

This is somehow a routine work in Dinver. For each target curve (more than one curve can be provided), a few modes can be attributed, generally a single one. A "mode" is defined by the velocity type (Group or Phase), the polarization (Rayleigh or Love) and the mode index (0=fundamental, 1=1st higher,...). If there are several modes for a curve, the misfit components (for each sample of the target curve) are computed with the closest mode, the mode that best fit. It is a way of automatically identify modes, whenever you have a doubt: a mode branch that can be the 2nd or the 3rd higher mode for instance.

Best regards,

Marc
fabioca
Posts: 9
Joined: Tue Jun 07, 2022 4:55 pm

Re: joint inversion Rayleigh and Love

Post by fabioca »

Hi MArc,
thank you. So, if I pass a target with two dispersion curves (one R0 and one L0) to dinver, I can run one inversion to fit both curves? I thought dinver wated only one polarization (Rayleigh or Love).
admin
Site Admin
Posts: 811
Joined: Mon Aug 13, 2007 11:48 am
Location: ISTerre
Contact:

Re: joint inversion Rayleigh and Love

Post by admin »

Exactly.
For a particular curve, it is non sense to have both polarizations. The user should know the polarization of each curve. It would add useless uncertainty to the results.
fabioca
Posts: 9
Joined: Tue Jun 07, 2022 4:55 pm

Re: joint inversion Rayleigh and Love

Post by fabioca »

Dear Marc., thank you.
I have seen now your last answer, but I am probbaly missing something. My question was referring to the possibility of running a joint inversion of R0 and L0 for isotropic structure (assuming VSV=VSH), so to exploit the different sensitivity of R0 and L0 with depth. is this possible to do with geopsy?
Of course I can run separate inversion for VSV and VSH, but it is not not the same thing.

Any feedback is appreciated. Thank you very much.
Fabio
admin
Site Admin
Posts: 811
Joined: Mon Aug 13, 2007 11:48 am
Location: ISTerre
Contact:

Re: joint inversion Rayleigh and Love

Post by admin »

Hi Fabio,

Sorry for the delays... Summer holidays and then I was busy with other topics.
If you try to invert at the same time a Love and a Rayleigh curve, there is a single Vs value for each layer (in all directions, isotropic).
I'm unsure to fully understand your question.

Marc
fabioca
Posts: 9
Joined: Tue Jun 07, 2022 4:55 pm

Re: joint inversion Rayleigh and Love

Post by fabioca »

Marc, thank you very much. I was able to run the inversion i wanted, so everything is OK. I really appreciate the possibility to run everything automatically from command line. Another question.. probably not really on the joint inversion.
I would like to not invert for VP and density, rather to fix them based on the VS values (i.e. the classical Vp=VS*sqrt(3) and Brocher or Christensen relationship for density. Density is not really an issue, but when I run the inversion, obtained VP tends to 'participate' in improving the fit and sometimes results in values that are unphysical (weird Vp/VS ratio). I would like to avoid this. I was wondering if in dinver there is this option.
Thank you again for the excellent work you have done.
admin
Site Admin
Posts: 811
Joined: Mon Aug 13, 2007 11:48 am
Location: ISTerre
Contact:

Re: joint inversion Rayleigh and Love

Post by admin »

Poisson's ratio profile is designed for that purpose. If you do not consider the relation between Vp and Vs, leave a single layer with reasonable limits like the default one (0.2 to 0.5). If you want to get more control over the relationship, reduce the range, eventually with particular ranges for each layer. With very narrow ranges for Poisson's ratio, it may be difficult or impossible to find a valid model. The initialization of the inversion may take ages or it may fail to start.

Best regards,
Marc
Post Reply