Can Dinver deal with this kind of dispersion curve?

This forum is dedicated to discuss all problems and suggestions related to the inversion software
Post Reply
leedward
Posts: 5
Joined: Wed Sep 12, 2007 7:48 pm

Can Dinver deal with this kind of dispersion curve?

Post by leedward »

Hi, Marc,

Can Dinver deal with the dispersion curve with dominant higher mode? Thanks.

admin
Site Admin
Posts: 446
Joined: Mon Aug 13, 2007 11:48 am
Location: ISTerre
Contact:

Higher modes dispersion curves

Post by admin »

Hi,

No problem, when you specify the target dispersion curve, the fundamental Rayleigh mode is set by default. At your option, you can change it to first or second higher mode, or even switch it to Love modes. Group and phase velocities are also supported.

There's also the possibility to specify more than one mode for a particular branch of a dispersion curve. For instance, fundamental, first and second higher modes. In some situations, modal identification is subjective or simply impossible. This way the inversion will scan for all possibilities and will output all families of models.

Marc

leedward
Posts: 5
Joined: Wed Sep 12, 2007 7:48 pm

Post by leedward »

I am very interested in specifying more than one mode for a particular branch of a dispersion curve. Because I have gotten some dispersion curves with big jump to higer mode. So how can I do to define this part as the higer mode and the rest as the fundamental mode? Thanks

admin
Site Admin
Posts: 446
Joined: Mon Aug 13, 2007 11:48 am
Location: ISTerre
Contact:

Post by admin »

If your curve shows obviously two modes, the best is to cut it, split the two parts and provide two curves to the inversion, one identified as the fundamental and the other as first higher mode.

Load your curve two times in dinver target module, for the first curve cut the fundamental part to remove the high mode (or what's currently identified as higher mode). Switch to the second curve, remove the fundamental mode part and set the mode index to 1.

Usually, there is a branch of curve in between the two modes. I think that the best option is to remove this transitional part from the inversion. Some authors claim that it is possible to invert the "effective" mode (based on the energy partition of modes). I think that it may corrupt the final results because the resolution power of the array (linear or 2D) used for the measurements is never included. We have many experiences where a small array cannot well resolve the two modes (we have something in-between, or a longer transitional range) and a bigger one resolves them nicely. In fact, there is no intermediate mode in the case of a perfect 1D structure. For real cases, this is another story, but the effective mode approach is still for 1D, so apples are inverted with a parameterized model of pears.

Identification of higher modes is still not a straight forward processing. If you cannot get a good fit and if the inverted structure does not correspond to a realistic model, one has to go back to the identification of modes and change it. In some cases, it may be interesting to invert the modes separately (only fundamental) to test the mode identification. I remember a case where the higher mode was in fact the second higher mode. Once modes are clearly identified, you can run the inversion with the modes together.

Post Reply