Laterally constraint inversion

[zibi]

Hi Marc,

I am running dinver for Rayleigh wave dispersion curve for a series of points along the profile. I am calling it from the command line to automate the process. However, sometimes the best-fitting model is significantly different from the adjacent points.

Is there any possibility to constrain the inversion so it takes into account the neighboring DCs or models? I know that you can add weights to data points, but I am not sure what is the weight used for. Do you think creating a target with many adjacent curves, with the weights decreasing with distance, is a way to constrain it?

Looking forward to your answer
Zibi

[admin]

Hi Zibi,

The weight is used to combine misfits of distinct natures (e.g. based on dispersion curve, ellipticity curve or HV peak). It is not assigned individually to each dispersion curve, so it will be of little help in your case.

Some ten years ago, I investigated how two neighbouring dispersion curves can mutually constrain each other. Unlike claimed by some authors, I could not find any robust improvement of the inversion results by running a simultaneous inversion. Some tools were developed at that time but everything remained at a draft level.

If you want smooth modifications of Vs across the various profiles, you should define how two neighbouring profiles are connected: same number of layers, variation rules for Vs, thickness,...

[zibi]

Hi Marc,

Thank you for your reply. I was hoping that lateral dependence of dispersion curves would help constrain the inversion. I see that this is not the case.

Best regards,
Zibi