Hi Marc,
One question, what is the correlation that dinver uses when the parameterization correlates the density and the Vp to the Vs.
Doubts in parameterization
Doubts in parameterization
- Attachments
-
- Parametrization.png
- (261.32 KiB) Not downloaded yet
Re: Doubts in parameterization
Hi,
When you link a layer from a profile to a layer of another profile, the bottom depth of the layer of the first profile is forced to be the same as the bottom depth of the second profile. No parameter for depth or thickness is added for the first profile. Usually layers from Vp and density are linked to those of Vs.
If the density is held constant and equal in all layers, it is useless to add many layers for density. Just a single layer is fine in your case.
For Poisson's ratio, this is not a parameter profile. It only adds new conditions between parameters on Vp and Vs layers. Hence if all layers have the same boundaries (0.2 and 0.5) it is useless and not efficient to have multiple layers for Poisson-s ratio. A single layer is sufficient. If you want to change the limits on Poisson's ratio with depth, introduce new layers but always link their depths to Vs or Vp. If not, a new parameter for the depth of the changing limit of Poisson's ratio will be added. Not necessary to add that such parameter is usually not constrained, useless and slowing down the inversion process.
For a simple parameterization, to summarize: one layer for Poisson's ratio and density, 2 to N layers (start from a low number of layers and increase the number until getting an acceptable fit). Keep the same number of layers for Vp and Vs, link them for the depth. Set depth parameter rather than thickness.
Best regards,
Marc
When you link a layer from a profile to a layer of another profile, the bottom depth of the layer of the first profile is forced to be the same as the bottom depth of the second profile. No parameter for depth or thickness is added for the first profile. Usually layers from Vp and density are linked to those of Vs.
If the density is held constant and equal in all layers, it is useless to add many layers for density. Just a single layer is fine in your case.
For Poisson's ratio, this is not a parameter profile. It only adds new conditions between parameters on Vp and Vs layers. Hence if all layers have the same boundaries (0.2 and 0.5) it is useless and not efficient to have multiple layers for Poisson-s ratio. A single layer is sufficient. If you want to change the limits on Poisson's ratio with depth, introduce new layers but always link their depths to Vs or Vp. If not, a new parameter for the depth of the changing limit of Poisson's ratio will be added. Not necessary to add that such parameter is usually not constrained, useless and slowing down the inversion process.
For a simple parameterization, to summarize: one layer for Poisson's ratio and density, 2 to N layers (start from a low number of layers and increase the number until getting an acceptable fit). Keep the same number of layers for Vp and Vs, link them for the depth. Set depth parameter rather than thickness.
Best regards,
Marc