9.5.3. 3D Inversion¶
In this section, we will invert the simulated data in 3D with three strategies:
Link to DCIP3D documentation
184.108.40.206. Step by step¶
220.127.116.11. Case 1: Unconstrained Inversion¶
As a first step, we invert the simulated data with a purely unconstrained approach.
18.104.22.168. Case 2: Sensitivity weighted inversion¶
The result obtained with the unconstrained approach appears to be dominated by the source-receiver position, with most of the conductivity anomalies recovered near the survey lines. In order to reduce this geometric bias, we will incorporate sensitivity-based weights.
Load the sensitivity file generated by
This solution is an improvement over the purely unconstrained as lower conductivity anomalies are recovered at the electrodes, while the conductive kimberlites are better recovered
22.214.171.124. Case 3: Inversion with 2D starting model¶
In the third case, we will incorporate the stiched 2D model in the 3D inversion through a starting and a reference model.
To create a full 3D model from the merge 2D models created in the merging step, the steps are:
Right-click on the merged model
Edit -> Fill/Interpolate no-data-value(see the documentation)
Keep the default parameters and enter
no-data-valueand choose the
Then you can use that model as starting and reference model:
We have once again improved the solution, and the iteration process is a lot quicker since we are starting with a model closer to the final solution.