9.5.2. 2D Inversion

In this section, we will invert the simulated data in 2D as a pre-processing step for the 3D inversion.

9.5.2.1. Setup for the Exercise

• Download the demo

• Open GIFtools

• Set the working directory

Tip

• Requires at least GIFtools 2.26 (login required)

9.5.2.1.1. Extract 2D data objects

In the previous simulation section, we have generated a DC3Ddata object, which we first want to invert in 2D:

• Assign simple uncertainties

  • % = 15

  • floor = \(0.05\)

• Set the data/uncertainties

• Copy the LineID from the DC3Ddata object

• Seperate the survey lines

  • Export the 2D topography at 20 m resolution.

9.5.2.1.2. Explore the model space

Before attempting to invert all data, it is good practice to test different assumptions on a single line. First let’s test different reference conductivity values by running a series of 2D inversions.

• Create a DC2Dinversion object to serve as a template

  • Set \(\alpha_s=0.0025, \alpha_x=\alpha_z=1\)

  • Select data from Line 7 (directly above the conductive kimberlite)

• Create a Model Space object

• Edit the Model Space inversion options and set mref over a range \([1e-5,\;1e-2,\;4] in log space\)

• Write All and Run

• Upon completion, Load the results First to target and Extract model section

Reference: 1e-5 S/m Reference: 5e-5 S/m Reference: 1e-4 S/m Reference: 5e-4 S/m Reference: 1e-3 S/m Reference: 5e-3 S/m

[OPTIONAL]

-Create a DOI using two of the recovered models

Note

Changing the reference conductivity value can drastically change the solution at depth, which can be used to estimate the Depth-of-Investigation (DOI) of a geophysical experiment.

9.5.2.1.3. Run a Batch Inversion

We have a total of 10 DC2Ddata objects that would like to invert. Rather than manually inverting each line, we will make use of the Batch Inversion object to speedup the process.

• Create a DC2Dinversion object to serve as a template

  • Set \(\alpha_s=0.0025, \alpha_x=\alpha_z=1\)

  • Set the reference conductivity to \(1e-4\) S/m

• Create and edit a Batch Inversion

• Run all

• Upon completion, load all results First to targets

9.5.2.1.4. Merge and interpolate models

While we can view each inversion result on their respective 2D mesh, in this section we will bring together the 2D models into our 3D mesh for later use.

• Merge results using the DCSurveyFull lineID property as reference

  • Right-click on the ` BatchInversion

  • Click Merge all

  • Select the 3D original mesh

  • Select the DC3Ddata object generated in the DC3DForward

  • Check that line IDs are correct

  • Click OK to proceed

Note

• We have recovered conductive anomalies consistent across lines.

• The chosen best-fitting half-space conductivity might be slightly too high due to the thin conductive overburden. The user is invited to repeat the experiment with different background conductivity values.