9.4.2. Laterally Constrained 1D Inversion

Here, time-domain data are inverted using the laterally constrained 3D inversion code developed by the open-source community in Python. Just like the AtoZ Frequency EM1D example, individual 1D inversions are constrained laterally such that the set of recovered 1D models are smooth horizontally and can ultimately be constrained and interpreted in 3D. The collaborative work invested in SimPEG and empymod has improved the lateral 1D inversion in many ways:

• Avoids the slow read/write of the legacy EM1DTM file format

• Parallelized forward modeling

• Efficient digital filtering scheme (empymod)

As part of this exercise, the user will:

• Set relevant inversion parameters

• Invert the field observations using a set of laterally constrained 1D conductivity models (susceptibility models can be included)

9.4.2.1. Setup for the Exercise

If you have NOT completed the previous tutorial and would like to start here, complete the following steps:

• Download the demo

• Open GIFtools

• Set the working directory

• Import EM1DTM formatted data from the file Assets/TKCdata_VTEM_EM1DTM.dat

• Import 1D mesh (layers file)

• Import the topography data (3D GIF format)

• Create elevation from surface topography

  • Set elevation at 40 m above topography

  • Set i/o header for Z to the elevation column you just created.

• Import and set waveform

• Assign uncertainties from file Uncertainties.dat

-dBzdt at t = 0.00009 s -dBzdt at t = 0.00030 s -dBzdt at t = 0.00190 s

-dBzdt over DO-27 -dBzdt over DO-18 -dBzdt over till

9.4.2.2. 1D TEM Inversion

Here, the set of TEM data are inverted using the laterally constrained 1D approach.

9.4.2.2.1. Setup the inversion

• Create an em1dtm inversion object and set the output directory

• Click on the newly created em1dtm inversion object to set the output directory

• Set any necessary em1dtm inversion parameters under edit options:

  • Global tab:

    • Mode panel: set to laterally constrained 3D

    • Make sure the mesh, observed data and topography are properly set!

    • Trade-off mode panel: Use ‘discrepancy’ mode

    • Other parameters left as default values

  • Conductivity tab:

    • Leave the initial and reference conductivity to best-fitting halfspace

  • Click Apply (NOT Apply and write)

Note

You do NOT need to write all files, as the data and inversion parameters will be passed on to Python as HDF5 file. This will save time by avoiding to read/write the legacy EM1DTM file format

9.4.2.2.2. Run Inversion and Load Results

• Run inversion (SimPEG)

  • Inversion progress will be displayed in a command prompt

  • Results will be loaded automatically at the end of the inversion

• View the model

• View the misfit map

The lateral constraints strategy comes with many advantages:

• Neighboring 1D conductivity models are more consistent

• Conductivity structures are interpolated in 3D, possibly highlighting trends in the model and easing the interpretation.

• Possible to employ a \(\beta\)-cooling strategy similar to the 3D inversion code.

Congratulation, you have recovered two pipe-like bodies by inverting Time-Domain EM data in 1D with the open-source inversion routines SimPEG + empymod! You are invited to try the Python algorithm on the AtoZ FEM1D example