[Question] Converting Loop 3D Models to Tomofast-x

[vinicius-anp]

Hi,

I’m a PhD student working under Mark Jessell, using geophysical inversion and modeling. I’d like to know if there’s a way to convert a model created in Loop 3D to Tomofast-x.

I have a mesh grid where each cell is structured as follows:

$N$
$X_{\text{min}}^1 \quad X_{\text{max}}^1 \quad Y_{\text{min}}^1 \quad Y_{\text{max}}^1 \quad Z_{\text{min}}^1 \quad Z_{\text{max}}^1 \quad m^1 \quad i^1 \quad j^1 \quad k^1$
$...$
$X_{\text{min}}^N \quad X_{\text{max}}^N \quad Y_{\text{min}}^N \quad Y_{\text{max}}^N \quad Z_{\text{min}}^N \quad Z_{\text{max}}^N \quad m^N \quad i^N \quad j^N \quad k^N$

$N$ is the number of model cells, $X_{\text{min}}^1 \quad X_{\text{max}}^1 \quad Y_{\text{min}}^1 \quad Y_{\text{max}}^1 \quad Z_{\text{min}}^1 \quad Z_{\text{max}}^1 \quad m^1 \quad i^1 \quad j^1 \quad k^1$ are the min/max coordinates of the X, Y, Z planes of the cell (rectangular prism), $m^i$ is the model value at this cell, $i^i j^i k^i$ is the 3D cell-index (integer).

The mesh consists of cubic cells at the core, transitioning to larger rectangular prisms toward the edges.

Would it be possible to ‘slice’ in cubes/prisms a geological model using these geophysical mesh coordinates? My goal is to use a geological model as a constraint in Tomofast-x inversions. If this exact feature isn’t available, is there a way to incorporate my geophysical mesh coordinates when generating a geological model in Loop? That would allow me to process the data accordingly.

In addition, what formats can I use to export my model from Loop? I could be wrong, but it seems to use a tetrahedral mesh instead of a rectangular one. Sorry for the many questions—it's just that Loop seems like an amazing tool, and I want to check if it will be compatible with my work before jumping in.

Best regards,

Vinicius

[lachlangrose]

Hi @vinicius-anp,

You certainly can do this with LoopStructural, you can export/query the model in any way you would like.

LoopStructural models/interpolators can be queried for the values using either model.evaluate_model_value or model.evaluate_feature_value methods. So you should just need to calculate the cell centres from your tomofast model and calculate the values on this.

Regarding formats, what format do you want? Loopstructural models are format agnostic - you can use a tetrahedral support for interpolating properties but export the model onto a rectangular one if you want. Regarding file formats, vtk, geoh5, omf, gocad, pickle, numpy etc are all possible.

Feel free to get in touch if you are having trouble.

Cheers

[vinicius-anp]

Hi @lachlangrose,

Thank you for the answer. Since Tomofast-x allows prior models in ASCII format separated by spaces (as in my previous message), I will probably export into NumPy and do some processing to get the bound cell coordinates, which consist of cells with a cubic size (100 x 100 x 100) and then increase in size with an expanding factor of 1.15 for the next adjacent cell in the padded area.

I will get familiarized with the software to run some tests, but glad to hear it is format-agnostic.