Hi@SneakerShot,

the advice I gave you was quite general. I built a model North-East of yours and had that issue there (that's why I was familiar with the Walloon geological survey).

Looking at your data, there is a more urgent fix required: It seems as if you picked points along the exterior of the polygons of your map.

With the green layer, I am not sure which sampling strategy was used. But what you should be aiming for, is the bottom contact of each layer. So something like this:

My drawing is obviously without proper knowledge, what makes sense in your case (edit: in terms of summarizing your current data into a surface to be modeled). After doing this, you might notice, if some layers are unconformaties and if you should utilize more serieses than one you are currently using.

I hope this helps
Nils

This is the default pyvista plotter (after installing gempy-viewer, rungpv.plot_3d(geomodel) or check the associated video :) If you want to use it, you might have to install python locally and not use collab. To my knowledge, pyvista allows only the static view, that you already saw in the collab notebook.

Please ignore my initial tip. Please focus on the time regarding the sampling from the map. Gempy tries to connect surface points of the same surface. If you have the outline of a geological unit in your input data, Gempy will still follow what it was supposed to do and draw mostly horizontal layers or some very weird stuff.

GemGIS was meant as an interface between GIS software and Gempy; reading shapefiles and so on turning them into tables of the required input shape... But it also includes functions to sample from DEMs :)

Thank you for your advices, i will try it today !

Enjoy! Feel free to share the result, if you want some feedback :)

When selecting surface points, it helps to ask yourself, if the selected points represent the true bottom contact of the geological body you are modeling. If not, is it a fault or an unconformity? Faults would be their own structural element.

Hi@SneakerShot,
happy to see, that you got some results! Are you familiar with raster files? You can extract the information needed to create a structured grid from the geotif file from the meta data. If you have the x, y from the meta data, you get corresponding z components and set the topography from that array. Check out the examplehere, where we set the topography from an array.

The rasterio code could look something like

with rasterio.open(geotiff_path) as src:        band = src.read(1)  # Read the first band        transform = src.transform        # Get rows and cols of all pixels        rows, cols = np.meshgrid(            np.arange(band.shape[0]), np.arange(band.shape[1]), indexing='ij'        )        # Convert pixel coordinates to spatial coordinates        xs, ys = rasterio.transform.xy(transform, rows, cols)        xs = np.array(xs)        ys = np.array(ys)        # Flatten arrays        xs = xs.reshape(-1,1)        ys = ys.reshape(-1,1)        zs = band.reshape(-1,1)

That code was created with Chat GPT, but at first glance looks correct. Just have to stack them in to a single array, set the topography as in the tutorial and compute :) (edit: I asked if you are familiar with the format, because you could achieve a similar solution, going through the bounds and getting the resolution, but I want to encourage you to verify the result of either approach by using a scatter plot with the generated coordinates first.)

Aaaah! I found the issue: your raster has 4 bands and I believe none of them has the actual elevation data stored! Please revise how you got your DTM. The values range only between 0 and 255 because they correspond with RGB values and values 1-3 are the same -> grayscale. (edit: tbh, your DTM looks like the hillshade, so definitely check your GIS)

Hello, thank you for your fast and clear answer. I resolved the issue ! Now, I'll start to discover how to use FloPy :)

You are very welcome :) Feel free to come back if you need more help, like extracting output from GemPy