export method for the disu, disv, npf etc. does not create .nc files

[giovannifi]

I defined a model grid using voronoi cells. I built the model using the DISV package in MF6.
When I try to use the following commands in python:

gwf.disv.export(workspace + '\\' + 'disv_mf6.nc')
gwf.npf.export(workspace + '\\' + 'npf_mf6.nc')

I get the following error

Traceback (most recent call last):

  File "H:/scripts/python3.6/modflow/post_process_mf.py", line 78, in <module>
    gwf.npf.export(workspace + '\\' + 'npf_mf6.nc')

  File "C:\Program Files\Python36\lib\site-packages\flopy\mf6\mfpackage.py", line 1762, in export
    return export.utils.package_export(f, self, **kwargs)

File "C:\Program Files\Python36\lib\site-packages\flopy\export\utils.py", line 468, in package_export
f = NetCdf(f, pak.parent, **kwargs)

  File "C:\Program Files\Python36\lib\site-packages\flopy\export\netcdf.py", line 175, in __init__
    self.model_grid.grid_type))

Exception: Grid type vertex not supported.

everything works fine if I try to export shape files.
Thanks in advance

[christianlangevin]

The last we checked there wasn't a stable standard for writing netcdf files for unstructured grids. Maybe this has changed by now, but it has been low on our priority list. Is there a reason you need netcdf files for your model?

[giovannifi]

No worries. I have no urgencies. Just wanted to notify the bug if it is a bug. It would be nice to have the option. I wanted to load the model with paraview and create some nice 3d pictures for reporting purposes. I also wanted to produce some cross sections. I managed to write a script code for the last one but it is not really efficient.

[christianlangevin]

If you want to make plots with Paraview, then VTK is probably a better option. You should be able to export a DISV model to VTK output files.

[giovannifi]

I had the impression that the vtk commands/methods work only for the structured grids. I have not figured out how to use them for an unstructured grid with voronoi cells. If you say that it is possible, I will investigate more. Thanks a lot for that.

[christianlangevin]

Sorry, you're right. I don't think we have VTK support yet for unstructured grids.

[giovannifi]

I manged to create a function to create the VTK file for unstructured grids using the module vtk that has to be installed in python with the command:

pip install vtk

The code can be found here: https://www.dropbox.com/s/n2p9b0w7lef038f/vtk_grid.zip?dl=0

My script generates the VTK file for a horizontal unstructured grid built for MF6 with the DISV package. It also adds the hydraulic conductivity to the vtk file as a cell scalar.
The key function of my script is called MakePrism, that, I would reckon, could also work for a fully unstructured grid provided that "Faces" and "p_list" will be defined accordingly.

• “P_list” is a list of the coordinates of all the vertices
• “Faces” is a list of lists whose length is equal to the number of cells. For each cell you have to define a list of vertices for each face of the cell

I hope this could help anyone.

[jdhughes-dev]

@langevin-usgs is this something DJ could work on?

[christianlangevin]

@etiennebresciani, do you have any interest and time in implementing VTK support for unstructured grids?

[etiennebresciani]

I would be interested in principle, but time is indeed an issue here... I would have a hard time justifying it as I am not working with unstructured grids at this stage.

Just some thoughts: it may not be a bad idea to indeed use the genuine vtk python module, as @giovannifi did. It could be used throughout (i.e., even for structured grids). This would add a dependency, but on the other hand it would reduce the amount of code to maintain.

[marcowhites]

Hi, any update on this topic? I'd like to export a DISU model outputs into vtk format

[jlarsen-usgs]

@marcowhites, @giovannifi

I've started working on updating/rewriting the VTK code to be able to export DISV and DISU models. @giovannifi thanks for sharing your code, I've been looking through it and will probably adapt some if it for the flopy VTK module.

[marcowhites]

@jlarsen-usgs
Thanks! For now, since I am using Flopy and Gridgen, I am using a piece of code to paste the modelling results on the VTK file created by Gridgen when I generate the DISU grid. It's clearly a rough patch, but it works