[Paraview] netcdf

Tue May 16 10:42:31 EDT 2017

Thanks very much Ken and others.

I just tested the warp by scalar filter (I didn't know about this before)
and it looks like it can do what I need with the vertical coord, like you
suggested.

What I will do is this:
I will modify the netcdf file to add a 1D x and y coord variable with the
same name as the x and y dimensions.  I will add a 3D coord var for z.  If
the Paraview netcdf interface can read this in then great!  If the Paraview
netcdf interface cannot handle the 3D z coord then I will use the warp by
scalar filter to apply the z coord var, and will also let you know in case
you want to modify the functionality.

BTW I found this info about coordinate vars on more than 1 dimension:
http://cfconventions.org/cf-conventions/v1.6.0/cf-conventions.html#_two_dimensional_latitude_longitude_coordinate_variables
I'll try the coordinates attribute as suggested in the link above, and will
send you an example netcdf file that I believe to be consistent with CF if
it doesn't work in Paraview.

Regards,
Rupert Gladstone

On Tue, May 16, 2017 at 3:39 PM, Moreland, Kenneth <kmorel at sandia.gov>
wrote:

> Rupert,
>
>
>
> It has been multiple years since I really took a look at the CF
> convention, so I don’t remember the details of whether it supports varying
> coordinates across multiple dimensions or whether the ParaView reader
> supports that. If the ParaView reader is missing some corner of the CF
> convention, let us know and we can work on that.
>
>
>
> In any case, you can displace a 2D or 3D image by a height value that is
> stored in a field by simply running the “Warp by Scalar” filter.
>
>
>
> -Ken
>
>
>
>
>
> *From: *Rupert Gladstone <rupertgladstone1972 at gmail.com>
> *Date: *Tuesday, May 16, 2017 at 12:31 AM
> *To: *"Moreland, Kenneth" <kmorel at sandia.gov>
> *Cc: *Aashish Chaudhary <aashish.chaudhary at kitware.com>, "
> paraview at paraview.org" <paraview at paraview.org>
>
> *Subject: *[EXTERNAL] Re: [Paraview] netcdf
>
>
>
> Hi all, thanks very much for your replies so far.  I must say I am finding
> the Paraview community very helpful.
>
> Having considered my data further, I would be able to make a 1D coordinate
> variable for my x and y dimensions but not for my z dimension.  The ocean
> model uses a hybrid coordinate which can be transformed into depth, but the
> resulting depth coordinate will be spatially varying across my data set.
> In other words my z coordinate variable will need to be a 3D variable.
> From my understanding of COARDS and CF this is allowed.  But I think from
> your email that Paraview will not be able to associate this 3D z coordinate
> variable to another variable (e.g. ocean temperature) for the purposes of
> plotting.  At least, not using the Paraview CF interface.  Is that correct?
>
> I don't currently have a handy netcdf file (I have a rather large one).  I
> could generate a smaller netcdf file to demonstrate the issue, but I am not
> sure it is needed here, as the problem is not that Paraview fails to read
> the file as expected, but rather than I am trying to clarify what I can
> reasonably expect from Paraview with its current functionality.
>
> Do you have a suggestion for how to read in a (topologically) structured
> data set on a rectangular cartesian projection in which the the vertical
> coordinate variable varies across all three dimensions?  Should I ask
> Paraview to read it in as unstructured data?  I think I saw that Paraview
> offers some kind of interface for unstructured data in netcdf files.  Are
> the requirements of this documented somewhere?  Or do you still feel that I
> need to provide an example netcdf file for you to better understand the
> problem?
>
> Thanks again for your help.
>
> Regards,
>
> Rupert
>
>
>
>
>
> On Tue, May 16, 2017 at 6:36 AM, Moreland, Kenneth <kmorel at sandia.gov>
> wrote:
>
> Rupert,
>
>
>
> As Aashish said, it might be easier to diagnose the issue if you sent us a
> file. But if your file is not following the CF or COARDS convention, then
> the reader will simply interpret the arrays in the file as uniform grids
> with spacing of 1. This is not likely to conform with the coordinates you
> want.
>
>
>
> The official documentation for the CF convention is maintained here:
> http://cfconventions.org/. In summary, you specify coordinates by using
> arrays of the same name as the dimension. So for example if you have a 3D
> array with dimensions named “Z”, “Y”, and “X”, then you also make a 1D
> variable named “X” on the “X” dimension that has the x coordinate for each
> grid point in the dimension. Likewise, you have a “Y” variable on the “Y”
> dimension for y coordinates and a “Z” variable on the “Z” dimension for the
> z coordinates.
>
>
>
> Or, if you don’t want to mess with your file format, you can do as Sam
> suggested and transform the data once it is loaded into ParaView.
>
>
>
> -Ken
>
>
>
>
>
> *From: *ParaView <paraview-bounces at paraview.org> on behalf of Aashish
> Chaudhary <aashish.chaudhary at kitware.com>
> *Date: *Monday, May 15, 2017 at 9:08 AM
> *To: *Rupert Gladstone <rupertgladstone1972 at gmail.com>, "
> paraview at paraview.org" <paraview at paraview.org>
> *Subject: *[EXTERNAL] Re: [Paraview] netcdf
>
>
>
> Rupert,
>
>
>
> Would it be possible for you to send us a sample file and what the list of
> VARS you expect to see in the paraview?
>
>
>
> Thanks,
>
>
>
> On Mon, May 15, 2017 at 3:37 AM Rupert Gladstone <
> rupertgladstone1972 at gmail.com> wrote:
>
>
>
> Hi, I have a question about netcdf formats.  I am developing a coupled ice
> sheet - ocean model.  Currently both models run in the same cartesian
> coordinate system.  The ice model outputs unstructured .vtu files, which
> paraview reads just fine.  The ocean model outputs structured netcdf
> files.  If I naively select the "generic and CF conventions" option when
> reading the ocean netcdf file then the data display ok, but not to scale.
> It seems like the structured fields have been read in just fine, but the
> coordinate variables have not.  I don't think the netcdf files are CF
> compliant.  I would like to be able to read in both .vtu files and netcdf
> files and display the data sets together on the same scale.
>
> Do you know what I need to do to read in the netcdf coordinate vars
> correctly?  I am hoping that I can simply insert a post-processing step to
> implement some minor manipulation to the ocean model output files so that
> Paraview can read them in to scale.  Is it simply a case of renaming the
> coordinate variables in the netcdf file such that they have the same names
> as the corresponding dimensions?  Note that this is not in general possible
> as some of the coordinate variables are two-dimensional variables (in the
> horizontal plane), though for most of the simulations we plan in the near
> future I can enforce that the coord vars will have a one to one
> correspondence to the dimension vars (i.e. I can enforce that coord vars
> are 1D).
>
> Is the ordering of dimensions important?
> Is the ordering of coordinate vars important?
> Is the naming of dimensions important?
> Is the naming of coord vars important?
> Can Paraview cope with 2D coord vars?  If so, how does Paraview know which
> coordinate var applies to which var?
>
> Thanks very much for your help.
>
> Regards,
> Rupert Gladstone
>
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at http://www.kitware.com/
> opensource/opensource.html
>
> Please keep messages on-topic and check the ParaView Wiki at:
> http://paraview.org/Wiki/ParaView
>
> Search the list archives at: http://markmail.org/search/?q=ParaView
>
> Follow this link to subscribe/unsubscribe:
> http://public.kitware.com/mailman/listinfo/paraview
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/paraview/attachments/20170516/afc7f0ba/attachment-0001.html>