The good news: I’ve successfully imported an AerChemMIP netcdf file into the Vapor GUI
The bad news: I cannot replicate my success with two other meteorological netcdf files - one from WRF and the other from MONC (Met Office and NERC Cloud Model)
WRF file error
I am unable to create a render from this file. I’ve noticed that the file does not display anything under ‘Variable Metadata’:
However an ncdump reveals that there definitely are variables (I have screenshots, but as a new user apparently I can only attached one image in this post)
MONC file error:
I get a specific error when I try to create a volume render of the pressure field:
Error performing paint event
Failed to read region from variable/timestep/level/lod (zn, 0, -1, -1)
nc_get_vara_float(65536, 11) : NetCDF: Start+count exceeds dimension bound
Error retrieving variable data
Under ‘Variable Metadata’ the variables are showing with a plus button next to them, but if you click on the plus button there is no additional metadata showing. Again, an ncdump reveals that there are definitely variables in this file.
In both files, the dimension ordering of the variables does not adhere to the Time,Z,Y,X convention.
I saw another post on this forum that mentioned the dimension ordering of coordinate variables, but it sounded as though re-ordering didn’t achieve anything. I wondered if anyone has come across this before and knows of a solution? Thanks in advance!
Hi @annebarber , sorry to know that you run into this problem. I wonder if it’s possible to share a sample data file (through Google Drive, or shared filesystem, etc.) that renders this error for us to inspect? It’s often helpful to look at the data directly to know what needs to be done. Thanks!
Hi @pearse (tagging @sam for reference) yes that fixes the problem. Thank you - I wasn’t aware of the differences in these import options and assumed because all my files are NetCDF I could use NetCDF-CF. I’ve since watched the training video for more information.
Unfortunately I still have the issue with my other file (from MONC) - expanding each variable’s ‘plus’ button in the metadata section shows nothing:
I’m not sure that I can share this file as it is proprietary data but I will check with the owner. I have a feeling this means that the data are not fully CF-compliant which will clearly be important when importing data using this reader. I will continue looking into this myself, but thought I would flag in case anyone has seen this exact error before and has any useful pointers!
Hi @pearse thanks so much, I was waiting to confirm with the data owner that they were ok with me to share, which they have just done.
A bit of background: I spoke to the file owner and he noted that MONC by design retains some of the information in the model configuration file in the resulting netcdf file. This meant that I was seeing three additional ‘dimensions’ that didn’t explicitly describe the spatial/temporal nature of the data (you can see them in the screenshot a few posts up). I thought this might be causing an issue so I used Python to make a copy of the file and drop these extra dimensions along with any variables that referenced them. I can now read the file in Paraview (which I could not do before) but unfortunately have the same problem in Vapor that I did before.
For some reason my modified file (~7GB) is more than twice the size of the original (~3GB). Due to its size I’ve uploaded it to GoogleDrive - if you could let me know when you’ve downloaded it I’ll delete from the account. If this file is too large to deal with let me know and I can send you the original 3GB version (but in theory this file should be better/have fewer issues): MONC_file - Google Drive
It looks like your file doesn’t have a time dimension or coordinate variable. Also, your 3D variables may (or may not) need to be ordered by (time, z, y, x) but I would first try adding time. We have more documentation on our NetCDF-CF data requirements at the link below. Please let me know if the time dimesion/coordinateVar is the fix!
