Variables on the velocity grid (xq, yq) show netcdf error

[winnie-chu-aos]

Hello,

I have been encountering an issue where datasets including velocity variables are not able to be combined. As a test, I ran a simulation where I output only zonal (u) and meridional (v) velocity in 3D, where u has dimensions (xq,yh) and v has dimensions (xh,yq). In the regional configuration, these dimensions are output as (xq_001,yh_002) and (xh_002,yq_001), respectively.

I get this error when the reader is working on a file near the edge of the model region (so the tile is smaller than the rest.)

[rank 000] NetCDF copy of yh_sub02 from 19930101.ocean_daily_budgets_1993_01.nc.0023
[rank 000] ERROR in NetCDF /mppnccombine-fast/mppnccombine-fast.c:71 -40 NetCDF: Index exceeds dimension bound

It works fine for all the previous coordinates

[rank 000] NetCDF copy of xh_sub01 from 19930101.ocean_daily_budgets_1993_01.nc.0023
[rank 000] NetCDF copy of yh_sub01 from 19930101.ocean_daily_budgets_1993_01.nc.0023
[rank 000] NetCDF copy of xq_sub01 from 19930101.ocean_daily_budgets_1993_01.nc.0023

but seems to get stuck at yh_sub02. I'm confused becaue this doesn't seem to be an issue for tracer points, and the dimension that it's stuck on is a tracer point, albeit one used for the velocity variables. Additionally, given the particular sub-region where the issue arises, I am wondering if the shortened dimensions of the sub-region itself has something to do with the error?

Attached below is an example of a region that works (19930101.ocean_daily_budgets_1993_01.nc.0022) and a region that does not (19930101.ocean_daily_budgets_1993_01.nc.0023).

Thank you so much for your help!

19930101.ocean_daily_budgets_1993_01.nc.zip

[aidanheerdegen]

Glad to hear this has been useful for your work. The team that maintained this software (of which I was a member) no longer exists as the centre that funded it has closed.

There is a possibility ACCESS-NRI could take over responsibility for mppnccombine-fast as it will likely still be required for high resolution ocean modelling, but I defer to @chrisb13 on this.

@ScottWales wrote the program, so he will need to be consulted about the future of this project.

[winnie-chu-aos]

I see, thanks for the info I really appreciate it!

[ScottWales]

Hi, it looks like the variable yh_sub02 is only present in the *.0022 file, it does not exist in the *.0023 file.

It's expected that dimensions are consistent across the different files, e.g. *.0022 has

        float volcello(time, z_l, yh_sub01, xh_sub01) ;
        float u(time, z_l, yh_sub02, xq_sub01) ;
        float umo(time, z_l, yh_sub02, xq_sub01) ;

while *.0023 uses different dimension names for these variables

        float volcello(time, z_l, yh_sub01, xh_sub01) ;
        float u(time, z_l, yh_sub01, xq_sub01) ;
        float umo(time, z_l, yh_sub01, xq_sub01) ;