Custom topography in Held-Suarez model

[drchessgremlin]

Right now I am just trying to modify the default FHS94 run with a topography file. I have simply followed the instructions described here Dynamical Core | Held Suarez and have zeroed out the PHIS variable just to make sure things run under ideal conditions. PHIS = 0 everywhere should run exactly as though no topo file is being used. However, I get the following error:

SHR_REPROSUM_CALC: Input contains 0.10000E+03 NaNs and 0.00000E+00 INFs on process 0
ERROR: shr_reprosum_calc ERROR: NaNs or INFs in input

I am somewhat new to working with CESM and wondering if anybody has some experience with this issue. Thanks!

 

[brianpm]

Seems like your file might contain invalid values. Have you checked that PHIS is actually zero in the file and is a float type? Another possibility is that you saved the modified file as a netCDF4, which the model does not seem able to use (at least that was the case not too long ago when I ran into it). You can check that with "ncdump -k".

 

[islas]

There are some issues with reading in custom topography in some of the recent versions of CESM2. I think if you download the latest release CESM2.2, you should find that this is fixed. Is that an option for you?

 

[drchessgremlin]

Seems like your file might contain invalid values. Have you checked that PHIS is actually zero in the file and is a float type? Another possibility is that you saved the modified file as a netCDF4, which the model does not seem able to use (at least that was the case not too long ago when I ran into it). You can check that with "ncdump -k".

I'm pretty sure that every point has PHIS = 0 and is a float but I'll double check. I do believe the file is being saved as a netCDF4 file though, but this seems to be consistent with the instructions I cited. Once Cheyenne is back up tonight I'll check these things, thanks.

 

[drchessgremlin]

There are some issues with reading in custom topography in some of the recent versions of CESM2. I think if you download the latest release CESM2.2, you should find that this is fixed. Is that an option for you?

I considered this. I've been using CESM2.1.1 which I thought was the most recent version available on Cheyenne. Once Cheyenne is back up tonight I'll check this out. Thanks.

 

[drchessgremlin]

There are some issues with reading in custom topography in some of the recent versions of CESM2. I think if you download the latest release CESM2.2, you should find that this is fixed. Is that an option for you?

So Cheyenne is back up and I see that 2.1.1 is the latest version in /glade/p/cesm/releases/. I'd prefer to use a version available as a prebuilt on Cheyenne. Is there a topo file in /glade/p/cesmdata/inputdata/atm/cam/topo/ that is known to work with FHS94? I've been trying to run with a 0.9x1.25 finite volume grid, which I know is unsupported but runs fine without topography, so I have used SGS-gtopo30_0.9x1.25_remap_c051027.nc. If there are alternatives that are known to run out of the box I could start there though.

 

[drchessgremlin]

There are some issues with reading in custom topography in some of the recent versions of CESM2. I think if you download the latest release CESM2.2, you should find that this is fixed. Is that an option for you?

So I decided just to checkout 2.2.0 from the repo. It seems to be running fine. At very least it doesn't fail immediately. So thank you for the help!

 

[islas]

Great. Glad it's working. I think the issue is there in all CESM2.1 versions, but fixed in CESM2.2. I think it also works in CESM2.0, but I'm not 100% certain. Thanks for raising this issue - I have added a note on the webpage you pointed toward to advise user to use CESM2.2 instead.