search

NOAA-GFDL / FMS

GFDL's Flexible Modeling System
Other
93 stars 135 forks source link

Integer indexing for sub-regions in diag_table #1209

Closed kshedstrom closed 6 months ago

kshedstrom commented 1 year ago

I swear I ran my regional Arctic domain with a bunch of cross sections saved, using the i,j indices of the cross sections. It no longer works, giving me the error: FATAL from PE 0: diag_util_mod::get_index: array NOT monotonously ordered

I get this error with the lat/lon locations as well, since the lat/lon values are indeed not monotonously ordered.

Would it be possible to get the i,j indices working again, for instance like:

"ocean_model_z", "volcello", "volcello", "ocean_Bering_Strait", "all", "mean", "321 330 184 184 -1 -1",2

thomas-robinson commented 1 year ago

The current diag manager does not support the use of indices for a subregion. It only supports the use of lat/lon. The new diag manager being developed will support the use of indices. If this entry worked in your diag table, I am not sure how.

If you have a problem with a lat/lon subregion, then we will be more concerned with that. What does the diag table entry look like for that?

kshedstrom commented 1 year ago

The lat/lon examples look much the same: "ocean_model_z", "volcello", "volcello", "ocean_Bering_Strait", "all", "mean", "176.75 178.1056 67.9 67.9 -1 -1",2

They fail the same way since my domain covers the north pole and latitude is not monotonic across the domain, plus of course longitude wraps around at +/-180.

nikizadehgfdl commented 1 year ago

@kshedstrom Would you be able to provide us with a workdir on gaea to reproduce the error? I think we need to have your grid to investigate this issue as it appears specific to the ocean_hgrid.nc

FYI , for a small global test case of MOM6-SIS2, when I use the 360 complements of the lons you used in diag_table the test runs:

"ocean_model_z", "thetao","thetao",  "ocean_Bering_Strait",    "all", "mean", "-183.25 -181.8944  67.9 67.9  -1 -1",2

but when I change it to your choice of longitudes I get an error

"ocean_model_z", "thetao","thetao",  "ocean_Bering_Strait",    "all", "mean", "176.75 178.1056  67.9 67.9 -1 -1",2

I get

FATAL from PE    26: diag_util_mod::get_subfield_size: can not find gstart_indx/gend_indx for thetao, check region bounds for axis  1

It could be that FMS diag_manager is looking for bounding longitudes commensurate with the ocean_hgrid bounds (-280 to 80 in this test case).

FYI, in the OM4p125 diag_table for sections Bering Strait bounds appear as:

"ocean_model_z", "volcello", "volcello", "ocean_Bering_Strait",    "all", "mean", "-171.4 -168.7  66.1 66.1 -1 -1",2
kshedstrom commented 1 year ago

Sure, my most recent run is in /lustre/f2/scratch/gfdl/Katherine.Hedstrom/work/fre/run23

nikizadehgfdl commented 1 year ago

I cannot access the ocean_hgrid.nc and probably the rest of the INPUT files to run your test:

 ncdump: /lustre/f2/scratch/gfdl/Katherine.Hedstrom/work/fre/Arctic_12k/INPUT/ocean_hgrid.nc: Permission denied
kshedstrom commented 1 year ago

Sorry, did I just fix it for you?

uramirez8707 commented 6 months ago

Fixed in #1505 Integer indexing for sub-regions is now available in the diag_table.yaml https://github.com/NOAA-GFDL/FMS/blob/main/diag_manager/diag_yaml_format.md#26-sub_region-section