dimension order in example 6

[taylor13]

I just noticed that I messed up the ordering of the dimensions for example 6 (Appendix 3) of our model output specifications document. Assuming here the C convention for dimension ordering, the x dimension in the ncdump of the sample output should appear after the "y" dimension, not in the reverse order. So, for example,

hfls(time, y, x)
latitude(y, x)
vertices_longitude(y, x, vertices)
.
.
.

I hope @sol1105 can check that I'm correct about this before I proceed to correct the model output specifications document and before @mauzey1 modifies example 6 in the CMOR documentation.

[sol1105]

hfls(time, y, x)
latitude(y, x)
vertices_longitude(y, x, vertices)

Yes, that is how we expect it.

[durack1]

It seems CMOR 3.15.2 is being directed to flip the coordinates, the history provides the hint

hfls:history = "2026-08-07T00:46:52Z altered by CMOR: Reordered dimensions, original order: time y x." ;

Would be ideal to ascertain what is triggering the reordering, which is causing the problem

[mauzey1]

The issue is due to how cmor_set_grid_mapping gets the names of grid dimensions. For this specific example, it is getting the projection_x_coordinate and projection_y_coordinate for the lambert_conformal_conic grid mapping from the following.

cmor/Src/cmor_grids.c

Lines 460 to 469 in 33c8b0f

	} else if (strcmp(name, "lambert_conformal_conic") == 0) {
	*natt = 5;
	strcpy(att[0], "standard_parallel");
	strcpy(att[1], "longitude_of_central_meridian");
	strcpy(att[2], "latitude_of_projection_origin");
	strcpy(att[3], "false_easting");
	strcpy(att[4], "false_northing");
	*ndims = 2;
	strcpy(dims[0], "projection_x_coordinate");
	strcpy(dims[1], "projection_y_coordinate");

The ordering of the list of dimensions at the bottom is treated as the ordering of the dimensions for the file. This causes CMOR to flip the dimensions if they are in [y, x] order. These dimensions were added to CMOR in May 2025 to support CRS.

[taylor13]

Thanks @mauzey1 . It looks to me that when lines 468 and 469 were added last year (along with similar entries for a number of other types of grids), the order already used in defining the "mercator" grid was followed. I think the "mercator" ordering was wrong, however, so all of the the orderings of projection_x_coordinate and projection_y_coordinate are wrong. We should change the order to:

     strcpy(dims[0], "projection_y_coordinate"); 
     strcpy(dims[1], "projection_x_coordinate"); 

I'm not absolutely positive about this, but why not try making this change, and then I'll look at the ncdump and see if it fixes the problem. (I want to check the actual coordinate and variable values along with the order of the dimension names.)

[mauzey1]

Aside from projection_x_coordinate and projection_y_coordinate, there are also latitude and longitude for the label latitude_longitude.

cmor/Src/cmor_grids.c

Lines 479 to 483 in 33c8b0f

	} else if (strcmp(name, "latitude_longitude") == 0) {
	*natt = 0;
	*ndims = 2;
	strcpy(dims[0], "latitude");
	strcpy(dims[1], "longitude");

and grid_latitude and grid_longitude for the label latitude_longitude.

cmor/Src/cmor_grids.c

Lines 525 to 532 in 33c8b0f

	} else if (strcmp(name, "rotated_latitude_longitude") == 0) {
	*natt = 3;
	strcpy(att[0], "grid_north_pole_latitude");
	strcpy(att[1], "grid_north_pole_longitude");
	strcpy(att[2], "north_pole_grid_longitude");
	*ndims = 2;
	strcpy(dims[0], "grid_latitude");
	strcpy(dims[1], "grid_longitude");

Do we want the order of these reversed as well? They follow the same order as projection_x_coordinate and projection_y_coordinate.

[mauzey1]

Adding to the above post, the file generated by test_cmor_crs.py has the following.

        double latitude(lat, lon) ;
                latitude:standard_name = "latitude" ;
                latitude:long_name = "latitude" ;
                latitude:units = "degrees_north" ;
                latitude:missing_value = 1.e+20 ;
                latitude:_FillValue = 1.e+20 ;
        double longitude(lat, lon) ;
                longitude:standard_name = "longitude" ;
                longitude:long_name = "longitude" ;
                longitude:units = "degrees_east" ;
                longitude:missing_value = 1.e+20 ;
                longitude:_FillValue = 1.e+20 ;
        char type(strlen) ;
                type:long_name = "surface type" ;
                type:standard_name = "area_type" ;
        float baresoilFrac(time, lat, lon) ;
                baresoilFrac:standard_name = "area_fraction" ;
                baresoilFrac:long_name = "Bare Soil Percentage Area Coverage" ;
                baresoilFrac:comment = "no future, CMIP6_Lmon.json comment: Percentage of entire grid cell  that is covered by bare soil." ;
                baresoilFrac:units = "%" ;
                baresoilFrac:history = "no history 2026-09-01T23:16:30Z altered by CMOR: Converted units from \'\' to \'%\'. 2026-09-01T23:16:30Z altered by CMOR: Treated scalar dimension: \'type\'. 2026-09-01T23:16:30Z altered by CMOR: Converted type from \'d\' to \'f\'." ;
                baresoilFrac:cell_methods = "area: mean where land over all_area_types time: mean" ;
                baresoilFrac:cell_measures = "area: areacella" ;
                baresoilFrac:coordinates = "type latitude longitude" ;
                baresoilFrac:missing_value = 1.e+20f ;
                baresoilFrac:_FillValue = 1.e+20f ;
                baresoilFrac:grid_mapping = "crs" ;

Reversing the order of latitude and longitude in cmor_grids.c creates the following.

        double latitude(lon, lat) ;
                latitude:standard_name = "latitude" ;
                latitude:long_name = "latitude" ;
                latitude:units = "degrees_north" ;
                latitude:missing_value = 1.e+20 ;
                latitude:_FillValue = 1.e+20 ;
        double longitude(lon, lat) ;
                longitude:standard_name = "longitude" ;
                longitude:long_name = "longitude" ;
                longitude:units = "degrees_east" ;
                longitude:missing_value = 1.e+20 ;
                longitude:_FillValue = 1.e+20 ;
        char type(strlen) ;
                type:long_name = "surface type" ;
                type:standard_name = "area_type" ;
        float baresoilFrac(time, lon, lat) ;
                baresoilFrac:standard_name = "area_fraction" ;
                baresoilFrac:long_name = "Bare Soil Percentage Area Coverage" ;
                baresoilFrac:comment = "no future, CMIP6_Lmon.json comment: Percentage of entire grid cell  that is covered by bare soil." ;
                baresoilFrac:units = "%" ;
                baresoilFrac:history = "no history 2026-09-01T23:21:07Z altered by CMOR: Converted units from \'\' to \'%\'. 2026-09-01T23:21:07Z altered by CMOR: Treated scalar dimension: \'type\'. 2026-09-01T23:21:07Z altered by CMOR: Reordered dimensions, original order: time lat lon. 2026-09-01T23:21:07Z altered by CMOR: Converted type from \'d\' to \'f\'." ;
                baresoilFrac:cell_methods = "area: mean where land over all_area_types time: mean" ;
                baresoilFrac:cell_measures = "area: areacella" ;
                baresoilFrac:coordinates = "type latitude longitude" ;
                baresoilFrac:missing_value = 1.e+20f ;
                baresoilFrac:_FillValue = 1.e+20f ;
                baresoilFrac:grid_mapping = "crs" ;

[taylor13]

Not sure where test_cmor_crs.py came from, but I think there are a number of problems with output produced by it:

1. setting grid_mapping = "crs" is supposed to mean there is a variable in the file by the name of "crs" that provides multiple coordinate system properties. For a lonxlat coordinate system, this is needed only if you want to specify such things as the semi-major axis of the ellipsoid representing earth. In climate models that's not needed.
2. There is no reason to include longitude(lat, lon) and latitude(lat, lon) in this file because baresoil is a function of the cartesian coordinate lon and lat.

If this file were compliant with CF (and CMIP), we would prefer that the coordinates be ordered (time, lat, lon), not (time, lon, lat).

I'm not very familiar with the grid_mapping options, and I'm pretty sure that CMOR has not been thoroughly tested in this realm.

How can I help?

[taylor13]

Regarding "latitude_longitude" and "rotated_latitude_longitude", I'm guessing those also need reordering, but let's see what happens when you reorder projection_x_coordinate and projections_y_coordinate first.

[mauzey1]

Here's what I get from example 6 after doing the project coordinate order reversal.

netcdf hfls_tavg-u-hxy-u_mon_glb_g999_ACCESS-ESM1-6_amip_r1i1p1f1_197901-197902 {
dimensions:
        time = UNLIMITED ; // (2 currently)
        y = 3 ;
        x = 4 ;
        bnds = 2 ;
        vertices = 4 ;
variables:
        double time(time) ;
                time:bounds = "time_bnds" ;
                time:units = "days since 1979-01-01" ;
                time:calendar = "360_day" ;
                time:axis = "T" ;
                time:long_name = "Time Intervals" ;
                time:standard_name = "time" ;
        double time_bnds(time, bnds) ;
        double y(y) ;
                y:bounds = "y_bnds" ;
                y:units = "m" ;
                y:axis = "Y" ;
                y:long_name = "y coordinate of projection" ;
                y:standard_name = "projection_y_coordinate" ;
        double y_bnds(y, bnds) ;
        double x(x) ;
                x:bounds = "x_bnds" ;
                x:units = "m" ;
                x:axis = "X" ;
                x:long_name = "x coordinate of projection" ;
                x:standard_name = "projection_x_coordinate" ;
        double x_bnds(x, bnds) ;
        int lambert_conformal_conic ;
                lambert_conformal_conic:grid_mapping_name = "lambert_conformal_conic" ;
                lambert_conformal_conic:standard_parallel = -20., 20. ;
                lambert_conformal_conic:longitude_of_central_meridian = 175. ;
                lambert_conformal_conic:latitude_of_projection_origin = 13. ;
                lambert_conformal_conic:false_easting = 8. ;
                lambert_conformal_conic:false_northing = 0. ;
        double latitude(y, x) ;
                latitude:standard_name = "latitude" ;
                latitude:long_name = "latitude" ;
                latitude:units = "degrees_north" ;
                latitude:missing_value = 1.e+20 ;
                latitude:_FillValue = 1.e+20 ;
                latitude:bounds = "vertices_latitude" ;
        double longitude(y, x) ;
                longitude:standard_name = "longitude" ;
                longitude:long_name = "longitude" ;
                longitude:units = "degrees_east" ;
                longitude:missing_value = 1.e+20 ;
                longitude:_FillValue = 1.e+20 ;
                longitude:bounds = "vertices_longitude" ;
        double vertices_latitude(y, x, vertices) ;
                vertices_latitude:units = "degrees_north" ;
                vertices_latitude:missing_value = 1.e+20 ;
                vertices_latitude:_FillValue = 1.e+20 ;
        double vertices_longitude(y, x, vertices) ;
                vertices_longitude:units = "degrees_east" ;
                vertices_longitude:missing_value = 1.e+20 ;
                vertices_longitude:_FillValue = 1.e+20 ;
        float hfls(time, y, x) ;
                hfls:standard_name = "surface_upward_latent_heat_flux" ;
                hfls:long_name = "Surface Upward Latent Heat Flux" ;
                hfls:units = "W m-2" ;
                hfls:cell_methods = "area: time: mean" ;
                hfls:missing_value = 1.e+20f ;
                hfls:_FillValue = 1.e+20f ;
                hfls:cell_measures = "area: areacella" ;
                hfls:grid_mapping = "lambert_conformal_conic" ;
                hfls:coordinates = "latitude longitude" ;

[taylor13]

That looks right to me. So, if "latitude_longitude" and "rotated_latitude_longitude" are similar, we need to reverse those too. Thanks!

[taylor13]

To be sure, could you post the full ncdump of "hfls" (or point to it)? I'd lilke to actually look at the ordering of the values of some of these variables.

[mauzey1]

Here's the full dump

ncdump output/MIP-DRS7/CMIP7/CMIP/MOHC/ACCESS-ESM1-6/amip/r1i1p1f1/glb/mon/hfls/tavg-u-hxy-u/g999/v20260901/hfls_tavg-u-hxy-u_mon_glb_g999_ACCESS-ESM1-6_amip_r1i1p1f
1_197901-197902.nc 
netcdf hfls_tavg-u-hxy-u_mon_glb_g999_ACCESS-ESM1-6_amip_r1i1p1f1_197901-197902 {
dimensions:
        time = UNLIMITED ; // (2 currently)
        y = 3 ;
        x = 4 ;
        bnds = 2 ;
        vertices = 4 ;
variables:
        double time(time) ;
                time:bounds = "time_bnds" ;
                time:units = "days since 1979-01-01" ;
                time:calendar = "360_day" ;
                time:axis = "T" ;
                time:long_name = "Time Intervals" ;
                time:standard_name = "time" ;
        double time_bnds(time, bnds) ;
        double y(y) ;
                y:bounds = "y_bnds" ;
                y:units = "m" ;
                y:axis = "Y" ;
                y:long_name = "y coordinate of projection" ;
                y:standard_name = "projection_y_coordinate" ;
        double y_bnds(y, bnds) ;
        double x(x) ;
                x:bounds = "x_bnds" ;
                x:units = "m" ;
                x:axis = "X" ;
                x:long_name = "x coordinate of projection" ;
                x:standard_name = "projection_x_coordinate" ;
        double x_bnds(x, bnds) ;
        int lambert_conformal_conic ;
                lambert_conformal_conic:grid_mapping_name = "lambert_conformal_conic" ;
                lambert_conformal_conic:standard_parallel = -20., 20. ;
                lambert_conformal_conic:longitude_of_central_meridian = 175. ;
                lambert_conformal_conic:latitude_of_projection_origin = 13. ;
                lambert_conformal_conic:false_easting = 8. ;
                lambert_conformal_conic:false_northing = 0. ;
        double latitude(y, x) ;
                latitude:standard_name = "latitude" ;
                latitude:long_name = "latitude" ;
                latitude:units = "degrees_north" ;
                latitude:missing_value = 1.e+20 ;
                latitude:_FillValue = 1.e+20 ;
                latitude:bounds = "vertices_latitude" ;
        double longitude(y, x) ;
                longitude:standard_name = "longitude" ;
                longitude:long_name = "longitude" ;
                longitude:units = "degrees_east" ;
                longitude:missing_value = 1.e+20 ;
                longitude:_FillValue = 1.e+20 ;
                longitude:bounds = "vertices_longitude" ;
        double vertices_latitude(y, x, vertices) ;
                vertices_latitude:units = "degrees_north" ;
                vertices_latitude:missing_value = 1.e+20 ;
                vertices_latitude:_FillValue = 1.e+20 ;
        double vertices_longitude(y, x, vertices) ;
                vertices_longitude:units = "degrees_east" ;
                vertices_longitude:missing_value = 1.e+20 ;
                vertices_longitude:_FillValue = 1.e+20 ;
        float hfls(time, y, x) ;
                hfls:standard_name = "surface_upward_latent_heat_flux" ;
                hfls:long_name = "Surface Upward Latent Heat Flux" ;
                hfls:units = "W m-2" ;
                hfls:cell_methods = "area: time: mean" ;
                hfls:missing_value = 1.e+20f ;
                hfls:_FillValue = 1.e+20f ;
                hfls:cell_measures = "area: areacella" ;
                hfls:grid_mapping = "lambert_conformal_conic" ;
                hfls:coordinates = "latitude longitude" ;

// global attributes:
                :Conventions = "CF-1.12" ;
                :activity_id = "CMIP" ;
                :area_label = "u" ;
                :branded_variable = "hfls_tavg-u-hxy-u" ;
                :branding_suffix = "tavg-u-hxy-u" ;
                :creation_date = "2026-09-01T23:08:11Z" ;
                :data_specs_version = "MIP-DS7.1.0.0" ;
                :description = "Simulation of the climate of the recent past with prescribed sea surface temperatures and sea ice concentrations." ;
                :drs_specs = "MIP-DRS7" ;
                :experiment = "Simulation of the climate of the recent past with prescribed sea surface temperatures and sea ice concentrations." ;
                :experiment_id = "amip" ;
                :external_variables = "areacella" ;
                :forcing_index = "f1" ;
                :frequency = "mon" ;
                :grid_label = "g999" ;
                :history = "2026-09-01T23:08:11Z ; CMOR rewrote data to be consistent with CF-1.12 and CMIP7 data requirements." ;
                :horizontal_label = "hxy" ;
                :initialization_index = "i1" ;
                :institution = "Met Office Hadley Centre" ;
                :institution_id = "MOHC" ;
                :license_id = "CC-BY-4.0" ;
                :mip_era = "CMIP7" ;
                :nominal_resolution = "100 km" ;
                :physics_index = "p1" ;
                :product = "model-output" ;
                :realization_index = "r1" ;
                :realm = "atmos" ;
                :region = "glb" ;
                :source = "ACCESS-ESM1-6: aerosol: classic; atmosphere: um7-3; land-surface: cable3; ocean-biogeochemistry: wombatlite; ocean: mom5; sea-ice: cice5" ;
                :source_id = "ACCESS-ESM1-6" ;
                :table_info = "Name: CMIP7_atmos.json; Creation Date:(2026-07-21 13:16:24) MD5:28339aa355908b9331150e1536f5e384" ;
                :temporal_label = "tavg" ;
                :title = "ACCESS-ESM1-6 output prepared for CMIP7" ;
                :tracking_id = "hdl:21.14107/838992f0-f786-45fb-a9c4-84cdcf48148e" ;
                :variable_id = "hfls" ;
                :variant_label = "r1i1p1f1" ;
                :vertical_label = "u" ;
                :license = "CC-BY-4.0; CMIP7 data produced by MOHC is licensed under a Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0). Consult https://wcrp-cmip.github.io/cmip7-guidance/docs/CMIP7/Guidance_for_users/#2-terms-of-use-and-citations-requirements for terms of use governing CMIP7 output, including citation requirements and proper acknowledgment. The data producers and data providers make no warranty, either express or implied, including, but not limited to, warranties of merchantability and fitness for a particular purpose. All liabilities arising from the supply of the information (including any liability arising in negligence) are excluded to the fullest extent permitted by law." ;
                :cmor_version = "3.15.3" ;
data:

 time = 15.5, 45.5 ;

 time_bnds =
  0, 31,
  31, 60 ;

 y = 0, 10000, 20000 ;

 y_bnds =
  -5000, 5000,
  5000, 15000,
  15000, 25000 ;

 x = 0, 10000, 20000, 30000 ;

 x_bnds =
  -5000, 5000,
  5000, 15000,
  15000, 25000,
  25000, 35000 ;

 lambert_conformal_conic = _ ;

 latitude =
  10, 8, 6, 4,
  20, 18, 16, 14,
  30, 28, 26, 24 ;

 longitude =
  280, 290, 300, 310,
  282, 292, 302, 312,
  284, 294, 304, 314 ;

 vertices_latitude =
  5, 6, 15, 14,
  3, 4, 13, 12,
  1, 2, 11, 10,
  -1, 0, 9, 8,
  15, 16, 25, 24,
  13, 14, 23, 22,
  11, 12, 21, 20,
  9, 10, 19, 18,
  25, 26, 35, 34,
  23, 24, 33, 32,
  21, 22, 31, 30,
  19, 20, 29, 28 ;

 vertices_longitude =
  275, 285, 285, 275,
  285, 295, 295, 285,
  295, 305, 305, 295,
  305, 315, 315, 305,
  277, 287, 287, 277,
  287, 297, 297, 287,
  297, 307, 307, 297,
  307, 317, 317, 307,
  279, 289, 289, 279,
  289, 299, 299, 289,
  299, 309, 309, 299,
  309, 319, 319, 309 ;

 hfls =
  80, 82, 84, 86,
  88, 90, 92, 94,
  96, 98, 100, 102,
  81, 83, 85, 87,
  89, 91, 93, 95,
  97, 99, 101, 103 ;
}