NetCDF File with Rotated Lat/Lon Grid Causing Error

[balbrightwpc]

Hello,

I am trying to run pcp_combine on some experimental West-WRF runs we received from a partner that are in netCDF format. When running the command, I receive the following error:

ERROR :
ERROR : NcCfFile::get_grid_mapping_rotated_latitude_longitude() -> Rotated latitude longitude grid not handled in MET.
ERROR :

Here are the ncdump details of one of the netCDF files that are being used (I did not include all variables to save space):

ncdump -h west-wrf_gfs_d02_2023-01-09_12_00_00.nc
netcdf west-wrf_gfs_d02_2023-01-09_12_00_00 {
dimensions:
time = UNLIMITED ; // (1 currently)
west_east = 717 ;
south_north = 753 ;
variables:
double time(time) ;
time:standard_name = "time" ;
time:long_name = "Time" ;
time:units = "hours since 1970-01-01 00:00:00" ;
time:calendar = "standard" ;
time:axis = "T" ;
float lon(south_north, west_east) ;
lon:standard_name = "longitude" ;
lon:long_name = "Longitude" ;
lon:units = "degrees_east" ;
lon:_CoordinateAxisType = "Lon" ;
float lat(south_north, west_east) ;
lat:standard_name = "latitude" ;
lat:long_name = "Latitude" ;
lat:units = "degrees_north" ;
lat:_CoordinateAxisType = "Lat" ;
float west_east(west_east) ;
west_east:standard_name = "projection_x_coordinate" ;
west_east:long_name = "x coordinate of projection" ;
west_east:units = "degrees_east" ;
west_east:axis = "X" ;
float south_north(south_north) ;
south_north:standard_name = "projection_y_coordinate" ;
south_north:long_name = "y coordinate of projection" ;
south_north:units = "degrees_north" ;
south_north:axis = "Y" ;
int Rotated_Pole ;
Rotated_Pole:grid_north_pole_longitude = 180.f ;
Rotated_Pole:grid_north_pole_latitude = 51.f ;
Rotated_Pole:grid_mapping_name = "rotated_latitude_longitude" ;
float precip(time, south_north, west_east) ;
precip:standard_name = "total_precipitation_amount" ;
precip:long_name = "Accumulated Total Precipitation Over Simulation" ;
precip:units = "mm" ;
precip:grid_mapping = "Rotated_Pole" ;
precip:coordinates = "lat lon" ;
precip:_FillValue = 1.e+20f ;
precip:missing_value = 1.e+20f ;
// global attributes:
:CDI = "Climate Data Interface version 2.3.0 (https://mpimet.mpg.de/cdi)" ;
:_NCProperties = "version=2,netcdf=4.7.4,hdf5=1.10.7," ;
:Conventions = "CF 1.6, Standard Name Table v19" ;
:source = "/expanse/lustre/scratch/cw3eprod/temp_project/NRT_gfs_backfill/output/2023010900/wrf_2dom/wrfout_d02_2023-01-09_12_00_00" ;
:institution = "CW3E - Scripps Institution of Oceanography" ;
:title = "/expanse/lustre/scratch/cw3eprod/temp_project/NRT_gfs_backfill/output/2023010900/cf/wrfcf_gfs_d02_2023-01-09_12_00_00_temp.nc" ;
:notes = "Created with NCL script: wrfout_to_cf.ncl v2.0" ;
:created_by = "Daniel Steinhoff - [email protected]" ;
:creation_date = "Mon Oct 7 19:07:48 PDT 2024" ;
:history = "Fri Oct 11 11:47:57 2024: cdo -w -L -O merge temp1.nc temp2.nc temp3.nc temp4.nc temp5.nc temp6.nc /expanse/lustre/scratch/cw3eprod/temp_project/NRT_gfs_backfill/output/2023010900/west-wrf_2023010900/west-wrf_gfs_d02_2023-01-09_12_00_00.nc\nFri Oct 11 11:47:34 2024: cdo -w -L -O chname,Z_p,Z500 -setzaxis,surface -sellevel,500 -selname,Z_p wrfcf_gfs_d02_2023-01-09_12_00_00.nc temp1.nc" ;
:CDO = "Climate Data Operators version 2.3.0 (https://mpimet.mpg.de/cdo)" ;
}

Do you have any advice on how to handle this file? If you need any other information or if you need me to upload one of the files, please let me know. Any help would be appreciated, thank you!

Ben Albright

[CPKalb]

Hi Ben:

What version of MET are you using? Support for rotated lat/lon grids was added in MET version 10.1.1. So, the first thing to check is to make sure you are running MET version 10.1.1 or higher.

Assuming you are running 10.1.1 or higher, would you be able to upload a file using the instructions so we can take a closer look?

Thanks,
Christina

[CPKalb]

There are 1d lat/lon variables in the file. In this case, they are called west_east and north_south. I was originally confused since the long names say projection_x_coordinate and projection_y_coordinate. But, looking at the values, I think they are latitude and longitude.

[JohnHalleyGotway]

I was able to make some more progress... but issues still remain.

Based on this line of nccf_file.cc, I see that we're looking for NetCDF variables whose standard_name attribute is set to grid_latitude and grid_longitude when reading the grid information. So I used ncatted to modify the standard_name attributes of the west_east and south_north variables in the sample file:

ncatted \
-a standard_name,west_east,o,c,"grid_longitude" \
-a standard_name,south_north,o,c,"grid_latitude" \
west-wrf_gfs_d02_2023-01-09_12_00_00.nc \
west-wrf_gfs_d02_2023-01-09_12_00_00_MOD.nc

Plotting that does run, but the plot includes no background map data, which usually means that it's not in the correct spot on earth:

plot_data_plane west-wrf_gfs_d02_2023-01-09_12_00_00_MOD.nc LandMask.ps 'name="LandMask"; level="(0,*,*)";' -v 4
DEBUG 4: Grid information:
DEBUG 4:    Projection: Rotated Lat/Lon
DEBUG 4:    Nx: 717
DEBUG 4:    Ny: 753
DEBUG 4:    rot_lat_ll: 30.343
DEBUG 4:    rot_lon_ll: 133.155
DEBUG 4:    delta_rot_lat: 0.027
DEBUG 4:    delta_rot_lon: 0.030
DEBUG 4:    wrapLon: false
DEBUG 4:    true_lat_south_pole: -51.000
DEBUG 4:    true_lon_south_pole: 0.000
DEBUG 4:    aux_rotation: 0.000

Here's the image:

Then I ran regrid_data_plane to regrid this to a global grid (NCEP Grid 4) and plotted it to see where MET thinks this data lives:

regrid_data_plane west-wrf_gfs_d02_2023-01-09_12_00_00_MOD.nc G004 LandMask_G004.nc  -field 'name="LandMask"; level="(0,*,*)";'
plot_data_plane LandMask_G004.nc LandMask_G004.ps 'name="LandMask"; level="(*,*)";'

So data the should clearly be over the western United States is in the middle of the Pacific Ocean. Clearly, there's a problem. However I don't know exactly what it is. Is it a problem with the projection information itself or is it a bug in MET's processing of that grid information?

[balbrightwpc]

Hi John,

Appreciate the reply and you continuing to look into the issue. I ran into the same issue when trying to plot the data as well after renaming the data. I believe one of my co-workers was able to plot the raw data correctly. I have asked him to either post here or relay to me what he did so stay tuned.

[tomerburg]

Hi all,

The metadata in these files are different than files I normally work with - the 1D coordinates "east_west" and "south_north" are technically in degrees longitude and latitude, respectively, but don't actually correspond to the coordinates of the data. That would be the "lon" and "lat" 2D arrays. When I plotted it with Python/Cartopy, using the "lon" and "lat" arrays as the x and y coordinates, I was able to get the data correctly plotted:

I don't have enough MET expertise to know whether it's a bug on MET's side or with how the data is provided into MET, but hopefully this is enough information to guide you in the right direction.

[CPKalb]

@JohnHalleyGotway, correct me if I'm wrong, but I think MET needs the latitude and longitude for rotated lat/lon grids to be 1d variables. However, from the previous post, those variables don't correspond to the actual data in the file. I ran some testing on this. Specifically, I edited the netCDF file so that the dimensions are lat and lon (rather than west_east and south_north). I also edited the lat and lon variables to have standard name attributes of grid_latitude and grid_longitude. However, the file still does not go through MET. I get the following error:

ERROR  : 
ERROR  : NcCfFile::get_grid_mapping_rotated_latitude_longitude() -> Coordinate variables don't match dimension sizes in netCDF file.
ERROR  : 

I'm guessing you will need to read this file in using a python embedding script, as I don't think it can be read into MET in the format it is currently in.

Tina