CDO

Hi Brian,

I have never done it, but did you try the command isosurface (see 2.3.10 selsurface)?

Best regards,

Estanislao

Hi Estanislao,

Thanks for the quick reply. I didn't know about the selsurface operators. :-)

But it unfortunately fails with:
@cdo isosurface (Abort): No processable variable found!

I though it was because my variables are 4D (time,height,lat,lon) and not 3D as the documentation states. But i removed the time dimension to test the idea, but it still fails with both bottomvalue and topvalue

I have uploaded two small sample files that can be used for testing. I can't see why it shouldn't work, at least with the one with only three dimensions (but that wouldn't solve my issue anyway).

cdo bottomvalue dimension_height_test.nc out.nc

cdo    bottomvalue (Abort): No processable variable found!

ncdump -h dimension_height_test.nc 
netcdf dimension_height_test {
dimensions:
    height = 1 ;
    lat = 51 ;
    lon = 51 ;
variables:
    double height(height) ;
        height:standard_name = "height" ;
        height:long_name = "height" ;
        height:units = "m" ;
        height:positive = "up" ;
        height:axis = "Z" ;
    double lat(lat) ;
        lat:standard_name = "latitude" ;
        lat:long_name = "latitude" ;
        lat:units = "degrees_north" ;
        lat:axis = "Y" ;
    double lon(lon) ;
        lon:standard_name = "longitude" ;
        lon:long_name = "longitude" ;
        lon:units = "degrees_east" ;
        lon:axis = "X" ;
    float sh(height, lat, lon) ;
        sh:standard_name = "specific_humidity" ;
        sh:long_name = "specific humidity" ;
        sh:units = "kg/kg" ;
        sh:param = "0.1.0" ;
        sh:_FillValue = -9999.f ;
        sh:missing_value = -9999.f ;
        sh:cell_methods = "time: mean" ;

Cheers,
Brian

Hi Brian,
I totally forgot about sellevel. That command should do the job.

Hi again,

That unfortunately doesn't remove the dimensions it just extracts them. That is part of the command we created the file to begin with.

cdo -O -s --pedantic \
      -f nc4 \
      ...
      ...
      ...
      -settunits,seconds \
      -setmissval,-9999 \
      -remap,${GRID},${WEIGHTS} \
      -merge \
      -selltype,102 -selvar,pres ${SRCFILE} \
      -selltype,103 -selvar,2t,q,10u,10v ${SRCFILE} \
      -sellevel,0 -selvar,cc,tp ${SRCFILE} \
      tmp1_${SRCFILE}.nc

/Brian

Hi Brian,

I think the best way is to do the following:

# display the time stamp of the file
cdo -s -showtimestamp dimension_time_height_test.nc

# reduce one element dimensions
cdo --reduce_dim -copy dimension_time_height_test.nc tmp.nc

# add the time dimension
cdo -setreftime,2024-02-01,00:00:00,1sec -settaxis,2024-02-02,19:00:00,1sec tmp.nc outfile.nc

Result:

dimensions:
    time = UNLIMITED ; // (1 currently)
    lon = 51 ;
    lat = 51 ;
variables:
    double time(time) ;
        time:standard_name = "time" ;
        time:units = "seconds since 2024-2-1 00:00:00" ;
        time:calendar = "proleptic_gregorian" ;
        time:axis = "T" ;
    double lon(lon) ;
        lon:standard_name = "longitude" ;
        lon:long_name = "longitude" ;
        lon:units = "degrees_east" ;
        lon:axis = "X" ;
    double lat(lat) ;
        lat:standard_name = "latitude" ;
        lat:long_name = "latitude" ;
        lat:units = "degrees_north" ;
        lat:axis = "Y" ;
    float sh(time, lat, lon) ;
        sh:standard_name = "specific_humidity" ;
        sh:long_name = "specific humidity" ;
        sh:units = "kg/kg" ;
        sh:param = "0.1.0" ;
        sh:_FillValue = -9999.f ;
        sh:missing_value = -9999.f ;

Hi Karin,

Thanks for the solution, but that sort of only comes halfway and increases the complexity (by having to extract the timestamp and split it with some bash commands and insert the axis again).

The current NCO commands does the job just fine, but I would like to be able to do it without using temporary files to reduce both CPU load and disk IO.
With the above commands I only get one temporary file instead of two, which is of course better than nothing. I haven't tested the performance difference yet.

I still think the option to just name the dimensions to reduce would be a nice feature.

But thanks for the quick replies and the different solutions.

/BRian

You can remove all vertical dimensions with size=1 with the CDO function setzaxis,surface.
surface is a predefined (undocumented) zaxis description for:

zaxistype = surface
size      = 1
levels    = 0 

Hi Uwe,

Awesome, it works exactly as I wanted. :-)

I just went down a rabbit hole to try and optimize the job as much as possible.

Thanks for the amazing support and great tool in general.
Brian

Ah, that is really great. The undocumented things are always cool. 👍