Can ncwa (NCO) understand missing_value

Question

I'm trying to use NCO's ncwa script to find out ranges for some data variables in netcdf files. However it looks like it cannot decode missing_value attribute by default and that is used to define the results.

For instance, I am trying to run:

ncwa -y min -v hs ./test.nc /tmp/min.nc 

to find out the minimum value for variable hs in file ./test.nc. The output is -9 as shown below which is the missing_value attribute of that variable.

I tried with different files with different missing_value values and this is still the case. However the _FillValue attribute does seem to be decoded correctly by ncwa when present, with associated values being ignored in the results.

Any help would be appreciated. Thanks.

% ncdump /tmp/min.nc
netcdf min {
variables:
    float hs ;
        hs:standard_name = "sea_surface_wave_significant_height" ;
        hs:units = "m" ;
        hs:long_name = "Significant wave height" ;
        hs:missing_value = -9.f ;
        hs:cell_methods = "time, lat, lon: minimum" ;
data:
    hs = -9 ;
}

% ncwa --version
ncwa 4.6.9

Answer 1

For reasons described here, NCO pays attention to _FillValue not missing_value. The recommended solution is to add a _FillValue attribute with the same value as the missing_value attribute:

ncatted -O -a _FillValue,,o,f,1.0e36 inout.nc

or to rename the missing_value to _FillValue:

ncrename -a .missing_value,_FillValue inout.nc