Re: [netcdfgroup] how to extract data

• To: sh zhou <sh.maeam@xxxxxxxxx>
• Subject: Re: [netcdfgroup] how to extract data
• From: Russ Rew <russ@xxxxxxxxxxxxxxxx>
• Date: Mon, 18 May 2009 15:33:26 -0600

> I have a NetCDF file and the description is in accessary,   I use  " ncdump
> -v layer_absorber profile.nc > out.txt" to extract the data, but there are
> so many data extracted.
> For instance, The description of "layer_absorber" as following:
> #
>     double layer_absorber(n_profiles, n_absorbers, n_layers) ;
>                 layer_absorber:long_name = "Layer absorber" ;
>                 layer_absorber:units = "Variable (see Absorber_Units_ID)" ;
>                 layer_absorber:_FillValue = -999. ;
> #
> Could you  tell me how to extract the data of " layer_absorber(1,1,1:100) "
> ?

In addition to the ncview, NCL, or NCO solutions already mentioned, you
could just use ncdump options "-v" (variable) and "-f" (full annotation)
and pipe the output into grep.  For example:

  ncdump -v layer_absorber -f c profile.nc | grep "// layer_absorber(0,0,.*)"

will output an annotated list of values such as

    1.091546e-06,  // layer_absorber(0,0,0)
    1.054535e-06,  // layer_absorber(0,0,1)
    1.078923e-06,  // layer_absorber(0,0,2)
    1.016285e-06,  // layer_absorber(0,0,3)
  ...

If you really want only the indices in the range 1 to 100 for the last
dimension, a longer grep regular expression is required or use a pattern
matcher that knows arithmetic, such as awk.

To use 1-based Fortran column-major indices rather than 0-based C
row-major indices, instead use

  ncdump -v layer_absorber -f f profile.nc | grep "// 
layer_absorber([0-9]*,1,1)"

to get output such as

    1.091546e-06,  // layer_absorber(1,1,1)
    1.054535e-06,  // layer_absorber(2,1,1)
    1.078923e-06,  // layer_absorber(3,1,1)
    1.016285e-06,  // layer_absorber(4,1,1)
  ...

--Russ

• References:
  • [netcdfgroup] how to extract data
    • From: sh zhou