Concatenation of ROPP files

[Ian Culverwell]

We should investigate the use of ncks to glue ROPP files together (should DMI want to include T in their BUFR output, for example).

[Ian Culverwell]

Example solution (thanks to Chris Burrows for the data)

We seek to transplant the observed refractivity into the 1dVAR output, which by default contains the refractivity obtained by forward modelling the retrieved {T, q, p}.

This is how to do it.

OBS=atm20110912_000312_M02_1230427200_N0019_XXXX.nc # input file, containing refractivity derived from bending angle

ANL=1dvar_sol.nc # 1dvar solution, containing forward-modelled refractivity based on 1DVAR retrieval.

OUT=1dvar_sol_with_obsN.nc # desired output file

ncks -a -x -vrefrac $ANL -O $OUT # extract all but refrac from the retrieval file

ncks -vrefrac $OBS -A $OUT # append observed refractivity to the output file.

That's it. The 3 files are attached. As a check:

ncks -H -F -ddim_lev2a,1 -vrefrac -Q $OBS

refrac(1)=151.03

ncks -H -F -ddim_lev2a,1 -vrefrac -Q $ANL

refrac(1)=151.812574607

ncks -H F -ddim_lev2a,1 -vrefrac -Q $OUT

refrac(1)=151.03

This shows that the observed refractivity has been transplanted into the output file.

[Ian Culverwell]

Alternatively, we could start transplant the 1dvar retrieval ({T, q, p}) into the observation file, and then convert that into BUFR, as DMI would do in practice. This procedure is carried out in the attached transplant.sh script. The nc and bufr output from this script are also attached, as are the "cdl" and "dh" summaries of the contents of the original and transplanted obs files.

[Ian Culverwell]

Needs passing by DMI customers to see if it does what they want.

[Ian Culverwell]

No response from DMI after asking for feedback twice, so closing ticket.