Generalise outputting from ROPP PP

[Ian Culverwell]

Stig Syndergaard (DMI) reminds us, in connection with Action 3 of the PCR-RE1 review held in April 2016, that:

I haven't seen the final formulation of the recommendations and 
action (Kent did not forward it to me (yet)), but I think it's only 
related to a much older issue and change that we did here: 
https://trac.romsaf.org/ropp/changeset/3308/ropp_src/branches
/dev/Share/dmi_trunk_6.0/ropp_pp/config (sec. 0 of the config 
files).

I think at the time you preferred not to make that particular 
change in ROPP, and I didn't bug you about it again. But when Axel 
at the review saw the config file at the end of the ATBD (see other 
email), he couldn't see that Tdry was output. Thus the issue came 
up that we have different config files in the dmi_trunk. I don't 
think the recommendation is that the values of the parameters 
should be the same, although that was also discussed, but I think 
it was dismissed.

Of course, there are code changes to go along with the config file 
change, I'll try dig them up when I get back.

So we should probably try to do this for ROPP9.0.

[Ian Culverwell]

Stig has provided the following details.

The following is taken from our last merge of the 
dmi_trunk with ROPP 8.1, where we copied tags/8.1 to 
dmi_trunk_8.1 and then merged everything from our 
dmi_trunk_8.0 into that.

Necessary changes to ropp_pp/config files can be found 
here (only the first block of changes in section 0, the 
rest is related to BAROCLIM etc):

https://trac.romsaf.org/ropp/changeset/4735/ropp_src/branches/dev/Share/dmi_trunk_8.1/ropp_pp/config

Corresponding changes to ropp_pp_read_config.f90:

https://trac.romsaf.org/ropp/changeset/4735/ropp_src/branches/dev/Share/dmi_trunk_8.1/ropp_pp/common/ropp_pp_read_config.f90

Corresponding changes to ropp_pp_types.f90:

https://trac.romsaf.org/ropp/changeset/4735/ropp_src/branches/dev/Share/dmi_trunk_8.1/ropp_pp/common/ropp_pp_types.f90

And sections 13-14 in ropp_pp_invert_tool.f90:

https://trac.romsaf.org/ropp/changeset/4735/ropp_src/branches/dev/Share/dmi_trunk_8.1/ropp_pp/tools/ropp_pp_invert_tool.f90

And sections 21-22 in ropp_pp_occ_tool.f90:

https://trac.romsaf.org/ropp/changeset/4735/ropp_src/branches/dev/Share/dmi_trunk_8.1/ropp_pp/tools/ropp_pp_occ_tool.f90

As you see there are many other changes in 
ropp_pp_invert_tool.f90 and ropp_pp_occ_tool.f90, but I 
think you have most of them already (related to 
BAROCLIM and new sf_method). Some newer ones are from 
our combination of ropp_pp_invert_tool.f90 and 
ropp_pp_occ_tool.f90 (I don't think I forwarded that 
one to you yet).

We have made a number of changes to the ropp_pp config 
files since that merge and added a few more files. If 
you want exactly the files we have in our dmi_trunk 
(for ROPP 9 or 10; which I wouldn't mind, but it's not 
something we decided should be), then you can take them 
from here:

https://trac.romsaf.org/ropp/browser/ropp_src/branches/dev/Share/dmi_trunk_8.1/ropp_pp/config

The parameters in these files are the ones we plan to 
use in our reprocessing.

Let me know if you have any questions.

[Ian Culverwell]

The main difference, then, is a change in the output choice from

output_tdry = .true.        ! Flag to output dry temperature 

to

output_lev1b = .true.       # Flag to output level 1b data 

output_lev2a = .true.       # Flag to output level 2a data

The removal of output_tdry is likely to be more problematic than the inclusion of output_lev1b and output_lev2a. There is also an overhead in updating the documentation. On the other hand, we have been mandated to do it by the PCR-RE1 (​https://trac.romsaf.org/wiki/ticket/194).

[Ian Culverwell]

test_occ.sh

[Ian Culverwell]

makecf_occ.sh

[Ian Culverwell]

test_invert.sh

[Ian Culverwell]

makecf_invert.sh

[Ian Culverwell]

After much toing and froing, too tiresome to document here, Stig and I agreed on the version of the changes implemented at r5106. The attached test scripts check that all 2⁴=16 combinations of the four output options {output_lev1a, output_lev1b, output_lev2a, output_diag} are doing the right thing.

Tentatively, submit to the test folder ...

[Ian Culverwell]

Works OK, so close ticket.