Upgrade Rttov13 + WRF changes

[lkugler]

Description:

• Mostly changes to allow the use of RTTOV13

Edit (@hkershaw-brown) WRF changes moved to https://github.com/NCAR/DART/tree/wrf-thm

• WRF temperature variable changed from T to THM in WRF's model_mod
• Allow interpolate to access the uppermost level of the WRF state.

Fixes issue

• WRF model_mod - Why not allow last level? #366
• Bug: clouds flipped up/down in RTTOV for IR/VIS #365
• RTTOV12->13 upgrade has no issue

Types of changes

• Bug fix (non-breaking change which fixes an issue): Bug: clouds flipped up/down in RTTOV for IR/VIS #365
• New feature (non-breaking change which adds functionality): RTTOV upgrade
• Breaking change (fix or feature that would cause existing functionality to not work as expected): RTTOV-SCATT is not tested! Code missing!
• Documentation update

Documentation changes needed?

• My change requires a change to the documentation.
• I have updated the documentation accordingly.

Tests

• rttov13 changes: comparison to my own script using the python interface of RTTOV
• WRF allow accessing the top level of the WRF state in interpolate: ./perfect_model_obs gave the expected results
• cange to THM as type_t variable: ./perfect_model_obs did not crash

Checklist for merging

• Updated changelog entry
• Documentation updated
• Version tag

Testing Datasets

• Dataset needed for testing has been provided:
  wrfout state, expected output in netcdf format, obs_seq.in

[lkugler]

Here I have some before-after plots comparing to my script that uses the python interface to RTTOV. I create observations close to every gridpoint of the model, DART computes the radiance from an interpolation of the state. python-RTTOV uses gridpoint values/columns as input, so the left is naturally smoother.

The images indicate that the changes make sense. For WV there is still a visible difference for some gridpoints.

Before-image VIS

After-image VIS

After-image WV

[nancycollins]

by the way, really nice catch on the inverted column(s).

[hkershaw-brown]

Hi Lukas,

This is great work to get the rttov13 interface up and running.

For the obs_def_rttov13_mod.f90, at the moment git is interpreting this as a git mv so
obs_def_rttov_mod.f90 no longer exists in this branch.

Because we have people using rttov 12, for this pull request I'd like to have

obs_def_rttov_mod.f90
obs_def_rttov13_mod.f90

I think this makes clearer if we do depreciate rttov12 in a future release.

The mkmf.template can have a VERSION variable so we do not need two mkmf.template.rttov for each compiler:

e.g.

RTTOV_VERSION = 13

RTLIBS = -lrttov$(RTTOV_VERSION)_wrapper

I can go ahead and make these two changes if you are comfortable with me editing your branch.

Cheers,
Helen

[hkershaw-brown]

note, T vs THM is a wrf v3 to v4 issue. T is prognostic in v4.

Inside WRF, use_theta_m=0 for T= THM

[lkugler]

You mean THM is prognostic in v4. The WRF userguide says

The prognostic variable for temperature is moist potential temperature by default.

And if you set use_theta_m=0 then THM is the dry potential temperature and happens to be the same as T.
But if you update T by assimilation then this update is not used by the forecast model since only THM is prognostic but not T.

[lkugler]

We can also remove the commit related to the levels (25efb6c) from this PR?! Since its a separate thing.
Following the discussion in #366

[hkershaw-brown]

I'm splitting the WRF and RTTOV parts of this pull request.

The WRF changes are on https://github.com/NCAR/DART/tree/wrf-thm

[hkershaw-brown]

Approved.

Splitting the WRF updates to a separate pull request to support wrf3 and wrf4 users.