Add observation YAMLs for AMSU-A, ATMS, SSMIS, SEVIRI, and AVHRR

[emilyhcliu]

Add observation YAMLs for AMSU-A, ATMS, SSMIS, SEVIRI, and AVHRR
Pass benchmarks are provided in the YAMLs for unit tests.

Please see the following JEDI-T2O issues for documentation and validation results:
JEDI-T2O#42 --- AMSU-A (100% match)
JEDI-T2O#44 --- ATMS (Difference: 1 point for NPP and 5 points for N20)over south sea ice area from hydrometer check)
JEDI-T2O#55 --- SSMIS with Azadeh (100% match)
JEDI-T2O#53 --- AVHRR with Azadeh ( Difference: 1 point for N19)
JEDI-T2O#52 --- SEVIRI with Xuanli (100% match)

Summary plots for validation will be posted later.

[ADCollard]

@emilyhcliu Apparently the develop version was not pointing to the right UFO build after all. I just pushed a commit to update CMakeLists.txt to point at:

  ecbuild_bundle( PROJECT ufo   GIT "https://github.com/jcsda/ufo.git" BRANCH feature/gdasapp_ufo )

[ADCollard]

@emilyhcliu run_ufo_hofx_test.sh appears to still be pointing at Cory's data directories. Can you update, please, for both Hera and Orion (there are failures for AVHRR and Seviri).

[ADCollard]

@emilyhcliu I'll work on the final AMSU-A/ATMS QC in the morning. There will not be exact agreement, but we should be screening out obs with missing values anyway. We don't want to be pointing to a version of CRTM that is not under change control.

[ADCollard]

@emilyhcliu I'll work on the final AMSU-A/ATMS QC in the morning. There will not be exact agreement, but we should be screening out obs with missing values anyway. We don't want to be pointing to a version of CRTM that is not under change control.

.... actually I am going to see if I can make a modified CRTM tag....

[emilyhcliu]

@emilyhcliu run_ufo_hofx_test.sh appears to still be pointing at Cory's data directories. Can you update, please, for both Hera and Orion (there are failures for AVHRR and Seviri).

Yes

[ADCollard]

With the latest changes, I get the relevant tests to pass on Hera. I will look in detail at the summary output plots and will also test on Orion:

> export machine=hera
> export GDASApp=/scratch1/NCEPDEV/da/Andrew.Collard/git/GDASApp_Emily 
> ./test_yamls.sh 
aircraft.yaml Fails (boo!)
amsua_metop-a.yaml Passes (yay!)
amsua_metop-b.yaml Passes (yay!)
amsua_metop-c.yaml Passes (yay!)
amsua_n15.yaml Passes (yay!)
amsua_n18.yaml Passes (yay!)
amsua_n19.yaml Passes (yay!)
atms_n20.yaml Passes (yay!)
atms_npp.yaml Passes (yay!)
avhrr3_metop-a.yaml Passes (yay!)
avhrr3_metop-b.yaml Passes (yay!)
avhrr3_n18.yaml Passes (yay!)
avhrr3_n19.yaml Passes (yay!)
satwind.yaml Fails (boo!)
seviri_m08.yaml Passes (yay!)
seviri_m11.yaml Passes (yay!)
sfcship.yaml Fails (boo!)
sfc.yaml Fails (boo!)
sondes_ps.yaml Fails (boo!)
sondes.yaml Fails (boo!)
ssmis_f17.yaml Passes (yay!)
ssmis_f18.yaml Passes (yay!)

[CoryMartin-NOAA]

Did not test, but assuming it will be all good once @ADCollard finishes his tests. Thanks a lot @emilyhcliu for all of your hard work on this and getting it in GDASApp!

[ADCollard]

This IODA PR broke the eva scripts which meant I could not produce the output plots. A one line change to `ush/eva/gen_eva_obs.yaml' fixes this and was pushed here.

[ADCollard]

@emilyhcliu @CoryMartin-NOAA

I am happy with the PR now. I am including the plots that show the biggest discrepancies below, but I think these are all within reasonable bounds.

[ADCollard]

@CoryMartin-NOAA I approve this request but to be doubly sure (as I have been making multiple changes to this PR myself) can you do a clean checkout and build of this branch and run ush/ufoeval/test_yamls.sh as in this comment?

Note that there should be no need to change any files as long as the GDASApp and machine environment variables are set correctly.

[CoryMartin-NOAA]

note that I also had to set $STMP to get this to run but so far it says:

./test_yamls.sh 
aircraft.yaml Fails (boo!)
amsua_metop-a.yaml Passes (yay!)
amsua_metop-b.yaml Passes (yay!)
amsua_metop-c.yaml Passes (yay!)
amsua_n15.yaml Passes (yay!)
amsua_n18.yaml Passes (yay!)
amsua_n19.yaml Passes (yay!)
atms_n20.yaml Passes (yay!)

I'm impatient and going home now but I think this is sufficient to say it is ready to merge, thus I will do so :-)