Convert SSMIS from bufr to ioda

[XuLu-NOAA]

This is to document the bufr2ioda conversion for SSMIS:
Modifying the sample script from
https://github.com/JCSDA-internal/ioda-converters/blob/da03e61950cd522c2ca3afa3c45ebebd0ebb380b/test/testinput/bufr_ssmis.yaml

[XuLu-NOAA]

The initial test with GSI diagfile converted data works fine with the yaml modified from
https://github.com/JCSDA-internal/skylab/blob/2c3b7a65e1975aab62e4a35a8b387521c7b25604/obs/ufo/ssmis_f17.yaml
/scratch1/NCEPDEV/hwrf/scrub/Xu.Lu/SampleCase_SSMIS/jedi.yaml
The increment comparison between JEDI and GSI are here, it looks fine for now before we can tune any vertical localization:
https://docs.google.com/presentation/d/1-aSqes1FXG7QPHAGP4MnjYw26bBT4_O394rHJaFQAOI/edit#slide=id.g2d48c00eff3_0_0

However, modifying the bufr_ssmis.yaml to generate the ioda from bufr has experienced issues.
/work/noaa/hwrf/save/xulu/hafs_jedidual_inter/parm/analysis/jedi/yaml_templates/bufr2ioda/bufr_ncep_ssmisu.yaml
The sensorZenithAngle SAZA, solarZenithAngle SOZA, and solarAzimuthAngle SOLAZI are not contained in the original bufr data. But they are required by CRTM. If I manually assign 0 values there, JEDI will fail when obs fall in the DA domain.

In GSI, the SAZA is fixed to be 53.0, and SOZA/SOLAZI are calculated inline using a function called zensun based on only the lat/lon/time of the day.

[XuLu-NOAA]

Bufr-query is used to generate the IODA file using the bufr. A new issue is opened here for the purpose:
NOAA-EMC/bufr-query#51
With the converted obs, the DA now looks reasonable:

But the differences between GSI and JEDI is larger than those from the GSI diagfile:

Needs some further investigations.

[XuLu-NOAA]

Single observation test using GSI diag file shows the inconsistency between GSI and JEDI is likely due to the BC. The Hofx without BC is reasonably more consistent

After discussing with Emily, the differences in the BC is likely due to the CRTM version difference. Also, the all-sky capability in JEDI might also increase the difference as compared to the HAFS-GSI, which by default has no hydrometeor for all-sky. So I'll leave the differences as of now.