Add ZTD bufr2ioda python tool and "obs space" yaml in rrfsv2-workflow

[HuiLiu-NOAA]

This PR adds a bufr2ioda python tool and an "obs space" of jedivar yaml for ZTD obs.

The ZTD bufr2ioda python tool converts RAP hourly gpsipw Bufr data into the ZTD IODA files for use in the RRFSv2 workflow.  The .json file is generic with analysis time only and should work for all bufr2ioda python converters.
The initial "obs space" part is for assimilation of ZTD observations using the existing UKMO local operator. More QC filters would be added later.

[SamuelDegelia-NOAA]

Could we add distribution and obs localizations sections to gnss_zenithTotalDelay.yaml? Those will be needed for LETKF. See here and here for examples.

[delippi]

Thanks for this, @HuiLiu-NOAA! It is my understanding that this PR is meant to get the yaml established in RDASApp and that there will be more updates to come. I think there are more QC filters that will need to go in, but this is a great starting point! I do have a few comments/suggestions for its current form.

• Can you refer to some of the validated yamls in RDASApp and follow that style? This is just my own personal preference, but I would like to see a single line (possibly at most two lines) comment before each filter. For example, you can call your step 1 # Pre-online regional domain check (latitude) and # Pre-online regional domain check (longitude) respectively.
• I believe we're going to stop (as of yesterday or so) naming any netcdf files with the .nc4 extension. We should just use .nc.
• The jdiag name will be renamed, but we need to hash that out with PR284.

[ShunLiu-NOAA]

@HuiLiu-NOAA It appears the goal of this PR is not clear enough. It is trying to solve a couple of issues in one PR. Let's discuss this and find a better way to commit these changes.

[HuiLiu-NOAA]

@HuiLiu-NOAA It appears the goal of this PR is not clear enough. It is trying to solve a couple of issues in one PR. Let's discuss this and find a better way to commit these changes.

Moved the slightly revised SATWND python tool to a new separated PR.

[delippi]

Are you planning to add more QC filters? I can think of a couple that you may want to try.

[HuiLiu-NOAA]

Are you planning to add more QC filters? I can think of a couple that you may want to try.

Yes, more QC filters will be added as the evaluation continues. Your suggestions are very welcome.

[delippi]

Are you planning to add more QC filters? I can think of a couple that you may want to try.

Yes, more QC filters will be added as the evaluation continues. Your suggestions are very welcome.

You don't have to add these right now since you're just getting the yaml established and you plan to do more testing, but here are a few of my thoughts. These are based on what I have in the conventional obs yamls which I think should still apply.

• What kind of QualityMarker info comes with this data? I see you reject all with QM > 10. For the other data we reject QM > 3.
• How is this data spaced in time? You might consider some time window filter that selects data only within say 15min of the analysis time.
• Duplicate checking
• Error inflation
• A sanity check to make sure the data falls within some expected range
• More robust gross error checks