Ship and Sondes PS analysis

[rtodling]

It seems to me there is something undesirable in the PS Ship and Sondes analysis. I see Jo(a) being larger than the Jo(b) ... this is illustrated more simply by looking at a 2-middle-loop case (for PS sondes); see fig below.

I have not done a GSI corresponding comparison at this point (9/25/25).

we should check to see if this gets addressed w/ the changes that come in here:

https://github.com/JCSDA-internal/ufo/pull/3779

[rtodling]

Turns out the issue above is an FGAT problem. FGAT is JEDI makes no sense: only the center background is updated.

[Dooruk]

The following PR below following JEDI changes introduce additional potential issues, in terms of relying model output versus observation sources like it was done before, (or vice versa) but I'm writing this as a reminder here:

#637 (comment)

[Dooruk]

More information from @gmao-msienkie

If the obs file dont have temperature JEDI will use background value. However the file should have a temperature field even if it's empty (or filled with missing values). That way code can look at background instead.

FYI @metdyn

[metdyn]

@Dooruk, @gmao-msienkie, Thanks for update. In my opinion, it is a bug in

• https://github.com/JCSDA-internal/ufo/pull/3779
  to require this temperature field in the file. If it is not bug, we need to have file changes to accommodate this PR. I will run this to see the issue.

I will check if @rtodling was testing UFO PR with his unmerged branch in

• https://github.com/JCSDA-internal/fv3-jedi/pull/1381