Oct 2024 Variable Change Sprint #457
Conversation
|
General comment about what will be required in terms of getting the UFO test working -- we'll need to go over the variable name changes seen in the yaml; look for the geovals that changed names, and rename those variables in the geoval files of the corresponding instruments. This is a relatively simple task ... I might work on that later today. |
…ng - now it causes a crash
…... I wonder if this was the case in the original code, but had gone unnoticed - I am checking
|
So I am confirming that the problem I see w/ SSMI/S is already in develop: Quadratic cost function: JoJc( 0) = 2904.536994156287 So the changes introduced in this PR are not related to this issue. I will open an issue regarding this. |
|
At this stage, I say that all work to merge w/ Sprint (19Nov2024 version of JEDI) has been done for VAR. Work needs to be done for: SOCA and HOFX/UFO |
|
Can you share the November build you are using so I can also test with it? |
|
I have now a script that converts the variables in the geovals files according to the Sprint changes. I have triggered the UFO testing for all obs types ... except for aircraft and gps which seems to give me an unrelated error - all seems to work. I think I might be able to figure out the gps one ... the aircraft relates to a bias correction issue - not clear what's going on. The only thing I have not tested - cause I don't know how - is SOCA. But for atmosphere I now believe to have all needed changes. Somebody w/ super-powers will need to run the script I created over the geovals files stored in the original place. |
|
(Doruk) This is done now:
|
…rint_atm_sondes' into feature/rtodling/sprint_atm_sondes
… feature/rtodling/sprint_atm_sondes
|
@Dooruk why do we need all these requests for review? |
It's more of a heads up for everybody but we can go ahead and merge it |
|
I ran swell-convert_ncdiags-suite and the ncdiag file for aircraft that was generated still has the 'windUpward' name in the metadata and not the new 'instantaneousAltitudeRate'. I guess that converter needs an update as well. |
Sorry, SLES12 build wasn't updated to 11/19 but it is now 👍 |
Description
All this is using JEDI from 19Nov2024. So this is following JEDI build update to November 19th
Working to address changes from October 2024 Variable Change Sprint.
fields_metadata.yamlchanged for SOCAThese now work for all of the atmospheric obs systems:
However, in the SSMI case, I am seeing the Jo got down to negative numbers ... this means something is odd somewhere - bias correction, or something. I am not sure whether this was the case in the original code (before all these changes) ... I am currently checking on that.
SOCA notes:
There are major changes in terms of what internal variable names SOCA would be using and hence the configuration yamls needed to be updated. In terms of MOM6 outputs or obs files, there are no changes. See more details in JCSDA-internal/soca#1082. For instance instead of
tocnwe setsea_water_potential_temperature.It also conforms with JCSDA-internal/soca#1085,
BkgErrGODASis moved out to the generic block. Most of the default balance values were what we have been using but there is an additionalOceanSmootherthat we should be careful about.