Update JEDI to march 6 build#753
Conversation
|
@mranst can you share the comps plots for the 100 iteration case? |
|
|
Are we waiting for @rtodling's input here regarding the differences? |
Yes, though at this point should we target a later date? |
|
@mranst I think we should still merge this one, would be hard to track 4 months of differences. |
|
^ @Dooruk are these for the march build or february? |
Nope, it was February 🥲 Yikes, March 6th vs January 15th looks really different with incerements. Do you see similar results? I'm wondering if I made a mistake:
|
|
@Dooruk Huh, mine doesn't look that bad, this is for the 01152026_bufr and 03062026_bufr, not sure why that would make a difference though |
|
Ok, my march6 iteration was 100, others were 10. I will run it with 100. Sorry about that.. |
|
Yep, I'm getting the same results as yours with the same iterations, sorry again, I'm rusty. I will try to get @rtodling's eyes over these plots on Wednesday: Zonal wind 1000Pa:
|
|
Can you remind me, with comparison plots does EVA pick and plot only the observations that are actually different? |
No, it generates ALL plots. I can look into adding difference detection for obs |






This pr updates the pinned versions of JEDI to march 6, which in this build includes the addition of
bufr-query. See GEOS-ESM/jedi_bundle#114 for more details.This build does pass the tier1 tests, however there are differences in the residual norm for

3dfgat_atmos:I can share the full 100-iteration diff if necessary