Turn shuffle option on in soca -> cice.

[shlyaeva]

This turns on shuffle option in soca -> cice and changes seaice edge to 40%.
For all background ice concentrations < seaice edge when shuffle is on, the code finds a nearest background point whose total ice concentration is closest to the soca analysis total ice concentration, and copies that background ice profile as the new analysis at the current point. For all background ice concentrations > seaice edge ice concentrations are rescaled, and some ice fields recomputed.

Some results from a first cycle of high res experiment:
Background:

soca analysis (no postprocessing; I am cutting off values outside of [0, 1] when plotting):

analysis after Soca2Cice on develop (where shuffle is off, and seaice edge for the arctic is 0.8) As expected, the analysis looks like background for ice concentrations < 0.8

analysis after Soca2Cice in this branch (shuffle is on, seaice edge is set to 0.4). Note that these results are with JCSDA-internal/soca#1108

This branch can be merged in develop regardless of JCSDA-internal/soca#1108, since gdasapp soca hash already includes the bugfix that allows shuffle to be turned on. However, I would recommend running experiments that include JCSDA-internal/soca#1108 which further improves on shuffle.

I ran C384mx025_3DVarAOWCDA for 2 cycles with this setup, and forecasts succeeded (cice was OK with these analyses). We'll do better evaluation when we run the next marine candidate experiment.

[guillaumevernieres]

🎉

[shlyaeva]

FYI to repo maintainers: It looks like I have the rights to merge here, should I have those rights?

[guillaumevernieres]

FYI to repo maintainers: It looks like I have the rights to merge here, should I have those rights?

Yes, for sure you should too!