Revised-EVP option instead of legacy solver in OM3?

[adfraser]

@anton-seaice and I just had a discussion about this, and Anton suggested to document it.

Anton encountered a stability issue with a c-grid run, and I suggested to try the revised EVP option in CICE, from Bouillon et al 2013. It went some way toward fixing the instability.

Then I remembered @aekiss we discussed this with Tian for her OM2-based paper - she was encountering grid-aligned artefacts in OM2-01-derived divergence/convergence/shear maps, and you mentioned that revised EVP would have likely fixed this issue.

So - what are we thinking for OM3? @JFLemieux73 might have some nice insights here :) I think if it improves the grid-aligned artefacts, and possibly increases stability in weird geometry cases, it's worth considering, but I don't know enough about the drawbacks.

@dpath2o has also tried some fast ice runs with CICE6.4 and there were some minor changes to fast ice. I think Dan is favouring using revised EVP, right Dan?

[dpath2o]

Thanks @adfraser . Yes, my work using CICE6-standalone with revised-EVP bottom figure; top figure revised EVP disabled/default. Linear kinematic features are minimised without an effect on internal ice stress.

[JFLemieux73]

Hi @adfraser,

I am pretty sure the instability is not related to the EVP. I think it is related to the ridging scheme and to the explicit treatment of the ice strength in the momentum equation. I think the problem is also present with the B-grid but has been unnoticed (it is possible the B-grid is less sensitive to this issue). Using the Hibler formulation for the ice strength instead of Rothrock mitigates the problem. Have a look at this issue for details:

CICE-Consortium/CICE#992

[adfraser]

Thanks for the comment JF :)

While we are taking about revised-EVP though, what is your opinion on using it for our ACCESS-OM3? It will likely fix the grid-aligned DKPs we were seeing in OM2. Are many other groups using revised EVP?

[anton-seaice]

While we are taking about revised-EVP though, what is your opinion on using it for our ACCESS-OM3? It will likely fix the grid-aligned DKPs we were seeing in OM2. Are many other groups using revised EVP?

Summary from Elizabeth was along the lines of she thinks its probably a better solution but was implemented and tested initially in other models, and hasn't been tested / proven enough to be set as the default for CICE (but might be in the future if us/other modelling centres show it works well).

[JFLemieux73]

Hi @adfraser,

Sorry for the delay. It is true that revised-EVP has been less tested than standard EVP in CICE. We use the standard EVP in all our operational systems at ECCC. My suggestion would be to use the standard one, making sure you set ndte to a sufficiently high number (240 is ok for 0.25 degree but I would increase it for higher resolution).

Note that Koldunov et al. 2019 report interesting result with revised EVP in FESOM.

Koldunov et al., Fast EVP Solutions in a High-Resolution Sea Ice Model, JAMES, 2019.