question about e_yieldcurve and e_plasticpot

[Jieshun Zhu]

When running CICE6 with default values (=2) for e_yieldcurve and e_plasticpot, I got too thick in the Beaufort Sea and too thin near the North Pole. I happened to see the paper of Miller et al. (2005; doi:10.1029/2005GL023622), and found that they got the similar bias as mine and their ice thickness simulations were significantly improved by decreasing the ratio of the elliptical yield curve's major to minor axes (e) from 2 to sqrt(0.5).

However, by carefully comparing the CICE6 document with Hibler (1979), it looks like e_yieldcurve is different from e. Could any CICE developer please help verify my following understanding?

1) The e in Hibler (1979) is equivalent to e_plasticpot**2/e_yieldcurve in CICE6, right?

2) Both e in Hibler (1979) and e_yieldcurve in the CICE6 document are described as "the ratio of the major and minor axes of the elliptical yield curve". It is clearly opposite to my understanding about 1). So my 1) is wrong?

Thanks for your help in advance!

 

[dbailey]

Good questions. We are all involved in the CICE Consortium meeting, but will look at this next week.

 

[dbailey]

@Philippe Blain can you answer this?

 

[JF Lemieux]

Hi Jieshun,

The plastic potential is a new feature. I am testing it right now. For the moment I suggest you stick with the standard VP rheology. If you set e_plasticpot = e_yieldcurve = 2 you get the standard elliptical yield curve of Hibler 1979.

Having too thick ice in the Beaufort and too thin close to the North pole is a common problem in sea ice models. Playing with the ellipse could help but these biases could also be related to the atmospheric forcing you are using.

I would not use e=0.5 (i.e, e_plasticpot = e_yieldcurve = 0.5). It is not realistic to have a shear strength larger than the compressive strength. However there are many papers that show that a smaller e (but e larger than 1.0) improves the simulation of landfast ice (Dumont et al. 2009, Lemieux et al. 2016) and of linear kinematic features (e.g. Bouchat et al. 2022).

I would suggest a value of 1.4 or 1.5.

So basically you just have to set e_plasticpot = e_yieldcurve = 1.4.

Note that a smaller e would reduce the ice velocity. You might have to slightly modify the air drag coefficient.

I hope this helps!

Jean-Francois

 

[Jieshun Zhu]

Hi Jean-Francois,

Thanks for your reply.

From your reply, it looks we usually set "e_plasticpot = e_yieldcurve". But from Ringeisen et al. (2021; Non-normal flow rules affect fracture angles in sea ice viscous–plastic rheologies), I read "for e_plasticpot > e_yieldcurve, the absolute value of the divergence is smaller and the shear strain rate is larger compared to a normal flow rule (e_plasticpot=e_yieldcurve)".

• Does e_plasticpot vs. e_yieldcurve control shear strength vs. compressive strength?
• or the absolute value controls the difference between two types of strength?

Thanks again,
Jieshun

 

[JF Lemieux]

Hi Jieshun,

e_yieldcurve controls the shear strength vs. compressive strength. e_plasticpot affects the shape of the curve for the flow rule (the plastic potential). Again, as this quite new, I suggest you simply set e_plasticpot = e_yieldcurve for the moment. I am doing simulations right now with e_plasticpot not equal to e_yieldcurve and plan to write a paper about this. I could keep you informed about the results.

Best regards,

JF

 

[Jieshun Zhu]

Thanks, Jean-Francois. I will look forward to your results.