Standalone, regional CICE run

[Mks]

Hello everyone,
I'd like to run CICE (last version) standalone at a rather fine resolution in the Ross sea as a test experiment. Even though the documentation states that CICE can be run on regional grid, I've been unable to configure it in such a way so far: the regional namelist option seems to be linked to the popgrid subroutine, yet the provided input data include global grids only. Is there some sort of POP grids dictionary somewhere, or a way to generate said grids? As for the resolution, I read in the doc that the CICE driver supports gx1, gx3 and tx1 grids. In the CESM CICE (5) doc, tx0.1 is also mentioned: has support for it been dropped in CICE6?
I had a quick glance at the rectangular grid type. Spacing can be set as a namelist option, but not the edges of the domain which seem to be hard-coded. My understanding is that it works with the ICE_GRID variable limiting the options to a 80x80 or a 128x128 domain, is that correct? Resolution would then be determined by the spacing. Would that be a way to address my problem?

 

[eclare@lanl_gov]

Hello Mks,
The CICE model was designed to be used on global grids and so most of the configurations are global. gx1 and gx3 grids are global with the north pole rotated into Greenland; tx1 and tx0.1 are also global but "tripole" (which isn't really 3 poles but rather one regular pole and one pole stretched between two points in the northern hemisphere). In the code, we support grids for testing within the CICE Consortium, and let users design and implement their own, otherwise. For instance, the tx0.1 grid that CESM uses is a high-resolution version of the tripole grid, and we do not need to test it in the Consortium as part of our processes, so we do not offer it or forcing data for it. The RASM project runs CICE on a regional Arctic grid in a fully coupled model configuration, and they have a lot of infrastructure to be able to do that, including things like boundary forcing for the ice grid. Likewise, the rectangular grid option is there for testing purposes, but you could configure it for your case. We have code "hooks" (i.e. very basic capability) for restoring along the boundaries, but the restoring capability needs to be exercised and refined in real problems. You are free to change the code for different grid sizes and grid spacing, or to use altogether different grids! One thing that has come up in recent discussions is a desire for better support for regional configurations, so this is something that the CICE Consortium will look into doing in the future.

 

[Mks]

Thanks for the detailed reply! I broadly came to these conclusions. For the moment I will conduct my tests globally, and I'll see how to work around it in the future.
Cheers

 

[dpath2o]

Hi Elizabeth,

I'm wanting to investigate fast ice dynamics of Antarctica through the lens of CICE6. To do this my thought is to run a regional standalone version of CICE6 and then begin to look at building off the work, in broad terms, here ( https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/2016JC012006 ). I'm very much in the nascent phase of this sandbox having just downloaded CICE6 and successfully building and the running the three grid types. So I'm now looking at how I might 'regionalise'. I realise this not a trivial problem, but possibly in the past couple of years since this post/thread was left, there are tools that I'm not seeing from the GitHub release or elsewhere, that I could be pointed towards.

Thanks in advance.

Cheers,
Dan

 

[eclare@lanl_gov]

Hi Dan,
Setting up a stand-alone, regional configuration is on our to-do list, but we have not done anything yet. As mentioned above, there are "hooks" available in the code but it will still take some work to set up. We generally do not recommend using the code (other than testing) in any stand-alone configuration since the coupled feedbacks with the atmosphere and ocean are extremely important, and so offering a regional configuration has not been high on the priority list. If you make progress on it, we'd be happy to help you to share it with the rest of the CICE modeling community!
e

 

[dpath2o]

Thanks Elizabeth. I'll consult with my supervisors, but I'm keen to give it a red hot go.