Southern Ocean bottom water and sea ice in CMIP6 models

Southern Ocean bottom water and sea ice in CMIP6 models

This is a topic that Lettie Roach and I are proposing together at UW.

Scientific Motivation:

The circulation in the Southern Ocean is important for global climate, meridional overturning circulation, and stratification. Its capacity for heat uptake and its influence on global overturning make it an important driver for heat and carbon redistribution in the global ocean system. Climate models often poorly simulate bottom water formation in the Southern Ocean (see figure below), creating bottom water with densities that differ from climatology or getting the right amount of bottom water with the right properties for the wrong reasons. In the CMIP5 models, no models produced deep Antarctic water on the shelf and allowing it to spill over and convect, rather, models generate deep water through open ocean deep convection. Also, those models with strong sea ice seasonality have spurious deep convection (Huezé et al, 2013). Diagnosing the deep convection process and the bottom water formation in CMIP6 models is an important step in identifying models that can accurately simulate future Southern Ocean properties as well as future global climate changes.

The first goal of this project is to determine whether CMIP6 models form bottom water around Antarctica with properties and in quantities similar to observations as well as if they do it in the right way. The second goal is to determine whether deep convection impacts sea ice variability across models or vice versa and if there is an easily identifiable physical mechanism.

Proposed Hacking:

We hope to diagnose the Southern Ocean bottom water formation in CMIP6 models and the impact this formation has on overturning and transport in the region. We also hope to see whether there are links between bottom water formation and sea ice concentration and whether there are any mechanisms that link the two, perhaps through overturning or transport. The following proposed can serve as a guideline as to what we hope to build.

  • Determine the water mass properties in the abyssal ocean around Antarctica
  • Look at mixed layer depths in the Southern Ocean
  • Calculate residual mean overturning streamfunction in the Southern Ocean
  • Look at ACC transport in the Southern Ocean
  • Plot seasonal sea ice concentration

Anticipated Data Needs:

Variables: potential temperature (thetao), salinity (so), density (h_rho), velocity (uo and vo), mass transport (umo and vmo), eddy mass transport (uhGM and vhGM), mixed layer depth (mlotst), sea ice concentration (siconc), and sea ice thickness (sithick)

Experiments: piControl, historical

Anticipated Software Tools


Desired Collaborators

Anyone at all interested in the circulation in the southern ocean, sea ice near Antarctica, bottom water formation would be welcome. Experience with pangeo, python and/or xarray is a plus.


This is a great idea!! I think this project can have a lot of synergy with the project on Water masses of the future, and Plotting ocean variables in density coordinates (using xhistogram).

I am interested in participating in this.

If you are interested we could also try to extend the scope to the other water masses of the Southern Ocean too, as done in studies like
Downes et al 2015 (, Sallee et al 2013 (, and few other references within.

1 Like

Hi all,

I am working on wrapping an analysis across 26 CMIP6 models analyzing Southern Ocean circulation - providing a framework to understand why a model does / does not accurately represent the ACC structure. This is essentially an update to my paper where I did this in 31 CMIP5 models: to be able to assess whether models have improved / not across generations CMIP3 - CMIP6.

After seeing this project proposal this seems like this could potentially be a good collaboration. I do not look at bottom water properties in particular - or water mass properties either - just the large-scale density gradients across the current etc. However, something interesting to me is when looking at the long-term behavior of the ACC across the piControl and historical simulations - many models exhibit long-term (decadal to 100 yr) variability in the ACC - which I assume is linked to variability in dense bottom water formation via open-ocean convection near Antarctica.

I will be at the NCAR location - how can we collaborate?


I would be happy to join in and do a pre-hacking discussion today afternoon/evening, if others are interested? Get to know each other and figure out where different peoples strengths are, and what people anticipate as the main coding challenges that we need to overcome to get the analysis going.

1 Like

Becki, that sounds interesting and potentially could be a great collaboration. I’m not entirely sure how easy it will be to facilitate cross-location collaborating, but I think it would be really good to at least chat a few times, maybe via a project-specific slack channel in the hackathon workspace?

Druv, it would be great for you to join! You’ll be at the UW hackathon, right? I’m around this afternoon/evening, I can reach out to other interested folks as well to see if we can get a quick meeting set up. I met with Lettie and Brian Green last week (very briefly) but I think it would be useful to check in with each other again.

Yes, I’ll be at UW hackathon. I am around today afternoon/evening, let me know if something can be setup. Maybe around 3 or 3:30 (after LuAnne’s group meeting)? Or I could stop by your office and you could catch me up (might be easier).

I think for cross-location collaboration slack would be great, also we can use gists or github to share code and notebooks. Slack also has a video chat option that can be used from time to time.

Sure thing! How about I just catch you up after group meeting and then we can organize a meeting with other collaborators if we deem it necessary.

I’ve created a slack channel in the cmip6hackers workspace for this project, named #so-project. Becki, I added you to it, so hopefully it can help us communicate with you as you work on things at NCAR and we work on things over here at UW!

1 Like

This project is great!
I will attend the event at NCAR - so I’m not sure if it will be possible to join.
If it is, I’ll be happy to!

Hi @Ron-Maor, @BeckiBeadling will be working on this from NCAR too, so maybe you guys can work together there.

@Ron-Maor To add to what Dhruv said, you are welcome to join this project! Please join our slack channel in the cmip6hackers workspace (it is titled #so-project) to keep up with communications during the hackathon. There will be information in the channel on where to find our git repository and shared Google drive.