Model mixing

Back from the cruise, back to the normal life. For a researcher, normal life means spending a while thinking about a problem, and writing about it when you have found a solution, or at least something interesting to say. You then send your text, and wait while two to three experts in your topic read it, comment on it, and send the comments to you so that you can improve the text (that is the peer-review process). And when you receive the email saying “Dear Dr Heuzé, We are pleased to inform you that your following manuscript was accepted…”, you can afford a little victory dance in the office. And talk about this paper to advertise it. So, dear reader, how much do you know about Southern Ocean deep convection in climate models?

1) Wrong processes in the models

When satellites started observing Antarctica in the late 1970s, they saw a large hole in the sea ice: the Weddell Polynya. For three years, even in the middle of winter, a large area of the Weddell Sea (the one south of the Atlantic) stayed unfrozen. Researchers that went on a cruise in the area in February 1977 observed something unusual in the water: “deep convection” had occurred there in winter. That is, in winter, the water from the surface sank down to 4000 m depth!

Can you see the Weddell Polynya in the upper left area? Credit: Claire Parkinson (NASA GSFC)

Can you see the Weddell Polynya in the upper left area? Credit: Claire Parkinson (NASA GSFC)

In the “real” Southern Ocean, that has never been observed since. But I discovered during my thesis that in climate models, it happens every winter, over very large areas, and continues happening during climate change simulations. So I decided to investigate into more details what causes this phenomenon using one model in particular, and that is the topic of the paper that got accepted this week.

2) How to fix your climate model

A climate model is a computer representation of the world as we understand it. It is not perfect, first because we still don’t understand everything about the climate system, but also because we are limited in how detailed our representation can be. We do not have enough computer power to model the fine scale processes, and instead have to “parameterise” them. For example, we cannot model all the individual droplets of rain, and instead talk about precipitation over a 200 x 200 km² region. Similarly, deep convection and mixing are very complex processes in the ocean and are parameterised. My basic idea was that deep convection = too much vertical mixing (the water sinks too easily). So I created a few variations of the model, and in each of them I only changed one thing: I increased or decreased vertical mixing.

I was wrong. The problem with deep convection is not the mixing per se, but what happens a few years before deep convection. I won’t bore you with the details (only 40 pages of my thesis…), but if you have waters that are anomalously warm or salty at the surface, you have a problem. And if the model has low vertical mixing, the problem stays, whereas if the model mixes strongly, the problem will be “defused” (like a shaker mixing all the cocktail ingredients). This is why, opposite to what I first thought, it is the version of the model with strong mixing that does not do deep convection in the Southern Ocean!

3) Why does the representation of Southern Ocean deep convection in models matter?

Well, it was the overall topic of my thesis! More seriously, deep convection makes surface waters sink to the sea floor at the wrong location in models. In reality, this process happens much closer to the coastline, and then this water travels to the rest of the ocean. These models are used to estimate climate change up to 2100. It is a fact that our climate is warming. But not all areas are warming and will warm at the same rate. In particular, coastal Antarctic waters are much warmer by 2100 than waters in the middle of the Weddell Sea. So the water that is sinking in models (because of deep convection) is probably less warm than the water that will actually sink by the Antarctic coast in the real ocean during this century. Warmer in reality = more sea level rise than represented by models.

By correcting the models, we improve our ability to estimate current and future sea level rise, and hence build better coastal defences (among other things).

Sea level rise in Europe if all the ice melts... sorry Norwich! Credit: National Geographic

Sea level rise in part of Europe if all the ice melts… sorry Norwich!
Credit: National Geographic

For more information:
On the Weddell Polynya and polynyas in general, I can recommend the book “Polynyas: Windows to the World” by Walker Smith Jr. and David Barber.

Deep convection in the Weddell Sea was revealed by Arnold Gordon’s 1978 paper “Deep antarctic convection west of Maud Rise”.

I will add the link to my paper “Increasing vertical mixing to reduce Southern Ocean deep convection in NEMO3.4” as soon as it is available online. In the meantime, you can read the version from six months ago here.

And if you really love reading me, have a look at what climate models do compared to the real ocean, what they think will happen by 2100, and much more in my thesis.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s