Cryospheric Sciences

Ice-Ocean interaction

Image of the Week — Cavity leads to complexity

Aerial view of Thwaites Glacier [Credit: NASA/OIB/Jeremy Harbeck].


A 10km-long, 4-km-wide and 350m-high cavity has recently been discovered under one of the fastest-flowing glaciers in Antarctica using different airborne and satellite techniques (see this press release and this study). This enormous cavity previously contained 14 billion tons of ice and formed between 2011 and 2016. This indicates that the bottom of the big glaciers on Earth can melt faster than expected, with the potential to raise sea level more quickly than we thought. Let’s see in further details how the researchers made this discovery.

Thwaites Glacier

Thwaites Glacier is a wide and fast-flowing glacier flowing in West Antarctica. Over the last years, it has undergone major changes. Its grounding line (separation between grounded ice sheet and floating ice shelf) has retreated inland by 0.3 to 1.2 km per year in average since 2011. The glacier has also thinned by 3 to 7 m per year. Several studies suggest that this glacier is already engaged in an unstoppable retreat (e.g. this study), called ‘marine ice sheet instability’, with the potential to raise sea level by about 65 cm.

Identifying cavities

With the help of airborne and satellite measurement techniques, the researchers that carried out this study have discovered a 10km-long, 4km-wide and 350m-high cavity that formed between 2011 and 2016 more than 1 km below the ice surface. In Figure 2B, you can identify this cavity around km 20 along the T3-T4 profile between the green line (corresponding to the ice bottom in 2011) and the red line (ice bottom in 2016). According to the researchers, the geometry of the bed topography in this region allowed a significant amount of warm water from the ocean to come underneath the glacier and progressively melt its base. This lead to the creation of a huge cavity.

Fig. 2: A) Ice surface and bottom elevations in 2014 (blue) and 2016 (red) retrieved from airborne and satellite remote sensing along the T1-T2 profile identified in Fig. 2C. B) Ice surface and bottom elevations in 2011 (green) and 2016 (red) along the T3-T4 profile. C) Changes in ice surface elevation between 2011 and 2017. The ticks on the T1-T2 and T3-T4 profiles are marked every km [Credit: adapted with permission from Figure 3 of Milillo et al. (2019)].

What does it mean?

In order to make accurate projections of future sea-level rise coming from specific glaciers, such as Thwaites Glacier, ice-sheet models need to compute rates of basal melting in agreement with observations. This implies a correct representation of the bed topography and ice bottom underneath the glacier.

However, the current ice-sheet models usually suffer from a too low spatial resolution and use a fixed shape to represent cavities. Thus, these models probably underestimate the loss of ice coming from fast-flowing glaciers, such as Thwaites Glacier. By including the results coming from the observations of this study and further ongoing initiatives (such as the International Thwaites Glacier Collaboration), ice-sheet models would definitely improve and better capture the complexity of glaciers.

Further reading

Edited by Sophie Berger

David Docquier is a post-doctoral researcher at the Earth and Life Institute of Université catholique de Louvain (UCL) in Belgium. He works on the development of processed-based sea-ice metrics in order to improve the evaluation of global climate models (GCMs). His study is embedded within the EU Horizon 2020 PRIMAVERA project, which aims at developing a new generation of high-resolution GCMs to better represent the climate.


Image of the Week – (Un)boxing the melting under the ice shelves

Image of the Week – (Un)boxing the melting under the ice shelves

The Antarctic ice sheet stores a large amount of water that could potentially add to sea level rise in a warming world (see this post and this post). It is currently losing ice, and the ice loss has been accelerating in the past decades. All this is linked to the melting of ice – not at the surface but at the base, underneath the so-called ice shelves which form the continuation of the Antarctic ice sheet over the ocean. These floating ice shelves (represented in color in our Image of the Week) are melted by ocean water from underneath. How can this process called ‘sub-shelf melting’ be included in ice-sheet models? One simple way is to divide the ice-shelf cavity into a number of ocean boxes. Let’s briefly see how it works.

How to model sub-shelf melting in ice-sheet models?

There are three main ways to do so – which way is most suitable depends on the application:

  1. The most elaborated approach is to use ocean models that resolve ocean dynamics underneath the ice shelves. However, they need a lot of computational power.

  2. As an alternative, simple parameterizations in which melting is a function of the depth of the ice-shelf base can be used. However, such parameterizations are for many applications too simple…

  3. Recently, intermediate approaches that include the basic ocean dynamics have been developed (e.g. Lazeroms et al., 2018; Pelle et al., in review). One such approach is the ocean box model (Olbers and Hellmer, 2010) that we extended for the use in an ice-sheet model. Our extension is called Potsdam Ice-shelf Cavity mOdel (PICO, Reese et al., 2018).

In the following, we take a closer look into the approach of PICO…

“Boxing” the cavity circulation

In Antarctic ice-shelf cavities (i.e. the water below the ice shelves), in general, an overturning circulation transports ocean water from the sea floor along the ice-shelf base towards the calving front (see Figure 2). It is driven by the “ice-pump” (Lewis and Perkin, 1986): ice melting near the grounding line (separation between the grounded ice sheet and the floating ice shelf) reduces the density of the ambient water. It becomes buoyant and rises along the shelf base towards the ocean. Through this process, new water from outside of the ice-shelf cavity is “pumped” along the continental shelf towards the grounding line. This leads to the typical pattern of highest melting near the deep grounding lines and lower melting towards the calving front.


Figure 2: Schematic showing the ocean boxes following the ice-shelf base, with the first box B1 near the grounding line, and the last box Bn at the calving front. The arrows indicate the overturning circulation. The ocean water enters the cavity from box B0 which is at depth of the continental shelf, in front of the ice shelf. [Credit: Fig. 1 of Reese et al. (2018)]


By dividing the ice-shelf cavity into 2 to 5 ocean boxes, the transport of the overturning circulation is simplified while the sub-shelf melt pattern is captured. The open ocean conditions are simply represented by the ocean reservoir box B0 (Figure 2). And the circulation is driven by the differences in water density between the ocean reservoir (B0 in Figure 2) and the first box near the grounding line (B1 in Figure 2). The model computes sub-shelf melting successively over the ocean boxes, starting near the grounding line.

Sub-shelf melting with PICO

Sub-shelf melting can vary a lot in-between ice shelves (Figure 1). Antarctic ice-shelf cavities can roughly be sorted into two types (Joughin et al., 2012). The first category are the cold cavities in which the ocean water is close to the freezing point and in which sub-shelf melting is generally low, about 0.1 meter per year. The second category are warm cavities which have a temperature of about 1 degree – that does not sound like much, but for an ice shelf, this feels like being in a sauna – and sub-shelf melting can easily exceed 10 meters per year. Small changes in ocean temperatures can hence have large effects on sub-shelf melting. An increase in sub-shelf melting thins the ice shelf, as for example observed in the Amundsen Sea region in West Antarctica (see this post). The ice shelves there are examples for warm cavities, and a cold cavity is, for instance, underneath the Filchner-Ronne Ice Shelf (see Figure 1 for the specific locations).

In reality, of course, things are much more complicated than simulated by our PICO model. For example, the Coriolis effect can influence ocean circulation in the cavities, sills in the bed can block access of warm water to the grounding line and so on…

Applications of PICO

To summarize, PICO is a simple and efficient modeling tool that can capture the general pattern of sub-shelf melting observed in Antarctica today. Being implemented in the Parallel Ice Sheet Model, it is openly available, so if you got excited about what it can do and want to use it yourself, you’re welcome to download it!

Further reading

Edited by David Docquier

Ronja Reese is a postdoctoral researcher at the Potsdam Institute for Climate Impact Research, Germany, in the group of Prof. Dr. Ricarda Winkelmann. She investigates ice dynamics in Antarctic with a focus on ice-ocean interactions and ice-shelf buttressing. She developed and implemented PICO together with Ricarda Winkelmann, Torsten Albrecht, Matthias Mengel and Xylar Asay-Davis. Contact Email:

Image of the Week — Think ‘tank’: oceanography in a rotating pool

Miniature ocean at the Coriolis facility in Grenoble. [Credit: Mirjam Glessner]

To study how the ocean behaves in the glacial fjords of Antarctica and Greenland, we normally have to go there on big icebreaker campaigns. Or we rely on modelling results, especially so to determine what happens when the wind or ocean properties change. But there is also a third option that we tend to forget about: we can recreate the ocean in a lab. This is exactly what our Bergen-Gothenburg team has been doing these last weeks at the Coriolis facility, in sunny Grenoble.

How to build your own miniature ocean

Take a 13m diameter (circular) swimming pool. Install it on a rotating platform, and start turning to simulate the Coriolis force, i.e. the impact of the Earth rotation on the flow. Fill it so that the water level reaches 90cm. Actually, the exact value does not matter and can be changed; just make sure that your tank width is an order of magnitude larger than your depth, and that you do not overflow everywhere on the lab floor. Congratulations, you have an ocean! But for now it is a bit boring.

Let’s add some stratification and density-driven currents. As we explained in a previous entry, all you need to do for that is change the temperature and/or salinity of your water. The people here at the Coriolis facility say that changing the salinity is easier than the temperature, so ok, put a source somewhere in your tank that will spit out salty water. Make it even more realistic: have some trough, underwater mountains, solid ice shelves etc. Or rather, some Plexiglas of the corresponding shape. Now you have a beautiful part of the ocean with realistic currents!

But how do you observe it? You can lower probes into the water at specific locations, as if you were doing miniature CTD casts in your miniature ocean. Or you can visualise the whole full-depth flow: add tracer particles to the water flowing from the source (in our case, biodegradable plastic), shoot lasers at it at various depth levels, and take high resolution pictures as you do so. Then, you can track the particles from one image to the next to infer their velocity, using a method called PIV.


By the way, it looks way neater than on this image – that one is just from our overview camera, for fun. [Credit: Céline Heuzé]

What does it look like when you fire lasers at a large rotating tank?

In a nutshell, it looks like this:

The water flows from the source on the right of the image, towards the ‘ice shelf’ on the left. We are watching the scene from above, from our office that rotates with the tank. The laser successively illuminates several levels from the bottom of the channel to the water surface, revealing the changing structure of the flow with depth. In our real experiment, it took more than 10 minutes for the water to reach the ‘ice shelf’ – here, I have slightly accelerated it.

It is surprisingly peaceful and relaxing to watch. Well, there is tension and suspense regarding what the flow will do since this is, after all, why we are here. But otherwise you are in the dark, with particles shining all around you, in the silence except for the low-squeeking noise of the rotating tank, gently rocked by the vibrations of the platform, and there is not much you can do but wait and enjoy the view. You can also count how many undesired bubbles and dead insects floating at the surface you can see!

Why do we need rotating tank experiments?

As we explained in this blog, the future of the Antarctic ice sheet is unknown due to marine ice sheet instability. We do not know under which conditions the floating ice shelves that block (‘buttress’) the big land-based ice sheet may collapse. In particular, we do not know what controls the flow of comparatively warm waters that melt the ice shelves:

  •  under which conditions do these waters penetrate under the ice?
  •  at which depths do they sit?
  •  what are the impacts of stratification and the shape of the ice shelf itself?

These questions cannot easily be answered by going in the field. We would need access to many ice shelves, year round, and the ability to observe the flow everywhere –including under the ice– synoptically. Instead in the lab, we just need to adjust our flow speed, or the rotation speed of the tank, or the amount of salt in the source, and we are ready to observe!

Further reading:

The blog of the team:

Our blog post about the video game Ice Flows!, illustrating the marine ice sheet instability

Edited by Sophie Berger

Image of The Week – The Pulsating Ice Sheet!

Image of The Week – The Pulsating Ice Sheet!

During the last glacial period (~110,000-12,500 years ago) the Laurentide Ice Sheet (North America) experienced rapid, episodic, mass loss events – known as Heinrich events. These events are particularly curious as they occurred during the colder portions of the last glacial period, when we would intuitively expect large-scale mass loss during warmer times. In order to understand mass loss mechanisms from present-day ice sheets we need to understand what happened in the past. So, how can we better explain Heinrich events?

What are Heinrich Events?

During a Heinrich event large swarms of icebergs were discharged from the Laurentide Ice Sheet into the Hudson Strait and eventually into the North Atlantic Ocean. This addition of fresh water to the oceans caused a rise in sea level and a change in ocean currents and therefore climate.

We know about these events by studying glacial debris that was transported from the ice sheet into the oceans by the icebergs and eventually deposited on the ocean floor. From studying ocean-sediment records we know that Heinrich events occurred episodically during the last glacial period but not on at a regular intervals. Interestingly, when compared to temperature records from Greenland ice cores, it can be seen that the timing of Heinrich events coincides with the cold phases of Dansgaard–Oeschger (DO) cycles – rapid temperature fluctuations which occurred during the last glacial period (see our previous post).

the timing of Heinrich events coincides with the cold phases of Dansgaard–Oeschger (DO) cycles

What do we think causes them?

A new study, published last month in Nature, uses numerical modelling to show how pulses of warm ocean water could trigger Heinrich events. Our image of the week (Figure 1) illustrates the proposed mechanism for one event cycle:

  • a) Ice sheet at it’s full extent, grounded on a sill (raised portion of the bed, at the mouth of the Hudson Strait). Notice the sill is around 300m below sea level at this time.
  • b) A pulse of sub-surface water (purple) warms by a few degrees, encouraging iceberg calving at the glacier front and causing the ice begin to retreat from the sill.
  • c) As the ice retreats, it becomes unstable due to an inwards sloping bed (see our previous post on MISI). This leads to sudden rapid retreat of the ice – characteristic of Heinrich events.
  • d) Due to ice loss and thus less mass depressing the bed, the bed will slowly rise (Glacial Isostatic Adjustment), eventually the sill has risen to a level which cuts off the warmer water from the ice front and the ice can slowly advance again.

Once the ice has advanced back to it’s maximum extent (a) it will slowly depress the bed again, allowing deeper, warmer water to reach the ice front and the whole cycle repeats!

The authors of this study used this model to simulate Heinrich events over the last glacial period and were able to accurately predict the timing of Heinrich events, as known from ocean sediment records. Check out this video to see the model in action!!

Why is it important?

This study shows that the proposed mechanism probably controlled the onset of rapid mass-loss Heinrich events in the past and more generally that such mechanisms can cause the rapid retreat of marine terminating glaciers. This is important as it adds to our understanding of the stability (or instability) of present day marine terminating glaciers – such as the West Antarctic Ice Sheet! If such rapid mass loss happened regularly in the past we need to know if and how it might happen in the future!

such mechanisms can cause the rapid retreat of marine terminating glaciers.

Check out the full study and the news article summarising the findings here:

Image of the Week — The ice blue eye of the Arctic

Image of the Week — The ice blue eye of the Arctic

Positive feedback” is a term that regularly pops up when talking about climate change. It does not mean good news, but rather that climate change causes a phenomenon which it turns exacerbates climate change. The image of this week shows a beautiful melt pond in the Arctic sea ice, which is an example of such positive feedback.

What is a melt pond?

The Arctic sea ice is typically non-smooth, and covered in snow. When, after the long polar night, the sun shines again on the sea ice, a series of events happen (e.g. Fetterer and Untersteiner, 1998):

  • the snow layer melts;

  • the melted snow collects in depressions at the surface of the sea ice to form ponds;

  • these ponds of melted water are darker than the surrounding ice, i.e. they have a lower albedo. As a result they absorb more heat from the Sun, which melts more ice and deepens the pond. Melt ponds are typically 5 to 10 m wide and 15 to 50 cm deep (Perovich et al., 2009);

  • eventually, the water from the ponds ends up in the ocean: either by percolation through the whole sea-ice column or because the bottom of the pond reaches the ocean. Sometimes, it can also simply refreeze, as the air temperatures drop again (Polashenski et al., 2012).

Melt ponds cover 50-60% of the Arctic sea ice each summer (Eicken et al., 2004), and up to 90% of the first year ice (Perovich al., 2011). How do we know these percentages? Mostly, thanks to satellites.

Monitoring melt ponds by satellites

Like most phenomena that we discuss on this blog, continuous in-situ measurements are not feasible at the scale of the whole Arctic, so scientists rely on satellites instead. For melt ponds, spectro-radiometer data are used (Rösel et al., 2012). These measure the surface reflectance of the Earth i.e. the proportion of energy reflected by the surface for wavelengths in the visible and infrared (0.4 to 14.4 μm). The idea is that different types of surfaces reflect the sunlight differently, and we can use these data to then map the types of surfaces over a region.

In particular for the Arctic, sea ice, open ocean and any stage in-between all reflect the sunlight differently (i.e. have different albedos). The way that the albedo changes with the wavelength is also different for each surface, which is why radiometer measurements are taken for a range of wavelengths. With these measurements, not only can we locate the melt ponds in the Arctic, but even assess how mature the pond is (i.e. how long ago it formed) and how deep it extends. These values are key for climate change predictions.

Fig. 2: Melt pond seen by a camera below the sea ice. (The pond is the lighter area) [Credit: NOAA’s]

Melt ponds and the climate

Let’s come back to the positive feedback mentioned in the introduction. Solar radiation and warm air temperature create melt ponds. The darker melt ponds have a higher albedo than the white sea ice, so they absorb more heat, and further warm our climate. This extra heat is also transferred to the ocean, so melt pond-covered sea ice melts three times more from below than bare ice (Flocco et al., 2012). This vicious circle heat – less sea ice – more heat absorbed – even less sea ice…, is called the ice-albedo feedback. It is one of the processes responsible for the polar amplification of global warming, i.e. the fact that poles warm way faster than the rest of the world (see also this post for more explanation).

The ice-albedo feedback is one of the processes responsible for the polar amplification of global warming

But it’s not all doom and gloom. For one thing, melt ponds are associated with algae bloom. The sun light can penetrate deeper through the ocean under a melt pond than under bare ice (see Fig. 2), which means that life can develop more easily. And now that we understand better how melt ponds form, and how much area they cover in the Arctic, efforts are being made to include more realistic sea-ice properties and pond parametrisation in climate models (e.g. Holland et al., 2012). That way, we can study more precisely their impact on future climate, and the demise of the Arctic sea ice.

Edited by Sophie Berger

Further reading

Image of The Week – Plumes of water melting Greenland’s tidewater glaciers


Figure 1: Simulation of a plume at a tidewater glacier in a general circulation model (MITgcm). Left – water temperature and right – time-averaged submarine melt rate in metres per day. Shown are face-on views of a tidewater glacier, as if you were under the water in front of the glacier, looking towards the calving front. 250 m3/s of fresh water emerges into the ocean from a channel at the bottom of the glacier, forming a plume. As the plume rises towards the fjord surface it mixes turbulently with warm ocean water, causing the plume to warm with height. Further details of this simulation can be found here: Slater et al. 2015.

Loss of ice from The Greenland Ice Sheet currently contributes approximately 1 mm/year to global sea level (Enderlin et al., 2014). The most rapidly changing and fastest flowing parts of the ice sheet are tidewater glaciers, which transport ice from the interior of the ice sheet directly into the ocean. In order to better predict how Greenland will contribute to future sea level we need to know more about what happens in these regions.

Tidewater glaciers meet the ocean at the calving front (Fig. 2), where ice undergoes melting by the ocean (“submarine melting”) and icebergs calve off into the sea. In recent decades, tidewater glaciers around Greenland have retreated (due to increased loss of ice at the calving front) and started flowing faster. This in turn causes more ice to be released into the ocean, contributing to sea level. Understanding the cause of these changes at tidewater glaciers is an ongoing topic of research.

Figure 2: Kangiata Nunata Sermia, a large tidewater glacier in south-west Greenland. The expression of a plume originating at the base of the calving front is visible on the fjord surface as turbid sediment-rich water. [Credit: Peter Nienow]

Figure 2: Kangiata Nunata Sermia, a large tidewater glacier in south-west Greenland. The expression of a plume originating at the base of the calving front is visible on the fjord surface as turbid sediment-rich water. [Credit: Peter Nienow]

One possible cause of change is an observed warming of the ocean around Greenland (Straneo and Heimbach, 2013). A warming of the ocean is likely to lead to increased submarine melt rates, which may in turn influence iceberg calving if, for example, melting results in instability of the ice at the calving front. Submarine melt rates are thought to be increased further by upwelling of warm water at the calving front (Fig. 1 and Fig. 2).

This upwelling water, called a plume, may be initiated by submarine melting of the ice, or by fresh glacial meltwater from the ice sheet surface. This fresh glacial meltwater penetrates to the base of the glacier and flows into the ocean from beneath the glacier, which may be hundreds of metres underwater. Once in the ocean, the meltwater rises buoyantly because of a density difference between the meltwater and ocean water, forming a plume. In order to better understand the effect of plumes on submarine melting, we can model plumes using a numerical model (e.g. MITgcm). Our image of the week (Fig. 1) shows such a model, which we can use to estimate submarine melt rates. In combination with simpler analytical approaches (Jenkins et al., 2011; Slater et al., 2016), we can estimate how submarine melt rates may change over time and from glacier to glacier (Carroll et al., 2016), and begin to assess the effect of submarine melting on tidewater glaciers and ultimately on future sea level rise.

Edited by Teresa Kyrke-Smith and Emma Smith


Donald Slater is a PhD student in the Glaciology and Cryosphere Research Group at the University of Edinburgh. His research focusses on understanding the effect of the ocean on the Greenland Ice Sheet. For more information look up his website or follow him on twitter @donald_glacier.

Image of the Week — FRISP 2016

Image of the Week — FRISP 2016

The Forum for Research into Ice Shelf Processes, aka FRISP, is an international meeting bringing together glaciologists and oceanographers. There are no parallel sessions; everyone attends everyone else’s talk and comment on their results, and the numerous breaks and long dinners encourage new and interdisciplinary collaborations. In fact, each year, a few presentations are the result of a previous year’s question!

The location changes every year, moving around the institutions that are involved with Arctic and Antarctic research. The 2016 edition just occurred this week, 3rd – 6th October, in a marine research station of the University of Gothenburg, in the beautiful Gullmarn Fjord.

Each year, a few presentations are the result of a previous year’s question!

Fjord at the sunset [Credit: Céline Heuzé]

Gullmarn fjord at the sunset [Credit: Céline Heuzé]

70 participants from 37 institutions:

  • Attended 49 talks on model results, new observation techniques, and everything in between;

  • Spent more than 15h discussing these results, including 2h around 15 posters;

  • Drank 50 L of coffee, 60 L of tea, 20 L of lingon juice… and a fair amount of wine!

Poster session at the FRISP 2016 meeting. [Credit: Céline Heuzé]

Poster session at the FRISP 2016 meeting. [Credit: Céline Heuzé]

I can’t really choose THE highlight of the conference.
As an organiser, it was a real pleasure to simply see it happen after all the long hours of planning.
As a scientist, it was a great and productive meeting, giving me new ideas and the opportunity to discuss my recent work with the big names of the field in a friendly environment.
And as a human, I enjoyed most the under-ice footages, and in particular the general ”ooooh” that came from the audience.

It was a bit sad to say goodbye to the participants, old friends and new collaborators. But I know that I will see them again during FRISP 2017… and I hope to see you there as well!

 Edited by Sophie Berger and Emma Smith

Water Masses “For Dummies”

Water Masses “For Dummies”

Polar surface water, circumpolar deep water, dense shelf water, North Atlantic deep water, Antarctic bottom water… These names pop in most discussions about the ice-ocean interaction and how this will change in a warming climate, but what do they refer to?

In our second “For Dummies” article, we shall give you a brief introduction to the concept of “water mass”, explain how to differentiate water from more water, and why you would even need to do so.

Global heat budget and the need for an ocean circulation

The global climate is driven by differences between the incoming shortwave radiation and the outgoing longwave radiation (Fig. 1):

  • In the tropics, there is a surplus of energy: the Sun brings more heat, all year-round, than what is radiated out;
  • At the poles in contrast, there is a net deficit: more energy is leaving than is coming from the Sun (who is absent in winter).

The global ocean and atmosphere circulations act to reduce this imbalance, by transporting the excess heat from the tropics to the pole. Here we will focus on the global ocean circulation only, since this post is written by an oceanographer, but similar principles also apply to atmospheric circulation.

Fig 1 :Earth’s latitudinal radiation bugdet, The tropics show a surplus of energy that compensates the Poles’ deficit[Credit: National Oceanograpy Center

Fig 1 :Earth’s latitudinal radiation bugdet, The tropics show a surplus of energy that compensates the Poles’ deficit [Credit: National Oceanograpy Center].

The global ocean circulation

In a nutshell, surface waters bring heat towards the poles where they cool down, sink to the abyss, and return towards the tropics as deep waters where they can go back to the surface..…

We talk about “the global ocean circulation” because although the Earth officially has five oceans, they are not totally separate bodies of water. In fact, the Arctic, Atlantic, Indian, Pacific and Southern oceans are interconnected, with water circulating and moving between them. How does this happen?

The global ocean circulation has two components:

  • The wind-driven circulation, fast but limited to a few hundred metres below the surface of the ocean (read more about it here for example);
  • And the thermohaline circulation (shown on Fig. 2), slower but which affects the whole depth of the ocean.

Today’s post focuses on the latter, since we will talk about water properties. The thermohaline circulation, also called density-driven circulation, depends on two water properties:

  • The temperature (‘thermo’) is mostly controlled by heat exchange with the atmosphere or the ice. Cold water has a high density.
  • The salinity (‘haline’) can be modified by evaporation, precipitation, or addition of fresh water from melted glaciers/ice sheets or rivers. Salty water has a high density.
Fig 2- The global thermohaline circulation shows warm surface currents in red, cold deep currents in blue. Deep waters form in the North Atlantic and Southern oceans. [Credit: NASA]

Fig 2- The global thermohaline circulation shows warm surface currents in red, cold deep currents in blue. Deep waters form in the North Atlantic and Southern oceans [Credit: NASA].

Roughly speaking, a water mass is any drop of the ocean within a specific range of temperature and salinity, and hence specific density. Some water masses are found at particular locations or seasons, while others can be found all around the globe, all the time. Since density sets the depth (density MUST always increase with depth), water masses will lie and travel at particular depth levels.

A quick and dirty oceanography guide

Water masses are formed.

Some are the result of the mixing of other water masses. The others start at the water surface, where they exchange gas (notably oxygen and carbon) with the atmosphere. When a water mass becomes denser than the waters below it , for example, if it is cooled by the wind or ice, it sinks to its corresponding depth within the ocean.

Fig 3- The bathymetry of the Arctic Ocean forces dense (deep) water masses to enter the region via Fram Strait whereas lighter (shallower) waters can go through the Barents Sea [Credit: adapted from IBCAO bathymetry map, Jakobsson et al., 2012 ].

Water masses move all around the globe…

…provided their density allows it. The vertical distribution of density in the ocean must be “stably stratified”, which means that the density increases with depth. In practice, that means that dense waters cannot climb up a shallow bathymetric feature but have to find a way around it. For example to enter the Arctic Ocean (Fig 3), a dense water mass has no choice but to go via Fram Strait, whereas a less dense one can go via the Barents Trough. Similarly, there is a depth limit of about 500 m to reach the northwestern Greenland glaciers.

Water masses retain their properties

Or rather, not all these properties change considerably with space and time. We are not talking only about temperature and salinity, but also about gas and chemical concentrations. It is then possible to track a water mass as it travels around the globe or watch its evolution with time.

You should use T-S diagrams

Visualising water properties can either be done with one graph showing how the temperature varies with depth plus another one for the salinity (multiplied by the number of locations to be observed at the same time); or all of this information can be combined on one image (as done on Fig. 4). This image is called a T-S diagram it and shows how the temperature (T) varies as a function of the salinity (S). It is customary to also draw the lines of constant density (the ‘isopycnals’, black on Fig. 3). These isopycnals give information about the types of mixing happening and the stratification, but we will talk about that in another post.

Fig 4 - an example of how to combine several profiles (top) into a T-S diagram, for one of the randomly selected Arctic historical points that I work with.[Credit: C. Heuzé]

Fig 4 – an example of how to combine several profiles (top) into a T-S diagram, for one of the randomly selected Arctic historical points that I work with [Credit: C. Heuzé].

Because each water mass occupies a very specific region of the T-S diagram (see Fig 5 for an example in the Atlantic), identifying them is relatively easy once you have plotted your data on such diagrams.

Fig 5 – example of a reference T-S diagram with the different water masses of the Atlantic Ocean. Water massed are labelled by their acronym (e.g. AABW= Antarctic Bottom Water) [Credit: after Emery and Meincke (1986)]

Why do ocean water masses matter to the cryosphere?

  • Marine ice sheet instability, and more generally basal melting, is caused by warm dense waters melting floating glaciers from below; how dense the water mass is determines whether it can even reach the glacier.
  • Sea ice formation and melting can be largely affected by water masses moving up and down, especially is those going up are warm.

But there’s a reason why we always talk about “ice-ocean” interactions: it’s not just the ocean acting on the ice, but also the ice impacting the ocean:

  • The densest water mass in the world, Antarctic Bottom Water, forms in the middle of winter if a hole in the sea-ice cover opens (that is called a polynya), suddenly exposing the relatively warm ocean to the extremely cold atmosphere. The resulting strong heat loss and the increased salinity as sea ice reforms make this water sink straight to the bottom;
  • On the other hand, deep water formation can be stopped by the cryosphere: paleorecord evidence showed that it happened in the North Atlantic due to surging ice sheet / marine ice sheet instability (so called Heinrich events) or meltwater floods (Younger Dryas);
  • Less dramatically, icebergs, ice shelves or even sea ice, can cool or freshen water masses they meet, forming “modified” water masses (for example “modified Atlantic Water”),

Each aspect of these interactions is already experiencing climate change and is much more complex than this brief overview… but that will be the topic of another post!

Further reading

 Edited by Sophie Berger and Emma Smith

Image of the week – Our salty seas and how this affects sea ice growth

Image  of the week – Our salty seas and how this affects sea ice growth

Earth’s oceans are not simply just water, they are a complicated multi-component fluid consisting of water and dissolved salts (ask anyone who has tried to drink it!). The existence of these salts has a significant impact on global ocean circulation. Nowhere is this more significant than in the polar oceans where it is one of the key factors influencing sea ice formation. In this week’s image of the week we are going to show you how freezing ocean water is a little more complicated than you may think!

The salinity and temperature of ocean water affect its density; essentially how much it weighs. Typical ocean densities are around 1000 kg/m3  and, depending on the temperature and salinity may vary by up to 1 %. This seems tiny, but these small changes in density are what drive the thermohaline circulation, the dominant large-scale ocean circulation. The density of sea water, as a function of temperature and salinity, is expressed in terms of the equation of state  (a mathematical way of describing the density of sea water in relation its temperature and salinity). Contours of the equation of state of seawater are shown in this week’s Image of the Week. The figure is from a recent paper by Mary-Louise Timmermans and Steven Jayne, who try to understand how changes in Arctic temperature will influence the density, and therefore the circulation, in the Arctic Ocean. The y-axis is temperature, and the x-axis is salinity. The black lines are density contours. The dashed line plots the freezing point of water.

Sea Ice Formation

Sea ice begins to form when ocean water is brought to this freezing point. If one was to put a cup of tap water into a freezer, ice would begin to form at 0 °C. But talk to a group of polar ocean modellers, and they will tell you the freezing point of water is about -1.8 °C. How can this be?

Let’s got back to our figure for some clues. Looking at the dashed line representing freezing point of ocean water you will notice that as the salinity increases, the freezing point decreases. So an increase in salinity of sea-water suppresses its freezing point. Just like how salt is used to melt ice in winter, it prevents the water from reaching its freezing point until the water reaches roughly -2 °C.

How does this all link together?

When the ocean gets cold, the influence of temperature on density changes, affecting how rapidly sea ice can form. Take a look at the bending of the black contours as the temperature is reduced to zero and below. Whereas in “normal”, warm contexts, a decrease in temperature leads to an increase in density, this changes as the temperature approaches 0 °C. As the ocean cools, the top-most, coldest water typically sinks, and is replaced by warmer water from below, driving ocean circulation convection. It therefore can take a long time to bring the surface of the ocean to near 0 °C. Since there is salt in the ocean, the water can reach colders temperatures where something very different happens. As the water continues to cool, the coldest water no longer sinks, and may even float, with sea-ice formation happening rapidly.

The formation process of sea ice, and its relationship to the ocean it forms out of is an extremely complicated and rich phenomenon, and it all depends on salt!

Further Reading

  • Mary-Louise Timmermans and Steven R. Jayne, 2016: The Arctic Ocean Spices Up. J. Phys. Oceanogr. 46, 1277–1284, doi: 10.1175/JPO-D-16-0027.1.
  • For more on sea ice check the National Snow and Ice Data Center (NSIDC) website – All About Sea Ice!

Edited by Emma Smith

Image of the week – The winds of summer (and surface fluxes of winter)

Image of the week – The winds of summer (and surface fluxes of winter)

Antarctica is separated from the deep Southern Ocean by a shallow continental shelf. Waters are exchanged between the deep ocean and the shallow shelf, forming the Antarctic cross-shelf circulation:

  • Very dense waters leave the shelf as Antarctic Bottom Water (AABW) that will then flow at the bottom of all oceans.
  • Meanwhile, relatively warm water from the Southern Ocean, Modified Circumpolar Deep Water (MCDW*) comes on the continental shelf and brings heat to the ice shelves.

That is, Antarctic cross-shelf circulation influences the water mass that transports heat, carbon and nutrient all around the globe in very large volumes (Purkey and Johnson 2013), and the basal melting of Antarctic floating ice (Hellmer et al. 2012), hence the stability of the whole Antarctic ice sheet.

Although critical for both the ocean and the cryosphere, very little is known about the mechanisms behind cross-shelf circulation. We know that the mechanisms that control it vary on a seasonal time scale (Snow et al. 2016b). However, most hydrographic observations around Antarctica are taken in summer, when there is less sea ice and when the Southern Ocean is the least stormy. This means that we have very few measurements of the seasonal variations of the cross-shelf circulation itself.

Why does it matter that the cross-shelf circulation varies between summer and winter?

Three words: sea level rise.
Nearly half of the world’s population lives in coastal areas (
UN report). Antarctica contains enough ice to raise the sea level by 60 m, and although a total melting is very unlikely, current rates could raise the sea level by 1m by 2100 (read more about it on To project future sea level rise and design relevant coastal defences, we need models to predict when and where the Antarctic ice will melt.

However, models are only as good as the observations that were used to constrain them. Having only summer observations in an area of Antarctica that has notable differences between summer and winter ocean circulation means that until now, models could not represent accurately the transfer of heat from the ocean to the ice shelves

A better observation strategy is needed if we want our models to correctly represent Antarctic basal melting and the global ocean circulation.

Antarctic cross-shelf circulation: summer vs winter

In summer, the circulation is mostly controlled by the strong katabatic winds blowing from the interior of the Antarctic continent towards the ocean. All the surface water masses go in the same direction, simply following the Antarctic coastal current. Nothing really happens at depth.

In winter, the circulation is also controlled by buoyancy forcings, that is changes in temperature or salinity at the surface of the ocean. Here, these mostly occur in a polynya (a hole in the sea ice cover) where the “warm” ocean is cooled by the very cold atmosphere, and where the surface becomes very salty as sea ice reforms (a process called “brine rejection”: salt is expelled from the new ice as water freezes). These buoyancy forcings form dense water (DSW), which sinks to the abyss and off the shelf as AABW. Mass conservation means that something else (here MCDW*) needs to come to the shelf to compensate for that outflow. You can notice that MCDW now flows in the opposite direction than it did in summer.

Take home message

Summer data is better than no data. But always be aware of the limitations of your model if you don’t have the datasets to test it– you may have a surprise when you do!


Snow, K., B. M. Sloyan, S. R. Rintoul, A. McC. Hogg, and S. M. Downes (2016), Controls on circulation, cross-shelf exchange, and dense water formation in an Antarctic polynya, Geophys. Res. Lett., 43, doi:10.1002/2016GL069479.

Edited by Sophie Berger and Emma Smith