In 2016 two influential new publications raised the possibility of a rapid acceleration of sea level rise in the 21^st century – to ±2 metres (DeConto & Pollard) or more (2-5m, Hansen et al).

In this background article we take a look at both these studies – but also at 30+ other publications that we think are helpful to show the broader context: an increasing pile of evidence, coming from a world of science, indicating ice sheet dynamics are never linear and can be surprisingly rapid.

Looking at the key processes we show that the Hansen and DeConto studies add to a list of no less than 13 different (yet highly entangled, often synergistic) ice sheet melting feedbacks, feedbacks that can promote a rapid acceleration of global sea level rise:

Hats off to these people… A meltwater river on Greenland in the summer of 2014, with climate researchers from the University of California around them. Picture: Mia Bennett, cryopolitics.com.

Understanding sea level rise is understanding the two extremes of the story – and then properly tying them together. This article is about the processes that form that connecting part (and coincidentally that’s also where the scientific debate is at)

To develop a proper understanding of future sea level rise, you first have the two extremes of the story: the observational current speed (global average in the order of 2-3 millimetres/year) and the final sea level rise for a certain amount of created warming (in the order of 29 metres for +2 degrees, and going as high as 55 metres for a runaway warming scenario – see our previous article).

Now if you want to project future sea level rise and you would extrapolate the current speed and draw a straight line until you get to 29 metres or more (well over 10,000 years) you would be a bit silly (dear politicians) – and ignoring a world of science, science about ice sheet behaviour, melting feedbacks – science about acceleration, and one word to emphasize above all: non-linearity.

To help you develop an understanding and link the two extremes of the sea level story, in this special fourth article of the series we have decided to make a list of the various suggested ice sheet melting feedbacks – and what the newest research has to say about them. (Finally. It exists!) Read it, and hopefully you’ll learn more about ice sheet dynamics and mechanisms of accelerated melting. (And if you’ve learned from it, please care to pass it on to others.)

You have probably heard about James Hansen and his latest sea level publication, in which he – together with 18 co-authors – shows why the Greenland and Antarctic ice sheets might be far less stable than previously assumed, raising the possibility of 2-5 metres sea level rise, within this century.

Perhaps a less familiar name (because he is a lot younger) is Robert DeConto, a climate scientist from the University of Massachusetts Amherst, who together with paleoclimatologist David Pollard of Pennsylvania State University this year also published a study suggesting a mechanism for accelerated melting of the Antarctic ice sheet – which could raise global sea levels to 2 metres by 2100.

Now what’s with these studies? In one word: feedbacks. Positive feedbacks sadly – feedbacks that lead to acceleration of the climate response of large ice sheets and therefore to acceleration of sea level rise. In this article we’ll try to take a closer look. [If you want to compare the DeConto and Hansen studies to older SLR projections, please take a look at our short chronology of 2100 sea level rise forecasts – in which we’ve fitted both.]

We understand the height of Anthropocene sea level rise. Big question is the speed

In part 3 of the series we’ve looked at the final extent of the sea level rise our carbon emissions cause. Depending on how fast we manage to reduce these emissions we’re looking at anything between 29 to 55 metres of sea level rise.

Although few people are familiar with these numbers and the sheer magnitude of changes to system Earth they represent – within the scientific community there’s little dispute about this final sea level rise. The paleorecord is clear enough.

What is far more uncertain though is what will happen in the near(er) future – how steep that convex shaped first part of the global sea level rise graph will actually be. In other words: decimetres this century – or perhaps already multiple metres?

A research-based overview of melting feedbacks and other mechanisms for ice sheet collapse:

A sea level rise speed of multiple metres per century would be possible if positive melting feedbacks are activated. And throughout years of research many have already been suggested – and are all subject of ongoing investigation, through modelling, paleoresearch and in situ observations:

1: ALBEDO FEEDBACK
‘Albedo Effect’ – reflectivity-absorption-melting feedback. As the temperature goes up, more snow melts, decreasing summer albedo, increasing warming – accelerating melting

The so-called albedo effect is one. It’s an important climate feedback (both positive and negative) for any place on Earth that undergoes a surface colour (brightness mostly) change.

We know of the albedo effect mostly as an important positive feedback to (summer) Arctic sea ice melting and tundra (permafrost) thawing. When snow melts it unveils an underlying surface – and be it soil, ice or open water, this surface will always be darker and therefore have a lower albedo (reflectivity) – leading to increased absorption of solar energy, converting to heat and therefore increasing warming (that in turn increases melting, fuelling the feedback loop).

This albedo feedback however also applies to glaciers and large icesheets. When the snow cover melts in summer it can show ice underneath, ice that can be surprisingly dirty and dark (due to so-called cryoconites on the ice – bonded dust and soot particles, that are increasingly deposited from fossil fuels combustion and (boreal) wildfires – an albedo-operating positive climate feedback on its own). Or, what’s worse – because looking from space they’re even darker: small ponds of meltwater, that stay on top of the ice sheet – leading to extra heat absorption during the polar summer, when the sun is up 24 hours per day.

You’re already looking at two ice melt accelerating feedbacks in practice, in this image of the Greenland ice sheet: a meltwater river (set to form a moulin and disappear inside the glacier) and cryoconites, bonded dust and soot, increasingly darkening the glacier surface as snow cover shrinks. Picture: Mia Bennett, cryopolitics.com.

Formation of such darkening meltwater ponds is already being observed in Greenland and Antarctica. In May 2016 a group of researchers led by Marco Tedesco of Columbia University and NASA’s Goddard Institute of Space Studies (GISS) published evidence in The Cryosphere that on Greenland the overall albedo had already significantly declined between 1996 and 2012 – and that they found weaknesses in forecasts of continued albedo decline, stating these are underestimated, as models fail to reproduce (darkening) increases in water, grain size and aerosols on the ice sheet surface. And sadly, very recently, in August 2016, a group of researchers led by Emily Langley of Durham University presented evidence in a Geophysical Research Letters publication of a similar albedo feedback, already active on the East Antarctic ice sheet, where (based on satellite observations between 2000 and 2013) they had found a rapid increase of (dark-coloured, heat-absorbing) meltwater lakes (about 8000) on top of the ice sheet.

2: MELTWATER LUBRICATION FEEDBACK
‘Meltwater lubrication feedback’ – warming of the atmosphere causes ice sheet surface melting. Before this meltwater runs to the sea, it can also speed up glacier movement, accelerating sea level rise

Another fairly well-known ice-melting feedback is (suggested) meltwater lubrication. As the top layer of the ice sheet heats up in a warmer atmosphere, billions upon billions of litres of meltwater flow off, first on the surface, collecting in ponds and flowing off in meltwater rivers across the ice surface – and where these find cracks, continuing in so-called moulins, eroded vertical meltwater tunnels. Some of these moulins it has been suggested – especially around the edge of the ice sheets – reach down to the bedrock. And there (but also within fractures in the ice itself) the meltwater may act as a lubricant, increasing the glacier flow, and by doing so the transportation of ice towards the warm ocean, where icebergs are formed and melting takes place. Now the science on this one is somewhat sketchy (because it is difficult to look inside the glaciers) – although large-scale moulin-forming is indeed being observed on Greenland.

In 2013 a group of British researchers led by Sarah Shannon of the University of Bristol said they had tried to model the meltwater lubrication feedback and in their PNAS publication wrote that they had found it would only increase melting by 5 percent. Unwanted extra sea level rise, but no runaway scenario there. These researchers suggest part of the meltwater may refreeze or (the other extreme) the meltwater may collect in large rivers on the bedrock that flow directly to the ocean.

Newer, observational research by an American group brings these model calculations into question. Led by Kristin Schild of Dartmouth College and publishing results in Annals of Glaciology in April 2016, these researchers managed to actually track sediment-rich meltwater leaking from multiple spots underneath the Rink Isbræ, a fast-flowing West Greenland glacier – meltwater that also led to increased glacier front erosion, they found.

In 2013 a group of researchers led by Andreas Peter Ahlstrøm of the Geological Survey of Denmark & Greenland reported in the open access journal Earth System Science Data that they had found that meltwater (through moulin formation and lubrication of the glacier base) did indeed increase glacier velocity, but that this was a seasonal effect, that only occurred at the onset of the melting season, when meltwater (as a lubricant) was a limiting factor. Their results were more or less confirmed by an American-Dutch research group led by Twila Moon of the University of Washington one year later with a publication in Geophysical Research Letters that distinguished three different seasonal glacier velocity patterns, two of which were influenced by the presence of meltwater at the glacier base.

In August 2015 a large group of researchers from the Switzerland-based World Glacier Monitoring Service reported general acceleration of meltwater formation and glacier movement for glaciers in Greenland, West Antarctica, the Canadian and Alaskan Rocky mountains, the Himalayas and glaciers in European mountain ranges in a special publication in the Journal of Glaciology.

Another group led by University of Washington’s Ian Joughin, looking more closely at another specific glacier on the west side of the Greenland ice sheet, Jakobshavn Isbræ, found it had doubled in speed, twice. Building on earlier observations by NASA, showing the Jakobshavn glacier had first doubled its speed between 1997 and 2003, their new observations, published in The Cryosphere in 2014, showed another doubling from 2003 to 2013 (again observing large seasonal fluctuations, in which meltwater could play a key role). [Finding two doublings in a row is of course important observational evidence to distinguish between a (possible) single speed increase and (possible) real acceleration, continuing exponentially in time (representing Moore’s Law of doubling times).]

The mechanism behind the meltwater lubrication feedback and observed glacier speed increase for the Greenland Ice Sheet. Image first published in New York Times by Andrew Revkin.

Now of course when it comes to these actual glacier speed observations the mechanism is not fully clear: the observed accelerating glacier velocity does not necessarily have to do be due to meltwater lubrication – or meltwater alone. It could also be down to other feedbacks. Hence the list goes on:

3: ICE HEIGHT FEEDBACK
Diminishing ice sheet altitude feedback working on temperature, meltseason lengts & snow accumulation – easily overlooked, but a sensical Grand Final to full collapse(!)

A less-known ice sheet melting feedback (suggested by three researchers from the Potsdam Institute for Climate Impact Research, led by Alexander Robinson and published (2012) in Nature Climate Change) has to do with the sheer volume of the ice sheet – or actually it’s height. As the ice sheet slowly melts from the top down (where it has contact with the above-zero atmosphere) the altitude of the ice mass surface declines. This altitude is an important climate factor in itself – a 3-kilometre high ice mass is in fact a proper mountain range. Yet every 100 metres of declining altitude can raise year-average air temperatures over the ice mass by another degree, thereby accelerating the warming, the ice loss – and the caused sea level rise. In the meanwhile net precipitation over the ice sheet (in the form of snow, leading to accumulation) would also decline as the ice sheet loses height – just like ultimately the snow-rain ratio. (The decreasing ice sheet height feedback could create a full-melting tipping point for the Greenland ice sheet as low as 1.6 degrees – the above authors have suggested.)

4 & 5: ICE SHELF COLLAPSE FEEDBACKS
Here we take two in one: the (basic) ocean warming and sea level rise* ice shelf feedback. (There are more complicated ice shelf feedbacks, as we’ll find out later!)

Both ocean warming and sea level rise itself [*) this one is more complicated – to properly understand, also see next one, about bipolar seesaw mechanisms] can also create powerful positive ice sheet melting feedbacks, in this case more (exclusively) for the Antarctic ice sheet – that extents into the Southern Ocean, where it’s bordered and protected by ice shelves – very thick plates of sea ice (or rather ‘floating glacier ice’ – not to be confused with actual (seasonal) sea ice).

On some key locations (the ‘grounding line’ – official border between ice shelf and ice sheet) these extending ice shelves touch the ocean floor – for instance over the Ross Sea (at the border of the West and East Antarctic ice sheets) and the Amundsen Sea (that is covered by the two-least stable Antarctic glaciers, the Pine Island Glacier and the Twaites Glacier). If these ice shelves are lifted up and or melted in warmer waters they will disintegrate – and uncork an increase in glacier flow towards the oceans, leading to accelerated sea level rise.

In 2014 a group of scientists led by Nick Golledge of the Antarctic Research Centre of Victoria University of Wellington published findings in Nature, explaining they had found evidence of massive pulses of rapid sea level rise during the Pleistocene-Holocene that they attributed to similar ocean-ice sheet feedbacks. In the observed real world clear examples are found of collapsing protective ice shelves around the West Antarctic Peninsula, for instance the iconic collapse of the enormous Larsen B shelf into the Weddell Sea in 2002 (following the collapse of the smaller Larsen A ice shelf in 1995).

In this Weddell Sea the ocean warming feedback could play a key role in ice shelf collapse, as here increasing circumpolar winds (again, another positive climate-melting feedback in itself – one that relates to both climate warming and ozone recovery) create a sea current that leads the warmer surface waters directly under the ice shelf.

In May 2015, a research group led by Rolf Jansen of the AWI Helmholtz Centre for Polar and Marine Research, published evidence in The Cryosphere, showing the larger Larsen C ice shelf (closer to the South Pole) was becoming unstable due to a combination of surface melting and warm water melting at its base – and that its full disintegration seemed a matter of time, a disintegration that the researchers predict will be followed by further recession of the feeding glacier. In August 2016 the same researchers published an update stating the crack in the Larsen C shelf had grown to 22 kilometres longer over the length of the Antarctic winter night (the period they could not observe it – also the period that should be very cold and stop melting(!)) and is now 130 kilometres long at the onset of the southern hemisphere spring.

On the opposing side of the Antarctic Peninsula there has also been a (partial) collapse of a large ice shelf in recent years: the Wilkins shelf in 2008 – a collapse that also continued in the Antarctic winter, the European Space Agency reported.

The wind-enhanced warm water melting feedback is also observed under the Amundsen Sea ice shelves. In 2014 a research group led by Sunke Schmidtko of the University of East Anglia published in Science that they had observed the waters under the ice shelves of the West-Antarctic Bellinghausen and Amundsen Sea have steadily become both warmer and saltier over a timeframe of 4 decades, adding that shelf basal melt caused by this ocean water is of larger importance than surface melting. The ice shelves in the Amundsen Sea are important, because they are connected to two large glaciers (Pine Island & Twaites Glacier) that have dramatically increased in flow velocity.

In 2014 a research group led by Eric Rignot of the University of California and NASA’s Jet Propulsion Laboratory explained in Geophysical Research Letters how the Pine Island and the Twaites Glacier, together with four other large West-Antarctic glaciers had picked up speed – stating the retreat and thereby final collapse of the entire West Antarctic ice sheet had become ‘unstoppable’. Key to the entire West Antarctic Ice Sheet would be the Twaites Glacier, that partially holds all others in place. Also in 2014 another research group, led by Ian Joughin of the University of Washington, had modelled the Twaites Glacier and confirmed the NASA study – stating in a separate publication in Science that full collapse of this key West-Antarctic glacier could ‘potentially already be underway’ and could happen in two centuries time.

Natural Antarctic ice sheet and ice shelf dynamics. Image made by Hannes Grobe of the Alfred Wegener Institute for Polar and Marine Research.

The sea level rise ice shelf feedback of course first requires sea level rise to occur – possibly an acceleration. If that happens other climatically stabler margins of Antarctica could also be prone to ice shelf collapse, for instance the very large shelves over the Ross Sea. Following ice shelf collapses around the Antarctic Peninsula in 2002 researchers of the US National Snow and Ice Data Center have investigated the larger shelves over the Ross Sea and found in summer ice sheet conditions are only ‘a few degrees too cool’ for the start-up of a similar disintegration process – also noting that (around the edges) warming had been very rapid, at +0,5 degrees Celsius per decade.

In February 2016 a study led by Yusuke Yokoyama of the University of Tokyo was published in PNAS that presented sedimentary evidence that in our modern geological period the Holocene the Ross ice shelf had in fact been a far larger ice sheet, stretching across the sea bottom to the very edge of the continental shelf, until it had been lifted up by the sea and eventually become unstable and partially collapsed. The big similarity between this late-Holocene collapse and the present condition is that the furthest extent of the Ross ice shelf is about a 1000 kilometres from the grounding line, again opening a giant wedge of sea water and a potential to lift it up from the locations where the ice sheet is now pinned by some minor protective morphology on the sea floor. Other research confirms previous collapses (also Pleistocene and Pliocene) of the Ross ice sheet & ice shelves, for instance this climate-model study led by David Pollard and published in Nature in 2008.)

Now here, when we’re suggesting ice shelf collapse following sea level change, it’s good to add a note about the gravitational effect of ice sheet melting – the pull of ice masses on surrounding ocean waters due to the immense mass of multi-kilometre thick glaciers. As ice sheets decline in volume and mass (height in particular), their gravitational pull on surrounding ocean waters declines, which could in fact lead to local sea level lowering (possibly increasing shelf stability).

Why ice shelf uplifting due to sea level rise is possible:
In the long run Antarctic melting will cause a sea level drop around the Antarctic coasts – and sea level rise everywhere else across the globe (most notably on the northern hemisphere). In the initiating phase however very different processes could come into play:

First a potential local height increase of the East-Antarctic icesheet (the higher the mass, the stronger the gravitational pull). Research shows that although the Antarctic icesheet as a whole is losing mass, this does not (yet) apply for the East-Antarctic ice sheet: in 2012 a big group of researchers led by Andrew Shepherd of the University of Leeds published in Science an elaborate Antarctic mass balance for the period of 1992-2011: in that time the West Antarctic ice sheet had lost (best estimates) 65 gigatonnes of ice per year and the Antarctic Peninsula another 20Gt/y – but the East Antarctic had in fact gained mass, albeit a smaller number: around 14 gigatonnes. What’s possibly more important (for gravitational pull) is that this extra mass is not gained around the edges, but on top of the ice sheet, due to an increase in snowfall. And indeed that’s what another study found, led by Curt Davis of the University of Missouri-Columbia and published in 2005 in Science: a thickening of the East Antarctic ice sheet of 18 millimetres/year between 1992 and 2003 (while West Antarctica thinned). This ice sheet height increase could favour a local sea level increase, and local ice shelf uplifting.

Secondly: if Greenland melts quickly, while Antarctica comes as a delayed response, sea levels will first rise rapidly around Antarctica (as it’s furthest from the Greenland, where the gravitational pull declines the fastest). The above-mentioned Shepherd study in Science indeed also finds the Greenland ice sheet has lost a far larger amount of ice mass over the same period: an estimated 142 gigatonnes per year (precisely twice the net amount for Antarctica).

Thirdly Antarctic sea level rise can be promoted by locally rapid ocean warming, leading to local amplification of thermal expansion, fourthly by an ITCZ shift towards the southern hemisphere (trade wind water build up) and fifthly by a decrease in bottom water turnover (due to meltwater increase, cooling decrease or disturbed saltwater formation).

These three are elaborated below in our summary of possible mechanisms behind hemispheric asynchrony in ice sheet dynamics and sea level rise:

6-8: BIPOLAR SEESAW FEEDBACKS
Greenland-Antarctic bipolar seesaw, Dansgaard-Oeschger & Heinrich Events: hemispheric asynchrony as a mechanism for accelerating net ice sheet disintegration and sea level rise

Here we pile together some different terminology for additional paleo-climate-deduced mechanisms for (rapid) ice sheet collapses, some of which are more empirical evidence of the possible nature of rapid ice sheet changes and fluxes in sea level rise, and not conclusively linked to specific climate feedbacks.

There is a chance that Earth’s climate change will show increasing hemispheric asymmetry. In fact there is already asymmetry developing under the current greenhouse gas-induced warming – that measurably happens much faster over the northern hemisphere land masses and (due to the sea-ice albedo feedback (‘(North) Polar Amplification’)) the rapidly warming Arctic Ocean.

In ultimate form this asymmetry can manifest itself as an opposing phase-difference called the bipolar seesaw – whereby the north and south pole have alternating climatic phases – fluctuations that may be stronger or weaker than the underlying trend (which is a global trend of net warming, ice mass loss and sea level rise).

An interesting analysis by Jeffrey Severinghaus of the Scripps Institution of Oceanography, published in Nature in 2009, found that if you’d filter out the greenhouse gas-forced warming trend (that can of course be witnessed globally) in the global temperature dataset, you uncover a bipolar seesaw pattern in the 20^th century.

These seesaw patterns work through the Meridional Overturning Circulation (MOC) – also referred to as the thermohaline circulation, the world’s connected system of ocean currents. Through formation of North-Atlantic Deep Water and Antarctic Bottom Water (formation that can both increase and decrease due to changes in temperature and salinity of the ocean water, hence thermo & haline) – there’s a suggested North-Atlantic MOC switch around Iceland & Greenland and a separate Southern Ocean MOC switch.

The Meridional Overturning Circulation, or thermahaline circulation, driven by temperature & salinity around the poles – and by trade winds in the tropics. This beautiful system of warm and cold, subsurface and deep ocean currents is also referred to as the ‘Great ocean conveyor belt.’ Possible changes in this system will likely have very large climatic consequences.

Factors that increase deep and bottom water formation are relative cooling (of poleward surface currents, like the North Atlantic Gulf Stream) – remember water is densest/heaviest around 4 degrees Celsius – and salinity increases (once temperatures drop further, to below -4 degrees Celsius and sea ice formation starts in the ocean top layer, a process that produces brine, very heavy water).

Of course the climate warming decreases both factors – first a decrease of relative cooling (due to a decreasing temperature gradient between the poles and the tropics) and then also a decrease in salinity, due to declining sea ice formation and an increase of seasonal meltwater influx, both from sea ice and runoff from glaciers.

The bipolar seesaw is also recognised in paleoclimatic research – also at a longer timescale (±1.500 years). Here bipolar mechanisms are associated with abrupt climate change and pulses of rapid sea level change, with large geographical changes – the Younger Dryas as a relatively well-known example.

The Younger Dryas was a climatic hiccup at the boundary of the Pleistocene and Holocene – possibly caused by gradual warming coming from the Last Glacial Maximum that led to a large-scale (threshold) influx of fresh melting water in the North Atlantic, shutdown (either abrupt or gradual) of the Gulf Stream and strong weakening of the Meridional Overturning Circulation. As a result, within a very short time (mere decades – according to the disputed Lake Agassiz hypothesis possibly less, years) the northern hemisphere returned to ice age conditions – that continued for another 1,500 years (a so-called ‘stadial’ if you want to know the correct climate term).

A Gulf Stream shutdown can enhance Antarctic ice shelf collapse
Point for now is something else: that following a Gulf Stream shutdown the southern hemisphere did not cool – in fact to the contrary: in 2013 a special publication by Randall Carlson of Oregon State University in Encyclopedia of Quaternary Science presented an overview of evidence (from various studies) of wide-spread Younger Dryas warming in the southern hemisphere.

Possible reason is that the direct consequence of Gulf Stream shutdown (rapid cooling of the North Atlantic) leads to an increased tropical-polar temperature gradient, that can shift the average location of the Intertropical Convergence Zone (ITCZ) and thereby the trade winds, that are another engine (apart from the polar deep/bottom water formation) to the Meridional Overturning Circulation. We’ve written about the mechanism before here at Bitsofscience.org, for instance in our piece that connects current Arctic warming to Amazon drying (where a decreased boreal temperature gradient sucks the ITCZ to the northern hemisphere).

In 2015 a research group led by Paul Goddard of the University of Arizona reported in Nature how indeed when trade wind-driven accumulation continues while deep/bottom water turnover decreases, this can lead to very rapid local sea level rise – as witnessed on the US East Coast over the years 2009-2010 – with a sudden jump of the coastal seas north of New York by 128 millimetres, following a temporary decline in the Atlantic Meridional Overturning Circulation (AMOC).

A long introduction to illustrate what can happen due to a (meltwater-induced) slowdown of MOC: it can push the tropics to the southern hemisphere, creating a trade-wind-driven build-up of water that will ultimately also influence the Southern Ocean – leading to possible sea level rise (due to accumulation, local thermal expansion and (thirdly) possibly decreased bottom water turnover) around Antarctic ice shelves, for instance under the Ross Sea, promoting collapse scenarios. (Meanwhile of course ocean temperature Antarctic melting would also increase.)

Paleoclimatic evidence of asynchronous ice sheet melting:
Dansgaard-Oeschger Events and Heinrich Events are other observed examples of hemispheric asynchrony in ice sheet behaviour – and opposite phases to the above-described seesaw – so with rapid warming/ice sheet melting on the northern hemisphere.

Heinrich events are moments of rapid northern hemisphere ice sheet disintegration during ice ages, that occurred after pauses of several thousands of years – and have been deduced from very large-scale iceberg formations in the North Atlantic (and the sediments these left behind). The causes of Heinrich events are not yet fully understood. Some suggest ice sheet may have internal physical limits to size and weight – naturally collapsing beyond certain thresholds (the so-called “binge-purge” model).

Other authors, for instance Mark Maslin of University College London in Geophysical Monograph in 2001 suggest a hemispherical seesaw pattern for Heinrich events too – and more specifically a sea level rise ice shelf feedback, noting on the northern hemisphere melting (ice sheet disintegration) would have preceded (albedo-feedback) warming. [Please note in the Pleistocene the northern hemisphere still had ice sheets bordered by ice shelves – for today’s world this again serves as an analogue for the Antarctic situation.]

Moreover Maslin thinks that Heinrich events, following AMOC collapse, lead to local North Atlantic cooling and thereby ITCZ shift away from the equator towards the southern hemisphere. In a separate publication in the Journal of Quaternary Science in 2001 together with lead author Dan Seidov of the Pennsylvania State University they call this ‘(Atlantic Ocean) heat piracy’ – and a very clear example of a bipolar climate seesaw – which they state did not only occur during Heinrich events, but also during Dansgaard-Oeschger events.

Dansgaard-Oeschger events are similarly rapid climate fluctuations that happened 25 times during the last glacial period and that seem to correspond to the Holocene Bond events (recurring every 1,000-1,500 years, with the Medieval Warm Period followed by the Little Ice Age listed).

D-O events manifest itself as a rapid (decades) onset of northern hemisphere warming, followed by slow (centuries) northern hemisphere cooling. Richard Alley of the Pennsylvania State University wrote in PNAS in 2000 that the warming phase was both rapid and strong over the Greenland ice sheet: a warming of 8 degrees, is three subsequent steps of 5 years time(!)

On the southern hemisphere the fluctuation is far less extreme – and could again show a seesaw phase difference, as suggested by Thomas Stocker of the University of Bern in Paleoceanography in 2003.

The causes of Dansgaard-Oeschger events are still unclear and debated – but they serve to remind of episodes of very rapid climate and ice sheet instability, possibly following just small threshold triggers.

Concluding: there is large paleoclimatic evidence against linear ice sheet-climate interaction – abrupt, non-linear and asynchronous responses of ice sheets, leading to net acceleration of warming-induced sea level rise should be considered

We draw the conclusion from bipolar seesaw mechanisms and paleoclimatic (opposing or single hemisphere) rapid climate shifts and sea level changes like the Dansgaard-Oeschger and Heinrich events that following a structural globally forced long-term climate change (Pleistocene-Holocene, possibly Holocene-Anthropocene) long-term ice sheet melting and sea level rise can be sped up by local disturbances, including paradoxically relative (single hemisphere) cooling following (A)MOC changes – as these could amplify warming and sea level rise effects elsewhere, leading to further collapse, fuelling the structural change.

As the events are paleoclimate-deduced and not fully linked to feedback mechanisms in today’s world’s climate system, historic seesaw and single-polar sea level events are indicative only, another illustration to show how ice sheets and global sea levels can show non-linear and also rapid responses following possible thresholds in structural global climate change.

9: VOLCANIC ICE MELTING FEEDBACK
Isostatic rebound volcanology melting (and CO2) feedback

Rule of thumb: once you start disturbing stuff on a global scale, also expect responses on a global scale. This is why we included a possible geological climate-ice melting feedback, one proposed by Carolina Pagli of the University of Luxembourg and Freysteinn Sigmundsson of the University of Iceland in Geophysical Research Letters in 2008.

The largest ice sheet (8000km2) of Iceland, Vatnajökull, has lost 10 percent of its mass over the last century. Modelling work by Pagli suggest the isostatic rebound this causes (land under and around melting glaciers and ice sheets rises, also causing earthquakes) leads to a significant increase of magma production (0.014km3/y) under that ice sheet – and ‘unusual magma movements’.

Now Iceland is of course formed on the mid-Atlantic ridge – but what about the world’s larger ice sheets? Greenland in its current state is not very volcanically active – Antarctica could be a different story though. Here no less then 37 volcanoes have been distinguished below the massive ice sheet, of which two are currently active. If Antarctic ice sheet mass loss would also lead to an isostatic rebound volcanic feedback, that could in turn trigger other melting feedbacks, possibly through meltwater production – and lead to acceleration of global sea level rise.

Now it’s time to get to the new kids on the ice sheet feedbacks block, the DeConto Nature study – and the Hansen study in Atmospheric Chemistry and Physics, both dating (the peer-reviewed versions that is) from 2016:

10 & 11: ICE CLIFF COLLAPSE (& HYDROFRACTURING OF ICE SHELVES)
Again, we pile together two ice sheet melting feedbacks (as they’re never as clearly defined) – both illustrating a mechanism for increased glacier flow and ice melting at the ice sheet marine margin.

Here we get to the publication by Robert DeConto of University of Massachusetts Amherst and David Pollard of in Nature of March 31 2016. In their introduction DeConto and Pollard first briefly point to the paleoclimatic sea level record – both the most recent interglacial (see our coverage of Eemian sea level rise) when temperatures where about as high as they are today (and sea levels about 6-9 metres higher) – and the most recent longer geological time period without major glaciation, the Pliocene, during which the CO2 concentration was close to the current level (±400ppm) but temperatures were 2-3 degrees higher (than pre-industrial Holocene!), and sea levels 12 to 32 metres higher [please read our special about Pliocene paleoclimate and Earth System Sensitivity climate inertia to understand why at 400ppm CO2 global average temperatures are likely to keep rising (by 1-2 degrees Celsius extra).

Taken from another study by David Pollard, Robert DeConto and Richard Alley (2014, Earth and Planetary Science Letters) this image serves to show the (reverse-sloped) Antarctic basins, large ice shelves over coastal seas (for instance Ross Ice Shelf) and the location of relatively vulnerable Antarctic glaciers, mentioned elsewhere in this article – like the Twaites and Pine Island Glacier.

Looking for mechanisms to explain such large-scale sea level responses to seemingly small temperature fluctuations, DeConto and Pollard tried to refine existing ice sheet models – incorporating two previously overlooked/underestimated factors for ice sheet dynamics, as warming-ice melting feedbacks. When including the two mechanisms, they can replicate both interglacial and preglacial ice sheet and sea level temperature responses:

Buttressing ice shelves, a stabilising factor
In our special paragraph about ice shelf feedbacks (both ocean warming and sea level rise induced, see above) we’ve addressed the specific morphology of Antarctic ice shelves, the portions of large glaciers that extent into the Southern Ocean, where essentially they start to float (yet can continue for a long time, for instance up to 1,000 kilometres over the Ross Sea between the West and East Antarctic).

The place where the ice shelves definitively touch the ocean floor is called the grounding line, which is also the official border between (floating) ice shelf and (land-based) ice sheet (a bit artificial, as both stem from the same glacier).

Ice shelves however can touch the ocean floor at other points over the shallow continental (land) shelf of Antarctica – and these places (also discussed in the ice shelf paragraph) can also be of special interest, as such sea floor morphology can pin the ice sheet, stabilising both the ice sheet and its feeding glacier.

Now of course any stabilising factor, once decreasing, becomes a destabilising mechanism – and when that mechanism is climate warming or ice melting induced, by definition also a positive feedback, leading to acceleration of warming, sea level rise – or both:

Ice shelf meltwater hydrofracturing & speed increase
At places where the ice shelves are not just ‘free floating thicker versions of sea ice’, but actually create friction (with their size and mass on the water, possibly attachments to sea floor morphology, they create friction – and in that sense buttressing to the glacier flow of the ice sheet behind, slowing its flow to the ocean. DeConto and Pollard in their model simulations find that these buttressing ice shelves can be subject to yet another climate warming-melting feedback – one that has a bit of several we discussed higher up in this article: an element of albedo effect, ice shelf thinning & collapse – and another key role for meltwater somewhere in between, not as a glacier lubricant, but as a fracturing agent.

A factor that is often overlooked in modelling of marine ice sheet instability, the authors state, is vulnerability to direct atmospheric warming – as summer temperatures (also on the East Antarctic margins) are increasingly reaching values above the freezing point of 0 degrees Celsius,

The meltwater that further ice shelf surface thawing (and increased summer rainfall) is set to produce, will perculate through the ice sheet, and [once it refreezes – remember freezing of water is an exergonic phase transition] warm the ice sheet. That in turn will reduce the viscosity of the ice and speed up the flow. Rain and meltwater will also increase hydrofracturing, as has been observed during the Larsen B ice shelf collapse.

Ice sheet cliff height increase + collapse feedback
Once vulnerable ice shelves are thinning, speeding up ice flow and collapsing – the authors predict another melting feedback will pick up. Where buttressing outer lines of ice shelves disappear, ice sheet glaciers will form increasingly high calving fronts at the marine edges – and these glacier fronts are subject to specific laws of physics, DeConto and Pollard argue, stating they collapse at points where ice shelf thickness reaches 1 kilometre, which would imply more than 90 metres of height from the sea level – “because longitudinal stresses at the cliff face would exceed the yield strength of the ice”.

Worryingly, such conditions exist at many places around the current grounding line of Antarctic ice shelves and once (following cliff collapse) these glaciers would retreat, the feedback loop could become even stronger, as currently many Antarctic ice shelves are grounded on reverse-sloped beds, meaning there are (sometimes water-filled) basins deeper under the ice sheet – and that ice cliff height would generally increase further (over the critical limit of 90 metres) under an ice sheet retreat scenario, creating a true runaway melting loop.

[The authors compare future Antarctic marine ice sheet margins (structural cliff collapse over areas with reverse-sloped beds) to the (rapidly eroding) glacier fronts of Helheim and Jakobshavn glacier on Greenland today.]

Incorporating meltwater-ice-shelf feedbacks and structural ice cliff collapse feedbacks not only enables DeConto and Pollard to replicate paleoclimatic changes to the Antarctic ice sheet – they are also surprisingly concrete about the 21^st century implications, stating Antarctica could contribute 114 centimetres of additional sea level to the ‘business as usual’ RCP8.5 emissions scenario [using conservative Pliocene sea level rise reproduction, of >10m (remember research shows Pliocene sea levels were 12-32 metres higher(!))]. Under that same scenario the onset of major ice sheet retreat would then occur around 2050.

“When applied to future scenarios with high greenhouse gas emissions, our palaeo-filtered model ensembles show the potential for Antarctica to contribute >1 m of Global Mean Sea Level (GMSL) rise by the end of this century, and >15 m metres of GMSL rise in the next 500 years.”

[You can add another 1 metre to that 2100 projection coming from Greenland, thermal expansion, and smaller glaciers – to see a +2 metre scenario playing out.]

12 & 13: SOUTHERN OCEAN ‘COLD LID’ MELTWATER FEEDBACK
We get to the new Hansen study and can again pile two additional amplifying feedbacks together: a Southern Ocean meltwater surface ‘cold lid’ that promotes heat build up directly under ice shelves (and increases the Earth’s energy imbalance, amplifying the net warming)

James Hansen argues currently used climate models are still not properly tuned to predict ocean current and ice sheet behaviour following a global temperature rise of several degrees within this century. He states – again, together with 18 co-authors behind the publication of March 22 2016 in Atmospheric Chemistry and Physics, people including his Columbia University colleague Makiko Sato – that including more detailed feedbacks (that they discovered) to the equation illustrates far larger ice sheet vulnerability, again bringing models better in line with paleoclimatic sea level rise reconstructions (like the Eemian interglacial) and lifting 21^st-century sea level rise forecasts by as much as 2 to 5 metres.

Just like DeConto, Hansen argues the key is the Antarctic ice sheet – around which we risk to ignore yet another important (marine) ice melting feedback.

We’ve discussed in this same article (see the bipolar seesaw paragraph) how changes to the Meridional Overturning Circulation (thermohaline circulation, connected ocean currents) can simultaneously promote local cooling and ice sheet collapse.

An overturning slowdown in the Southern Ocean around Antarctica will have larger consequences still, Hansen explaines – as we should also include the heat storage factor:

Normally around Antarctica ocean water (coming from the warmer, central Atlantic) cools strongly – and then partially freezes creating sea ice on top and leaking brine in the remaining fluid. Both factors (cooling compaction and increasing salinity) increase the weight of ocean water around Antarctica, creating the strongest engine of the entire Meridional Overturning Circulation: the formation of Antarctic bottom water, water that sinks straight to the bottom of the Southern Ocean, where it feeds other ocean (bottom) currents across the globe. The bottom water formation around Antarctica is so strong however, that it is also compensated by columns of warm, light ocean water rising to the surface (that in turn feed surface ocean currents, also running away to smaller latitudes).

You can compare this to the general circulation in the atmosphere. On the marcoscale rising air in the tropics (dominant low pressure systems of the ITCZ) feed the Hadley Cell. Up high this air flows to the higher latitudes of the subtropics (where high pressure systems form. However: look at any low pressure system in detail (for instance your own backyard) and you find the process also partially compensates at microscale: columns of sinking air next to thunderstorms, blue sky next to a (convection, cumulus) rain cloud, etc. Imagine similar pillars of sinking and rising water in the Southern Ocean.

Now this warm surface water is important for the climate system, Hansen explaines: if heat is high in the oceans a portion is transferred to the atmosphere and from there a portion is leaked to the cosmos – essentially a natural escape route for solar energy.

In the warming climate however, both compacting factors will likely decrease, less ocean water cooling and less sea formation (as we’ve discussed in our paragraph about Greenland Deep Water formation) – and more importantly: once the Antarctic ice shelves and glaciers start melting, which has clearly begun, a rapidly increasing influx of fresh (light) meltwater (that is also cold).

This process not only weakens the bottom water turnover (as many have described) it also places ‘a light cold water lid’ on the Southern Ocean – reducing heat exchange with the surface and increasing heat build-up in the ocean, “raising the temperature of ocean water at the depth of ice shelves” – an amplifying feedback to ice shelf collapse and meltwater formation.

Actual observations of sea level rise – does Moore’s law apply?
In their new studies Hansen and his colleagues have not only looked to replicate the paleoclimatic ice sheet sensitivity with their adjusted ice sheet-ocean climate model, they’ve also tried to link their findings to the currently observed global sea level rise.

This is a dataset that is relatively young and noisy, filled with regional fluctuations (see 2009-2010 New York coast as described above) so uncertainty margins are high. However in the observations a clear acceleration is visible, the authors state – especially for the Greenland ice sheet.

Fitting within the observational data Greenland mass loss could have a 10 year or a 20 year doubling time. As it is mostly surface melting (also promoting glacier speed, see our paragraph about meltwater lubrication feedback) the characteristic of Greenland melting is very different from that of Antarctica – where the buttressing ice shelves play a key role – and their disappearance leads to possible strong acceleration of ice sheet mass loss. If Antarctic ice sheet mass loss has a 10-year doubling time, ‘the metre scale sea level rise would be reached in about 50 years,’ Hansen says – ‘and multi-metre sea level rise a decade later.’ [Yes, that’s how amplification works in practice(!)]

If Antarctic ice sheet loss would correspond to a 20-year doubling time, the metre scale would take about a hundred years.

Worryingly Hansen and his colleagues also provide evidence to show that the real world seems to be structurally ahead of the climate models. One important factor they say is that models are poorly equipped to represent ocean temperature stratification – creating excessive ocean heat mixing. We can see from our observations [please include the La Niña-dominant ‘temperature plateau,’ a clear example – and (as Hansen notes) the North-Atlantic cold blob].

When climate models assume easy & uniform ocean mixing they are also likely to underestimate the importance of the cold lid meltwater ice sheet melting (& Earth warming) feedback – Hansen concludes.

To be continued…

[If you’ve read this lengthy article to the end, we salute you – and are thankful for your time. If you think (despite the inescapable imperfections on our side) you’ve learned something from it and think others can too, please care to pass it on. Thank you very much.]

© Rolf Schuttenhelm | www.bitsofscience.org