Monday 25 July 2011

Arctic sea-ice free this decade?

As I outlined in my post In Flux, a key finding concerning the Arctic has made me wonder about whether we are indeed on a fast-track to a seasonally sea-ice free Arctic this decade. However on reflection that is in part an emotional reaction to an observation that unsettles me. I had been going to wait until September to post on this subject, so I could see what happens this year: I still see 2011 as having the makings of a spectacular season. But in the light of Maslanik's latest paper, and of the points raised by two commenters on the 'In Flux'page, I thought it might be as well to state in detail why, on balance, I don't think we are on a fast-track, and why I think key observations are being misinterpreted.

Update - I have since changed my view on this as outlined in this post.

In the In Flux post I've explained my basic disagreement with the extrapolation of volume method and what it implies, so I won't go over that again here.

To start, it's as well to go over the paper by Tietsche et al which was raised by Kevin O'Neill on the In Flux comments.

What the authors do is use a climate model to examine the factors behind the recovery of sea ice in a climate model. They remove all the sea ice on 1 July at several points in the 21st century, the study finds that the factors are similar for each period, so the study concentrates on what happens after the sea ice is removed on 1 July 2020.

Figure 3 of Tietsche et al, 2011, "Recovery mechanisms of Arctic summer sea ice."

The graph deals in net fluxes, so taking the example of sunlight, the yellow line (known as shortwave radiation); immediately upon removal of the sea-ice there is an increase in shortwave. This is not due to an increase in incoming sunlight but due to the darker ocean being exposed. Because the ice is white it reflects back most of the shortwave that hits it, so it reduces net shortwave. However the ocean absorbs most of the shortwave that hits it, this means less shortwave is reflected back into space, so it increases net shortwave. Much of the ocean warming is due to the absence of ice, normally melting that acts as a heat-sink. This leads to an increase in net energy absorbed by the Arctic region (black line). Gradually as September is aproached the insolation falls, so that by September fluxes are near zero. But once the sun sets for the long Arctic night the ocean, which had warmed substantially during the summer loses much of it's heat.

Into the autumn the ocean emits sensible heat (heat you can feel or 'sense'), latent heat (heat due to evaporation and the phase transition from water to ice), and long-wave (or infra-red) radiation. By November sufficient heat has been lost by the ocean for it to cool to freezing and begin to freeze.

Despite the extreme nature of the experiment, removing all sea ice abruptly at a peak season in the summer. Tietsche et al find that after only two years the sea surface temperatures have returned to range of natural variability, and the sea-ice rebounds. For surface air temperatures they find a large increase in temperature between October and February that mainly affects the lower troposphere, this is similar to the observational findings of Serreze et al 2009 for recent changes in the Arctic.

Tietsche et al answer the question: What happens to the energy gained by the Arctic following a simulated loss of sea-ice? They find that the energy is mainly lost to the atmosphere in the following freeze period and thence radiated to space, or in warming the atmosphere reduces atmospheric heat fluxes into the Arctic.

Tietsche et al also note that:
"Between 2000 and 2040, when the rate of decline is
maximal, Arctic summer sea-ice extent exhibits strong year-
to-year fluctuations. As noted by Holland et al. [2008] and
Notz [2009], this increase in variability is mostly due to
changes in the ice thickness distribution and does not
necessarily indicate proximity to some critical threshold."

I think that what we have seen in reality is similar to the process Tietsche et al found in their model. 2007 and a massive loss of ice, in both area/extent and thickness implied by loss of MY ice. Then 2008, 2009, 2010, all offset to a lower extent by 2007 yet not showing similar losses i.e. Cryosphere Today Area. And the energy gains of those years causing - not further declines in area - but a delay to the re-freeze as shown by the low anomalies after the summer melt season i.e. Cryosphere Today Anomaly. It's tempting in view of this study to view the large late year negative anomalies as being the process identified playing out through 2007, to a lesser degree 2008 and to an even lesser degree 2009 and 2010. Although it should be accepted that we've had only one major drop (2007), and there were also anomalously high extents of open water through the following summers.

The continuing volume decline shown by PIOMAS does tell us that some of the energy gained in recent years is being used to melt ice, and this at present apears to be accelerating. However if I'm correct in what I outline, what we will see in the coming years will not be a collapse to a seasonally sea-ice free state, but a reduction in the rate of volume loss. That said, due to certain modelling studies I am expecting further losses like 2007, that will be lead by weather events. I'm in the process of writing about three key modelling studies at present.

It still bothers me that the accelerating volume loss is occurring as the MY ice is stabilising, but there is one personal quality the Arctic has taught me since 2007 - patience.

So on balance, as I see it, the main problem is that people seem to be confusing the effects of thinner ice cover with the suggested intial symptoms of a rapid transition to a seasonally sea-ice free state.

Maslanik's recent study, discussed in the preceding post, states:
"The recovery in multiyear ice extent through March 2011
from the extreme reductions in 2007 and 2008 along with
the continued aging of the surviving ice through multiple
melt seasons is consistent with an ice pack that has not
passed a tipping point across the Arctic Ocean as a whole..."

As I've said before, formerly the ice-pack had a large amount of multi-year (MY) ice at it's core. This core, stabilised the ice pack by virtue of it's bulk, mechanical strength, and the lack of brine inclusions in the ice. As the ice ages it loses it's brine inclusions which both strengthens it and increases it's resistance to melt, and it ridges forming ice much thicker than that which grows thermodynamically. That old ice-pack is now gone as I have shown in the earlier posts on MY sea ice.

So what has happened in the first decade of the twenty first century is that the ice has transitioned to a younger thinner ice pack which is less stable than the old pack was, (e.g. Maslanik et al, 2007). This process has been ongoing for decades before the final transitional period, but we have now gone through the transition itself, which did turn out to appear like a non-linear threshold (i.e. Nghiem 2007 - see here) From my perspective 2007 is an outcome of this process of transition, not a herald of a rapid transition to a seasonally sea-ice free state. What happened in 2007 was an episode of weather 'taking advantage' of the decades long process of loss of MY sea ice, and in the process removing more of the MY ice.

In terms of what is going on in the Arctic I think that the prognosis is the single most important aspect. It's crucial too in terms of the wider public debate about Climate Change, not the faux-debate about whether it's real and ongoing - there is no such debate, but in terms of how serious Climate Change is and how we should respond. It is also crucial that claims about the future should not put us in a position whereby the public gets tired of overstatement and flawed prognostication. It is only by discussing the process and it's mechanisms that we can better understand what is going to happen and what public statements to make about that.

If the idea that the Arctic will be sea-ice free by around 2016 becomes widespread, and I think it already has: Come 2016 when there is no sign of a crash to sea-ice free state, what are people going to say about it to a wary public, tired of overstatement and perceived alarmism?

I know what I'll say if it does happen - I was wrong, this is way worse than I thought it was.

Maslanik et al, 2007, "A younger, thinner Arctic ice cover: Increased potential for rapid, extensive sea-ice loss."
Maslanik et al, 2011, "Distribution and trends in Arctic sea ice age through spring 2011."
Serreze et al, 2009, "The emergence of surface-based Arctic amplification."
Tietsche et al, 2011, "Recovery mechanisms of Arctic summer sea ice."


Peter said...

The Tietsche paper shows that after a random weather-driven loss, extent recovers back up to the long-term trendline within 2-3 years. It tells us next to nothing about the processes shaping that long-term trendline itself, and how fast it will head towards zero.

Put another way, after a "bad year" for Arctic melt, you get all the recovery you're ever going to get within the next 2-3 years. This is what we've seen - summer extent did indeed recover from 2007 through to 2009, all the way back up to the trend. However, the trend itself is downward, and accelerating.

2010 was lower than 2009: 2011 will be lower than 2010 unless something unprecedented happens in the next few weeks. In fact, it looks very likely that 2011 will be within spitting distance of the 2007 record, exactly in accordance with the quadratic and/or Gompertz predictions.

What qualified as a record low only 4 years ago is now the average, middle-of-the-road expectation from an average year.

Peter said...

To add: it probably tells us something about how quickly we may or may not transition from a seasonally ice-free state to a year-round ice-free state (i.e. not likely for a long time).

Dan Satterfield said...

Good analysis and inter4esting comment by Peter as well.. I always thought that the models forecasting a slower trend toward a seasonal ice free state were being dismissed a bit abruptly. That may just be the meteorologist in me talking.

You are right that statements about the ice disappearing in the summer by 2016 will be used by skeptics against the science and with significant success.

Chris Reynolds said...


Thanks for your considered comment.

What the Tietsche paper shows is that after a perturbation of ice extent, even one as extreme as the total removal of ice, the absence of sea-ice in the autumn leads to a loss of most of the energy gained that year. In my opinion this is a very important message for what we are seeing now. Although there is a lot of talk of the powerful ice-albedo feedback, a positive feedback. There is much less talk about the ice-thickness feedback. Tietsche shows it's power.

I agree that the long term trend is downward. As I noted in my earlier post on Maslanik's paper, according to Bitz sea-ice experiences net melt when it's thickness is above the equilibrium thickness and net growth when it's thickess is below equilibrium thickness. As the equilibrium point is moving towards a warmer Arctic this is driving the process of ice loss.

The question is how fast will this process proceed?

I posted 2 quotes that show that there is evidence that the process we are seeing is not part of a rapid transition. The statistical methods are risky because they don't take into account the physics of the process, I commented on Tietsche to try to show that the physics argue against statistical extrapolation. In any case given the error ranges involved (i.e. Tamino's recent update to his projection - see end of reply) it is virtually impossible for one more year of data to invalidate those methods.

My bet is that in the coming years, along with the relaxation of loss of MY sea-ice we will see a relaxation of the volume loss as shown by PIOMAS. I'm prepared to wait for this, but if there is some angle I've not spotted that allows us to ascertain whether we are on the fast track ahead of time, I'd like to know now.


Chris Reynolds said...

Dan Satterfield,

Thanks, I should point out, I'm an Engineer by profession (of the hands-on electronics/electrical sort) and amateur climate science reader by hobby. Far from a proper scientist.

The models generally don't do a good job with regards the Arctic. There is good reason to be sceptical of them - I should be in a position to post on that shortly. However as with my use of the Tietsche paper, I don't believe in 'throwing the baby out with the bathwater'. I still think their use of a model is reasonable and defensible. I take what I can from modelling studies, but only when I've read enough to be aware of the caveats.

I expect a virtually sea-ice free Arctic next decade, probably later in the decade. But I expect it to be as a result of at least a decade more preconditioning with a weather event, as in 2007, so it won't mean the transition has fully asserted itself. That will take time, and crucially, much more CO2 in the atmosphere. On the latter count we seem to be trying really rather hard, so we should get there. ;)

Kevin O'Neill said...


I find your reasoning to be a well thought out synthesis of the evidence at hand. After some thought I came to similar conclusions regarding what the Tietsche paper meant: the Arctic isn't ready yet to be ice-free on an annual basis. The abrupt wake-up call isn't likely to happen.

Two questions that leave me a little restless given the trend of decreasing extent: What are the likely changes to polar currents? and, Will there be a significant change in the predominant weather patterns?.

The Beaufort Gyre seems to be the Arctic's main stabilizing force. If decreasing extent disrupts the Beaufort Gyre, then we'll transition to a new regime - much like 2007 seemed to herald a new regime.

Chris Reynolds said...


I think you're right, the issue of ocean currents and atmospheric circulation are key.

The Arctic Ocean is dominated by what's termed a 'cold halocline' Basically the water under the ice is fresher than the water below, so the first 50m of water sits on top of the saltier (denser) deep water. This allows the top water to cool to below the temperature of the deep water.

However with increasing first year ice that may be changing. As the ice forms it rejects brine and makes the sea below the ice saltier, the brine makes the water more dense, so the halocline fails as the ice sinks to be replaced by relatively warmer deep water. There is also the issue of increased storms in the Arctic, which can mix the upper layers of the ocean. These factors may cause changes in vertical mixing and hence to the ocean currents taking water into the Arctic.

Then there is the Arctic Dipole (on which I shall write a post in due course).

I agree that changes in the atmosphere and ocean may further hasten the move to a seasonally sea-ice free state. That's why in my In Flux post I refer to the paper by Abbott & Tzipperman - they refer to a possible explanation for temperate climates in the Arctic's past (around 50M years ago, suggesting deep cloud cover and increased atmospheric heat transport). These issues are at present not known with certainty.

As for the Beaufort Gyre. I tend to think of it now as a destabilising force (sorry to be so contrary - it's not deliberate). I think you're correct, in the past it provided an ageing route for sea-ice as it cirled round before being picked up in the transpolar drift, some to be ridged and thickened against the Canadian Arctic Archipelago, some to be lost through the Fram Stait.

It's useful for a moment to consider why the sea ice isn't retreating as rapidly from the Atlantic sector, this is because here there is a lot of warm water from the East Greenland Current that readily melts the outflow through the Fram Strait, so the ice is more or less in equilibrium there.

However in the Beaufort/Chucki/East Siberian sectors we've seen most of the recession of the ice, because here sunlight on water provides the balance of the energy. And it's into precisely this region that the Beufort Gyre sends the MY ice it picks up from the central ice-pack. Where a lot of it melts each year. So now I see the Beaufort Gyre as enhancing the loss of MY ice and aiding the decline.

Kevin O'Neill said...

Chris - have you looked at the Beaufort Gyre Exploration Project website?

I think the idea of the sea ice acting as a flywheel between the atmosphere and ocean is an interesting one.

Whether the gyre is still acting as a stabilizing force or not may be an open question, but I think the MYI on the western edges of the icepack have delayed what otherwise might have been a quick melt-out all the way to the center of the icepack.

From my perspective, if what I wanted was someone to always agree with me I'd just talk to myself :)

Chris Reynolds said...


Thanks for the link, that's a fascinating website, just finished reading the associated pages from the one you linked to. I'd never heard of a pool of fresher water due to the Beaufort Gyre.

It's worth noting that at the start and end of that article they use the term 'healthy'.
i.e. "In the "healthy" oscillating Arctic climate conditions..."
I'm taking that to imply that the BG may not necessarily be a stabiliser in the current conditions, possibly for the reasons I outlined above.

When I talk to myself I sometimes find myself in disagreement...

Gray-Wolf said...

Hi Chris R! This is the first time I've had the pleasure of someone sharing my concerns about the Halocline layer! The Catlin's findings certainly point to a rapid exchange down to 200m of cold meltwater and warmer saltier bottom waters. Studies in the Pacific show swells propogate up 200m through the water column and the 'open water' we see across the summer Arctic now allow these swells to pass across the basin even under 2m of ice!(prof Barber had swells from Siberia smash his paleocryistic island back in 09').
Warmer oceans may well have started the 'mass reduction' in the Arctic ice but now other changes are making any return to the 'Old Arctic' more and more unlikely (without a return to an ice age to allow the unique depth of Halocline to reform again)

Rob Dekker said...

Niels Bohr one said "predictions are hard to make, especially when pertaining to the future".

Everyone 'predicting' the future uses a model. Now, I completely agree that simple 'extrapolation' (of any model) is not a good idea, since it ignores the physics.

So using a computer model that uses the laws of physics, as constrained by empirical observation, should be much better than extrapolation of trends.

Now, you mention Tietsche et al, which uses GCMs (general circulation models) to project recovery to a long term trend.

Let me point out that the long term trend for GCMs projects a, what was it, 6.5 million km^2 or so for the 2011 September minimum:

So my question is : is this realistic ?

Rob Dekker said...

Also, could you argue which physics could overcome the accellerated decrease in volume shown by PIOMAS (or accellerated decrease in extent as shown by the AMSR observations from NSIDC, UniBremen and JAXA) ?

The question really is : Does the Arctic amplify underlying trends or dampen it ?

Chris Reynolds said...


I must confess that the changes in the halocline are something I'm aware of but haven't until recently given much thought. It's on my list to read more about now that Kevin O'Niell has raised my interest.

Chris Reynolds said...

Rob Dekker,

The reason I used Tietsche et al was purely because it supported my contention that much of the energy gained due to summer open water is being lost to atmosphere (and thence to space). In the extreme example of Tietsche's study this loss of energy means the ice takes only 2 years to recover after an event as severe as the total loss of sea-ice. That's why I think we've not seen the subsequent crashes many expected after 2007. Although my reading of modelling studies and understanding of the physics involved (thinned ice pack) means I expect another crash, possibly a series of them this decade.

So my answer to your question about which physics could overcome the decline in volume is the loss of heat by extended periods of open water in the autumn (fall), combined with the fast re-growth of first year sea-ice in the winter.

I would suggest that this year and the previous 3 years show that the sea ice is nowhere near being able to melt regularly to such an extent that we see regular sea-ice free (< 1million extent of the Canadian Arctic Archipelago).

The Arctic on balance amplifies, a lot of this is due to ice-albedo, not just the exposure of dark ocean, but the darkening of wet ice/snow. Although the Arctic Dipole looks like another positive feedback.

Yes I am aware of the model's shortcomings and will be posting on that in the coming weeks. However that doesn't automatically invalidate Tietsche's findings. The findings I'm using are summed up in the figure I use, their figure 3. I do disagree with the overall trend of ice loss in the ECHAM run they use; summer ice free by 2050ish IIRC. But I consider the use of models as diagnostic tools useful.

Kevin O'Neill said...

The models are not perfect - but anything else is just a guess (even if it's an informed guess). GCMs are not intended as short-term predictors; even ten years is too short a period when talking about climate.

That said, if the PIOMAS data is correct and we see similar volume results from CRYOSAT II, then obviously the models have missed something.

With 6 months of night, the arctic ocean is always going to produce a lot of winter FY ice. The yearly battle is how much MYI - and especially thick MYI - is retained thru the melt season (and winter transport).

I really enjoy looking at the new PIPS 30 Day animated gifs of Ice Thickness. It gives a good idea of ice movement and where the real battles are being fought.

Rob Dekker said...

Chris R
When I first read Tiestche et al I felt the same 'comfort' that the Arctic may have a way to 'recover' from excess loss of ice, due to the effects of heat loss of open water after the melting season.

However, then I realized that physically it cannot be possible for that heat loss (cooling) in fall/early winter to be larger than the heat gain (including heat supplied by albedo changes) that caused the open ocean to occur in the first place, unless there is something else that changes.

Just think about it : the ocean will simply freeze over after it lost the heat initially supplied (initial heat plus heat supplied by albedo effect of open ocean).

I think Tietsche et al did not explain why the GCMs seem to assume that more heat will be lost during refreeze than the heat that was supplied to melt the ice in the first place.

And wouldn't that be a presequisite to claiming 'recovery' of ice extent in future years ?

Rob Dekker said...

I may not have been entirely clear in my reasoning.
What I tried to say is that Tietsche et al shows that the Arctic is not entire instable. That a small perturbation will not cause an imminent collapse. But Tietsche et al do not show that the positive feedbacks of the Arctic (such as albedo effects) are reduced.

Any initial downward trend (such as increased heat inserted from lower latitudes) will still be amplified by albedo effects and other positive feedbacks in the Arctic system. This means that Tietsche et al does not show that any downward trend (such as initiated by GHG increase) would not be amplified in the Arctic system, and thus any downward trend in Arctic sea ice will still be accelerating.

Chris Reynolds said...

Think about what's going on in the water. During the summer it's warming, but the warm water - being more bouyant - stays in a layer on the surface, unless mixed downwards, e.g. storms. Whereas in the freeze season open water loses heat to the atmosphere and sinks. Once ice forms it rejects brine, and the saltier water sinks. So yes, in the adjustment period more energy is lost. So it can lose energy from layers that weren't warmed by the summer.

I'm not trying to be difficult, but I have found that considering evidence often helps with theory.

Let's use Cryosphere Today's area. In 2007 the summer extent dropped by 1/4, it's been around that ever since. That's an extra 25% more insolation absorbed, componded over 4 years (07 to 10). Where has this energy gone to?

PIOMAS anomaly shows: A drop in 2007 but a recovery back to previous levels, 2008 drop in the melt season but no overall change. 2009 2010 overall equal drops, with the sharp drops in anomaly during 2010 and 2011 both being in the mid/early, not the late part of the season - with the most open water.

From the absolute PIOMAS series drops in volume were not apparent in 2008 & 2009, both years were over 2007. Only in 2010 is there a substantial drop. Yet in winters 2009/10 and 2010/11 both drop significantly.

Can you explain these observations in terms of your paradigm?

Rob Dekker said...

Think about what's going on in the water. During the summer it's warming, but the warm water - being more bouyant - stays in a layer on the surface, unless mixed downwards, e.g. storms. Whereas in the freeze season open water loses heat to the atmosphere and sinks. Once ice forms it rejects brine, and the saltier water sinks. So yes, in the adjustment period more energy is lost. So it can lose energy from layers that weren't warmed by the summer.

I appreciate your comment on this effect, and I'm sorry if I am asking too many questions.
But my first response is, can you quantify this effect of negative feedback due to brine rejection ?
Specifically, your assertion that "in the adjustment period more energy is lost" is troubling me.
How does that really work ? When the salty water sinks, where does the heat go ?
As far as I remember, the Arctic is fairly strongly stratified, and thus the upper layer of the Arctic ocean is actually colder than the lower (and saltier) layers. So wouldn't brine rejection actually cool down the warmer lower layers of the Arctic ocean ?

Please understand that we are all here to find out what's going on in the Arctic.
But I think what's going on right now is that the Arctic is showing us a response that we did not anticipate, and we are all scrambling to find reasons why this would be just a temporary thing, and that there are this-and-that negative feedbacks that were not taken into account, and that soon things will recover to normal (being the 7 million km^2 Sept minima that the IPCC models project).
What if it's not ? What if the Arctic indeed showed the positive feedback that we knew would be there all along, and we are really faced with an accelerated trend ?

As for other model projections (like PIOMAS), it seems that we are still running at the exponential rate of decline that would be anticipated if indeed Arctic amplification exists and is real, as witnessed by Wipneus' PIOMAS graphs :

Chris Reynolds said...

Sorry for the long reply.

Part 1.

No problem with asking questions, when I decided to start blogging I knew I'd have to answer them.

I haven't seen the issue of brine rejection as a negative feedback and I haven't read any research that quantifies it's impact in that way. As the Arctic ocean is tending to first year ice and the areas of open water at the end of summer have increased it has been observed that the halocline is failing. The standard explanation is that as first year ice grows it rejects brine which makes the surface layers more salty, and dense, than they previously were. Obviously if the upper layer gets salty enough it will be more dense than the below and will sink. When the water sinks because it's more salty than water below, it mixes with the water below, and is replaced by the deep water. Yes colder surface brine would cool the deeper water, however as the halocline is only around 50m deep (at the surface) the cooling effect from mixing into possibly 200m or more would be small. The observed decline of the halocline implies that the Arctic is less stratified than it was, the loss of Arctic ice implies that this will continue and the halocline will decrease.

The source for my assertion "in the adjustment period more energy is lost" is the Tietsche paper, see the figure in my post above, that's what their results show. They do not discuss that aspect. My suggestion with regards brine rejection is my best guess at what is going on in the model.

Chris Reynolds said...

Part 2.

Sorry but I'm am completely unimpressed by the curve fitting exercise in Wipneus' graphs. As I have said previously if we accept what those graphs show for the summer surely we must accept what they show for the winter. And that is a totally sea-ice free Arctic by the 2020s/2030s. We cannot appeal to novel atmospheric or oceanic heat transports, or cloud radiative effects to support that contention because the statistical projection technique is based on the processes as they are over the period the model fits to, and cannot account for novel processes. To me, based on current trends, the idea that the Arctic will be warm enough in winter to keep the ocean surface liquid just doesn't add up. The evidence from the past suggests that CO2 levels of the order of 1000ppm would be needed for that.

Furthermore if we are to take PIOMAS's volume as indicating a near ice free state as early as 2015, what are we to make of the fact that PIOMAS projections (using high-end global warming) show a near ice free state in 2035? See here, owning page here.

If the model is correct in terms of applying curve fitting it cannot be correct in those projections. I'd suggest that what the curve fitting misses, and the source model gets, is that as there is more open water at the end of the year the freeze season extends and more energy is lost from the system. So each year's gains are partially decoupled from following years by the large amount of gained energy being lost. What carries the 'signal' of reduction through each melt season is not so much the sensible heat, but more the latent heat implied by thinner winter sea-ice (less growth through the winter) which melts more easily come the following year. Of course without increasing CO2 these losses would turn into recoveries in a relatively short time. As for PIOMAS, they don't show a time series unfortunately, but the snap shots seem to me to reflect what's seen in other models - a slow start at first, a period of rapid loss, then a reduction of loss rate with a more gentle move into the seasonally sea-ice free state.

Personally I'm not seeking reasons not to believe that we face a rapid transtion and near seasonally ice free state this decade. Indeed, that prospect would excite me greatly!

As it is I'm resigned to a long slog. And I suspect that if I'm wrong about next decade it'll be because I'm calling it too early.

Rob Dekker said...

Thank you Chris, for taking the time to address my questions and comments.
Also thank you for a scientific discussion, where the purpose is to learn, rather than make a point, while we probably don't even disagree with each other.
I may have been around denier (oops, I mean self-proclaimed 'skeptic') blogs too much, so this discussion is very refreshing.

Got a couple of comments, and (unsurprisingly) a few more questions for you to ponder over.

Regarding part 1. Brine rejection and heat loss.

I'm sorry. I was confused. I thought that you were describing a physical effect that sustains Tietsche et al's conclusions of recovery to the mean.
But instead, it seems that you describe the findings from the Catlin Arctic Survey (Boxell et al)

That effect however, suggests a positive feedback, and thus seems to only amplify any melting trend (increasing the exponential decline part of any initial downtrend forcing).

Regarding this effect, you write as the halocline is only around 50m deep (at the surface) the cooling effect from mixing into possibly 200m or more would be small.

Let's try to quantify how 'small' this effect would be with a very rough first back-of-the-envelope calculation :
Since heat of fusion of water is 80 times it's heat capacity (per degree C), if we assume that water below the thermohaline is 1 C warmer than above, then a 150 meter decline of the halocline would suggest enough heat would be released to melt about 2 meter of ice above it.
I would not call that effect 'small'. In fact, if this effect is real, it would constitute a significant positive feedback effect that had not previously been included in the GCMs on which Tietsche et al are based.

Now, at this point, let me note that you are absolutely right that this additional positive feedback does not invalidate Tietsche et al's recovery to the trend line. However, it DOES show that the trend line itself may be falling off more steeply than the GCMs (and thus Tietsche et al) suggested.

Rob Dekker said...

Which brings me to the second part : exponential decline and curve fitting.

Here, I completely agree with you that curve fitting into the future is a bad idea.
We need to take physics into account before we should project into the future.

So let's look at the physics :
We recognize a downtrend in ice extent and volume (observations and modes) caused by an increase in hear input into the Arctic.

Since the Arctic is the coldest place around, the only way by which it can get rid of that heat is by radiation to space (Stefan Bolzmann). However, for that to happen, the Arctic needs to incease it's surface and/or atmosphere temperatures.
And it does. During winter, temperatures are now some 4-6 C higher than just a couple of decades ago.
However, during summer, it can't increase it's surface temperature as long as it's covered with ice.
So the ice first needs to melt before the negative feedback from Stefan Bolzmann radiation can kick in.

Then here is my reasoning : For the Arctic to achieve radiative equilibrium in summer. enough ice needs to melt so that the average temperature increase by something like 4-6 C on average (assuming that the same amount of excess heat continues to flow into the Arctic during summer as in winter). Now, the question becomes : how much ice (surface) needs to melt until that (negative feedback) become strong enough to dampen the trend line ?

Until that equilibrium is achieved, we should see a continuing downtrend in ice extent, which is meanwhile amplified (with an exponential factor) until enough (warm) open water exists to loose more heat than the Arctic is taking in.

So I think we are not debating the recovery to a trend line, but we are debating the trend line itself, and WHEN it will start to show that negative feedback of increased radiation to space due to warm open water that Tietsche et al is presenting..

Please tell me Chris.. What's wrong with that reasoning ?

Chris Reynolds said...

Likewise Rob, I just want to get this issue out in the open. As for the denialists - stuff 'em, I'm sick of them being allowed to frame the debate.

Firstly let's be clear that the feedbacks aren't 'included' in the models. They are emergent behaviours of the model physics. Issues like resolution (a concern of Maslowski's) and treatment of the ice physics, are relevant as to what feedbacks appear in model runs.

I had forgotten about the Catlin survey, but IIRC yes that is the basic idea I have that could explain the great loss of heat in autumn in the Tietsche study. Your back of envelope reasoning seems OK. But what it means is that enough energy to melt 2m of ice would be vented to the atmosphere, either rapidly through open water, or slowly by retarding the growth of ice where ice is above the ocean.

The problem here is that we can't get any further simply counting feedbacks against each other, only numerical models can help assign importance.

I disagree with your statement that the Arctic must warm to increase radiation to space. Stefan-Boltzman is not the only player. The major heat loss associated with the loss of heat suggested by Tietsche is due to open water into the freeze season. Sensible and latent heat can cause convection and this causes localised warming of the lower Arctic atmosphere, the radiation increase will be from that plume, so will be localised in the atmosphere and as Tietsche et al suggest is transitory. Examination of NCEP reanalysis data shows the 'plume' I'm taking about in the contemporary arctic atmosphere, e.g. Francis et al / Overland et al.

Furthermore Carl Wunsch ("Abrupt climate change: An alternative view"), in considering the ERBE and his disagreement with the thermohaline shutdown hypthesis, asserts that if ocean heat flux decreases atmospheric will increase to achieve a balance. Likewise in the Arctic between 1980 and 2006 there has been a decrease in atmospheric heat transport into the Arctic (NCEP based analysis - Smedsrud et al 2008 "Recent & Future changes of the Arctic sea-ice cover"). It is only at the global average that SB applies.

I agree that summer warming is severely limited by ice. In this respect one can view the PIOMAS volume loss as a proxy for an increase in the region's latent heat.

PS I actually think that the negative feedback from open water is what has lead to the years following 2007 not seeing a successive crash.

Chris Reynolds said...

I think our disagreement may not be that large. And yes, I think any talk of a recovery to the long term post 1979 trend is baseless drivel.

In "A sea ice free summer Arctic within 30 years" Wang and Overland 2009, the authors use close agreement between modelled annual range and observations to ween out a series of models. They find that those models project more rapid declines in Arctic sea-ice. They also note that once the ice gets to around 4.6M kmsq there are increased rates of loss (I'm not wholly covinced by this claim). They find that a virtually sea ice free state (< 1M kmsq) may occur as early as the late 2020s. I should note that I read this paper about a year after I'd decided that <1M sqkm off the Canadian Arctic Archipelago/Greenland is as close to ice free as to make no difference, and by guestimate that we'd not see that until the 2020s. PS That paper is on my list to blog on, if you're interested I'll bring it forward and post on it before I finish the series on the Arctic Dipole.

Maybe we could start to get a framework by each saying what general form we think the decline will take. then each explain why we think that curve will apply.

I think it'll be similar to a logistic curve, but reversed in the time axis (so going down), with the first <1M kmsq year next decade, but that year very likely being followed by at least a year of >1M kmsq (based on 2007 aftermath - bolstered by Tietsche), figures are in extent. I think 2007 was the start of the period of steeper slope.

If you're interested in pursuing this approach and explaining your 'overview' I'll explain why I think that graph is a good general form - although you may already have gathered that.

Rob Dekker said...

Thanks Chris, for your response and the many good references.
Regarding your remark about the possible trend curve into the future, I think that the curve projected by GCMs is likely to be fairly close to the actual trend. These models take well-known feedbacks into account and I don't think that any simple physics model that I myself would conceive could do better than that.

The question is only, what is the timeframe of that curve ?
The IPCC shows this curve :

with a seasonally ice free Arctic no sooner than 2070, and the mean showing still 2 million km^2 by 2100 (and 7 million km^2 for 2011).

Tietsche et al show their projection in their fig 1, with seasonally ice free state as early as 2050, mean hitting bottom by 2070 and 2011 projection of 5 million km^2 or so.

Incidentally, since both Tietsche et al and the IPCC plot are based on (the same?) models, why do they show such a large discrepancy ?

But the key indicator, I think is that ice extent as well as ice volume data still show exponential decline, which suggests that the negative feedback from Tietsche et al Oct/Nov did not really materialize yet. Maybe if the ice extent declines even further we could see that dampening effect, and we could start to see that 'logistic' curve you project, but the question is, when will this curve start to show up ? And will it be soon enough to prevent a near whipe-out of the remaining 4-5 million km^2 of Sept sea ice ?

Chris Reynolds said...

I agree with the general form, which is why I chose to point to the logistic curve.

The models are very variable in what they project for sea-ice. Holland has asserted that the difference between observations and the model spread/mean may be natural variability. However I think some just aren't doing a good job at all and shouldn't be used for Arctic studies.

Firstly any individual model run like Tietsche will show differences from other model runs: Each model run has to be considered as a 'seperate earth' because things like El Nino and NAO/AO won't be exactly the same in each model run.

Secondly the IPCC runs and the model means are collections of various models some of which I would have doubts about performance.

Those are the major reasons there are differences (as far as I'm aware).

If the effect Tietsche et al identify has not materialised then why has extent not declined further since 2007? Why have the intervening years shown the offset of 2007 with variability added? The simple fact of extensive open water into the freeze season demands that energy is being vented into the atmosphere. NECP reanalysis shows this regional lower troposphere warming. See top animated gif here.

If we are indeed to see a sea-ice free summer within the next 8 years (we already know it won't be this year). What will happen to CT's area graph? I'd be interested in your take on this.

Rob Dekker said...

Thanks for your thoughts on GCMs.

It is not just that they vary so much among themselves, but also that they seem to have a bias towards resilient ice extent which contradicts observations. Thus, I agree with you that some of them should not be used for Arctic ice projections, and that Holland's assertion that the differences are caused by natural variability don't explain the issues with these models, but I would like to go one step further :

I think that the high-bias of GCMs and the inter model discrepancy is an indication that Arctic sea ice may be much more sensitive to climate changes than any of the models assumes.

In defence of that statement, Serreze et al 2007 notes that various products such as NCEP/NCAR and ERA40 show a heat budget differences in the order of 10 W/m^2 per year, which translates to a gray area of observation of a meter of ice per year.
a net surface flux of 1 W m 2 over a year is equivalent to melting approximately 0.1 m of sea ice at its melting point.

So, our observations are not accurate enough to estimate the thickness of ice each year, or, in other words, the thickness of ice heavily depends on the influx of heat (and thus are very sensitive to climate changes).

So it may be that the multi-year trend for ice thickness battle is fought out at the single Watt/m^2 meter level. If that's the case, it is not surprising that GCMs can't really get an accurate projection for ice extent or area projected into the future. It also means that it may be that ice thickness trends are much more sensitive to heat input than we anticipated, and (at the level of single watt/m^2) below the level of detection of Arctic energy budget assessments.

OK. That said, I still recognize Tietsche et al's Oct/Nov open ocean feedback as a force to drive the ice extent back to the 'baseline', as it did in 2008 and 2009 after the 2007 negative anomaly. But if indeed ice thickness is sensitive to single-digit forcings, of the order of 10cm per Watt/m^2, then surely the ice now will be thinner than in 2007 and thus more sensitive to natural variability.
After all, you can't keep on thinning ice and expect the ice area to stay the same. Thus, 2010 and 2011 show again a steep decline, even though Fram Stait ice export and cloud cover anomalies like 2007 did not occur.

So my take on trends in a nutshell : Ice extent is a battle between long term ice thickness trends (which are determined at single W/m^2 multi-year averages) and natural variability (which is an order of magnitude larger).
Either way, short term excursions above or below the multi-year trend are dampened by fall open-ocean negative feedback (Tietsche et al) and the trend line itself is amplified by albedo and other known positive feedback effects.

In the end, I do not expect an ice free Arctic in the next 8 years, but I expect a continuation of accelleration of sea ice extent decline, possibly leading to 3 million km^2 by 2016.

Does that help ?

Chris Reynolds said...

Before I answer - I've missed an obvious source of the large difference between energy gain in the summer and later emission during the early freeze season: The latent heat of fusion means energy is emitted as sensible heat as the ice freezes.

I generally agree, with what you say although I see even the trend as being strongly influenced by the balance between positive and negative feedbacks.

The issue you raise of sensitivity to forcing is one of the reasons I think we cannot rule out that the changes we are seeing may be the prelude to a rapid transition. The major issue here is the trend in PIOMAS volume, as I outlined in my In Flux post. However on balance, the post above, and the position I've outlined in our discussion is my what I think.

All we would need to lose another 1M kmsq (or more) is another summer like 2007. In view of the Arctic Dipole's role in 2007 that is quite feasible, even likely.

I agree, the evidence, limited as it is, supports the claim that there has been further thinning since 2007.

Rob Dekker said...

Hi Chris,

One (important) note on the latent heat issue.
I think you had it right from the start : you cannot really take the latent heat of ice freeze in fall as part of the 'negative feedback'.
After all, the heat released during fall freeze must have been supplied during summer melt.

Figure 3 in Tietsche et al (which you show in your post above) assumes that ice is already taken away before the 'net' calculation even starts. After that, they present latent heat anomaly (of late ice melt) as a 'net' negative anomaly, which is only correct for the case where some magical positive anomaly removed all the ice (as they assumed in the first year). Of course, figure 3 is thus not correct for the second and subsequent years, where latent heat of ice melt/freeze will balance out (where they occur).

So for a complete annual cycle, the 'feedback' described by Tietsche et al is the balance between the albedo effect in summer, and the increased IR radiation + sensible heat (reduced heat absorption from lower latitudes) and should NOT include the latent heat plot.

So that reduces the true 'negative' feedback to a fight between the albedo effect in summer and the loss of IR/sensible heat due to late melting in fall. The residual will be ice thinning in the next year.

I had a long explanation typed in here, which shows that despite the negative feedback of fall energy, the delay in freeze after a positive anomaly causes a thinning of the ice in the next year (and visa versa ; a positive anomaly causes a thickening of the ice) but it was getting too complex for a simple post. For the moment though, if you assume that statement is correct, then it may be plausible to accept that a constant, low energy (single digit W/m^2) positive anomaly (such as a GHG effect) causes thinning of the ice on multi-year time frames, while not necessarily causing a reduction in sea ice area. But it seems that the winter temperatures are key, in determining the ice thickness for the next year.

The reduction in sea ice area then comes only after ice thinned enough for natural variability to melt out that thin ice faster than normal (such as what happened in 2007). After such an event, the trend-line is reset (due to now increased albedo positive feedback and fall freeze negative feedback being shifted to a new latitude) and 'recovery' (thickening) or reduction ('thinning') only happens if the long term trends (the low energy heat influx) either strengthen or weaken.

Now, assuming that long term energy influx trend (and local GHG effect itself) continues to increase slowly, this would suggest a step-wise reduction of sea ice area, where the main trend of ice thickness is the best indicator of which direction the next step will occur. This obviously is highly affected by changes in natural variability like wind, cloud density, cloud altitude, pressure distribution, ice export etc etc. On top of that, my main finding is that winter temperatures are extremely important for multi-year ice thickness trends.

OK. Maybe I used too much text, and should put out calculations. If I have a chance to clearly summarize my findings in calculations rather than text, I will certainly do so.

Chris Reynolds said...

Hello Rob,

I disagree with regards the latent heat issue. The latent heat anomaly is the anomaly from average latent heat implied by the lack of sea ice. However the lack of sea ice also leads to far higher sensible warming of the ocean than would be the case if ice had to melt to leave the ocean open. Within the terms of the Tietsche experiment the latent heat anomaly is fine IMO.

I agree that the late freezes we're seeing, and the implied thinner ice is critical. In fact I think this is a major factor that 'carries over' heat gain from one year to the next - the carry over is in lost volume of ice. Granted the heat gain is in terms of implied latent heat (of fusion) of this lost ice, but it is nonetheless real due to the physics of freezing and (more to the point) melting ice.

What you describe as the 'trend line is reset' seems pretty much how I interpret 2007. During 2007 so much old thick ice was lost that we have ended up with subsequent years displaying natural variability superimposed on a step offset caused by 2007. We are clearly living with a post-2007 ice pack that is markedly different (younger & thinner) from pre-2007.

I agree that winter temperatures are key to sea-ice growth, along with the delayed freeze. A lot of the warming we've seen in recent years seems to have been related to warm inflows of air into the Arctic. However the lower tropospheric warming from late season open water is also an issue. The last 2 winters' extreme -ve AO events have added to the recent winter warming of the Arctic. Furthermore the increase in occurrence of the Arctic Dipole causes warm air influxes.

Please feel free to post your calculations. I rarely try to apply numbers because I find myself mired in the question of what reasonable assumptions I can make, but it'll be interesting to see how you've gone about it.