"Our greatest responsibility is to be good ancestors."

-Jonas Salk

Friday, November 13, 2009

Arctic Sea Ice




Scary plot from treehugger

Less scary plot from NSIDC

Non-scary plot from Meteorology News

All show the SAME DATA.

A nice presentation by CalTech postdoc Ian Eisenman today on the possibility of a genuine Arctic Sea Ice tipping point. Not a "tipping point" but a tipping point. The stuff that was once known as "catastrophe theory" but which really just comes down to hysteresis.

A hysteresis is like driving around Madison Wisconsin: the way from A to B has no resemblance to the way from B to A. If Greenland melts due to a little warming, it will take a LOT of cooling to grow the ice sheet back. Or as Lech Walesa once said, it is easier to make a fish soup from an aquarium than to make an aquarium out of a fish soup. (By which he was referring, of course, to Poland's emergence from communism, which has an unfavorable hysteresis.)

Does something like this apply to sea ice? The argument is that a little more CO2 will melt enough ice that the Arctic albedo will DECREASE (was "increase", corrected. See comments.) enough to melt more ice, such that once one crosses a certain threshhold, the Arctic would suddenly warm all of its own, and it would take a huge cooling push to get sea ice back.

Eisenman has a simple energy balance model which has convinced him that 1) yes, it's possible and 2) no it's a long way into the future, above 4xCO2. He finds that most of the Winter sea ice needs to be gone before a small increase in forcing flips the Arctic Ocean really into a warm state where it becomes impossible to restart the ice cycle without a very large reversal of the forcing.

He made two very interesting points related to the sea ice trends.

The charts we see show much more decline in summer sea ice than in winter sea ice. But it happens that the latitudes of winter-only sea ice are mostly covered in land. If you plot, instead of ice area, the latitudinal extent of sea ice, you actually get very similar declines for all seasons.

Secondly, the 2007 anomaly that everybody got all excited about wasn't as spectacular as it looks. Five or six comparable anomalies have occurred on the latitude metric, just in other seasons. It's simply the fact that the 2007 anomaly occurred in the season of the sea ice minimum that it jumps out visually. It's a display of quantitative information question. Plots of annual ice minimum looked really scary in in late '07, but other ways of looking at the data much less so.

None of this is to say the ice isn't in decline, nor that we won't have ice free summers soon. It is to say that the scare of 2007 was a bit overblown and that though there may well be a tipping point in the Arctic, it's fairly far off at present, at least in Eisenman's opinion.

The guts of the talk are here at PNAS if you can access that. He made many other interesting points in an engaging and enlightening talk. A first rate presentation and in my opinion a first-rate hire for the department that gets him.


Scruffy Dan said...

hmmmm looks like I am going to have to issue a correction in regards to my 2007 sea ice loss post. I am now convinced (not only by this post, but also from stuff from Connolley and elsewhere) that I was far too 'alarmist' after the dramatic decline in sea ice in 2007.

Very interesting stuff, and it is nice to see that things aren't as bad as I thought they were.

Hank Roberts said...

> same data
except the first one ends in 2007 at the low point, and the other two don't.

It might be clearer to replace the latter two with current charts for comparison and just draw a line on them to show where they line up with the end of the first chart.


Lou Grinzo said...

I would feel much better about this analysis if more than just albedo was considered as a feedback. From what I can see based on a quick scan of the paper, methane hydrates and no-longer-permafrost weren't taken into account. I realize that's a very different mechanism--adding to GHG levels instead of directly trapping heat--but warming is warming, and the planet doesn't much care about the details of the source.

Michael Tobis said...


Those are separate "tipping points". The argument here is that an albedo switch probably does exist and probably isn't close at hand.

Archer and Buffet put the clathrate blowup on a very long time scale, thousands of years, but also believe that it is there. We may already have tipped that one but we won't live to know it.

As for tundral releases of carbon, I don't know. I haven't seen anything clear on it. But that wasn't the question Eisenman was addressing. I would put that under geochemistry, not climatology. I don't really know the state of the art on that one.

skanky said...

It's been a long time since I read it, but didn't WC co-author a paper that said essentially the same thing? If the conclusion is similar, but the details separate, then I apologise.

I'm sure he'll be here to tell us soon....though I wonder if he'll try not to bring attention to this post as it could affect his income. ;)

Gravityloss said...

"But it happens that the latitudes of summer sea ice are mostly covered in land."

Do you perhaps mean winter sea ice?
Ice retreats poleward in the winter, and there's no land near the pole.
For example over here at a land latitude about 60 deg N, there's some sea ice in the winter (except in 2007-2008) but not in the summer...

Alastair said...

You wrote "But it happens that the latitudes of summer sea ice are mostly covered in land." I think you meant to write "winter sea ice", or are you meaning the Antarctic?

I have always thought that the extent of the latitude of winter sea ice was determined by a dynamic process. C.E.P. Brooks pointed out that the coldest points on the Earth's surface (Siberia) are where the surface is solid. So, the Arctic sea ice growth each winter is based on its area, which does not change. In other words, when the recovery of the ice equals the same extent as that of previous years at that date, it will continue to expand allong the typical curve to the typical winter maximum.

The tipping point will happen when the summer ice area is so small that the winter ice cannot recover. Obviously this did not happen in 2007.

But the ice has been thinning, and it is possible that next year we will see a new summer minimum below the tipping point. In fact the ice recovery is showing a minimum for this time of year, but that could well change.

Gravityloss said...

Argh, meant to say ice retreats poleward in the summer!

Michael Tobis said...

Thanks for the correction. Yeah, I meant to say NON-winter or summer-only. Sorry.

Hank Roberts said...

Another one, a bit toward the scary:


Pangolin said...

I find this graph: http://nsidc.org/data/seaice_index/images/daily_images/N_timeseries.png

a bit more scary than all but the top one. Why isn't the sea ice recovery happening as it did in the past? What is this going to mean for weather patterns? What is going to happen to the sea ice next summer?

Living in California it doesn't much matter to me immediately if there is ice in the arctic or not except for the fact that it will probably change the weather. When the weather gets too strange normal patterns of life can't be sustained.

It's the suspicion that the rate of change, and nasty surprises, is accelerating that keeps me awake nights.

Michael Tobis said...

Bah, shouldn't read this stuff before I've had my coffee.

No, I was right the first time. The high arctic is entirely ocean buth the middle-arctic is mostly Siberia, Alaska, Canada and Greenland. So the advances and retreats of winter ice take place in a constricted zone and contribute less to total area, though they may contribute substantially to latitude.

Of course, the edge of the ice is always new, seasonal ice by definition.

Steve Bloom said...

"Of course, the edge of the ice is always new, seasonal ice by definition."

Actually no, recalling that a big factor in the loss of recent years has been older, thicker ice flushing out through the Fram Strait, so at least in that region the ice edge has a lot of older ice. Elsewhere, in summer it's common for the older ice to be exposed at the ice edge after the thin seasonal stuff melts.

Michael Tobis said...

Steve, that's an important phenomenon, but in terms of the latitudinal extent of ice, a small term, especially outside the peak melting season.

Anonymous said...

"The charts we see show much more decline in summer sea ice than in winter sea ice. But it happens that the latitudes of (1 winter-only)/(2 year round) sea ice are mostly covered in land. If you plot, instead of ice area, the latitudinal extent of sea ice, you actually get very similar declines for all seasons."

This does make sense for interpretation 1:
Hmm so the problem was that winter sea ice area (large extent) had not gone down much. *But* it has gone down a lot when you measure by latitude. And this makes sense - even a large ice edge latitude move shows up as a little change since there is so much land and so little ice.

Interpretation 2: For the year round ice this doesn't make sense - the summer sea ice area has gone down, hence no "buts", since latitude extension has gone down as well? Ie whichever method you measure with... Here though you get larger latitude changes for less area because of less area per latitude near the pole of course - but climate should roughly follow latitudes, not areas...

Confused? The original post now says winter-only sea ice...

There is just some contradiction, and I don't know which one of the claims I misunderstood.

Michael Tobis said...

I am definitely trying to say what you said in #1.

I still don't understand your #2 interpretation and so naturally I don't see how I implied it.

By using area weights rather than latitude weights, one sees an emphasis on variation in warm season retreats that in the latitude view is not there. That's the main idea.

EliRabett said...

Bah, Eli has been saying since 2007 that hysteresis determines Arctic ice futures but would you listen? Now you hear it from some Cal Tech hotshot and you believe.

Michael Tobis said...

But Eli, he had charts 'n graphs 'n stuff.

Dan Satterfield said...

I like. Good stuff and some real hope that even if it takes 60 years or so to change to a non carbon based economy, we will still have polar ice. (In winter at least).

Can only imagine how the denial crowd will rework this.


Anonymous said...

Ok, I had misunderstood the comment "Bah, shouldn't read this stuff before I've had my coffee..." interpreting it so that you backtracked (but hadn't yet updated the post back).
Yes, the post is clear.

EliRabett said...

Eli leaves the scut work to others.

More seriously there are base issues that you only have to think about a bit.

Unknown said...

"The argument is that a little more CO2 will melt enough ice that the Arctic albedo will increase enough to melt more ice..."

Er, you meant "decrease enough" surely?

wag said...

I think it's also a good explanation to economists why discounting doesn't make sense in the context of climate change. Free market types argue that "if future generations are richer, they'll be able to solve the problem with their greater wealth." That, of course, assumes that the problems they'd be solving are 100% reversible.

Aaron said...

Dr. Demming was the master of statistical behavior of feedback systems.

He is well worth rereading with a spreadsheet application and the Arctic sea ice data close at hand.

In class, Deming made us work with a series of physical (feedback) models then apply his tools to what the models actually did. We had to do this until we trusted his tools.

Number 2 is the scary graph. However, by taking an annual average, they discarded the short term "wobbles" that demonstrate how far the Arctic ice system is out of control. The behavior of the Arctic ice system in the past no longer predicts the behaviour of the ice system in the future. That is: we can expect to set new records frequently.

Knowledge of the global climate system may (or may not) provide insights.

Unknown said...

Yes, I was a student under big names as Aaron was, and they showed us that the recent wiggles in the ice plots are symptoms of ecosystems coming under stress, being less resilient, and this is what the metrics look like as collapse approaches.

Jus' sayin'. Not that society will take action to stop it or anything.



Michael Tobis said...

Dano, there is some research into whether there is a universal mathematics of the approach to a regime change in a dynamic system. I think that there won't be one.

Sea ice is in decline. Multiyear ice is in decline. Sea ice volume is in decline. But the whole point here is that there ISN'T any sign of an abrupt change you can point to, and the generally proposed albedo-forced abrupt change is far in the future if it exists.

This is different from an albedo-feedback amplification, which certainly exists and is starting to kick in. This will result in a gradual acceleration of the decline, which we are seeing.

But if you're talking about onset of dynamic regime changes, it is exactly my point that there isn't any sign of one, not that there is.

It is not the obligation of the environmentalist to find every proposed scare realistic. To the contrary, it is the obligation of the serious environmentalist to make a realistic assessment.

This is because we are into the territory in parameter space where tradeoffs are necessary. If everything looks like a huge disaster the problem becomes indeterminate, and all there is left to do is panic. Let's actually try to move beyond scaring people and into trying to get things right.

Michael Tobis said...

Wulina, yeah, sorry. I am forever forgetting whether albedo of 1 is white or black. I seem to have to look it up every time.

Fixed, and thanks.

Dano said...

I posted elsewhere about scaring people, Michael, and the resultant fatigue. You'll come across it surely and see how that does not comport with what I wrote above.

That is: ecologists who develop and track metrics to try and understand ecosystem dynamics find similar patterns to the Arctic graph when ecosystems become less resilient.

Can we graph, on the ground, what it looks like when they flip? I don't think we're there yet for most systems including the Arctic, but I'd say bigger names like Carl Folke or Gunderson or Peterson would disagree, esp wrt policy formation.



Michael Tobis said...

Sigh. Dano, you're great, but...

Sea ice has very different dynamics than an ecosystem, first of all. Secondly, there is a big difference between mathematics and intuition. Both have their place, but the best thinking occurs when they stay within bounds.

Again, I wrote an article stating that there is quantitative reaosning that sea ice is NOT near an abrupt transition. You write saying that 1) you agree with me and 2) based on some vague resemblance between these graph and some systems of very different character, sea ice IS near an abrupt transition.

You cannot have it both ways. It is OK to disagree if you want to. Nobody says we have to agree on everything. Once someone makes a quantitative argument it's not very convincing to rely on intuition, though, but sometimes it's worth sticking to your guns and looking for the counterargument that you know must be there.

It's not at all within bounds of reasonable discourse to say you agree and then advocate the diametrically opposed position. I have to call foul whether I want to or not.

Dano said...

That's not what I wrote, Michael.

I said the ice plot looks similar to metrics of ecosystems before they flip. I said, implied, hinted at nothing more and clarified the 'scary' bit with context of something I wrote elsewhere. 'Transition' appeared nowhere.



Michael Tobis said...

Is there a distinction between a 'flip' and a 'transition' that I am missing?

Unknown said...


What I wrote:

the recent wiggles in the ice plots are symptoms of ecosystems coming under stress, being less resilient, and this is what the metrics look like as collapse approaches.

No wonder you said something, Michael.

What I thought I wrote & didn't verify:

the recent wiggles in the ice plots are similar to metrics of symptoms of ecosystems coming under stress, being less resilient, and this is what the metrics look like as collapse approaches.

Hopefully this is clear and I don't have to alter the caffeine dosage and workload.

Apologies for the confusion.



Michael Tobis said...

Alas we are not out of this pickle, as that's what I thought you were trying to say.

In WHAT WAY are they similar and WHT conclusions can we draw from it? You seem to be implying that "collapse" is approaching in the sea ice.

You can think so if you want, but in that case you are not agreeing with me. On account of the talk, I am of the opposite opinion, which is that the signs are NOT of an imminent collapse.

Alastair said...

Abrupt climate change happens without warning by definition, so using the slow decline as a reason for thinking we are safe from an abrupt change is illogical.

Both the Arctic sea ice and an Ecosystem are dynamical systems with positive and negative feedback loops. So they respond to the same maths, and if both show violent oscillations before final collapse then that should be no surprise.

Putting it another way, if the Arctic ice was showing a slow monotonic decline in minimum ice extent, then it should be possible to predict that the final collapse is a long way away. OTOH, what we have with the Arctic sea ice is not monotonic, so it is wrong to predict a slow steady decline.

IOW, the next sudden drop, one similar to that of 2007, may be the last.

Michael Tobis said...
This comment has been removed by the author.
Michael Tobis said...


The fact that we have no apparent "warning" cannot, on the other hand, be taken as evidence in favor of a pending abrupt change.

More to the point, though, I am not using the steadiness of the past as evidence that a tipping point is far off. I am simply making reference to Dr. Eisenman's quantitative analysis.

If your intuitions tell you otherwise, you should try to find a quantitative problem with his quantitative analysis. Otherwise you are blowing smoke. My suggested approach is called "the scientific method" or "science" to its friends.

Dano said...

In WHAT WAY are they similar and WHT conclusions can we draw from it? You seem to be implying that "collapse" is approaching in the sea ice.

OK, I need to make dinner and maybe I can find some of these tomorrow, but when looking at various metrics (choose one) when monitoring ecosystems for collapse, what you see before a flip is increasing chaoticality that is expressed graphically as similar to the last few years in Arctic ice, fish catch and so on.

I'm not implying collapse, I'm stating the metrics are similar. This is essential in decision-making and our knowledge is limited by what we measure. We have limited input in decision-making and can only go with what we know, and then go with the other ways of knowing that individual decision-makers use.

So. What conclusions we draw from it are we don't have a whole lot to go on, and the limited knowledge we have with what we measure is that the data we measure on the Arctic aren't promising. If I were a staffer and my elected asked me I'd have to use key mammal population data, temps, CH4 release, permafrost melt extent, all that and try to use that to make a case. Whatever that is that was asked for. If I were asked about system collapse, I'd state that what we know right now says it isn't imminent but a better chance in the future under BAU.

That's how it is often done in that world. If you are asking about conclusions in science, I take the point in your post and I'd encourage you to think about emergent properties a bit more before concluding something.

I have a couple projects cooking hot right now & a little scattered, else I'd agree to chat or e-m or something, but not this week. Apologies.