More on melt ponds

After you've read the blog post below, make sure to check out this web page by one of the co-authors of the Nature paper, and also this page with lots of other Arctic sea ice goodies.

---

A really good paper has been published online a couple of days ago on Nature, called September Arctic sea-ice minimum predicted by spring melt-pond fraction. It's really good because it's interesting, short, and it confirms what I've been suspecting for a while now. And when a paper confirms what one is suspecting, it must be really good, right?

All joking aside, the paper by Schröder et al. presents evidence that melt ponds play a very important role at the start of the melting season, to the point that it can heavily influence the September minimum. The last two melting seasons actually proved to be a great lesson in this respect. 2012 had a really good* start to the melting season, so good that when bad weather showed up, it didn't really slow down sea ice loss, the trend lines just kept dropping (low sea ice volume also played a role, of course). The reverse was true in 2013: cold and cloudy weather during the first half of the melting season caused a lagged response during the short periods when the Sun and higher temps finally got to the ice.

What Schröder et al. did, was develop a melt pond model and incorporate it into the larger Los Alamos sea-ice modelcalled CICE. Here's what they came up with:

Our simulations show that melt ponds start to form in May, a maximum extent of 18% is reached in the climatological mean at mid-July, and there are hardly any exposed ponds left by mid-August. The strong interannual variability and the positive trend are striking. Whereas in 1996, the year with the highest September ice extent since 1979, the maximum pond fraction reaches only 11%, in 2012, the year with the lowest September ice extent, up to 34% of the sea ice is covered by ponds.

This is accompanied by the following figure:

 

 Based on their results, Schröder et al. conclude:

[T]he melt-pond fraction in May seems to have the strongest impact on the sea-ice state in the subsequent September. Our results confirm that the early melt season is decisive for the strength of the summer ice retreat.

(...)

We conclude that the inclusion of a realistic melt-pond model will transform future forecast and climate models in the Arctic regions and beyond.

Although they didn't participate in last year's Sea Ice Outlook, they did make a prediction:

For September 2013 we forecast a mean ice extent of 5.55 0 ±44 million km2, which is closer to the observed mean value of 5.35 million km2 than any of the 23 statistical,model and heuristic predictions presented at the Arctic Sea Ice Outlook webpage in July (median value of 4.0 million km2).

As we all know, last year's ASIB community prediction for the SEARCH Sea Ice Outlook was overly pessimistic, and off the mark by almost 2 million km2, much more than the other predictions. Because of low volume and a record amount of first-year ice, I personally thought the September minimum couldn't but end up somewhere between 2007 and 2012, at the very least. Most of us would probably have guessed differently, knowing just how low the melt pond cover fraction was in May 2013, and how influential this can be.

Which goes to show how incredibly handy it would be to have near real-time melt pond cover fraction data at our disposal, which we could then compare to data from the last 5-6 years. It wouldn't tell us anything conclusive about the melting season's final outcome (weather determines this), but combined with ice age distribution, and data on thickness and volume, it would give us a good idea of where things could or couldn't head.

The image above comes from a blog post from 2 years ago, wherein I mentioned melt pond data having been developed by the University of Hamburg's KlimaCampus for the 2000-2011 period (also referenced in the Schröder et al. paper), using MODIS satellite data. Hopefully something similar comes along next year, as May 2014 is just around the corner.

Still, we now know that the start of the melting season can make it or break it.

---

* When I say 'good', I mean 'good for melting', not that Arctic sea ice loss is a good thing in itself.

The image at the top of the blog post was found here, with a nice quote by Don Perovich:

When Arctic melt ponds are sufficiently connected, as pictured here, they exhibit a property called universality that researchers believe is common to all complex, correlated systems.

Comments

[Artful Dodger]

Hi folks,

Toward the end of the melt season, melt ponds cover up to 50% of the sea ice area, decreasing the value of the surface albedo by up to 20%. Total melt is proportional to total solar energy absorbed, which is the time integral of sea ice albedo. Hence, earlier melt ponds, more melt.

Things aren't just black and white, not in the Arctic, and not with our energy choices. It's time for "Less Denial, More Action" by reducing our reliance on fossil fools.

Cheers,
Lodger

[Artful Dodger]

Neven wrote:

"Which goes to show how incredibly handy it would be to have near real-time melt pond cover fraction data at our disposal, which we could then compare to data from the last 5-6 years."

Our blog statistic "Cryosphere today Per Ijis Extent" (CAPIE) summarizes in melt pond fraction in numerical terms. Understand that CT ice area is derived with the ARTISAN sea ice algorithm, which is tuned to classify melt ponds as open water. On the other hand ( or denominator ;), IJIS extent uses an algorithm which classifies melt ponds as sea ice. Thus CAPIE also includes a measure of the total fraction of melt ponds.

Indeed, blog all-star Wipneus may be ideally situated to create the exact metric we'd be interested in seeing. With his access to daily IJIS AMSR2 data, by processing the data twice (once with each alternative algorithm), we would see JUST the melt pond fraction. And the geographic distribution of those melt ponds could be plotted on Wipneus' signature graphs and animations. :^)

Voor niets gaat de zon op!

Cheers,
Lodger

[Micheltsamados]

Thanks Neven,

great summary of our work.

Have a also a look at another piece on this at http://www.micheltsamados.co.uk/september/

Hopefully we will have some interesting feedback and useful discussions about this at EGU2014 next week.

Three of the authors Daniel Feltham, Daniela Flocco and me will be there. Sadly David Schroeder (the first author) will not make it this year.

Cheers,
Michel

[Neven]

Have a also a look at another piece on this

That's great. In fact, the rest of your website is also great. I've added links to the blog post.

Like you say, incorporating this stuff and ice drift into models, should really help improve forecasts. Ocean heat flux would be nice too, but I think that's the hardest thing to observe, and thus model.

Hopefully we will have some interesting feedback and useful discussions about this at EGU2014 next week.

Darn it, darn it, darn it! I already felt bad about not being able to go to EGU (even though I don't live that far from Vienna), now I feel even worse. :-(

If I would be there, I'd ask the following:

1)How difficult is it to create near real-time melt pond cover fraction data that can be compared to other melting seasons in the 2007-2013 period? May is just 5 days away. Easy, right?
2) Will you do another prediction this year, and make it known in June (preferably through the Arctic Sea Ice Blog ;-= )?

[Kevin O'Neill]

Neven asks - "How difficult is it to create near real-time melt pond cover fraction data that can be compared to other melting seasons in the 2007-2013 period?"

We may not be that far away. One of the papers currently in discussion at The Cryosphere Discuss is Sea ice melt pond fraction estimation from dual-polarisation C-band SAR – Part 1: In situ observations (R. K. Scharien, J. Landy, and D. G. Barber).

[Corrected the link, N.]

[sofouuk]

as Steve Bloom says on the other thread, the amount of melt ponding is obviously a reflection of spring weather, but what is the relative contribution of temperature and insolation? can the amount of melt ponding be predicted from general weather conditions (or even forecasts)?

[Neven]

Thanks a lot, Kevin! That's a completely different method, it seems. Hopefully something comes out of all this research that allows us to better assess the start of the melting season.

Our blog statistic "Cryosphere today Per Ijis Extent" (CAPIE) summarizes in melt pond fraction in numerical terms. Understand that CT ice area is derived with the ARTISAN sea ice algorithm, which is tuned to classify melt ponds as open water. On the other hand ( or denominator ;), IJIS extent uses an algorithm which classifies melt ponds as sea ice. Thus CAPIE also includes a measure of the total fraction of melt ponds.

That's true, but a direct measurement would be better, as CAPIE is also influenced by compactness of the ice pack.

[Kevin McKinney]

Exciting! It would be great if seasonal-scale constraints could be arrived at...

[wayne]

Melt ponds should mirror the ice bottom topography. But the larger reason for their appearance has something to do with complex thermal interactions luckily made simple by observing the horizon. I deal with what may be ice models Achilles' heel, the very reason why they fail might have something to do with the remarkable phenomena of ice cooling faster than air:

http://eh2r.blogspot.ca/

A simple example may help reveal the error, of which thermal conductivity may be greater in sea ice, or one of the reasons why it melts and also accretes so quickly.

[Pjie2]

Hi Wayne. In the second sentence on your site, you say "...sea ice must warm or cool faster than surface air, this is impossible with a standard physics interpretation."

This is wrong, it's perfectly possible for a surface to warm or cool faster than the overlying air, as anyone who's ever stood in the sun or found frost on their car windshield in the morning can tell you!

In particular, you can get frost forming on nights where the air temperature never goes below freezing. The thermal capacity of the air is low, as is the efficiency of heat transfer between (solid) surface and (gas) air. Thus the heating or cooling of the surface itself is dominated by the radiative flux rather than by conduction. Since the surfaces are radiating straight up to the night sky and the cosmic microwave background (-270 degrees C), this means they can lose heat rapidly and drop well below the ambient air temperature. You end up with a very steep temperature gradient above the surface, with the surface being below freezing, and the air temperature even a few centimetres higher being above freezing.

[wayne]

"who's ever stood in the sun or found frost on their car windshield in the morning can tell you!"

Hi Pjie2

In the morning after several hours of over night radiative cooling the ground is just as cold as the air, in fact it may be cooler than the air. Standing in front of the sun is an impression best sensed in a warmer atmosphere, try standing in the sun at -20 C and see how warm it is.

Or try an experiment, place water in a freezer and see how long it takes for water to change temperature, if you want it fancy, place water in a container half filled with water and air. See which one cools first.

[Pjie2]

"in fact it may be cooler than the air"

Yes, precisely. So why does the second sentence on your website say that ice cooling faster than air "is impossible with a standard physics interpretation"?

It's perfectly ordinary physics.

[wayne]

Pje2, , after several hours, perhaps 8 or more, that is not a nearly instant thermal response as suggested on my blog. Its all physics, but sometimes physics is extraordinary and requires more study.

[Kevin McKinney]

"…cooler than the air…"

Indeed. Readers may be interested in the investigations of William Charles Wells, who investigated the phenomenon of dew in the 1810s, winning the Royal Society's Rumford Medal for his considerable pains. (Also, later, the admiration of none other than John Tyndall.)

Having considered these “circumstances” affecting the formation of dew, Wells examines the “circumstances” affecting the cold which is dew’s precursor: on clear, still nights, the grass would become much cooler than the air temperature (customarily measured at 4 feet); long grass would cool more than short grass; and wind or cloud tended to eliminate this difference in temperature—although high cloud “would yet frequently allow of the grass being several degrees colder than the air.” The ground beneath the grass would be warmer; metal lying upon the grass tended to be warmer, too.

http://doc-snow.hubpages.com/hub/Global-Warming-Science-In-The-Age-Of-Washington-And-Jefferson-William-Charles-Wells

[wayne]

Hi Kevin, that is an easy familiar one, now try find one about sea ice.
Not at all the same as land in the morr'n. I believe that mixing land effects same as sea ice, as part of the larger problem of understanding, or rather misunderstanding sea ice interactions.

[George Phillies]

Readers interested in seeing melt ponds, open water, floating ice, landfast ice, and melting frozen soil, all in one picture in which you can see small details, may find the Barrow webcam http://feeder.gina.alaska.edu/webcam-uaf-barrow-seaice-images/current/image
to be of some interest.

[wayne]

Yes I added Theory Vs Observation discussion, its not as simple as morning dew, or feeling the noon sun at 70 degrees elevation. Simple physics may be warped into something more exotic and interesting, but first one must observe it. http://eh2r.blogspot.ca/

[P-maker]

A new observation-based Nature paper is out:

http://www.nature.com/news/climate-science-understand-arctic-methane-variability-1.15196

trying to make the link between melt ponds and methane emissions.

It turns out that 2007 had the highest growth rate in atmospheric methane concentrations.

[Joffan7]

You're quite right to be clear about how you are using "good" in "a really good start to the melting season, so good that...". Even better than footnoting might be to use a different adjective; "strong", perhaps?