CAPIE hits record low

CAPIE stands for Cryospheretoday Area Per IJIS Extent (another more scientific term is 'compactness'), the index which we collectively created last year and that is regularly mentioned in the weekly SIE updates. I'll quote from last year's blog post that saw the birth of CAPIE, called Area vs Extent, to explain what CAPIE means and what it is useful for:

It tells us something about how much the ice pack is spread out. When the pack gets compacted area and extent will come closer together. But in the melting season the gap gets greater. This is because ice is melting, gets spread out and melt ponds start to form.

If you look at the ice pack as a whole, the holes that form away from the edges are not counted as sea ice area, but they do get counted as sea ice extent. That's because for extent every grid cell that contains more than 15% ice is counted as 100% ice. In consequence area will always be lower than extent.

Because of the difference between the ways area and extent are calculated, we can divide the first by the second and this gives us a percentage that, especially in comparison to other years, tells us something about how much the ice pack is compacted or spreading out. If the ice spreads out, there will be more holes, area will drop faster than extent, and thus the area/extent ratio will drop. If the ice gets compacted by winds, holes will disappear, the difference between area and extent becomes smaller and the percentage goes up.

Well, due to a huge drop in CT sea ice area CAPIE has hit an all-time low: 57.39%. As can clearly be seen on our CAPIE graph:

This is the yearly record low in previous years:

• Aug. 16th 2005: 69.53%
• Sept. 26th 2006: 67.69%
• Aug. 19th 2007: 60.01%
• Aug. 13th 2008: 59.57%
• Aug. 16th 2009: 63.91%
• Sept. 3rd 2010: 60.28%
• Aug. 12th 2011: 57.39%

But I don't think this is just the ice pack spreading a lot. I think that in a significant part of the Arctic - the Pacific side to be exact (the Beaufort, Chukchi and East Siberian Seas) - the ice is, to quote Professor Peter Wadhams from this BBC interview a few years back, "just melt[ing] away quite suddenly". Or to link back to the title of the last SIE update: flash melting. And it's showing up in area numbers first.

Comments

[Andrew Xnn]

Wow; not unexpected since it's been low all year, but this record came on rather suddenly.
Seems like the pack is becoming more and more like a house a cards.

Wonder how low CAPIE can go?

[Neven]

If it goes any lower, I'll do my impression of Chicken Little. I think/hope this was it. Extent should start catching up.

[crandles]

128k extent drop & 205k area drop would not mean much change in CAPIE but that isn't what happened so those days don't match up. Yes, that makes sense, the 128k extent drop hasn't been included in the calculation yet. So for a further CAPIE fall tomorrow, area would have to fall by more than 213k - not likely. Extent should start catching up seems a safe bet for tomorrow.

I doubt we are at area minimum yet but rate should begin to slow as we approach the area minimum. The Area minimum does usually come before extent minimum. So there probably isn't much time left to beat this CAPIE record by much more this year.

[Rlkittiwake]

Last year, the Cryosphere Today area looked like it was easing into a gentle landing, but in the last half of August it dropped roughly 500K below where it was initially set to hit. In the last few days, it has also re-inflected downward, but this year doesn't even seem to have passed go or collected $200.

How does that one Tom Petty song go?

[Arcticio]

About 500,000 sqkm are below 30% concentration and above the 15% extent threshold. The upcoming weather could easily do anything in both directions with these areas. There is still room downwards.

[Lucia (The Blackboard)]

I think it's worth nothing using data since 1979, I find no statistically significant trend in the difference (extent-area) with time. The scatter plot looks like this:

This suggests that the amount of mushy/slushy/ melty/ spread out ice might be pretty constant given the time of year.

Meanwhile the area has a large trend:

To some extent at a given time of year, it turns out the best fit curve for extent looks pretty much like:

extent= 1*area+C1+ noise.
where C1 is a constant for that time of year.

So ignoring the "noise" (i.e. scatter for different years.)
extent-area= C1

Area/Extent= 1- (Extent-Area)/Extent = 1 - (Extent-Area)/(Area+C1)

But looked at this way, the ratio decreasing because Area decreases but (Extent-Area) stays constant.

[Seke Rob]

Lucia, here My Version ... No statistical significance? Winter has no change due the geographical situation. It just in-fills. Melt season slam dunks what's going on.

[Lucia (The Blackboard)]

Seke--
I think we are plotting different things. Mine (extent-area) doesn't have dimensions %, it has dimensions "area". I think this might have been clearer if I'd added units to my graph. Sorry-- I did it quickly.

I'm sure if I plotted "area/extent", that will have a statistically significant trend.

I think it's interesting that (extent-area) does not. To me, this suggests there is a tendency for a specific amount of ice area to become all slushy and spread out over the summer as a result of warming and the size of that area is neither growing nor declining as we see a downward trend in NH sea ice over the years.

[Seke Rob]

To add to my previous post, in absolute terms, the August data through 2010 in absolute terms, shows little divergence based on NSIDC data:

Arctic SIA+SIE August

Because the smaller Area is calculated to an equally smaller Extent, the percentage decreases. Does that convey?

[Lucia (The Blackboard)]

Seke--

Does that convey?

No. I now have no idea what you are talking about or what point you are trying to make relative to what I wrote.

To try to learn what point you are trying to convey to me, my questions would would start with:
"When you write, "shows little divergence based on NSIDC data:" you must mean mean "shows little divergence in something based on NSIDC data." I haven't been able to guess what you want me to fill in for that "something". What is the "something"?

I'd also mention that I don't disagree that the ratio of (area/extent) is declining. Are you under the impression I think or said this ratio is not declining? I think what I wrote is entirely consistent with recognizing that it's declining.

What I intended to convey is that I think it is interesting to note that if we pick a specific day of the year, the difference (not ratio) (extent-area) is remaining nearly constant. I don't see anything you are posting to contradict my observation. I still think this observation is interesting and worth noting.

Maybe you think it's not interesting? Or something? The point you are trying to make eludes me.

[Seke Rob]

Now, for Global that's different, but those are 2 physically different systems combined, with a 2 season shift at that. Then you get this picture: Global SIA+SIE July

Fun with statistics.

[crandles]

Please ignore my post, it was complete rubbish - I should know area is smaller than extent, d'oh I'm an idiot.

(Seke's points have lost me too.)

[crandles]

>Lucia's "I think it's interesting that (extent-area) does not. To me, this suggests there is a tendency for a specific amount of ice area to become all slushy and spread out over the summer as a result of warming and the size of that area is neither growing nor declining as we see a downward trend in NH sea ice over the years."

If we model the area of ice as a circle (only roughly I know) then as the ice retreats the circumference is reducing so it appears the vunerable ice (Extent-area) should reduce. That there is no trend suggests that the ring of vunerable ice is remaining the same area by increasing the width of the ring.

Various explanations for this could be put forward. To suggest a few:

The ice is thinner so you get a wider band of melting ice.

Waves reach further into the pack because thinner ice doesn't have the same dampening effect.

With thinner ice the extra ice melt where ponds of water have been will be a certain depth and this depth represents a larger proportion of thinner ice thickness.

This doesn't seem to bode well for the ice in that if it gets a lot thinner the melting will be taking place at a higher rate further into the ice pack.

[Daniel Bailey]

"Waves reach further into the pack because thinner ice doesn't have the same dampening effect."

And remember, the further the waves reach into the ice, the thinner the mixing layer under the ice gets, bringer the deeper warmer waters closer to the surface, helping melt even more of the undersides of the ice, thinning it even more...

[Seke Rob]

Lucia, you do something fast, I do something fast, woosh. The something in the second chart is SIA and SIE i.e. Area and Extent plotted side by side. Linear Regression for the 2 for July, including 2011 give 0.7572 for Extent (Area+Water) and 0.7817 for Area hardly significant... the relative spread out being the theme here.

Yes, on the dailies we agree [I missed that you were talking dailies]... looking at the month of year series gives it meat. Either way, Area is on decline, and that's the bit that counts [and it's thickness].

Pity we cant see additional replies in preview, while writing a post, but see in other tab that crandles has picked up. Yes, my chart shows it that the watery part in extent is quasi constant, an obscure physical/mechanical property. Would, as suggested somewhere, wind have less hold on thinner ice? Thicker ice on the move certainly has inertia.

[Daniel Bailey]

Seke, thinner ice has less freeboard (elevation above the water line) than does thicker ice. Makes sense when you think about it.

Less "sail" to catch the wind.

[Bob Wallace]

Thicker ice has more "keel". Lots more sticking down into the water which means a lot more wind needed to shove it around.

[Andrew Xnn]

Could somebody help educate me on the theory behind CAPIE?

In 2005/2006, CAPIE was over 70% for most of the year.

In 2007/2008, big melt down and CAPIE dropped to near 60%.

2009 early 2010 CAPIE recovers to a range of 63 to 70%.

Late 2010 2011, CAPIE begins to drop again and eventually reaches all time low.

So, why the 2009 early 2010 recovery in CAPIE?

What weather trends or other factors came into play?

Long term, I can only guess that CAPIE will establish lower lows. Theoretically, it could reach near zero, but the pragmatist in me suggest that it won't and instead we will see it reaching values seen in seasonally ice free bodies of water like the Great Lakes.

[Seke Rob]

NOAA and NSIDC do the Monthly/Seasonal Reports, Report Cards and the like. Those are a good summary read-back resource: Go to

http://nsidc.org/arcticseaicenews/2011/032311.html

http://nsidc.org/arcticseaicenews/2011/030211.html

For the March 2010 updates. For more, select date / month / year of interest in right margin.

[Rlkittiwake]

Any calculus gurus on here?

I'm wondering if some of the answers to Andrew's questions might be hidden in the CAPIE derivatives. In other words, if the highs and lows might not be as interesting as the periods of relative stasis versus periods of change.

If I had to provide a rationale for this hunch, it would be that things like insolation on a square kilometer of open water are constant from year to year, but it's the deviations from these constants that are worth closer analysis. Taking a derivative might filter out the stuff we're not interested in.

Sincerely,

rlkittiwake (who would be a hydrologist today were it not for the horrors of Calc 2)

[Bfraser]

SEARCH September Sea Ice Outlook: August Report

is now available

http://www.arcus.org/search/seaiceoutlook/2011/august

bill

[L. Hamilton]

I hadn't updated the area-to-date graph in a few days, but if you look now you'll see that 2011 area is now below the 2009 annual minimum.

[D]

Has anyone correlated CAPIE with sea ice extent 1, 2 and 4 weeks later?

This seems like a simple test that would give us a better idea of the extent to which CAPIE values might predict future ice melt.

[Neven]

What weather trends or other factors came into play?

Could it be thinner ice? Like I said in the latest SIE update: "Where holes in the ice pack were the most striking feature of last year's melting season, I would say this year is increasingly characterized by a sea of slush puppy ice in the Beaufort and East Siberian Seas."

Theoretically, it could reach near zero, but the pragmatist in me suggest that it won't

That all depends on how large the difference between SIA and SIE can become/remain. Perhaps towards the very end when SIA is reaching zero (meaning all of the ice is practically gone), but SIE is slightly higher, might we see CAPIE ratios lower than 50%. But a very small ice pack that is calculated as 30K SIA and 50K SIE still has a CAPIE of 60%.

Again, CAPIE tells us something about how spread out or compacted the ice pack is, not how much ice there is left. And in this period of the melting season it tells us how much compaction potential there is, which of course is very interesting. If from now to the end of September we would get to see serious compaction (like it happened in 2007), all records would be broken by a large margin.

[Seke Rob]

Pete Dunkelberg posted over at RealClimate link: 14 Aug 2011 at 11:29 PM a reference and excerpt of a paper by Rampal et al 2011 on observed to modeled extent differentials, indicating that thin ice 'does' get transported faster, which would support the relative ease of extent increase, thus decrease of the CAPIE value. The value than indication, in present weeks, a powerful predictive indice of what can happen to extent with the weather maps in hand. Clear.

MASIE so happens to show an extent increase of 46,949kmsq from the 12th to the 13th... any way the wind blows, after all in thinned conditions.

[Neven]

On the question whether thin ice moves around faster than thick ice: I think it's useful to keep in mind that we are talking about a large collection of ice floes.

These floes bump into each other. I would guess smaller, thinner floes are much more mobile because they slow each other down less when bumping into each other. When the wind turns, the floes will turn almost as fast. Larger, thicker floes need more time to realize what's going on.

In other words, when the pack consists of mostly thin floes, it will spread out more easily and thus faster than the opposite.

In other words still, the thinner the ice gets, the more important it allows atmospheric patterns to become.

[Rob Dekker]

Interesting new paper in GRL coming up, which attibutes 50% of ice loss to Antopogenic sources :

http://www.miamiherald.com/2011/08/14/2359373/new-study-blames-human-beings.html

It explicitly suggests the influence of thinner sea ice as one of the reasons that ice extent may be more volatile in the near future. Thinner ice, more break-up, more extent even though area reduces...

[FrankD]

Had a little free time today, so I've updated my IJIS SIA/SIE graph.

For newer readers, same idea as CAPIE, but based on the two IJIS datasets (area interpolated from their SIA graph).

Dipped below 2007 on 2nd August and has continued heading south ever since. On 10th August, reached the lowest ever ratio (75.8%), beating the 76.5% set on 21st August 2007. Still going down.

Same bloke, different hair cut.

Even though 2011 extent and area are still above 2007, it's worth taking a look at 2011 -v- 2007 with the compare tool on CT. 2011 has a lot of lower concentration reds and greens and yellows - the pink area of 90%+ concentration for 2011 is not much more than half that for 2007. We're melting through...

[Neven]

Rob, that paper is discussed in the comment section of this blog post.

Funny, the NSF headline is "Arctic Ice Melt Could Pause in Coming Decades". The headline in the Miami Herald is "New study blames human beings for half of Arctic ice melt".

[Rob Dekker]

Regarding the conditions in the Western Arctic right now, as I pointed out before, the atmosphere tends to tilt to a negative radiative balance around this time (halfway August). Surface melt thus starts to become dependent on influx of heat from lower latitudes. However, since the wind currently blows from the north (due to the arrangement of high and low pressure areas) there in not much heat coming on by air. Thus, as the Obouy 5 video already confirms) top melt in the Beaufort has come to a halt, and melting ponds start to reluctantly freeze over.

With pressure zones the way they are, remaining melt will be bottom melt, caused by ocean water heat. The open ocean water in the Beaufort must have accumulated massive amounts of heat during July (and June) which will now determine how fast this slushy ice will desintegrate. Flux bouys suggest there is still 50-75W/m^2 heat coming in by ocean water, which suggests a 1-2 cm/day bottom ice melt.

That's not a whole lot, but with 100 meters of sea surface warmed up over the past two months, this bottom melt will continue for at least another month. That may or may not be enough to melt out the remaining already battered ice in the Beaufort area beyond the 2007 boundaries. And if the winds change, we can see some help from air heating (from Northern Canada and Alaska) which is currently absent.

It's still gonna be a very interesting end of the melting season..

[Conrad Dunkerson]

There are a few statements above to the effect that area must always be less than extent. While this is historically accurate it is not necessarily so from a mathematical viewpoint.

Every year there is a portion of the ice pack below 15% concentration which is effectively treated as zero extent, but non-zero area. Thus, considering only that portion of the ice pack, area is greater than extent. Whether we will ever see a year in which a significant enough portion of the pack is at less than 15% concentration that the total area exceeds extent is an open question. I'd have said it couldn't happen, but this year is showing that it IS possible for large portions of the pack to exist as 'slush' with an effective extent of zero.

Note that this also means the 'CAPIE' figures are in a sense overstated. If extent were calculated on >0% sea ice instead of >=15% sea ice the extent values would be larger and the CAPIE values correspondingly lower. However, since every identifiable speck of ice counts for area, but only >=15% counts for extent there is a degree of inspecificity in the factor. That is, identical CAPIE values can be reached for different ice concentration amounts.

This divergence is probably very small... except in years like the current where there is a significant amount of <15% concentrated sea ice.

[Neven]

Conrad, I believe SIA has a cut-off of 15% as well, meaning that everything below 15% isn't counted. But I could be wrong, I keep mixing things up when it comes to SIE and SIA.

[Seke Rob]

In past the area was inferred from the extent, literally pixel counting those that meet predetermined thresholds. This wiki http://en.wikipedia.org/wiki/Measurement_of_sea_ice discusses the method and mentions PIOMAS, suggesting it's rather current. This also popsup in search of NSIDC discussing the calculation: http://nsidc.org/data/smmr_ssmi_ancillary/area_extent.html . Probably not a new item linked on this blog.

[crandles]

Re "Theoretically, it could reach near zero"

Umm, Theoretically, it could reach near 15%.

But I don't see that happening with large areas/extents. If there are some areas at 15.1% then there are likely other areas at different%. So only when we get down to something like a mere 10000km^2 extent 1560km^2 area will it be near 15%. Of course, at the other extreme the last moment before the ice disappears could be 156km^2 of extent and area for 100% CAPIE.

(I am with Neven in thinking that everything below 15% isn't counted.)

[Kevin McKinney]

Yes, I believe crandles is right about the 15%--IIRC, we hashed this out last year.

On another track, I was struck by Neven's point way up top on this thread that the current CAPIE dive suggested the 'suddenly melting away' scenario (not melting completely this time, of course.) Obviously, there are basically two ways CAPIE can fall; area declines or extent increases (at least relatively.)

Speaking crudely, the former case means melt, and the latter divergence, no? So if the drop is not an artifact of different area and extent algorithms--an idea also suggested up top--then Neven's suggestion seems supported.

I'll be watching to see what happens with CAPIE over the next couple of days. The state of the ice leads me to think there's some degree of significance to the CAPIE record--but maybe that's my confirmation bias showing.

[Lucia (The Blackboard)]

Extent loss is at a relative high. The curves below are icomputed by taking the difference of 7 day smooth extents separated 7 days. Black is 2011:

(Note: the final dot is computed assuming the daily extent loss is sustained. It just lets you see if the daily loss happens to be higher or lower than the recent smoothed value. Today, it happens to be very near the smoothed values, but this is not always so.)

[DeWitt Payne]

Neven,

You may have named the index, but I've been using since 2008 when I started following CT ice area as well as JAXA extent. Back then I had to access CT daily to get the numbers because I didn't know the URL's for the digital archives. In fact, I may have been the first to post those URL's publicly after I received them from William Chapman. I should probably search CA to see when I mentioned it first. I'm pretty sure it will be earlier than last year since the last CA ice thread was in 2009.

[Patrice Pustavrh]

Well Kevin, regarding the "drop" being artifact of alghoritms (and also, using different sensors, as far as I know IJIS is using AMSR-E and CT uses SSMI), it is also true that both IJIS and CT area are rougly on the same path - yes, sudden drop is most probably artifact of algorithms, but couple of day running average is not, as these errors are canceled out. And we are seing area/extent ratio quite low for a couple of days and also we have a 2007 like decrease in area too, so there is a lot of melt in progress. And it is safe to say that ice is more dispersed in 2011 than in 2007.

[Neven]

Hi, DeWitt. Yes, I remember seeing you over at the Arctic sea ice threads over at CA in 2008 and at Lucia's in 2009. I always appreciated your posts and those of a certain Andy.

[George Phillies]

With respect to wind, air volume in and out of the Arctic must nearly match -- "nearly" because air pressure does fluctuate. Thus cold air blowing out in zone A must be balanced by some temperature air blowing in, in zone B.

[crandles]

I think air generally sinks in polar areas so there cold well be net wind blowing out horizontally replaced by air from above.

[Seke Rob]

Sinking air compresses and releases heat if there is some drive behind it.

[Stevemosher.wordpress.com]

nice digs Neven.

I love your animations.

back to lurking and learning.

[Neven]

CT SIA shot up with +63K and IJIS SIE was -76K for that date, so CAPIE is back in safe haven with 60.29%.

[Kevin McKinney]

An area *increase?* Isn't that a bit, er, unexpected at this point? How do we interpret that?

[Seke Rob]

A bunch of <15% compacted?

[crandles]

Not really that unusual at all eg

2010.6083 -1.3569711 4.0605297 5.4175010
2010.6110 -1.2773265 4.1104975 5.3878241

2009.6083 -1.0783485 4.3391523 5.4175010
2009.6110 -0.9626269 4.4251971 5.3878241

2008.6165 -1.6043165 3.6525917 5.2569079
2008.6191 -1.5570961 3.6814303 5.2385263

2007.6110 -1.8015233 3.5863006 5.3878241
2007.6136 -1.7220306 3.6162455 5.3382764

seems most years have upward moves. Perhaps 63k is on the large side but as you see, I found larger in 2009 at similar time.

[crandles]

It surely takes a lot of <15% ice extent to compact to make 63k area especially if (more likely when) some >15% is also melting.

So perhaps sensor noise has just happened to be downward yesterday and upward today?

[Phil263]

May be these erratic movements are telling us that SIA is not a very reliable measure and that we should stick to SIE as a proxy indicator for the general state of the ice!
Also it looks like temperatures in the High Arctic are now starting to dip below zero freezing up all those melt ponds.

[Seke Rob]

August 15, 2004 +130,655, August 5, 2001 +148,681... but that was in a different Arctic era so to speak. :D

It's the sensory uncertainties that make me look at more consecutive days [in numbers] before looking really up.

[crandles]

>"May be these erratic movements are telling us that SIA is not a very reliable measure and that we should stick to SIE as a proxy indicator for the general state of the ice!"

That would be a perplexing conclusion when I find that CT area at 15th Aug is a better for predicting extent minimum than extent itself on 15th Aug. (By 31 Aug Extent is better but extent obviously has to get better at some point before the minimum.)

Perhaps (rather weidly), the average of the last two days is more reliable than usual because there is a good change that the noise cancels out more than it usually does.

[Kevin McKinney]

Thanks, crandles. Interesting, though I'm still not too sure what to make of it. . . if anything beyond noise. At the least, though, it's good context to have.

[Lucia (The Blackboard)]

An area *increase?* Isn't that a bit, er, unexpected at this point? How do we interpret that?

There could be resolution issues. I seem to recall JAXA's stated reason for highlighting extent in graphs is they believe they can measure extent more accurately than area.

As we approach the area minimum, even the 7-day average has wobbled up and down a little. So, don't be too surprised to see these.

crandles: I agree the area predicts the upcoming minimum extent better than current extent. So, whatever the relative error in measurement might be, the area predicts better.

[Artful Dodger]

Q. Have you ever seen a friend in a room, come back the next day and they were in a different room? Did you assume they have a twin?

A. No, you go back to the first room to see if they've left!

Some background:

AMSR-E flies onboard NASA's Aqua bird, in a Sun-synchronous polar orbit, 14.5625 times per day (98.4 min/orbit). AMSR-E's sensor swath is 1445 km wide, and is always advancing to the West over the Earth's surface. The last orbit of each day hugs the International dateline/Greenwich Meridian.

As a consequence of these orbital parameters, when there is a westward advection of sea ice crossing a swath boundary, that the area can be counted twice in a single day. Further, when this happens near 180E longitude, the sea ice can have over 22.5 hours to move, but still be counted in that day.

The solution to this observational quirk is as simple as it is obvious: go back and check the other room.

Here's how SIA in crandles examples above behaved the following day:

Year.Frac Delta-SI
2007.6136 +29,945
2007.6165 -95,341

2008.6191 +28,839
2008.6219 -46,257

2009.6110 + 86,045
2009.6136 -100,286

2010.6110 +49,968
2010.6136 -66,414

Watch the 2-day average SIA. Tomorrow's CT data will tell the tale of sea ice displacement. Or, maybe it's Evil Twin!

[Artful Dodger]

Hi folks,

The key use for CAPIE is that it provides a means for a numerical explanation of sea ice loss rates during the Arctic Summer. CAPIE directly quantifies the effect of the albedo flip on solar heat gain within the ice pack. Let's see.

The following chart compares insolation reaching the surface at 76 N latitude (in watts/m^2) to the effective heat gain (Solargain) after compensating for the proportion of open water within the pack ice, and the different albedo of sea ice and water.

First, some notes:

• 76 N latitude was chosen as a median value for the pack ice. Obviously other values, or more refined methods, could be employed, but using 76 N in this example allows us to see the trends in a useful way
  
• the chart above uses a 2-day average for CT SIA, to match IJIS practice for SIE, and to smooth some of the observational issues in sampling SIA
  
• I refer to this 2-day averaged SIA/SIE as CAP2E (pronounced CAP-TOO-EE)
  
• Note that I have also created a 1-day index I call CAP1E, which employs 1-day SIE data derived from IJIS daily reports, but this method is beyond the scope of this comment
  
• this data neglects daily cloud albedo feedback, for which I have no data source
  
• keep in mind there are both positive and negative cloud albedo feedbacks. Empirical studies have shown the balance point is somewhere near 70% sea ice concentration
  
• this treatment is not an energy budget, so also neglects oceanic heat flux (which increases throughout the Summer).
  

Now for some observations:

• Solar heat gain on June 20, 2011 is estimated to have been 157.8 watts/m^2 within the domain of the Sea Ice Extent (the interior of the pack, not the peripheral seas which have already melted).
  
• solar gain peaked on July 2, 2011, reaching 108.2% of the June 20 value
  
• the first day in 2011 that solar gain was less than 100% of the June 20 value was July 28 (day 209)
  
• this is 39 days continuously @ 100%+ of the Solstice max solar heat gain!
  
• solar gain on Aug 8 was 90.8% of June 20, and 80.8% on Aug 14
  
• on Aug 17 (last day of available data), solar gain was still 74.1% of June 20, or 117.0 w/m^2. Top melt season isn't over at 76N!
  
• since July 28, insolation has gone down 45%, but solar gain has gone down only 26%
  
• this last ratio is the literal fingerprint of the albedo flip, directly causing the lengthening of the Summer melt season
  

This is a long comment already, so I will post a data table separately. Questions and thoughts are welcome.

Cheers!
Lodger

[Bob Wallace]

Suffer another of my newby questions, please.

The IARC-JAXA extent graph is based on larger cell sizes than is the Bremen extent graph. The JAXA shows 2007 lower than 2011, Bremen shows them tied.

Does this not mean that there is more open water to ice than the JAXA-sized sampling indicates?

Is the IJIS extent based on larger or smaller sample cell sizes? (Is it 25 km vs. 12.5 km?) If CAPIE is based on the 'larger square' data wouldn't that tend to throw it off?

[Artful Dodger]

Here is the data table promised above:

Date: Day CAP2E: Insol: Gain: %Sols:
Jun20 171 82.89% 386.90 157.8 100.0%
Jun21 172 82.45% 386.85 158.8 100.7%
Jun22 173 82.04% 386.69 159.8 101.2%
Jun23 174 82.03% 386.42 159.7 101.2%
Jun24 175 82.42% 386.04 158.6 100.5%
Jun25 176 82.45% 385.55 158.3 100.3%
Jun26 177 82.19% 384.96 158.7 100.6%
Jun27 178 81.51% 384.26 160.0 101.4%
Jun28 179 80.30% 383.45 162.6 103.1%
Jun29 180 78.34% 382.53 167.0 105.8%
Jun30 181 77.19% 381.50 169.3 107.3%
Jul01 182 76.55% 380.37 170.3 107.9%
Jul02 183 76.14% 379.13 170.7 108.2%
Jul03 184 76.29% 377.78 169.8 107.6%
Jul04 185 76.30% 376.33 169.1 107.2%
Jul05 186 76.73% 374.77 167.4 106.1%
Jul06 187 76.12% 373.10 168.1 106.5%
Jul07 188 75.31% 371.33 169.2 107.2%
Jul08 189 75.20% 369.45 168.6 106.8%
Jul09 190 75.34% 367.47 167.3 106.0%
Jul10 191 75.55% 365.38 165.9 105.1%
Jul11 192 76.11% 363.18 163.6 103.7%
Jul12 193 76.34% 360.88 162.1 102.7%
Jul13 194 75.92% 358.48 161.9 102.6%
Jul14 195 75.57% 355.97 161.6 102.4%
Jul15 196 75.25% 353.36 161.1 102.1%
Jul16 197 74.97% 350.65 160.5 101.7%
Jul17 198 74.76% 347.83 159.7 101.2%
Jul18 199 74.33% 344.92 159.2 100.9%
Jul19 200 72.90% 341.90 160.9 102.0%
Jul20 201 70.73% 338.78 164.1 104.0%
Jul21 202 69.46% 335.56 165.2 104.7%
Jul22 203 68.69% 332.24 165.2 104.7%
Jul23 204 67.63% 328.82 165.7 105.0%
Jul24 205 66.79% 325.30 165.6 105.0%
Jul25 206 65.91% 321.69 165.6 105.0%
Jul26 207 65.89% 317.98 163.7 103.8%
Jul27 208 66.89% 314.17 159.8 101.3%
Jul28 209 67.07% 310.27 157.4 99.8%
Jul29 210 65.91% 306.27 157.7 99.9%
Jul30 211 65.01% 302.18 157.3 99.7%
Jul31 212 64.45% 298.00 156.1 99.0%
Aug01 213 63.49% 293.73 155.7 98.7%
Aug02 214 62.03% 289.36 156.0 98.9%
Aug03 215 60.76% 284.91 155.9 98.8%
Aug04 216 61.24% 280.36 152.6 96.7%
Aug05 217 61.81% 275.73 149.1 94.5%
Aug06 218 61.34% 271.01 147.3 93.4%
Aug07 219 60.54% 266.21 146.0 92.6%
Aug08 220 60.60% 261.32 143.3 90.8%
Aug09 221 61.51% 256.36 139.1 88.1%
Aug10 222 61.16% 251.30 136.9 86.8%
Aug11 223 60.29% 246.17 135.4 85.8%
Aug12 224 59.15% 240.96 134.3 85.1%
Aug13 225 58.52% 235.68 132.3 83.8%
Aug14 226 59.73% 230.31 127.5 80.8%
Aug15 227 61.09% 224.88 122.6 77.7%
Aug16 228 61.16% 219.36 119.5 75.7%
Aug17 229 60.76% 213.78 117.0 74.1%

[Artful Dodger]

Bob: IJIS uses uses a 12.5 km grid. The grid size does not significantly affect Area/Extent as long as the same concentration cutoff (ie: 15%) is used for both metrics.

[idunno]

Hi Lodger,

Looking at the 76N on Uni Bremen today, it seems that only about 25% of it is actually ice covered. The rest is either land, or is open ocean that is not measured as part of either extent or area. How are you accounting for this?

[Artful Dodger]

Hi idunno,

You're quite right, there are regions within the main pack that have lower local sea ice concentration, especially in the E. Siberian and Laptev seas, as per Uni-Bremen/AMSR-E charts.

However, I'm attempting to characterize the effect of the albedo flip on the remaining pack as a whole, using (relatively) simple methods. The next logical step is what Maslowski does with his energy balance model, which is why I will leave this treatment the way it is! Useful for the intended purpose ;^)

I'm trusting that the averaging effects of the distribution of sea ice will be represented by the mean. 76 N l attitude is a useful median value for this comparison. Picking another latitude, say 80N, and you'll find insolation actually goes up. Using a high-resolution (24-steps/day) routine for Aug 17, we see the minimum insolation occurs at around 77N, with higher values both to the North and South:

Lat: Aug17:
70 232.1
75 218.6
76 217.1
77 216.2
78 216.9
79 217.7
80 218.4
85 220.9
90 221.8

As I mentioned, you could also specify a better representation of insolation. For example, you could solve the definite integral from Aug 17 00:00 to Aug 17 24:00 hrs for this equation:

altitude = asin(sin(LAT) * sin(DEC) + cos(LAT)* cos(DEC) * cos(H))

and then integrate solar altitude from the (variable southern) latitude of the pack ice domain to the N. Pole (blimey!)... But you'll find it doesn't change the answer an appreciable amount after June 20, and it is a serious PITA. ;^)

So for the purpose of characterizing the effect of the albedo flip across the domain of the remaining ice pack, I think this approach is appropriate, and gives useful results.

[FrankD]

After the discussion on albedo changes earlier in the year, I'd had the nagging feeling I was waiting for something to come of it, without knowing quite what. Now I know what I was waiting for - and all I can say is "wow!".

I dips me stovepipe hat to you, Dodger, you loveable scamp. That's some quality work you've done between filching gentlemens handkerchiefs... :^)

[Artful Dodger]

yeah, been working on this since Jan 13... Next I think it's time to do some correlations between SIA or SIE and my 'solargain' estimate. I'm on the River until Thursday, so please, carry on without me!
Cheers,
Lodger

[Leif Ratinger]

Area per extent ratio using real data from a single consistent source:

http://deponie.bplaced.net/arctic/rat.svg

This is calculated using the IUP Bremen GeoTIFF files and pixel area data from NSIDC. Missing data are interpolated up to a 13-day gap and filled by persistence up to 7 days. Pixels around the pole that were ever undefined are always counted as 100%; remaining missing data (very few) too. Data includes lakes.

I also removed every pixel that is less than 12.5 km away from a land pixel to remove most of the spurious coastal ice. 2007 had more such spurious ice and that's the reason that 2011 is at 2007 level in IUP Bremen graphs.

Extent: http://deponie.bplaced.net/arctic/ext.svg

Area: http://deponie.bplaced.net/arctic/area.svg

To compare, same without coasts masked out:

Ratio: http://deponie.bplaced.net/arctic/ratc.svg

Extent: http://deponie.bplaced.net/arctic/extc.svg

Area: http://deponie.bplaced.net/arctic/areac.svg

Current data with marked coasts: http://deponie.bplaced.net/arctic/2011-08-21.png

Coast and pixels ever undefined after interpolation and persistence (marked red): http://deponie.bplaced.net/arctic/mask.png

[Neven]

Wow, that's impressive, Leif! How did you do all of this? Amazing. If you want, we could turn this into a guest blog post together.

[Seke Rob]

On record low CAPIE, NSIDC's Extent-Area ratio, including the 0.31 million square km blind spot counted as 100%, they hit an August record too of 59.96%. Their previous low stood at 61,72% set in 2008. [See Chart] Almost slipped off the scale. The monthly NSIDC report hopefully has comments on this topic of concentration (expecting Sep.3-5 for it to come out)