Cryosphere Today 2013 Arctic SIA daily minimum: July poll

[Neven]

THIS POLL WILL RUN FOR TWO WEEKS (until July 20th). YOU CAN CHANGE YOUR VOTE. MIND THE DIFFERENCES WITH THE NSIDC SEA ICE EXTENT MONTHLY/SEPTEMBER MINIMUM. THIS IS CT SEA ICE AREA DAILY MINIMUM.

Same text as last month's poll:

It should be 'fun' and interesting to hold a poll each month, to see how opinions change according to sea ice rhythm and pace. These monthly polls can then also function as a dedicated thread for everything concerning Cryosphere Today sea ice area. Daily updated data can be found here.

Again, this poll is for Cryosphere Today sea ice area daily minimum (unlike the other popular measure: NSIDC sea ice extent monthly minimum), so we're looking for the lowest total Arctic sea ice area number on any given day at the end of the 2013 melting season.

Here's how things are currently looking based on data up to June 4th:



These are the daily minimums for the last 7 years (in millions km2):

    2005: 4.092
    2006: 4.030
    2007: 2.919
    2008: 3.004
    2009: 3.425
    2010: 3.072
    2011: 2.905
    2012: 2.234

Again, try to use this thread to discuss CT SIA mainly. There are other threads for the other topics, and if there isn't, feel free to open one.

I will open a new thread for the 2013 CT SIA daily minimum August poll once this month is done. Keep voting. It'll be interesting to see if there are shifts in voting patterns as the melting season progresses.

[Neven]

I find voting very difficult this month. For some reason or other 2013 is lagging 2012 by a million km2, and so I would be inclined to go two bins up this month (voted for 2.25-2.5 last month). But 2012 slowed down a bit in the coming week, as can be seen on the graph I posted, and I'm expecting 2013 to drop like crazy (which will be the subject of the next ASI update). Therefore I will move just one bin up to 2.5-2.75. Depending on what happens the rest of this month I might go back down in August.

[OldLeatherneck]

I jumped up just one click to 2.0 - 2.5.  So much of the arctic is just a bowl of slush waiting to be demolished. 

[Tor Bejnar]

I've gone up one bin - still expect a record low CT Arctic sea ice area (SIA), just not so impressive (depressing?) of one.

[Richard Rathbone]

Up 1 bin for me too. Though in round numbers I'd still say 2.5, I'm going above the line rather than below it this time.

June was consistent with the May picture of generally later melt than 2012. The later it gets, the more sunlight has bounced off into space never to return. There will be catch up because there will be areas getting sun in July/Aug 2013 that melts ice where it just heated open water in 2012, but no sign yet that there's a chance of an overtake, and the clock is ticking.

[dree12]

I'm staying steadfast at 2.75 to 3.00, pending July data. Although I expected a slight recovery from 2012, I bumped my prediction up last month due to the even slower-than-expected melt. As June picked up the pace once again, I'm likely sticking with this unless either a phenomenal drop or an unprecedented slowdown occurs.

[icebgone]

For me weather remains the key.  The current cold trend for north of 80 degrees remains in place and currently is cooler than ever.  I would not be surprised to find that everything north of 80 continues as is and may even begin refreezing prior to the start of September.  If the weather changes then I will look at coming down a bin or two but either way this is a brief recovery year.  Given the strange weather and the warmth lurking in the Atlantic and now the Pacific too, a new low maximum ice area and volume for winter 2013 may result.  Perhaps even an early bottom melt beginning as early as March 1, 2014?

[DavidR]

1.75 - 2
Consistent with my estimate for SIE assuming SIE is 50% greater than SIA at the end on the season.  Essentially assumes the oceans are warming more consistently than the air and with average weather patterns over the entire melt  season we'll see a little more melt  every year. 

The current rate of SIE decline, which intersects with 2012 SIE by 15 July, suggests that anything is possible this year.

A March estimate for sea ice thickness suggested it was 8% less than last year.  So while the air may  be keeping the surface frozen warmer water may  be removing the underlying support.

[ChrisReynolds]

My previous prediction had been 1.75 to 2.00 as the peak of a probability distribution weighted towards the high side. In other words my expectation had been that if wrong the real outcome would have been between 2.00 and 2.23 (last year's record). This had been a judgment based on data, not a numeric method.

I had to see what the June Cliff brought before amending my prediction. Now that is over I am able to do so.

From 30 June to the daily annual minimum of CT Area the trend is flat.



In other words area loss from 30 June is unchanging. Using the average area loss, and its standard deviation (sigma) it is possible to make a numeric prediction. The hindcast predictions for past years are shown in the graph below, with the actual minima the purple trace (mislabelled series 4).



These are calculated as the average area loss from 30 June applied to the actual area at 30 June for each year. With the standard deviation applied to give 1 SD bounds. When using 1.25 sigma  the bounds are widened and the actual values fall within the bounds 76% of the time, so as a trade off between size of bounded area and % of the time that actual falls within hindcast, I have chosen 1.25 sigma for a prediction.

In only one instance has a 2 sigma deviation occurred, this is what would be needed to meet 2011's CT Area record. A sigma deviation of -2.90 would be needed to at least meet 2012's record, this has never occurred.

The numerically calculated prediction is: ~75% probability of this year being between 3.22M km^2 and 4.15M km^2, centred on 3.68M km^2. This is shown on an anomaly plot below.



As a concession to sea ice state in Laptev to Pole, and thinner ice than in any year that has a minimum between 3.68M and 4.15M km^2, in expectation of a slight fall of anomalies this year I have narrowed the range to the lower half, so my prediction is 3.68M to 3.22M km^2, mid point 3.45M km^2.

So I have voted for 3.25 to 3.5M km^2, but some of my vote technically falls in +3.50M. The probability of this prediction is less than 76% but by how much I am unable to calculate due to the heuristic element.

[Steven]

I voted for 2.75 - 3.00.

Much will depend on the shape of the curve.  If we get a steep ending like last year, then we could end up close to a new record.  If we get a flattened curve, then we could end up way above last year.

The abundance of FYI should start playing a major role at some point.  But the slow start and the lack of insolation may counteract that.  Weather will still be the decisive factor, IMHO

[Richard Rathbone]

With a projection of area anomaly equal to end June coming in that far above what I am getting from a projection of volume anomaly equal to end May (as spotted by crandles), I'd better double check where the difference is coming from.

One is I am doing too much arm waving and not enough calculating so lets put that right.

PICT thickness is following the post 2010 pattern closely, and I see no reason to put it anywhere but 1.4m when area is minimum in 2013.

I project end season volume to be up by 0 to 2e12 m3 because it was up 1 at end May and thats the historical range of variation in volume loss from end May to end season. That gives area up by 0 - 1.4 from last year, with anything from 2.2 to 3.6e12 m2 being plausible, and the actual value depending on just how good or bad the conditions are for volume melt.

So calculation is 2.9, and then its arm waving.

1) more FYI means more favorable for melt.
2) June weather wasn't good, but compared to the way I thought it would develop wasn't bad either. I think the normal catch-up mechanisms that tend to balance the anomaly will deal with the amount it fell behind in June and I'm still thinking FYI will push it towards the high melt part of the historically plausible range.

I'm not revising 2.5-2.75, but I am weaselling to 2.75 - delta, rather than 2.5 + delta which was what I first said.

The difference between 2.9 (from volume anomaly) and 3.7 (from area anomaly) is right at the edge of the plausible range. A central projection hit from one would just about be plausible under the uncertainty range for the other, but there's a high chance that come September I regard at least one of these projection methods as broken. Both will break eventually, and I reckon at least one must have been lucky to have stayed in its ballpark post 2010.

[Neven]

Third century break in a row, will break through the 7 million barrier tomorrow, but still a massive 834K behind 2012. Should be less than 500K by the end of the month, if 2013 follows the 2009 pattern. Perhaps less if the weather is favourable during the second half of June.

[Alistair]

From day 185 to 190 2012 only lost 240k (6.216 -> 5.976) is this when 2013 starts to catch up?

2013 has been on the cool side for the past couple of months but has still been losing area at a consistent pace.  I can't see this changing any time soon.

[TerryM]

3.0 - 3.25
The slow spring melt won't be made up for in the short time we have left (I don't think). This year may answer questions about how important early melt is and how much attention we should pay to Arctic temperatures.
I'd been thinking 2015 + 2 years for melt out. This year's out of the running, but there are still 4 yrs left. Every time I think I know what the Arctic will do next it throws me a curve.
Terry

[Tor Bejnar]

I am appreciating Chris's analysis. 
I have previously noted that 2009 had a relatively large late June SIA and then had a spectacular 3 M km^2 loss in 26 days (per Pettit's CT SIA bar graph showing days per 1M declines).  The other two years on Chris's bar graph of prediction - actual final area with largest low predictions (greatest neg. sigma) are 1984 and 2008.  1984 had a respectable 3M loss in 28 days around this time; 2008's 31 days is not impressive.  (Looking at these days per 3M drop makes me think how precious a few days, one way or the other, are!) [The data I'm looking at doesn't tell me anything about 1997, the year with the greatest sigma.]

I can therefore draw no clear conclusion, but do suggest that relatively large end-of-June SIA supports the possibility of greater subsequent loss that year.  But a sigma of 2.9 would indicate a 2007-like new game in town.  I, for one, would not be shocked.
 

[Gray-Wolf]

Hi TerryM!

I wish I had your confidence!

 Could it not be that the 'slow start' had a large element of 'melt processing' going on instead of just slow melt?
 could it also be that this 'processing ( mechanical errosion)  is now about to make it's impacts (back then) know by the speed of losses of the ice thus 'processed' over the next period of 'melt favourable' conditions?

[TerryM]

GW
Certainly didn't intend to exude confidence. I have very little faith in my WAGs.


Terry

[icebgone]

Yes, I would expect 2013 to shrink the area and extent differences with 2012 over the next seven days.  After that, 2012 ramps up area loss over the next seven days which 2013 will have difficulty matching.  Expect area difference of approximately 750k between 2012/2013 by Mid July.  IMHO Compaction from High Pressure will reduce 15-25% of area and extent difference by August 1.

[jdallen]

Yes, I would expect 2013 to shrink the area and extent differences with 2012 over the next seven days.  After that, 2012 ramps up area loss over the next seven days which 2013 will have difficulty matching.  Expect area difference of approximately 750k between 2012/2013 by Mid July.  IMHO Compaction from High Pressure will reduce 15-25% of area and extent difference by August 1.

And yet, some sites are reporting a drop of less than 16K in SIE yesterday.

(O.o)

It is simply maddening!

[Neven]

CT has had two slow days in a row now, whereas the trend lines on all the extent graphs are dropping like rocks. Last year it was the other way around, with 2013 taking the lead on the CT SIA graph, and IJIS SIE lagging big time and only dropping to first place after GAC-2013 hit.

I know there were more melt ponds last year than this year, but I don't think that's it. Does anyone know what could be causing this discrepancy?

[Sourabh]

I know there were more melt ponds last year than this year, but I don't think that's it. Does anyone know what could be causing this discrepancy?

One of the reasons I think is how the extent is defined itself. Some charts assume above 15 percent or 25 percent. So, it is possible that small melt in ice area reduce that percentage coverage below that baseline. Therefore, extent is dropping faster than area.

[Steven]

CT has had two slow days in a row now, whereas the trend lines on all the extent graphs are dropping like rocks. Last year it was the other way around, with %2013 taking the lead on the CT SIA graph, and IJIS SIE lagging big time and only dropping to first place after GAC-2013 hit.

I know there were more melt ponds last year than this year, but I don't think that's it. Does anyone know what could be causing this discrepancy?

For what it's worth, I think the timing plays an important role.

Hudson, Baffin and western Kara melted rapidly over the last weeks.  At the same time most of the other regions have been colder than usual.  Some of the old melt ponds drained and are showing up as high concentration ice again.  The cold weather also delayed the formation of new ponds.

It is striking that many regions have a high ice concentration now (purple colors):

http://igloo.atmos.uiuc.edu/cgi-bin/test/print.sh?fm=07&fd=07&fy=2013&sm=07&sd=07&sy=2012

This may be due to the slow top melt.  But it may also suggest that PAC-2013 and other weather events compacted the ice in some of the regions surrounding the Pole Hole.  I would expect PIOMAS to confirm that in its next regional update.

It seems that a further compaction of the ice is taking place now, including the Pole Hole itself, so the present situation may persist for a while.

[AndrewP]

I will stick with the 3.00-3.25 bin.

I am now leaning towards the higher side of that bin but it could fall anywhere between 2.6-3.9 (narrowed from 2.25-4 in June).

I would also like to reiterate that the bounce from last year should have been apparent to all by early June for the same 3 primary reasons I have already stated:

1. Increased area (area loss from early June to minimum has low variation)
2. Increased volume
3. 2012 had very bad weather which should not have been expected to repeat itself. 2009 and now 2013 represent cooler summers in our new climate. 2007 and 2012 represent warm summers. 2008,2010,2011 represent more normal summers. 2012 is not yet the new normal.

The bin below me had some merit and some objective arguments could be made for that bin in June. However the guesses in the 2.5-2.75 and especially 2.25-2.5 and lower bins wreaked of a lack of objectivity. I am glad to see most guesses are now above the 2.25-2.5 bin.

Also I am glad to see, Chris, that you are now using the relationship between early summer CT_area and minimum CT_area as your primary method now. When I used this in June you stated that the burden was on me to show why this relationship would continue. I never responded at the time, but actually the burden is on others to show why this longstanding statistical relationship will change. I am glad to see you are now accepting this logic. I am still guessing a bit lower than this method yields alone due to the exceptionally low volume to area ratio (IE thickness).

[ivica]

Uncertainty is unbearable for me and I have no desire to vote, choosing one bin do not express my opinion.
Possibility to choose k consecutive bins (expressing uncertainty that way) would be much better.

We had majority of votes in a couple of bins previously, now that is smeared over many bins.
Not much votes for now, how many of you will pass this round because of uncertainty being too big?

Voted for 2.25 - 2.5 to flatten graph & emphasize uncertainty.

[Steven]

I know there were more melt ponds last year than this year, but I don't think that's it. Does anyone know what could be causing this discrepancy?

Forgot to mention (FWIW): the low concentration areas in 2012 are hinting at the advanced melting and floe formation that were taking place, especially in the large neighborhoods of the polynya's, which were gathering solar heat over a longer period.  Much of these regions appeared to be in a further advanced stage of melting last year than they are now.  Not surprisingly IMHO, with the slow start of the current melting season.  Of course, things may change quickly.

The floe formation in the Laptev-North Pole region is currently getting a lot of attention, quite rightly, but the total ice area has been stalling for a while now, it seems, at least as far as the area graphs are concerned.  There seems to be still a lot of ice volume over there.

[ChrisReynolds]

Also I am glad to see, Chris, that you are now using the relationship between early summer CT_area and minimum CT_area as your primary method now. When I used this in June you stated that the burden was on me to show why this relationship would continue.

As I have used 30 June I can't be using the same method as you were, because when we were discussing that method it was before end-June results. When you used it that was (IIRC) before we had seen how the June Cliff would turn out. Nobody could tell how far that was going to go. In terms of correlations through the season with end of season min, and in terms of the June Cliff, that month presents a barrier to numeric methods.

I would have stuck with my previous method, using melt distributions by April thickness in PIOMAS, but that method has been invalidated by the very odd behaviour of this year.

I'll try to find our discussion when I get back from work this evening.

[ChrisReynolds]

Found the final post of our discussion, in which I said what you say. Forgot about using post history, once I remembered finding that was easy.
http://forum.arctic-sea-ice.net/index.php/topic,357.msg7839.html#msg7839

That was on 21 June, it was only on 24 June that I called an end to the June Cliff, a judgment which has turned out to be proven by subsequent events (lucky). More than any other factor it's the June Cliff that's persuaded me. Note that June 2013 volume loss (I used area above) was the greatest excluding 2010 to 2012. Seeing the anomaly levelling in the final week of PIOMAS June data has since persuaded me that we will see a similarly increasing then levelling behaviour of anomalies, as in recent years.

I cautioned back in May that it was too early to say what effect the late start would have, so I stuck to my former prediction. I said I'd revise at the earliest statistically possible time: Early to mid June, that's what I did. From that point of view, I'm afraid I think you've been lucky, that is all.

[AndrewP]

I find it highly improbable that so much of the melt season depends on 1 week in June. Our conversation was on June 21 at which time you were defending a 1.75-2 prediction and now you are at 3.25-3.5. Not that much has changed in less than 3 weeks. Volume and area melt have been a bit slower than recent years which might be enough to justify a 100-300k increase in one's guess.

Part of the problem is heavy reliance on one single method and the fact that you have switched from one method to another from June to July.

The fact is volume was already well above the 2010-2012 average by the end of May. Thus the starting point for any prediction should have been above the 2010-2012 average which is 2.73. So nudge that up to 2.8  or 2.9 because it was above the average. Melt for the first 2 weeks of June was clearly slower than those 3 years so you can nudge that up another 100k. This assertion is validated by the fact that by mid-June PIOMAS had risen above all 3 of those years including 2010 which had a minimum SIA of 3.07.

This doesn't even begin factor in the fact that 2010-2012 were all exceptionally warm arctic summers. And before somebody says that is the new norm, a sample size of 3 is not nearly big enough to assert that especially when 2009 was so much cooler and had a minimum of 3.43. Any probabilistic analysis should factor in the possibility of a 2009 repeat of such a cool summer.

Thus by mid-June (say June 15) the evidence clearly pointed to the 3-3.25 bin (I could accept one bin lower). Prior to June 15 it pointed to the 2.75-3 bin. Prior to June 1st you could have argued that the FYI albedo difference pointed to the 2.5-2.75 bin, although I don't find those albedo differences entirely conclusive. The melt during June has been slow but not so slow as to justify more than a 1 bin increase from June 15th to present and a 2 bin increase from early June to present.

Anybody switching bins by more than 1, possibly 2 bins if they guessed prior to June 15th, had issues to begin with.

The biggest problem I think is that many here including you Chris are relying on a PIOMAS to minimum CT_Area relationship from 2012 alone. 2012 lies on the very low end of the distribution. Most years with similar pre-season volumes to 2012 will not produce similarly low CT_areas come September.

For a prediction in June I would use the following method:

1. Establish statistical relationship between years with similar volume and September SIA and use to make a prediction based on current year volume.

2. Establish statistical relationship between years with similar area and September SIA and use to make a prediction based on current year area.

3. Average method 1 and 2 weighting method 1 at 67% and method 2 at 33%.

4. tweak slightly (100-200k) downwards based on factors such as FYI proportion and location of thickness anomalies. As June progressed I would argue these factors became less relevant since their theorized mechanism (SIA feedbacks) should have already taken much of its effect.

5. tweak slightly (100-200k) upwards due to the fact that 2010-2012 (the years used in parts #1 and 2) were exceptionally warm summers

On June 1 this analysis yielded 2.75-3. By June 15 it yielded over 3 because SIA loss had been slow and SIV loss was likely slow (an assertion which turned out to be correct).

While you might be able to improve on this method slightly I doubt that it would yield anything below 2.5 on June 1 and 2.75 on June 15th.

[AartBluestoke]

I estimate 2.5, almost all in the arctic basin. This assumes that the arctic basin loses about 10% more area than it lost last year, because the thickness is 8% less, and assuming that regions which melted out for more than a month will melt out this year also.


Furthermore i estimate that at the end of this month we will still be above the 2012 value, but i would place less certainty on that prediction than the final number.

All numbes done by eyeball estimate based on regional trends from http://forum.arctic-sea-ice.net/index.php/topic,382.0.html


Direct regional trends

	2012 now	2012 1st aug	2012 min	2013 now	2013 1st aug   minimum
Total	7.15	5.15	3	7.4	5.35	2.5
arctic basin	4.2	3.9	2.8	4.2	3.9	2.4
Greenalnd Sea	0.3	0.2	0.1	0.18	0.1	0.05
Hudson Bay	0.15	0.05	0	0.15	0.05	0.05
East Siberian Sea	0.8	0.5	0	0.8	0.5	0
LapTev Sea	0.4	0.1	0	0.4	0.1	0
kara Sea	0.15	0	0	0.3	0.1	0
Barents Sea	0	0	0	0	0	0
Baffin/Newfoundland Bay	0.15	0	0	0.17	0	0
St Lawrence	0	0	0	0	0	0
Canadian Archipelago	0.5	0.2	0.1	0.55	0.25	0
Beaufort Sea	0.2	0	0	0.4	0.2	0
Chukchi Sea	0.3	0.2	0	0.25	0.15	0
Bearing Sea	0	0	0	0	0	0
Sea of Okhotsk	0	0	0	0	0	0

[Rob Dekker]

Maybe a silly question, but does anyone have a list of the "Cryosphere Today 2013 Arctic SIA daily minimum" over the past 10 or 20 years ?

[Wipneus]

Maybe a silly question, but does anyone have a list of the "Cryosphere Today 2013 Arctic SIA daily minimum" over the past 10 or 20 years ?

   year   minctd
1  1979 5.306725
2  1980 5.507712
3  1981 4.956492
4  1982 5.139060
5  1983 5.386929
6  1984 4.695892
7  1985 4.992847
8  1986 5.381843
9  1987 5.288995
10 1988 5.144891
11 1989 4.815916
12 1990 4.628935
13 1991 4.460384
14 1992 5.026778
15 1993 4.472953
16 1994 4.816096
17 1995 4.410301
18 1996 5.238185
19 1997 4.899706
20 1998 4.262403
21 1999 4.204499
22 2000 4.168766
23 2001 4.533619
24 2002 4.034710
25 2003 4.141665
26 2004 4.282973
27 2005 4.091798
28 2006 4.016919
29 2007 2.919439
30 2008 3.003556
31 2009 3.424597
32 2010 3.072129
33 2011 2.904740
34 2012 2.234009

[Rob Dekker]

Wipneus, you are awesome.
Thanks !
Tomorrow, I'll try my correlation method on these numbers.
http://neven1.typepad.com/blog/2013/07/problematic-predictions-2.html
and see if there is a pattern there that makes sense in terms of physics of energy input.

[Neven]

I have a stupid question as well: what satellite sensor is used for CT SIA, and what is the resolution?

Another century break (117K) reported today, which isn't all that much, especially after  after two slow days (26K and 66K). And so I'm still a bit surprised at this 'lull', with extent dropping like crazy (IJIS reported the 8th century break in the first 9 days of July). Last year it was the other way around.

This can be seen clearly on the CAPIE graph, with 2013 trailing 2012 by 5%:



Yes, 2013 has been much slower up till now, temps are lower, PIOMAS says there's more volume, but still I'm surprised how some of you guys are so confident that 2013 has absolutely no chance of breaking records and will even end up above 2007/2011.

This is the Arctic we're talking about. Nothing is a dead certainty there.

I'd wait for a week or two to see how CT SIA develops, because I'm not sure things have been portrayed all that accurately so far.

[Pmt111500]

the yearly min/max/averages (5-day smoothed) in graphical form for those so inclined  expanded to 2d-flat world disappearance of ice. (note that according to the functions in year 1BC  30% of the earth surface was covered with Arctic ice in summer, this melted in the next winter to cover only 20% of the earth.)

[Frivolousz21]

I voted for Between 2.75 and 3.0 million km2.  My official vote is 2.9 mil km2.

I have been with that since sometime around June 20th.  The turn to a very good to great pattern for ice melt was already anticipated to occur between the end of June and July 20th and be in this mode or a similar one 60-70% of the time from it's arrival until September 1st. 

The general mode of set up has to be a dominant High Pressure over the Pacific side from the Eastern/Northern ESS to the Southern Beaufort/Western CAB.  Different positions and strengths can vary.

But in my opinion the location/compaction is not as big of deal breaker or even that much of a difference vs the frequency.

My original prediction made back on March 8th on Americanwx.com was 2.5-2.75 mil km2 for CT.  So I am glad so far I haven't had to make any large jumps.  My reasoning was the large ant-cyclone that sent so much MYI Westward. 

[werther]

I’m losing sight over which vote I did and which not. Even so, where I voted, I held on to my original margins between 3.2 and 4.0 Mkm2 SIE mean September. Going from there it corresponds with 2 – 2.5 Mkm2 SIA.  So I voted in the 2 – 2.25 box.
This season will stand on its own. Most important is the state of the ice, being prepared for several years. Thin, weak, vulnerable. It is no use to compare to weather/patterns earlier years. Interesting to reconstruct the decline. Not fit to support a forecast.

[dree12]

My prediction (~2. is continuing to be squeezed by slower-than-expected melt. Regardless, I'm sticking with it as I believe Beaufort & friends should eventually melt out.

I do believe beating 2012 will be a very unlikely (p<0.05) event.

[Neven]

I'd wait for a week or two to see how CT SIA develops, because I'm not sure things have been portrayed all that accurately so far.

Forgot to add: also because big day-to-day changes can clearly be seen on the UB SIC maps, and Wipneus' work with Uni Hamburg data in another thread shows that drops seem larger than CT reports.

[ChrisReynolds]

Andrew,

You are arguing with the benefit of hindsight. The fact remains that at the time of our discussion the CT Area anomaly cliff was at best in a pause (having paused previously), and could conceivably have continued to carry CT Area down close to 2012's area at the same time. Furthermore before the PIOMAS data was in it was conceivable that it would end on a rapid drop, opening up the opportunity for further anomaly declines in July. In both cases I am not aware of any data that could have settled the matter to a high probability before it was clear that the CT Area anomaly cliff was over, and that PIOMAS would definitely level.

And incidentally, yes June really is a critical month. Between June 9 and July 9 correlation of detrended ice area (1979 to 2012) from those dates with the area at minimum jumps from 0.30 to 0.60, in plain English that's from a negligible to a working predictive quality. For 2007 to 2012 the jump is from 29 June to 9 July, a jump of 0.34 to 0.76, creating a strong inflection in correlations around that time of the year.

It also might help if I explain how I view and approach prediction. Prediction is a tool for testing hypotheses, no more no less. If one is right one moves onto more interesting territory (the unknown), if wrong it is time to consider the implications. I'm afraid that I consider the satisfaction I may feel at being right as infantile, and something to be suppressed, not indulged in.

In this sense my changing predictions for 2013 represent a shift in hypotheses. Prior to having enough data by end June, my hypothesis had been that thinner ice than 2012 would result in a crash to below 2012's record. By end of June both atmospheric and sea ice data confirmed that the weather of June 2013 had been highly anomalous and had caused a substantial retarding of the melt season. At that point it was obvious that 2013 would not be a good test of the thin ice hypothesis. So I fixed upon a test of whether the late summer melt would follow previous form, or would behave differently, even giving a method of statistical judgment of just how anomalous 2013 ends up being.

[Bob Wallace]

Yes, 2013 has been much slower up till now, temps are lower, PIOMAS says there's more volume, but still I'm surprised how some of you guys are so confident that 2013 has absolutely no chance of breaking records and will even end up above 2007/2011.

This is the Arctic we're talking about. Nothing is a dead certainty there.

I'd wait for a week or two to see how CT SIA develops, because I'm not sure things have been portrayed all that accurately so far.

What might be telling would be to plot out total extent/area for all regions except Beaufort and Kara.  With the two "late to come to the party" regions out of the picture I suspect the race would look very tight.

And then ask the likelihood of those two areas not melting to ~zero by the season end.  I would guess low.

[ChrisReynolds]

For what it's worth, and bearing in mind that a decline in anomalies is feasible, here's my previous prediction with current anomalies.



I mentioned above the idea of prediction to test hypotheses...

As described in one of my June blog posts, June was a very weird month in the context of the post 2007 period.
http://dosbat.blogspot.co.uk/2013/07/june-status-part-2a-atmosphere-slp.html

Here is July so far.



As with June the 2007 to 2012 pattern has still failed to appear!!!

I'm pondering whether the fall in anomalies throughout July and August in most post 2007 years is due to the dipole anomaly set up by the summer atmospheric pattern of 2007 to 2012. If, as I suspect, that pattern fails all this summer then the anomalies shown in the first graph should fail to fall.

[Rob Dekker]

I'm voting in the 2.5-2.75 slot.

The method I used is the same as the one I posted here :
http://neven1.typepad.com/blog/2013/07/problematic-predictions-2.html

Method briefly explained again :

I take 3 variables known in June : (snow-cover in June, SIE in June, and SIA in June) that I think best represent how much energy is getting absorbed into the Arctic, and thus should serve as  good predictors for later state of the ice (including predictions about area and extent in September).

I then run linear regression (and essentially principal component analysis) to determine the parameters for these variables that obtain the best correlation over the past years data.

Don't want to go back too far, since before 2000, the state of the Arctic was very different than now (and thus resp, and thus no combination of parameters obtains a correlation better than simple extrapolation of trends.
But with 3 variables, I need at least 10 data points or so, or else I will surely be 'over fitting'.

I found that going back to 2002 works well and seems to be representative to how well the Arctic responds to these 3 variables.

For SIA daily minimum, the formula for energy that gives the best correlation over the past 12 years or so is this one (june_area normalized (to 1.0) :

 june energy = 0.38 * june_snowcover - 1.57 * june_extent + 1.0 * june_area ;

 *************** Linear Regression Analysis *****************
 11 years analysed (2002-2012) :
 Mean energy -6.289610 Mean ice cover 3.466049
 SD energy 0.660019 SD ice cover 0.622146
 Correlation (R): 0.990791 
 Beta(slope delta-icecover/delta-energy) 0.933936
 Alpha (ice-cover at energy 0) 9.340144
 ice cover = 9.340144 + 0.933936 * energy
 *************** Prediction and confidence intervals *****************
 2002: energy -5.78, area 9.13, extent 11.70, predict 3.94, final 4.03, delta 0.09
 2003: energy -5.59, area 9.05, extent 11.77, predict 4.12, final 4.14, delta 0.02
 2004: energy -5.27, area 9.19, extent 11.52, predict 4.42, final 4.28, delta -0.13
 2005: energy -5.63, area 8.74, extent 11.30, predict 4.08, final 4.09, delta 0.01
 2006: energy -5.85, area 8.34, extent 11.06, predict 3.87, final 4.02, delta 0.14
 2007: energy -6.78, area 8.15, extent 11.50, predict 3.01, final 2.92, delta -0.09
 2008: energy -6.77, area 8.53, extent 11.37, predict 3.02, final 3.00, delta -0.02
 2009: energy -6.38, area 8.92, extent 11.47, predict 3.38, final 3.42, delta 0.05
 2010: energy -6.72, area 8.02, extent 10.83, predict 3.07, final 3.07, delta 0.01
 2011: energy -6.73, area 8.20, extent 10.99, predict 3.05, final 2.90, delta -0.15
 2012: energy -7.69, area 7.71, extent 10.97, predict 2.16, final 2.23, delta 0.07
# -2013: energy -7.33, area 8.57, extent 11.58, predict 2.50, final ?
Standard Deviation for de-trended prediction : 0.084

Note the amazing correlation factor R of 0.99, and resulting Standard Deviation on the prediction of only 84 k km^2.
Now, gotta take that SD with a grain of salt, since I included all known data in the analysis.

So I ran the entire process again, this time excluding individual years and then see how well this method can predict SIA that it did not know about. With that method, I get R=0.987 and resulting SD of about 100 k km^2. Which is still remarkably good and much better than straight line extrapolations of area and extent alone.

Which brings my prediction for 2013 SIA to 2.4 - 2.6.
No slot for that, but my got feeling is that 2013 will go to the high end of this range, so vote into the 2.5-2.75 slot.

Sorry for the long post, but after many experiments, I'm getting some confidence in this method.
We'll see in September if the Arctic agrees )

[Frivolousz21]

The sea ice getting roasted.

[Pmt111500]

Checking  the average losses through Julys since considering changing my vote from last month. I guess I delay my vote almost to the closing date, at least it looks like July 17th is the date I'll decide. Then there's the (about 7 day) period in the end of the month where catching up on 2012 may happen. In other words, (on this week) not confident enough to vote for a new record. (Note the early years lost much ice during this week, but that was from marginal seas.)

[werther]

Just saw CT reporting -136K again. So we're back in the big ones after a short hickup.
While '12 had a couple of real blows through the coming days, it 'll be interesting to see whether '13 can keep up.
When you look at ECMWF, it should not be impossible. Big wind now over the Arctic Basin Bight (credits to Lodger). The Chukchi is losing. So does the ESS, ice is thinning over there. Likewise in the Beaufort. More holes, deeper greys over the old fractures.

[Jim Pettit]

Just saw CT reporting -136K again. So we're back in the big ones after a short hickup.
While '12 had a couple of real blows through the coming days, it 'll be interesting to see whether '13 can keep up.
When you look at ECMWF, it should not be impossible. Big wind now over the Arctic Basin Bight (credits to Lodger). The Chukchi is losing. So does the ESS, ice is thinning over there. Likewise in the Beaufort. More holes, deeper greys over the old fractures.

As was noted a day or two ago, over the next five days, 2012 lost 813k km2 of area. But then, over the next 15 days after that, it fell by just 867k, If 2013 is to catch 2012--or even come close--I believe it'll happen during that latter stretch.

11 days ago, 2013 trailed 2012 by 1.12 million km2. That gap has been cut nearly in half at just 607k.

[helorime]

Being new to this I keep hesitating to vote.  It still seems to my like the extent should crash, but it hasn't yet.  Comparing thickness from July 10 2012 to 2013 the current Ice seems way more vulnerable and the slight area differences trivial by comparison, but maybe I am completely off track....

[johnm33]

Took the bottom bin, here's why. I think we'll lose practically all the ice, mainly by bottom melt. The PAC drew the Atlantic water into the central arctic, replacing the huge volume of ice shunted south, where it caused much of the melt, then it swung tight round the pole and exited Fram leaving that relatively ice free. This looked like a slow start, and it was cool but all that cloud/mist took a lot of energy to evaporate and looking towards Norway the Atlantic waters are anomalously warm as they are off mainland USA.  Now with a more 'normal' flow those waters have regained their usual route and are flowing beneath Beaufort and on to the archipelago. As a consequence of the cracking event the ice is still fragmented leaving the underside smoother, more rubble than ridges, so has less of an insulating freshwater layer beneath it, and what there is will be pushed toward Fram/NWP/Nares by the pressure of the inflow, and if it is indeed smoother this will flow quicker.
My guess is that we'll see lots of sudden in-situ melt soon and as the fresh water leaves Fram more ice in the Greenland sea, given present conditions Beaufort could be ice free in 2 weeks leaving plenty of time for the Multi-year ice around Lincoln to succumb this season.

[BornFromTheVoid]

A drop of 123.5k, larger than the long term average and the average of 07, 11 and 12.

[dree12]

A drop of 123.5k, larger than the long term average and the average of 07, 11 and 12.

And yet, every day, I expected more to have melted. Sometime...