NSIDC 2015 Arctic SIE September minimum: August poll

[Neven]

ATTENTION: There are two polls on the ASIF. One is for NSIDC sea ice extent monthly/September average minimum, the other is for Cryosphere Today sea ice area daily minimum. Make sure you are aware of the difference before voting. You can discuss various extent/area data sets in this dedicated thread.

-----

This NSIDC extent poll will run for 12 days (until August 13th). Until then you can change your vote. This is this year's last poll.

Here's how things are currently looking based on data up to July 30th:



These are the September minimums for the last 8 years (in millions km2, found here):

    2005: 5.57
    2006: 5.92
    2007: 4.30
    2008: 4.73
    2009: 5.39
    2010: 4.93
    2011: 4.63
    2012: 3.63
    2013: 5.35
    2014: 5.28

You can use the comment thread below to motivate your choice, but discuss various SIE/SIA data sets in this dedicated thread.

[Neven]

I went down one bin compared to last my month: between 4.5 and 4.75 million km2. Perhaps a bit conservative and I might change my vote in 10 days or so if July behaviour is extended into August.

My votes for all the polls this year:
June: Between 4.75 and 5.0 million km2
July: Between 4.75 and 5.0 million km2
August: Between 4.5 and 4.75 million km2

This means that I think 2015 will end up 3rd to 5th.

---

PS I hope at least 100 people vote again! 

[Subjectivist]

ATTENTION: There are two polls on the ASIF. One is for NSIDC sea ice extent monthly/September average minimum, the other is for Cryosphere Today sea ice area daily minimum. Make sure you are aware of the difference before voting. You can discuss various extent/area data sets in this dedicated thread.

-----

This NSIDC extent poll will run for 12 days (until August 13th). Until then you can change your vote. This is this year's last poll.

Here's how things are currently looking based on data up to July 30th:

These are the September minimums for the last 8 years (in millions km2, found here):

    2005: 5.57
    2006: 5.92
    2007: 4.30
    2008: 4.73
    2009: 5.39
    2010: 4.93
    2011: 4.63
    2012: 3.63
    2013: 5.35
    2014: 5.28

You can use the comment thread below to motivate your choice, but discuss various SIE/SIA data sets in this dedicated thread.

Neven FYI the link you included took me to the 2014 dedicated thread instead of the 2015 dedicated thread.

[Neven]

Thanks, fixed now.

[greatdying2]

Same vote as July: 4.25-4.5 . Perhaps a bit on the low side, but certainly within reach. I also wouldn't be surprised if it turns out to be even lower.

[werther]

I'm not really interested in competitive guessing, but as I'm going over today's MODIS tiles and making a CAD comparison, why not submit my opinion?
As it occurred to me that there is continuing extent loss on the Chukchi and ESS front, which doesn't express itself in an exiting way in the IJIS-numbers, I have a strong hunch that this year's melt will continue right into September.

For my pleasure, I took the CAB limit in my CAD-drawing and deformed it according to what can be seen on the MODIS tiles. I disregard all other disconnected ice in the periphery, because it isn't important (though it will probably add up to some 200K in the end).
That's how I got to a daily minimum of 4,17 Mkm2.

My boundary does extend into the Beaufort Sea, because the larger MYI floes are not going to melt out. It isn't warm enough out there either.
The Atlantic side will probably hold some 100 km N of Svalbard and Frantsa Yosefa.
Part of the ESS-Arm will hold out.
The big unknown is the 'Laptev Bite'. Today, I see the weaknesses into the direction of the Pole shaping up further. One is within 200 km from the Pole, another is now developing into an 800 km2 polynia, about 600 km from the Pole.

Because NSIDC does a September average, I voted for 4,25-4,5 Mkm2.

[Tor Bejnar]

Neven,
The 'remove vote' option doesn't appear above.  (It does for the CT poll.)

[Neven]

Neven,
The 'remove vote' option doesn't appear above.  (It does for the CT poll.)

Thanks, Tor. Fixed now.

[Tor Bejnar]

My guess is up 1 bin (again), now 2 bins above 2012's minimum (e.g., 4-4.25).

[seaicesailor]

4.5 - 4.75. I might go lower, but the o-buoys #10 to #12 show sustained temperatures below zero and snow accumulation.

[jplotinus]

3.75-4.00

Baffin still has a bit to contribute and CAB looks to continue with large losses.

[jdallen]

Same bucket I've used for the last two - 4.25 to 4.5

I expect we'll pass 2011 but not quite make 2007.

[icefisher]

I moved down one bucket to 5-5.25 as July temp's have remained consistently warmer than I anticipated.  However, unless a big cyclone or major dipole develops, surface melt should slow down as air temperature cools.  Bottom melt is largely dependent on cyclonic mixing over the next week or two. 

[AySz88]

Down two bins (to 4.50-4.75) for now thanks to the super-July. Back to the same bin as my June guess. Seems like average losses would be right around there.

[ChrisReynolds]

As previously...

Sxxxx are the 'scenarios' which are the results from taking the losses from 31 July to the NSIDC monthly minimum, and subtracting those from the 31 July 2015 extent for the Arctic Ocean (which includes Greenland Sea and the CAA). Scenarios are for years 2007 to 2014.

These are then ordered in with the NSIDC Sept extents for 1979 to 2014 which are in order from lowest to highest.

2012   3.62   
S2012   3.82   0.41
S2008   4.01   0.54
2007   4.29   
S2007   4.36   0.44
S2011   4.57   0.45
2011   4.61   
2008   4.72   
S2010   4.74   0.54
S2014   4.85   0.55
2010   4.92   
S2013   5.01   0.51
S2009   5.03   0.51
2014   5.28   
2013   5.35   
2009   5.38   
2005   5.56   
2006   5.91   
2002   5.95   
2004   6.04   

The third column of numbers against each scenario is the compactness for the peripheral seas of the Arctic Ocean Basin (Beaufort round to Laptev). Compactness for that region on 31 July 2015 is 0.43. A scatter plot of periperal seas compactness for 31 July 1979 to 2014 and the losses from 31 July to the NSIDC monthly minimum 1979 to 2014 gives a linear fit with an R2 of 0.56, this is statistically significant.

This suggests that it is reasonable to use the 31 July compactness to narrow down the range from that for the 8 scenario calculation results.

Accordingly I select the scenarios with a similar compactness.

2012   3.62   
S2012   3.82   0.41
2007   4.29   
S2007   4.36   0.44
S2011   4.57   0.45
2011   4.61   
2008   4.72   

So I am voting 4.0 to 4.25M km^2 as the middle range of 3.75 to 4.5.

I think that there is a very good chance that we'll see a new second lowest set this year. However compactness is higher within the Central Arctic, so a new record low is not likely. But a lot depends on the weather.

All of this stems from the weather in July, until the Arctic Dipole set up arrived perhipheral extent and area showed little indication of above average falls, once the pattern established such falls have happened through July. My suspicion is that the damage done will now drive strong falls in August.

[cesium62]

In the various regions, 2015 will be about the same as 2012 except for the CAB and Beaufort and maybe the CAA.  The other places will either melt out or are the Greenland sea.  The weather in the CAA looks warm and windy.  The last thickness map I looked at made the CAB and Beaufort look kinda weak.  The CAB looks vulnerable to strong melting around the Laptev bite and north of the Atlantic where some warm temperatures should be occurring this week.  The CAB looks worse in 2015 than it did in 2012.  And Beaufort is looking kinda thin too, with wind and rain coming this week.

So, 2015 is on track to meet 2012 except for maybe in the Beaufort, so I'll be real conservative and guess 3.75 to 4.0.

[cesium62]

Same vote as July: 4.25-4.75 . Perhaps a bit on the low side, but certainly within reach. I also wouldn't be surprised if it turns out to be even lower.

Did you vote for two buckets, or is there a typo you want to fix here?

[Bruce Steele]

I will stay with my July 4.25-4.5 vote. I think there is plenty of surface water heat in the Beaufort Gyre to keep melt extended this year and also a long lived Microcat deployed with ITP 66 is showing surface warming to 6 meters at 80 degrees North.

[jdallen]

As previously...

<snippage>

So I am voting 4.0 to 4.25M km^2 as the middle range of 3.75 to 4.5.

I think that there is a very good chance that we'll see a new second lowest set this year. However compactness is higher within the Central Arctic, so a new record low is not likely. But a lot depends on the weather.

All of this stems from the weather in July, until the Arctic Dipole set up arrived perhipheral extent and area showed little indication of above average falls, once the pattern established such falls have happened through July. My suspicion is that the damage done will now drive strong falls in August.

What's this?!  I'm a bucket more conservative in my estimate than Chris?  This is a moment worthy of note

Actually, I think you've just done a better job of laying out your logic, which mine tends to follow albeit via slightly different methods.  My range is a bit narrower, but, I'm expecting the numbers to fall out at the lower end of my range (4.25-4.5), so in fact, we really aren't that far apart.

Gratifying to see my analysis standing up well compared with better expertise;  now we wait and see how skillful we were....

[cats]

I am sticking with my 4.5-4.75 bin from last poll - but I may change to a lower amount before the poll closes.  Still watching to see what the next couple of weeks brings for the Beaufort and whether things pick up on the Atlantic side.

[greatdying2]

Same vote as July: 4.25-4.75 . Perhaps a bit on the low side, but certainly within reach. I also wouldn't be surprised if it turns out to be even lower.

Did you vote for two buckets, or is there a typo you want to fix here?

Whoops, fixed thanks (4.25-4.5).

[ChrisReynolds]

JD Allen,

It might end up being too low...

Having just done my July blog post I suspect that MYI might impact Chukchi and Beaufort, as it seems to have done so far. Chukchi has stalled for 11 days now, Beaufort has been only slightly above average losses all summer, both are likely due to MYI. Scenario S2011 would suite me, it would mean my SIPN April PIOMAS prediction would succeed, and it would be 3rd place which would make the denialists look silly.

[kingbum]

I went 4.75-5.0 might go one lower but I'm expecting a little earlier refreeze on the Atlantic side in late August

[seaicesailor]

JD Allen,

It might end up being too low...

Having just done my July blog post I suspect that MYI might impact Chukchi and Beaufort, as it seems to have done so far. Chukchi has stalled for 11 days now, Beaufort has been only slightly above average losses all summer, both are likely due to MYI. Scenario S2011 would suite me, it would mean my SIPN April PIOMAS prediction would succeed, and it would be 3rd place which would make the denialists look silly.

Yeah, Chukchi stalled and buoys around the near part of the CAB and Beaufort show really cold conditions, which is good for MYI. Melting it requires sustained temps above zero right?. But by August 13 we should know what the stormy weather will be doing to the ice and in what sense, if anything.

[ChrisReynolds]

SeaIceSailor,

Well, if I loose on this prediction my SIPN prediction from April stands a chance of success, either way it's been a really exciting melt season. Predictions should be a test of ideas, and ideas should be changed according to the evidence otherwise they're beliefs. I don't have any time for the 'look at me I was right' types, that's just infantile.

It wouldn't be the first time MYI has caught me out, and I wouldn't be the first person to be caught out by MYI...

This is not an easy subject.

[seaicesailor]

SeaIceSailor,

 I don't have any time for the 'look at me I was right' types, that's just infantile.

Chris,

I'm sorry, I didn't imply anything like that.

I was just commenting that August, I heard, brings storms, with their coldness, snow, waves, dispersion, upwelling, and whatnot).
Ha, I say this with the same ease as if I was enumerating the damn symptoms of flu, that I know well without knowing much of medicine.

I won't be throwing arguments or references to you because ultimately I don't know much about it. I just enjoy the ride, and rely on people as Neven, Hunt, and you to learn one thing or two. I don't spend time to go myself to the sources, as you guys do, for whatever my reasons are. I use the www to safely throw opinions about things that I am never sure about (however lately I am becoming more cautious which is good).

Anyway, hope there is no misunderstanding cause I really respect your blog and comments (hell, it's one of the few sources I use for learning)

[anthropocene]

Staying with original June vote of 3.5 - 3.75 (i.e. roughly matching 2012). Most probably will be below final value but will stay with original guesstimate for consistency. long term weather forecasts tend to show cyclone activity in pacific side so may still get that low: If the final value is higher the main reason is less cyclone activity than expected.  Looking forward - 2016 is looking like it could challenge for a new minimum. Is it too early for a freezing season thread? ;-)

[seaicesailor]

 I closer to going one bin down:
Wip and slow wing show very dramatic effect of our WAC SAC or PAC. Many MYI floes, evaporating. Chukchi stall, unblocked.
One more day of this on schedule.

[Neven]

And there we have it, 100 votes. Great stuff, everyone, thanks.

And remember, you can still change your vote until August 13th.

[ChrisReynolds]

SeaIceSailor,

 I don't have any time for the 'look at me I was right' types, that's just infantile.

Chris,

I'm sorry, I didn't imply anything like that.

I didn't think you were, I was just clarifying my approach to prediction. Of course, in private when I'm right I feel good, but luck always plays a part in it. 

[Richard Rathbone]

4.5-4.75
Up a bit from 2007 on the basis of the shape of the volume decline being the best match I see, 2007 volume being a bit higher, but 2015 being more spread out.

[slow wing]

My previous guesstimates have been:
in June poll: 3.7 +/- 1.0 million km^2
in July poll: 4.2 +/- 0.9 million km^2
Each with 1-sigma uncertainties.

The upward change reflected the slow June melt. But that is behind us now, with a continuation of July's strong melt.

My guesstimate now is 3.6 +/- 0.6 million km^2.

So I am voting in the bin '3.50-3.75'.

The reason is that I expect the extent to collapse inside the peripheral seas, with virtually no ice South of 80 degrees North - only a small stub against the Canadian coast, a bit in the CAA and a bit on the East coast of Greenland. And I expect that to happen by the beginning of September and to persist for most of September, so the September average should reflect that.

The area inside the 80 North circle is only 3.88 million km^2, and that includes some land as well.


So I expect it to be challenging 2012 for the lowest September extent - ending up either first or second lowest on record.


Here is the current concentration map (Aug 03 2015) that largely leads me to that conclusion...

(EDIT: corrected link to correct map; HT to greatdying2.)

[greatdying2]

Here is the current concentration map (Aug 03 2015) that largely leads me to that conclusion...

(Archived map first - but it doesn't seem to display the whole area - so also the daily map which will update.)

Here's the archived map with the full image:

[smeagol]

New to this, gave the issue long hard consideration and concluded I had no idea whats going to happen in the next four weeks so just went for a guess of 4.25-4.5. & 2.25-2.5 for area

[Tor Bejnar]

Welcome smeagol. 
You are either being truthful or modest.

[OSweetMrMath]

Here it is, August 4th, and I already feel terribly behind on participating in this poll. My prediction last month was that July would come in at 8.1 million sq km, August would be 5.7, and September would be 4.8.

I hedged a little, saying that since the melt at the end of June had been slow, if it continued to be slow in July, the actual extent would be higher than predicted. Well, it was slow, at the beginning of the month, anyway. The July extent is 8.77 million sq km, for a prediction error of 0.67 million sq km, the largest since I started predicting extent last April. Going back further, this is larger than any of my model residuals since August 2012.

The bad news for my model is that extent loss has been high in later July, and my model not only completely fails to recognize that but treats the low loss in the first part of July as a strong signal. Consequently, my model now predicts that the August average will be 6.1 and the September average will be 5.1 million sq km.

I straight up do not believe my model. The pattern of melt has been sufficiently different this year, with low melt in June and early July, but high melt in late July, that my model doesn't understand the data. (The model is also limited by the fact that it only considers monthly averages.) As a result, I think the previous prediction, that September will be 4.8 million sq km, is more credible than the current prediction.

Nonetheless, I am voting as my model instructs, in the 5.0-5.25 bucket.

[OSweetMrMath]

Other people have posted distributions for their predicted extent (see ChrisReynolds, above, or OldLeatherneck, predicting the daily minimum measured by ADS-NIPR (formerly IJIS/JAXA) at http://forum.arctic-sea-ice.net/index.php/topic,1112.msg58997.html#msg58997). So here is mine, based on my extent model for NSIDC September monthly extent.

Note that even though I've been making predictions from this model for over a year, I have not validated it. This has not seemed like a big deal for point estimates of the extent, but it feels more significant for interpreting the predicted distribution. In other words, take this prediction with a grain of salt. The prediction probabilities are only meaningful to the extent that the model is correct.

As an additional caveat, my model does not understand the extent loss in July. The fact that melt was slow at the start of the month and high at the end means that the July average was (relatively) high, which means that this model predicts that the September extent will also be (relatively) high. This is probably precisely backward, as the high melt at the end of July should be predicted to continue into August, resulting in a relatively low September extent. The listed probabilities are probably somewhat lower than the "true" probabilities. As an approximate correction for this, subtract 0.25 million sq km from the extent while leaving the probabilities unchanged.

The prob column is the probability that this year's minimum will be below the extent (measured in millions of sq km) listed in the ext column. I have included both some standard probabilities and comparisons to September extent in previous years.

prob	ext
0.0006%	3.32	(minimum possible under this model)
<0.01%	3.62	(2012 minimum)
1%	4.24
1.3%	4.29	(2007 minimum)
5%	4.51
8.6%	4.61	(2011 minimum)
10%	4.64
14.3%	4.72	(2008 minimum)
20%	4.79
33.4%	4.92	(2010 minimum)
50%	5.05	(median and mean prediction)
76.7%	5.28	(2014 minimum)
80%	5.31
83.3%	5.35	(2013 minimum)
85.7%	5.38	(2009 minimum)
90%	5.44
94.8%	5.56	(2005 minimum)
95%	5.57
99%	5.83
99.5%	5.91	(2006 minimum)
100%	6.60	(maximum possible under this model)

[Paul Beckwith]

This video may help in your guess on remaining Arctic sea ice this year...

Will #Arctic #Sea #Ice Vanish this Summer 2015? https://t.co/RXE69M5p0Z

My guess: zero 

[Gonzo]

I think volume will be lowest on record

[A-Team]

I'm not thrilled with the arbitrary and capricious focus on the Julian calendar in the NSIDC definition. It's apparent from the graphical blowup of September (days 244-273 in 2015) that it begins too early and runs too late into freezeup. The box in the graphic picks out September, the faint colored dotted lines the SIE monthly averages.

They could have gotten rid of a lot of silly variance just averaging ten days centered on 255. Or just take the minimum from a rolling ten day average. Either might slightly change ordering of years.

However as a lad taking 'scholastic aptitude' tests in school, I soon learned that the right answer was not the correct answer but rather whatever some C student majoring in psychology who later writes multiple choice exams for the K-12 education market thought was the correct answer. This selects for people good at gaming the system rather than for scholarly propensities. Just saying I don't want to be predicting how fast Sept freezeup will be when the question was Sept minimum.

At any rate, watching 2015 wander back and forth between 2007 and 2012, this year seemed destined to end between them and rather closer to 2007, which worked out to 4.52 so I went with that bin though this is a wash with the upper part of the next lower bin.

September was originally the seventh of ten months on the oldest known Roman calendar, with March (Latin Martius) the first month of the year until perhaps as late as 153 BC.  After the calendar reform that added January and February to the beginning of the year, September became the ninth month, but retained its name. It had 29 days until the Julian reform, which added a day. September is mostly in the sixth month of the astrological calendar (and the first part of the seventh), which begins at the end of March/Mars/Aries.

[DavidR]

4 - 4.25:  I  am expecting  a larger than average melt over the remainder of the season because of high July temperatures and low compaction of the ice, particularly on the Pacific side of the CAB.
So I am estimating 1 sd below the 'average post 2006' melt  for both polls.  My SIPN prediction will be 4.1 +- 0.2.

[Neven]

134 votes already. That's great, thanks, everyone.

[crandles]

This video may help in your guess on remaining Arctic sea ice this year...

Will #Arctic #Sea #Ice Vanish this Summer 2015? https://t.co/RXE69M5p0Z

My guess: zero

The images in the video seems to suggest to me that the purple and blue ice under 1.5m thick can melt out in Aug and early Sept; so does some thicker ice in Beaufort but it is a fairly limited area.

That is a long way from suggesting that the 'ice [will] vanish this summer 2015'.

Don't you think it is damaging to suggest something dramatic like the ice is going to vanish 'this summer'. When this (predictably) doesn't happen, don't you think this creates a history of environmentalists crying wolf?

[Peter Ellis]

To amplify crandles' point - extent is currently 6 million, give or take a few.  There are about 40 days left in the melt season. Getting to zero would require an average melt of 150,000 per day, sustained right up until the very last day of melt.  It's not just unbelievable but grotesquely so and serves only to make everyone concerned look stupid.

[kynde]

I also think we're on the safe side of not totally collapsing this year, but to be fair for those making the lower guesses, to me it also seems clear that the year we're going to hit zero or close enough it will be an outlier, likely a favourable early season followed by a strong weather phenomenon like the GAC12 or worse, possibly something we haven't seen yet. Wasn't the collapse that the GAC12 induced, and frankly the GAC12 itself, quite unprecedented at the time, too?

In that sense betting low continuously should pay off one year, yet equally it will never be the most likely outcome making it somewhat of a wrong choice every year.

Forgive me for reiterating some of the points obvious to most and for outright being wrong on the rest

[iceman]

I'm not really interested in competitive guessing, but as I'm going over today's MODIS tiles and making a CAD comparison, why not submit my opinion?
  ....

To me it's more in the spirit of "wisdom of the crowd."  Pretty sizable crowd this round.

I'll stick with last month's guess of between 4.5 and 4.75 million km2.  Though a slowdown appears to be coming in the next week or two, I still expect a late decline in volume anomaly to pull extent below its average August/Sept trajectory.

[Subjectivist]

I keep reminding myself that 2011 was more or less a return to the 2007 low and that preconditioned the ice for the new record low in 2012.  2015 is much like 2011 IMO, right now the trace for ice Area on CT is closely paralleling 2007/2011.  Even if nothing else goes wrong we are looking at a low remaining ice year, and couple that with the El Nino going on through this winter and we might have a new massive loss record year 2016.

[ChrisReynolds]

I have used a rough method to outline prospects for the NSIDC September Extent recently. I last used it on 2 August using 31 July extent. This method just answers the question - what if the extent losses from now to the September Extent happened from current extent, where would it take us?

Sxxxx are the 'scenarios' which are the results from taking the losses from 5 August to the NSIDC monthly minimum, and subtracting those from the 5 August 2015 extent for the Arctic Ocean (which includes Greenland Sea and the CAA). Scenarios are for years 2007 to 2014.

These are then ordered in with the NSIDC Sept extents for 1979 to 2014 which are in order from lowest to highest.

2012   3.62
2007   4.29
S2012   4.34
S2008   4.52
2011   4.61
S2007   4.71
2008   4.72
S2011   4.80
2010   4.92
S2010   4.96
S2014   5.03
S2009   5.06
S2013   5.19
2014   5.28
2013   5.35
2009   5.38

The order has changed significantly since I last did this, as is to be expected, it's a rough method. This gives a range of 4.3 to 5.2M km^2, or 2007 to just under 2014.

Taking the average loss from 5 Aug to the Sept 2015 and calculating the difference from that average and standard deviation of those losses, 2012 was over 2 standard deviations from the average. For 2015 to reach 2012 levels would require a loss that is 4.6 standard deviations from the average!!!

Anyway. given ice state as of the curent data, weather, and given the above calculations, I am going up one bin to 4.25 to 4.5, which also covers the bins above and below in my expectation. Over the last week I just haven't seen the sort of aggressive loss I had expected, but I still expect stronger than average losses this month.

[OSweetMrMath]

Chris,

It's striking that half of your scenarios end up in 0.36 million sq km gap between 2010 and 2014. I can't help feeling that's in agreement with my model, which puts a probability of 43% of being in that same range.

A difference is that in your case, that's the upper 50% of outcomes, while in my case, it's the middle of the outcomes. But I also think my model is likely to be too high, and the center of the "true" prediction of my model is very close to the center of your scenarios.

It's also striking that after you raised your vote by one bin, you are still two bins away from the median of your scenarios.

[Paul Beckwith]

In my video, notice that I do not make a guess on sea ice minimum. I let the observer decide what they want. I thought the ice would all go a few years ago (2013), so I am reluctant to come up with any logical number.

However, I think that when it goes it will be extremely rapid, and totally amaze everybody, including people that actively follow it; none more intensely than those on this forum.

To have a bit of fun; I picked zero for this year (again!). A few long duration cyclones could rip everything to shreds, and very warm water could extend the melt season a week. Who knows; it is a total guess. We are trying to come up with a number on a highly nonlinear process. The sea ice will be there, until it is suddenly not:)