Polls August 2012

ATTENTION: new polls in the right hand bar, closing August 20th.

I guess the latest SEARCH Sea Ice Outlook hasn't come out yet because participating forecasters want to see the full effect of this storm on the sea ice extent numbers before submitting their final estimate for the September average sea ice extent as calculated by the National Snow and Ice Data Center. But we need some new polls over here, so I'm putting out this post.

As you all know, we're doing two polls, because only betting on the September average is kind of boring, as we'd have to wait until October to see the final result. It's more fun to track the daily numbers, and so there's a poll for Cryosphere Today sea ice area as well.

I chose this dataset because IJIS switched from the AMSR-E to the WindSat sensor, and their SIE numbers are oscillating a bit too much to trust comparisons to other years. There are not any other daily updated, freely accessible data sets out there (NSIDC has one now, but I had already started the polls by then) and as Cryosphere Today is an old and trustworthy player in the whole sea ice cover calculation business, I've decided to go for a daily minimum sea ice area poll.

Here are the results from last month's poll:

The NSIDC minimum monthly/September sea ice extent received 238 votes. The average* of all votes was 4.24 million square km.

These are the NSIDC September monthly average SIE minimums (data here) in the 2005-2011 period:

• 2005: 5.57 million square km
• 2006: 5.92 million square km
• 2007: 4.30 million square km
• 2008: 4.73 million square km
• 2009: 5.39 million square km
• 2010: 4.93 million square km
• 2011: 4.61 million square km

The average* for the Cryosphere Today minimum daily sea ice area (279 votes) was 2.93 million square km.

These are the Cryosphere Today daily SIA minimums (data here) in the 2005-2011 period:

• 2005: 4.09 million square km
• 2006: 4.03 million square km
• 2007: 2.92 million square km
• 2008: 3.00 million square km
• 2009: 3.42 million square km
• 2010: 3.07 million square km
• 2011: 2.90 million square km

There are new polls in the right hand bar, that will close on August 20th. Because of recent developments I've changed the options. If you want, you can re-vote later on by going to the polls directly (here for NSIDC minimum monthly SIE, and here for CT minimum daily SIA). For an explanation of the differences between the polls, go to this blog post.

* This is my complex method for processing the poll results and calculating the average:

I've calculated the number by multiplying the percentage number with the average of an option. For instance 'between 4.0 and 4.2 million' = 4.1, and for 'less than 2.8 million km2' I've used 2.7 million km2. I've added up all multiplications and divided the result by 100.

Comments

[Al Rodger]

*An alternative calculation method for the average would be to use the 'median' (rather than the 'mean') which is the vote with half the ballot higher & half lower.
(Within each bucket of votes, the votes can be considered spread evenly between the limits.)
Using the median, arguably all votes do still effect the result as even an ultra-high ice vote will shift the median up, just not as far as in the 'mean' calculation (which arguably is perhaps too far).
The result using 'median' yields 4.18 extent (compared with the mean's 4.24) and 2.86 area (mean's 2.93).

[Chris Reynolds]

I still haven't much of a clue about the extent minimum. Getting the feeling I should download extent and start using it again, but for the moment I'm sticking to area (CT).

Now it's just deciding how far down it'll go.

I've 2 methods of making projections:
1) subtract the annual loss from a selected date from the figure at that date. Repeat for 10 day periods from day 150 to 220 (latest). Take the average of all the 10 day predictions over that period

2) Use the daily changes (day's area minus previous days area) from each post 2007 year (2007 to 2011) to create a series of 'scenarios'. Using the starting point of the most recent CT area data (day 221) and add the changes to the area on that day for each day onward - as if the year's (2007 to 2011) melt profile were followed from day 221 onward.

Method 1 gives 2.6M kmsq, and for previous years seems to gives good agreement with actual results with a stdev of error at under 4%. Roughly 2.5 to 2.7 M kmsq.

Method 2 only takes the most recent figure into account so is much lower. Figures in the range 2.2 to 2.5 M kmsq.

I suspect we'll see a substantial reduction in melt rate during and after the last week of August. This happened in 2007 and 2011. Applying this assumption to method 2 gives figures mainly in the range of method 1.

So I'm saying that the minimum for 2012 will be between 2.5 and 2.7 M kmsq, CT daily area.

This leaves me with a problem in terms of the pole - the demarcation between the last two categories splits my range. I've just tossed a coin - so find myself voting for the lowest category.

[Seke Rob]

For reference, JAXA recorded the fastest 1 million drop on record going from 7 to 6 million km square, 12 days [with little prior years competition]. Just 6 days into the 6 to 5 million segment, the 5.3M Km^2 prelim report for the 11th, *leads* any prior year in daily average decline, but whom am I telling... http://bit.ly/IJISMD

[Andre Koelewijn]

I'ld say the way you calculate the average is pretty conservative, going down only half of the scale of the previous steps at the minimum.

Early May, I made a layman's guess of 2.4M for CT SIA minimum and 3.4M for SIE. Pretty bold figures, which I based on the general decline over the past decades and the notion that the continuing decline in volume according to the PIOMAS model should by now come in on the area and extent numbers: just too much thinned ice around in parts of the Arctic which used to have THICK multi-year ice.

So where you, Neven, counted '2.7M' for my vote on 'less than 2.8M', it wasn't what I meant (obviously, you couldn't see). And that may count for more votes in that category.

Looking at the present figures and forecast, that SIA number of 2.4M still seems achievable. A SIE of 3.4M on average for the whole of September is less likely, but who knows...

[Tor Bejnar]

I will very much miss sea ice area polls next year if "area" is left out as Neven intends. A common definition of virtually ice-free Arctic is "sea ice area < 1M km2".

[Neven]

I know what you mean, André, it's far from perfect, but it's the best I can do, and either way, we shouldn't take predictions by polling too seriously.

Like I said last year if you take the WUWT and ASI Blog poll results and average them, you will probably get close to the actual number. Especially now that over at WUWT they are voting really low, for fake skeptics that is.

[Stevemosher.wordpress.com]

I'm gunna stick with my june numbers.
2.9 on area, 4.29 on extent.

[Twemoran]

Neven & Larry H.

Very nice addition to the graphics page!

Terry

[Neven]

Thanks, Terry. We did the same thing last year at the end of the melting season. Those bar graphs that Larry makes are great visually.

[crandles]

>"I will very much miss sea ice area polls next year if "area" is left out as Neven intends. A common definition of virtually ice-free Arctic is "sea ice area < 1M km2"."

+1

I will miss it too. I also like to point out that area is a better guide to predicting both area and extent minimums and is therefore IMO worth the attention.

[Neven]

Thanks for sharing your feelings about this, gentlemen. :-)

We'll see about his next year. It's just that the large differences between the polls (daily vs monthly, extent vs area) can be confusing for newcomers, I think.

[Greg Wellman]

I finally pulled the trigger on the August poll. 4.25 and 2.85, so I'm betting on two new records, but by small amounts. If a compaction event occurs I'll be quite a bit off though.

[Greg Wellman]

I changed my CT vote down one bin, so I'm going for 2.75 now. Probably conservative as CT currently has 2.92(!)

[Otto Lehikoinen]

Since here we are mostly discussing arctic processes and their details I'll note here somewhat off topic that El Nino -phase of ENSO has begun (though most probably know it already, http://www.esrl.noaa.gov/psd/enso/mei/table.html
This has a lagged influence on atmospheric temperatures (was it by 4 months?), so in October-November-December there should be a rise in global atmospheric temperatures. What does this mean for Arctic isn't too clear as the Arctic is in the dark period and warm air rises. Should the Nino conditions continue until December that is an alltogether different matter.

[crandles]

Last day for voting. CT area is already below the top four options which have gathered 11% of the votes.

For monthly average extent, 3 options are above current level attracting 2.4% of votes. Extent does increase late September so these 2.4% votes are not mathematically out of it but they are pretty unlikely.

I am getting predictions in the region of 2.4 for CT area daily and about 3.8 for September average extent.

[crandles]

CT Area 2.594492

Only one poll category left. 35.3% of voters were right - more votes than any other category.