NSIDC 2017 Arctic SIE September average: August poll

[Neven]

ATTENTION: There are two polls on the ASIF. This one is for NSIDC sea ice extent monthly minimum or September average (which is also used for the SIPN sea ice outlook), the other is for JAXA sea ice extent daily minimum (provided by ADS, previously by IJIS). 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 11 days (until August 12th). Until then you can change your vote. Here are the results for the June poll and July poll. There will be a new poll next month.

Here's how things are currently looking based on data up to July 31st:



These are the September averages for the last 12 years (in millions km2, found here):

    2005: 5.59
    2006: 5.95
    2007: 4.32
    2008: 4.74
    2009: 5.39
    2010: 4.93
    2011: 4.63
    2012: 3.63
    2013: 5.35
    2014: 5.29
    2015: 4.68
    2016: 4.72

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

[Richard Rathbone]

Up 1 bin, now thinking there's rather more risk of it being a little higher than last year rather than a bit lower, based on the current state of the Slater projection. http://cires1.colorado.edu/~aslater/SEAICE/

[gerontocrat]

I need all the help I can get. So I am waiting for the PIOMAS for July, especially Neven's analysis, plus to see if the Arctic weather does turn nasty over the next week or so. Only then will I put my incorrect guess in.

[Ned W]

Based on today's JAXA and the past decade's JAXA-daily-vs-NSIDC-monthly comparison, I'd expect 4.64 +- 0.74 (95% CI). 

A difference of only 0.12 would separate 3rd place (just below 2011) from 7th place (just above 2008).  It's quite crowded there in that 4.63 - 4.74 range.  The actual predicted value of 4.64 would mean squeezing into the very small gap between 2011 and 2015, for 4th place.

For now I'll go with 4.25-4.75, but I might revise that depending on how the next 10 days unfold.

[Ned W]

Here's the graph showing the relationship between current (Day 213) JAXA extent and the September monthly NSIDC extent:



Interpret it as you will.  The red dot shows the expected value based on the average difference between Day 213 and Sept. monthly.

I note that over one-third of the early responses to the poll are outside the 95% confidence interval for that estimate (all on the low side).  Just sayin'.

[Ned W]

OK, here are the a priori probabilities of each bin, based on the 2007-2016 historical record of JAXA Day 213 vs NSIDC Sept mean:

Over 5.75:   <1%
5.50 to 6.00:    1%
5.25 to 5.75:    5%
5.00 to 5.50:   15%
4.75 to 5.25:   33%
4.50 to 5.00:   48%
4.25 to 4.75:   47%
4.00 to 4.50:   31%
3.75 to 4.25:   14%
3.50 to 4.00:    4%
Under 3.75:   <1%

Total is more than 100% because of the overlapping bins. 

Disclaimer: If one believes that this year's "conditions" (thickness? weather?) are radically different from previous years, one might then expect that this model would not be accurate.

[Paddy]

4.25 to 4.75 for me.

[jdallen]

4.5-5.0

I think the ice will be more dispersed, which follows the poor quality we have seen.

Much still depends on weather.  I am keen to see what the next storms do, as badly placed, the could radically change the outcome.

[Jim Pettit]

This is a current list of the September NSIDC monthly SIE average were the remainder of the 2017 melting season were to exactly follow the behavior of any of the last 12 seasons (2005-2016):

2016: 4.5M km²
2015: 4.429M km²
2014: 4.845M km²
2013: 4.892M km²
2012: 3.858M km²
2011: 4.56M km²
2010: 4.755M km²
2009: 4.904M km²
2008: 4.02M km²
2007: 4.298M km²
2006: 5.158M km²
2005: 4.988M km²

That's a minimum of 3.858 (2012), a maximum of 5.158 (2006), and a 12-year average of 4.601.

Based on all of that, September 2017 SIE should come in between 4.35 and 4.65. That is, options  #3, #4, or #5. I went with #4 for now.

[Daniel B.]

4.25 to 4.75.  6th lowest overall.

[RealityCheck]

Extent holds up while volume declines significantly. But we're entering a more volatile phase of life for arctic sea ice IMO. Relatively smaller events (e.g. cyclones) have greater potential to affect the extent figures disproportionately relative to previous years' impacts. If we get one or two reasonably strong storms, extent could well trend lower than other years' data would suggest. But not a new low this year it seems...

[RealityCheck]

So 4.25 to 4.75 I think.

[magnamentis]

3.75 - 4.25, second lowest over all, no change to my earlier votes

[Bruce Steele]

Up one bin 4.00- 4.50

[Neven]

I've added the graph based on the latest data.

I'm going to stick with last month's vote, 4.25-4.75, but I might change before the poll expires.

Edit: I actually voted between 4.0-4.5 last month, as I did in June, so I've gone up a bin.

[icefisher]

Missing the major storm that my initial guess included.  Extent 4.665, but volume may reach an all time low with a little help from weaker cyclones on the Pacific side of the CAB.

[greatdying2]

3.5 - 4. Up one bin because the extent has been slower to respond to the ice thinness than it might have. I still think we have a good chance at 2nd, or possibly even a new record depending on how bad the weather gets.

[Archimid]

Disclaimer: If one believes that this year's "conditions" (thickness? weather?) are radically different from previous years, one might then expect that this model would not be accurate.

 Maybe not "radically different" but volume is about the same as the lowest value. I guess we are all waiting to see how significant that might be.

[meddoc]

3- 3,5 million km2.
But mostly Rubble.

[deconstruct]

3.75 to 4.25

I guess that 2017 will come in second.

The overall weather was just too cold and not favorable for ice melt than in other record years, the condition of the ice is still bad nonetheless. But it still would be a huge deal IMO, if 2017 would come in second (or even third), despite the ice-friendly weather over most of the melting season and the lack of precondition due to more snow on the ice earlier and only little ice-export through Fram strait.

That would just show, how thin the ice is and would make one wonder, how low extent could get, in a strong melting year.

[Juan C. García]

I'm voting 4-4.5 million km2.
I still believe that the low volume will make the 2017 daily value to be under 2007, but a freeze on the second half of September and the NSIDC algorithm could make that the NSIDC official monthly average could be above 2007.

I have gone up 2 bins.

[EgalSust]

4.5-5 - same as July. Might change it before the deadline though if there is remarkable melting action...

[RikW]

the numbers of the last couple of days make me doubt again if I shouldn't choose for a new record.

I still have the gut-feeling it will be worse then most of us, including myself, expect, but I can't put my finger on it...

[Sterks]

Voting 4.25 to 4.75 mill. Km2 despite the hiccup of today

[gerontocrat]

Also going for 4.25 to 4.75 million km2, based on a daily Jaxa extent minimum of 4.1 million km2 + 0.4 million km2 as the average difference with the NSIDC month average.

[RikW]

I guess the daily minimum will drop below 4.00, and it will hover most of september around that point, so the average will be slightly above 4.00.

For now i'm going for 3.75 - 4.25, but i'm doubting to change to 4.00 - 4.50. Luckily I still have 3 days to make the final decision

[Archimid]

To attempt to capture the fact that the atmosphere is cold but the ice is thin I voted low on the daily poll but I voted high on the monthly poll, 4.25-4.75. I think that after a near record daily low, the arctic will undergo rapid growth for at least a few weeks, maybe months.

[Juan C. García]

Less than 2 days, to make your vote or change it! 

[epiphyte]

IIRC I didn't vote last month - but then I would have been lower than where I am now, which is 3.5-4.0. I think that the minimum might well come in early, and bump along the bottom for a while due to ongoing melt in some regions long after that...

[seaice.de]

My sea ice outlook contribution: 4.2 +- 0.4

[Jim Hunt]

My sea ice outlook contribution: 4.2 +- 0.4

No new record then Lars?!

[seaice.de]

Well, not entirely ruled out. The lower bound of the error bar scratches 2012 which leaves some possibility. But just from the statistics I would say that a new record is not likely.

My sea ice outlook contribution: 4.2 +- 0.4

No new record then Lars?!

[Neven]

Less than 2 days, to make your vote or change it! 

I'm sticking with 4.25-4.75.

[Pavel]

I change my previous bets rising it by 0.75 mln. Now 3.75-4.25 lowest daily and 4.25-4.75 September average. Likely another mild winter or warm summer needed to break records

[magnamentis]

be patient, the main difference this year is thickness and logically it will show on the last mile, means during the last few weeks of the melting season. it's already hinting and if the weather conditions won't see to a very early minimum and a calm reminder of the season i still think we're getting close ( opted for second lowest from the beginning and stick to it )

BTW, talking about being close at the end of the season, IMO we are permanently close to the lowest at almost any time this year or below earlier. i opt for a sudden death over wide areas for the reason mentioned above.

[iceman]

I'm guessing September's graph will be more V-shaped than usual, and therefore the resulting rank for the month's average could be as low as 6th even if the daily min (other poll) is higher.  So going with 4.25 to 4.75.

[Peter Ellis]

I'm guessing September's graph will be more V-shaped than usual, and therefore the resulting rank for the month's average could be as low as 6th even if the daily min (other poll) is higher.  So going with 4.25 to 4.75.

I was all prepared to argue this, then I realised that your 'higher' actually meant lower amounts of ice, and vice versa.

[EgalSust]

Forum median for August prediction was 4.0-4.5, up two baskets from July and three from June.

SIPN/SIO median was 4.25-4.75, same as in July and June.

[Richard Rathbone]

SIPN just recycles the predictions from earlier months, unless its updated (e.g. Rob Dekker's contribution is specific to June, but SIPN counts him as having made the same prediction in July and August too.) This exaggerates how stable the SIPN predictions are from month to month, since quite a few of the August ones are really June or July predictions.

[Ned W]

SIPN just recycles the predictions from earlier months, unless its updated (e.g. Rob Dekker's contribution is specific to June, but SIPN counts him as having made the same prediction in July and August too.) This exaggerates how stable the SIPN predictions are from month to month, since quite a few of the August ones are really June or July predictions.

That seems like a bad idea, unless the person making the prediction explicitly chooses to retain the same prediction in later months.

[Dharma Rupa]

SIPN just recycles the predictions from earlier months, unless its updated (e.g. Rob Dekker's contribution is specific to June, but SIPN counts him as having made the same prediction in July and August too.) This exaggerates how stable the SIPN predictions are from month to month, since quite a few of the August ones are really June or July predictions.

That seems like a bad idea, unless the person making the prediction explicitly chooses to retain the same prediction in later months.

It does seem to me they are conflating two kinds of prediction then -- though I don't know if that is bad or good.  My gut says I'd like to know both the static early prediction and the running late prediction; which would mean that simply tossing the two together would not be to my tastes.

[Feeltheburn]

Too bad I missed this poll. However, I checked my vote from June, which was 4.5-5 million km2, with only 1 vote above me. Unless there is an cyclone, it should finish somewhere between 4.5-5 million km2 (although it upticked slightly above 5 million today).

[EgalSust]

Daily values until now:

2017   9   1         4.758
2017   9   2         4.791
2017   9   3         4.801
2017   9   4         4.782
2017   9   5         4.723
2017   9   6         4.643
2017   9   7         4.635
2017   9   8         4.697
2017   9   9         4.641
2017   9   10         4.628
2017   9   11         4.646

Any estimates for the monthly average? (How was it again, it's not going to be an average of the daily values, right?) Goes above 5M or not?

For comparison, here are the daily values for 2016, average ended up being 4.72M.

2016   9   1         4.278
2016   9   2         4.211
2016   9   3         4.263
2016   9   4         4.258
2016   9   5         4.201
2016   9   6         4.203
2016   9   7         4.083
2016   9   8         4.104
2016   9   9         4.137
2016   9   10         4.160
2016   9   11         4.209
2016   9   12         4.236
2016   9   13         4.341
2016   9   14         4.357
2016   9   15         4.378
2016   9   16         4.363
2016   9   17         4.309
2016   9   18         4.434
2016   9   19         4.495
2016   9   20         4.533
2016   9   21         4.593
2016   9   22         4.627
2016   9   23         4.814
2016   9   24         4.899
2016   9   25         4.916
2016   9   26         5.071
2016   9   27         5.143
2016   9   28         5.174
2016   9   29         5.175
2016   9   30         5.187

edit. made a crude calculation:

2017 first eleven days average: 4.70
2016 first eleven days average: 4.19

2016 monthly average: 4.72
2017 monthly average, if the ratio between monthly and first 11 days will be the same as in 2016: 5.23

[oren]

EgalSust, it is not a simple ratio. If I am not mistaken, NSIDC monthly "average" takes all pixels that had 15% ice through the month, so it will tend to focus on the last days of the month (as they have more ice). In addition, more ice now compared to 2016 could also mean less ice growth to the end of the month. So the numbers are simply not in yet.

[Sterks]

@Egalsust
Unless there is an unexpected fast refreezing, like last year but even faster, September extent can't be above 5 million km2, something is wrong in your assumptions (two things I'd say).

[Tealight]

For the monthly extent the NSIDC calculates the average sea ice concentration for every pixel and if the concentration is over 15% for the entire month the pixel is counted as ice covered. This is completely independent from daily values.

If high concentration areas move around by wind/currents then the monthly extent can also be higher than any daily value.

Using this method I calculated that the monthly extent until 11th September is 4.712 million km2.

Edit: Even if an asteroid would smash into the North Pole and wipe out the entire ice pack, the monthly September extent won't go below 4.029

[Feeltheburn]

For the monthly extent the NSIDC calculates the average sea ice concentration for every pixel and if the concentration is over 15% for the entire month the pixel is counted as ice covered. This is completely independent from daily values.

If high concentration areas move around by wind/currents then the monthly extent can also be higher than any daily value.

Using this method I calculated that the monthly extent until 11th September is 4.712 million km2.

Edit: Even if an asteroid would smash into the North Pole and wipe out the entire ice pack, the monthly September extent won't go below 4.029

I bet the average will be above 4.712 million km2 since it's bound to go up during the second half of the month.

[Daniel B.]

For the monthly extent the NSIDC calculates the average sea ice concentration for every pixel and if the concentration is over 15% for the entire month the pixel is counted as ice covered. This is completely independent from daily values.

If high concentration areas move around by wind/currents then the monthly extent can also be higher than any daily value.

Using this method I calculated that the monthly extent until 11th September is 4.712 million km2.

Edit: Even if an asteroid would smash into the North Pole and wipe out the entire ice pack, the monthly September extent won't go below 4.029

I bet the average will be above 4.712 million km2 since it's bound to go up during the second half of the month.

Agreed.  The last few data points indicate the ice is at or very near its minimum.

[Tealight]

Update for the provisional monthly extent until 22nd September.

Not much happened in the last few days. The monthly extent is at 4.722 million km2, a mere 10k km2 increase from the 11th. This means the icepack wasn't very mobile and any refreezing  happened in the low concentration areas which already counted towards extent.

Edit: 4.594 million km2 is guaranteed

[Richard Rathbone]

How much of that is already baked in (qualifies even if all ice vanished for the rest of the month)?