Record dominoes 8: NSIDC daily sea ice extent

There are several scientific organisations that keep an eye on the Arctic sea ice cover and put out graphs to inform us of the amount of ice that is left. You can see most, if not all, of them on the ASI Graphs webpage. I expect the record on most of these graphs to be broken in weeks to come.

---

The daily sea ice extent graph of the National Snow & Ice Data Center in Boulder, Colorado, one of the foremost and best-known organizations observing the Arctic, is showing a new record. Actually, the graph went crazy and is showing new record year 2012 only (will be fixed soon, I'm sure; edit: smart crandles says "it performs exactly what was asked i.e. display current year and year with the record low."):

Edit 2: the graph has been fixed and updated, so I have made an animation that shows how the trend line dropped from the Arctic storm onwards:
But we have the numbers this time, because NSIDC decided - kudos to them - to release their daily numbers (hat-tip to Larry Hamilton):

-----------------------
Year | min(extent)
----------+------------
1979 | 6.89236
1980 | 7.52476
1981 | 6.88784
1982 | 7.15423
1983 | 7.19145
1984 | 6.39916
1985 | 6.4799
1986 | 7.12351
1987 | 6.89159
1988 | 7.04905
1989 | 6.88931
1990 | 6.0191
1991 | 6.26027
1992 | 7.16324
1993 | 6.15699
1994 | 6.92645
1995 | 5.98945
1996 | 7.15283
1997 | 6.61353
1998 | 6.29922
1999 | 5.68009
2000 | 5.9442
2001 | 6.56774
2002 | 5.62456
2003 | 5.97198
2004 | 5.77608
2005 | 5.31832
2006 | 5.74877
2007 | 4.1607
2008 | 4.55469
2009 | 5.05488
2010 | 4.59918
2011 | 4.30207
2012 | 4.0892 (and running)
-----------------------

Another big domino has fallen. All the records on daily extent and area graphs have been broken now, except for IMS, right in time before my holiday (starting tomorrow). Thank you, Al Gore, for destroying the ice so I can go!

We now await how low all the daily graphs will go, we await the PIOMAS update in the first week of September, and then the final mega-domino: NSIDC September average minimum (which possibly could remain standing if the daily numbers stall and the melting season ends really early). As an interlude I have a piece coming up about consequences, which will be posted tomorrow. New ASI update will be up later today.

Enjoy while it lasts!

Comments

[Seke Rob]

Sept. 30 is in and the numbers tally to 3.58m single day average for September and 3.57m using the 5 day trailing average.

2012 9 26 3755110
2012 9 27 3809650
2012 9 28 3884540
2012 9 29 3894390
2012 9 30 3929450

Don't know what September CAPIE will work out to be. The worst or close to that [no rocket science models needed to guess that]

[crandles]

2007 9 Goddard N 4.30 2.78
2008 9 Goddard N 4.73 2.99
2009 9 Goddard N 5.39 3.47
2010 9 Goddard N 4.93 3.07
2011 9 NRTSI-G N 4.61 2.89
2012 9 NRTSI-G N 3.61 2.11

[Seke Rob]

That is the fastest I [moi] has seen for NSIDC monthlies. Including the 0.31m Area adjustment an NSIDC break up index of .6704 or in reverse 1.49, a new September record.... The last 6 years:

71,86%
69,77%
70,13%
68,56%
69,41%
67,04%

http://i137.photobucket.com/albums/q210/Sekerob/Climate/ArcticSIA-SIERatio2.png

In a picture... what else.

The difference to the 0.03M we derive from the dailies is to small to ponder on.

[crandles]

Didn't they change to trailing averages? That presumably helps them get the answer out faster than in previous years.

It would seem that IJIS averages the percentage coverage over 2 days for each cell before applying 15% coverage test. But the NSIDC data is out so rapidly after 30 September single day value seems to suggest NSIDC do not use 1st of October information at all. They could still be doing muliple day averaging of percent coverage for each cell perhaps after eliminating outliers but it is presumably all done on analysis of days prior not after the day in question.

[Seke Rob]

Tried 4 different ways including centered and not getting their number, but as said too small to get the shorts in a knot.

[Artful Dodger]

NSIDC has likely gone back to their previous practice of estimating the next 2 days, while retaining the 5-day centered average for consistency with previous years.

IMHO they should have just waited for Oct 2nd data to become available, and then announce the Sept monthly mean.

But as S/R says, not worth kvetching about. It's the min daily SIE that's the real star & the headline grabber.

Kinda shocking to see the 2.11 M km² Sept sea ice area, wot?

Cheers,
Lodger

[Neven]

I guess that makes it time for yet another record domino: 11

Thanks for posting all the info. I'm a bit slow lately, extremely busy and in a bit of a blogger existential crisis. More on that later.

[Espen]

Neven,

Crisis? What Crisis?

http://www.youtube.com/watch?v=iIyVDHlJgSE&feature=fvwrel

[Artful Dodger]

Neven, you'd be less than human if all this bad news didn't affect you. Remember these days of innocence when you look back in 2030. If only we had known...

[Wipneus]

Okt 2:

2012 9 28 3,88454
2012 9 29 3,89439
2012 9 30 3,92945
2012 10 1 4,04797
2012 10 2 4,11772

September average with true (midpoint) 5 day averages can now be calculated: 3.585 Mm2

How to calculate from daily data to get 3.61 stays unknown.

[Seke Rob]

Wipneus,

As was eluded by Lodger, up in this or other NSIDC thread somewhere, they use a guesstimator at end for Oct1/2 and then compute a centered 5 day average... to publish their number on the 1st. The diff is too small to worry about... but M&M would and issue an FOIA, to "demand" clarification and to code and the whole database and every email ever exchanged on the matter ;O)

[Seke Rob]

Well, already checked the centered before, but with actuals Oct1/2 in, it's still 3.585... same as single day average. Wonder if I can submit an FOAI from Italy to a USA Fed. institution (Just kiddin') :P

[Wipneus]

They would have to check their "guesstimator", values over 5.2 for 1&2 Okt would be required for that.
Maybe it is not an important difference, and I am certainly not going to investigate any further, but it is far out of range with the spread we get with single day values, 5 day averages trailing or centered: 3.579 3.573 3.585.
I guess another sort of difference must be causing it.

[Peter Ellis]

The documentation is here:

http://nsidc.org/data/docs/noaa/g02135_seaice_index/#monthly_data_files

The text says that the monthly extent and area data are derived from monthly gridded data. That is, they average each individual grid cell for each month, and then compute extent/area.

This is not quite consistent with their figure 2 at the top of the page, which has the monthly data being calculated directly from the daily data. I think that instead of coming from the yellow "Daily Gridded Concentration" box, the line should come from the "Monthly Gridded Concentration" box two places below it.

[Peter Ellis]

To see why this matters, consider a toy system with only two grid cells, with values as follows:

Day 1: 25%, 25% (daily extent = 2)
Day 2: 25%, 0% (daily extent = 1)
Day 3: 0%, 25% (daily extent = 1)
Day 4: 0%, 0% (daily extent = 0)

If you simply average the daily extent figures, then you would say the average extent is 1.

However, if you make a gridded average, then both cells have an average of 12.5% across the four days, which is below the extent threshold, and thus you would say the average extent is zero.

[Peter Ellis]

In practice it'll only make a very slight difference in the low decimal places - differences likely to be bigger in months with a lot of mobile, low concentration ice.

[Derek]

Has anyone been watching how long the Northwest passage and Northern sea routes have been open. Another two dominoes perhaps?

[Artful Dodger]

NSIDC Sep Monthly SIE for 2011 is 4.61

Sep daily average (either daily or 5-day centered) is 4.56

So monthly data is not computed with a simple average of daily data.

Maybe SEARCH has some more info?

[Peter Ellis]

Lodger! LOOK UP! I just posted the explanation for this, with links to NSIDC's own documentation.

Christ on a crutch, is this turning into a post-only forum?

<harrumphs like a grumpy schoolmaster>

[Wipneus]

Peter, that looks very reasonable.Whether the 15% rule is applied before or after the monthly averaging is likely to make that difference.

[Artful Dodger]

Hi Peter,

I did read your comment, interesting speculation, but far from confirmation.

My comment is more directed to folks like Chris Randles and Larry Hamilton who are subscribed to the SEARCH email list. They may have more information, since SEARCH specifies Sep Avg SIE.

We should be looking for independent confirmation, not a cheering section per se. Not that your explanation is incorrect. I hope it is.

Also hope you didn't play yourself out with Paul and friends, we need your ideas here ;^)

Cheers,
Lodger