This week was a true roller coaster ride with Arctic Sea Ice. It is best summed up by looking at the JAXA graph for extent, shown below:
Below, see the area of interest magnified.
I’ve added the 5 million square kilometer line for reference.
The roller coaster ride actually looked for a day like it might cross the 2009 line, but soon turned down again, ending this week at 5,142,813. Here’s the recent JAXA data
08,28,2010,5342656 08,29,2010,5352500 08,30,2010,5348281 08,31,2010,5329375 09,01,2010,5332344 09,02,2010,5304219 09,03,2010,5245625 09,04,2010,5192188 09,05,2010,5142813
Source: http://www.ijis.iarc.uaf.edu/seaice/extent/plot.csv
JAXA sea ice area has dropped to 2008 levels:
JAXA AMSR-E Sea Ice Area – click to enlarge
Sea ice concentration from JAXA:
While JAXA shows extent now lower than 2009, DMI and NANSEN plots show it to be about even. The differences in observing sensor/platform AMSRE -vs-SSMI and methodologies at agencies are in play.
Above: Danish Meteorological Institute Arctic Sea Ice Extent – 30% or greater. Note that while this graph shows 30% concentration at the cutoff point, it is valuable to compare.
Above: NANSEN Artic ROOS- Sea ice extent 15% or greater – click for larger image
The differences appear to be in the low end of concentration, the 15% to 30% range. It suggests that the brief gains we saw may be wind related, blowing floating ice around, compacting it when winds are strong versus allowing expansion when winds are weak.
Temperature, after holding near freezing, now appears headed sharply downward.
Above: Danish Meteorological Institute – Mean Temperature above 80°N
Some light refreezing may take place before the end of September, which could minimize the ability of wind to sharply change extent like we saw recently.
With all these variables in play, choosing a winner will be as much a game of luck as of skill. Based on what we’ve seen, it seems probable that it will come from the middle of the pack between 2008 and 2009.
From SEARCH:
The estimates from the scientific community range from 4.0 to 5.6 million square kilometers, with 8 of the contributors suggesting a September minimum below 5.0 million square kilometers, 3 contributors suggesting a minimum of 5.0 million square kilometers, and 5 contributors suggesting a September minimum above 5.0 million square kilometers. Two contributors forecast a September minimum below that of 2007 at 4.0 million square kilometers and 3 contributors suggest a return to the long term downward linear trend for September sea ice loss (5.5 to 5.6 million square kilometers). None of the contributors indicate a return to the climatological sea ice extent of 6.7 million square kilometers.
Including all 18 contributions gives a September ice extent minimum of 4.8 +/- 0.77 million square kilometers, with a range of 2.5 to 5.6 million square kilometers.
Individual responses were based on a range of methods: statistical, numerical models, comparison with previous observations and rates of ice loss, or composites of several approaches.
Günther Kirschbaum says:
September 8, 2010 at 8:34 pm
IJIS extent has just gone below 5 million km2 and currently stands at 4,977,344 km2. I wonder when extent will stop dropping like a rock. Should be any day now.
_____________________________________________________________
Will you let me take a SWAG on this one?
I think it has a lot to do with the freezing point of salt water (~28F) versus freshwater (~32F) and the SST in areas of low concentration ice versus areas of high concentration ice.
It also has a lot to do with the distribution of sea ice throughout the Arctic. Particularly the lower concentration areas (e. g. 15% concentration can go down as 30% concentration goes up as the areas < 15% are not moving into the 15%-30% concentration areas as fast as the 15%-30% concentrations are moving into the 30%+ concentration areas).
It also has a lot to do with boundary conditions and associated forcings (primarily wind). Think of these constraints in the context of a constraint based random walk model in 2D. If done right, than the low concentration ice will have net movements to higher concentration ice areas.
Areas of higher concentration ice are assumed to be no longer melting and might also have lower SST's and should freeze first as these SST's approach the freezing point of saltwater (~28C).
Areas of lower concentrtion ice might have higher SST's and given the amount of open water and wind conditions, the upper layer of the Arctic Ocean is well mixed and will not freeze this saltwater until the SST's in these areas hit 28F.
You also have to consider the freezing temperature density hump that exists for saltwater (for freshwater the density hump is at ~4F). In other words there will be movement (mixing or turnover) due to saltwater stratification at this time. The coldest water is not the highest density water so it will move down the water column at this time of year (assuming equal salinity with depth in the lower concentration areas).
So, in closing, the current conditions are ripe for consolidation, but are not ripe for freezing throughout the Arctic Ocean at this time.
Correction: ~4F should be ~7F (~4C). Ignore 2nd to last paragraph as freshwater must be present at the top of the water column for this argument to have any weight at all (bass ackwards as written)
Correction Part Deux: OK, I found it, I knew I saw this somewhere before.
http://en.wikipedia.org/wiki/Arctic_ocean#Climate
” The density of sea water, in contrast to fresh water, increases as it nears the freezing point and thus it tends to sink. It is generally necessary that the upper 100–150 meters of ocean water cools to the freezing point for sea ice to form.[12]”
Reference [12];
http://nsidc.org/seaice/intro.html
“The process by which sea ice forms is also different from that of lake or river ice. Fresh water is unlike most substances because it becomes less dense as it nears the freezing point. This difference in density explains why ice cubes float in a glass of water. Very cold, low-density fresh water stays at the surface of lakes and rivers, forming an ice layer on the top.”
“In contrast to fresh water, the salt in ocean water causes the density of the water to increase as it nears the freezing point, and very cold ocean water tends to sink. As a result, sea ice forms slowly, compared to freshwater ice, because salt water sinks away from the cold surface before it cools enough to freeze. Furthermore, other factors cause the formation of sea ice to be a slow process. The freezing temperature of salt water is lower than fresh water; ocean temperatures must reach -1.8 degrees Celsius (28.8 degrees Fahrenheit) to freeze. Because oceans are so deep, it takes longer to reach the freezing point, and generally, the top 100 to 150 meters (300 to 450 feet) of water must be cooled to the freezing temperature for ice to form.”
Enjoyed your post EFS Junior…Caught a thread from last year which sort of reminds me of this year allthough we have had different results..http://anhonestclimatedebate.wordpress.com/2009/06/25/%E2%80%9C%E2%80%A6arctic-temperature-is-still-not-above-32f-the-latest-date-in-fifty-years-of-record-keeping%E2%80%A6%E2%80%9D/
Looking at the warmer than normal waters in the Bering Strait/north..Our only hope right now is the prediction that this year would see a large ice melt but not as large as 07 but with a dramatic refreeze made by Joe Bastardi back in may I think.
Air Temps are averaging just below freezing at the outer ice edge areas that have seen the most melt as far as I can tell…
EFS_Junior says:
September 8, 2010 at 11:22 pm
Correction: ~4F should be ~7F (~4C).
Should be ~39ºF.
Scott:
… a while back you ran a sort of 2-stage forecast … sort of a “learning” function.
It gave an ABSURD result:
4.5 million (extent) …. on OCTOBER 6.
Run it again.
You really ought to send this in – I mean the WHOLE METHOD — to the Sea Ice List at the Arcus site.
Of course they might already be doing that, but it sounds right up their alley.
And it Might be Correct.
You HAVE to make the Bizarre predictions BEFORE they are reasonable, for True Science.
To TEST a theory, it has to be on the table BEFORE the result is in. I know _I_ looked silly with 1.0 m km2, but PARTS of my theory worked, I will have the bad parts fixed Next time.
JAXA revised upwards to 4989375..Whew!??..still a 38,000 drop but hopeful. My Temps at 77 north have dropped to an ave of 30.8…
Confirmed JAXA 15% extent for Sept 8th: 4989375. Updated charts …
15-day: http://img688.imageshack.us/img688/594/15day20100908.png
7-day: http://img839.imageshack.us/img839/4430/7day20100908.png
Updating the ensemble Arctic sea ice extent slopes (km^2/Day) of Bremen (8/26-9/9), JAXA (8/25-9/8), and NSIDC (8/25-9/8), the ensemble (combining the data from all three indicies) slope is -36K/Day (R^2 = 0.92).
Bremen, JAXA, and NSIDC all use the same 15% concentration cutoff definition, so combining all three should increase the confidence of the resultng ensemble slope (at least in the very short term of a few days).
For each individual dataset;
Bremen slope is -29K/Day (R^2 = 0.90)
JAXA slope is -34K/Day (R^2 = 0.94)
NSIDC slope is -46K/Day (R^2 = 0.99)
NOTE: Fitting a quadratic to either the ensemble or any of the individual datasets results in concave down trend lines at this time (i. e. no change from yesterday). The whole purpose of the ensemble was to try to detect a concave up composite trend line (in other words, another method of determining the magnitude and timing of the 2010 minimum).
The following sequence of hypothetical daily losses for the JAXA dataset is of very low probability (~3.3% at this time) but is consistent with the current ensemble slope;
-30,275 (9/8)
-29,238
-28,050
-27,160
-26,599
-26,269
-25,583
-23,984
-21,329
-18,563
-15,374
-11,371
-6,749
-2,253 (9/22)
Final hypothetical extent = 4,707,812 km^2 (beating the 2008 JAXA minimum by one km^2)
On another note, the set of JAXA minimum dates (2002-2009) are;
9/9/2002 (included even though there is missing data)
9/18/2003
9/11/2004
9/22/2005
9/14/2006
9/24/2007
9/9/2008
9/13/2009
The average or mean (or p = 0.5 or 50% or even money chance) of the above is 9/15 with a standard deviation of 6 days, giving a p = 0.68 (68%) chance of the minimum date falling between 9/9 and 9/21.
Note that the two dates that fall outside these bounds were 2005 and 2007, and both fall outside on the high side for whatever reason (I noticed this a few days ago, in looking at the time series, and thinking to myself that 2010 had visual minima characteristics most similar to 2005/2007).
Charles Wilson said
September 9, 2010 at 7:06 am
To TEST a theory, it has to be on the table BEFORE the result is in. I know _I_ looked silly with 1.0 m km2, but PARTS of my theory worked, I will have the bad parts fixed Next time.
_____________________________________________
I don’t think any parts of your theory worked when you are so far out. I don’t think there will be a next time either for you to be brutally honest, I think the scientists will have put you in the “bit of a crank” bucket and will stop it happening again.
Anyone who mixes up capitals and lower case in a poorly written out blog post should not be having the chance to submit items to scientific web sites. That’s the problem with the Internet, no peer review.
I don’t really like being so blunt, but somebody had to say it.
Andy
Anu says:
September 8, 2010 at 8:10 pm
“Non affecting a downward trend” ? They are part of that downward trend.
See for yourself:
http://img204.imageshack.us/img204/3219/septemberminimums.jpg
Interesting curve indeed. Being a humped curve, both ends have a downward trend. Extrapolating backwards, I guess your point here was that ice was increasing prior to 1980? How exactly does that fit in with CAGW? (The GCMs in your link show conveniently flat Arctic ice prior to 1980 but that’s not quite data).
Keep watching the Arctic this decade
I will indeed …
The powerpoint by Maslowski attributed Arctic ice decline to ocean warming; however looking at ocean temperature trends, the central and western Pacific upper 300m heat content appears on a slow decline over the last 10 years (NOAA) as does the PDO expressed as a temperature anomaly (JISAO) over the last 30 yrs.
( http://www.climate4you.com/ )
plus North Atlantic OHC appears on a downward trend (Bob Tisdale’s blog). Also, the rate of sea level rise – an indicator of OHC change – was much smaller for the last 2010 el Nino than for the previous one in 1998 (Colorado Center for Astrodynamics Research), although the air temps in both were more similar.
Where is the ocean warming coming from going forward?
I just pulled up the PIPS2 page for a lark. I can’t believe some of the stuff they show- (http://www7320.nrlssc.navy.mil/pips2/ithi.html) – ice thickness of >2.5m off Banks Island (see sat. image – http://exploreourpla.net/explorer/?map=Arc&sat=ter&lon=-133&lat=76,9&lvl=7&yir=2010&dag=251) and ice > 4m N of Svalbard (image – http://exploreourpla.net/explorer/?map=Arc&sat=ter&lon=53.8&lat=83.1,9&lvl=7&yir=2010&dag=251), which is just delusional. Wonder what else is wrong there.
AndyW says:
September 7, 2010 at 10:31 pm
Anyhow, agree with Anu, 4.xx is very much on the cards now.
Yup, the cards cooperated – 4.xx is in the history books now.
http://img713.imageshack.us/img713/1296/jaxalatestseaiceextent.png
The recent slope is pretty steep – it’s unlikely to V up in the next few days, but it could flatline. It’s also possible that it could drop below 2008, but without Cryosat-2 data on the thickness of the broken up ice margins, and better temperature data of the upper 100 meters of ocean in those areas, it’s just a guess at this point.
http://img838.imageshack.us/img838/4530/arclatestlargecircles.png
Some of those yellow circles could include sea ice that is merely 10 or 5 cm thick now – bottom melt this next week could be 1 to 2 cm per day. It’s a race to the finish line.
I hear a lot of talk about the importance of ice drifting out the Fram Strait, but I don’t see evidence of more than 5,000 sq km or so making it to the “melt zone” per day. Perhaps someone has more accurate, timely data.
phlogiston says:
September 9, 2010 at 12:07 pm
Interesting curve indeed. Being a humped curve, both ends have a downward trend. Extrapolating backwards, I guess your point here was that ice was increasing prior to 1980?
Don’t get confused by the ‘infinite in both directions’ properties of a quadratic curve – mathematics is just a tool. If you fit a sinusoidal curve to the annual changes in distance from the Earth to the Sun, that curve doesn’t imply the Earth existed 8 trillion years ago. Try to focus on the data of interest.
A sigmoid function might make an interesting fit to this dataset – try it if you think it will give some insight into the future. But the data doesn’t have to follow any easy mathematical function – it can just drop off the cliff in an ugly, jagged way.
(The GCMs in your link show conveniently flat Arctic ice prior to 1980 but that’s not quite data).
You’re right – prior to 1980 the satellite data already showed a slight decline:
I’ve previously posted on WUWT scientific papers that looked at this satellite data starting in 1972. “Skeptic speculation” that data before 1980 would show increasing Arctic sea ice extent are unfounded.
The powerpoint by Maslowski attributed Arctic ice decline to ocean warming; however looking at ocean temperature trends…
Look at the Argo data for the upper 2000 meters of the worlds oceans. This has been hashed out on many previous threads at WUWT – Bob Tisdale doesn’t understand how surface ocean heating can move down in the ocean in small area vertical currents and then back up at a later time and place. The oceans have complicated underwater-surface topology, salinity profiles, currents from winds, the spinning planet, heat differentials, etc. Looking at a small region or upper 700m gives an incomplete picture.
The oceans are warming, and the Arctic summer sea ice is disappearing:
Where is the ocean warming coming from going forward?
Ultimately, the Sun. The increasing fraction of CO2 in the atmosphere is causing the radiation budget to become imbalanced, and the overall temperature of the Earth will rise until the energy budget is balanced again. This will warm the oceans, the lithosphere (rocks and dirt), the atmosphere, the cryosphere, etc.
Watch the Arctic – the skeptics will be proven hopelessly confused there much earlier than for the rest of the planet.
Things will be pretty clear for most people by 2020, except for the most [snipped, no personal ridicule please. Robert]
Anu is a skeptic of cosmic rays.
Andy W.
In 2007, Everyone’s prediction was FAR off.
So every Professional in the World is a Crank ?
My Outlook was conditional: _IF_ Sunny, like 2007
— no one could tell me whether it would be.
Next time: Someone will have figured it out.
Me.
I predicted the Dipole’s return right. I predicted High Pressure’s return & when.
NOAA predicted Record High Hurricanes.
_I_ predicted Saharan Dust Storms would kill the HURRICANES – – Result: record Low Hurricanes. Are they ALL cranks ? – – And I was Right, again
When NOAA’s Method predicts a Disaster, they have to come out with it even if it is unceertain – – the typical Hurricane prediction is 1-in-4, thus the Governor cancelled the Katrina evacuation, killing an extra 1000. Because all the Experts were Cranks. PS Look at the Aerosol Daily map: http://www.osdpd.noaa.gov/ml/air/index.html – – a wind shift sent the dust south, the “hurricane Nursery” is clear, just like before Earl : LOOK OUT East Coast. Will I be Right or Wrong ?
>> Yet you are right in this: if Scott keeps his Method secret
.. he cannot be Wrong & no one will call him a Crank.
This is why I praised Steve for putting up an Outlook. Guts. (excuse the use of Capitals for Emphasis).
Anu says September 8, 2010 at 7:21 am
Believe it or not I do know what the density and latent heat of fusion of water are thanks. Interested to know where your volume estimate has come from though, and what the tolerance is.
Look at any of these MSPPS Sea Ice concentration images : http://www.osdpd.noaa.gov/ml/mspps/seaiceprd.html
WOW.
Looks like under 3 million km2. Implies 2 m km2 area.
Close exam suggests 10-25% concentrations are not shown so it is a bit like the 30% Extent of DMI, even so, it is a LOT less. Anyone know why ?
JAXA provisional 15% extent for 9th = 4965781, a 23.5K loss.
Charles Wilson says:
September 9, 2010 at 4:59 pm
Sorry to burst your bubble Charles, but I explained all my methods in Sea Ice News #19. They are all very simplistic and statistically-based. Nothing an undergrad student in a stats class couldn’t do. If I remember right, the method you refer to is highly dependent on the endpoint data and consequently it was just random chance that it gave an interesting prediction on that day.
Also, others have commented on it, but if you could format your posts to be a little more readable (an extra line break between thoughts would do wonders) would make them much more reader-friendly.
In regards to the ice, the past week’s losses have been nothing less than brutal. It’ll be exciting to see what kind of records are going to be broken and see if we can catch back up (or down) with 2008’s minimum.
-Scott
Scott:
I’ll have to take a Stats class. Thnx for the endpoint tip.
Scott said (Sea Ice #19)
… I was looking for an approach that wouldn’t need previous years’ data to work. As a first shot, I tried plotting daily loss on the y-axis with date on the x-axis. I then used a linear fit to this to predict future performance, extrapolating to the minimum. Again, I arbitrarily started the plot on Aug 1. …it’s predicting a minimum of 4.52e6 km^2 on Oct 6,.
… How does a Linear extrapolation produce a Late Minimum … by 2 weeks more than the Record ? … and without using previous years … it is the Time element I was shocked by, as Sea temps imply a very late melt, but not How late. I figured you’d stumbled onto something that indicated that. Can you explain your derivation of minimum DAY ?
Confirmed JAXA 15% extent for Sept 9th: 4972656. Updated charts …
15-day: http://img811.imageshack.us/img811/6962/15day20100909.png
7-day: http://img255.imageshack.us/img255/5944/7day20100909.png
Anu says:
September 9, 2010 at 2:06 pm
Look at the Argo data for the upper 2000 meters of the worlds oceans. This has been hashed out on many previous threads at WUWT – Bob Tisdale doesn’t understand how surface ocean heating can move down in the ocean in small area vertical currents and then back up at a later time and place.
I dont think Bob Tisdale is the only one who has problems with warm low density water sinking down into colder higher density water. OK you have wind and surface currents and choppiness, and on the small scale convection can overcome thermodynamics and density. The new Argos data about previously unavailable complex spatial structure of water temperature which are interesting. New data of this sort will always throw up unexpected complexity and structure – this makes experimental science interesting and rewarding.
But you yourself have admitted that data that goes even as far down as 2000m (sometimes) is incomplete. Just as many surprises await us in the 2000-4000+ m ocean depth in terms of temperatures and movement patterns.
If we are talking about multidecadal and even longer duration oceanic cycles, then what we are talking about is THC – the currents going all the way to the ocean floor that are largely unaffected by what is happening at the surface. Even undergraduate oceanography makes this clear (or at least did 20 years ago). THC is driven by large scale downwelling at places such as the Norwegian Sea and some locations in the Southern Ocean. The other side of the coin of downwelling is of course large scale oceanic upwelling such as in the south eastern Pacific off Peru (driving the ENSO cycle and fortunes of the Peruvian anchovy fishery). It would be nice if downwelling and upwelling were tidy, constant and predictable but of course the reality is natural quasi-chaotic complexity with emergent patterns such as cyclic rise and fall in upwelling. It is often repeated on this site that the oceans hold the vast majority of climate heat energy. So subtle changes in THC and upwelling spatio-temporal pattern could easily exert a controlling influence on global temperature oscillations. I dont think the temperature sea surface structures revealed by Argos tell the whole story concerning global THC.
When a new technology arrives in a scientific field e.g. Argos floats, it is human nature for young practitioners of the new tools to dismiss all work done by the older generation of scientists with supposedly “out-dated” tools. The problem with this is that the old-timers are generally proved right in the end. Young scientists sometimes dont like reading papers more than 10 years old and only condemn themselves to re-discovering repeatedly what was already known.
The oceans have to be considered as a whole system where THC plays the leading role, in order to understand its role in climate, rather than selectively looking at parts of it which paint a picture to one’s liking.
Finally – next year we will be in a full on La Nina and some such as Joe Bastardi are predicting that cooler water will gate-crash even the Arctic. JB accurately predicted a near-2007 ice minimum now, against many on WUWT including myself, but predicts strong recovery for 2011-12. It would take courage to bet against that with JB’s track record. A 2011-12 uptick in minimum would spoil the stats on Tamino’s quadratic curve.
But the data doesn’t have to follow any easy mathematical function – it can just drop off the cliff in an ugly, jagged way.
..or an equally ugly jump-up (beauty is in the eye of the beholder 🙂
The oceans are warming, and the Arctic summer sea ice is disappearing:
One can only admire your faith. Preach it sister!
Things will be pretty clear for most people by 2020
Once again, here we agree – who could argue with that? You say death spiral I say recovery, lets call the whole thing off. Till 2020. (or maybe not)
Updating the ensemble Arctic sea ice extent slopes (km^2/Day) of Bremen (8/26-9/10), JAXA (8/25-9/9), and NSIDC (8/25-9/9), the ensemble (combining the data from all three indicies) slope is -37K/Day (R^2 = 0.93).
Bremen, JAXA, and NSIDC all use the same 15% concentration cutoff definition, so combining all three should increase the confidence of the resultng ensemble slope (at least in the very short term of a few days).
For each individual dataset;
Bremen slope is -31K/Day (R^2 = 0.91), current extent is ~4.70E6 km^2
JAXA slope is -34K/Day (R^2 = 0.95), current extent is ~4.97E6 km^2
NSIDC slope is -47K/Day (R^2 = 0.995), current extent is ~4.72E6 km^2
This makes the JAXA extent significantly higher than the other two sites, by an average of ~0.26E6 km^2, suggesting that JAXA might drop some more, following a pattern similar to what occured a little over a week ago (but only time will tell).
NOTE: Fitting a quadratic to either the ensemble or any of the individual datasets results in concave down trend lines at this time (i. e. again, no change from the past two days). The whole purpose of the ensemble was to try to detect a concave up composite trend line (in other words, another method of determining the magnitude and timing of the 2010 minimum).
The following sequence of hypothetical daily losses for the JAXA dataset is of very low probability (~2.1% at this time) but is consistent with the current ensemble slope;
-29,305 (9/8)
-28,126
-27,263
-26,737
-26,446
-25,789
-24,208
-21,558
-18,802
-15,614
-11,597
-6,948
-2,450 (9/22)
Final hypothetical extent = 4,707,812 km^2 (beating the 2008 JAXA minimum by one km^2)