Yesterday, I promised to add the bets to the projection graph. I grabbed the bets from the database, culled duplicates from identical people and added dots to the graph. (No guarantees I made no mistakes. This was done manually.) Below, you can see the projected value September NH Extent anomaly (solid blue) based on a linear fit to past JAXA values at this time of year along with the ±2sigma uncertainty intervals. The mean of “late” NH ice bet (ignoring the bet for a negative extent) and along with the ±1sigma range are show in with a vertical purple trace. Dots for each individual bet are shown just to the right of that trace. Since it’s my blog, my bet is shown in red. Just to the right of that trace, I show the bets that closed in June in lavender.
Data of interest. (Ice extent values in * 10^6 sq. km)
- Most recent Jaxa: 3.95 (Black dot and dashed line.)
- Current ‘best’ estimate based on extrapolation of JAXA: 3.28 . [3.00. 3.55]
- Mean Bet (Late) : 3.95
- Mean Bet (Early) : 4.70
Note that we can pretty much already decree the winner of the “early” bets. Well… unless Mars attacks causing temperature in the Arctic to plunge permitting ice to refreeze almost immediately. Meanwhile, it appears that my “late” bet is too high, but owning to the gap in bets around what appears to be ‘most likely’ value, I might come in 2nd. (Carrick got his bet in about 40 minutes before I did and would take the quatloos if we are closest. (Why didn’t I bet what 3.5 as my projection actually recommended instead of getting optimistic? Oh. Well. Still, this is only based on JAXA extent. I haven’t checked to see if that model is “best”. It’s probably not and the error bars ought to be larger.)
It is nice to see that with the late bets, the spread still includes the projections. So, we still have some excitement in that outcome. If catastrophe strikes, procozio who occupies the “effectively zero ice” position will clean up. But for now, it looks like we can’t say for sure who will take the quatloos.
I am doing my, “in the Quatloos” early dance!
Be warned: Mars could attack causing an uber-cold September. . .
That’s why we sent probes to monitor the traffic…
The rate of decline I have found which most closely matches the actual figures from June 1st., is the 2008-11 mean x 1.24.
On the assumption that continues, the projected average for September would be 3.652 million km^2, with a minimum of 3.488 million km^2 on Sept. 10th.
However, that assumes that the rate of increase, after the minimum, will also be 1.24 x the mean, which may not be valid.
Based on a rate of the mean x 1.0 after the minimum, the average would be 3.628 million km^2
Both figures would of course, make SteveF the winner, not yourself or Carrick!
Underneath that quiet, sandy landscape, Marvin is assembling his fleet, all armed with the Uranium PU-36 Explosive Space Modulator.
Meanwhile, given it’s been an odd year in the Arctic (early ice extent higher than usual; ice broken up by storms and winds) might not the minimum and the recovery also be unusual? Like sooner and faster?
Interesting times.
But opportunity again to ask – is it evidence of warming?
Intuitive model A says “warming is melting the Arctic ice”.
But DMI melt season temps are normal and were higher fifty years ago:
http://ocean.dmi.dk/arctic/meant80n.uk.php
And Arctic ocean heat content ( perhaps dubious ) indicate a cooling Arctic ocean:
http://www.climatewatch.noaa.gov/wp-content/uploads/2012/07/OceanHeatAnom_2011_720.jpg
Model B says: “circulation is driving ice out of the Arctic”
That’s certainly something we can see:
http://upload.wikimedia.org/wikipedia/commons/6/64/Decrease_of_old_Arctic_Sea_ice_1982-2007.gif
And is consistent with heat loss because the exposed and thinly covered Arctic loses more heat during winter than the thickly covered one.
But one would expect a dynamic loss to fluctuate eventually to a circulation which allowed accumulation, and that hasn’t happened over the course of the record.
My short term quatloos are on persistence of the trend, but my long term quatloos are still on the full range of Model B.
cui
Sure. That’s why it ain’t over ’till it’s over.
Well… at a minimum, if it’s warming you do tend to expect melting.
My long term bet says there will be less ice in the 2020a than in the 2010s.
The annual HadCRUT3 anomaly figure for the N.H. Extra Tropics region (north of 30 degrees), up to July, is 0.913c, only slightly below the highest annual figure on record, of 0.921c for 2007.
Ray,
I do understand that melting takes a lot of latent heat to accomplish,
but melting takes place only during the melt season, and Arctic temperatures don’t seem to be changing that much during the period when melting occurs. And there is much more heat stored in the Arctic waters, which appear to be losing heat, than in the air anyway.
I’m still watching NIC, which is still slightly above 2007, with a drop coming in the 2007 record. Meanwhile air temperature has dropped below zero. Interesting days ahead.
http://www.natice.noaa.gov/products/ice_extent_graphs/arctic_daily_ice_extent.html
Select start year 2006 and month of August.
cui bono (Comment #102261)
August 28th, 2012 at 10:43 am
It’s been unusual because there is so little of the thick, permanent sea ice left, that the thin stuff we have now can be broken up by something as simple as a storm.
Extent is not telling the underlying story, the sea ice mass, which appears to be permanently changed. All we have now is the stuff that freezes from year to year over winter. It freezes over quickly, and is gone just as quickly.
“…which appears to be permanently changed.”
Yikes. Permanently? Forever…and ever…and even after ever?
Lucia, it seems I mistyped my bet on the latest round – I must have typed 4.2 when of course I meant to type 3.2 – could you correct it for me please?
Thanks.
I see that talk of the Arctic Sea Ice “Death Spiral” has returned at Open Mind:
http://tamino.wordpress.com/2012/08/29/arctic-sea-ice-death-spiral/
lucia (Comment #102264),
August 28th, 2012 at 1:10 pm
Is that based on the entire planet continuing to warm, or is it something unique to the arctic?
Skeptical–I think the planet will continue to warm owing to ghgs. The arctic will continue to warm along with the rest of the planet.
Actually, the arctic appears to be warming considerably faster than the rest of the planet.
Ray (Comment #102275),
Which is why I asked my previous question. There does appear to be something unique happening in the arctic.
@Skeptikal: The entire planet warms, but due to a phenomenon known as ‘Arctic amplification’ the Arctic in particular warms faster. See http://www.skepticalscience.com/What-causes-Arctic-amplification.html
Benjamin, I think they’ve decided declining ice is less important than the wintertime inversion in reducing heat-loss and increasing amplification. As you probably know there is little arctic amplification during summer months, and is maximum in the months where the solar isolation is the greatest.
Figure.
I don’t think it’s fully understood yet, but may have more associated with boundary layer physics than the driving associated with ice loss. [In a way that’s a more satisfying explanation because it doesn’t depend on “Rube Goldberg” models to explain why you get a flip between when the ice gets lost and when you see maximum amplification at the surface.]
But here’s the thing – Arctic Amplification is supposed to be from albedo feedback. We should have observed albedo feedback now because of the decline in ice area ( and more importantly, the decline in multi-year to first year ice ). But the ocean heat content ( which contains more energy than the air ) demonstrates heat loss, not heat gain:
http://www.climatewatch.noaa.gov/wp-content/uploads/2012/07/OceanHeatAnom_2011_720.jpg
With less and thinner ice, the Arctic Ocean is much more able to lose heat to the atmosphere, which is what we observe – Arctic Ocean cooling with Arctic Atmosphere warming. There is evidence that more than half the sea ice loss has been dynamic:
http://seaice.apl.washington.edu/IceAge&Extent/
http://www.agu.org/pubs/crossref/2004/2004GL019492.shtml
The question is when, if ever, should we expect those dynamics to revert? Should they revert and sea ice accumulate, the atmosphere may again be cut off from the surplus of energy it has been receiving from the Arctic Ocean.
Examine the history of the Arctic, and see that it may have occurred before: http://data.giss.nasa.gov/gistemp/time_series.html
Carrick,
Hm. Solar isolation – not to be confused with solar insolation (an oxymoron anyway which people still use).
What goes on in the Arctic winter (besides it’s night all the time)? Is there much weather to speak of? With an iced-over ocean, the heat exchange between the liquid water below the ice, and the air above air has to be pretty slow (?). Does the water even move around much? The inversion you speak of – warmer air is aloft due to better horizontal mixing at height? The inversion inhibits both convective and radiative losses? No “solar insolation”? It’s a wonder it doesn’t get colder than it does.
Well, sorry for all the questions, but I’m ignorant.
BillC,
Regarding “solar insolation”, I think you might mean it’s “tautological”.
BillC,here’s the reference.
Long day so I hope this is coherent:
Briefly, an inversion gets set up during the polar winter (temperature increases with height). This is a result of the fact that the ground is a more efficient radiator than the atmosphere above it and thus loses heat more quickly.
There are four mechanisms for exchange of heat energy, convective and expansion, convection and change of state (heat of fusion or vaporization), radiation and conduction.
Convection can either be forced (e.g, wind sheer) or it can be due to heating. Large scale heat-driven convection is damped at night due to the inversion… higher altitudes are warmer and hence less dense, so if you move a packet of air upwards, you are moving a more dense packet into a less dense region, and it will want to return to it’s original position… it’s stable against vertical perturbation.
[Day time, with a positive lapse rate gamma = -dT/dz, you have the opposite picture, higher altitudes have higher density because they are colder, and if you move a packet of air upwards, it will want to keep moving… classic thermal instability.]
Anyway, with an inversion, three of the four mechanisms are suppressed, both convective mechanisms and radiative heat loss near the surface. (Net heat energy will be exchanged from the warmer layer of air to the cooler layer, hence heat energy absorbed by the atmospheric layers in the inversion, will tend to retransmit that heat energy back towards the surface.)
I think their argument is a warmer surface equals more heat energy transmitted back to the surface, which is a positive feedback. I am unsure why people think that inversions amplify radiative heat loss. (They make it more important because convective heat losses are generally nil, except on windy nights, but still…)
I think insolation is correct btw (as opposed to “insulation”.)
Arctic ice extent for August 29 from NIC:
2007. 5.04 M sq miles
2012. 5.93 M sq miles
My “model”, using the 2008-11 mean change x1.24 is still on track to produce a September average of about 3.6 million km^2.
At the end of August, the predicted extent using this method was 3.812 million km^2, compared to the actual of 3.801, a difference of only 10623 km^2, or less than 1 day out after 3 months.
However the figure for the 31st will probably be revised upwards, so the prediction may be even closer.
Of course, past performance is no guide to the future, (as IPCC temperature models demonstrate), but I wish I had used this method to arrive at my original forecast for the bet.
Ice is growing again:
29/08/2012 3.75251
30/08/2012 3.79271
I know!!
My bet is back in the ±2sigma using only extent. I’m adding in area and getting ready to show the horserace as we enter September.
I assume your ‘uncertaint’ interval is a 95% confidence interval for the 95% conditional mean of ‘September Extent’ versus ‘Jaxa. The 95% prediction interval (for a single future observation) would seen to be the one of interest — and is more Lucia bet friendly.
pfffft!.. splelling…awlays apblom..sigh
mwg– I meant it to be a prediction intervals… but I actually screwed up and forgot a little chunk of the uncertainty. I’ve got a new post up.
steveta– whose are those? Jaxa is
242 8 29 3912500
243 8 30 3877031
244 8 31 3801406
It’s not growing in any of the indexes I’ve looked at but temperatures have plummeted in recent days to -5°C, well below the freezing point of even sea water.
In just a few days, the melt ponds are nearly frozen over, and I’m expecting a “knee-point” to emerge in the data associated with the change in albedo as a clear feature (for the record I think it’s already there as the “bend” in recent days, but I think you need at least a weeks data to firmly say “yes the trend has changed”).
Carrick–
Yes. I’d say the loss rate has slowed — but it will oscillate up/down, down, down,/up/down, down… before it really bottoms out. When I read that I was excited because the rate has slowed enough that my projection rose. 🙂
Lucia,
I was too brief in my earlier comment today. However, if the interval (band) in the ice predictor is a prediction interval, as it appears to be in todays latest and greatest, you should call that instead of a confidence interval (band). I suggest this because in an earlier regulatory sector phase of my work it quickly became apparent that ambiguous, casual use of statistical terms inevitably lead to confusing discussion–getting worse as more parties are involved. My experience has been that as a group scientist and engineers are not as tuned in with statistics as one might expect–afterall for many of us it may have been more of distraction to what we were really interested in our formative years–or we followed procedures by rote.
I should have been more explicit earlier–the term ‘uncertainty interval’ is ambiguous; common use of confidence interval is in reference to an interval for the [conditioned] mean; prediction interval is for prediction of (specified) one or more future observations, and a tolerance interval is for a future predictions.
Clearly the interval is of minor interest here, so I am really just encouraging use of consistent clear language in support of the cause–running a gambling site, and I do not wish to distract for the joy of your online patrons.
Finally, while I thoroughly enjoy the tone and content of your postings, but I do think you should have more pictures of cats, and seek out their contributions. … More cats.
mwg–
Yes. I should. I’m being sloppy. I do mean the prediction intervals.
I’m an engineer. 🙂
Well.. I have Grizelda below watching out for image scrapers and “The General” above greeting visitors. ‘Mo has made an appearance from time to time.
I’ll try to get ‘Mo and Grizelda to pose more. Alas, The General loved to have his picture taken and would pose (as he did with the chipmunk), but he is no more.
And I a (sloppy) scientist. It has been pretty clear to me what you have done along the way in the postings and certainly easy enough to piece things together, i.e., reproduce, given the info you provide. (The 8/31 graph suggested the PI band–very little curvature and all of the data point contained in the band–and looks very much like one I have calculated from digitize 8/28 data–other than one data point has visibly changed between 8/28 and 8/31. And of course the earlier 8/28 CI had the hourglass shape.)
So sloppy in approach, not in a serious or even nit-picky way. And it is clear that I was not telling you something that you and many of the seasoned contributors to this blog do not know. My advocacy is to use the correct language from the beginning for the benefit of those readers that have not had to delve into statistics beyond mean, standard deviation, and maybe simple linear regression. Maybe it will stick with some, and that would be a good thing.
I say this good-naturedly–don’t reply that you were sloppy–we all are. Say, ‘Yeah, I think we should try to tighten our language here in blogworld so let’s work toward it.’ There’s a lot of crap in climate blogworld and some good insight stuff here–in digestible size chunks. Tweak it this way and make it even better.
Enough on that, but do expect me to keep the pressure up in regards to the cats. I know cats. I think they could do more.
The average rate of loss normally slows at this point in the cycle.