SEARCH Outlook Contribution May 2012

My SEARCH contribution follows. However, I want to say this is the best statistical scheme I have found and doesn't necessarily mean I believe it. If I was submitting a huristic estimate then I would be saying:

I find myself concerned at the rapid drop in area and the appearance of the ice. High area in March April and May in southerly locations should have reflected more solar radiation and kept things cool. Despite this Chukchi isn't far behind normal. The volume is about the same as last year but spread more thinly into southerly locations. The time for that to slow down retreat of areas like Chukchi has almost past and it seems to me as if the thinner ice at more northerly locations will now cause faster retreat so that areas like Chukchi can catch up if not overtake. On the Atlantic side, so far the weather has prevented fast retreat in these weak ice area. That is unlikely to continue indefinitely and despite the weather the Barents and Kara are still looking very weak compared to past years. I think there has been high outflow through Fram strait, and high Atlantic water heat content inflow recently and low snow cover in April and May around the arctic basin.

Weather may be able to overcome these factors but it could also enhance the melt. With average weather I fear record low ice levels could occur. I usually try to avoid being too alarmist but I don't think things look good for the ice.

So although the statistical scheme below gives 4.29 which is in the 4.25 to 4.5 range, it is close to the border and in the poll I have gone for 4-4.25 for NSIDC average extent and 2.8 to 3 for area. i.e next to bottom categories for both.

If I have such views why stick to the statistical scheme?

One problem is how much do you adjust for your feelings. Another reason to leave as this naive statistical estimate is that my series of estimates hopefully providing some sort of guide to how much more accurate estimates need to be as we get closer to the minimum in order to demonstrate skill over a naive statistical model like I am using.

 Anyway here is my contribution:

1.        Extent Projection 4.3 +/- 95% confidence interval of 1.12 M Km^2 based on past performance of this technique.

 

2.        Methods/Techniques - Statistical

 

3.        Rationale

 

Towards the end of May Extent and Area tend to show little variation. They also show little skill for predicting the minimum. The methods I have presented last year worked on extent and area showing how the season is progressing and gain more skill as the season progresses.

   

The second method of predicting falls from current postion using area, extent or culmulative energy that could be captured by ocean that was formerly ice does not show any skill at this length of time before the minimum.

 

The first adjusted gompertz fit method at this time of the season barely shows any skill over Hamilton’s Gompertz fit method. This method attempts to predict the residual from a Gompertz fit of NSIDC average September Extent by linear regression using the residual from a Gompertz fit of Cryosphere Today area at 31May.

 

4.        Executive Summary

   

I attempt to predict the residual from a Gompertz fit of NSIDC average September Extent by linear regression using the residual from a Gompertz fit of Cryosphere Today areas at 31 May.

5.        Estimate of Forecast Skill (if available)

            The standard error arising in the linear regression is .44 M Km^2.

            However standard errors of a method tend to underestimate the errors likely in practice. So I have used only information up to 31 May of the year being predicted to predict each of the last 10 years. As shown in the table this gives a RMSE for this method of 0.56 M Km^2 which has been doubled to give a 95% confidence interval of 1.12 M Km^2. This is hardly any smaller than the RMSE of 0.57 M Km^2 for Hamilton’s Gompertz fit method. Consequently, it is not clear whether there is any skill at this lead time compared to just using a Gompertz fit.

            Estimate at 31 May of year concerned       

 

Year	Actual	Gompertz Fit	Gompertz & Area Fit	Gomptz Fit Err	Gomptz & Area Fit Err	Gomptz Fit Err^2	Gomptz & Area Fit Err^2
2002	5.96	6.44	6.44	-0.48	-0.48	0.23	0.23
2003	6.15	6.17	6.11	-0.02	0.04	0.00	0.00
2004	6.05	6.06	6.06	-0.01	-0.01	0.00	0.00
2005	5.57	5.95	5.89	-0.38	-0.32	0.14	0.10
2006	5.92	5.65	5.42	0.27	0.50	0.07	0.25
2007	4.30	5.62	5.59	-1.32	-1.29	1.75	1.67
2008	4.73	4.74	4.54	-0.01	0.19	0.00	0.04
2009	5.39	4.40	4.61	0.99	0.78	0.97	0.60
2010	4.90	4.58	4.48	0.32	0.42	0.10	0.18
2011	4.61	4.46	4.41	0.15	0.20	0.02	0.04
2012		4.28	4.29

                                  MSE   0.33         0.31

                    RMSE   0.57         0.56

Comments

[Ned Ward]

Very nice, crandles.

Over at WUWT, Anthony Watts has also posted his official prediction for ARCUS:

"Since there was a tie between the 4.8-4.9 and the 5.0-5.1, values, the median value of 4.9 million square kilometers was chosen for submission."

In comments, Peter Ellis rightly points out that the median of those two ranges is actually 4.95, not 4.9.

Frequent WUWT commenter Pamela Gray is having trouble following Peter's math:

"hmmm. I see three numbers Anthony listed when deciding which one to send it, 4.8, 4.9, and 5.0. The median of those three numbers (not the mean, and not the mode), would be 4.9. So Peter, where do you see 4.95?"

FWIW, I like Peter's suggestion that Anthony chose to round the number down in an attempt to reduce the embarrassment of once again over-predicting sea ice extent.

But it's also possible that Anthony just miscalculated the median.

[Neven]

Good to see you submit again to SEARCH, crandles. You and Larry are the pride of the ASI blog. :-)

So although the statistical scheme below gives 4.29 which is in the 4.25 to 4.5 range, it is close to the border and in the poll I have gone for 4-4.25 for NSIDC average extent and 2.8 to 3 for area. i.e next to bottom categories for both.

I voted one notch higher on both polls (4.25-4.5 for NSIDC monthly minimum extent, and 3.0-3.2 for CT daily minimum area), but considering the latest PIOMAS update I might change those before the end of the month.

[Stevemosher.wordpress.com]

hehe.

nice crandles. 4.29

That's the bet I have in over at Lucia's

[crandles]

Great minds or fools never differ? ;-)

[Account Deleted]

http://psc.apl.washington.edu/lindsay/Prediction/2012/September_ice_extent_2012.html
2012 Prediction Season

End of May 2012: This prediction is made with model data from May 2012. The best single predictor is the fraction of the area with open water or ice less than 1.0 m thick, G1.0. This predictor explains 79% of the variance. The predicted extent in September is 4.06 +/- 0.42 million square kilometers (orange star below). If the prediction is accurate it will be a new record low (the 2007 record low was 4.3 M sq km), however the previous record is within the error bars of the prediction. The region most influential in making the prediction this year is an area stretchng from the eastern Beaufort Sea and also in the Kara Sea (right map in the figure) where there is a greater than normal fraction of thin ice (middle map) and the G1.0 variable has a significant correlation with the September ice extent (left map). The top figure shows the time series of the observed September ice extent (solid line), the predictions of the satistical method for past years (cyan diamonds), and the prediction for this year (orange star and error bars). The error bars are the standard deviation of the error in the fit of the regression. The trend line (dashed) and the prediction of the trend line (black star) are also shown.

[Lord Soth]

I find it interesting that even the climate deniers are even reporting numbers below the 5 million sq km mark.

There has been a large export of ice out of fran strait so far this year, and if it keeps up, the north pole could find itself under first year ice, by the end of the summer.

Any bets on a melt out to the pole, for 2012 ?

[Stevemosher.wordpress.com]

ha crandles, i like to think great minds.
Last year sure looked like a record breaker in the making and I considered it to be fluke that the record didnt fall. This year the ice looks worse, so I cant imagine fluking out two years in a row. With the "right" weather the record could be crushed.

At some point I want to chat about your regression as I want to steal it for some unrelated work.. see you at the finish line

[Daniel Bailey]

"Any bets on a melt out to the pole, for 2012?"

Given that the ice thicknesses in the vicinity of the NP have been about sub-2 meters thicknesses of late, and given that melt-out typically is about 1.6-1.8 meters (IIRC), than I put it at 50/50.

With the vagaries of weather reigning fickly omnipotent.

[Kalle GZ]

3.0-3.2 Area 4.5-4.75 extent.

I don't see a big melt year coming for the ice. The Arctic overall looks healthier then the last 2 years.

However, those open water areas in the Canadian Archipelago, the Beaufort Sea, and the Laptev Sea are making me a bit scared, if they continue to grow huge and affect the shape of the Arctic I see a lower minimum then what I predicted.

[Daniel Bailey]

"The Arctic overall looks healthier then the last 2 years."
Is this the same Arctic the rest of us are watching? Or are you just Poe-ing us?

[Jim_pettit]

If really pushed, I'd suggest that SIA will be right at 7 million km2 by the first of July, 4 million km2 by the first of August, and 3 million km2 by the first of September. Add a drop from there of anywhere from 100k to 400k km2, and there's your record. SIE should, of course, follow a similar curve.

It fills me with a mix of awe, wonder, and dread that as soon as, say, five years from now we may be sitting here predicting a new SIA record of a few hundred thousand km2, or an SIE record of a million km2 or so. That is, if there's that much ice left by then...

[Apocalypse4Real]

I went with SIA of 2.8-3.0 and SIE of 4.0-4.25. I anticipate that current warming and SIA decline will accelerate.

The Hudson is under a torch right now, the Bering and Chukchi melt out seem on track (temps range from 7 to 19 C in the area at the moment).

It is currently 16 C at Pevek Airport, Russia on the East Siberian Sea, and there is no drop below freezing through 16 June.

Tiksi airport on the Laptev is expected to remain above freezing through 16 June.

[Mike Constable]

I see a slab of ice has broken off by Barrow, just leaves a narrow coastal strip with melt pools? Wonder if the break-up forecast for the fast-ice will start before it all goes!