# Arctic Forecast Verification Update

By Steve Goddard

On June 28, I generated a forecast projection for Arctic ice during the remainder of the summer, seen below.

So how is it doing so far? The image below shows that my forecast has been too conservative. The actual JAXA path (red) is above my forecast (dashed.)

Does this mean that I need to increase my forecast minimum to a higher number?

No, I don’t have any plans to do that, because I expect melt to occur faster during the remainder of the month. NCEP is forecasting warmer temperatures over the East Siberian and Chukchi Seas for the next two weeks.

We are already seeing evidence of this in the DMI graph of 30% concentration ice.

http://ocean.dmi.dk/arctic/icecover.uk.php

I expect to see that JAXA will move closer to my 5.5 million km² forecast for the summer minimum.

Article Rating
Inline Feedbacks
July 20, 2010 12:04 pm

Hey Steve!….are you “The Spy who came from the cold”? Brrrrrrrrrrrr!

phlogiston
July 20, 2010 12:35 pm

My guess is 5.65 M km^2

Dinostratus
July 20, 2010 1:18 pm

Anothony, how did you come up with your forcast? It seems that you are implying you have some model or at least a data based projection but I have not seen any equation. My ‘prediction’, i.e. 5,3001,000sqkm, is entirly data based. I made plots of dA/dt vs. A, looked at multiple years and determined that the average gradient, when the extend goes from 10e6sqkm to 8sqkm, correlates to the minimum. I didn’t propose a mechanism. I’m interested to see if you have some quantitative mechanism.

July 20, 2010 1:34 pm

Dino – maybe this prior post will help you understand his forecast method…
http://wattsupwiththat.com/2010/07/07/take-a-right-turn-to-ice-station-zero/

July 20, 2010 1:58 pm

I have a very scientific forecast :
Take all SEARCH june forecasts. Correct them using the same error ratio they demonstrate in 2009 june forecast.
The result is 5 627 000 km²
The details are available (in french) here : http://www.lepost.fr/article/2010/07/16/2154103_surface-de-la-banquise-une-approche-originale-pour-la-prevision.html
Anybody who want to PROVE that I am wrong will have to wait until mid september.

nevket240
July 20, 2010 2:05 pm

http://www.bloomberg.com/news/2010-07-20/arctic-ice-melting-fast-may-reach-all-time-low-this-year-russia-says.html
don’t you folks listen??? there is funding to be had…. scream and be rich..
regards

Pamela Gray
July 20, 2010 2:08 pm

PapyJako, statisticians would love your method. It forms an important component of statistical models.

geo
July 20, 2010 3:08 pm

I hope you are wrong, Steve. 🙂
Again, I’ll call anything above 5.5M km/2 a victory for my prognosticating skills, as that puts me (just joe sixpack on the internet, paying moderately close attention) closer to reality than the median guess of the panel of 16 experts. And actually, that’d be two years in a row, and thus a trend! Who knows, if I hit, maybe Dr. Walt will ask me to participate in next year’s panel. He certainly *should* ask you.

rbateman
July 20, 2010 4:19 pm

I am one who has said your 5.5M km2 is on the conservative side.

July 20, 2010 4:20 pm

Two cubes in mine Steve, and have one yourself. Thanks for dishing up the good cheer.

Charles Wilson
July 20, 2010 4:44 pm

Anyone else try looking at the DMI Website ? ?
Steve: Everyone will expect you Photoshopped out 2010 regaining 2nd Place BUT … I expanded your version of DMI’s Chart, Steve – – you have all the lines “Dashed”
— well, DMI don’t do dashed lines.
The tiny gap wipes out Today’s vertical drop.
MAYBe it’s an artifact of your image system. But if you consult JAXA data, you find 2010 just had a 100,000 day (first in weeks). Gee, even yesterday was higher than its been in weeks – -beat 2007 for the second day in a row. Guess we are in for some SUNSHINE. And Pips 2.0 Displacement Chart, shows the near-random Drift of little tiny arrows has gone to good-sized ones — and MAYBE is sorting out into a pattern similar to 2007’s. Or Maybe NOT.
http://www7320.nrlssc.navy.mil/pips2/archive/retrievepic.html?filetype=Displacement&year=2010&month=7&day=21
Note if you put 2007 into the web address instead of 2010, 2007 had a few days wandering, but the next 8 are Heavy Melts.
Don’t sweat the Daily Up & Down, Steve – – it’s the WEATHER.
PS: If You are Me, of course, with the POSSIBLE DOOMSDAY FORECAST
… being right or Wrong might be FATAL – – so I always hope I’m wrong.
Current Odds I give for 300 mph winds … are 10% and dropping every day.
But I still sweat. A lot.

Jhwa
July 20, 2010 4:45 pm

Have you looked at the buoy temps?
07/20/1800Z 86.516°N 0.212°E -13.4°C 1011.3mb
07/20/1700Z 86.518°N 0.228°E -13.0°C 1010.9mb
07/20/1600Z 86.520°N 0.243°E -13.8°C 1010.1mb
07/20/1400Z 86.526°N 0.255°E -12.6°C 1012.6mb
07/20/1300Z 86.528°N 0.249°E -11.5°C 1012.5mb
OMG
JHwa

Gail Combs
July 20, 2010 5:04 pm

Jhwa says:
July 20, 2010 at 4:45 pm
Have you looked at the buoy temps?
07/20/1800Z 86.516°N 0.212°E -13.4°C 1011.3mb
______________________________________________________________
Jhwa, is that a dash 13.4°C or is it negative 13.4°C?, yeah I know a dumb question but please answer anyway.

Jhwa
July 20, 2010 5:13 pm

It looked to me as a -XX.XDeg C

Jhwa
July 20, 2010 5:14 pm

Is far as i’m concerned there is no such thing as a dumb question.

kwik
July 20, 2010 5:48 pm

Steven Goddard, I am impressed, to say the least, of your work.
Is it a coincidence that your surname is the same as a certain rocket engineer who was ridiculed for so long by, was it the NYT?
Anyway, hopefully it will show young, inspired, wannabe scientists, how to approach a “problem”, instead of just being a parrot. I dont like parrots.

Daniel M
July 20, 2010 5:58 pm

Let’s hope that you’re estimate turns out to be conservative, Steve. We can only imagine the cries from the warmistas if we approached or shot past the 2007 low, no doubt accompanied by a flood of tears for the poor polar bears.
Meanwhile, in the Southern Hemisphere…
Seems like global cooling is killing the unfortunately less adorable penguin: “Scientists are investigating whether strong currents and colder-than-normal waters have hurt populations of the species that make up the penguins’ diet”

Mike Maxwell
July 20, 2010 6:48 pm
Mike Maxwell
July 20, 2010 6:49 pm

…err, that hyperlink was supposed to be titled “Paul the octopus”

ianpp
July 20, 2010 6:55 pm

Paul the psychic octupus predicts 5,602,287.23705km2 +/- 18 km2.

Frederick Michael
July 20, 2010 8:36 pm

The preliminary JAXA figures for 7/20/10 show 132k sq-km lost in a single day. 2009 had 5 days in a row like this starting on 7/21/09.

FergalR
July 20, 2010 8:50 pm

PIOMAS volume anomaly shows a slight uptick: http://psc.apl.washington.edu/ArcticSeaiceVolume/IceVolume.php

David W
July 20, 2010 11:44 pm

Frederick Michael says:
July 20, 2010 at 8:36 pm
“The preliminary JAXA figures for 7/20/10 show 132k sq-km lost in a single day. 2009 had 5 days in a row like this starting on 7/21/09.”
But 2009 didnt have the extent of early season ice loss that 2010 has experienced. It was still losing ice in Hudson Bay during the latter part of July, although the July ice loss in the East Siberian and Kara seas are probably looking fairly similar.
Much will depend in the next 7 days on whether the rate of loss in those areas matches what happened during the same period last year. If so I think we will end the year with a similar extent to 2009.
I dont think the ice loss is likely to exceed what happened last year for this period but I think there’s still a possibility the loss could be slightly lower given the Hudson Bay ice is already gone.

Perry
July 21, 2010 1:52 am

“The ice may melt fully in summer and freeze in winter,” Frolov said. “This is one possible scenario.”
To contact the reporters on this story: Maria Kolesnikova in Moscow at mkolesnikova@bloomberg.net,
What other scenarios spring to mind? Ice freezing in summer & melting in winter?

Mooloo
July 21, 2010 3:08 am

What other scenarios spring to mind?
Actually I think he might be putting the idea that there will be no winter ice into the minds of readers. Which would just be ridiculous.
As you say Perry, his “possible scenario” is the only possible alternative to what we have now.

Alexej Buergin
July 21, 2010 3:25 am

” ianpp says:
July 20, 2010 at 6:55 pm
Paul the psychic octupus predicts 5,602,287.23705km2 +/- 18 km2.”
I understand your joke, but if the standard deviation is 18 km^2, Paul would have left out anything to the right of the decimal point: 5 602 287 +/- 18

Frederick Michael
July 21, 2010 3:26 am

David W says:
July 20, 2010 at 11:44 pm
Much will depend in the next 7 days on whether the rate of loss in those areas matches what happened during the same period last year. If so I think we will end the year with a similar extent to 2009.
I don’t think the ice loss is likely to exceed what happened last year for this period but I think there’s still a possibility the loss could be slightly lower given the Hudson Bay ice is already gone.

If the next 7 days match the ice loss of 2007 (but without the melt coming from the Hudson Bay), the Arctic Ice is in big trouble. We simply should not be getting this much melting. This is encouraging though:
http://ocean.dmi.dk/arctic/icecover.uk.php

Casper
July 21, 2010 5:32 am

I also updated my forecast. The minimum should occurs on 20th August. Its value is 6708821 km^2.

Charles Wilson
July 21, 2010 5:57 am

JAXA Data: at
Comparing _______2007___ to___ 2010_______&___________2009
Ahead June 28______ no________ 679,531 Square Kilometers
Ahead July 17____ 458,437________ no __
Daily Loss:________2007_________2010__
July17-18 ______ – 71,406 _____ – 81,562
July18-19 ______ – 63,907 _____ – 108,906__ (was 100,000 preliminary)
July19-20 ______ – 92,187 _____ – 131,719preliminary
July20-21 ______ – 103,438 _____ ? ?
July21-22 ______ – 100,250 ____
The weird thing is the last month Preliminaries were corrected down, about half the time — Yesterday’s was the first time I’ve seen one corrected UP.
Sats show skies clearing — but for how long? Again, cold air = La Nina = SUNNY = more Melt.
See Mr. Watts new Post on the New La Nina.

July 21, 2010 8:00 am

Frederick Michael
You can take a deep breath. Even if all of the remaining ice melted in the East Siberian Sea, that is only half a million km^2, which doesn’t change anything dramatically. My forecast assumed that would happen.

EFS_Junior
July 21, 2010 9:02 am

Well my latest forecast is 4.44E6 km^2 (standard deviation = 0.38E6 km^2), so it’s definitely less likely (than more likely) that 2010 will set a new low versus 2007.
The new PIOMAS numbers are out there (in graphical form) as of yesterday;
Sea ice volume anomaly is -10,600 km^3, 1979-2009 average volume for this time of year is 22,800 km^3.
Thickness using JAXA data suggests an average thickness of 1.5 meters.
The September PIOMAS volume data (minima) should be quite entertaining, could be as low as 7,000 km^2 based on current trend lines.
The paint appears to be drying a bit faster at the moment.
Been staring at the paint too much lately via MODIS. The Arctic had been very cloudy over the past 2-3 weeks, currently it appears to be less cloudy, don’t have a clue if this will continue though.

EFS_Junior
July 21, 2010 9:05 am

Correction: 7,000 km^2 should read 7,000 km^3 in previous post.

Roger Knights
July 21, 2010 12:28 pm

EFS_Junior says:
July 21, 2010 at 9:02 am
Well my latest forecast is 4.44E6 km^2 (standard deviation = 0.38E6 km^2), so it’s definitely less likely (than more likely) that 2010 will set a new low versus 2007.

Then there’s free money for the taking at http://www.intrade.com where \$55 will get you \$100 if 2010 is lower than 2009 (which was about 5.4 million square km). That should be a no-brainer.

Pascvaks
July 21, 2010 1:27 pm

Steve
It’s the Economy!
PS: Now if the graph goes ‘Flat Line’ at say 148×10^6km2, that means Mann & Co. were wrong, right;-)

Dinostratus
July 21, 2010 5:46 pm

Ken Finney – Well I reread it. There’s mention of this, mention of that, etc. etc. but nothing that says something like, “Ice thickness is 120% of normal right now and the average wind speed is 30kmph CCW at 80deg so I think the summer extent will be 105% of the average based upon this trend of X vs. Y.”
By providing sources then giving a number there is the implication that the sources have a relationship to the number. Yet, I don’t see anything specific.
Can someone point me in the right direction?

David W
July 21, 2010 10:02 pm

Frederick Michael says:
July 21, 2010 at 3:26 am
“If the next 7 days match the ice loss of 2007 (but without the melt coming from the Hudson Bay), the Arctic Ice is in big trouble. We simply should not be getting this much melting….”
I would think the chances are very low. More likely we will see a short upward trend in the level of ice loss and then the level of loss will reduce through August. I’m now very confident we will either match or exceed the 2009 minimum extent (more ice – less ice loss).
The ice loss for 21st July was back down to 60,000 sq km btw. Cryosphere Today is showing the East Siberian Sea about 200,000 sq km above where it was at this time last year. Even if that gap was to disappear by the end of the month it would still only add 20,000 sq km a day.
Looking at Cryosphere Today on a region by region basis I simply cannot see any thing pointing to the likelihood of the ice loss returning to a 80-100k per day running average. Both the Chuckchi and the Kara Seas have little ice left to lose and the Arctic Basin shows no indication of repeating the 2007 rate of loss.

July 22, 2010 2:11 am

FergalR said on July 20, 2010 at 8:50 pm:

PIOMAS volume anomaly shows a slight uptick: http://psc.apl.washington.edu/ArcticSeaiceVolume/IceVolume.php

That’s not an up-tick. Apparently that was someone finally noticing that the unprecedented long dead-straight drop was an error, blowing off the posted 3-5 day update schedule for over a month (previous update was June 14), then “fixing” the problem. The previous graph ended at about -10.7 anomaly units (by eyeball). Between June 14 and July 17, according to this new graph, the volume continued the dead-straight drop for two weeks ending at -11.2, then suddenly reversed course and shot back up to the June 14 amount in two weeks. Yup, in the middle of the melt season the volume suddenly started increasing apparently as fast as it was previously decreasing. Extent was dropping while volume rose.
Or maybe not. The more I try to think about that chart, the more I think it yields nothing useful. Notice this from that linked main PIOMAS page:

Anomalies for each day are calculated relative to the average over the 1979 -2009 period for that day to remove the annual cycle.

So on June 14 the graph was showing -10,700 km^3 relative to that day. By the average X volume was expected on that day, Y volume was calculated instead, the difference is what goes on the graph. Thus the graph says nothing about the actual volume, only the daily variations from the daily averages.
Imagine how that would work in business accounting. According to the averages derived from the previous thirty years of data, for the first week of July the company’s total assets should be \$10 million. For 2010, total assets were only \$9.5 million. It does not matter if in the last week of June the company had only \$9.3 million thus it had gained \$200,000 from the one week to the next, what gets reported is a half million loss for that first week of July. Can you imagine any company doing that? Can you imagine the resulting graph showing anything useful about the long-term health of the company?
And where did they get those daily volume numbers for the 1979 to 2009 period? PIOMAS. PIOMAS said what it thought the volume should be on each of those days, now it says what the daily difference is from the average of the numbers it calculated.
GIGOGG – Garbage In, Garbage Out, Garbage Graphed.

EFS_Junior
July 22, 2010 8:53 am

PIOMAS is an anomaly graph, not an ice volume graph per se.
Also, no data has changed from the June 17, 2010 update, two new data pointe were added though.
The last three data points are as follows;
Timeframe, anomaly (km^3), volume (km^3)
mid-June,-10.70K,16.25K
June-July,-11.20K,14.01K
mid-July, -10.63K,12.195K
Volume will continue to decrease to the annual September minima of 13.63K (1979-2009 average), which when coupled with the current large anomaly, if it continues alongthe current trend line, could go as low as 7.0K, or ~1/2 the 1979-2009 average minima.

July 22, 2010 7:17 pm

From: EFS_Junior on July 22, 2010 at 8:53 am

PIOMAS is an anomaly graph, not an ice volume graph per se.

Yup.

Also, no data has changed from the June 17, 2010 update, two new data points were added though.

June 14 actually. And with a posted 3-5 day update schedule, to cover over a month they added only two data points? Since the graph only shows lines without points, are you sure you haven’t missed a few?

The last three data points are as follows;
Timeframe, anomaly (km^3), volume (km^3)
mid-June,-10.70K,16.25K
June-July,-11.20K,14.01K
mid-July, -10.63K,12.195K

Well, the daily anomalies are calculated relative to a certain day. Care to share how you obtained the volume numbers for those specific points?

Casper
July 23, 2010 4:40 am

Steve,
by comparing the curves from last years, have you noticed a significat change in their progression? I think it is not due to sun high…

Kevin Stanley
July 26, 2010 1:42 pm

Anomaly is really a very useful way to think about this type of data. You know that mass will be lost as spring turns to summer, but you don’t care about the normal amount of loss. So, for each observation you are essentially calculating the difference between what’s normal for that point in the annual cycle and what’s observed in this specific instance.
So an uptick in anomaly would happen any time melt slows down sufficiently, which will obviously happen from time to time as the weather changes.
I have to strongly disagree with your criticism of the PIOMAS anomaly data in particular, which reads like a criticism of using anomaly data in general:
“Yup, in the middle of the melt season the volume suddenly started increasing apparently as fast as it was previously decreasing. Extent was dropping while volume rose.
Or maybe not. The more I try to think about that chart, the more I think it yields nothing useful. Notice this from that linked main PIOMAS page:
Anomalies for each day are calculated relative to the average over the 1979 -2009 period for that day to remove the annual cycle.
So on June 14 the graph was showing -10,700 km^3 relative to that day. By the average X volume was expected on that day, Y volume was calculated instead, the difference is what goes on the graph. Thus the graph says nothing about the actual volume, only the daily variations from the daily averages.”