Not sure if I should wait to see if NSIDC publishes updated Arctic sea ice age data for September (to check on a Goddard prediction).  But in the meantime, have a look, and let me know if there are any other specific Arctic sea ice predictions I should include.  I know SEARCH has a bunch, but I don't want the post to become unwieldy.  It's already fairly long.

Lessons from Past Climate Predictions: Arctic Sea Ice Extent

I had downloaded the volume data myself just a couple days ago, the record minimum from last year was achieved just under a month early this year.  Crazy stuff.

>>>Goddard's final revised prediction was made 15 months after tamino's, and was still far less accurate.

Eh, what?  Tamino posted his in July 2010, or so you said just before.  Goddard posted a prediction for 2010, this December?

Whoops!  I confused the 2010 and 2011 predictions.  Kept going back and forth.  Good catch Alex.

Also, Tamino had kept updating his prediction.  The most recent one was between 4.2 and 4.6 million square kilometers (4.4 best fit?), and he said that with Jaxa data from July and August being used as a predictor of September minimums, the minimum this year would be 4.17.

So, the early minimum threw him off a bit, but I think it would be more honest to at least mention the most updated prediction and how it was off by more than the older one.

Here it is.

Hmm is that worth mentioning?  He went from 4.63 to about 4.41, but the actual value was 4.53.  So really just off by 0.1 million too low rather than too high.  Seems kind of trivial.

Hi Dana1981, no need to wait for after September!  I haven't dug out the daily SSMI data that NSIDC uses, but you can see why there might be confusion... 

The points are daily, the divisions on the bottom axis are decimal fractions of a year, the divisions being weeks in this case, rounded down to two decimal places.  It's a quick graph.

I was talking about waiting in order to get Arctic sea ice age data (1-year, 2-year, etc. old ice).  Steve Goddard made a prediction that multi-year sea ice would increase from 2009 and 2010.  But I haven't found that data anywhere - it seems like NSIDC publishes it at random times, but they might in their October news update.

Dana: multi-year ice is even now still being flushed out of the central pack via Fram Strait and will continue to do so for some time.

Much of the current growth in ice extent is due to the winter expansion of the amount of ice hugging Greenland's east coast - ice from the main pack.

Baffin Bay, which - post 1818 - used to be accessible for only about two weeks in late August is still almost entirely open water as I write these words.

An earlier start to melt and a later start to significant refreezing does not bode well for the climate systems of the northern hemisphere.  We can expect a lot more weather-weirding.

Dana 1981, sorry, my mistake. I’ve given some recent detail on age of ice in The Canary in the Coal Mine (which is pending review!). To save you ploughing through the references, the one you are probably after is Maslanik 2011. The movie, which Jim Maslanik and Chuck Fowler kindly updated to 2010 for me, is also useful, but obviously doesn’t give the numbers directly

I like this - thumbs up

This might be a dumb question but others might be wondering as well: what's the significance of the numbers on the horizontal axis where it says 2011.62, 2011.64....? And why is there a 2011.75 but no 2011.76 (even though the distance between 2011.74 and 2011.75 is the same as the one between eg. 2011.72 and 2011.74?

On which figure, Baerbel?  On my figure it's the months of the year, but I couldn't figure out how to get Excel to properly display the months, so I just left them off.  I don't see any 2011.64, etc. though.

Sorry, in the graph Peter included in his post at 6:14am (should have specified that right away)

Oh well it's fractions of a year (i.e. 01 July 2011 is 2011.5).  The .75 is probably a typo - looks like it should be .76.

Erm, Yes it is fractions, but to give nearest intervals to weeks rounded down to 2 decimal places, mad eh? I had the same problem with Excel as Dana, but sorry to confuse the issue.

Please provide an explanation or excise it.

I didn't use the figure in question.  As noted above, my version doesn't have labels on the x-axis.

I'm going to update the post tonight to include Dikran's (sort of) predictions.  After-the-fact 'predictions' he made based on past data without first checking the results :-)

Dana1981, Apologies for diversion away from the point of your post.  Chart is snapshot of daily data from different Ice concentration algorithms through September, two using the same AMSR-E sensor data, JAXA and Bremen. Bremen is higher resolution and shows most difference in the melt season. The MAISIE data is a multi-sensor assimilation and is probably more accurate.  My point, poorly expressed, and with hindsight not helpful, was I'm not sure the skeptics appreciate that in the predictions we have to stick with one time series, usually the SSMI data from NSIDC, so we compare like with like, otherwise we could get 10% or so errors. Watts has commented that NSIDC should switch to AMSR-E for example, which would introduce a downwards step he would then no doubt criticise.  I'm happy to delete the chart altogether if you wish!

& post is good

Thanks Peter, no worries about your graph.

I actually just came to a new realization too.  Goddard's sea ice age claim was for September (end of summer) data, but the graph he used was of March (end of winter) data.  So actually his claim that sea ice thickness has increased every September since 2008 (a whopping 2 years) isn't even true - in fact there was a very large decline in September sea ice thickness from 2008 to 2010.  Thickness increased a bit in March, but who cares?  That's the end of the winter, and not what Goddard is talking about anyway.  He just plain screwed up.

I'll wait a few days to publish this post, because last October 4th, NSIDC updated the sea ice thickness data.  Hopefully they'll do the same this year, but I added a discussion of the data to totally demolish Goddard.  Not only are his predictions consistently wrong, but so are his claims about past Arctic sea ice data!  Goddard can't get anything right.

The other posts of the series featured (more or less) reputable scientists, I don't know if WUWT and Goddard deserve to be part of the "lessons from past climate predictions" series. This could well be a standalone post on arctic sea ice minimum forecasts.

In the discussion on thickness/volume, Goddard's claim is on the amount of multi-year ice but you reply with ice thickness (right after fig. 6). Also, you do not provide a graph or reference.

Riccardo - I view sea ice thickness and age as the same thing (if it's older, it's thicker).  But I can clarify that.  Like I said, I'm waiting for NSIDC to update the ice age data.

This post actually started out as an evaluation of Bastardi's prediction, who's not a reputable scientist either.  Ultimately the purpose is to see how predictions have fared and what we can learn from them.  So they don't necessarily need to be made by reputable scientists for us to learn from them.

You'll want to check out "Farewell to the Arctic — as We Know It" posted on Time.com before you finalize your article.

Ha, nice ending too. What a bunch of maroons "skeptics" are.

Just to re-iterate and clarify Riccardos point, one that sceptics may/will pick up on.  The relative and absolute extent of end year 2 and 3 yr ice did increase 2009/2010 (of course, after 2007 wiped out so much) but the overall thickness is estimated to have deceased.  Ice of all ages is on average thinning.  What appears to be happening is that the melt/reformation rates are changing, more taken away, less added, year on year.  One of the reasons for the rapid 2011 melt even in relatively benign (in terms of extent reducing factors other than temperature, such as wind) environment.  Check the (free) Maslanik 2011 reference above, fig 3, both March and Sept multiyear ice extent are shown.  Goddard has been rather selective, but I don't want you taking a hit on this.  However, his prediction on 5 yr ice looks highly dubious after the extensive 2011 overall melt, see same fig 3.   

looks good to me

Note my prediction for 2011 is what would have been predicted in October 2009 rather than October 2010, adding the datapoint for 2010, the 2011 and 2012 projections are as follows:

prediction for 2011 = 4.764138 (+/- 1.050547)
prediction for 2012 = 4.605450 (+/- 1.077168)

“Canada’s Arctic ice shelves, formations that date back thousands of years, have been almost halved in size over the last six years, Canadian researchers said on Tuesday.

“Researchers at Carleton University in Ottawa, who regularly analyze satellite images from the region, also found that a major portion of the ice shelves split in half this summer and other pieces covering an area roughly one and a half times that of Manhattan have broken off since the end of July.

“Consistently higher temperatures in Canada’s Arctic, the researchers said, were the main cause of the dramatic decline.”

Source: “Arctic Shelves Have Lost Half Their Size in Six Years,” New York Times, Sep 28, 2011

Just out: the NSIDC Sept update.  Go get the graphics for your post!

Thanks Peter.  They improved it to include 3-5 year old ice.  Good stuff.

Dana, where you say, "Since the total amount of ice (volume) has declined rapidly (more on this below), it's not necessarily accurate to claim that the amount of multi-year ice has increased (as Goddard has) just because the percentage may have increased."

The chart for the age of the ice is for ice coverage, so volume should be changed for "extent".

Dana,

Thanks for doing this-- important to expose the misinformation on this IMHO.

One small nit pick.  In Fig. 5, theposition of the blue and orange dots do not actually coincide with the minima on the relevant trace.  I'm assuming that the current locations coincidee with the climataological minimum?

You/we got into "troule" recently for digitizing a graph, how confident are you that the trace made by Bastardi is an accurate representation of what he drew?  Just making sure that the everything is as accurate as possible.  Pity that he did not state a specific extent-- but that is how they manage to give themselves "wiggle room".

In the eighties, more than 30% of the ice in the Arctic basin was more than 5 years old, now only ~5% is that old.  Good God, what a mess....

And the ice shelves are disappearing as John noted above.  I would encourage you to inlcude that (I know Luke he does good work) as it is relevant and a notable feedback from the reduced sea ice (i.e., increased wave action and warmer ocean temperatures taking their toll on the ice shelves).

Tom - good point, thanks.

Alby - the Bastardi reproduction is reasonably close.  He did state a specific extent (5.5 million sq km), which I noted after the graph.  So even if the graph isn't 100% accurate, the minimum extent prediction is.

I put the dots where the sea ice minimum is usually located.  This year was a bit odd.  But actually I think the offset makes the difference (between predictions and data) easier to see.

Will try to fit in the ice shelves somewhere...actually it makes for a decent ending.

Dana, I want to draw your attention to what I consider to be errors in this blog post that need correction

Prediction 1: "My forecast is that come the end of September, the amount of multi-year ice will again  increase relative to last year..."

Post:  Prediction not directly  adressed.

Outcome:   Multi-year (2 year plus) ice increased from approximately 2.29 million km^2 to 2.32 million km^2.

Discussion:  Although the prediction is not directly adressed, you talk around the point in a way giving the clear implication that Goddard is in error.  Indeed you state that he has made errors further on, which only makes sense if this is an error, which it was not.

Prediction 2: In 2013 there should be an increase in the amount of five year old ice, because that is when the 2008 ice will have aged five years.

Post:  Goddard's prediction that 5-year-old ice will increase in 2013 looks particularly ill-advised, considering its very steady decline, and the fact that it is rapidly approaching zero.

Potential Outcome:  This analysis ignores the fact that each year, 2 year ice of the previous season becomes 3 year ice of this season, and so on.  Therefore, the large percentage of 3 year ice in this season will become 5 year ice in 2013.  Not all ice is preserved like this, of course, with much being melted.  Based on the 2010/11 figures only, about 50% loss per year is a reasonable conservative estimate.  That means 1.08 million km^2 (the current 3 year plus ice extent) * 0.25, or approximately 0.25 million km^2 of 5 plus year ice will survive in 2013, compared to the current 0.18 km^2

The fly in the ointment for Goddard's prediction is the spur of old ice from the North Pole to Russia.  Similar spurs have formed recently in 1986 (dissipated 1989), 1996 (dissipated 2003),  and 2007 (current spur).  Thas is based on observing this video  

If that spur dissipates in either 2012 or 2013, Goddard's prediction will be in serious danger.

I adress these points in my post #14 under the blog post, but thought it better to draw your explicit attention to the issues here rather than in public.  If you have better figures than mine (which are based on pixel counts) which support your claims, please ignore.  Otherwise it would be better to correct the post.

Sorry I did not pick this up in the review process.

Yeah fair points Tom.  I made some revisions to that section.