« NSIDC Arctic sea ice news June 2012 | Main | Stronghold »


Feed You can follow this conversation by subscribing to the comment feed for this post.


Following that graph, by mid-July we'll have passed all volume minima from before 2007...


Wipneus, can you just remind us - does your "expected" curve assimilate new data each month, or is that a fixed curve calculated at some point in the past.

Does the fact that we are currently tracking just a little below the projection force that projection lower than the curve displayed on last months graph?



2012 "expected" points on any date are calculated from (and only from) the 1979-2011 data for the *same* date. Like the September minimum curve, but for other dates of the year.

So the expected curves are fixed from month to month, unless the people that actually produce the data file make a change to old data.

Hope that I was clear.

James Benison

This is not simply bad.

It really is an unqualified disaster.

Al Rodger

A closer look at the PIOMAS anomaly graph is probably in order. Is that a minimum anomaly for the year at Day #174 ? It could well be.
Link to graph here (usually 2 clicks to 'download your attachment'.
(I'm not long messing around with this graph of annual anomalies. It will probably evolve into something at MARCLIMATEGRAPHS in coming months.)

Surprised not to see anybody else saying it - If 2012 now parallels 2010, the summer minimum PIOMAS volume will be 2,700 cu km. Assuming PIOMAS is not grossly exagerating ice loss, it is difficult to see the Arctic Sea Ice coping with many more of these new minima in volume each summer without a big crash in Area & Extent.

Seke Rob

@ 2,700 Km^3 there is not doubt the old ice factories have to be fired up again. The positive [in Wattsian logic], the more melts, the more *surface* that can refreez... Stand back for Watts/Goddard et al reporting on absolute records of shiny surface to have formed in the 2012/2013 autumn/winter season, a say "see, we knew there was nothing wrong... all normal".

Account Deleted

Seke Rob, skeptics have little optimism. They predict a second place of the September minimum.

July sea ice outlook submitted to ARCUS
Posted on July 5, 2012 by Anthony Watts
The WUWT Extent Projection: 4.55 million square kilometers.


Though I haven’t posted any comments on june PIOMAS yet, the thought is ‘always’ in my mind. As I imagined this morning, 11000 m3 / 8,5MK = 1,29 m, and given a daily melt of 1 cm, the rest of this melt season should produce 63CM(days) of extra melt, overall mean.
With a 3.7MK minimum extent that would stand for 3,7 x 0,66 = 2442 km3 at the minimum.
What will that look like? And what formidable, pattern-disrupting heat release would that produce next autumn, before SekeRob’s glistering winter sheet appears?
If it appears…

Seke Rob

AIL80, Cant's go there... on my blocked IP/address list [always have an instant "brain going to mush" feel going there and the zombies racing in with their straws to suck the cranium empty]. From what I remember, their collective prediction [collective lying to the collective self] was still *only* 3rd highest on SEARCH... is that what you were saying? http://www.arcus.org/search/seaiceoutlook/2012/june. Not seen the July report yet (next week I suppose).

A different take on PIOMAS, planning a seasonal and boxcar too. The present all time max anomaly occurred a week ago, the minimum anomaly pretty much matched 2011, and happened mid Jan, when we were about closest to the sun. The cumulative melt line will have to wait to somewhere in September, might even go into October.


Seke Rob

Neven, glistering or glistening... the intend of the word close enough ;?

And on my previous post, of course the maximum anomaly is *till* day 181 of 2012 :(.

Seke Rob

Oops, that was Werther, not Neven. Mea Culpa.


One thing bothers me about PIOMAS plots: Their average always includes all data from 1979 to the year before. Why don't they choose a base period of 1979-2000 like NSIDC does with extents? I also can't see how their standard deviations can hold up; they've got both 2010 and 2011 more than 2σ below the mean for all of both years. Since there is only ~5% probability that any given measurement will fall outside of 2σ, how are their standard deviations getting huge?

Nightvid Cole


I'm not sure why they don't fix their base period as 1979-2000 or whatever. But as far as standard deviation is concerned, it is simply calculated the usual way (i.e. square root of the mean squared deviation). The fact that there are >2σ deviations indicates only that the distribution is highly likely to be non-Gaussian, not that they made a mistake in the calculation.

Espen Olsen

Breaking News:

O-Buoy #4 Project Web cam is now very close to open water!



Tim & Nightvid/

Aren't these huge swings we have seen lately on PIOMAS-graphs there because the annual cycle itself has changed? I remember Tamino discussing something similar.

Plotting anomalies doesn't remove the annual cycle but the average (1979-2011) annual cycle. That means the anomaly curve will very likely go back up and be within 2sigma soon again.


See this discussion on Taminos blog: http://tamino.wordpress.com/2011/06/23/sea-ice-3-d/#comment-51789

Wayne Kernochan

@Neven: OK, this is maybe your second-best post :) I've been dying for this kind of volume/thickness analysis. It's just great.

Since, of course, we're never satisfied, I would be interested in the yearly figures on *thickness at minimum area*. This would remove the decrease in thickness due to new ice during the refreeze, more or less, and (until area really takes a nose-dive) tell us how close we are at minimum to a full melt.

@Tim: using the average of all the years before this one bothered me as well, until I sat down and thought about it. It does tend to understate the difference from 1979-2000, but I think it's more appropriate for volume, which has been decreasing during most years since 1979 much more than area or extent (at least until now :)). The issue with area and extent until relatively recently has been more "is this going down"; with volume, it's "how fast is it going down", so a "cumulative average" understates the anomaly today in some ways but also makes the fact the downward trend started in the 1980s data more obvious.

Seke Rob

Set the chart up last year to see if I could reproduce, which took a while, but mine turned out identical. With that also set it up so that the I could enter any period end for the base line, and the picture remains the same, except the anomaly itself gets bigger, the more the base line period is shortened.

For day 181 (last avail.date)

-10,538 base 1979-2011
-10,834 base 1979-2010
-11,316 base 1979-2008 (the ever popular 30 years of climate world)
-12,615 base 1979-2000

The illustrious Dr. Roy has a different trick up the sleeve to make it appear to the ignorati, too clueless to know what they're looking at. He uses a sliding 30 year baseline [now 1981-2010, and surprised he's not moved yet to 1982-2011], which helps to make the present anomaly show smallest possible, whilst home free not to get finger pointed.

Seke Rob

On my last bit, this is not to infer that PIOMAS is tricking. As Wayne notes [saw entry only after posting], their method is more appropriate and with the uncertainties, probably prudent.

Myself, longing for a POIMAS-Cryosat-2 comparative write up since all their calibration work. Would cause some red steaming ears in places.


Wayne: I would be interested in the yearly figures on *thickness at minimum area*

Maybe you like this one. It is minimum volume/area plotted against volume:


The interesting point is that ice thinning rate is accelerating after 2009. Only two data points, but yet.


Oops that was extent, this is area:



Thanks for the earlier explanation, Wipneus, but I wasn't very clear. The question I was posing is whether:
1. the 2010 and 2011 "expected" points are actual projections on the data to that point (1979-2009 and 1979-2010 respectively), or
2. they are earlier points on the same curve that is generating the projection for 2012 (using 1979-2011 data).

Differences between expected and actual highlight different things depending on which you are doing - either the predictive value of this model (however "naive"), or the quality of the fit to past data of the exponential curve you are using. I'm wondering which it is.

With regard to the two new graphs, I bet you could get a better fit on that trend than linear... ;-) ...especially when passing through the origin (0,0) is mandatory - after all, if volume is 0, volume / extent must also equal 0...




It is 1, I am interested in the prediction one year ahead.

Your last point is incorrect. Volume and extent/area can both "go to" zero but still have finite ratio.


Volume 30 Jun 11.163

Largest decline in volume from 30 June to minimum 11.317. That is 1981 and recent declines have been smaller. So not very comparable really but thought I would throw it in to give you a fright.

My prediction of the 30 June to minimum decline for 2012 based on regression with volume and volume residual from gompertz fit of 30 June volumes gives 8.67 K Km^3 decline.

11.16-8.67 = about 2.5 K Km^3 but there is still pretty large error bars. Compared to previous record low of 4.017, that is down 38% !

On top of this, early July weather is likely enhancing volume decline.

It is pretty remote that the ice will practically all go this year but if the volume is about 1K Km^3 less than PIOMAS says and we get a 2 sigma weather and other unknown factors towards more volume decline that also increases albedo feedback then you can perhaps see a 1 or 2 % possibility of it all going this year.

Even if we end up at 3K Km^3 per Wipneus' prediction, the probability in each future year seems likely to increase noticably. (Even if the volume returns to the same maximum volume as this year, a higher proportion is FYI and we still seem to be trending to more areas melting out sooner increasing the albedo feedback.)

My prediction of CT area 1 day minimum is around 2.6M Km^2. That doesn't necessarily give average thickness of less than 1m because the dates of minimums are different. Though presumably volume will start increasing earlier from a lower minimum to bring the minimum dates more towards each other.


Hello, I was wondering if I can have your opinions on a matter of some considerable importance.

The U.K.'s parlimentary Environmental Audit Committee chaired by several MP's is currently examaning the need for protecting the arctic.

Following the presenation of the PIOMAS data to the committe by Prof. Peter Waddhams of Cambridge Unviersity, in which the high possibility of an ice free arctic by 2015 was discussed, the committee asked Prof. Julia Slingo (head of the U.K. Met office, which rus the Hadley center) for her opinion on the matter, here is her reponse:

“Our assessment, based on the latest climate model simulations that have been performed for the Fifth Assessment Report, suggests some time between 2030 and 2080. Our own model would say between 2040 and 2060, and it is fair to say that our view is that the earlier dates-in other words, the more pessimistic outlook for the Arctic-are associated with models that we believe are more credible, in terms of their capability to reproduce the observed seasonal cycle in sea ice extent, and also the variations in sea ice from year to year. Our expectation is certainly not in the next few years, as I think you have heard from some evidence, but within, say, 2025 to 2030 would be the earliest date.”

I personally find this quite remarkable, she is using the output of models which clearly don’t fit the empirical data of the sudden loss of ice to dismiss our best empirical measurements.

Perhaps more importantly, Prof. Waddhams finds it remarkable too; he wrote to the committee following Prof. Slingo’s testimony to state is astonishment at her remarks:
“The data that Prof. Slingo rejected are part of PIOMAS, which is held in high regard, not only by me, but also by many experts in the field. From my position of somebody who has studied the Arctic for many years and has been actively participating in submarine measurements of the Arctic ice thickness since 1976, it seems extraordinary to me that for Prof. Slingo can effectively rule out these PIOMAS data in her consideration of the evidence for decreasing ice volume, when one considers the vast effort and diligence that has been invested over such an extended period in collecting data under the ice by both British and US scientists. Prof. Slingo offers no reason whatsoever for dismissing this extremely pertinent set of measurements and their associated interpretation, arguing that "the observational estimates are still very uncertain". This is not the case.”
A copy of his letter can be found here: http://www.publications.parliament.uk/pa/cm201012/cmselect/cmenvaud/writev/1739/arc26.htm

What do you all make of this? Am I missing somethig important here? Why would Prof. Sligo tell the U.K. parliment that the PIOMAS data isn't credible but that the climate models are? Is this really the best scientific advice from the head of an august institution?

If you agree that the advice is missleading then what could be done to remedy the situation?

Thanks in advance for taking the time to reply.

Robert Greer


I'm similarly perplexed. When I say that the Arctic could be ice-free in just a few months, people who don't closely follow the region give me a pitied look, as if I've succumbed to some kind of hysteria. But the most sensible projections for an ice-free Arctic (at least seasonally) are no more than three or four years away.

What's worse, these sensible projections are nearly-invariably undercut by the hard data coming in every year. When one compares the grim models with the even-grimmer data, then compares THOSE with broad theoretical explanations of the accelerating trend (e.g., albedo or methane release feedbacks), it seems entirely reasonable to say that the Arctic ice cap could melt entirely as soon as this year.

I believe this: the Arctic will be ice-free sometime before October of this year. I defy anyone to tell me this is impossible or even unreasonable given the state of the ice, its trajectory, and the theorized causes of its accelerating demise.

Someone in another thread (I believe it was crandles) asked why we weren't focusing on what we would do in the event of a complete ice collapse in the Arctic. I don't understand why there hasn't been a chorus of affirmations to this sentiment.

Kevin McKinney

Robert--Essentially this conversation was had over at RC about 2 months back. Quite a few researchers with solid track records still think that the sorts of dates that Prof. Slingo thinks most probable are the best ones to bet on. The RC post on that topic essentially said that statistical projections are in general less reliable than physical modeling; and fitting a extrapolative curve forward from PIOMAS is definitely in the category of statistical projection. So it's not that PIOMAS is not to be trusted; it's that in this view, physical modeling is likely to prove a better 'prophet'.

Of course it's also true that GCMs have consistently under-predicted the observed decline in sea ice--and it's true that Dr. Maslowski's projection is based to some degree upon *regional scale modeling*. He argues that GCMs can't resolve oceanic heat flows with sufficient detail, and that is why they suffer from a 'prediction gap.'

FWIW, I'd bet on Maslowski at this point, and I think your prediction of an ice-free Arctic this melt season is very unlikely to come true--though it looks to me as if the table has been set for another potential record low this year.

If it happens, I really hope it serves as (yet another) wake-up call to the seriousness of the situation we have created for ourselves.

Patrice Monroe Pustavrh

@Kevin: Am I correct that Maslowski's prediction for nearly ice free summer Arctic sea is 2016+/-3 years ? That would mean he is correct if Arctic is summer ice free as 'late' 2019. This is actually quite close to predictions for 2025 and actually, very close to "linear rate" decrease from PIOMAS.
On the other hand, I personally think that even prediction for ice free Arctic in 2040 or even 2060 or even the "outdated" 2100 is extremely alarming. We must be aware of this.

Eli Rabett

It is an old tradition in forecasting that when your model is in error consistently in the same direction, you apply a statistically derived correction to it (e.g. if your model was always 1 unit low, for forecasting you discount by one unit).

This neatly captures the difference between trend analysis and modeling in purpose. The former is a pure forecast, the later seeks to understand the system, the two together provide the superior forecast because each captures things that the other misses.

It looks like Arctic ice is one of those cases.

Oh yeah, people may be using different definitions of ice free:)

Rob Dekker

Regarding the dispute between Dr.Slingo and Dr. Wadhams, there are two issues which seem unresolved :
(1) Which model(s) did Dr. Slingo use to assess the development of average thickness of ice in the Arctic over the past few decades, and
(2) do both parties agree on the definition of an "ice free" Arctic, and if so, what is it ?

Rob Dekker


Yes, as far as I know, Maslowski projected 2016 +/- 3 yrs as the likely first 'ice-free' Arctic during minimum extent, with 'ice-free' defined as 1 million km^2 or less.


Note that this matches Gore's projections of an "ice free" Arctic by "as easly as 2013".

Personally, I have put my bets on a 3 million km^2 minimum extent by September 2016 :

Either way, these projections make the IPCC look like "deniers", but then again, the IPCC projects a minimum of some 6.5 million km^2 this year,
while self-proclaimed 'skeptic' blog WUWT just put in their best estimate of 4.55 million km^2.

Does this make any sense to anyone ?


This is actually quite close to predictions for 2025 and actually, very close to "linear rate" decrease from PIOMAS

"Linear PIOMAS" extrapolates to "zero ice" in about 2034. 2019 is about 4 sigma out of that estimate:


Of course the graph is shouting loudly "this is not linear, ignore any predictions made here".

Kevin McKinney

"On the other hand, I personally think that even prediction for ice free Arctic in 2040 or even 2060 or even the "outdated" 2100 is extremely alarming."

Roger that, Patrice.

Al Rodger


You ask what we make of the Stingo/Waddhams interchange.

I would say that the 'scattergun' letter from Waddhams will carry little weight. He should have stuck to one criticism (eg Stingo's ignoring of PIOMAS).

For me, the interesting points in Stingo's evidence were (a) Talking as "we" all the time. Is this because she is not totally in agreement with "we"? Or because "we" are all in total agreement? (b) Her own recent findings of volume loss not being compatable with the 75% PIOMAS figure. So what are her findings? Is this from the models that still cannot force an ice-free summer before 2040 while Stingo is happy to offer 2025 for such an event? (c) Her assertion "...there is still a lot of multi-year ice..." Surely the reason why the summer ice volume is down 75% is because we know Area is down quite a bit, we know FYI is somewhat thinner and, most importantly, we know the old thick MYI has pretty much disappeared.
The basis for the PIOMAS finding isn't solely some artifact of a model. Its summer minimums can be pretty much reproduced using Ice Age on the back of a fag packet.

But then I have long wrestled with the mismatch between the current scientific projections of Arctic Ice Area loss & the evidence. I presently explain it to myself as 'scientfic inertia', a phenomenon with a shorter timescale than scientific denial. But the speed of decline of Arctic Ice will probably be quick enough to outpace any changes in the accepted scientific wisdom, that is unless some mechanism arises very soon to slow the decline in the ice.

Espen Olsen

Ice quake at cam # 4:



I didn't know whether to be more surprised by the lack of logic on offer, or that they weren't called on it by a committee that included several lawyers.

For the best example, let's just consider the two premises offered by Dr Slingo:
P1: extent is decling quickly
P2: thickness is declining (by an uncertain amount).

Now you don't have to be Aristotle to figure out that there is only one conclusion you can draw from those premises: volume is declining faster than extent (by an uncertain amount).

Dr Slingo's facts only support the contention that summer ice volume is declining by some unknown amount greater than 12% per decade. Dr Wadhams offer of the PIOMAS-derived figure of 75% over 30 years (37% per decade) is not ruled out by this data.

So how in the name of Nansen's pants did this get passed off as rebutting Wadhams argument, and how did the comittee accept it as such?

I'm not the sharpest tool in the shed, and I saw that hole before I reached the end of Dr Slingos first sentence...(Oh, and she wasn't Robinson Crusoe when it came to logic bypasses, either).

The whole exchange just left me facepalming...

Doug Bostrom

Rob Dekker: Either way, these projections make the IPCC look like "deniers", but then again, the IPCC projects a minimum of some 6.5 million km^2 this year...while self-proclaimed 'skeptic' blog WUWT just put in their best estimate of 4.55 million km^2.

An excellent point-- "alarmist" shoe on the other foot.

If WUWT's prediction is below that of IPCC for year after year, what are we to conclude?

Seke Rob

Whilst doing a PIOMAS seasonal volume,


it only now struck me as there being something counter intuitive of PIOMAS volume versus sea ice extent [not looked at area yet]. Generally extent reaches its peak in March and volume reaches its peak in April. Any ideas why that is?

Did not see a plot of day in year maximum, so added that to bottom... no trend, but that was obvious already from the various daily volume plots.

Christoffer Ladstein

Extent is naturally stopped by the spring gearing up in the more southern latitudes,thus nibbling at the edges, volume continues to build further north where the heat of the spring is not particularly present until may. Some of the same dynamics are behind the summer and winter months, the darkest month isn't necesseraly the coldest...

Seke Rob

Thanks... that was a right before me eyes 101.

Christoffer Ladstein

Ha! Seke Rob, you're the wizard(one of many!)in here, no doubt about that, keep up the good steam, just as far as you don't spill the excess heat into the Arctic;-) ...

Aaron Lewis

For Climate Science to admit that Arctic sea ice is near its end, would mean that they have to admit that their research and peer review process was flawed. Since sea ice affects other aspects of the global circulation models, it means that all of their estimates of global warming are understated - in particular, they likely miss the speed of sea level rise, and failed to consider the possibility of "sea level rise events."

I think "Seasonal to decadal scale variations in the surface velocity of Jakobshavn Isbrae" doi:10.1029/2011JF002110 shows that the changes and impacts of current sea ice loss are not being taken seriously be the climate science community.

With loss of sea ice, the GIS will be the cold sink for the Arctic, and that will change NH atmospheric circulation.

Kevin McKinney

"For Climate Science to admit that Arctic sea ice is near its end, would mean that they have to admit that their research and peer review process was flawed."

I've seen no indication that many working scientists think anything else. After all, why would they work to replicate research if they thought it was already "flawless?"

And as for peer review, remember the Energy and Environment affair, which sparked the resignation of half the E & E editorial board due to flawed peer review? That's just one (albeit exceptionally egregious) example of dodgy peer review; others are not hard to find--nor to find acknowledged, either.


Aaron: I saw the Ice Crash happen in 2007, reading news and blogs eagerly to see if the models were questionned. They weren't. Soon after Maslowski issued his famous interview at AGU, "it might happen as soon as 2013". He really wasn't taken seriously. It's probably because it's so frightening to think that the prediction of Arctic thawing by 2100 would be erroneus and happening much, much faster.

SEARCH and PIOMAS were the only new things appearing after 2007, but no new model happened to really take this event in account.

So I'm quite pessimistic about seeing new models, going as fast as reality - for arctic ice, Greenland thawing, Larsen iceshelf in Antarctica, and if course temperatures.

Lawrence Jones

Aaron, etc: “…their research and peer review process was flawed.”

Not fair. Seems to me that climate scientists are appropriately reticent to forecast possible or plausible outcomes. When faced with anomalous data (e.g., Arctic Sea thawing faster than GCM suggest), a scientist will gather more data and investigate causes before taking a position. Protecting the public is not the purpose of science.

GCMs cannot back-model abrupt climate changes in the geological record (because causes are complex and data sparse; am I wrong on this?). And so, GCM forecasts cannot be expected to predict or track an abrupt climate change if and when one begins. Worse, if an abrupt climate change has begun, GCMs will suggest it’s not happening and that we are seeing temporary anomalies.

It seems unfair to blame climate scientists for not warning that the Arctic Sea ice may disappear rapidly, or that heat waves may rack North America, or that floods and storms may set new records. They have done so, but in terms appropriate to scientists. We must look to brave political leaders, poets and others to move the nation.

This old engineer fears we have already triggered an abrupt climate change, that nature will educate the public, and that the political climate on climate will change.

Account Deleted

Thanks for sticking up for the climate scientist Lawrence.

I also think they do a good job, with tools that have to be developed for an ever-changing situation. I'm also impressed that many of them take the time to make their data available/answer question from us.

Account Deleted

As to the flaws in the peer review process. Yes there can be problems/issues - but at the moment it is the best system we have. The alternatives i.e. Post-publication review, Open peer review haven't really taken off in anyway.

James Benison

Despite claims that the "best predictor" is the physical climate models, available evidence suggests that the physical models are not accurate.

First, no climate models (to my knowledge) have been able to reproduce the ice loss that has been observed in the last few decades.

Second, no climate models (to my knowledge) have been able to reproduce the conditions of the Pliocene. That is essentially the climate forcing we have in place right now today. The models say ice in the arctic. Paleoclimate data says NOT.

Given that an important criteria for validating a model is that it should be able to reproduce observed data either the data is wrong or the models are wrong. So far the data is winning two points to one.

Account Deleted

This Annual Review article covers on-going work to improving sea ice predictions.


They acknowledge they have a problem in modeling sea ice loss.

Sea ice is undergoing rapid decline; however, the skill in multimodel averages is relatively poor, the uncertainty in multimodel ensembles is large, and both are subject to model selection.

James Benison

Unfortunately things in the arctic seem to be changing much more quickly then they can be modeled and predicted.

I cannot imagine anyone in the year 2002 predicting that we would observe the changes that have occurred in the last ten years.


I've been on the road for ten days, so have been a little late updating things; just wanted you to know that the PIOMAS "death spiral" chart has been updated to include the June data: (Click for full-size):


It's the tightening up of the spiral that's most alarming, I think. Given the way the ice melt is speeding up, I can't imagine how small the orange ellipse denoting the current decade's average will be by the time the decade is complete. (Or maybe I can, and am afraid to...)

(I inadvertently posted this to the wrong thread earlier. Sorry about that; I intended it to only show up here.)


Jim Pettit: The whole climate seems to move from one mode to another. The 80s curve shows almost no "New Year's bump" from comparing ice volumes on Jan 1st 1980 and Dec 31st 1989. The following curves don't only show it, but are drawn further and further inwards....

Account Deleted


You are right about that and it is a point they make in the Annual Review article (I can send a copy to you if you want). I just felt that Aaron was being a bit harsh on the climate scientists. It's expect that they take a bashing on Watt's site, Neven's site is meant to be relatively Climate Scientist friendly (review/critique the science - but substantiate your claims).

Al Rodger

Just for a bit of fun, here's a prediction for the PIOMAS.v2 average for September.
Graph of Ice Volumes using Ice Age
Lasy year I had been playing with PIOMAS & Summer % Ice Age from NSIDC on a spreadsheet, wondering if it would give some indication of ice thinning over the decades. Very quickly the Ice Age data started giving PIOMAS-type figures so I gave up on the thinning.
I thought now to dust it off & see what happened when I put in some guestimates for % Ice Age & September Ice Area. Come the end of September, we can see how silly it looks.
The eventual number, exceedingly 'input sensitive', came out at 3,690 cu km/


3690 is more than 2 standard errors above my 2200 +/- 1 std err of 550. Are we both going to look silly or just one of us?


std err of 550

The people of PIOMAS estimate absolute uncertainty of +/- 1350 km3 (or +/- 750 for anomaly uncertainty).

I have used the 1350 figure, taken as 2 sigma, into my own calculations. With added fitting uncertainties I get a 2 sigma error of about 2300.


Yes 550 is certainly too low. I was taking it as a estimate of the PIOMAS figure (in the way they are currently estimating volumes) not the actual ice volume so I haven't included the 1350 km^3.

Forecast errors are likely to bigger than the standard error which is measuring the fit assuming the method used is perfect so there is nothing else other than gaussian noise. My method is certainly not perfect and was done in a hurry. The last factor added changed the forecast from 2500 to 2200 and there could easily be larger factors which I haven't used.

So doubling 550, adding 1350, adding further method fitting uncertainties, a sensible 2 sigma error for the ice volume (rather than PIOMAS figure) from me is quite possibly over 2300 so your 2300 looks sensible and perhaps we are not too far apart.

Perhaps an outcome somewhere around 2950 would leave neither Al nor I looking too silly.

I feel my 2200 is putting me out on a bit of an alarmist limb. But I trying to keep it to where the statistical method I investigate lead me.

2300 2 sigma range is more than my 2200 estimate so I cannot rule out the ice disappearing this year. That seems pretty unlikely looking at MODIS and makes me wonder if the uncertainty distribution is skewed or something for such a dramatic ice disappearance to look possible albeit as a romote possibility from the numerical farecasts.

I do welcome discussion of differing forecasts. I think it helps to assess how far out of line with most views I am.


So doubling 550, adding 1350,

Ignore the following if it is already obvious. Independent, normal distributed errors add in a quadratic way. To combine errors 1100 and 1350 as:


about 1740.

Al Rodger

Well I had another go at predicting the 2012 POIMAS minimum with the thought that the melt season to date should give some indication of the season's final outcome. I was also hoping to do better than my previous prediction based on changing "Ice Age."

The model I arrived at is ridiculously simple - Minimum Volume = Day 150 Volume - 14,800 cu km. For 2012 it works out as 3,300 cu km. The variance on calculating past Minimum Volume is s.d. = 800 cu km. (Volume in the linked graph (2 clicks to "download your attachment") remains Ave Sept Vol.)

So I seem to be creating a track record for myself as a "High Ice-ist." This is entirely unintentional.

(And as for the suggest that folk adventurous enough to set forth a reasoned prediction for this year's minimum could be branded as "silly" if their prediction proves to be wildly wrong - surely not. That would be like saying that the supercilious grin on the face of forecaster who just happens to get it right would be well deserved!)

The comments to this entry are closed.