« What it's all about | Main | Meanwhile, in Greenland... »


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

Chris Reynolds

Last night my attempt to submit was rejected twice. I will have to try again tonight.

Jim Hunt

Not being one to take any "Real Science" headlines at face vale "Snow White" has been doing her own due diligence on northern hemisphere snow cover. What she discovered seems to me to be relevant to any forthcoming SIPN predictions:

Alaska May Snow Cover At Record Low Levels

Here's something else she found flapping around in the cold wind blowing across the floor of the Rutgers Global Snow Lab:


Never if I recall correctly we are talking about the lowest 5 day running average (nsidc), value in September for Sea Ice Extent.

If that's the case I'll go for

3.9 +- 0.5

Method. WAG/years of observation.

Let's see how wrong I can be... Might as well start the ball rolling.

I just had a look at the NSIDC Arctic page, haven't been there in quite a while. It seems that 2015 is -2sd below average for most of 2015 and May stands in 3rd place overall and the lowest in the post 2010 record.

Worth a guess.

Nightvid Cole

I went ahead and sent in the snow-based estimate of 3.26 M km^2 based on TOPAZ4 and explained in more detail on the forum.

I have stuck to June 13th as the date to use for snow cover, first for consistency, second to avoid dependence on long-range weather forecasts, and third, because it seems the correlation begins to degrade after that due to the slow-melt years "catching up".


I would love to know Rob Dekker's prediction because it is thoughtful and on physics, and he gets the tendency very nicely. But I am in no position to demand he disclosing it here or submitting it to SIP

Helen Wiggins (ARCUS)

Hi all - I wanted to leave a quick note to encourage submissions from this great citizen-science community. If you are emailing to the [email protected] address from a personal account, our spam filter might catch it, but we'll look at all of the filtered emails to catch any submissions. Also, if you have any problems, you can email your submission directly to: [email protected].

Helen, ARCUS and Sea Ice Prediction Network

Chris Reynolds


SIPN uses the NSIDC September average.

Bill Fothergill

"SIPN uses the NSIDC September average"

As Chris pointed out above, SIPN uses the NSIDC September average as the baseline against which they measure the various submitted estimates.

Unfortunately, the wording used on the SIPN site can be slightly ambiguous. If one looks at the link provided by Neven in the Original Post, one will find, for example, the following...
"share ideas about the monthly September minimum sea ice extent"

That has fooled more than one person, and I remember last year having a backwards & forwards on-line debate with one of the SkS team before finally convincing him that the monthly average was the metric of interest - not the actual minimum.

Chris Reynolds

Well, after several failed attempts my submission has not been thrown back from the Arcus email address, so it must have gone in. And I didn't need to bother Betsy.

Nightvid is 3.26M km^2, based on snow cover.

I have put in 5.15 million km^2 +/- 0.64 million km^2, or 5.79 to 4.51M km^2, based on April PIOMAS volume.

Rob Dekker has yet to declare.

David Rennie has yet to declare.


Chris, is that the average for the entire month or is that the running 5 day average figure they post on the arctic seaice news page?

I should have said

#3.9M km2 +- 0.5M km2


I've gone in at 3.8 M +/- 600K km^2. Would have been significantly higher except for the solid start even without warm weather over the past couple of months.

Chris Reynolds

Neil T,

My prediction is in terms of monthly extent for September, not the 5 day average.

From the SIPN Guidelines:

Point number 4 reads.

"4.*September monthly average projection (extent in million square kilometers).
To be consistent with the validating sea ice extent index from NSIDC, if possible please first compute the average concentration for the month and then compute the extent as the sum of area of all cells > 15%."

The monthly extent is available here:
And September is available here:

Chris Reynolds

Thanks for that David,

That's two low and one high put in for the SIPN.

Two of us have ideas we're testing using prediction. Nightvid is testing an approach using snow cover. I'm testing April volume as the driver of the extent decline.

Rob Dekker

Sorry to be late to the party (some trouble with NSIDC area numbers, and some uncertainty on snow numbers), but I just submitted my projection to ARCUS :

4.9 M km^2 with SD 470 k km^2.

This estimate is based on linear regression of how 'dark' the Northern hemisphere was during April and May, as estimated by two variables :
(1) Rutgers snow cover numbers from April and May and
(2) NSIDC May numbers of (extent-minus-area) to estimate dark (leads and melting ponds.

If I do the same linear regression excluding April snow numbers, I end up with a projection of 4.5, and a standard deviation of 520 k km^2, and honestly speaking, I think the April snow numbers just give a better SD because Northern Hemisphere snow numbers (including April) are following the normal linear down trend that can be expected from a globally warming planet.

So I think my 4.9 is conservative, and I would not be surprised if I need to down-adjust that number steeply next month.

Also note that even a simple linear trend obtains a SD of about 550 k km^2 or so, and thus I take my projection with a VERY LARGE grain of salt.

June projection should be much better.


Your estimate increases as compactness A/E is greater, doesn't it? Meaning that if compactness drops from current record values to more normal ones, your prediction will probably go down a lot (other factors apart). Thanks

Rob Dekker

navegante, yes, that is correct, especially since BOTH extent and area are running at record lows already.


Thank you.
This time predictions are well aligned around 5 M km2, Chris Reynolds' method giving 5.15 M, and the "melt pond guys" Schroeder et al 5.1 M km2, all with uncertainties about +/-0.5 M km 2.
I am feeling that it ll fall on the lower half of those ranges.

L. Hamilton

A new article on surveys tracking public beliefs about Arctic sea ice was just published online in Witness the Arctic (ARCUS):


There should be a larger piece in Polar Geography coming out next week.


Rob Dekker

Larry, thank you for posting these public belief findings.
I find your statements to be spot-on, at multiple levels :

On this and other factual questions, it seems likely that many people chose answers derived from their more general beliefs

I find evidence in that also on other subjects, such as MH17, Keystone XL, evolution theory and AGW.

On a positive note, mis-beliefs often just go away once evidence is overwhelming. In that regard, I find it encouraging that your survey shows, that the opinion of "climate is changing due to natural causes" is for the first time dipping below 50%.

Kevin McKinney

Rob, yes--too few time points for any statistical conclusions to be drawn I'd think, but the overall curves had, I thought, some promise according to the ol' eyeball.

Jim Hunt

The SIPN June report is now available:


The median Outlook value for September 2015 sea ice extent is 5.0 million square kilometers with quartiles of 4.4 and 5.2 million square kilometers. The overall range (excluding an extreme outlier) is 3.3 to 5.7 million square kilometers.

The "extreme outlier" is 0.98, from "Wadhams (SIPOG)"

The comments to this entry are closed.