Ice Free Arctic

Xnn

I used all the data, and found that a linear trend does not fit very well.

Since 1970 or so, it does appears to be linear with the exception of the last couple years. However, there is no reason I can think of to throw out the last few years.

Excel has a nice function where different types of trend lines and projections can be drawn. So, from just playing around with it, a 4th order polynomial looks to fit the best.
Of course, this isn't a physics based approach, but the data is clearly trending lower at an accelerating rate.

mugaliens

Amazing! About the time we'll run out of oil...

Comming at 'ya!

Skyhunter

Sun spots for recent activity, (last 400 years) and beryllium-10 for further back

Be that as it may...it is a huge mistake scientifically to base ones hypothesis on fifth order evidence that is contradicted by first and second order evidence.


First. The oceans would not get colder due to tidal mixing, whatever that means. Moving the heat around does not change the overall temperature.

Second. What evidence do you have that suggests such a phenomenon occurs?

sylas

Trying to project the data series is not a particularly good guide, no matter what order projection you use, unless you know something more about the underlying relation. You need some kind of physical model for relating ice and climate, and some kind of physically based expectation for what climate might do, before you can really say anything much about when the Arctic might be free of sea ice in summer.

Projections well beyond the range of the dataset are only useful if you have some independent theoretical basis for thinking that the underlying trends follow the kind of function you are using for making the projection.

Cheers -- sylas

mheslep

Well said. That's why any time we read a prediction somewhere, we would all gain more clarity if the prediction was attached to a "based on the XYZ model/theory" qualifier. Even the popular press could spare a half dozen words to heed that guidance.

Skyhunter

The Arctic oscillation (AO) has been in an extreme negative phase this winter. High pressure over the Arctic and low pressure in the lower latitudes, leading to an anomalously warm Arctic winter. This has slowed the formation of winter sea ice, however if it persists through winter it could lead to a slight recovery of the sea ice due to a decrease in wind blown perennial ice into warmer waters.

I personally think that as the ice gets thinner and thinner, the Arctic sea ice will disappear suddenly some summer in some not to distant future.

mheslep

So much for posts 21 and 22.

Xnn

Here is an abstract of the 2009 Wang paper that predicts a nearly sea ice free Arctic by September 2037. It is based on 6 IPCC physics based models. Also, the the top quartile prediction is 2028.

http://www.agu.org/pubs/crossref/200...GL037820.shtml

I understand that predictions made without a physical model is risky.
A problem is that one might be unwittingly cherry picking the data.
However, when I put the minimum sea ice data from the Cyrosphere site
into an Excel spreadsheet and fit the best curve to it,
I just so happen to get about the same result. So, I feel confident
that my projection is about the same as that obtained from the physics based models.

sylas

This is more credible than a projection only, IMO, with the proviso of course that sea ice modeling is hard.

We're going to have to finish up here soon, unfortunately, but it may be worth leaving the thread with a link to msg #23 of thread "State of the Climate", where mheslep gave a link to the Sea Ice Outlook project at the "Arctic Research Consortium US". This was basically a competition between research groups to predict sea ice change one year in advance. A short term prediction is a rather different beast, subject to larger variations with local effects of wind and current obscuring the trends that are in many ways easier to model: but the failings of the models were intriguing.

In brief; loss of summer ice by 2040 is reasonably well founded; but given the uncertainties it might be rather more, or less, time required.

Yes, that gives a lot more weight to the projections. I'm sure all the modelers will be continuing to look at the matter and refine their models for some time yet. The one consistent feature of applying physics to the problem is that the loss of ice is not merely an aberration, but a real physically well founded consequence of changes that will be continuing to take place for many years yet.

Cheers -- sylas

hamster143

I look forward to fishing at the North Pole in July 2040.

Xnn

Here's an image with a projection drawn using the Cyrosphere data.
It is in rough agreement with Wang's published prediction of 2037.

Attached Thumbnails

 

sylas

Can you be more specific as to what data source you are using? Different groups define the extent in different ways, and the numbers in your graph are not from the definitions I've normally used.

Your graph pretty obviously omits the 2009 figure, which recovered quite a bit from the record minima of 2007/2008, which show up on the graph pretty clearly; but 2009 is not there. Including the data from Sept 2009 makes the match with Wang rather less good... and note also that Wang's group was part of the prediction group I mentioned in the previous post. They underestimated the extent in 2009, as did everyone else; although as usual the number in a given year is in part due to short term changes at the particular weeks of the minimum, not only long term trend. Note that the effort reported must have used different definitions from the numbers in your graph.

Another common source is the National Snow and Ice Data Center (linking to the Oct 2009 report), in which minimum summer extent shows as follows:
(link to full size original)

I wouldn't discount Wang's expectations, but neither would I bet a lot of money on their proposed date. The trend has been substantially faster than estimated in the IPCC 4AR, but Wang's prediction is currently looking to be bit too far the other way.

It's best, I think, to acknowledge that Arctic summer minimum ice extent is hard to predict, and that any specific prediction is tentative, given the limits of models being used. The models do indicate that the decline is real physically founded trend that will be continuing into the future; but quantifying the models to the level of a specific year is... rash.

Cheers -- sylas

Xnn

I use the UIUC Cryosphere site. Specifically the following diagram (summer):

http://arctic.atmos.uiuc.edu/cryosph....1900-2007.jpg

I see they report different values than the NSIDC and don't have a value for 2009.

For 2007, NSIDC reports 4.3 Mkm^2
For 2007, UIUC reports 5.5 Mkm^2

The NSIDC chart shows a decline of about 2.5Mkm^2 over the last 30 years.
The UIUC chart shows a decline of about 3Mkm^2 over the last 30 years.

From the above data, it does not appear that the Arctic would be ice free within 30 years. So, for Wang's and others physics based models to be correct there will have to be an acceleration from the rate of ice loss over the last 30 years.

While the 2009 minimum extent was more than 2007/2008, it does not appear to be substantially greater than the trend line.

The differences between the 2 sources of sea ice extent data suggest to me that they may have different definitions of sea ice.

mspelto

For Wang and others to be correct we need an amplification. And an amplification does appear to be underway. I was impressed with this paper that I was a peer reviewer for.
http://www.the-cryosphere.net/3/11/2...-3-11-2009.pdf. The emphasis is on observed temperature anomalies and increased heat flux from the open ocean from late summer-early winter.

sylas

I believe the cryosphere today website defines "extent" as the area that has 50% cover or more of ice -- could be 30%, I am not sure. NSIDC uses "extent" as area with 15% cover or more. It may just be me, but I have found it hard to get the information from cryosphere today, by comparison with NSIDC.

I did a bit of private analysis on this last year when the blogosphere suddenly went wild with a bunch of claims about scientific malfeasance based on an exercise by Steve Goddard in the UK counting pixels on images to support some wild accusations. Goddard himself learned from the experience and accepted useful corrections from the NSIDC. But the whole brouhaha did show very clearly just how quick many people are to jump on claims about how climate scientists are all involved in some hoax given even the silliest pretext.

For background on the story, see this poster session on the whole event by Walter N. Meier, Stephanie Renfrow, and Mark Serreze of the NSIDC. Striking back: A case study in addressing a skeptic’s public assertions about sea ice data.

Quite right! You can see indications of an acceleration in the data, very roughly around 1996/1997. The acceleration is not quite as strong as Wang's proposal, I think, but some acceleration is a consistent feature nearly all physically based models. The graph I provided from NSIDC has a single regression line from 1978, which if extended as a linear trend would indicate no summer sea ice in the Arctic by the end of the century.... but there's an inflection in that graph which would mean any non-linearity makes it likely to be a lot sooner. I wonder if Wang's work relies too much on the 2007/2008 minimum. The abstract states that this minimum is used as a "starting point", but I think the data shows them to be anomalous, even given the accelerating trend.

So... an essentially ice free summer as soon as 2037? Perhaps. This estimate is an average of a number of models, some of which suggest even sooner and others which suggest a bit later. By 2070? Very likely. By 2100? Bet on essentially ice free summers before this. Ice free in this context still allows for isolated floating sea ice and for ice around some coastal margins.

Cheers -- sylas

Bill Illis

I've spent a bit of time working with the sea ice extent numbers.

The satellite sea ice extent data actually goes back to 1972 but for whatever reason, it has not been matched up/reconcilled to the satellite data starting in 1979 provided by the newer post-1979 satellites.

I've matched up the daily satellite-estimated sea ice extent estimates provided by Jaxa with the Nasa Team algorithm daily data which goes back to 1979. Jaxa used the Nasa team algorithm in developing their algorithm so they are reasonably consistent.

http://www.ijis.iarc.uaf.edu/en/home/seaice_extent.htm

The main point is that, throughout the year, there is some variation in sea ice extents but there is a seasonal range that has not varied as much as the September minimum numbers make it look.

This chart is not the last word and some adjustments were required to match up the two records, but it does provide a little better perspective on the Arctic/NH sea ice extent. The two lowest sea ice extent years (2007 and 2008) and the two highest sea ice extent years (1980 and 1996) are highlighted as well as 2009 (the red line).

There is still a lot of change required before the day 255 (September 12th, the average date of the minimum) sea ice extent hits Zero.

sylas

I think there are problems with your match up, and that you may be giving values for sea ice minima prior to 2002 that are too low. The lack of an existing match up record is very probably because there are all kinds of difficulties with doing a match up across different satellites and time periods, and such a graph is known to be misleading.

Given the rapidly rising temperature records in the Arctic, it is surprising to see bits of the graph dropping below the 2009 minimum. I think this indicates inaccuracies in how you have merged data from different sources.

Cheers -- sylas