.....

[bluewave]

......

[AvantHiatus]

In the grand scheme of things, a cold NW atlantic may be irrelevant outside of its respective regional influence and impacts upon the North Atlantic Oscillation. It's possible to lose significant ice cover with a +NAO depending on other parameters.

 

However, this change may protect Greenland from severe melting in the short-term. Additionally, whatever gains are made are supplanted by the much larger Siberian Ridge pattern.

 

[The_Global_Warmer]

No.  I would be shocked if that is sea ice low thru the year 2020.

[ORH_wxman]

I'd give it about a 50/50 shot or so...hard to say. It took 5 years to break 2007's record...though the ice has rebounded better than it did after 2007 relative to the very next year....2014 will tell us a bit more. IF the N atlantic actually does cool for the rest of the decade, then it will have a good chance to be the low...but I'm not convinced of that yet.

 

 

One thing I'd be quite certain of, is that we won't have "ice free" before 2020...prob not before 2030 either.

[skierinvermont]

I was going to say 50/50 too. If I had to guess I'd lean towards seeing one lower year.

[PhillipS]

Those articles are interesting.  Thanks for sharing them with us.  However, the empirical data don't support the hypothesis that the decline of Arctic SIE or SIA is slowing since 2012, much less stopping.

 

Here are the Arctic SIE monthly long-term trends (%/decade) as reported by NSIDC since 2009:

 

             J     F     M     A     M      J      J       A        S      O      N      D

2009  -3.1, -2.8, -2.7, -2.8, -2.5, -3.3, -6.1,   -8.7, -11.2, -5.9, -4.5, -3.3

2010  -3.2, -2.9, -2.6, -2.6, -2.4, -3.5, -6.4,   -8.9, -11.5, -6.2, -4.7, -3.5

2011  -3.3, -3.0, -2.7, -2.6, -2.4, -3.6, -6.8,   -9.3, -12.0, -6.6, -4.7, -3.5

2012  -3.2, -3.0, -2.6, -2.6, -2.3, -3.7, -7.1, -10.2, -13.0, -7.1, -4.8, -3.5

2013  -3.2, -2.9, -2.5, -2.3, -2.2, -3.6, -7.4, -10.6, -13.7, -7.1, -4.9, -3.5

2014  -3.2, -3.0

 

Hopefully that table won't be too hosed up when it posts.  As you can see, only the Spring months of April and May show significant improvement and even there the trends are still negative.  September, the month of the SIE minimum, shows an accelerating decline, even with the milder melt season in 2013.  Here is the NSIDC plot for September:

 

 

I give empirical data more credibility than model projections (particularly since the Arctic sea ice models have seriously overshot the actual melting) so in my opinion the odds of 2012 being the lowest minimum SIE for the 2010s is one chance in ten.

[skierinvermont]

Extrapolation of empirical data is not the same thing as empirical data. Empirical data is historical.

 

Extrapolation of it introduces a very shaky assumption: historical trends will continue into the future. It also assumes the trend is linear.

 

 

I would correct your statement to read "I prefer reliable empirical data to model hindcasts." Even that statement might not be true all the time in some situations where model hindcasts can be considered more reliable than less reliable empirical data itself.

 

When talking about the future a blanket statement of empirical data > models is just silly. Any prediction of the future should attempt to identify causal mechanisms and whether these mechanisms will persist. Doing so is the very essence of modeling. We don't always need computers to 'model' because our brains are very powerful modeling tools themselves. But the use of computers, or even just simple physical models, or pencil and paper calculations, allows for the introduction of more variables than our brains can simultaneously process and also is more objective.

 

Every projection of the future is 'modeling.' Even simple linear extrapolation is a form of 'modeling' technically but it relies on a very shaky logical assumption (all the mechanisms of the historical trend will continue to evolve exactly as they have historically). It makes no attempt to identify what these mechanisms actually are and if they will persist.

[PhillipS]

Extrapolation of empirical data is not the same thing as empirical data. Empirical data is historical.

 

Extrapolation of it introduces a very shaky assumption: historical trends will continue into the future. It also assumes the trend is linear.

 

 

I would correct your statement to read "I prefer reliable empirical data to model hindcasts." Even that statement might not be true all the time in some situations where model hindcasts can be considered more reliable than less reliable empirical data itself.

 

When talking about the future a blanket statement of empirical data > models is just silly. Any prediction of the future should attempt to identify causal mechanisms and whether these mechanisms will persist. Doing so is the very essence of modeling. We don't always need computers to 'model' because our brains are very powerful modeling tools themselves. But the use of computers, or even just simple physical models, or pencil and paper calculations, allows for the introduction of more variables than our brains can simultaneously process and also is more objective.

 

Every projection of the future is 'modeling.' Even simple linear extrapolation is a form of 'modeling' technically but it relies on a very shaky logical assumption (all the mechanisms of the historical trend will continue to evolve exactly as they have historically). It makes no attempt to identify what these mechanisms actually are and if they will persist.

 

I agree with all your points - I should have been clearer in my post that my lack of confidence is the Arctic sea ice models which have a consistent track record of optimistic projections.  As an engineer involved in R&D I use models daily - some simple pencil and paper, some finite element analyses which require lots of computer power.  But I wasn't speaking in general terms, I was addressing the thread topic.

 

Since the topic of this thread is whether a new SIE minimum will be set within the next six melt seasons (through summer 2019) I feel that the observational record is relevant, and reliable.  The period of satellite observations is over thirty years and is therefore long enough for trends to have some predictive power.  You're correct that linear trends and linear extrapolations assume that whatever has been causing the decline of Arctic sea lice will continue to occur - at least for the next six years - but I don't understand why you feel those are shaky assumptions for an admittedly rough projection.  But six years isn't a horribly long time to extrapolate from a relatively consistent thirty year record, and  I don't know of any major new factors affecting Arctic sea ice during the summers - OHC is still rising, global temps are still increasing, according to PIOMAS the SIV is still a fraction of what it was at the start of the start of the observational period and, as I posted earlier, NSIDC reported that 2013 showed a worsening of the SIE decline over 2012 and not an improvement.  What do you feel I'm missing?

 

Looked at from a purely probabilistic standpoint for the sake of discussion, with the assumptions that each year's melt season is independent and any one year in a decade would have a 10% probability of setting the minimum, the probability of a new minimum being set in the next six years would be:

 

P = 1 - (1 - 0.1)⁶  or about 47%, which is close to the fifty / fifty odds some posters favor.

 

Of course the erroneous assumption is that each year has an equal chance of setting a new SIE minimum.  With the observed -13.7%  per decade trend, each year has about a 1.3% greater chance of setting a new minimum than the year before.  And remember, we're starting the odds calculation two years after the 2012 minimum.  So a more realistic probability equation would be:

 

P = 1 - [(1 - 0.126) * (1 - 0.139) * (1 - 0.152) * (1 - 0.165) * (1 - 0.178) * (1 - 0.191)] or about 65%, closer to two / one odds.

 

As I wrote in my earlier post, my opinion is that the likelihood of setting a new SIE minimum this decade is even greater than 65% - that it's around 90%.  I based that on the non-linear factors of ice melt playing a greater role as the sea ice decline continues.  Those factors include, but aren't limited to, thin ice having a lower albedo allowing sunlight to penetrate to the water below, thin ice having less structural integrity allowing large areas to be crumbled by waves and winds (crushed ice melts faster than big blocks) and lower spring SIE maximums mean more open water at the start of the melt seasons.

 

Well, I've gotta run but I hope that this longer explanation has shown that I don't think we're far apart on our viewponts.

[ORH_wxman]

 NSIDC reported that 2013 showed a worsening of the SIE decline over 2012 and not an improvement.  What do you feel I'm missing?

 

 

 

I'm not sure why you continue to think this. Only the linear trend per decade got worse going back 30 years...not the actual sea ice itself. 2013 had higher volume and extent the entire melt season than 2012 did. We could flat line another 2-3 years and the linear decline would still worsen because linear regression would start to place more weight on the post-2007 type values than they did previously when the sample was only 4 or 5 years.

 

To show how the linear trend is affected by the weight of earlier years: If you use a 10 year trend...then stopping at 2012 is worse. 2002-2012 has a slope of -0.21 while 2003-2013 has a slope of -0.16. 2002-2013 has a slope of -0.15 just to show that its not biased to exclude 2002.

 

 

Essentially, no matter what happens this melt season within reason, the 1981-2014 trend will be worse than the 1981-2013 trend...even if we finish with a value higher than 2013. So, using that logic, it doesn't matter what 2014 does...the ice will be worse no matter what.

 

That type of logic doesn't really make any sense if we are judging the state of sea ice versus recent low years (such as 2012).

[PhillipS]

I am surprised that the UK Met Office is touting accuracy of their HadGEM1 model.  Here is the iconic graph of the model projections versus the Arctic sea ice observational data:

 

We have all probably seen this graph at one time or another.  There is another version which was updated to include the 2012 data but for some reason this forum throws up when I try to post it so I've provided the link to it.  Notice that the UKMO HadGEM1 model projection is shown.  I encourage everyone to examine for themselves whether that model "has projected Arctic ice loss with reasonable accuracy up to the present day".  I haven't done a comprehensive comparison between the HadGEM1 model projections and the observational data, but from my cursory examination it appears that HadGEM1 projections have been just as bad as many other models.  Or maybe I'm just unreasonable.  So my question is - can anyone give me good reason to give the HadGEM1 model projections any credibility?

[skierinvermont]

Bluewave, didn't you delete your thread on HadGem1 because you no longer thought it was accurate after I pointed out it shows rapid warming for the near future?

[Jonger]

I am surprised that the UK Met Office is touting accuracy of their HadGEM1 model.  Here is the iconic graph of the model projections versus the Arctic sea ice observational data:

 

We have all probably seen this graph at one time or another.  There is another version which was updated to include the 2012 data but for some reason this forum throws up when I try to post it so I've provided the link to it.  Notice that the UKMO HadGEM1 model projection is shown.  I encourage everyone to examine for themselves whether that model "has projected Arctic ice loss with reasonable accuracy up to the present day".  I haven't done a comprehensive comparison between the HadGEM1 model projections and the observational data, but from my cursory examination it appears that HadGEM1 projections have been just as bad as many other models.  Or maybe I'm just unreasonable.  So my question is - can anyone give me good reason to give the HadGEM1 model projections any credibility?

 

Considering our previous knowledge of the arctic pre-1979 is so bad, I would question many of those previous plots as being overly exaggerated high. That would bring it more inline with the poorly performing GCMs.

[skierinvermont]

Considering our previous knowledge of the arctic pre-1979 is so bad, I would question many of those previous plots as being overly exaggerated high. That would bring it more inline with the poorly performing GCMs.

 

Actually those plots are too low relative to our pretty decent knowledge of sea ice extent back to the early 20th century.