« EGU2016, my impressions | Main | PIOMAS May 2016 »


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

Chris Reynolds

Crandles, John,

I'll get back to you tomorrow, when I'm not so tired.

Crandles, Rob,

Having downloaded some data this morning: Whilst warm air seems to explain a lot of the 2010 to 2012 increased spring volume loss, it doesn't seem to explain much of the rest of the series. I've converted some measures of both temperature (different heights and regions and thickness/volume(from PIOMAS) together with the Spring Volume Loss (roughly equal to May minus June volume) to a 1981 to 2010 baseline - equal to NCEP/NCAR baseline. Neither ice state nor air temperature provides a complete explanation. I'll get back to it later in the weekend when I'm not so tired.

PS I have checked - the large 2010 to 2012 SVL remains regardless of the 30 year baseline, so NCEP baseline ending in 2010 plays no role.

Anyone know what proportion of variance a given R2 explains?

Bill Fothergill

"Anyone know what proportion of variance a given R2 explains?"

Chris, that might be a bit like "how long is a piece of string?".

The following link might help...

Susan Anderson

Bobcobb: this is a wonderful community, unusually rich in sharing knowledge, learning, observing, collecting data, working together, and building on each others' work. I am a fringe lurker and have little to contribute, as a moderately informed layperson, but have been welcomed and encouraged when I do drop in. I catch up on comments to the main articles pretty much daily, and sometimes look through the forums (great resource). It is a fine demonstration of how to be encouraging and grow as a group without being kidnapped by attack.

You seem to me to not be in any way a climate science denier (unskeptical skeptic) and to make sense. You are, however, quick to attack and to overinterpret what other people say, while slow to moderate your own voice. You even did it to me, which startled me and forced me to repeat myself with greater clarity.

In the world at large, we need more than the choir to accomplish the necessary, and the time to do so is slipping away. Unnecessary anger, even if only in language and not intended to sound as sharp as it comes across, puts people's backs up. The people we need to be paying attention are not the people we agree with, but all of society. It's easy to find fault, but difficult to build bridges. Third parties observing the infighting can draw the conclusion that there is more distance between us than is actually the case.

Yes, there is some exaggeration, but there is much more understatement, ignorance, apathy, and denial in the world at large. Attacking people's religion, telling them they are flat-earthers, Dunning-Kruger, or making otherwise unfounded characterizations, especially if the accusations are incorrect, will drive away people beginning to open their minds. It's better to use the golden rule (do unto others as you would have them do unto you) than to be righteous.

You may trust Neven and others to remove repetitive dishonest voices, and though the pace may be slow, there is sometimes some lurker who is observing the process of tolerance and learning not only from the material but from the manner in which it is presented.


Fair enough Susan.

Susan Anderson

Bobbcobb: Thanks, with warm greetings ...



Thanks for your "etiquette" comment. Most everyone on here follows this protocol exceptionally well especially considering the number of voices present. I appreciate the reminder.


Susan said, "You are, however, quick to attack and to overinterpret what other people say, while slow to moderate your own voice."

As a long time mostly lurker who reads virtually 100% of this blog I appreciate the civil dialog and the "I want to know more; please explain so I will understand better" attitude of nearly everyone on here.

Unfortunately, I stopped reading some of your posts because of the abruptness that Susan points out as well as the absoluteness of some of your statements. In my world all means "all" and never or none means "never or none." For example: a/b = c/d if a=c and b=d is in fact a false statement. To make this a true statement we need to state that b and d do not equal "0". So, when you make absolute statements I understand them in a similar fashion so it became pointless for me to continue reading your posts.

I will try reading your posts again. Much of what you say has a "ring of truth." Please take Susan's advice.


Of course, Susan. Vaughn, you make a fair point. I have a history of dealing with Mcphersonites and their ilk who have insulted me before on when I've brought up my points in Fractal Planet and elsewhere. So you might understand why I post the way I do. I will ease off on the abruptness and such. That being said, I think most on here would agree that there's virtually no chance of ice free this year.

Rob Dekker

Chris said

Anyone know what proportion of variance a given R2 explains?

The definition of R^2 is Explained variation / Total variation.
So there you have it : If R^2 is 0.9, that means 90% of the variation is explained with this model.
However, that does not mean a lot, and there are lots of 'if's and 'but's to that, as Bill's link explains.

Personally, I prefer to look at the standard deviation of the residuals.
That typically gives a better way to compare models (including comparing to the super-simple ones like a linear trend).

Chris said :

: Whilst warm air seems to explain a lot of the 2010 to 2012 increased spring volume loss, it doesn't seem to explain much of the rest of the series.

I am not exactly sure what you are trying to do here. Could you explain in more detail ? (Maybe in a blog post) ?

Susan Anderson

bobcobb, thanks again and I understand about MacPhersonites, but many of his followers are good people.

While I felt it was risky to embark on that extended comment, my "side" is prone to groupthink in making the accusations I described and others. So this was not meant entirely for you, but for others whose language tends toward sharpness.

I am afraid of what we are doing to our world, and that means we need to be pleasant to people we don't agree with, not suggest they are stupid (even when I secretly think they are). I fail to take my own advice at times.

[OT: In the US followers of two reasonable voices on climate are at each others throats and enabling the real enemy, egged on by outsiders with vested interests. They don't know they are being played. Pointed criticism, even if it doesn't descend to outright insult just plain doesn't help.]

I see Neven has a new article, so we can all move on.

Chris Reynolds


A high index winter AO (low pressure over the pole) produces more advection from the Siberian coast, which results in more ice production, a decrease of transport in the Beaufort gyre and an increase in Fram Strait export (Rigor & Wallace 2002). This means reduced winter thickness in the peripheral seas of the Arctic Ocean Basin, which during the time of Rigor & Wallace's study probably played a role in low minimae.

What you are looking at is autumn and Spring.

The spring connection probably has a role for open skies and preconditioning. But you might also be picking up early Arctic Dipole activity. Overland's paper on the early summer AD looks at June as a crucial time for AD impacts in the summer. A negative AD over much of the pack as early as April would favour cooling as the sun is so low net longwave emission overcomes shortwave absorption. At 250K (-20degC) assuming an emisivity of 0.8 (rough guess) the emission of ice is 177w/m^2, taking into account that the calculations we have are for TOA (so at low angle of incidence the incident insolation will be lower), and that the ice has a high albedo, absorbed insolation is likely less than emitted infra-red.

Autumn, I agree with, increased clouds will trap heat, probably also associated with winds which mix the column of summer warmed water.

I'm not convinced by your total of AO for autumn and winter, e.g. 2008 and 2009 were rebounds from 2007. I suspect that in using only 2006 onwards you're rather limiting your dataset, what does it look like for 1979 to 2015?


I'm not ruling out a continuation of warming in the late winter period, I just see data that is too nosiy to support that conclusion. As I said the integral of warming over winter (FDDs) is what dominates the thickening of ice, and this doesn't suggest an abnormal warming over the last few years. I now have a full set of CSVs of gridded NCEP surface temperature, 1948 to present (using Panoply) and to test my code handing it I'll calculate FDDs over ocean for the full period. I'll blog on it when I'm done. This means I'll be able to produce surface (0.995 sigma) FDDs since 1948, I'll let you know when it's done - but this week is going to be another week of long days at work.

Chris Reynolds


It'll be a while before I'm ready to post, from experience on other matters it could take the rest of the year to tie it all together (or I might crack it with a flash of inspiration whilst doing something else this week!).

I've decided to try looking at timing of the end of sub zero temperatures, so I've got a dataset of daily surface temperature from 1948 to a few days ago and will have to code the solution to that - not totally sure I know how I'll approach that - not yet anyway.

Anyway here's a hint of what I mean. I started off by using the May-June difference in volume as indicative of the spring volume loss SVL). With daily volume and temperature I could feasibly look at the situation on a day to day basis, but would have to have a more rigid definition of the onset of the SVL. The end is easy as the anomalies go upwards at the inflection.

Note that there are forcing files for PIOMAS
But they only go up to 2004 and IIRC are monthly.

So using the monthly PIOMAS SVL data (in blue on the following), and as an example I have shown temperature at surface and 700mb.

Oh, and both these are baselined as anomalies from the same period; 1981 to 2010. Changing the anomaly period to an earlier 30 yr period does not remove the SVL.

Surface temperature north of 70degN

700mb temperature north of 70degN

OK. So for surface, the SVL curve prior to 2010 bares little relationship to the May/June temperature anomalies.

While at 700mb we have a better relationship over that period, but for 2013,14,15 if temperature is the driver than why does the SVL anomaly remain so high after 2010 to 2012??? Lookin at that data was one of the things that led me to suspect thickness loss after the 2010 volume loss event was playing a role. But the 2010 volume loss mainly appears in the heff based caculations. Gice should be a better measure, and the loss of thick ice after 2010 in that data is far less pronounced.

My reading of the evidence suggests that this is far from simple.

There are other diagnostic PIOMAS variables:
But Oflux (ocean heat used to melt ice) only goes to 2004, and the top ten layers of the ocean temperature only goes to 2013 - might still be useful though.

I keep thinking back to the papers on sea ice modelling I've read. Maybe I could do a single column model to try to figure out the physics of the matter. When I last looked at it, the issue was horribly complex for a math dunce like me.

R2 of 0.9 = 90%, now that's what I thought, but someone knowledgeable told me it wasn't that simple. Ah, yes, looking at Bill's link (thanks Bill), I'm probably usually more interested in the explained amount of standard deviation than variance. Which explains my confusion.

Rob Dekker

Chris, I'm sorry, but I don't understand what you are doing.
But please let me know if I can help.

Chris Reynolds


I'm trying to figure out what has caused the enhanced spring volume loss in PIOMAS.

Rob Dekker

Thanks Chris. That makes sense.
Regarding your initial findings :

Whilst warm air seems to explain a lot of the 2010 to 2012 increased spring volume loss, it doesn't seem to explain much of the rest of the series.

that kind of was expected.
After all, temperatures in spring only affect melt (melt-ponding) if they are above 0 C. And if they dip below that after a while, ponds freeze over and un-do the melt and the albedo-effect.
So there is that non-linear effect of melt-ponds, but beyond that, melt in spring is also affected by open-water close to the ice. Think albedo difference between ice and open water : it warms the water quite nicely, which leads to ice melt later on.
And then regarding the "rest of the series" there are a lot of variables that would need to be considered.

So I'm not surprised that you conclude :

My reading of the evidence suggests that this is far from simple.

I couldn't agree more.

The comments to this entry are closed.