Arctic amplification dominated by temperature feedbacks in contemporary climate models

There’s a paper in Nurture GeoSci entitled Arctic amplification dominated by temperature feedbacks in contemporary climate models by Felix Pithan & Thorsten Mauritsen (Nature Geoscience (2014) doi:10.1038/ngeo2071). As far as I know the paper is entirely sensible, though I’ve only read the abstract. From which I quote:

Feedback effects associated with temperature, water vapour and clouds have been suggested to contribute to amplified warming in the Arctic, but the surface albedo feedback—the increase in surface absorption of solar radiation when snow and ice retreat—is often cited as the main contributor… we analyse climate model simulations from the Coupled Model Intercomparison Project Phase 5 archive to quantify the contributions of the various feedbacks. We find that in the simulations, the largest contribution to Arctic amplification comes from a temperature feedbacks: as the surface warms, more energy is radiated back to space in low latitudes, compared with the Arctic… the surface albedo feedback is the second main contributor to Arctic amplification and that other contributions are substantially smaller or even oppose Arctic amplification.

So far, so analysis-of-climate-models. As I’ve had occasion to say in the past: GCMs can be good tools for studying climate, but alas they are nearly as complicated as the real climate system, so it can often be pretty hard to work out why they are doing something, even if you know that they are doing it.

But if you squint at the paper a bit through “skeptic”-tinted googles, it looks a bit like it is saying that the models have got it wrong, and that something else is the real effect. And this is what NoTricksZone has tricked itself with: the headline there, “Climate Modelers Flub Again…Albedo Not The Number One Arctic Amplifier After All!” says it all. In the comments I’m patiently trying to explain to them what the paper is really saying. There is some hope – there’s an update to the post which rather plaintively says the claims made by the authors are based on “model simulations” which you might well have thought was the bleedin’ obvious: “Here we analyse climate model simulations from the Coupled Model Intercomparison Project Phase 5 archive to…” is more than just a hint to the wise; although There’s no light the foolish can see better by.

Speaking of which, WUWT falls into the same error. And for a wonder, my comment pointing this out passed moderation. Even more astonishingly, more than one person has, though gritted teeth, admitted that I’m right.

Comments there: NTZ

You haven’t understood the paper. What the paper discusses are the mechanisms for Arctic amplification, as seen in climate models. The models the paper discusses are largely the same as the ones used before. So its not possible for these results to falsify the models, because these results are based on taking the model results at face value.

What you’re missing is the distinction between *understanding* the model results, and the construction of the models.

So “This casts many of the assumptions made in earlier climate models deep into doubt” is entirely wrong. The assumptions made in building the models aren’t challenged at all by this study. What this illustrates is the way the models are so complicated, it can be hard to know exactly why they do things.

FWIW, the idea that ice-albedo feedback isn’t the full answer has been known for some time; e.g. discusses V A Alexeev et al, Climate Dynamics (2005) , which actually sounds distinctly similar to the new Nature paper.

got the reply:

Not really sure what you’re getting at. Wrong assumptions with right answers still means faulty models.

One thing is clear: 114/117 models have missed the barn so far. Models have been hopelessly inadequate, biased in one direction. In general the paper in my view is rather murky and, should it come out from behind the paywall, it would be interesting to get a closer look at it.

So there’s a regrettable failure of understanding, and a not-hard-to-predict attempt to deflect the conversation.

My reply took two attempts to get through:

> Wrong assumptions with right answers still means faulty models

There is no suggestion from the paper in Nature that the assumptions in the models are wrong, or that the models are wrong. The Nature paper presents a new analysis of the models, and that analysis suggests that “the surface albedo feedback… [which] is often cited as the main contributor… is the second main contributor to Arctic amplification”.

> what Spiegel writes

I can find nothing in the English translation of the Spiegel article which supports your assertion. I can see how a rather hasty (and from your point of view, hopeful) reading of the article, or the Nature paper, might suggest that.

I hope you can understand the point I’m trying to make: this doesn’t challenge the assumptions, or the results, of the models at all. It makes no suggestions that the models are in errors. It merely presents a new analysis of the model results.

I don’t think I’m getting through.

Comments there: WUWT

My opening bid was:

> climate models need to be reworked

You seem to have made that up. Its not in the paper.

REPLY: It’s an opinion. much like many of your Wikipedia entries – Anthony

since I didn’t feel inclined to waste my pearls of wisdom if they were going to be suppressed. But they weren’t (indeed, AW replied, I’ve just noticed that so I’ve added it in here. He doesn’t seem to understand that its not an opinion; this is a matter of the actual statements in the paper), so I replied to “Graeme W” with:

> Much as I hate to agree with Mr. Connolley,

Dr. But apart from that, yes, you’re correct. This is an analysis of model simulations. The article makes no claim at all that the models are flawed, all it is doing is analysing the actual causes of a certain affect *in the models*. The suggestion that “climate models need to be reworked” appears to be an interpolation by our host, possibly based on the post at NTZ, which has made a similar error.


which drew a reply

REPLY: Yet, CMIP5 models still don’t match reality. So yes, they need to be reworked on many levels until they can properly predict climate with accuracy. The dialing in Arctic albedo and feedbacks (plus many other things) aren’t quite there yet. If they were, we’d see better agreement in graphs like this one:

Unless of course, you’d like to argue that models are “good enough” and need no improvement whatsoever.

So I think the shape of the argument for the defence starts to become clear: ground will shift away from what this article actually says, onto “models aren’t perfect”, which is dull. I replied (I’ve corrected the spelling of CMIP here):

That the CMIP simulations aren’t perfect would be agreed by all who work on model development. But that’s not the point here: which is that *this study* provides no evidence for that assertion; its entirely orthogonal to that idea, since its an analysis of model output. I don’t know what you mean by “dialing in Arctic albedo”; that appears to continue your misunderstanding of the paper. The paper isn’t suggesting the models should “dial in” the albedo at all. All its doing is presenting an interpretation of the model results.

which drew in return

REPLY: And you are honing in on a headline, not the body, tough noogies if it upsets you. The fact remains that

1. CMIP Models still have a poor understanding of feedbacks
2. CMIP Models still don’t have a handle on real-world albedo changes
3. CMIPModels aren’t matching reality as measured

Hence, they need to be reworked. I’m not going to change the headline simply because you interpret it in your own special way. Now run along and write up your usual smear.

So, yes, the defence shifts, and he also attempts to use the “headline defence”, which is weird, because although headlines in the Meeja are often crap, that’s because they’re written by subs. In this case, he’s written it himself, so “blame the headline” is no defence at all.

> honing in on a headline, not the body

Not at all. Later on you include: “This casts many of the assumptions made in earlier climate models deep into doubt. It’s back to the drawing board (again) for the modelers.”

This is as wrong as your headline. As I said, that models can be improved is doubted by none, but what we’re talking about here is this study.

> Please provide a clarification which would remove the ambiguity in your post.

Mmmm, this is difficult. (a) and (b) are wrong, because the study doesn’t talk about errors in the models. (c) is wrong, because the study isn’t really talking about nature very much. I think you’re missing the basic point: this study is about the interpretation of model output; its trying to work out what processes in the models are responsible for a certain result in the models. Models are complicated things; its often not at all easy to work out why they do what they do.

(that last abc bit is in response to some drivel by richardscourtney. My response has been disappeared, but Windchasers says much what I would have).


* Harry Potter and the Polar Amplification of Global Warming
* Polar amplification, again – wherein I somewhat prophetically note that People constantly get polar amplification wrong. That post does cast some doubt on the novelty of the Nurture thing, though.

Another year in stoats

Subtitled As the days of my life are but grains of sand.

I’ve tried to not-choose the knockabout stuff, which is all great fun, but ephemeral. But some months were thin and I had little choice.

* Jan: On happiness
* Feb: The sleepwalkers
* Mar: Man slumped after hitting wall (see-also)
* Apr: North Korea ‘may not be performance art’, say experts
* May: Syria: the West makes the usual mistake – which I drag out not because its brilliant, nor because time has proved me wrong.
* Jun: Saturn’s hexagon. See-also Earth from Saturn.
* Jul: Up three to nine. See-also Happy Birthday to Watts’ paper!, an event none of us would wish to fail to mark (I look forward to its second unbirthday too; speaking of which, don’t forget Climategate 3.0 – well, how could you?)
* Aug: This year’s sea ice considered unexciting
* Sep: AR5: cursory review of chapter 4 (cryosphere) mass balance of Antarctica
* Oct: Wyatt and Curry part II: not waving but drowning
* Nov: Thrust
* Dec: Climate science is interesting and fun

Overall, a fairly thin year I’d say. Which was part of the reason for writing this.