Saturday, August 15, 2009

Another comment on McLean et al

It's been said before, that there is one way of getting things right, but an unlimited number of ways of getting things wrong. Thus, it is not entirely surprising that there is another comment on the McLean et al paper apparently submitted to JGR (found here on the arxiv). However, this appears to be a rather ham-fisted attempt at rescuing the main conclusion of the original paper (that ENSO, and in particular the SOI time series, can explain a large part of the global climate change signal over the last century) while contradicting the analysis on which it was based.

The manuscript is based on a linear regression in which various forcing time series are regressed against global temperature. The novelty here is that the authors propose to use the cumulative SOI series as well as the ordinary one. Nowhere do they justify this specific choice, (ie using two versions of SOI but only one of each other forcing), but I'll let that pass for now. Here are the relevant results, based on their "T2" equation for the full model (where all forcings are considered):

All I have done here is taken some of the forcing time series that S+C used, and multiplied them by the regression coefficients that they obtained. SOI and cSOI are the plain vanilla and cumulative Southern Oscillation Index data respectively, both with a 12 month smooth. AF is all anthropogenic forcings, which as well as the major contributions from GHGs and tropospheric aerosols, also includes more minor things like land use. Since S+C's paper explicitly discusses the the influence of GHGs alone, I have also separated out the well-mixed GHGs from the other anthropogenic forcings (which they did not show). The numerical values printed beside each line refer to the linear trend of each series over this interval 1960-2008 (which was their choice).

So, their results as plotted above suggest that GHGs are responsible for about 2/3 of the trend, with cSOI contributing about half that much (1/3 of the total trend). I don't believe for a minute that the cumulative SOI is that important, and the authors admit that the formal confidence intervals for their regression coefficients are much too small "due to collinearity" (I'd expect autocorrelation is the real problem here, the regression should already account for collinearity, but I have not checked their calculation). But regardless of that, what I'm really interested in is the bizarre way in which they try to interpret these results. They start by observing that their regression coefficient (of about 0.2) for the instantaneous response to anthropogenic forcing is much smaller than the typical equilibrium response (of about 0.8) of climate models which reproduce the recent warming. Then they speculate that this "difference" might be explained by a response to ENSO, which would mean that ENSO would have to be responsible for most of the warming. Of course, this flies in the face of their own results that reproduce the recent warming with ENSO contributing about 1/3 of the total (and clearly less than the anthropogenic effect). They finally conclude that
these results are clearly inconsistent with the claim that ”[M]ost of the observed increase in global average temperatures since the mid-twentieth century is most likely due to the observed increase in anthropogenic greenhouse gas concentrations” [IPCC , 2007]
Well, there's no accounting for arbitrary numerology and make-believe. Their own estimate, presented right there in the paper, is that despite giving the SOI two bites at the cherry by including two version of it in their regression, GHGs still account for about 2/3 of the total warming (and double the total contribution from ENSO). They might as well have said "we did a regression, but didn't like the results, so here are some conclusions we made up earlier".

I guess that McLean et al will welcome this "proof" that their conclusion was right even though their research was wrong (this new paper attributes a mere 0.02C of warming to the SOI itself). It seems that some people start with the answers they want and then try to find arguments to support them (as mt says "very not the IPCC" is really the ultimate aim).

11 comments:

Hank Roberts said...

Author! Author!
http://www.google.com/search?q="David+R.B.+Stockwell"+"Anthony+Cox"

David B. Benson said...

You would hope that those who have not studied enough statistics themselves would at least consult a professional statistician.

I do that even for my white noise, non-autocorrelated problem.

Erl Happ said...

If you had half a clue as to why the tropics warms and cools, where it warms and cools, what drives the flux in cloud cover that controls the amount of sunlight that gets to the surface of the ocean and what this has meant in terms of the historical rise and fall in temperature by latitude, then, and only then, should your speculations invite the attention of those who make it their business to be concerned with matters of this sort.

It is the change in the underlying forces that determine sea surface temperature that must be assessed if we wish to attribute the change near surface atmospheric temperature to any particular 'forcing'.

The atmosphere can not heat the sea. On the other hand the temperature of the sea very plainly drives near surface atmospheric temperature.

Steve Bloom said...

Congrats, James, you've been visited by a climate crackpot of the first water. Erl (Happ) has few peers.

James Annan said...

Thanks, I was wondering what on earth had hit me :-)

Erl Happ said...

From another place Steve Bloom visits, a comment that seems appropriate: "Apparently a crank/crackpot is anyone who does not agree with the current wisdom, no matter how reasonable his argument.

Ad hominems (attack the man rather than the argument) are usually more dangerous to the issuer than the recipient."

Fact of the matter is that until you can explain what makes the tropics warm and cool you can be as sophisticated as you like with the mathematics but you will be unable to quantify the contribution of tropical energy gain to surface temperature change elsewhere on the planet OVER ANY TIME INTERVAL. Until you do that, you are not in a position to say with hand on heart: "we can't think what is causing this warming other than greenhouse gases", and convince anyone. And that my friends is the situation you are in.

Don't like what I say? Well, the remedy is to address the issue.

Let me remind you, you have yet to find a direct link between greenhouse gas concentration and the temperature of the troposphere.

But, perhaps you have now moved to a place where observational evidence for what you say is simply unnecessary.

I, at least, am trying.

EliRabett said...

Yes Erl, you are very trying.

The point is that the atmospheric back radiation limits the sea surface cooling rate by returning a portion of the emitted IR energy.

Over the near term long run the temperature of the ocean results from the balance between absorption of solar energy and cooling by radiation.

Hank Roberts said...

> Erl

Of the blog formerly known as "CO2Skeptics"

Erl Happ said...

Who was it said "I don't like what you say but I will defend to the death your right to say it".

Obviously didn't have a big impact on the custodian of this blog.

James Annan said...

Cry me a river, kook.

I haven't even deleted any of your rambling comments, yet.

Erl Happ said...

James,
Well then, my corrections to Hank Roberts and the links that would have elucidated the situation must have gone into the aether entirely. Obviously I am both a crackpot and a kook.

So, lets try you on this one:
The atmosphere is not amenable to modeling that treats the globe as a closed system. Our understanding of atmospheric processes is elementary. Mathematicians who do not appreciate that the basic parameters driving climate are externally imposed and forever changing, are a hindrance to progress and best employed elsewhere.

For the full story see http://climatechange1.wordpress.com/

This story is called: Wherefore art thou NiƱo?'