Place your bets: Make your call on 2010 global average temperature

If you want an accurate explanation of temperature change, then you want the output of an ensemble of general circulation models tuned to 20th century conditions. Any thing else is half-baked nonsense.

In particular, if you want some simple explanation, some one-factor explanation, some snappy little one-sentence explanation, ditto. All you'll get is at best approximations and at worst the usual half-baked nonsense.

I gave you the cartoon version of it -- the tradeoffs between GHGs (warming) and aersols (cooling) .. but that's clearly not good enough for you.

OK, then focus on the real thing. You want a good explanation of temperature changes, you want what the IPCC puts out, the summary of the ensemble of models they consider good enough to be used in their analysis. So here it is, again:




Note the blue bar. Note that is rises in the first part of the century. Note that it is does not continue to rise in the second part of the century. Only the red bar does that. The blue bar is predicted temperature absent anthropegenic forcings. The red bar includes anthropogenic forcings.

As an extras for experts, the width of the bar shows the uncertainty level.

Now, do the models fully explain the warming prior to WWII? No, as I said above, they don't fully explain ocean warming (though, as I said above, I believe that the most recent data would remove much of that pre-WWII bump due to instrument calibration issues that were uncovered after the last IPCC report.)

Do the models account for most of the warming prior to WWII. Yes. How does the size of the unexplained warming prior to WWII (the height of the line above confidence interval band) compare to warming after 1970? It is small compared to that. Can climate scientists then reasonably conclude that you are unlikely to explain recent warming without anthropogenic factors? Yes, note how far apart the blue band and the observed black line fall by the end of the century.

Is that the only piece of evidence that points to anthropogenic factors, that one little line? No, of course not. Look up "attribution" or "fingerprinting" in the context of global warming and you'll see plent of additional evidence.

In terms of complete explanations, that's it.

You seem to keep wanting simple explanations of what is an inherently complex phenomenon. Well, that's usually not a good thing to do. What you have here is the output of a detailed, complex explanation of a complex phenomenon. This is the right thing to look at, if you want "an explanation" of 20th century warming.

Let's go one step further: Are there any complete models like these that explain late-20th-century warming without anthropogenic factors? None that I know of. If you know of a general circulation model that does so, please post it. So, as far as I know, manmade factors are both sufficient and necessary to explain recent temperature change.

Now, beyond this, do you also require some half-baked simplified text to accompany the actual explanation? That is, do you also require some nice "story" to tell here?

Well, tough. I tried gave you one and you spat on it, although as far as I know the tradoff between GHGs and aerosols is the correct explanation of post-WWII lack of warming. So, as of now, you're looking at the explanation. The correct explanation of warming is that it is accounted for by complex, physics-based general circulation models.

The full extent of unexplained 20th century temperature change is shown above. It's the section of the black line that sticks up above the uncertainty band around the ensemble of model estimates. (And, in fact, for the 3rd time, I think it's actually modestly exaggerated above, but that's hardly relevant.) The models, as they are, largely explain the data, as they are. And the unexplained rise is quite small relative to late-20th-century temperature rise, none of which would be explained without the anthropogenic factors (the blue band cycles back down after mid-century, while only the red band continues to rise.)

EDIT: Maybe this will be helpful in understanding why there is no simple, one- or two-dimensional explanation of the facts. These are the inputs to the NASA model, below. The story I tried to tell you was about the tradeoff between GHGs and S02-derived aerosols. Those are, basically, the top and bottom lines in this diagram.

So, the answer to "why did temperatures rise prior to WWII" appears to be that the positive forcings on that diagram outweighed the negative ones.

Obviously, temperature does not reflect the forcing instantaneously, so to do this right ... well, you need to look at the output of a general circulation model. But if you want some "eyeball" explanation of why temperatures might have risen prior to WWII, then here it is. I'm not sure I could even tell you in detail what all of those terms are. But GHGs and aerosols are the biggest ones.






At the endpoint (this next one), the net forcing is the sum of a lot small positive and negative manmade effects. I can attempt to give you some simple story (at such-and-such a time, factor X was more important than factor Y), but in fact, that's not really very helpful.

But that's all historical. Eventually, if we continue with business as usual, the persistence of C02 in the atmosphere means that, eventually, we will be able to tell a simple story. Eventually (but not in the past), the cumulative effect of C02 is going to be so big that it will dominate the changes. For now, though, not so much.




SECOND EDIT: BUT I'M GOING ABOUT THIS ALL WRONG. LETS SEE WHAT HAPPENS IF WE TAKE THE ALTERNATIVE VIEW SERIOUSLY.

The mainstream view of this issue is that manmade GHGs cause warming, that the warming is going to have costly consequences. Modest (but significant) expenditures to reduce GHG emissions (2% of GDP by 2050, per the CBO) would be less expensive than continuing under business-as-usual and attempting damage control later.

Instead of explaining the mainstream view, let's take the alternative view seriously. Put it down as an actual hypothesis, and evaluate it as such.

Alternative view: There is some yet-to-be-identified Factor X, that explains late-20th-century warming. The evidence for X is so strong that it is not worth taking measures to constrain GHG emissions.

Now let's evaluate that alternative view. What would we have to do, in order to demonstrate that the alternative view is better than the mainstream view?


1) Prove that the greenhouse effect does not exist.

You can’t have twice as much warming as the data show. So, in order to say that something else is causing the warming, you must first show why man-made GHGs don’t cause warming. Otherwise, you’d get twice the warming.

So, task #1 is to show where all the scientists running these general circulation models have gone wrong, and why all of them have so grossly overstated the impact of C02 and other warming gasses. If you could do that, you’d get the Nobel prize. Nobody has yet done that. I therefore judge it is unlikely anybody will be able to get past Step 1.

2) Identify a factor that fits the data, in detail, better than man-made greenhouse gasses. This factor must do the following:

2a) Cause greater warming over land than over ocean, as GHG-based warming has done. That rules out any type of ocean-based heat-exchange mechanism (e.g., it ain’t one of the long-term ocean-based trends).
2b) Cause warming of the lower atmosphere but cooling of the troposhpere, as GHG-based warming has done.
2c) Cause much more rapid warming in the arctic than elsewhere, and little to no warming in the antarctic, as GHGs have done (and were predicted to do, back in the 1980s). This rules out any “urban heat island” or other proposed causes of changes in the observed data.
2d) Cause more warming at night than during the daytime, as GHG-based warming has done. This, along with direct observation, rules out variation in solar output as the driver of climate change.

I could go on.

Then, once you’ve first explained why GHGs don’t cause warming (which no one has done or will do), and you have your candidate factor X that isn’t ruled out by the known observations on the system (which no one has yet done to any convincing degree, thought people have tried, e.g., cosmic rays), you have to show that X fits the data better than the current candidate, man-made GHG warming.

Finally, you have to decide whether the evidence on Factor X is so strong, relative to the evidence for GHG-based warming, that it is worth, in effect, gambling the future of civilization on it, rather than pay 2% of GDP (by the year 2050) to limit our GHG emissions.

So, by all means, I hope you will start a thread on this. Start by demonstrating why GHGs aren't warming the planet. Then proceed to offer plausible candidates that explain the facts better than the mainstream view. At the end, we can see whether we feel like gambling our descendents' futures that this alternative view is correct, rather than reduce their incomes by 2% at some future point.

 

And back to being on target, yes, my most recent estimate (June update above) is about a 93% chance of a record for the GISS series. For the first half of 2010, all three series that I am tracking are above their calendar-year records.

 

You're years out of date. Of course, if you only listen to denialist sites, you'd never know that. But if you got your information from a somewhat broader spectrum of sources, say, PriusChat, you'd know that Douglass et al. made a whopping error, and that paper(s) pointing that out have already been published.

By using the search function, I see that I've explained or referenced an explanation of the Douglass et al. error no fewer than four times, here on PriusChat.

So, here's a decent explanation of what Douglass et al did wrong, post 58 in this thread, more than two years ago:

http://priuschat.com/forums/environ...t-decade-may-see-no-warming-6.html#post612353

Here's the gist of it. They took the predictions from 22 models, and treated them like 22 data points. Predictions come with a (typically large) uncertainty attached to them, datapoints don't. They ignored that uncertainty, just tossed it away. That's something that would get you flunked in Econometrics 101, but which apparently few hard scientists were familiar with.

Here's my prior explanation:

"If I ask 1000 economists to predict the average rate of inflation in the year 2100, and take the average and standard deviation of those predictions, what I have is a very precise estimate of the average prediction. What I most assuredly do not have is a very precise prediction. In fact, it's still just a guess. I should have no expectation that the actual inflation rate in that year will be close to that prediction. And if I then asked 100,000 economists, I'd get ten times more precision in my estimate of the average prediction. But the prediction itself would be no more precise than the first one. (And no more accurate, for that matter.)

That's the big booboo Douglass et al. made. They assumed that if you had (e.g.) 1000 models, and averaged them, the resulting prediction would be 10 times more precise (ten times smaller error bars, ten times smaller standard deviation) than if you had only averaged ten models. (And therefore, but left unsaid, if I averaged a million model runs, I'd be claiming to have predicted temperatures with near-zero uncertainty). But that's just nonsense. What I'd have is a very precise estimate of the average prediction, to which I still need to add the uncertainty of the typical model. (Ie, a guess is still a guess.) Which Douglass et al. didn't do."

That's been said in the literature, not just by Realclimate and by amateurs like me.

Here's a critique as published in the literature, that does the calculation correctly. I believe it might have been published in the same journal and the original Douglass et al article.

https://publicaffairs.llnl.gov/news/news_releases/2008/NR-08-10-05-article.pdf

And here's a fact sheet, so you can get the basics without reading the entire paper:

http://www.realclimate.org/docs/santer_etal_IJoC_08_fact_sheet.pdf

Not that there isn't some genuine research interest in this issue. But it's just wrong to assert that the observations, at this point, lie outside the (properly calculated) band of uncertainty around those model predictions. First, you'd have to talk about the significant uncertainties in the data -- all the old stuff comes from radiosonde data, which had lots of problems. Then you have to address the statistical issues correctly. Which Douglass et al didn't do.

 

For what we were discussing -- historical temperature trends -- that's pretty much on the mark. You've got the particles embedded in ice, and once you're into the satellite era, you've got satellite-based measures of total aerosol load.

Once you get to the most recent years, though, the picture is less clear:

Global Aerosol System 2000-2007 : Image of the Day

That's total aerosol. Aerosols are complex. For the trend in reflective aerosols like sulfates, I'm pretty sure the Chinese require S02 scrubbers on their coal-fired plants, same as we do. That may be a new development, but I'm sure I read that somewhere.

Still, when it gets to assessing the tradeoff, I'm back to square one and/or that question has already been answered. The only way to do that is with a general circulation model. That's the forcings graph I posted above. Roughly speaking, the net manmade aerosol effect is the blue line net of the black line (reflective aersols less the impact of black carbon). The gray line is, I think, mostly aerosols from volcanic eruptions. In any case, the manmade aerosols did, in fact, increase toward the end of the period. But not enough to offset GHGs.

 

Bob,

RealClimate: Of buckets and blogs

Probably more detail there than you'd care to see, but it boils down to calibrating various antique and modern methods for measuring ocean surface temperature. As it turns out, different countries favor different methods, and therefore the mix of methods used to measure temperatures in the historical data shifted abruptly prior to and after WWII, as international shipping was disrupted.

I'm sure this is still being worked on, and the issues have been known for years (note that they reference the last IPCC report for a discussion). It's just that this particular quantification of the issue came out after the last IPCC report.

 

August update. If temperatures cool further, it won't be a record in the GISS series.

 

September update.

At this point, all three series (GISS and the two satellite series) are between 3 and 4 hundredths above the prior annual record. The current temperature in the GISS series remains just barely warm enough to be on track for a new record. Both satellite series show current temperatures well above the level needed to set a new record.

 

October update. All three series, to date, remain slightly above their respective all-time highs. Cooling relative to earlier in the year is evident, consistent with the current La Nina conditions.

 

November update. At this point:
The GISS series will almost surely set a record this year.
The RSS series probably will not.
The UAH series is too close to call.

Aside from the higher variance on the satellite series, my observation (among these series, but also looking at the NCDC and HadCRUT3v series), is that the omission or downweighting of polar data matters significantly this year. (RSS omits the poles. UAH has no specific mention of the poles, but the RSS writeup makes it clear that they get unreliable date near the poles).

The NCDC states that there was a strongly negative Arctic Oscillation in November. That tends to move cold air out of the Arctic and into the eastern US (among other places). Which we are feeling even now. And thereby move warmer air into the Arctic.

State of the Climate | National Overview | November 2010

That would tend to make omission of the Arctic important. Omit the Arctic and you only see the cold side of that exchange. I'll have a (half-baked) test of this when NCDC and Hadley release their final series. NCDC has appropriate inclusion of the Arctic, Hadley omits the Arctic entirely. Judging from what I see, I'd bet on a new record for NCDC and nothing close to a record for Hadley.

Anyway, the upshot is that the focus of this bet -- the GISS series -- will almost certainly set a new record this year.