Statisticians reject "global cooling" claim

[richard schumacher]

AP IMPACT: Global cooling? Statisticians reject claims that climate trend is shifting

By SETH BORENSTEIN , Associated Press
Last update: October 26, 2009 - 2:39 PM

WASHINGTON - Have you heard that the world is now cooling instead of warming? You may have seen some news reports on the Internet or heard about it from a provocative new book. Only one problem: It's not true, according to an analysis of the numbers done by several independent statisticians for The Associated Press.

[...]

"To talk about global cooling at the end of the hottest decade the planet has experienced in many thousands of years is ridiculous," said Ken Caldeira, a climate scientist at the Carnegie Institution at Stanford.


AP IMPACT: Global cooling? Statisticians reject claims that climate trend is shifting | StarTribune.com

[TimBikes]

Rather than look at trending, which says nothing in and of itself about the effect of CO2, the more interesting question would be to have statisticians look at the correlation between temperature and CO2 vs. temperature and other factors over the past century.

I have seen analysis that suggests co2 correlates at r^2=0.44 whereas TSI = 0.57 and AMO + PDO correlates at an r^2 = 0.85.

I wonder why they didn't undertake this sort of analysis? It probably wouldn't have given them the message they wanted to convey.

[Philosophe]

Originally Posted by richard schumacher

"To talk about global cooling at the end of the hottest decade the planet has experienced in many thousands of years is ridiculous," said Ken Caldeira, a climate scientist at the Carnegie Institution at Stanford.


AP IMPACT: Global cooling? Statisticians reject claims that climate trend is shifting | StarTribune.com

What about this one?

CO2 Is Green - More CO2 Results in a Greener Earth

What is the most ridiculous: the industry trying to make us believe this or people who do believe it? (I don't need an answer...).

[iota]

People will do and say almost anything to justify their own inaction or poor choices...

[chogan2]

"They" don't do that because it would be nonsensical to rely on single-factor non-physical (correlation) models when physics-based models are available.

"They" have models based on the laws of physics, accounting solar irradiance, atmospheric composition, heat uptake by the oceans, aerosols, clouds, water vapor, and so on. "They" also don't confuse cyclical variations in heat uptake by the oceans with actual inputs of energy to the earth.

They can use their physics-based models to parse out the contribution of changes in (e.g.) solar irradiance to overall climate forcing. See below: big bar at left is C02, little bar at right is change in solar irradiance, time period is centuries. If they'd done the same thing for the past 30 years, the solar bar would have been slightly negative on net, because we're at pretty close to the solar minimum right now.

Data @ NASA GISS: Forcings in GISS Climate Model

"They" appear to have shown that, within this context -- having all factors on the table -- manmade greenhouse gases are necessary and sufficient to explain 20th century warming, based on the underlying physics. Without them, you need some unknown "X" factor to explain the warming, with them, you get the warming. With them AND some "X" factor, you'd have to get twice the observed warming. (So, if someone thinks something else is causing the warming, someone needs to explain why C02 is not a greenhouse gas and why all those physics-based calculations about the impact of C02 are wrong.)

Amateurs, by contrast, love correlation analyses, because they don't have to know anything to be able to calculate a correlation. Just put the data in Excel and calculate. And they aren't bothered by having to check whether or not the correlation observed is even close to being physically plausible as a source of warming.

Even within the narrow context of purely statistical models, most amateurs don't understand the basics, which here would include omitted variables bias. (Translation: even if you're going to run a simple-minded statistical model, it at least needs to be a multivariate regression with all the relevant determinants included as the same time, not a one-variable-at-a-time correlation. If multiple factors are affecting temperature and you run correlations, you have no idea what the correlation coefficient actually means even in a purely statistical sense, let alone passing the much greater challenge of demonstrating a plausible physical mechanism that might generate such correlation.)

So the short answer is that "They" ignore correlation analyses because they have far better tools for looking at these issues.

So in this particular case, why ask a statistician? Because the question was purely statistical: can you infer a downward temperature trend from the most recent N years of annual temperature data. The question was purely one of curve-fitting.

The answer to that is no, you can't.

Why not? First, to assert global cooling, you have to cherry-pick the start year at 1998. If you start with 1997 or 1999, and stupidly play connect-the-dots between first and last year available, the resulting line still slopes upward. The only way connect-the-dots shows a downward slope is if you pick 1998 (or, if you rely on NASA GISS instead, 2005, since GISS shows 2005 as the warmest year on record. But, somehow, connecting dots ten years apart makes the argument seem plausible, while connect dots 3 years apart makes the argument seem ludicrous. Even though, statistically, they are not that far apart. So, Hadley it is, and 1998 it is.)

That means, if you're going to assert that the globe is cooling, you only have 11 years of data.

Next, assume that you're going to do this properly and actually fit a linear trend, instead of playing connect-the-dots. (With connect-the-dots, you only have two data points, first year and last year, which means you don't even have enough data to perform a statistical test.) If you fit a linear trend to the NASA GISS data, you get an upward slope, if you fit it to the British Hadley data, you don't (with the difference being due mainly to the superior way in which GISS gap-fills the arctic temperatures.)

Does either dataset have enough statistical "power" to be able to reject a zero trend, if the true trend were as projected by (e.g.) the NASA model, roughly 0.3 degrees F per decade? Answer: No. With either dataset, even if the true trend were as stated by NASA, zero would still be within the 95% confidence interval for the trend estimated from the annual data from 1998 to 2008.

Upshot: even if you wanted to focus on just the trend since 1998 (and ignore the fact that the decade average 1999-2008 was the warmest decade in the instrumental record), you would have had to have seen an enormous drop in temperatures on average in order to have found a statistically significant downward trend over the decade (that is, global cooling). The data aren't even close to showing that.

My prediction is that all this "global cooling" crap will disappear in the next couple of years anyway. Both NASA GISS and the Hadley center went out on a limb and suggested that as we move from La Nina to El Nino and the solar cycle bottoms out, we ought to be looking at new record annual highs in the next few years. Looking at the GISS data, if the rest of the year is merely as warm as September, 2009 would be the 3rd warmest year on the record, while if the with-year warming trend continues (due to the ongoing development of El Nino conditions), 2009 would end up being quite close to being tied with 1998. So those predictions seem plausible. Guess we'll see.

[Shawn Clark]

Originally Posted by Philosophe

What about this one?

CO2 Is Green - More CO2 Results in a Greener Earth

What is the most ridiculous: the industry trying to make us believe this or people who do believe it? (I don't need an answer...).

Don't tell the skeptics, but I've heard that website is a trap to ensnare the feeble minded so that they can be put before a "Death Panel." Rumor has it that Nielsen is in on the diabolical government plan as well...they are carefully tracking Faux News viewers for the Administration so that they can be sent to reeducation camps in Detroit. (Queue the Ron Paul supporters exclaiming: "I knew it!")

[TimBikes]

Originally Posted by iota

People will do and say almost anything to justify their own inaction or poor choices...

True - reminds me of the healthcare debate.

You've got more than 70% of the population who is overweight or obese and yet wants government to pay for their medical care.

[icarus]

Just to clarify a popular misconception.

Most of us are not asking "the government' (Or anyone else) to pay for our health care. What we are asking for, longing for, (and in desperate need of!) is health care that is, fair, affordable, and that covers everyone, that is no skewed in favour of Big Insurance, Big Pharma, Big for profit Hospitals etc!

You may love the health insurance you have now, but picture you and your family if you loose if due to job loss, or if they don't cover you due to "pre-existing conditions" even though they have gladly taken your premium dollars over the years! As long as our health care is in based on the ability of having a good enough job to get health care, were screwed!

You may not like the Canadian system, or the British system, or the French system, or the German system, or the Japanese system, or nearly every other system in the G-20, but the FACT is that on average we in the US pay about twice as much both as a percentage of GDP an on a per capita basis than the rest of the G-8 or G-20.

The damn GOP is already blowing BS about "how expensive" it will be to pass ANYTHING! Most economists agree that the situation we have now is the most expensive option! Doing nothing, just allows the situation to get worse!

How many people went bankrupt due to medical issues in the US this last year? Thousands! How many in Canada/France/Germany/Japan combined went bankrupt due to medical issues in the last year? ZERO!!!! The GOP is the party of "We don't want to pay for anything especially if it helps those that have less." This BS about people who are making $500k per person or $1million per couple not being able to pay a bit more makes me want to puke! Those that have gotten so much benefit from the misguided economic policies of the last 3 decades are just plain selfish! And I'm sorry but the middle class folks that continue to buy the line only have themselves to blame when their health care goes south along with their jobs as "corporate America" looks for the cheapest alternative!

Icarus

[Shawn Clark]

The conservative/GOP claims on nearly everything are complete BS/lies. Healthcare is certainly an example of that. They are running SCARED of a public option because the regional monopolies conservatives have favored would be EVISCERATED by a functional public competitor. Conservatives FEAR the free market result of folks actually having reasonable and informed choices.

If conservatives actually believed what they profess, then they would not fear a public option because their mantra is that govt. is inefficient and always inferior to the private sector, yada, yada, yada. Now they claim that the same "inefficient govt." would outcompete their sleek private system.

Personally, I think the Medicare plus 5 option sounds attractive.
If we actually get a universal system and real public option, health care costs in this country, and the expense to govt. will decline, not increase.

One things is certain, we can't do any worse than what we have at present, the least efficient health care system on the planet.

[TimBikes]

Medicare is a looming disaster, so I wouldn't want to base anything on that. That said, I agree there are problems (as Icarus rightly points out) such as the currently uncovered population, the rising costs, the linkage of insurance to job, problems with pre-existing conditions, etc. However, I doubt more government intrusion of the sort being proposed is going to solve the problem.

If government has a role, it would be (in my opinion) to create a more competitive market for insurance through better health insurance company transparency, setting a "floor" on what defines a basic policy of what should be covered and perhaps rating various policies (much like they rate energy star appliances or ratings on tires), and opening up insurance competition across state lines.

The problem of the uninsured is vexing. Certainly we don't want the status quo (emergency rooms for sniffles) to persist. Pre-existing conditions are tricky too. If there is no consideration of them however, why would I ever buy insurance until I was ready for treatment? Think about it. It's like buying fire insurance after my house burns down. Good for me, but that really isn't insurance.

As for rising costs - the medicare model of continuing to squeeze providers is not a valid solution. The solution has to involve people taking responsibility for themselves -- bearing some portion of the cost through high deductible plans for instance would greatly discourage mis-use of the medical system and strongly encourage personal responsibility with regard to personal health (food, fitness, etc). Cost control will also have to address tort reform.

I have ZERO confidence that the current 1400 pages of the health care bill are going to accomplish much of any of this. It is simply a large expansion of governement spending and control -- not a solution of any sort.

A better solution would be tackling each one of these issues one at a time, bill by bill. Not with a misguided "omnibus" bill that frankly, no one in congress has even read, much less understood.