From Alan Tomalty we have this question:

No one has been able to separate out the amount of warming due to natural causes and amount due to CO2. Therefore I propose looking at all temperature stations that go back to at least 1880 for their daily maximum temperatures or averaged maximum temperatures. Then you would have 140 years of data, 70 years before 1950 and 70 years after. Since the tangent on the exponential curve of man made CO2 emissions started to move upward violently ~ 1950, this means that before 1950, essentially CO2 had ~ 0 effect but after 1950 a constantly increasing effect if you assume it has an effect at all.

So I propose doing 3 separate regression graphs for each of the 140 yr data temperature stations. 1 graph before 1950, 1 graph after 1950 and 1 graph overall. Do a regressional analysis on each graph with dependent variable as the temperature and date as the independent variable.

The 2 other things that are allowed to be considered are natural warming as 1 other independent variable and CO2 warming as another , but these 2 variables are not graphed. They exist in analysis below. The regression coefficients you come up with for each station are then turned into C degree changes.

Here are the definitions for the 3 way regression of CO2 intervention analysis. Natural warming effect = regression coeff in 1950 graph. Overall effect = regression coeff in complete time series graph. Natural diff is the natural warming effect that you can project into the future ie: is that part of the warming after 1950 that was or will be caused by natural factors.

The CO2 diff is that part of the warming that was or is caused by net CO2 in the atmosphere. Add them together and you have the regression coefficient of the overall complete times series graph which stat you already know. Get an average of Natural diff and CO2 diff from all your temperature stations and then you have the global average difference that CO2 makes versus natural warming.

Here are the equations.

1) Natural diff = Natural warming effect- (CO2 warming effect – overall effect).

2) CO2 diff = CO2 warming effect – natural warming effect.

3) Overall effect = Natural diff + CO2 diff

Interesting idea, but there are some complications. First, regressions can’t prove cause, but when used to suggest cause it makes the assumption of fixed linear forces operating on or overriding other “unimportant” forces. To a first approximation, fixed strictly linear forces (in any system) might be workable—but cannot be convincing without external evidence of both the linearity and the abrupt *change* in linearity to a new linearity.

Here are some other considerations.

Pick a spot. Your residence, for instance. The best way to ascertain temperature changes at this spot, which is not necessarily a possible way, is to (1) gather as long a series of measurements from that spot as you can, (2) define what a *trend* or *change* is, then (3) just look. Either the definition has been met, or it hasn’t.

Chances are the definition of the trend will be too strict if it is formal-model based. A regression coefficient, for instance, is too formal. Again, why would we expect temperature to fit a strict regression from any arbitrary starting point (when your series started) to some arbitrary end point (now)?

Why not just look at what the temperature actually did at your location? Then we can say “Here is what the temperature did.” There is no notion of cause in this, just observation.

We have to remove classical statistical thinking from understanding change. With linear trends, we get the idea the trend is realer than Reality, that the departures from linearity are “noise”. This is not so. Assuming error-free measurements, the measured temperature happened, it was felt by whatever was there at the spot. The lived experience (if we may) of the things embedded in the air is the only reason we’re interested in air temperature in the first place.

We can always look and see what happened. Not that we want to avoid the difficulties of saying *why* changes occurred, but we want to be careful. For instance, many have pointed out how certain structures, even including cities with all their pavement and so on, have grown up around temperature measurement stations. The structural changes obviously affect the temperature.

Some then ask what the “real temperature” “was” in the absence of these structural changes. Well, that question makes a kind of sense, but it’s irrelevant. The things at the station experienced the temperature that was measured. And if the structural changes were man-made, then the climate (the statistical average) change is also man-made.

I stressed error-free measurements, because many temperature measurements do have error. That error can be accounted for by formal means, which thus *increases* the uncertainty of the measurements. This uncertainty must always be kept when analyzing any change.

Another form of “error” is change in measurement type. The way it used to be measured is not the same as now, for instance. Again, this can be analyzed formally, and it, too, increases the uncertainty of any change claimed, an uncertainty that must be kept in the final answer, which *must* then come with a plus-or-minus.

With all those warnings in mind, here is where I more or less agree with Tomalty’s strategy. Gather from as many places as possible the longest series of measurements, accounting for the types of error and changes in measurement types just mentioned. “Line” the series all up, and see how the changes (using the definition of change made above) happened at each station.

Some of the places might have trends or changes and some with have others, and some will have none. This is far superior to looking at some sort of global average temperature (GAT), which is experienced nowhere and which never has any notion of uncertainty attached to it.

And, incidentally, it has to be *observable* uncertainty, not *parametric* uncertainty, which is irrelevant and always much lower than observable uncertainty. This is how many fool themselves. See this link about removing parameters.

If it turns out that stations of interest, like your residence, did not change, or could not be said to have changed with sufficient certainty after accounting for error etc., then what has global cooling been to you? Nothing.

Sorry, I meant global warming. No: climate change. Climate disruption, rather. Or was it climate apocalypse? Whatever.

*To support this site and its wholly independent host using credit card or PayPal (in any amount) click here*

Categories: Statistics

Global Meshugas.

Interesting intellectual exercise. Naturally, we must use only the approved, properly adjusted data. Raw data is so… icky.

If you take a course in thermodynamics you will learn that temperature is an intensive parameter and intensive parameters do not scale. Therefore, global average temperature is physically meaningless.

Carbon dioxide should have no warming effect, where warming is an actual increase in the internal kinetic energy of a sample of matter. As an IR active agent, it will screen out incoming IR in the appropriate bands as well as screen out outgoing IR. IR is *light*, not heat; though you can covert heat to light and light to heat.

In the atmosphere, the IR effect of carbon dioxide is like that of water, though the effect must be smaller since there is *a lot* more water in the atmosphere and some of the carbon dioxide will be absorbed into suspended liquid water drops. Thus, it will shift the lapse rate (change in temperature with height above the ground) away from the ‘dry’ value toward the ‘moist’ one. That does not have to change the surface temperature, at all; which will be determined by the actual conditions at the surface at that location and at that point in time.

And, as Ray says, averaging the average that is temperature has no meaning physically.

Just a note to emphasize the radical — the fundamental — nature of Matt’s critique, which is only incidentally about ‘global warming’ or CO2:

it is provable — certain — that our root approach to many questions is seriously misguided.Matt’s work in the philosophy of statistical and logical analysis represents a deep and far-reaching improvement in the way we could do science, if we but will. We all owe him for this achievement, even though some or many of us will not live to reap its benefits.

It is not picking on Mr. Tomalty’s proposal to remark that his is the current ‘serious’, approved way to look at ‘serious’ questions. Which is to say, the way we have all been taught relies on a collection of inferences and assumptions that are just about as correct as the Doctrine of the Four Humours. Thus, our first step out the door is in the wrong direction. Our inferences, arguments, and conclusions — even our initial questions — become over-certain and unreliable, perhaps wildly so. When we get something right, it’s because of something else that we accidentally did as well.

I definitely want Matt’s regular readers to appreciate the depth of his accomplishment. It is nothing less than a radical and beneficial critique of how we approach problems, with a way forward that avoids known paradoxes and impossibilities, both mathematical and philosophical, in the assumptions, inferences, and logic we have been taught.

Of course, Matt as a person and a thinker is more than this giant step forward in how we can look at and know things. He has interests and passions beyond that, obviously.

Still. We knew him when.

Anybody with half a brain who spends any time at all looking at the supposed evidence purporting to underlie the “Catastophic/Dangerous Anthropogenic Global Warming” CONJECTURE can’t help but conclude that it’s a bunch of nonsense— yet one more episode belonging to “Extraordinary Popular Delusions and The Madness of Crowds.”

Ray,

And if you had taken a course in statistical mechanics you would know that temperature is itself an average.

Global warming hypothesis doesn’t depend upon existence or otherwise of global temperature. The models provide estimates of local warming.

It’s a nice sounding idea but, interpretation aside, the raw data doesn’t exist in a credible form so the whole issue is moot.

On first looking into the global warming idea many many years ago.. I used the microfiche collection at the U of Alberta in Edmonton to find temperature records in old small town newspapers using some in Texas and some in Alberta. I then compared the data I found to the official record. The newspaper record is, of course, both discontinuous and inconsistent, but for the dates and places for which I found data, I found no consistent relationship between the numbers published then and the officially recorded data “now” (where “now” = ca 1998).

More recently (last year) I found a proxy that I rather like: the raising and lowering of the boundary between the tropo and strato spheres. See http://winface.com/oldwin/amt/clim1.html The conclusion? No 100 year warming, no 100 year cooling.

Paul, thank you for your work. I have had similar problems even examining the data on the internet, once I could find the all time high for a given day from the states weather data, yet it did nearly always above the new record high that the media and NOAA reported, guess what, you can no longer find that data.

“And if you had taken a course in statistical mechanics you would know that temperature is itself an average.”

What has that got to do with intensive parameters not scaling?

What has the real temperature done in the last 20 years? I hope you write about this soon.