Numerical Models, Integrated Circuits and Global Warming Theory

American Thinker ^ | February 28, 2007 | Jerome J. Schmitt

[expatpat]

I think you have a good point there.

[edsheppa]

Historical data is perfectly good for validating a model

But not if the model is inferred in whole or in part from that historical data. That can easily happen even without the modeler's knowledge. The only sure way to avoid that problem is to test a prediction about data as yet unknown.

[lmailbvmbipfwedu]

Might I direct your attention to

Pikey, O.H., Pilkey-Jarvis, 2007, Useless Arithmetic, Columbia University Press, ISBN 0-231-13212-3.

For your reading enjoyment. BTW the subtitle of this book is "Why Environmental Scientists Can't Predict The Future"

[Tolik]

Can't claim the credit: I read a fascinating take on the modeling problem. Let's say that you have 100 parameters (I'd love to know if it a stretch to say that such complicated system as global climate can be described by 100 variables). And lets assume that we know each of the variables with 99% certainty (unbelievable certainty, but lets assume for the sake of this argument). The resulting certainty then is .99^100 = 0.3660 i.e. 36.66%.

37% is excellent if you play lottery. Is it good enough justification to spend trillions of dollars?.

BTW, if anybody can say that its wrong to argue this way, I'd like to be educated on why.

[Tolik]

Just wait 10 years and see if any model works. Its only prudent if the price is in trillions dollars, isn't it?

[Lonesome in Massachussets]

Not true. Suppose that you want to calculate , say, the co-efficient of expansion of a metal as a function of temperature. You could perform an experiment and vary the the temprature systematically and come up with an answer. OTOH, if someone had reliable observations, you could simply fit your model to these observations. If you knew what you were doing, and knew the limits on the accuracy of observations, given a temperature, you could predict the expansion, and state reliable upper and lower bounds.

It someone gave you a temperature outside the range of the observations however, you would be on shaking ground. The metal might melt at a very high temperature, for instance.

Astonomers have *nothing* but historical data yet they can make very accurate predictions about planetary ephemeris, with reliable limits on the expected errors. Sixty years ago astronomers were able to predict (to within a few blocks) which streets on Manhattan would see a total eclipse and on which it would be partial.

[Publius6961]

Another keeper!

I particularly loved these two paragraphs, Which should be self evident to anyone with just a conversant familiarity with the hard sciences:

Almost all semiconductor manufacturing processes occur in closed vessels. This permits the engineers to precisely control the input chemicals (gases) and the pressure, temperature, etc. with high degree of precision and reliability.

Closed systems are also much easier to model as compared to systems open to the atmosphere (that should tell us something already). Computer models are used to inform the engineering team as the design the shape, temperature ramp, flow rates, etc, etc, (i.e. the thermodynamics) of the new reactor.

Nonetheless, despite the fact that 1) the chemical reactions are highly studied, 2) there exists extensive experience with similar reactors, much of it recorded in the open literature, 3) the input gases and materials are of high and known purity, and 4) the process is controlled with incredible precision, the predictions of the models are often wrong, requiring that the reactor be adjusted empirically to produce the desired product with quality and reliability.

There was an article recently by a statistician, who pointed out that use of statistics analysis by amateurs can quickly lead to non-sensical results.
For example, if 3 assumptions are made which are 99% correct, the results can only be correct, significantly less than 99%.
When just a dozen factors are 99% correct, the results are only about 50% reliable.

With climate having dozens, perhaps hundreds of relevant factors, most of which are guessed at or ignored altogether, how reliable can these computer "model predictions" be?

[edsheppa]

You missed the point. I didn't say that models inferred from historical data are necessarily unreliable. I said it is unsurprising, and no indication of reliability, that a model fits data from which it was inferred. One should infer reliability only if the validating predictions are independent of the model.

[AFPhys]

As a scientist who has often used modeling in research, I concur that models have to be regarded with great suspicion -- unless there is LOTS of broad empirical data to back it up. It reminds me of the old saying about Magnetohydrodynamics calculations:

"It takes a genius to get computational results from these equations -- and a fool to believe them."
---<>---<>---<>---<>---<>---

And MHD of any earthborne system is child's play when compared with the vastly greater complexity of global climate/ weather. Even the Sun is more uniform and is probably more calculable than Earth, due to the Earth's continents and diverse terrain and diverse components.

By the way... is MHD of any use in describing auroras?

[expatpat]

I don't know. I'm basically a solid-state physicist (mostly theoretical) and digital signal-processing guy.

[neverdem]

thanks, bfl

[ggekko60506]

Great Post.

The author's analysis is devastating critique of the methodological fallacies that lay behind Climate Modeling as it is currently practiced.

There is also another major flaw in the current practice that the author did not mention that should warrant the summary rejection of the use these models in Public Policy.

The builders of the models on which the IPCC and other Reports have been based have refused to publish the input parameters and algorithms used to generate predictive results. Peer review is the backbone of scientific integrity and the lack of peer review outside the small coterie of model builders is a big of a red flag as can be imagined for any type of predictive model.

When the builder of the infamous "hockey stick" model was asked to publish his model assumptions he played the "political persecution" card. The highly and odd defensive reaction is indicative of how badly these models have been cooked for political purposes.

[IncPen]

... ping

[Former Proud Canadian]

Mark

[Lonesome in Massachussets]

Not true. The internal consistency of the historical data just needs to be available. Astronomers know pretty well how accurate their predictions are, because they know how good their observations are. Calibrations are just a special kind of historical data, selected to allow low variance estimates.

I agree, models are less reliable outside the range of inputs over which they are validated. Anyone who understands modeling and statistical technique must be hightly skeptical of the entire Global Warming ideology.

[listenhillary]

Bump!

[MosesKnows]

Perhaps the scientist uses a ROM and the engineer uses a WAG.

The Difference Between an Engineer and a Scientist

---

An engineering student asked his professor the difference between an engineer and a scientist. The professor explained it this way.

Ask a scientist and an engineer the following question. Imagine you are on a bus and seated exactly nine feet away beckons a beautiful naked lady.

How many bus stops will it take you to reach the beautiful naked lady if the distance between you is decreased by exactly one-half at each bus stop?

The scientist quickly answered that an infinite number of stops would not permit you to reach the beautiful naked lady because the distance is only decreased by half at each stop.

The engineer answered that after about ten stops you will be close enough for all practical purposes

[Reeses]

To solve the global-warming puzzle, we first need to learn much more about the precipitation-system puzzle.

Great article! Note they are only talking about natural clouds and snow. Using technology there can be cost effective man-made clouds and snow as well. That changes everything. We dam mighty rivers and build hundred-mile lakes so we aren't at the mercy of mother nature's fresh water sources. We can do similar with clouds and snow to control the climate.

Scientists don't seem to want to understand precipitation, at least publicly. The reason is that once we have a good cloud and snow model we can test inexpensive man-made methods to actively manage the climate to be whatever we want.

The end of the world is averted again by the use of technology. Next!

[MosesKnows]

Which should be self evident to anyone with just a conversant familiarity with the hard sciences:

The Al "the boob" Gore adherents have not yet learned that 2 does not equal 3; not even for large values of 2.

[edsheppa]

OK, then show me. Here's 45 years of historical data. C and T are observed (I've added a little random measurement error). Make a model that explains T. Then we'll see how well it does against the underlying system that produced the data over the next 50 or 100 years.

Year	C	T	Year	C	T	Year	C	T
1956	100.00	20.86	1971	115.98	22.25	1986	134.84	24.06
1957	100.75	20.85	1972	117.11	22.41	1987	135.91	24.31
1958	101.99	20.81	1973	118.45	22.66	1988	137.93	24.40
1959	103.09	21.19	1974	120.03	22.54	1989	138.83	24.76
1960	104.17	21.16	1975	121.05	22.87	1990	140.36	24.70
1961	104.52	21.05	1976	122.19	22.67	1991	141.57	25.06
1962	106.31	21.35	1977	123.58	23.11	1992	142.91	24.98
1963	107.04	21.46	1978	124.35	23.05	1993	144.24	25.12
1964	108.02	21.69	1979	125.09	23.32	1994	145.48	25.38
1965	109.64	21.66	1980	126.82	23.60	1995	147.49	25.29
1966	110.57	21.70	1981	128.73	23.53	1996	148.62	25.55
1967	111.61	21.90	1982	129.15	23.80	1997	150.86	25.79
1968	112.54	21.92	1983	131.07	23.85	1998	151.85	25.95
1969	113.70	22.08	1984	132.01	23.93	1999	153.78	26.22
1970	114.87	22.36	1985	133.36	23.92	2000	155.17	26.03