Are the global warming models "robust"?

Question

A "skeptic" posted this. "Then equations are hacked off so something can actually be computed in a reasonable time. The reasonable assumptions used to whack out the physics, mostly depend on the whim (sorry, educated guess) of the researcher."

This is somewhat true, though the process is far more rigorous, with assumptions tested, etc. And EVERY major climate model, not matter how the researcher chose to simplify parameters, shows that now, greenhouse gases are the most important factor driving climate.

In statistical terms the "signal" from greenhouse gases is very strong.

Models that give the same basic results even when assumptions are modified are called "robust", meaning strong. Are greenhouse gas models "robust"?

By the way, this was an unusually knowledgeable post by a skeptic, even if a bit strident. The skepticism was reasonable, and my response is intended to be respectful.

Answer 1

Correct as far as it goes--but you should add that such models almost always  rely on special types of mathematics such as fractal geometry--which are designed to handle such problems without sacrificing accuracy.  And, in fact, have to--computers are digital devices, while equations, unmodified, are analog statements.

A veryssimple example: you can write the quantity one third as "1/3" and it is absolutely precise.  But--although a computer will accept that as input, when it comes time to crunch numbers, ithe computer must express the number as a digital number--0.3333333---on to infinity.  Only it cant just list more and more "3s"--at some point the rest has to be "hacked off."

The same is true o fvirtually all the complex equations used in any kind of computer modeling.  But the mathematics isn't arbitrary--as I pointed out, very sophisticated systems of math have been developed to do this--and are used for avariety of applications. They areextremely robust.

Answer 2

Digital modeling is relatively new.  However, humans have been using analog modeling since the dawn of time; the sundial, Stonehenge, mechanical clocks, the abacus, mechanical calculators and the Antikythera mechanism are examples.  

Early computer models of the climate were able to show the direction of the change and some hints at the magnitude, and supported earlier theoretical predictions.  Later models confirmed the trend and narrowed the range of uncertainty.  Surely models are imperfect and are constantly improving, but this is not the point.  

As different investigators approach the problem with different computational methods, the different models all show the same trends and general outcomes.  As greenhouse gasses increase, average temperature increases.  Warming will be greater at the poles and at night.  As atmospheric moisture increases due to the ability of warmer air to hold more moisture, precipitation events will become more intense.  These effects are occurring now exactly as the models predicted.

Digital modeling is not some sort of alchemy.  Theoretical science is based on mathematics.  Mathematics is internally consistent and irrefutable.  A theoretician uses mathematics to prove that his theory is internally consistent.  The theory may have no practical application at this point, but it is a valid framework for further study and application.  Newton remains a giant because the calculus he invented has endless applications.  

Empirical scientists design experiments to test the theory.  After the theory and the mathematics have been proven to be correct by direct measurement through experimentation, the same mathematics from the theory is used to design the model.  

Models are used to design nuclear weapons, chart spaceflight, model astrophysical phenomena like star and galaxy lifecycles, model particle interactions in nuclear physics, model biochemical reactions for the design and efficacy of drugs, model population growth, model economic systems, model thermodynamic systems for combustion and engine design, model material behavior for structural design, and more I haven’t thought of.  

The point being that the models aren’t “tweaked to get the answer you want”.  They are used every day around the world by scientists and engineers to make things that work.  

Computer modeling has opened a new chapter in science.  Empirical scientists were once limited by that which they could construct and observe.  Now we can take models, which are proven to be internally consistent because they arrive at the same results obtained from direct measurement, and drive them beyond the directly obtainable.  We can drive them into the future, into the past, speed them up, slow them down, enter parameters that would be unrealistic or uneconomic or impossible for a researcher with practical constraints.  Sometimes the models produce nonsense.  Sometimes they produce results that the researchers are unable to explain, and thus open a new field of inquiry.  It is a new frontier.

It amuses me that people use weather models as proof they are no good.  Ã¢Â€ÂœWe can’t predict the weather next week, who is to say what will happen in 100 years?”  The point is we can predict the weather next week with some accuracy.  We can predict the weather tomorrow within a degree or two of temperature and an hour or two of when the precipitation will arrive.  That is an incredible achievement.

Answer 3

The computer models are pure BS. No disrespect meant, but the old farmers almanac is more accurate than these computer models.

Answer 4

I would say definately not for a number of reasons.  

Once a theory is developed, it is just a theory untill it is tested against real world data and can predict things before they happen, rather than just show an approximate fit to historical data.  

Modeling of climate is not something that we have the science for.  The studies into how CO2 affects the opacity of the atmosphere have huge error bands.  Estimating what the effect of increasing CO2 will be is highly speculative.  Trying to model feedback effects is also highly speculative.  Why should we be remotely surprised when the verticle temperature gradients (as investigated by Douglas, Singer Christie et al) don't agree with warming from a greenhouse effect and are instead more consistant with warming from the ground up (like increased incident radiation).  

The other thing is what else can be modelled with finite element modeling?  We can't model constant temperature, incompressible aerodynamic flow over a car or a building with much accuracy.  There are models available, but they're not about to replace wind tunnel testing.  Modelling the climate systems for the entire planet would seem to be something that we don't have the technology to do even if we did have the physics (and apparently we don't).

Answer 5

Sure.  Climate models have been able to accurately hindcast the warming of the 20th century.

http://en.wikipedia.org/wiki/Image:Clima...

And the various different climate models have similar projections of future warming, depending on various inputs like their future assumed CO2 emissions changes.

http://globalwarmingart.com/wiki/Image:G...

The important thing is that they all agree that global warming will not only continue, but accelerate in the future.

Answer 6

Are you asking a question?

Since water vapor is the most important greenhouse gas, then yes, I think everyone, even skeptics agree that greenhouse gases are the most important factor in the average global air temperature.  but the earth's rotation is also an extremely important driver of climate.

The issue with models, is that they don't model water vapor and clouds.  Water vapor tending to raise temperature, and clouds reflecting sunlight and lowering temperature.

Since water vapor is the most important greenhouse gas, and it is not modeled in the models, there is little reason to believe the models predict anything.

If you were modeling a different problem by computer, and could adjust the model using fudge factors, parameters, plugged numbers, etc. to make it fit, everyone would roll their eyes and politely say ... great model, and walk away.

There was a legal case on TV where bite marks convicted a man of murder.  To show the bite marks matched, the "expert" kept changing the position of the cast of the teeth, moving them, rolling them, until yes indeedy they eventually matched up with the holes in the corpse.  But did the teeth as is match the bite as is?  No.

Answer 7

If computer models are accurate at hindcasting, then they should be able to forecast equally as well.

That being the case Bob, can you tell us if this summer is going to be warmer or cooler compared to last year?

Can you show us how you came to this conclusion?

Why do I think you just can't do this yet?