Thoughts on climate modeling

This is cobbled together from a Buzz thread (yes, Google Buzz) about Global Warming. I think there are a lot of lay people who accept the science of Global Warming without understanding that most of the projected warming comes not from physics, but solely from computer models of the Earth's climate. Here's why these models (and thus Global Warming) are on very shaky ground.

Belief in Global Warming comes down to your faith in computer models.

The models are so important because CO2 itself will not warm the earth that much. Basic physics precits something like 0.5-1.5 degree Celsius over the next 100 years. That's the direct effect of the heat-trapping of CO2, and I think that's a reasonable forecast.

The much greater warming projected by groups like the IPCC comes from positive feedback effects shown in computer models. That is- that 1 degree warming will be amplified by other processes in the climate system, resulting in much larger overall warming- 3-4 degrees.

This is the dirty little secret of Global Warming - basic physics only gets you to 1 degree, you need the models (and their positive feedback) to get really bad warming.

What's wrong with models?

My chief objection to models, is that no one else uses them- and is taken seriously. In no other scientific domain has anyone demonstrated compelling statical modelling of complex systems capable of making confident predictions of future states. Not economics or biology. Even in Aeronautics where models are incredibly important, it's very, very difficult to make a system that properly models chaotic behavior like turbulent air flows. And Aeronautics has it easy- they can do trial-and-error and compare their models to real-life measurements.

The idea that a few dozen PhD groups have somehow created accurate computer models of the entire weather system, despite this system not being full understood, despite the system featuring chaotic, positive feedbacks, and being almost impossible to test (since there's only one, limited, and inaccurate set of data to ground truth it against) is absolutely incredible.

I given them credit for trying - and in the process learning more about the climate system and computer modeling. But the idea that they can have any confidence whatsoever about their system(s) making accurate predictions about the future is pure politics

Where's the proof?

After I posted the above, someone asked: "where's the proof that these models are wrong?"

And that's really the key question- how do you disprove a computer model?

Suppose some really smart researchers came out with a new "Economics" model and they claimed that unless we did X, GDP will shrink by YY% over the next 50 years.
How do you disprove their model?

They've fitted their model to some existing data sets- but there are thousands of free variables in their model they can tune. It more-or-less replicates the past 20 years of data, but again, they've tuned their model until it did just that- they wouldn't release a model that failed these sorts of tests.

Their model uses basic, and well-accepted micro & macro economics to model the economy, but it also makes "guesses" as to how various processes interact and how some particularly complex systems work.

Now as reasonably skeptical scientists, such a model is unconvincing. The lack of data to test makes it untestable. There are millions of possible models that will fit the limited data and be roughly consistent with known economics, but the vast, vast majority of them are wrong or useless.

This is in essence what climate researchers have done. They've dreamed up a computer model which is roughly compatible with known climate processes, makes guesses about other complex parts, and has been tuned to approximate the near-past.

It's almost impossible to "refute" such a model, but it's not a particularly compelling form of science, and it's certainly not worth sending trillions of dollars worrying about something this model tells you.

Are the models really that complex?

I make the claim that the Earth's climate is a complex process, which is difficult to model. Certainly there are Engineering fields where models are used with much success. Unfortunately, the answer is that the climate system simply is enormously complex. It involves interactions across an enormous space (the surface the globe) and the 3D volume of the atmosphere. The specific geography of each point makes a big difference, as do relatively small features, like the fact that the Mediterranean is open to the Atlantic (via a few mile gap), whereas the Isthmus of Panama completely separates the Atlantic and Pacific.

Further more, the system is more than just a "heat flow" equation. It has to model exchanges of gases (most notably evaporation, aka the water cycle) and the complex interaction of water vapor, dust, pressure and temperature (aka cloud formation). It also has to account for the various physical properties, such as the Sun/Moon gravity (tides), solar wind, and Coriolis effect.

And even the "heat flow" equations are difficult as they involve an entire geography of different surfaces with different specific heats, conductivities, and albedo. And we haven't even started talking about the effects of biomass on climate!

Many of these basic climate processes are still not well understood. Things like storm formation are still quite mysterious (you'll notice that tropical storm formation forecasts are very iffy even a few hours into the future). Basic processes like open-ocean cloud formation are not well understood and are typically modeled statistically (that is, the models use empirical statistical data to predict cloud formation rather than actually modelling a physical process). Other more complex phenomenon like El Nino and other large-scale ocean trends remain almost entirely mysterious. It's incredible that despite the enormous uncertainty around climate modeling, the modelers insist on the accuracy of their models!

Positive Feedback is a double-edged sword

The other major reason to be skeptical is that the models show the climate system to be dominated by positive feedback. This is extremely unusual for any long-term stable system. Positive feedback systems typically blow themselves up, which is why they are so rare. The difference between positive and negative feedback is the difference (literally) between 3 mile Island (which is still in operation producing nuclear power- at least the other core at the same site is) and Chernobyl. The Chernobyl's reactor featured positive feedback where increases in temperature* caused the reactor to produce more power, whereas 3MI (and all Western reactors) feature negative feedback. You'll note that 3MI got hot enough to completely melt the core and then simply shutdown. Chernobyl of course blew itself sky high.

The very thing that makes Global Warming so dangerous- positive feedback- also makes it very unlikely to be correct.

It's much more likely that when all physical process are accounted for, that the climate either has negligible feedback or is somewhat negative. There's a sort of "evolution" in complex systems - any system which sits at an equilibrium with "positive feedback" will with even a small push leave that equilibrium. This process will continue until the system finds an equilibrium from which it is hard to push it.

What about the Experts?

An overwhelming % of climate scientist believing in climate change, but I find that uncompelling:

1. here's a selection bias- anyone going into a field of science has to believe that their field is useful, otherwise why bother?
2. There's a cultural bias- questioning GW is social unacceptable (esp. among educated elites). This influences both the willingness of people to speak their minds and is internalized as a belief system
3. Recent history provides plenty of examples of well-known, well respected scientists and groups making outrageously incorrect predictions. Paul "Millions of Americans dying of famine in the 1980's" Ehrlich is probably the most famous- But such predictions were well respected and I've no doubt would have found widespread support (at the time), among the educated elite. Also don't forget the climate scientists in the 70's and early 80's who were certaitn that the Earth was cooling due to pollution. That includes prominent scientists like Carl Sagan.
4. Smart people believe all sorts of crazy non-sense, even when they shouldn't. Lots of smart people believed in Communism, and some still do! Plenty of very smart people believe in God and other weird stuff.

If something which is widely believed to be true, that's a good sign it probably is- but it's certainly not foolproof, even among a well-educated elite. In my opinion the clear facts over which I have knowledge (the unsuitability of the climate to accurate modeling), outweigh the fact that other smart people disagree. But I always was a narciscist.