What to make of the climate models?

The Great Global Warming Swindle is suspicious of the climate models. Merchant sums up their argument in slide 17 then gives his response

Climate models are useless

• Proposition:
• Climate models require many uncertain parameters, that can be "tweaked"
• Implication
• Climate modellers choose parameters that given them the results they want
• This is wrong because:
• Modellers impose various limits on what parameters are valid

Here Merchant seeks to justify the models. Fair enough. But all the program's criticisms of models are valid.

I think it is interesting that there is no adequate validation of the models. They are tweaked to make them fit the data. I don't object to this. But I don't see the tweaking as proof, I see subsequent data fitting the models (without further tweaking) as the beginnings of proof.

Slide 18

Model parameters

• Parameters in models often based on fundamental physics, chemistry etc
• Other parameters are tightly constrained by observations of the real system
• Parameters that are poorly constrained may be check via their effects

Slide 19

Valid models

• Sets of (uncertain) parameters can interact to give unreasonable models
• These are eliminated by requiring models used for prediction to be able to simulate
• A sensible stable climate like the present (constant forcing, realistic variability)
• The changes in response to forcing during the 20th C (within reasonable errors)

This is a major problem with this sort of science. Science is either observational or historical. This is a fundamental idea of the philosophy of science. Unfortunately it is poorly known, though relatively easy to understand. The 2 may be mixed, that is have features of both. The basic concept is:

• Performing repeatable experiments is observational science
• Reconstructing previous events is historical science

So one is limited in being able to prove historical science because of the nature of it being non-repeatable. This is not a failing, rather an intrinsic limitation of it. Modelling historical climate (even as recently as the near present) falls mostly under historical science. That means that in order for the models to be accepted they must be validated. One can validate by working with some of the data and checking it with the rest of the data. But if all the data is used, or the models are adjusted with subsequent data then the theory remains unvalidated. Observational data is validated but differently; and it is gives more definitive results as anyone can potentially repeat the experiment over.

Note Merchant does not mention water vapour in the variables of the models. This may be an oversight in the presentation but water is very important—the most important.

He says that modellers try to have realistic assumptions. Yes, but they are assumptions. And if you believe in global warming and your models predict this you are likely to think your assumptions are good.

He dismisses radiation by saying this can be accounted by a few assumptions. But that is the debate is it not? I could say the same about CO₂ . In fact CO₂ should be easier as one just allows for a slow increase, whereas radiation is less predictable.

How about minimising assumptions, stating them clearly, then validating your model.

He showed earlier models on slides 8 thru 10. These had actual temperature and predicted temperature versus time for 1850 to 2000. The first was natural climate forcing: solar and volcanic; the second was anthropogenic forcing : green house gas and sulphate aerosol; the third was both natural and anthropogenic factors.

These models are just too accurate (for something as difficult as climate). Clearly the parameters have been tweaked to get this. This is okay, but it is not proof. It is a model, follow the model for some years and see if it continues to be true. While you are waiting for this, you have not proven anything, nor can you expect people to change behaviour based on your model. (Never mind that the idea that one can be so certain of the distinction of natural CO₂ versus man-made CO₂ is just too absurd).

And the models have not predicted accurately. There is now a lot of talk about Arctic ice-melting. Could this be because the temperatures have not been as high as predicted for the last few years so they lessen the discussion on the temperature and start discussing some effect they claim is related to it?