Predictability

I love models. Yes, metoo ... But I'm talking about algorithms here. Those who are a bit into mathematics know that with a little curve fitting you can go all the way. Lets aside Fourier. If you do not understand this, then I advise you to look at this Homer (Simpson) orbit example:

Maybe now you understand that a climate model can be made for any outcome. And then you also understand that it is wise to let the value of a model depend on any proof of the correctness of a model by testing it against measurements. The models of the club of Rome, for example, have not (yet) come true while they already beat the already known dates.

Is this bad? Not at all. We do not learn much without modeling and testing. I do have a problem with the way the press deals with modeling. They publish a scientific research as fact or almost fact at the moment that it offers sensational outcomes. Again something like that today!

'Coming four years will be unusually warm'

Fortunately, the text does add nuance "Drijfhout and his colleague concluded that the probability that the earth is warmer than normal until 2022 is according to them a consequence of a kind of break in global warming between 2000 and 2014. In that period the oceans have absorbed more heat, which is now coming back, but that expectation is dependent on other factors, for example volcanic eruptions and a changing amount of CO2 emissions can affect the global temperature. "

For the time being, meteorologists can not produce a reasonable weather forecast much further than a week ahead. Everything else is fun, interesting, valuable science, but nothing more. Whatever the press dakes from it.