My modelling backing began in complex dynamics (e.g. turbulent flows) but most of my experience is in financial modelling. And I can say with a high degree of confidence that anyone in the financial world who actually bet money based on this modelling approach (employed in the recent Nature article on UK flooding) can be described with one word: bankrupt. No one in their right mind would have any confidence in this approach. No one would ever trust a model that has been hand-tuned to match retrospective data to be accurate going forward, unless that model had been observed to have a high degree of accuracy when actually run forward for a while (a test every climate model so far fails). And certainly no one would trust a model based on pure modelling without even reference to historical data.
Te entire emerging industry of pundits willing to ascribe individual outlier weather events to manmade CO2 simply drive me crazy. Forget the uncertainties with catastrophic anthropogenic global warming theory. Consider the following:
- I can think of no extreme weather event over the last 10 years that has been attributed to manmade CO2 (Katrina, recent flooding, snowstroms, etc) for which there are not numerous analogs in pre-anthropogenic years. The logic that some event is unprecedented and therefore must be manmade is particularly absurd when the events in question are not unprecedented. In some sense, the purveyors of these opinions are relying on really short memories or poor Google skills in their audiences.
- Imagine weather simplified to 200 balls in a bingo hopper. 195 are green and 5 are red. At any one point in time, the chance is 2.5% that a red ball (an extreme event) is pulled. Now add one more ball. The chances of an extreme even is now 20% higher. At some point a red ball is pulled. Can you blame the manual addition of a red ball for that extreme event? How? A red ball was going to get pulled anyway, at some point, so we don’t know if this was one of the originals or the new one. In fact, there is only a one in six chance this extreme event is from our manual intervention. So even if there is absolute proof the probability of extreme events has gone up, it is still impossible to ascribe any particular one to that increased probability.
- How many samples would one have to take to convince yourself, with a high probability, the distribution has gone up? The answer is … a lot more than just having pulled one red ball, which is basically what has happened with reporting on extreme events. In fact, the number is really, really high because in the real climate we don’t even know the starting distribution with any certainty, and at any point in time other natural effects are adding and subtracting green and red balls (not to mention a nearly infinite number of other colors).