The Canadian Press: Climate-change researchers should use more data to avoid errors: study
...Newman's team used two models and expected some level of variation in the results. But they did not expect contradictory data.
"We basically got opposite answers when we should have gotten the same answer," Newman said.
"What we've shown is if you use more than one model you can get entirely different results, so (based on studies that used only one model) maybe we have no view at all of what the impacts are going to be."
Posts
3 comments:
"Those models, while generally consistent at predicting climate, "
Kool-Aid, Kool-Aid, tastes great :)
"What we need is a whole array of models that all make different assumptions and then we look for conclusions that are reasonably robust."
They don't seem to understand that models must be verified first. Once verified, the one or two most correct models may be able to be used to provide somewhat robust conclusions. No prediction is reasonable until the models have been proven accurate.
John, how do you "prove" a climate model is accurate? The only obvious route toward that is to compare model hindcasts and past climate. The problem is that even if a model does well at that, there is no certainty that it will be good at predicting future climate because the climate forcings are changing much more rapidly and to a much greater extent than we have seen in the historical record. Also a model that doesn't do as a good a job with historical records might be great at future climates. Until the future happens, we can't "validate" these models in the way that I think you are envisaging....
Post a Comment