If going purely by statistics, and you have one model that has a skill score around 90%, and five models with a skill score around 80% that are all clustered in agreement, the odds that the outlier model, despite typically being more accurate, is actually correct, is only 1%. It is a 99% probability that the cluster of 80% accurate models are right in that scenario. Now obviously there is a lot more nuance and complexity to this situation, but I think that oversimplified example should illustrate the point of how compelling model consensus in a given scenario should be, even if the models that agree are statistically less accurate on average.

Merry Christmas!

Nonevent. You should stop wasting your time here and enjoy it with your family. I’ll page you when we have another weather event showing up on anything but the Euro.

Meh.

The Euro was the first one to notice it at all... If anyone here remembers.