P18 would surpass any severe spell since the early 90s IMO.
As in any winter model watching season, it’s all about sifting the genuine signals from the noise. You’ll always get a smattering of Stella runs in any ENS set, especially now there are 30 GFS perturbations and the ECM available. The trick is to identify if any of the more ‘favourable’ solutions (and remember, the Operational is always preferred to any of it’s ensemble siblings in isolation) are a genuine sign of a particular synoptic forecast or just ripples on the lake. And to assist that process you need to look at the background signals that may be influencing those solutions - MJO, AAM, u-wind trends, strat temps, etc. This is where you need to make judgements and assign a degree of confidence on whether the extended range offered by any particular model seems plausible or not whether it makes sense given the above - and other - known parameters. The computers of course take these and many other parameters into account to churn out their solutions but it’s rare to achieve above 85% accuracy at 5 days and much less at 10 days so anyone saying it will definitely be this or that in two weeks is at best guessing.
The best even a hugely experienced forecaster can hope to achieve is an ‘idea’ of the set up in two weeks and that’s where means can be useful - to spot rough trends. That’s why I council against taking individual runs in isolation. It’s nice art but very unlikely to be a solution.
Now I’ve forgotten where this post was heading 😂☺️
Originally Posted by: nsrobins