My understanding is that although there is no inherent bias or AI architecture in NWP algorithms, the resolution with height varies model to model (the EC having more layers than the other two top models). Therefore changes filtering down from above may not be modelled very well with respect surface output, and as said parameters are subject to almost hourly adjustment (primarily satellite data), the volatility is adversely affected.
For the resolution reasons I tend to take more note of the extended EC ensemble when we’re interested in 15 plus days in a SSW (ongoing with another warming) event.
This output is predicting a shift of the mean trough east in the next ten days and heights building to our N and NE and the reason why I’m still optimistic of a colder end of January than the GEFS for example is currently projecting.
Originally Posted by: nsrobins