Due to the definition of "No Regime" as "anomaly closer to climatology than any regime centroid", the ensemble mean forecasts drift to No Regime at long lead-times (without any post-processing).

Well, that’s the UK! We didn’t go fully metric: road signs are in miles, speed limits in mph.

This is standard in the UK

Yeah, tbh the main stumbling block today was my own stupidity, which is true with or without AI. I'm trying to learn to code like it's 2025 not 2019, as someone once told me, and this is working.

Fig from simonleewx.com/northern-ann...

If it beats 28 Nov 1968, yes

Yes! I’d love to do it if I had the time and resources

28 Nov 1968 (in ERA5 post-1958, which is considered reasonably reliable)

Around 30 km (the 10 hPa level)

Perhaps, but there’s no evidence of any changes to sudden stratospheric warming events yet, and models disagree on what might happen under future climate scenarios.

A few of the IFS’s 101 members have a zonal wind reversal too — but not the majority. 18Z GFS equivalent to a single ensemble member scenario charts.ecmwf.int/products/ext...