GFS 1 - 0 ECMWFAdded : Wednesday at 14:00 Eventually, the models all caught up with the GFS ► today with regards to forecasting snow across Southwestern and Southern areas from a frontal system which drifted across from the East :-
![]()
Eventually, even the ECMWF ► caught up reluctantly :-
![]()
But this raises a bit of a question with regards to the IFS Assimilation. You see, all the global models will have a way of assimilating the data, consider it their starting conditions.
NCEP, ECMWF, DWD etc all store their last run in their super computers, this is called the "First Guess". Then, as observations are sent in, these are then compared to the previous forecast and the model is then adjusted to give what's called a "Best Guess" model. The result of this is Global Assimilation and it's this which the physics are then run against to give us the numerical models and ensembles etc.
The IFS assimilation from the ECMWF ► has long been held as having the best starting conditions due to the fact that it takes far longer than most other global models and ingests a huge amount of data, but it got this one wrong.
When models disagree, then if the disagreement is in the near term, then it's usually the assimilation which is to blame, in the longer term it may well be the model physics and biases.
We did note that there was some issues with EFAS at ECMWF ► earlier this year with regards to forecast degradation due to missing observations, but we doubt this was the reason. But meteorologists who use NWP day in day out tend to have a "once bitten, twice shy" approach when models get scenarios like this wrong.
On Friday, the GFS ► brings in a front with snow across Western areas of the UK and Ireland :-
![]()
The ECMWF ► on the other hand doesn't :-
![]()
Will it redeem itself? Time will tell... At least you know what Fridays discussion will be about...
METEOROLOGIST : MARSH |
