Long Range Model Comparison (Photo Credit: Tropical Tidbits, S-J Lin, NOAA Geophysical Fluid Dynamics Laboratory)

 
As fall has begun and the leaves begin to change, so does the weather pattern on a large scale.  This is a very volatile time for weather models. This article will look at the upper air pattern the ECMWF, GFS, and FV3-GFS (updated version) all develop along with the similarities and differences.

It is known from verification scores that the ECMWF more often than not scores higher than the GFS (a big part of this resides in the superior data assimilation, but this is a topic for another article). This was made clear as the ECMWF predicted the track of hurricane Florence days in advance while the GFS stuck with the idea that Florence would progress farther north and stall along the coast. However, the FV3-GFS had Florence near the southern North Carolina coast well before the GFS caved to this narrative.  Is this sign of a significant improvement? 

Let’s compare the 500 mb height anomaly maps for North America via the 18Z GFS, FV3-GFS, and the 12Z ECMWF all valid the morning of Saturday October 6th. These model runs at this point initiated early to midday September 26th. Just glancing over the three maps below, there is one clear outlier within the three models, the GFS. Right away the overall pattern from the FV3-GFS and the ECMWF is rather similar, ridge over portions of the central US, western Canada, and into Alaska. A trough is shown in portions of southern Canada stretching into the northern US, as well as a possible tropical system to the southwest of Baja California. The locations of the ridge in the west are different, as well as the depth of the trough with the FV3-GFS showing a deeper more amplified solution. These differences are in relation to the pattern in the Pacific stretching into coastal portions of the West coast. Both models want to advertise an upper level low pressure system (possibly more than one of the FV3) somewhere in the CA/NV region. 

The GFS on the other hand shows a completely different pattern with a ridge in the east, and troughiness over the Northern Rockies stretching into Canada, along with a different Pacific pattern. ​

In retrospect, this is clearly a very complex pattern developing and these are only one run of each model (10 days in the future mind you) so although there are similarities in the two models, all three of these solutions are likely wrong on certain levels.

As winter approaches, it certainly will be interesting to see how the FV3-GFS handles the state of the atmosphere along with the tracks of these winter storms compared to the ECMWF. Something to keep an eye on and possibly become excited for depending on the verification of this model.

Be sure to stay tuned to GWCC for the latest on weather, climate, and winter weather here!

©2018 Meteorologist Joe DeLizio