COULD HURRICANE IAN'S FORECAST TRACK HAVE BEEN MORE ACCURATE?
Quick Answer: Yes
🌀 The GFS (American) model really botched the forecast track for Hurricane #Ian. The long range forecast by the GFS influenced the NHC & other meteorologists into considering the possibility for a landfall farther north. The ECMWF (European) & other global models, however, performed really well.
🌀The GFS had an average seven day forecast error of 248 miles (based on 12z runs). The ECMWF had a remarkedly better average error of only 89 miles. You can see in the graphic (issued seven days before landfall) that the GFS incorrectly had the storm going to the Panhandle. The ECMWF had the storm making landfall near Punta Gorda.
WHY WERE THE MODELS DIFFERENT?:
🌀 The four main synoptic factors controlling Ian's track were: a surface front over Florida, long wave trough over the NE US, two large high pressures (one over the Atlantic & one over the western US).
🌀How the models interpreted these large scale weather features determined the steering currents & ultimate track.
🌀 The ECMWF is computationally more powerful than the GFS. The raw math & ingestion of more high latitude satellite data also played a large role in Ian's track.
SO SHOULD WE IGNORE OTHER MODELS AND JUST GO WITH THE ECMWF?
🌀 No. The GFS was the most accurate model in 2021 after significant model improvements were implemented. The GFS had also handled storms earlier in 2022 better than the ECMWF.
🌀 The key is for meteorologists to continue to learn model strengths, weaknesses, & biases... much as I'm doing in this post.
🌀 In the days preceding the storm you may have noticed that I mentioned the possibility for a landfall slightly farther south than the NHC. I also made a post about how the models were suggesting a landfall farther south, more towards Sarasota than the western Big Bend. This was my intuition in going more with the European, German, Korean, & UK models over the lone American model. The intuition was in knowing that the ECMWF generally handles high latitude and large scale synoptic systems better than the GFS. Though, it's not kosher to issue forecasts far outside of the NHC official forecast. Too many cooks in the kitchen can lead to public confusion & from the official track.
COULD A CAYO COSTA LANDFALL HAVE BEEN CONFIDENTLY PREDICTED 4-7 DAYS IN ADVANCE?
Quick Answer: No
🌀The Florida Panhandle & western Big Bend forecast tracks were largely a result of the GFS & HWRF (GFS derived). Without forecast influence from the GFS, this area of the state never wouldve been in the direct crosshairs.
🌀 As to where on the western Florida peninsula (between Cedar Key and Fort Myers) the hurricane would make landfall is unfortunately beyond forecast skill more than 3 days out. The ECMWF actually confused matters as it started shifting toward a St. Petersburg landfall only a few days in advance of the storm.
NOTES:
🌀 The NHC has a cone of uncertainty for a good reason. Everyone in the cone should be on high alert. Fort Myers was within the cone with every update. But, the level of preparation & evacuation does demand a best estimate of landfall within the cone.
🌀 The brilliant scientists at the NHC & many meteorologists don't simply rip and read a forecast ... like the app on your phone. We look at the conditions around the storm, nowcasting, previous experience, & model biases/trends.
🌀 Forecasts have come so far! Florida had over seven days notice that a hurricane was coming for the state. Most forecasts were issued when Ian was a messy blob of rain off the coast of Venezuela.
🌀 Final Note: prayers continue for everyone in SW FL impacted by record surge... & through central FL where damaging winds & flooding wreaked havoc. Tampa Bay dodged a catastrophic bullet. A slight wobble farther north & many areas from downtown Tampa to St Pete would hardly be recognizable.
Comments