Select Page

Enlarge 12z European operational model landfall location for Hurricane Irma.

View more stories

If you closely follow hurricane forecasting, you know that in recent years, the European Center for Medium-Range Weather Forecasts has the best forecast model in the world based upon skill scores. Often, this intergovernmental organization of 34 nations produces the best forecasts for hurricane tracks. In some cases, during Hurricane Harvey, it even exceeded the skill of human forecasters at the National Hurricane Center.

For this reason, the European model now has an outsized influence on the forecasts for hurricanes around the world, including those in the Atlantic, and in particular Hurricane Irma, which presently threatens the Caribbean Islands as well as the southeastern United States. How much influence? Take a look at this plot of a bunch of different models from Wednesday morning. Note the dark blue line on the left-hand side of the forecast tracks—that’s the official track forecast from the National Hurricane Center that was issued at 5am ET. 

Enlarge 12z forecast model spaghetti plot for Hurricane Irma, with an official forecast track in dark blue.

Now, you may be wondering, “why is the official forecast so far to the left, when all of the other models had moved east?” The answer is the European model. This forecast system has superior hardware to run its calculations. But more importantly it has a method by which it better assimilates real-world data—observations from weather networks around the world, atmospheric soundings, reconnaissance aircraft, and much more—into its calculations.

The US analog to the European model is the Global Forecast System. It has a lower resolution, and it typically doesn’t perform quite as well (proposed NOAA budget cuts will make matters worse, too). However, this GFS model has some benefits: it runs four times a day, and NOAA freely makes the data available to anyone who wants it. This is why GFS data appears on plots like the one above, but the proprietary European model does not.

The European model runs every 12 hours, so when the forecast plot above was made, National Hurricane Center forecasters were working off of the 00z run, which typically comes out at about 2:30am ET. The ensemble forecasts (about 50 runs on a lower resolution model with slightly different initial conditions) typically follow about an hour later. Here’s what that ensemble data looked like for forecasters on Wednesday morning.

Enlarge 00z ensemble forecast from European model. Darker blue areas show most likely location of Irma’s center.

Critically, while the other forecast models this morning were suggesting Irma now appeared more likely to remain off the east coast of Florida, the European model was still holding to a “most likely” landfall position along the southern Florida coast.

After reviewing additional model data between 5am ET and 11am ET (but nothing new from the European model, since it hadn’t been run again), the National Hurricane Center adjusted its landfall track slightly to the east. “The NHC forecast has been shifted eastward to be in better agreement with the latest model guidance; however, it should be noted that there are numerous GEFS and ECMWF ensemble members that take Irma over and/or west of Florida,” senior forecaster Daniel Brown wrote in his discussion. “The updated NHC track is in best agreement with the latest ECMWF ensemble mean.” Translation: We’re not betting against the European model.

11am ET Wednesday official track for Hurricane Irma.

11am ET Wednesday official track for Hurricane Irma.

National Hurricane Center

The 12z Euro

Therefore, the big question facing forecasters heading into the afternoon hours was this: what would the European model do with the 12z run? Would it follow the other models offshore? The answer is “no.” The operational version (the highest resolution, best guess, single run) of the European model shows a landfall almost directly over Miami.

Enlarge 12z European operational model landfall location for Hurricane Irma.

This means that we should not expect big changes from the next official forecast from the National Hurricane Center, due out at 5pm ET. Despite a large portion of the model guidance now bringing Irma to the east of Florida, I anticipate forecasters will continue to defer to the European model.

Admittedly, this has been a rather technical discussion of forecast models. If you live in southern Florida, the key message is much simpler: a potentially catastrophic hurricane will approach your location later on Saturday night or Sunday. The time for making final preparations is now. If an evacuation is called for your area, go.

What we can confidently say from the modeling data is that there is a reasonable chance—perhaps 50 percent—that Irma will turn north toward the Bahamas and move far enough east of the Miami area to avoid dangerous winds and storm surge. Effects would be bad, but not devastating for southern Florida. But there also is a pretty good chance that catastrophic winds are coming to one of the wealthiest, most well-developed coastlines of the country, stretching from Miami to West Palm Beach.

For Floridians, it is absolutely a cliche to say, “Hope for the best, prepare for the worst.” In this case, it is also absolutely true.

More from the 2017 hurricane season

Source

https://arstechnica.com/science/2017/09/heres-what-the-worlds-most-accurate-weather-model-predicts-for-irma/