Hurricane Irma has regained Category 3 intensity, and additional strengthening is likely in the coming days as it tracks west across the Atlantic. Environmental conditions could even support Irma becoming a rare Category 5 hurricane at some point, the first since 2016’s Matthew. Within the next five days, the Leeward Islands, Puerto Rico, the Dominican Republic, Haiti and the Bahamas are most definitely at risk and should prepare for hurricane conditions (at least a close encounter if not a direct landfall).

Beyond then, forecast confidence drops dramatically. Some models curve Irma back out to sea before reaching the continental United States, but a significant percentage of models have Irma striking the U.S. East Coast as early as Saturday or Sunday. Therefore, anyone with interests from Florida to New England should monitor forecasts closely.

The center of Irma is still about 900 miles east of the Leeward Islands, 2,150 miles east-southeast of Miami and 2,150 miles southeast of Wilmington, N.C. The National Hurricane Center is predicting Irma to pass over or near the Leeward Islands on Wednesday, then to be near the eastern Bahamas by Friday. Beyond that, the spread in model tracks grows, but it usually does at such long lead times. Hurricane watches may be issued later Sunday for the Leeward Islands.

[img]https://img.washingtonpost.com/wp-apps/imrs.php?src=https://img.washingtonpost.com/blogs/capital-weather-gang/files/2017/09/145723_5day_cone_no_line_and_wind.png&w=480[/img]

In the longer range, beyond five days, we turn again to the model ensembles. Rather than relying on a single model or a single model run, we look at ensembles which generate dozens of model runs, each using slightly different initial conditions as input. The more similar the model tracks are to one another, the higher the forecast confidence. The larger the spread among the model tracks, the lower the forecast confidence. Watching the trends in these ensembles also helps wash out any erratic run-to-run changes.

Both the European and U.S. ensemble models below show similar forecasts with relatively high-probability tracks through about 168 hours (Saturday). Both then show a wide spread of track scenarios beyond that time, with landfall from Florida to New England possible, or staying out to sea. Note the European (ECMWF) ensembles show a little more potential for curving back out to sea than the U.S. (GEFS) ensembles, although even the European still shows a significant risk of U.S. landfall. On average, the European model is more skillful than the U.S. GEFS model for tropical cyclone tracks, and so far in Irma’s brief history, it does indeed have lower track errors.

https://img.washingtonpost.com/blogs/capital-weather-gang/files/2017/09/ensembles-693x1024.png

The trend over the past couple of days has been to keep Irma further south for longer, bringing it closer to the United States, but still turning toward the north near the coast. With this in mind, locations from South Florida up to New England are most definitely still in play and need to be paying close attention. With the shape of the coastline, you can easily see above how a slight difference in timing of that northward turn makes all the difference.

As far as impacts and timing, we can offer some hypothetical scenarios — that is, *IF* the hurricane were headed for Location A, it would make landfall on Date A. These are not official forecasts, only estimates based on current long-range model guidance. Dangerous tropical storm-force winds would typically arrive about a day earlier.

South Florida: Saturday/Sunday (Sept. 9/10)

South Carolina, North Carolina: Sunday/Monday (Sept. 10/11)

Delmarva, Long Island, Cape Cod: Monday/Tuesday (Sept. 11/12)

Looking through the records going back to 1851, 11 previous tropical cyclones passed within 200 miles of Irma’s current position as well as its three-day and five-day forecast positions (based on Sunday’s 5 a.m. Eastern Daylight Time advisory). Some infamous ones in that short list are the Great Miami Hurricane of 1926, Hurricane Gloria in 1985 and Hurricane Floyd in 1999. Note that six of the 11 recurved before reaching the U.S. East Coast.