A satellite image of Hurricane Harvey before landfall in Texas in August. (NOAA/NASA)

This was one of the worst hurricane seasons on record. We’ve seen six major hurricanes, Category 3 or stronger. Three of those made landfall with disastrous consequences. Hurricanes produced more than $300 billion in damages this year and left few places in the Caribbean, Gulf and Southeast United States untouched.

At one point in early September, the National Hurricane Center was managing simultaneous hurricane watches and warnings for three storms — Katia, Irma and Jose.

Despite the extreme season, the National Hurricane Center — which is a branch of the National Weather Service — turned out one of its best performances on record. When so much was on the line, its predictions for hurricane tracks, pinpointing where these monster storms were headed, were as accurate as they have ever been.

“The hurricane season still has a few weeks to go,” said Dennis Feltgen, the spokesman for the National Hurricane Center, which means all of this talk of how they performed is still preliminary. “That said, the data indicate the NHC Atlantic track forecasts in 2017 set record low errors at all time frames.”

None of their forecast seasons, which go back to 1970, were better than this year.


A forecast’s skill is measured in two important ways — did it get the track right, and did it get the intensity right. You can calculate these things for a particular storm, or, like in this case, for all of the storms combined in an entire season. The main goal of a forecast is to get it right, obviously, but what we really want to see is a forecast that’s better than a climatological coin toss. In other words, it needs to be better than saying “what would an average storm like this do?”

Beyond that, forecasters strive to reduce error as much as possible. In 2017, the National Hurricane Center had its lowest track error.

Among its best-forecast storms was Hurricane Irma, which existed in some form or another from Aug. 30 to Sept. 12, which is a significant amount of time for a storm to be on the Hurricane Center’s radar. The NHC issued 47 one-day forecasts, 39 three-day forecasts and 31 five-day forecasts. Impressively, NHC forecasters beat their own average track error by about 30 percent at every lead time on Hurricane Irma.


Hurricane Irma forecasts were particularly good. Each colored line represents a different track forecast made by the National Hurricane Center. The thick black line (which is covered by the colored lines most of the time) is the actual observed track that Irma took. The inset chart shows the track forecast verification; NHC’s forecasts for Irma are shown by the red curve (“OFCL”), their average track error over the past five seasons is the dotted black curve, and forecasts made just using climatology and persistence are indicated by the gray curve (“OCD5″). (NOAA/Brian McNoldy)

The intensity forecasts were — and always are — harder. The strongest hurricanes often become that way because they go through rapid intensification, which is exactly what it sounds like, but is very difficult to predict. Conditions have to be just right, and even then some storms intensify rapidly and some don’t. It’s still somewhat of an unknown in hurricane forecasts.

This season had several examples where a storm strengthened dramatically in less than a day. One of those storms was Hurricane Harvey, which — although it seemed to come out of nowhere — the National Hurricane Center predicted with accuracy.

After a hurricane-hunter mission showed Harvey was strengthening, the NHC issued a special update. Their strength forecast ended up being within 6 mph of what Harvey ended up being at landfall.

Still, other storms dragged this season’s intensity forecasts down.

“[Hurricane] Lee is a good example of unexpected genesis and then subsequent rapid intensification,” Feltgen said. “Roughly 40 24-hour rapid intensification events occurred during the 2017 season, and NHC correctly identified about six of them. This problem is, unfortunately, only slowly improving.”

The final verification statistics will be released by NHC early next year. Small changes are possible in that final report, but in all likelihood, the National Hurricane Center will be able to hang its hat on the 2017 season.