– The 2017 hurricane season will be remembered for its brutality and endurance, but it also goes down in history as having the most accurate storm track forecasts since modern mapping of tropical cyclones began nearly 50 years ago.

While official verification of forecast accuracy won’t be released until early next year by the National Hurricane Center, University of Miami storm researcher Brian McNoldy tallied preliminary results to find the smallest track error rates on record.

For every measurement, which includes days 1 through 5 before estimated landfall, the 2017 forecasts topped the hurricane center’s running five-year average by as much as 52 miles at the 120-hour mark.

Storm experts said the improvements likely saved lives.

“It’s certainly noteworthy that the track forecasts beat the previous record at every lead time,” McNoldy said. “It wasn’t just that the longer-range forecasts were good, they were all good.”

One of the most precise forecasts in terms of path was for the aberrant Hurricane Irma, which raged as a 185 mph Goliath for a record-shattering 37 hours.

The hurricane center’s average forecast track error five days out is a 225-mile spread. This season, the average during the same time period was 173.6 miles.

But for Hurricane Irma, the five-day track error was just 155 miles.

The 48-hour error rate as Irma closed in on Cudjoe Key was 58 miles, about 23 miles less than the overall average for that time period.

“Irma’s track forecast was very good. It was just a very well-behaved storm,” McNoldy said.

An estimated 6.5 million Floridians evacuated for Hurricane Irma, a storm that tormented the Caribbean for a breathless 11 days before the Sept. 10 landfall in the Keys as a dangerous Category 4 cyclone. For those who felt chased — first to the west coast, then back to the east coast or out of the state entirely — the track forecast might have seemed flawed.

But James Franklin, the National Hurricane Center’s former chief of the hurricane specialist unit, said the direction of Irma’s approach and Florida’s narrowness meant even the most minor track errors exacerbated evacuation confusion.

“Maybe expectations are a little unrealistic to think we can distinguish two days in advance between Palm Beach and Tampa,” Franklin said. “They’re not that far apart.”

At its widest, Florida’s peninsula is about 160 miles coast to coast.

After Irma’s second Florida landfall on Marco Island at 3:35 p.m. Sept. 10, hurricane-force winds extended 80 miles from its center, with tropical-storm force winds extending 415 miles.

“People can argue about uncertainty in the track and whether one model did better than another, but it misses the point that Irma was just such a physically large storm,” said Chris Davis, an associate director at the National Center for Atmospheric Research.

Davis reiterated Franklin’s concerns about expectations for pinpoint forecast accuracy being too high, but also that people still focus too much on the center of the forecast cone to make decisions.

“The center of Irma went over Naples; Jacksonville flooded,” Davis said. “The cone covered a huge part of the state, and there were hurricane-force gusts on each coast.”

The reasons for the increased accuracy in forecasting the tracks of the year’s 17 named storms are multiple.

Pieces of code that go into computer models representing processes in the atmosphere, such as radiation and clouds, have gotten better. Computers are running at higher resolutions. More sophisticated satellites circle the Earth beaming down images so clear that forecasters see towering cloud tops and fields of Saharan dust like never before.

And, simply put, 2017’s storm paths were easier to read.

“Deep, tropical long-track storms tend to be a little easier to forecast track for because they move, for the most part, west-northwest at a fairly constant speed,” McNoldy said. “The trick then is when and how quickly they will recurve.”