About a month before Hurricane Harvey slammed Texas with an amount of rain so immense forecasters said it could not happen more than once in a thousand years, a University of Miami scientist developing a new weather tool knew what might be in store for the Gulf coast.
“I can’t claim ‘problem solved’ or anything like that,” said Ben Kirtman, an atmospheric scientist at the Rosenstiel School of Marine and Atmospheric Science. But his experimental model could “preemptively improve your chances of not having a catastrophe.”
The 2017 Atlantic season will be remembered among the 10 worst on record, blamed for killing hundreds, costing billions and producing hurricanes with unprecedented fury. Harvey, the first major hurricane to hit the U.S. mainland in a dozen years, set a new U.S. record for rainfall. Irma followed, hammering Florida and Puerto Rico with fierce winds that made it the strongest hurricane ever recorded outside the Caribbean or Gulf of Mexico. Then Maria pounded Puerto Rico, further crippling it.
But for hurricane researchers, the season that ends Nov. 30 will also be remembered as a grueling run for prediction models and emerging forecast tools like Kirtman’s that up to now had not seen such monster storms.
So how did they perform? The good news is the chief model produced by the National Oceanic and Atmospheric Administration and used by the National Hurricane Center helped produce the best track forecasts since the center began issuing tracks. Earlier warnings came for storms nearing land, and new maps provided the arrival times for damaging winds.
“It’s more than just making a better mousetrap,” said Frank Marks, director of NOAA’s Hurricane Research Division at the Atlantic Oceanographic and Meteorological Lab in Virginia Key. “It’s actually pulling all the pieces together and making them work together like an orchestra.”
The bad news is forecasting intensity remains a problem. The hurricane center correctly predicted the rapid intensification for six out of seven storms, a forecast they have been reluctant to give in the past because they lacked confidence in the models, Marks said. But another 19 storms forecasters expected to strengthen also met the threshold of increasing wind speeds by 35 mph in 24 hours.
“There’s progress, but we have work to do,” he said.
Among the achievements worth noting:
▪ Hurricane center forecasters using NOAA’s ‘workhouse’ model made the most accurate track predictions yet. During Hurricane Irma, they beat their average by 30 percent with every forecast, according to UM hurricane researcher Brian McNoldy. An experimental version tested throughout the season improved forecasts another 10 percent. The model was created in 2007 to improve track and intensity predictions by 50 percent in 10 years.
▪ A UM ensemble model, used to create seasonal forecasts, accurately called for an increase in the number of major storms and a spike in storm energy for the season, which eventually produced 10 hurricanes in a row for the first time in more than a century.
▪ A Princeton-based NOAA model and the Climate Prediction Center correctly produced Harvey’s extreme rain days before the hurricane struck. Another experimental model that issues hourly forecasts for hazardous weather predicted the spot where Irma would make landfall 28 hours in advance.
All these improvements add up to better forecasts, which for emergency managers and the public can mean better preparation. In flood-prone cities like New Orleans and Houston, or states like Florida that are wrapped by a coast vulnerable to storm surge, a month’s warning about potential hazards could mean saved lives.
The advances also go beyond hurricane research. Kirtman’s new monthly model, dubbed SubX, extends forecasts out to 34 days. Coupled with the ensemble model that predicts seasonal weather, it could allow meteorologists to make forecasts further into the future.
“Suppose you’re planning a trip to Disney,” he said. “I can’t tell you definitively, but I can tell you there’s a 20 percent chance the whole five days it’s going to rain, or there’s a 70 percent chance.”
The models could also fill another critical gap: forecasting more imminent threats from climate change.
A warming planet that traps more moisture is expected to produce more intense hurricanes, but so far impacts like hurricanes and sea rise are generally forecast on global scales over years. Earlier this month, the U.S. government’s National Climate Assessment warned more intense hurricanes with heavier rain will likely increase in the coming decades. Oceans, which have risen globally by seven to eight inches since 1900 and continue to rise faster than at any time in the last 2,800 years of records, could also trigger more extreme flooding, the report said.
But that’s not the kind of information that could help ongoing work, like the $16 billion Everglades restoration project or a new voter-approved $400 million bond to address sea rise in the city of Miami. Impacts are not expected to rise along a straight line but bobble up and down between wetter and drier and colder and warmer years, like a roller coaster that steadily rises. Predicting the bobbles will be critical.
“It’s these shorter term fluctuations on top of the climate change signal where the real vulnerability kicks in,” said Kirtman, director of the Cooperative Institute for Marine and Atmospheric Studies, a research partnership between 10 universities and NOAA.
“You have this chronic problem, but then all of a sudden you have an acute problem on top of that chronic problem and that makes it catastrophic. It’s these risk multipliers and hazard multipliers and the interaction of timescales and that’s where we’re really trying to help.”
It also dovetails with the kinds of advances NOAA is trying to make in hurricane forecasting, Marks said.
The public, and emergency managers, now expect more precise predictions on when and how much force a storm will deliver, whether it’s heavy winds or storm surge. A good example is a map it began issuing that shows when and where hurricane-force winds will arrive. For each map, a model runs thousands of scenarios.
“It’s really where I think this whole process will go,” Marks said. “Having been here in Miami in Irma with two sisters in Naples and a cousin in Clearwater, those types of things are really critical.
“We’ve been flying into storms for decades, but not a lot of that data would get into the model and make an impact,” he said.
Researchers are also looking to explore areas with data gaps. At UM, a giant wave tank is looking at what happens when the ocean and air meet, a critical point of hurricane intensification. NOAA has also begun flying drones through hurricanes to get a better glimpse of conditions that storm sensors dropped from planes record at only a single point, using the drones on three successful tries during Maria, Marks said.
Underwater gliders also record ocean information to detect how hurricanes change the sea as they pass over. That information could help models detect changes during a busy season. Four deployed north and south of Puerto Rico collected data as Irma and Maria passed over, he said.
The next big hurdle will be finding a better way to convey that information, he said.
“We’re in an age of what we call the weather-ready nation. And we’re the tip of the spear,” Marks said, referring to the national appetite for weather information. “The nation can be ready, but not if they don’t understand what we’re saying.”
This story was updated to reflect the National Hurricane Center forecasts for 19 storms that met the threshold for rapid intensification.
Follow Jenny Staletovich on Twitter @jenstaletovich