For more than a decade, researchers at the National Oceanic and Atmospheric Administration have been working on ways to improve forecasts of two things boaters care deeply about: hurricanes and rip currents. Now, technology is finally catching up with these researchers’ visions—and making it possible to give boaters information so detailed, it represents a big leap forward.
“The technology is challenging, and anytime you’re promoting something new in the government, it’s challenging,” says Joe Cione of NOAA’s Atlantic Oceanographic and Meteorological Laboratory Hurricane Research Division. “But I feel very good about this.”
Cione is the principal investigator for NOAA’s work using drones to study hurricanes. Since the mid-2000s, he has been trying to figure out how to collect more data from deeper inside storms, including places manned aircraft can’t go. His original efforts were with a drone called an Aerosonde that weighed 25 or 30 pounds. His team, which included U.S. Navy and NASA personnel, would try to get themselves to wherever a hurricane was forecasted to make landfall. They’d launch the Aerosonde off a car and then wait for it to return with data. In seven or eight tries, they got it to work twice, with 2005’s Hurricane Ophelia and 2007’s Hurricane Noel.
“Ophelia was the first time a drone had ever been in a tropical
cyclone,” Cione says, lamenting that while it worked, it took 17 hours for the drone’s flight, with maybe two and a half hours inside the storm. “That’s not a very good return on your investment.”
However, the Navy folks had an idea. While NOAA used P-3 Orion aircraft to fly manned missions into hurricanes, the Navy used P-3s to detect submarines. The Navy would launch a probe out of the plane and let it drop into the ocean, where it would collect information. Why not use the same technique to launch a drone into the air?
In 2009, Cione’s team tested that idea with a Coyote drone on a clear day. The Coyote was much lighter than the Aerosonde—only 13 pounds. They managed to shoot it out of the plane and get it to fly, but still, the technology wasn’t quite right.
“Then, in 2012, a transformative event happened,” Cione says. “Hurricane Sandy hit in New York City. After that, Congress said we needed to understand hurricanes a little more. There was a big bill they passed through, and a very small part of it was for hurricane research. We got funding.”
With that funding, drone technology could improve faster. The first time
Cione’s team sent an air-deployed Coyote drone into a storm was 2014’s Hurricane Edouard. “It’s not easy to launch a drone from another plane,” he says. “It’s an engineering nightmare. It’s a 13-pound aircraft surviving 100-mph winds while getting spit out of a 200-knot aircraft. But we did it. We had proof of concept.”
Then, in 2018, Cione decided NOAA needed to build its own drones. He wanted drones that could travel farther and stay inside storms longer. So, he put out a call for small businesses to solve that engineering problem. The result was akin to the race the world is watching unfold with Covid-19 vaccines. Multiple companies are trying different approaches. If one works best, that’s great. If three of them work, then NOAA will have three new types of drones in its toolbox.
In January, the first of those new drones—the Altius—got a test in clear skies over Maryland. “It met almost every hurdle,” Cione says. “It was like getting a B-plus on an exam. Still very good, but we want to get to 100 percent before we go into prime time in a hurricane.”
Within about a year, NOAA should know whether the plan has fully succeeded—and the results should be substantially improved over current hurricane measurement methods.
On manned aircraft flying into storms, pilots are limited to flying at about 10,000 feet; they drop atmosphere-measuring “straws” into a few parts of a storm, and the straws collect data as they fall into the ocean. By comparison, the Altius drone should be able to fly around inside the storm as if it’s a doughnut, collecting data the entire time.
“It’s a snapshot versus a movie. The drone gives us a movie,” Cione says, adding that the detailed information will help with everything from precise evacuation orders to marina weather predictions. “Is your boat safe where it is now, or not? We should get better at answering that question. We should be able to say that even if a storm is 100 miles away, it might cause you problems. We can hopefully give boaters more confidence in these predictions going forward.”
Meanwhile, closer to land, NOAA has also taken a leap in the technology used to predict rip currents. NOAA Senior Scientist Greg Dusek started working a decade ago on how to better understand when and where rip currents would form. He had two basic problems: It’s really hard to get data out of a surf zone, where waves are breaking and sand is moving around, and even if he could get that data, there were no high-resolution, near-shore computer models available to process it.
So, he went lo-fi and sought the help of the Kill Devil Hills Lifesaving Station in North Carolina. The lifeguards in that location had been collecting records for many years about people they rescued from rip currents.
“That provided us baseline data for when we see rip currents that are hazardous to people,” Dusek says. “It was very lucky.”
His team at NOAA combined its ocean data with the lifeguards’ observations, and in 2013 created the initial model for better predicting rip currents.
“It looked like it worked really well in North Carolina, but we didn’t have a way to run it operationally,” he says. “We could look back in time, but we didn’t have any way to do an actual forecast. The computer models weren’t available yet, but the National Weather Service was working on the Nearshore Wave Prediction System. It’s a combination of computer-wave and water-level models at very high resolution near the shore all around the coastal U.S.”
That prediction system exists—and it provides information about things such as wave height, tide and surge for every kilometer along U.S. beaches, looking forward for six days. That was the data Dusek’s rip-current model needed.
“We ended up marrying the two things together and doing validation in Miami and San Diego, and working with other lifeguard agencies to collect similar kinds of information to assess the model’s performance,” he says. “This approach works pretty well every place in the U.S.”
In February, NOAA made the model operational. Today, the agency can predict the likelihood of hazardous rip currents on a scale of 1 percent to 100 percent—much as a weather app predicts the likelihood of rain for every hour of the day.
Boaters will start to see the rip-current information filtering into existing weather apps and forecasting programs, Dusek says, as NOAA feeds it out for local weather reports. And going forward, the team hopes to integrate data from webcams at places like waterfront hotels and surf shacks. “We can apply artificial intelligence to that video to try and predict rip currents,” he says. “That can potentially help improve our model even further over the next few years.”
