The quality of tornado warnings has improved dramatically since the early days of radar, a veteran weather official says, but success rates have leveled off in recent years.
It will take a series of steps – some relatively simple and others expensive and years from reality – to see a significant improvement in accurately detecting tornadoes and providing additional lead time for people in the path of tornadoes, said Don Burgess, research scientist at the Cooperative Institute for Mesoscale Meteorological Studies and adjunct professor of meteorology, both at the University of Oklahoma.
“We find ourselves on somewhat of a plateau” in what he called “skill scores,” Burgess said at the national storm chasers’ convention in Denver. “We have improved over where we used to be. ... Where do we go from here?”
Fearful of inciting public panic, the national weather bureau forbade the issue of tornado warnings until the 1950s. Even when the warnings began being issued, Burgess said, they “were not very skillful.”
As recently as the 1970s, tornadoes touched down only 20 percent of the time warnings were issued. Of the tornadoes that touched down, warnings were issued on less than 10 percent of them.
“We missed a lot of tornadoes and we issued a lot of poor warnings,” Burgess said while later discussing the data in more detail.
The arrival of Doppler radar technology gave researchers and meteorologists more tools to understand how the storms worked. During a project to evaluate the effectiveness of Doppler radar, lead times for tornadoes jumped several minutes and the false alarm rate improved by more than 20 percent.
That led to the deployment of Doppler radar nationwide, Burgess said, though that wasn’t completed until the late 1990s. Detection rates, lead times and false-alarm rates all improved as the radars came into use.
“But we need to do better overall,” he said. “So how do we improve?”
One way, he said, is to determine and duplicate best practices. Some weather service bureaus are consistently better at detecting threats and issuing warnings, he said, so the National Oceanic and Atmospheric Administration’s warning decision training branch is working to collect how they do what they do and making that information available throughout the branch network.
Another step is improving reports from storm spotters – whether they are from storm chasers, the public or the network of strategically positioned spotters.
“I know we get many good reports ... but we still deal with a significant noise level” – meaning reports that are poor or wildly contradictive to other reports from the same area, Burgess said.
“We have spotters and we have radar and we need to work together,” he said.
New smartphone apps are being rolled out that allow residents to report weather events such as hail or snow where they are, giving meteorologists “ground truth” for storms indicated on radar – and apps are being developed to send NOAA warnings to smartphones.
Researchers hope to develop and deploy smaller, lesser cost radars on towers that can provide data on what’s happening in the “boundary layer” – that space between the cloud base and the ground, Burgess said.
Vortex2, a two-year field study of tornadoes in the nation’s midsection, revealed that factors near the surface of the Earth play a much greater role in the development of tornadoes than previously realized, he said, and a network of radars scattered throughout Tornado Alley would help give meteorologists a clearer picture of the evolving conditions during potentially severe weather.
Attempts to have the radars placed on cellphone towers have met with resistance so far, he said, so researchers are looking for other ways to set up the radars.
“The private sector has to get involved,” Burgess said.
If a network of these boundary layer radars is to be established, “it’s going to have to be with a lot of private sector funding,” he said.
The next quantum leap in radar technology – phased array radar – has already been identified, Burgess said. It will allow meteorologists to scan the atmosphere much more rapidly than currently – and to aim multiple radar beams at the same storm during severe weather.
“It’s another world” in radar capabilities, he said. “Phased array will be a godsend to the warning community.”
But much work needs to be done in developing the delivery platforms for the new radars and building computers that can rapidly process the high volume of data being provided by the radars, he said. As a result, he doesn’t anticipate the phased arrays to arrive for more than a decade.
In the meantime, upgrades to the dual polarization radars now being deployed nationwide should serve as an effective interim measure.
By the year 2020, weather researchers hope to be able to accurately predict tornado formation 80 percent of the time and have a lead time of 45 minutes, Burgess said. The false alarm rates will always be higher than people will like, he said, because forecasters wisely error on the side of caution.
“We’re doing good things,” Burgess said. “We’ve made warnings a lot better.
“It’s going to take another long period of technology improvements and smart people doing smart things” to take radars and tornado warnings to the next level, he said.