I have conducted research on tornadoes for fifteen years. As an economist, I do not study the storms themselves, but rather the impacts of tornadoes on society. I mainly search for patterns in casualties or damages, which can be done safely away from the tornadoes. I have done some field research, and never wish for the opportunity to do field work in my neighborhood.
Last week’s tornado in Troy occurred without warning, meaning that the National Weather Service (NWS) did not issue a warning until after touch down. Is it still possible today for tornadoes to occur without warning? Yes, and nationally about 30 percent of tornadoes are unwarned. NWS forecasters face significant challenges in issuing warnings, which we must recognize to help limit tornado impacts.
Tornadoes form out of thunderstorms. If we think of thunderstorms as parents of tornadoes, several types of parents exist. Parent types interact with the technology and tools available to NWS forecasters. Last week’s tornado was a spin up that occurs in the bottom portion of the parent thunderstorm, perhaps the lowest 10,000 feet; thunderstorm clouds can rise to 40,000 feet or higher. In classic tornado conditions, rotation occurs much higher up in the thunderstorm. Doppler weather radars cannot see the lowest part of thunderstorms because of the curvature of the Earth. The phrase “flying under the radar” refers to a plane flying low to avoid detection by radar. Spin up tornadoes fly below the radar too.
Many thunderstorms will look to forecasters just like last Thursday’s storm. Warning for every such thunderstorm would likely produce a dozen or more false alarms for every tornado. To appreciate why, summer thunderstorms occur almost daily in Alabama, while summer tornadoes are quite rare. Last week’s tornado was the first in Pike County not associated with a tropical system since 1950.
The NWS could prevent unwarned tornadoes by warning for every thunderstorm. Yet this would raise the false alarm ratio dramatically, and already three out of every four tornado warnings are false alarms. My research has documented the cost of false alarms. People and businesses respond to tornado warnings. Walmart, for instance, monitors severe weather and orders stores into lockdown during warnings. Tornado warnings disrupt our lives – shopping, dining, television, homework – quite extensively. My research with Somer Erickson found that over 200 million person hours were spent under warnings nationally each year. The value of time under warnings exceeded other types of tornado impacts, including the property damage or value of deaths or injuries.
Increasing false alarms would also prove deadly. We are all remember the story of the boy who cried wolf too often, and so too many false alarms will lead people to ignore warnings. Kevin Simmons and I found in a study of twenty years of tornadoes and warnings that a higher false alarm ratio increases the lethality of tornadoes. Furthermore, Kevin and I showed that aggressively warning to prevent unwarned tornadoes would result in more deaths through false alarms than saved through warnings.
Last week’s tornado was rated EF-1 on the Enhanced Fujita Scale, and almost all spin up tornadoes are weak (rated EF-0 or EF-1). Over 75% of all U.S. tornadoes are weak, and yet account for just 5% of fatalities, 9% of injuries, and 10% of property damage. Issuing lots of warnings for potential weak tornadoes may lead people to ignore a warning for a far more dangerous tornado. And the precautions we should take during thunderstorms offer pretty effective protection against unwarned weak tornadoes: stay inside and away from windows. Falling trees represent a danger in thunderstorms and weak tornadoes, so stay out of rooms with big trees outside.
It would be nice if the NWS could warn for all tornadoes and never issue a false alarm. But Mother Nature does not allow this. Ignoring tradeoffs leads us to make bad decisions, a familiar theme in this column. The existence of this tornado warning tradeoff is just one of the many ways economists contribute to discussions of severe weather.
Daniel Sutter is the Charles G. Koch Professor of Economics with the Manuel H. Johnson Center for Political Economy at Troy University and host of Econversations on TrojanVision. Respond to him at dsutter@troy.edu and like the Johnson Center on Facebook.