The tornado warning system became one of the great scientific accomplishments of the 20th century.
But since then, tornado-warning lead time has regressed, even with improved technology, suggesting that there is a need for an independent review of the National Weather Service’s tornado warning procedures and training.
The pivot point came in 2011. Tornadoes, over and over, struck densely populated areas from the Plains to New England, killing 551 people and injuring thousands.
The season kicked into overdrive in April, when multiple rounds of severe weather swept through the South and Southeast during the middle of the month. Less than two weeks later, on April 27, the worst outbreak globally in nearly 40 years occurred.
On May 22, 2011, the Joplin, Mo., tornado became the deadliest of the tornado-warning era. As I document in my book, “When the Sirens Were Silent,” a combination of errors made by the Weather Service and local emergency management wrongly led Joplinites to believe that the tornado would pass north of the city. The tornado was rain-wrapped and invisible, so those in its path could not see the approaching storm for themselves. The people of Joplin were sitting ducks.
The failure of the warning system that day allowed the death toll to rise to 161, the worst single tornado death toll since 1947. Yet the Weather Service’s review of the event, or service assessment, overlooked the most critical issues that caused the unprecedented loss of life. It overly focused on overuse of sirens and not on:
- The excessively high rate of false tornado warnings issued for Jasper County, Mo., before the tragedy.
- The three times the Weather Service misstated the tornado’s direction of movement as northeast, which — if correct — would have meant the tornado would have missed Joplin.
- Three different tornado-warning polygons in effect for the Joplin metro area simultaneously, each providing at times different and/or conflicting information.
- The sounding of tornado sirens in areas not included in a tornado warning, while sirens were not sounded when the critical tornado warning was issued.
It is important to point out that the Weather Service’s assessment of the Joplin tornado did not include outside meteorologists who could independently develop findings and recommendations based on what was learned from that event. Four of the five main authors were Weather Service meteorologists; the lone author from outside the Weather Service was a social scientist from the University of Oklahoma.
Since 2011, the metrics intended to measure the “health” of the tornado warning system have deteriorated, an issue first covered by the Capital Weather Gang in 2017.
Unfortunately, the numbers have not improved since the 2017 story was written.
Probability of detection
2021 has had its own share of missed tornado warnings.
On March 13, a significant tornado southwest of Dodge City, Kan., went completely unwarned. When a high school student inquired about whether it was a missed event, the local Weather Service office replied with a cold “No,” but it later confirmed that the twister occurred and rated it an EF2.
Another tornado, this time an EF1, was missed by the Weather Service in Memphis on May 9 until well after touchdown. An obvious tornado signature was present on radar, including evidence of debris being lifted, prompting Twitter users to frantically tweet the local Weather Service office.
Days later, another tornado carved through metro New Orleans in the dead of night, the warning coming three minutes after the twister first touched down.
This spring, a television station in Oklahoma is running a promotional announcement that states, pertaining to tornado warnings, “We don’t wait for the National Weather Service.”
While stipulating that excellent individual tornado warnings continue to be issued, weather science has never been comfortable with discussing issues such as the deterioration in tornado-warning metrics. But it is unclear to me how we are going to rectify the situation unless these are discussed candidly.
What makes these declines especially puzzling is that the science and technology have continued to advance, including radar upgrades that make tornado detection easier, improved satellite imagery, field programs in tornado research and better computer-model simulations of thunderstorms.
Many people have theories, ranging from there being relatively more small, short-lived tornadoes — which are the most difficult to provide warnings for — compared with 2011, to insufficient training of meteorologists, to the retirement of experienced meteorologists. But counterarguments can be made for all of these explanations. For example, many of the young meteorologists entering the field are as bright and capable as ever.
I have reviewed several studies that have investigated trends in tornado-warning performance, including a 2018 analysis in the journal Weather and Forecasting, which suggested that forecasters may be missing warnings out of fear of false alarms. But I am unconvinced that they’ve fully explained and accounted for the lack of progress and the breadth of the problem.
What is needed to get the tornado warning program back to pre-Joplin levels? I think we need independent experts, not affiliated with the Weather Service, to evaluate this.
The service assessment process, as designed and implemented, remains an impediment to improving the quality of storm warnings. In fall 2012, after Hurricane Sandy, the Weather Service pledged that it was going to incorporate experts from outside the agency into the primary assessment teams. With small exceptions, it hasn’t happened.
The worst weather-related disaster since Joplin was California’s 2018 Camp Fire, which killed 85. While there were contributors from the outside, every member of that assessment team was an employee of the Weather Service’s Western Region. While I have no reason to believe the team operated in anything but good faith, the fact is that people within an organization may be reluctant to publicly point out deficiencies in that organizations’ performance and in the performance of people with whom they are friendly.
That is why I continue to believe the United States desperately needs a national disaster-review board modeled on the National Transportation Safety Board, which has been wildly successful at improving the safety of aviation and other modes of transportation. Such a review team would conduct independent investigations of weather science’s storm warnings and perform the day-to-day verifications of storm warnings and forecasts.
While warning accuracy is the key to reducing fatalities and mitigating damage, the success of the warnings can be judged only in a wider context, which an independent review team could scope. For example, the vast power failures that left more than 650,000 Alabamians in the dark before the April 27, 2011, tornado outbreak (due to storms earlier in the day) and unable to receive warnings, while mentioned, were not sufficiently analyzed in the service assessment of that event.
The weather community, including the Weather Service, is composed of extremely dedicated people. In the same way the NTSB guides the airlines and the Federal Aviation Administration to improve safety practices, a disaster review board would help guide the weather enterprise in pulling together its resources in a way that gets the tornado warning system back on track and helps us save even more lives.
Mike Smith is a retired fellow of the American Meteorological Society and certified consulting meteorologist. In 2019, Smith received the Special Award for Lifetime Achievement from the National Weather Association. He is the author of “Warnings: The True Story of How Science Tamed the Weather,” about the development of the United States’ storm warning system and its successes at saving lives.
Correction: The period described in the middle row in the left column in the tornado warning metric table originally was labeled 2011 to 2020. It should have been labeled 2012 to 2020 and has been updated.