Your blog post of June 13 “Overcoming the tornado false alarm problem” was of interest to me as I have spent most of my career studying the science of warning of tornadoes and other extreme weather and also communication methodologies for the same. In my book, Warnings: The True Story of How Science Tamed the Weather, I make the case that meteorology has made tremendous progress in warning of all types of storms the last ten years. But, there is still work to be done.
I agree with the opinions expressed in the article that it is essential for meteorology to find ways to decrease the number of false alarms of tornadoes. It is important to understand there are two types of false tornado warnings (meaning warnings issued when no tornado occurs):
* Unavoidable false alarms occur due to inadequate scientific knowledge, adequate measurements, or both
*Unnecessary false alarms are a result of sub-optimal warning communications or techniques
Several of the tornado warnings around Washington, D.C. on June were unavoidable false alarms. The thunderstorms had “hook” echoes, had rotation revealed in the Doppler wind data, and were moving to the right of the other storms in the area. Those are strong indicators the thunderstorms would cause tornadoes and I strongly concur with the NWS’s decision to issue warnings for them.
The sheltering of people in the basement of a government building during the June 1 tornado threat was an unnecessary false alarm because the District’s downtown was never in a tornado warning. This is presumably a communications issue. The National Weather Service’s more precise latitude and longitude-based “polygon” warnings are still not used by many institutions and by almost no individuals. I expect that to change rapidly as GPS-enabled smart phones can display these newer types of warnings.
The blog post mentioned the experiment the NWS is running on the citizens of Missouri, eastern Kansas, and southwest Illinois where they attempt to predict the impacts of severe thunderstorms and tornadoes as part of the warning messages.
I agree with the severe thunderstorm impact forecasts because weather science has the tools and knowledge to make accurate estimates in most cases.
Tornadoes are another matter.
There is no evidence we have the scientific skill to forecast or warn of tornado intensity. For example, the evening of April 14, the citizens of Conway Springs, Kansas, were told it was a “tornado emergency,” their town would receive catastrophic damage, and that their residents would be “killed” if they were not “underground.” What happened? Nothing. The tornado completely missed the town. How will the citizens react next time if a mere “ordinary” tornado warning is issued?
The converse occurred on May 19. The Wichita NWS told the citizens of Harper and Kingman counties that the tornadoes would be brief and weak. Turns out the tornadoes were up to F-3 (strong) intensity. Among other damage, one destroyed the home of the mother of one of my AccuWeather colleagues.
The day will hopefully come when we have the scientific skill to forecast tornado intensity in a consistently accurate manner. That day has not yet arrived and I believe this well-intended experiment is doing more harm than good.
One other aspect of attempt to forecast the intensity of a tornado is that we never know what is in the path. For example, on June 17, 1978, a weak tornado moved across Pomona Lake in Kansas and either struck or passed near the Whippoorwill dinner theatre showboat. It capsized, killing 16 and injuring three.
After years of doing tornado warnings very well, the National Weather Service seems to have lost its institutional confidence in the wake of 550 people being killed last year, a modern record. Part of the issue is the flawed “service assessments” done in the wake of the April 27 tornado outbreak in the South (330 dead) and after the Joplin* tornado (161 dead). The combination of omissions and unexamined major issues in those reports has confused what had been a straightforward warning mission. For years, I have recommended that NWS service assessments be conducted by an unbiased third party. I urge the new leadership of the NWS to adopt that policy for future service reviews.
Science moves toward when issues are openly and candidly discussed and evaluated. While this essay may seem negative, I wish to reiterate that tornado warnings on the whole have greatly improved, are usually quite accurate, and should be taken seriously.
* I cover these issues in my book, When the Sirens Were Silent.
Mike Smith is a board certified consulting meteorologist and Fellow of the American Meteorological Society. He is the Sr. Vice President and Chief Innovation Executive of AccuWeather Enterprise Solutions in Wichita.