There has been a lot of discussion in the Norman Meteorological Community in the past few weeks over the number of deaths in the devastating tornado season we have had so far this year. With over 550 lives lost and scores more injured it is sometimes hard to understand how these many lives are lost and people affected when we have greatly increased the capability to warn people in advance of these tragic tornadic events.
If you examine the three biggest outbreaks of the year (Alabama, Joplin, and Oklahoma) you quickly see that the National Weather Service did an amazing job of forecasting, and providing watches and warnings for these events. Days in advance of these events, the Storm Prediction Center had strongly worded convective outlooks, that in all three cases, provided excellent heads-up to the public. On the day of these events, strongly worded tornado watches were issued hours in advance, and tornado warnings had average lead times greater than 20 minutes for most of the tornadoes during these three events.
Today, there are many more ways that people can receive weather information then say even a decade ago with the ubiquity of Smartphones and the quality of graphics presentations by television meteorologists. So why did so many people die?
I believe that there are many reasons for the large number of deaths this year, the first is that there were a number of large violent tornadoes that happened to hit densely populated areas. These EF-5 and EF-4 tornadoes killed some people that did all the right things. Some killed had received the warning, gone to a safe place in their home and yet the tornado was so strong that it did not matter; they would have needed to be underground or in a safe room to have survived. There are others that received the warning and chose not to take any action until it was too late. And of course there were others that believed that “I have never been hit by a tornado before when there was a tornado warning, so why would it happen to me now?” The cried wolf syndrome is a big factor.
The reality is that the science of meteorology and the tools at the disposal of meteorologists (especially Doppler weather radars) have greatly increased the capability to warn the public. In fact, the average lead-time of tornado warnings used to be zero minutes before the NEXRAD program was implemented in the early 1990s, placing 144 Doppler radars around the United States. Today the average lead-time to tornadoes is 14 minutes, a great leap forward in 20 years.
But the percentage of tornado warnings that are false alarms has stubbornly stayed about 75%, meaning that 3 out of 4 tornado warnings do not have a tornado reported. The NWS continues to push the science envelope (in a positive manner) to gain more lead times for tornado warnings. Often tornado warnings are for a period of 45 minutes and cover a fairly large area. One thing that exacerbates the perceived false alarms is that many media outlets, smartphone apps, tornado warning sirens and other communications to the public warn for whole counties rather than the portion of the county that is actually warned for.
In my opinion, a big push needs to be made to only provide warnings for the area that is actually warned for by the NWS meteorologist, which is described by a polygon. And I also believe that we need to somehow provide more information on the timing of the event at each location. If what the public perceives when they receive a tornado warning, is that a tornado will occur only a small percentage of time somewhere in my county and sometime in the next 45 minutes, it does not provide the type of information they need to narrow the warning to determine if it is an actual risk to them and when. Thus, the natural tendency for humans is to not react at all, or to look at the skies watching to see if the warning is a real, but not reacting or moving to safety until actual observation of the tornado has occurred.
In my next blog, I will discuss how the smartphone revolution along with LBS can make a big difference in communicating warnings to the public.