Monday, March 25, 2013

Ill-Conceived Storm Warning Experiment Expands This Year

Dick McGowan
The National Weather Service, in part because of its deeply-flawed "Service Assessment" of the Joplin tornado, put an experiment in place to forecast the "impacts" of tornadoes in 2012. This was conducted at six of their offices in Kansas and Missouri.

The Wichita office issued two "catastrophic damage" warnings on April 14, 2012, and -- predictably -- they failed. Watch and listen to the "mass devastation" being forecast with this warning:

What happened? Nothing, absolutely nothing. In spite of a "tornado emergency" being issued for Conway Springs and "mass devastation" being forecast, Conway Springs was missed completely. However, if someone in Conway Springs had decided to flee the tornado based on the warning, they might have been killed by traveling east, south, or west as tornadoes moved around the city.

The NWS later issued a "tornado emergency" for east Wichita that also did not verify (the damage was light due to the tornado weakening).

It is often said that, in Washington, D.C., "nothing succeeds like failure." And, that is the case with the "impacts-based" warnings: They are spreading throughout the NWS Central Region this tornado season.
Map of NWS Central Region
There are multiple problems with this approach:
  • While we can forecast the general direction of movement of tornadic storms, the tiny movements (over say, a mile or two) cannot be forecast. Thus, Conway Springs gets missed.
  • Weather science has absolutely no ability to forecast the strength of a given tornado at a given location in time. Yes, we can say that "April 14, 2012, will be a day of big tornadoes somewhere in Kansas." We no ability to do that with a given storm on a mile-to-mile basis.
  • Even if we could do these things, it is the wrong approach. 

The basic tornado danger system -- warnings are broadcast, sirens sound and people take shelter -- stumbles over some frustrating psychology: Most people don't run for shelter at the first sign of danger but search, instead, for information to confirm it.

Experts say the Weather Service has often ignored behavioral research on delivering a "take cover" message that sinks in quickly.
So, trying to "scare" people into shelter will be counterproductive. Why? If you live in Conway Springs and you have been told "mass destruction" is imminent and nothing happens, how likely are you to take cover for an "ordinary" tornado warning?

What is the solution? It is to help people more readily internalize and confirm the warnings! 

I've prepared a paper on that subject that I will present at the American Meteorological Society's Conference on Storm Warnings and Communication in Nashville in June. 

In the meantime, take tornado warnings seriously this spring and seek shelter if one is issued for your location. But, take the "impacts" messaging with a huge grain of salt.

Addition: I heard from the NWS office in Omaha that they have opted out of this 'experiment." Good for them!

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.