There’s no question we’ve been desensitized to tornado warnings (see this blog post by James Spann). There are far too many warnings that turn out to be false alarms. Warnings can cover hundreds of square miles and even a large tornado would do damage over a miniscule section of the warned area. In fact at least 80 percent of warnings don’t even wind up verifying with any touchdown at all!
Part of the problem is that our technology is only so good. Extra doppler radars would cut down on the number of false alarms but there’s little chance we’ll double or triple the existing network. It seems to me, and this isn’t a criticism, that the National Weather Service tends to overwarn to cover their butts. I can’t blame them.
Given the public complacency the NWS has been issuing “tornado emergencies” within tornado warnings in rare cases where a large or violent tornado has been confirmed on the ground. The first one was issued in the May 3, 1999 Oklahoma City tornado and 82 have been issued since.
Here’s some interesting research from Patrick Marsh at OU. He found that the false alarm ratio is surprisingly high for something that is supposed to be issued only in the most extreme cases. In fact using significant tornadoes (EF2 or greater) the false alarm ratio is 46%!
So what’s the solution if tornado emergencies aren’t? How do we differentiate a run of the mill tornado warning for a storm that will be lucky if it can produce a 60 mph wind gust and a monster supercell dropping an EF3 or greater tornado in downtown Springfield? Should they all be covered by the same kind of warning?