Saturday, August 27, 2011

That's more like it?

Many years ago, I first heard a story, although I no longer recall how I heard it.  The story concerns Allen D. (Al) Pearson, who at the time was the Director of the National Severe Storms Forecast Center in Kansas City, MO (which eventually became the Storm Prediction Center and moved to Norman, OK).  The story goes that some old woman was known to call the NSSFC and complain about every tornado warning for her location that turned out to be a false alarm.  After a tornado outbreak in that area, Al got a call from the woman, whose home (the story goes) was utterly destroyed by a tornado.  Her words to Al?  "Now that's more like it!"  This story may be apocryphal.  I don't know for certain, but it illustrates something that has always left me puzzled about how "the public" responds to weather warnings.

This has been a year characterized by some of the worst tornado events in our nation's history.  After several decades of relatively low tornado fatality counts, the nation has surpassed 500 fatalities with several months yet to go in the year.  We've had a single tornado in Joplin, Missouri, kill more than 150 people for the first time since 1947 (the tornado that hit Woodward, Oklahoma).   It's evident that tornadoes remain a deadly threat, which is what many of us have been saying for decades.   There's already considerable whining and complaining that Hurricane Irene hasn't lived up to the "hype" generated by the warnings.

It seems that there are many folks who, like the old woman in my story involving Allen Pearson, seemingly would prefer to become storm casualties and have their lives and property devastated!  I completely fail to understand the "logic" of this attitude.  Of course, no one wants to alarm people needlessly, but the fact is that our ability to predict storms falls well short of perfection.  False alarms are an inevitable consequence of meteorological uncertainty, as are occasions where a warning fails to go out and yet the event happens.  The only way to never miss an event is to predict the event everywhere, all the time.  The only way to never issue a false alarm is to never issue any warnings at all.  Reality is such that the science leaves us somewhere between these two unacceptable extremes.  An objective evaluation of the tornado warnings show that a considerable majority of them turn out to be false alarms.   By the way, for most of its history, the USA's public weather forecasters never issued tornado warnings because they were forbidden to do so!  It was feared by many otherwise intelligent folks that such a warning would cause a panic!

Some people have complained that the National Weather Service issues too many needless tornado warnings.  There may be some validity to such complaints, especially when considering individual NWS offices, whose office policies regarding warnings can vary from one office to another.  The basic idea behind this concern is the so-called "cry wolf" problem, where the working hypothesis is that "the public" is desensitized by too many warnings.  Various proposals by various people have been made to remedy the problem, mostly impractical or unjustifiable from the point of view of the science of storms.

I've argued that the overwarning problem (if it is indeed a problem) is the direct consequence of one inescapable fact:  no one is ever killed (at the time) by a false alarm!  Forecasters are much more likely to be condemned for not issuing a warning, thereby missing a storm event that winds up killing someone, than by issuing a warning that turns out to be a false alarm.  This is beyond any doubt the primary cause for the overwarning bias.  I've argued that one way to reduce or eliminate this bias is to convert to probabilistic warnings.  I hear constantly about how and why that won't work, but that's drifting off-point in this blog.

Consider the "desensitization argument":  One comment that I hear quoted in the media from time to time is "Oh, we hear tornado warnings all the time, and nothing ever happens!"  All the time?  Really?  I don't think the objective evidence comes anywhere even close to that clearly hyperbolic declaration.  In any given location, even in "Tornado Alley", tornado warnings are relatively infrequent.  Are people seriously expecting to hit by a tornado every time a tornado warning is issued?  Why would people think that?  What shortfall is responsible for such a patently absurd expectation?  Surely all the weather broadcasters who are complaining about NWS warnings could have done a better job of explaining the reality of the warning challenge to their audiences!  Nor am I attempting to make excuses for the NWS - in fact, I believe the accuracy of their warnings certainly could use some improvement.  But someone needs to understand just why "the public" has this attitude and start a public information campaign to eliminate it!

This has been a terrible year for weather tragedies:  not just tornadoes,  but also high winds, floods, hurricanes, etc.  Surely people should be thankful when a warning for potentially dangerous weather turns out to be a false alarm!  There are hundreds of dead people from storm hazards so far this year, and likely thousands of their friends and families who, if given the option, would prefer that those events had not happened at all.  That they had simply been another false alarm.  Any of them happily would change places with some whiner grousing about false alarms.

If you choose to take the attitude that so many warnings are false alarms that you can ignore them indefinitely, there's a chance you may be wrong at some point in the future - dead wrong!  Not a high probability, though.  So you're most likely going to be able to live a long and productive life, complaining about false alarms to the end of your days.  Shouldn't you be grateful?

1 comment:

Pete Browning said...

Right on the mark. Of course, after this year I'm feeling pressure to improve the FAR. Thanks.