Posted at 1:53 PM on November 9, 2007
by Craig Edwards
Edwards, I just finished plowing two inches of partly cloudy! An associate in the snow plowing business couldn’t wait to chastise my lousy forecast. That was five years ago. This morning there was a half inch accumulation of light snow between Cambridge and Duluth that startled motorists and befuddled meteorologists.
NOAA uses an array of metrics to determine the value added by the meteorological staff over the computer models. For decades individual forecasters systematically made improvements on the forecast numerical guidance, i.e. temperatures and precipitation probabilities. In the 1980s, while serving as operations manager at the Indianapolis office, I wrote a paper on how good the computer models were becoming. It was increasingly difficult to make significant correct deviations to the computer guidance.
This morning, when flurries swept through the Twin Cities, the computer models tossed out precipitation probabilities for measurable moisture of less than ten percent. Forecasters knowing that snow displayed on radar was reaching the ground adjusted that guidance up to twenty percent.
The key issue here is measurable precipitation. With light snow, holding little moisture, you can have twenty minutes of flurries and no measurable precipitation. So, in effect, when the International Airport reported a trace of moisture this morning, the computer’s forecast of less than 10 percent was more accurate, on paper, than the 20 percent by the local forecaster.
Flurries, by definition imply little or no accumulation. My forecasts have improved somewhat. One day last winter my friend reported back that he just salted and sanded a half inch of flurries.