During the 1990s, East Coast forecasters had great success in predicting two huge snowstorms several days in advance: the "Storm of the Century" in March 1993 and the "Blizzard of January 1996." Much of that success resulted directly from continued improvements in weather computer models.

But last year on this date, those same models burned forecasters as an unexpected six to twelve inches of snow fell across much of the East Coast. These computer models had predicted that the storm would move harmlessly out to sea. The models did eventually catch on to the actual track of the storm, but only a few short hours before the white stuff began. So many folks from North Carolina to New Jersey awoke to a heavy snow that was largely unforecasted.

The lesson: these weather computer models were, still are, and will continue to be, imperfect. Weather forecasting is a blend of human experience, knowledge, and observation, all of those crucial to interpreting computer predictions - a combination consistently better than just computer forecasts alone.