Make sure forecasts don’t imply certainty

The US National Weather Service’s revamped flood forecasts are proving to be a key difference between 1997 and 2009 — and the change is for the better. The challenge is to remind the public of the uncertainties involved.

In ‘97, the service’s forecasts at times contained a single number that the public had a tendency to fixate on. In the case of the flood prediction for Grand Forks, that 1997 number, infamously, was a Red River crest of 49 feet. Forecasters always cautioned that the prediction could change as conditions changed.

But human nature being what it is, many people missed or dismissed those warnings, and took comfort in the 49-foot prediction — a comfort that proved hugely misplaced when, seemingly late in the game, the forecast was changed and the river surged to 54 feet. To its great credit, the service learned from its experience and changed its public forecasts significantly.

Today, the forecasts almost always are offered as ranges, not single numbers. So in the middle of last week, the outlook for Fargo called for a crest of 39 feet to 41 feet. Now, here’s the thing. By the end of last week, forecasters ratcheted up the projection and predicted a 41-foot to 43-foot crest in Fargo. Obviously, the first prediction was off; but that’s not the problem.

The problem is that like the single number of 1997, the “range” predictions of today still have an air of certainty about them, one that lay people (and, possibly, some city and state officials) often find misleading. In other words, when the forecasters said 39 feet to 41 feet, it was easy for Fargo residents and officials to fixate on that 41-foot number and assume that it was the worst-case scenario.

But it wasn’t “the” worst case scenario. It was a “likely” worst case scenario, with the word “likely” implying certain odds. And, in fact, as the weather conditions conspired against Fargo, the river surged — and on Thursday, the forecasters had to push up their crest prediction past 41 feet — the height people had come to believe was the worst case.

The new crest prediction caught many in Fargo by surprise; and late Thursday, the city had to scramble to raise the height of its sandbag dikes. But it shouldn’t have been a surprise — because in fact, these new ranges aren’t certainties at all, any more than the single number was. They’re simply predictions made with some high level of confidence.

Read more at http://www.grandforksherald.com/event/article/id/112756/