March 2009: Michael Stensgard used one of his family’s boats to go past a neighbor’s house on the way to his home as the Red River crested at Fargo-Moorhead .

Richard Tsong-Taatarii, Star Tribune

Weather Service backing off from fractional crest predictions

  • Article by: BILL McAULIFFE
  • Star Tribune
  • February 19, 2012 - 9:01 PM

Flood forecasting might be an advancing science, but flood forecasts themselves are on the verge of becoming less precise.

Concerned about floods that defy even sophisticated models, National Weather Service officials are moving toward incorporating levels of uncertainty into their forecasts, backing off from predictions of crests within fractions of a foot.

"People are focusing too much on the dot. We need to portray that it's not just the dot. It's a range, especially as you go farther out in time," said Diane Cooper, hydrologist with the National Weather Service in Chanhassen. "A tenth of a foot was portraying confidence and accuracy that's not necessarily true."

Forecasts might soon predict crests within a range of a foot or more, Cooper said.

The Weather Service was criticized in 1997 for forecasts that were lower than the ultimate crest of the Red River, where flooding devastated both Grand Forks, N.D., and East Grand Forks, Minn.

But even after years of advances, there have been surprises. Following heavy rains in September 2010, Cooper noted, forecasts for the Minnesota River at Mankato were accurate and caused little alarm, while just downstream at St. Peter and Henderson, the river reached an unexpected record height.

That's because the models were based on spring floods, Cooper noted. In September, with floodwater moving over crops, grassy banks and through shrubbery and trees with leaves, crests can be slowed and swelled.

"We're getting more informed about the factors that influence the forecast," Cooper added. "Because of that, we're recognizing that some of our previous assumptions are maybe breaking down."

The shift to range-based flood forecasts is part of an effort by the Weather Service to improve how it conveys often-complicated flood forecasts. It may look for ways to more clearly link crest forecasts with likely local impacts, for example.

Another option could be expanded use of interactive online "inundation maps," which predict where water will go when rivers reach various heights. Such maps are already used in many cities in the southeastern United States, and can be seen at

In St. Paul, meanwhile, officials are considering new strategies for dealing with flooding, even as drought continues to minimize flood risks this spring. "Last year was a big scare, and it got people thinking," said city engineer John Maczko.

One key feature would be a wall that could allow for a more rapid, less long-lasting and less expensive closing of Shepard and Warner Roads along downtown. Plans might also include a berm or levee to protect Lowertown from water that might back up out of the Bruce Vento Nature Sanctuary.

Shepard and Warner Roads have been closed three times in the past two years becasue of floods. The envisioned wall would have gaps for traffic that could be closed during flooding. It wouldn't keep Shepard and Warner Roads from flooding, but Maczko said it would have advantages. Closing the roads with dikes and sandbags costs as much as $1 million and can take as long as a week. Closing a gap in a wall would take 24 to 36 hours and cost far less.

This spring, the chance of even low-lying Lilydale Park flooding between now and May 20 is only about 2 percent, according to recent projections. Much of Minnesota has experienced one of the driest and warmest late-summer, autumn and winter stretches on record.

State officials have encouraged farmers to buy crop insurance ahead of a March deadline to protect against possible drought-caused losses.

Bill McAuliffe • 612-673-7646

© 2018 Star Tribune