They knew it was going to be a lot of rain. They had no idea it was going to be that much.
In fact, many scientists and others watching the Colorado weather forecasts for the week of Sept. 8 were happy that significant precipitation was on the horizon.
The coming rains were seen as a welcome respite from the state's drought, according to Klaus Wolter, a researcher at the Cooperative Institute for Research in Environmental Sciences (CIRES) at the University of Colorado. He works in the National Oceanic and Atmospheric Administration's Earth System Research Laboratory.
Volunteers help unload evacuees and their pets from an Army CH-47 Chinook helicopter used to reach communities in the foothills above Boulder, Colo., where roads and bridges had been washed out. Photo by Sgt. Jonathan Thibault, courtesy of Flickr.
"We were actually all looking forward to getting 2 to 4 inches of rain," Wolter said at a forum on the floods held yesterday on the university's Boulder campus.
In an area like Boulder, which receives only about 20 inches of rain each year, that is a lot. So in terms of predicting a week of heavy rains, early forecasts were accurate.
However, the jump from a lot of rain to "biblical" amounts, as one National Weather Service bulletin put it, was not forecast. The National Weather Service and others are reviewing the event to see whether they can improve their forecasting of major and damaging weather like this.
Nezette Rydell, the meteorologist in charge of the National Weather Service office in Boulder, said the biggest debate on the forecast floor during the event was where the rain would fall, and how much rain it would be.
Big questions hung in the air
"There was a lot of discussion," said Rydell. "Is it going to set up over Denver and Bear Creek, Turkey Creek? Is it going to set up over Boulder, Jamestown, Four Mile Canyon, Big Thompson? Those few miles make a big difference in the impact."
As to how much would come? The Weather Service did not foresee the record-breaking 24-hour precipitation in Boulder that lasted through 7 a.m. Thursday, Sept. 19.
"We did not forecast 8 inches of rain," said Rydell. "And if you are a meteorologist who did and if you have a email time stamp before 6 a.m. Wednesday, I'll be happy to talk to you about your forecast."
Rydell said the improvements in computer power that the Weather Service has implemented, as well as getting more observational data into models, could improve their forecasts of severe weather, although the most extreme events are always difficult to foresee because they are so rare.
In mountainous areas, topography can also make forecasting harder, she noted.
"In the mountain West as a whole, it's just more difficult."
Computer models flunked the test
What interested Kelly Mahoney, another CIRES/NOAA scientist who works on forecasting extreme weather, was how weather models, which had a fairly accurate medium-range forecast for the Colorado rains, became less accurate for the short-term forecast. Normally, weather models should get better the nearer in time they are to the event they are forecasting.
"As the event got closer ... things actually started to go off the rails a little bit and show actually less precipitation forming. The picture got a lot less clear," said Mahoney.
"To tell you the truth, none of [the models] did fantastic consistently across the board. It was very much a mixed bag in terms of a forecast."
That's a tough situation to be in as a weather forecaster. Since the Boulder rainfall was so far out of the norm, neither models nor forecasters were likely to anticipate it, said Wolter.
According to a report on the flooding led by Jeff Lukas, a researcher a CIRES and the Western Water Assessment, there have been three multi-day rainfall events in Colorado similar to the recent floods. They occurred in September 1938, June 1965 and May 1969.
Those storms had rain totals ranging from 10 to 16 inches. The recent record set in Boulder was 17.15 inches.
"In the context of the entire Front Range, this was a very rare precipitation event and in some respects unprecedented," said Lukas.
Scant help from the record books
As to whether the rainfall can be given the moniker "1,000-year," the researchers were more cautious.
That's because good weather records only go back to the late 1800s, so the probabilities are extrapolated from historical records using statistical assumptions that have a lot of uncertainty.
Researchers at the event did say it was likely that in some parts of the flooded area, the flows reached a 100-year flood stage, but the U.S. Geological Survey is still working on mapping out maximum flows in various drainages.
Nolan Doesken, Colorado's state climatologist, has an uncanny ability to recall extreme events throughout the state's history.
At the forum, he noted that while many Coloradans think of September as a time when summer's late afternoon thunderstorms are over and they can enjoy a month of sunny, dry weather, from time to time in the region's history, conditions are set up just right where moisture from the tropics makes its way north.
Rattling off several years in the past where this happened, Doesken pointed out that the severity of the recent floods depends on your perspective.
"Through history, we have monsters really pretty often somewhere in the state. They're not always here," said Doesken.
"What's a 1,000-year event at a certain point may only be a one-in-every-five-or-10-year return frequency for the area of the state. And it is something that happens every year over the area of multiple states," he said.
Reprinted from ClimateWire with permission from Environment & Energy Publishing, LLC. 202-628-6500.
E&E Publishing is the leading source for comprehensive, daily coverage of environmental and energy issues.
Click here to start a free trial to E&E's information services.