Are you better off guessing than asking a weathercaster? Missouri family finds out
By AARON BARNHART
The Kansas City Star
Monday night, as a line of clouds formed along the eastern edge of Kansas, a Missouri man was taking a popular New York Times blog by storm.
“How Valid Are T.V. Weather Forecasts?” was the headline on the Freakonomics site maintained by a Times editor and the co-authors of the Freakonomics best-selling book.
The author of the post, J.D. Eggleston, 41, who lives near Maysville, Mo., described how he and his fifth-grade daughter, Stefanie, last year spent several months charting the forecasts given out at 10 p.m. on all four Kansas City TV news stations, as well as forecasts from the National Weather Service. Then they compared the predictions to the weather that Kansas City actually got.
Their conclusion? Weathercasters “say they care about accuracy,” he wrote in the 2,200-word posting, “but their actions say they do not.”
What started as a simple elementary-school project — taking a few days’ worth of tem-perature and rainfall readings — morphed into a 220-day data quest that stretched into the summer and then sixth grade. When it was done, father and daughter produced a series of charts and graphs that measured how the TV meteorologists had performed in these measures: the ability to predict rain and the accuracy of their long-range forecasts.
On the whole, the Egglestons found that the four stations were more alike than not. All said the gold standard for next-day forecasts was to get within three degrees of the actual temperature — and all forecast highs and lows within that margin of error. But after that, accuracy dropped off quickly.
As J.D. Eggleston wrote: “The conclusion to be drawn here is not so much that one station is better than another, since all of them seem to be similar in accuracy … (but that they) did not do a good job by their own definition of plus/minus three degrees beyond two days out.”
The report, plus graphs, quickly became one of the Freakonomics blog’s most commented-on pieces of the year, with hundreds chiming in to support or take issue with the amateur weather critics.
“They should do this with the NYC/Jersey area,” was the comment echoed by other readers in various parts of the country. “The predictions are horrible.”
Gary Lezak, KSHB’s chief meteorologist, read the piece start to finish.
“I’m very pleased with the results,” said Lezak. “The first two days of the forecast are very important, and as tough a year as (2007) was for forecasts, ours were still the most accurate.”
Indeed, Lezak was rated best among local forecasters for his ability to predict the next 48 hours of weather. However, he also was one of the least accurate in his three-to-seven-day outlook, and the Egglestons reported that KSHB’s weekend weathercaster, Jeremy Nelson, “is better than Lezak longer range.”
But the tone of many comments underscored the no-win position that TV forecasters often find themselves in, especially during the severe storm months of April and May. They are paid to explain the weather in a friendly, confident way. The stations use words like “reliable” and “accurate” in their promotions.
And yet, as Lezak put it, forecasting is really “nowcasting.” It about looking at data available at that moment and deciding what will likely happen next.
The perils of nowcasting were never more obvious than the night that Eggleston’s critique went online. In the early hours of April 22, a dramatic thunderstorm swept through Kansas City — a tempest that none of the local TV meteorologists saw coming until after the late news had ended.
Next page >
To reach Aaron Barnhart, call (816) 234-4790 or send e-mail to TVBarn.com. | Aaron Barnhart
Join the discussion
Share your observations and experiences about news. Lively, open debate is the goal, but please refrain from personal attacks or comments that are racist, vulgar or otherwise inappropriate. If you see an inappropriate comment, please click the "Report as violation" link to notify a KansasCity.com editor. Thanks for your feedback.