Each of these media outlets would like you to believe that their weather is the best, that if you miss their forecast, you'll be left out in the rain -- or worse. But to those who've canceled cookouts that were supposed to be rained out but weren't, who dressed wrong because they believed a weather report, or who left their umbrellas at home and got soaked, it probably seems like a lot of the time, they just don't get it right. So why all the fuss?
The truth is that weather forecasting is one part science, three parts hype. For newspapers like the Charlotte Observer, which presents its forecast in an elaborate half-page spread with premier ad space below, weather is a necessity. For TV news stations like WBTV, WCNC and WSOC, weather can make or break a station's ratings.
Some of these media outlets had a better track record than others during the period we analyzed, and we purposely picked one of the toughest periods to forecast for each year -- from mid-February to mid-March -- when, as it turned out, anything from a blizzard to 80-degree temperatures is possible. With the exception of the Observer, we used the forecasts they display on the web to nail them down to a written prediction. Under these conditions, the television stations averaged a total of six to nine degrees off on the temperature per day and got the conditions -- rain, snow or cloudiness -- right about 67 percent of the time on average in their five-day forecasts. The Observer's forecast, prepared the day before, lagged behind at 56 percent.
Where do they get this stuff?
Before we tell you more about how they did, it's important to tell you a bigger story -- how they did it.
Most Americans probably think the forecasts they read or see on the local news or on The Weather Channel are free, but they're not. They cost every American citizen about $4 a year. The federal government is the main player in the weather forecasting business in this country, and if it suddenly stopped cranking out detailed forecasts every day, the weather would pretty much immediately disappear from local news and The Weather Channel would dry up.
Every day, the National Oceanic and Atmospheric Administration (NOAA) and its sub-agencies send up hundreds of weather balloons from fixed locations across the country and gather atmospheric data from satellites, radar and airplanes. It's a complex process, but the result is essentially reams of data that is then compiled into various computer models predicting the weather for most major areas.
Anyone who wants it can download it, and weather forecasters use those models, which often make similar predictions for the same day, to make their own forecasts. Usually what this means is that local forecasters pick the model they think is most accurate, or fiddle with the predictions of a few others. According to the NOAA website, the law allows them to use the National Weather Service's forecast verbatim and attach their own graphics to it if they choose without crediting the NWS.
One station, WSOC, buys its forecast from a private company called AccuWeather, which builds its forecasts off National Weather Service models as well, and prepares our local forecast in Pennsylvania. AccuWeather President Dr. Joel Myers says the company's meteorologists consult with local forecasters here in Charlotte on the forecasts. Everyone we talked to for this article said they put their own work into their forecasts, which is probably true. But when we lined up the forecasts by all five media outlets side-by-side and looked at who predicted what and when, an odd pattern began to emerge. Over 80 percent of the time, near-identical mistakes were made by at least four of the media outlets on the same days in their five-day forecasts. At the same time, they all got the precipitation or cloudiness correct or close to correct on the same days in their five-day forecasts over 70 percent of the time.
"They're putting out this perception that is really not true that they have this technology that differentiates them from the others," said Dr. Judah Cohen, a staff scientist at Atmospheric and Environmental Research, Inc. in Boston. "It's marketing. No one really has anything really different. It's just the packaging of it. All the information is the same and it's mostly coming from the government. If you have a person who has knowledge of the local area, he may have a feel that under this situation this model may do better, but really there is no one who has a technological edge over the other one."
That said, who really has the best local weather? It's hard to say for certain since they were all so close, but if we had to pick, we'd say The Weather Channel by a hair. The cable station correctly forecast local conditions -- cloudiness, precipitation, etc. -- close to 69 percent of the time, and was off by an average total of seven degrees on the temperature in its daily forecasts.
Two local stations, WBTV and WSOC, tied for the most accurate conditions forecasts. Both stations' forecasts for clouds and precipitation were on target 68 percent of the time. WBTV missed less often, forecasting the conditions wrong about 18 percent of the time compared to 26 percent of the time for WSOC. WBTV was off by almost eight degrees total on average. WSOC was off by just over nine.
WCNC correctly forecast conditions an average of 66 percent of the time. Like WSOC, it botched precipitation and clouds forecasts 26 percent of the time and was off by just over nine degrees on average on its temperature forecasts.
The Observer's five-day forecast is put together by WBTV's meteorologists and turned in the day before at 5pm. WBTV also uses a more conservative approach to the forecast, and with the time lag, it's noticeably less accurate than television news. The Observer's forecast for clouds and precipitation was wrong about 30 percent of the time. It was on target about 56 percent of the time and close another 13 percent of the time. On average, its temperature forecasts were off by just over 10 degrees.
A matter of personalities
In the ratings game, perception is everything and good weather broadcasts can take a television station a long way. In a 2003 survey by the Radio-Television News Directors Association (RTNDA), 72 percent of local news viewers said weather was the most important piece of information they tune in to local news for. Some 97 percent of news directors in the same survey said weather is the number one reason they believe people watch television news.
But accuracy isn't necessarily as important to viewers as presentation as long as a media outlet isn't significantly less accurate than its competitors. For TV newscasters, presentation is everything.
"People go to the station where they connect most with the anchors whether it be the weather anchor or the news anchor," said RTNDA Director Of Marketing and Communication Noreen Welle. "Since a lot of them rely on the same types of weather services available to each of them, a lot of it has to do with the personality and the longevity of the weather person."
WBTV Meteorologist Al Conklin seems to know what Welle is talking about. He says he and his colleagues at the station spend about three hours a day preparing the weather. But it can be how it's presented that matters the most.
"This is like stock markets or medical assessments," said Conklin. "It's an inexact science. It's not as easy as we make it appear. We've got to put an air of confidence on when we go out there. If you don't have any confidence, it will show and people will turn you off in a heartbeat. We just can't afford to have that."
WCNC Meteorologist Terri Bennett agrees that how much people like you can play a big role in how much viewers trust your weather forecast.
"Some people don't want to get their forecast from a redhead," she said. "Some people like me because I'm a mother."
One thing it's important to remember, says Cohen, a weather scientist with a PhD, is that most of the TV news meteorologists who crunch the numbers in the nightly news forecasts are television broadcasters, not PhDs.
Complicating matters, AER, Inc. scientist Mark Leidner said, is the time crunch the broadcasters are often under.
"They try to get through it and get their forecasts out as quickly as they can," said Leidner. "Therefore they rely on this guidance (federal weather reports) quite a bit and everyone's weather often looks about the same. It's an interesting story. It's a bit of pulling the covers back on the dirty little secrets of weather forecasting."
The big snow surprise
On Tuesday, Feb. 24, the Charlotte Observer ran an article headlined, "Meteorologists shudder through jinxed winter," which questioned why meteorologists seemed to be repeatedly botching their snow forecasts. That same day, the Observer's own forecast for Thursday, Feb. 26, read "light rain or snow possible," something of an ironic understatement given that by the end of the day on Thursday a large snowstorm had dumped 11.6 inches of snow and nearly an inch of rain on the Charlotte area.
Such is the state of weather forecasting. Two days before the storm, WCNC's website was predicting "flurries." Like them, nearly everyone was predicting snow, and three different models showed everything from a dusting to a wallop. In their five-day forecasts, everyone started off conservatively, going with the dusting or some combination of rain and snow, and then gradually adding inches as Thursday got closer. Though it seemed as if they didn't know what was coming -- and in terms of inches they probably didn't -- most local forecasters did a better job forecasting the snowstorm than they did with rainstorms and high temperatures later in the month.
It seems that forecasters take a lot of hits for predicting rain that never comes, but no one seems to notice the rainstorms they miss, which is actually the bigger problem. Altogether, CL found that it actually rained on more days when forecasters didn't predict rain than when they did.
If that weren't enough, the Charlotte regional area can often be a difficult place to forecast for. The mountains to the west do strange things to fronts that cross them. Unpredictable things blow in from the ocean. And strictly local information can be hard to come by.
Like Conklin, Bennett says it sometimes can be a challenge to adjust local weather models because the closest weather balloon that gathers local information is launched each day out of Greensboro and weather forecasters have to adjust the data to fit the Charlotte area. That can make it particularly tricky to predict snow and other cold weather events because it's difficult to tell where pockets of cold air detected high in the atmosphere over Greensboro begin and end. Both also say their forecasts may appear to be wrong if there was precipitation in some places across the 22-county region, but not at Douglas International Airport, where the weather is recorded.
Bennett says the large -- and growing -- amount of paved surface in Charlotte means that our temperatures are typically slightly higher here than in surrounding areas and have to be adjusted upward.
Viewers also want longer-term forecasts, a fact that some say pushes forecasters past the outer limits of science. Forecasts are usually pretty good at two days out, says Penn State Meteorology Professor Lee Grenci. At three to seven days out, they're not nearly as accurate, yet that, Grenci says, is what almost everyone is using.
Contact Tara Servatius at tara.servatius@cln.com