Is the 5-Day Forecast Bullshit? It seems they overpromise what’s possible. I don’t think we can currently predict the weather with reasonable accuracy more than one or two days into the future. I have to wonder if it was cutoff at 5 days because of some threshold of accuracy, if 5 sounded good to marketers, or whether that’s how many days of weather reports people can “chunk” at one time. By the time the 4th or 5th days role around, we’ve probably forgotten what was reported 5 days ago and only remember what was reported yesterday or the day before. But having a 5-day forecast is expected of our weather forecast providers.
To test this theory, I am running a little experiment here. Each day I’ll note what the forecast is for the New York, NY area 5 days from the current day, and after I’ve been doing this for 5 days I’ll also include the current day’s report for comparison. The reports are from Yahoo, which gets them from Weathernews, Inc., and which advertises they’re from the National Weather Service (which shouldn’t have anything to gain from marketing over-extended forecasts, right?).
Forecast for 1/20/2000: Thursday...variable cloudiness with a chance of snow showers. Low around 20. High in the upper 20s.
It occurs to me that the “extended forecast” – days 3 through 5 – have about half as much detail as the more immediate forecast. There’s “a chance of snow” but they won’t state that chance as a percentage. Still, if it gets significantly colder than 20 or warmer than 30 I’ll note that.