posted
I think the long term forecasts, 5 or 7 day are bumkus. Mainly because they said it wouldn't rain on my sister's wedding reception and it did. I think they should have to have accuracy ratings.
Posts: 11017 | Registered: Apr 2003
| IP: Logged |
posted
A lot of times the published weather is either the latest reading, which can be hours old because the infrastructure for tracking the weather was designed to collect data for analysis, it’s not made for real-time tracking, or is simply the predictions made earlier for that present time. Either way you’ll get inaccurate results, none of which are really the fault of the people publishing it, since they simply can’t afford to staff out as many people as it would require to get the current conditions right in every area.
Pooka, an accuracy rating would be rather meaningless. You could calculate the statistical likelihood of any weather report in general being correct, but the reason computer models are wrong is two-fold: the rules of weather development aren’t completely accurate (they can’t be) and more importantly, the data isn’t complete/ Small changes in weather in a localized region only take a short period of time to develop into big changes, and the likelihood of that happening is based on data that, by definition, those models don’t have (or it would be incorporated into the system). Predicting weather is so inaccurate because it’s one of the hardest endeavors of man-kind quite honestly. There’s no other system in the universe we know of that is complex and chaotic as our atmosphere, with the possible exception of Venus’s atmosphere.
but at the time I was checking the weather, the current weather said "Partly cloudy" and when I refreshed it said "Showers", neither of which was accurate. Someone's job should be to look out the window.
edit: It seems the weather wasn't inaccurate, just psychic.
*watches giant, black cloud roll over the mountain*
posted
I'm pretty sure the 7 day forecast is a marketing gimmick. They pull those numbers out of their... uh... back pockets.
Posts: 11017 | Registered: Apr 2003
| IP: Logged |
posted
I had a chance to speak with the Producer of the weather department at a local TV station. He was the brains behind the talking head you see at night on the actual newscast, having a degree in weather science and working on a masters in climatology. He filled in from time to time, but was happy actually creating the forcasts and maps, rather than being on TV every night.
I asked him about 7 day forcasts. His response: "You want to know the truth? 2 days out, we're 50% accurate. 3 days out, 40%, 4 days out 30%, 5 days out, 20%. When it comes to the 7-10 day forcasts, honestly, sometimes I just make them up."
He then went on to explain that the computer models have a hard time predicting that far out and they have to choose between several possibilities.
My personal observation is that they normally do one of two things with the extended forcast.
1. Go with the averages and predict based on climatological history.
2. Predict the worst case scenario based on the models.
Option one gives you greater accuracy. Option two keeps viewers/listeners checking back for updates to see if that really bad weather is really going to happen.
Posts: 995 | Registered: May 2003
| IP: Logged |