If you want to know what the weather will be in 10 days, is the 10-day forecast really any use? It often changes several times before the actual day arrives. (Oh yes, it hopes you don’t remember the “Sunny, 25˚” from a week ago, now that it’s hailing.) Should you bother reading it – or is it just a scientific-looking horoscope? Here’s an idea for a site to illustrate the uncertainty:

Weatherwiggle.com (yes, the domain is available) will graph how much the 10 day forecast is at variance with the actual weather, over time.

The dotted line in the graph represents the actual weather for a given day. The dots represent the predicted weather for that day, from 1 to 10 days ahead. Each frame of the animation is a different day.

As expected the 1-day forecast is mostly accurate and so it’s close to the dotted line. The 2 and 3 day forecasts are likewise fairly accurate. Beyond there however the predictions vary considerably from the real weather, and the animated graph ends up looking like a dog’s tail wagging. (Hence the site name.) In the fake example, you may as well ignore any forecast more than 3 days from the present.

Features to put on the site:

• choose your location
• today’s weather wiggle. A record of the predicted weather for today, from 10 days till the present, showing how accurately today’s weather was forecast.
• the next 10-day weather forecast – but with the diminishing accuracy of each future day shown. The accuracy should be computed based on historical data and would no doubt change based on local climate and time of year. An ‘ignore-beyond-here’ line would be good.
• animation of the daily weather wiggles in your locality for the last x amount of time
• cool data visualisations because people love that stuff

If you make the site let me know ok? (Or does something like this already exist?)