Predicting the Weather
January 11, 2012 in Editorial, Top
During these crazy months of winter, weather websites will be visited so that people can plan their days accordingly. With so many websites to choose from, it is hard to decide which to look at on any given day. But do all of these sites agree on a specific weather prediction? Do some predict worse weather than others?
Centives looked at different and well-known weather websites to see if they tended to agree or disagree on the weather for the month of November. Three were considered (AccuWeather, The Weather Channel, and Weather Underground), but Weather Underground only gave a range of possible temperatures, so we did not use this site in our first analysis. Both of the other sites were programed to give the weather in Bethlehem, Pennsylvania. In 9 out of the 30 days in November, AccuWeather and The Weather Channel predicted the exact same temperature. On 12 separate days AccuWeather predicted a higher temperature than The Weather Channel, while the Weather Channel predicted a higher temperature than AccuWeather on 9 days.
On average in November, AccuWeather predicted a temperature that was 0.37 degrees Fahrenheit greater than the temperature predicted on The Weather Channel. The greatest discrepancies between the two predictions occurred on November 15th and 21st, when AccuWeather predicted a temperature 5 degrees higher than the weather channel.
Centives then looked at how all three weather websites (including Weather Underground) differed in their qualitative predictions of the weather. For 18 days in November, all three websites described the weather in the same way. For example, on November 2nd, all three sites predicted sun. But on 7 days, The Weather Channel predicted bad weather in contrast to the other 2 sites that predicted sunny weather. There was only one instance of The Weather Channel providing a better weather forecast than another weather website.
Centive’s hypothesis is that weather websites operate as a business, just like any other website. They want the most viewers so that they can generate the most revenue. AccuWeather generates revenues of somewhere between $10 and $25 million a year. On the other hand analysts predict that The Weather Channel had revenues of more than $100 million in 2008. Perhaps the Weather Channel is a more popular website because they predict worse weather than their competitors. People who are concerned about planning for the worst case scenario may opt to visit The Weather Channel over any of the other weather prediction websites.
Of course, this is just a hypothesis, and can easily be proven wrong. But it is important to think about the information that you read almost daily, and whether or not there is a reason that it is crafted in the way that it is.
Find us on Twitter and read some of our other weekly editorials about The Muppets, Twilight and Real Steel
I actually ended up doing something like this for a city in Edmonton: http://www.mikerobe007.ca/2012/12/fall-weather.html
Wouldn’t it be most instructive to compare the forecast with the actual weather?
And if you re-run the analysis using celsius, are there any noticeable differences? In other words, is much of this merely an artifact of measurement?