Shaun Tanner has been a meteorologist at Weather Underground since 2004.
By: Shaun Tanner , 5:41 PM GMT on November 12, 2009
There are a couple of things I want to discuss that came up in the last blog regarding the WunderCast Competition.
First, we decided to do the precipitation the way it is done for simplicity. Right now, participants put in a probability of precipitation from 0 to 100 for what they think will happen during that day. If it does rain, then the score is 100, if it doesn't then the score is a 0. The farther you were away from those numbers versus what you forecast, the more points you will accumulate.
As a scorer, I want people to hedge their bets at times when they are unsure whether it will rain or not. People who say 100 and it doesn't rain really get punished. I believe the points used to punish when you do get the precipitation incorrect is appropriate for now.
Also, someone said that the forecast duration should be less than two weeks per city. Let me tell you my stance on this. This competition was set up as a learning exercise. Those with no forecasting experience can learn from those with considerable forecasting experience. Thus, I chose to use the two week duration to get the maximum amount of variation in any given city without completely boring people. If the duration is cut down to one week, we may not see any precipitation, no cold spells, or anything interesting during that timeframe.
In addition, since the valid forecast days are only Tuesday through Friday, there are only 8 forecast days per city. Thus, it isn't exactly two whole weeks, but rather 4 days followed by three days off.
Anyway, my idea for this blog was to get some ideas on forecast cities. What cities would you like to see in the competition?
Comments will take a few seconds to appear.