The Great Chicago Weather Test

[Tracy Timberlake]

Which Chicago TV station has the most accurate weather? To many, the first thought that comes to mind is 'none'. But Chicago TV weather forecasters have some amazingly accurate days...of course they have some where they miss by a mile too. For the month of November, we made this simple test. How close did they come predicting the temperature and could they predict precipitation. The test was done using their 9 & 10 o'clock 5-day forecasts. Using channel 2, here's an example of how it worked:

On Monday, Nov. 2, Ed Curren's 5-day forecast said Saturday's high temperature would be 62 degrees. On Tuesday he was still predicting Saturday to be 62. On Wednesday, Steve Baskerville predicted Saturday would be 60. On Thursday Baskerville's prediction for Saturday changed to 63 and on Friday Steve's prediction for the next day was 65. The actual high temperature for Saturday was 71. So the prediction was off by 9 degrees on Monday, 9 on Tuesday, 11 on Wednesday, 8 on Thursday and 6 on Friday....giving them a total 43. You got one point for every degree you were off. This, like golf, was a contest where the lowest score wins.

The other half of the test was for precipitation. Rather than get too complicated, it was done simply on the basis of 'did they predict rain/snow or not?...and did it'? Five points was given for predicting precipitation and not getting any...and just the opposite, five points for predicting a dry day and it wasn't.

These results were recorded Monday through Friday for the month of November, 2009. Three days were tossed out because not everyone had a nine or ten o'clock show every night.

After listening to 90 weather-casts, filled with words like might, should, may, could, chance, maybe, scattered, possible and possibly, these are the results:

WGN- 299
WLS- 362
WBBM- 387
WMAQ- 402
WFLD- 403

Channel 9 had two nights during the month (5th & 23rd) that their total was only nine points, while WMAQ had a 46 on the night of November 3rd. While channels 9 & 7 were very consistent, channel 5's problem was despite having some very accurate days, when they were wrong, they were REALLY wrong.

NOV.........2........5........7........9.......32
2..............36......44......32......19......42
3..............40......46......33......25......36
4..............32......38......29......18......31
5..............30......34......28.......9.......33
6..............20......14......17......17......17
9..............30......31......36......35......31
11............29......28......26......24......21
12............27......16......22......16......27
13............20......27......15......19......33
16............20......17......13......17......19
18............23......25......24......24......21
19............25......30......30......19......24
20............14......19......23......29......22
23............19......15......17.......9.......20
25............22......18......17......19......26

Once again, this test was rather unscientific. There is a chance that it might possibly and maybe could be done again...and should.