Not to mention that their forecasts are presented in a way that is statistically incoherent. Look at their hourly forecasts, then click to break down by 15 minute intervals. It is common to see things like "rain during 2:00-3:00 = 30%", and then also have something like 50% during one of the 15 minute intervals within that time range. I have yet to come up with a hypothesis for what those percentages could mean, and have that scenario be possible.
An unlikely, but theoretically possible way to get 50% inside of 15 minute interval but 20% over an hour would be if the model prodicted very widespread rain in a band moving over an area and time weighted the percentage.
Common misconception. The Chance of precipitation is actually stating that historically, given these conditions x% of days produced precipitation. It is a statement of the odds that it will rain at all.
http://www.straightdope.com/columns/read/854/what-does-it-me...
NOAA: http://www.srh.noaa.gov/ffc/?n=pop
http://www.usatoday.com/weather/resources/askjack/wadefine.h...
My real guess is that they have independent models for different time resolutions.
a few years from now: sells to Facebook for $1B ;)
I guess I will have to write a quick perl script to generate my own images now.