For specific point broadcasts, I like to use darksky and NWS hourly weather forecasts.
Darksky can be ridiculously accurate in the <12 hour range, before that, it's essentially an ensemble model of a lot of different providers. But, once you're under 12 hours, their own engine takes over, and it's pretty good.
NWS point forecasts are generally the best in the business for a few days out.
Here is Darksky's hourly forecast for Saturday at Folsom.
If you don't want to click, they show kickoff temperature in the low 60s, dropping to low 50s by the time the game is over. 25% chance of light rain during the game; 0% chance of rain that day before the game, and 0% chance after the game. Wind about 7 MPH.
Here is NWS' point forecast for Folsom (if you click through, pay close attention to the "48 hour starting period" text box at the bottom of the gray field - the way their links work, I can't get a static link to Saturday, so you need to adjust the 48 hour period if you click on this link any time after 1pm on Tuesday afternoon).
If you don't want to click, then as of right now they have kickoff temperature at 59º, dropping to 50º by midnight. Wind out of the north around 5 or 6 knots*. 24% chance of rain, and a "slight chance" (10-20%) of thunderstorms during the game. Both before and after the game, they have a 30% chance of rain and a "chance" (30-50%) of thunderstorms.
*I have the wind units set for knots, because I use this engine for my sailing forecasts - just be glad the engine is smart enough to not try and generate sea state forecasts for Boulder. Although it did put in "Fire Weather" options for Boulder, which I hadn't seen before - that looks interesting, and I will probably end up wasting 20 minutes of time playing with it.