Burns
Avatar of War Slayer
I only "trust" Weather Underground, as they get temperature and wind data from local people's weather stations that they allow the website to pull from. If one person's station seems out of wack too often, I switch to a different one.
AFAIK, for the 10 day, Weather Underground pulls from the national weather service American model, which gives a county by county forecast that they update multiple times a day. All they do is put in the current conditions, including the Jetstream, trade winds, and everything that effects the weather on the meta level. Then it spits out a best guess. They slightly tweak the variables of the inputs and then evaluate all the outputs for the final forecast. The farther out it is, the less and less accurate it is going to be.
If you live in an area that has highly variable weather patterns, it's just not going to be able to get any more accurate. For example, I think Houston has a built in chance of rain almost every day at around 4pm during the summer, depending on if the wind is blowing ocean humidity inland or not. That's not something the model can pick up multiple days in advance.
Finally, here is the official statement from the national weather service on what X% chance of rain actually means:
This guy from England explains it a slightly different way and also discusses how they evaluate numerous model outputs per released forecasts (there is an American model and a European model of weather forecasting):
www.rmets.org
AFAIK, for the 10 day, Weather Underground pulls from the national weather service American model, which gives a county by county forecast that they update multiple times a day. All they do is put in the current conditions, including the Jetstream, trade winds, and everything that effects the weather on the meta level. Then it spits out a best guess. They slightly tweak the variables of the inputs and then evaluate all the outputs for the final forecast. The farther out it is, the less and less accurate it is going to be.
If you live in an area that has highly variable weather patterns, it's just not going to be able to get any more accurate. For example, I think Houston has a built in chance of rain almost every day at around 4pm during the summer, depending on if the wind is blowing ocean humidity inland or not. That's not something the model can pick up multiple days in advance.
Finally, here is the official statement from the national weather service on what X% chance of rain actually means:
This guy from England explains it a slightly different way and also discusses how they evaluate numerous model outputs per released forecasts (there is an American model and a European model of weather forecasting):

What does a 30% chance of rain mean?
Will it rain today? Your weather app might say 30%, but do you know what that actually means? Ken Mylne explains why it's probably not what you're thinking...

- 1