I agree this is probably the case. Even my local rural weather service predicts 10 days out (although anything more than 2 isn't necessarily accurate). Payload probably contains at least a day or two in the future for the given location (zip code?).

What would be a good test, if you can determine how location is determined, would be to be offline and pick a location on the other side of the country or world and see if it is able to communicate and pull down the forecast.

Although you'd really have to know the specifics of the data. Simple thing like temp and a few types of weather (sunny, cloudy, rain, etc.) could be encoded very compactly. Given the type of bandwidth most people have, it's not inconceivable that updates could include the entire country or even world and be cached. Probably a drop in the bucket compared to, say, a Youtube video.

I agree this is probably the case. Even my local rural weather service predicts 10 days out (although anything more than 2 isn't necessarily accurate). Payload probably contains at least a day or two in the future for the given location (zip code?).

What would be a good test, if you can determine how location is determined, would be to be offline and pick a location on the other side of the country or world and see if it is able to communicate and pull down the forecast.