[seles]

I don't think ambient temperature and humidity sensors would work very well because people usually aren't outside and when they are, phone is usually in pocket.

I love this idea too though. I'm guessing pressure is one of the least effected by being inside or in a pocket, while still being highly useful for weather prediction.

[breckinloggins]

It would be interesting to study this. I wonder if there are ways to normalize the data and still extract some useful data even if the devices are in the users' pockets.

Longer term, I can see these sensors being embedded in glasses or contacts, which could be an easier problem.

[fleitz]

Individually you're right, in aggregate I'd expect the data to be very accurate.

[btown]

The problem is that the effect of being in a pocket is not zero-biased when it comes to temperature (and possibly also humidity). You'd see temperatures being pulled towards normal body temperature on average.

[fleitz]

Yes, but in my rudimentary understanding of weather the importance is the pressure differential rather than the absolute pressure.

I believe the data in aggregate will provide a pretty good map of the pressure gradient which could then be fixed to the more accurate dedicated weather stations. Think of it like the 10,000 year clock which uses a clock known to drift in conjunction with a solar time fix to calibrate.

[nieve]

You may be right about pressure, but btown is talking about temperature and that's going to have much worse systematic bias exactly as he says. Your comment and his are nearly completely unrelated.