how accurate is the altitude measurement in mobile phones

Question

How accurate is the altitude measurement from a mobile phone's GPS? I've gathered that the lat/long can vary by hundreds of meters but is that same level of uncertainty present in the altitude values?

In particular I'm working with Windows Phone 7 but I'm sure that this question applies to other mobile devices. I expect that there are only a few GPS chip manufacturers and the same chip would be used by different phones.

This question deals with how it is calculated but it doesn't mention anything about accuracy or reliability.

Answer 1

Since the triangulation hands back a point in space, the same inaccuracies would apply to the Z axis like they do to X and Y.

In other words, it's no more and no less accurate than the accuracy of the LAT/LONG.

Answer 2

I don't know specifically about the iPhone, but elevation is often much less accurate than X,Y information from a GPS. Here are some sources of information about this.

Answer 3

It requires fairly complicated math to understand fully, but no the altitude on a ANY gps is not as accurate the lat/long position.

http://weather.gladstonefamily.net/gps_elevation.html

http://gpsinformation.net/main/altitude.htm

http://www.sawmillcreek.org/showthread.php?83752-Do-civilian-GPS-unts-do-accurate-altitude

quote from the third link - "The altitude error is much greater because it is a satellite based system. If you think about it, the best satellite positions for a perfect read are going to be evenly distributed in an imaginary sphere surrounding you. Unfortunately, since you are standing on the earth, that rules out half the sphere because you need line-of-sight to the satellite. As a practical matter, it even rules out a constellation with satellites close to the horizon. So, generally speaking, your fixes will be overhead--which means that the cumulative error is mainly in the vertical plane. So, I think the offhand estimate is vertical error = 1.5x horizontal error."

Answer 4

Vertical error is specified as 1.5 * horizontal error. You must also allow for local deviation between the geodetic model and the planetary surface because the geodetic oblate spheroid model is an approximation only even when local correction tables are in use.