# Does barometer improve gps position accuracy?

It is clear that a barometer is more accurate in measuring altitude than GPS. However is it true that the barometer improves GPS position accuracy?

Would be a GPS device and a GPS+barometer device equally accurate if we didn't look at the altitude?

• "It is clear..."? You seem to have both a particularly low quality GPS device and a particularly magic barometer. What devices are you using, and what numbers do you have for their errors in altitude? GPS should be accurate to meters, barometers can be off by hundreds of meters. – user10851 Jun 7 '16 at 9:10
• That is just not so. You can buy GPS devices that have cm or even mm accuracy. The designers of your cell phone just didn't care about that. – CuriousOne Jun 7 '16 at 14:16
• @ChrisWhite Barometers have terrible absolute accuracy (because air pressure varies, obviously), but they can have rather good differential accuracy. So if air pressure isn't varying much and you have a good zero point, barometers can be very good. – tfb Jun 7 '16 at 14:19
• @Farcher Airplanes use barometers because they don't care about height above ground. The only thing that matters is the pressure at which they are flying. That reading is often converted to feet based on a standard atmosphere, but this is just tradition and that is certainly not an accurate measure of the number of feet between the plane and the ground. The only way to make a barometer accurate is to calibrate it (within the last couple hours, and not very far from where you are using it). That calibration is done with a GPS! – user10851 Jun 7 '16 at 17:16
• To get ~ cm's of accuracy from GPS you are talking about differential gps. This requires a local base station that's surveyed-in to cm's of accuracy, and permissions to communicate with it. – docscience Jun 7 '16 at 22:50

The science of sensor fusion provides, by the combination of multiple, seemingly redundant measurements, a reduction of variance in the combined measurement. Basically speaking, for independent, identically distributed noise processes, the signal to noise of combined $n$ measurements is increased by the factor of $\sqrt{n}$.