Your hand-held GPS is lying! Comparing GPS height with true ground height.

[@bmenrigh]

I’d be interested in the fourier transform of the error to look for specific frequencies. Perhaps the swings have a period of 5 minutes or some other “nice” period. You’d probably need to collect a lot more data though to get good estimates.

[@AnthonyDunk]

Interesting. Also note that Handy GPS computes and subtracts the geoid height offset using the EGM96 model. There is a setting to turn this off under the Advanced page in preferences and you can see the value being used by pressing the button to the right of the altitude on the main page (-30 metres where I am). This is a static offset though so will not account for the low frequency shift you've observed.

[@martinshreder]

Modern phones claim improved elevation measurement accuracy via use of integrated barometric pressure sensors. Would be interesting to see if these barometric corrections reduce the fluctuations shown here. ( i.e. RidewithGPS on a newish iPhone can log a hike, and allow export of a GPX file with and without barometric corrections)

[@robertsgeospatial]

I don't have a barometer on my phone. I wonder how well that would work... Not well, I suspect.

[@janismateuss]

@@robertsgeospatial Phone with a barometer is essential for accurate altitude. We use them for flight tracking, and FAI specify for certification barometer in the device is required for valid tracklogs. I've done my testing and I would trust my altitude to within 1-2m accuracy on my android + barometer device. As I study a lot of different tracklogs, on devices with no barometer I often see errors in the tens of meters, up to ~80m, and noisy signal, with erroneous spikes up and down.

[@sethwyma403]

Why is the z-axis specifically less accurate? Based on my limited knowledge of GPS I wouldn't expect errors to be meaningfully different on z- vs. x- and y-axes.

[@robertsgeospatial]

Look up GPS trilateration. For an accurate result you need a good spread of satellites about you. Horizontally, the birds are well distributed about you. Vertically, they are all above you! So it is a sub-optimal geometry for solving in the vertical axis.

[@PerMortensen]

@@robertsgeospatial Do you have any thoughts on why you would have these consistent low frequency errors, rather than just high frequency noise? Related to the positions of the satellites as they move through space?

[@robertsgeospatial]

@@PerMortensen Either space weather/atmospheric conditions, the changing positions of birds in the field of view, or the changing combination of birds in the field of view as they enter and exit. As birds move about the sky you might expect the solution accuracy to change slowly. As a bird enters or exits the field of view you might expect a step change in accuracy. Looking at the data, I think you could convince yourself that both of these effects are visible. Perhaps the atmospheric effects are responsible for the higher frequency noise. I am no expert on this - all just speculation!