peter wrote:
Mxsmanic wrote:

"Jim Macklin" writes:


they give altitude, accurate to within a few feet.

Unfortunately, no, they do not. GPS is accurate for lateral
navigation, not vertical navigation. GPS altitudes can easily be off
by as much as 200 feet at ground level in comparison to a correctly
set altimeter, and at altitude the disparity can reach 500 feet.

The reason for this is that the angles used for triangulation of
lateral positions are large and permit a high level of precision, but
the angles for triangulation of altitude are very small and it's very
easy to be off by a wide margin.


This indicates a basic lack of understanding of GPS technology. The
GPS receiver never deals with measurement of any angles nor with
triangulation. What is measured are the precise times of arrival of
the signals from the satellites. Since the satellites encode the
signals with timing information from their sychronized atomic clocks
and also send detailed orbital data to define their own positions, the
receiver is able to determine the relative distances to the various
satellites based on the speed of light/radio and the observed relative
signal delays. Using this distance information together with the known
positions of the satellites then allows for a determination of the
position of the receiver. Note that this never involves a measurement
of any angles.

It is true that altitude measurements are generally somewhat less
accurate than horizontal position measurements due to the basic
geometry of receiving satellite signals from only the satellites that
are above you. Ideal measurement of altitude would also involve some
satellites below you but of course their signals are blocked by the
earth. Similarly, east-west positions are a bit better accuracy than
north-south since the satellites are equally likely to be east and west
of you but there's a greater likelihood of them being to the south
rather than the north (at least from the northern hemisphere).

My long-term evaluation of GPS altitude accuracy has shown that I get
values within 35' of accurately surveyed altitudes at least 95% of the
time ever since Selective Availability was turned off. Using the WAAS
correction data improves this to get the accuracy down to 20' with 95%
confidence. Both of these are based on having a reasonably
unobstructed view of the sky (which generally isn't hard in an aircraft
unless the antenna is poorly positioned).

Your impression is supported by the published nominal accuracy of GPS:
+/- 10m horizontal, +/- 20m
vertical. The reduced vertical accuracy comes from the fact that
although the sats are at ~20,000 km
their horizontal spacing can be much larger than that. As you say, the
fix is 3D and it always is a 3D
solution once the minimum 4 sats are acquired for the solution. In
addition, if your receiver can hold more than 4 sats it may be able to
average the data to improve the fix. On my 12 channel boat GPS I see a
HDOP of ~1m these days. More important perhaps is that the aviation nav.
chart I use is not referenced to WGS84 which is strange as all my
sailing charts are. I understant that the "powers that be" are trying to
settle which geoid to use but I would imagine that the height
corrections could be quite large when it all gets sorted out. Is that
why the peak obstacle height over the sea is never marked as 0' (i.e.
MSL is not conforming to the geoid of reference)?

Cheers MC(student pilot)