In article ,
Wolfgang Schwanke wrote:

Bob Noel wrote in
:

as a rule of thumb, regular SPS GPS altitude error is roughly
50% greater than the horizontal error.

That would be minimal. But IME, while most of the time it's fairly
accurate, it can sometimes go wild. For example I was cross-countrying
at 3,500ish ft, when suddenly the GPS went through 3,000 2,500 ...
4,500 4,000 and then settled down at 3,500 once again as if nothing had
happend. It all lasted just a couple of seconds. It had not lost
coverage, that would be indicated by an error screen and the map
animation stopping which it did not; model GPS Pilot III. I understand
this is normal behaviour. If so, GPS altitude measurements are
completely unreliable.

A properly operating GPS receiver, with 4 SVs in view in an
appropriate geometry, should not report such extreme altitude
variations.

What is unknown: exactly how the GPS Pilot III selects SVs,
whether or not it will require 4 in view or let altitude "drift"
in order to maintain a 2D solution by using just 3 SVs, whether
or not it will report outages/drops within a few seconds, whether
or not it performs reasonableness checks on solutions.

I assume the GPS Pilot III doesn't have RAIM or FDE