Re: errors in GPS reported height

Bill, thanks for the nice post. Was working on my thoughts on the
problem...

If you think of it as strictly a geometry problem, where each GPS
satellite give you your position as a sphere, with the radius being your
distance from the satellite. The satellites with the best geometry for
giving you the best Latitude and Longitude positions would be all the
satellites that are along the horizon (assuming good signal). The circle
of the horizon. None of these satellites along the horizon will give you
any good information about altitude.

While the best position for a satellite to give you altitude is straight
up or straight down. Satellites that are straight up or straight down
would give no useful information about your latitude and longitude.

Another way to look at this would be to look at the number of satellites
that are "close" to the "best" geometry. Let's say within 45 degrees of
the optimal. So the "close" geometry satellites for determining altitude
would be a cone around a line straight up or straight down. The "close"
geometry satellites for Latitude and Longitude would be everything else
above the horizon. An area 2-3 times as large, containing 2-3 times as
many satellites to choose from, assuming equal distribution. 2-3 times the
number of satellites to choose the ones that have the best geometry, to
give the most accurate solution.

Also, think about how spheres can intersect each other. Latitude and
Longitude have the full spectrum of intersections. From, a sphere within a
sphere, the intersection being the sphere itself. To, a sphere touching a
sphere, the intersection being a point. And, all the circular
intersections in between the two. With altitude, the Earth blocks half the
possible intersections. There can never be the point intersection, or
anywhere close, of two spheres.

So, you can probably see why it's a harder problem to determine altitude.
It's the Earth, it's constraining the geometry of the problem for altitude
more than it is Latitude and Longitude.


--
Craig Fink
Courtesy E-Mail Welcome @ WeBeGood@xxxxxxxxx
--

.

Quantcast