Re: GPS to calculate mileage

["Duncan Fletcher" <dfletche@gte.net>]

Yes and no. I don't believe the biggest problem with altitude is WGS84 datum
to local reference datum but the same thing that affects all measurements -
selective availability (the fudge factor the military throws in the time
data to deliberately make it inaccurate). The second major problem is
satellite geometry and resulting signal strength (the best geometry for
altitude involves 3 of the satellites being on the horizon which is not the
case for horizontal - satellites on the horizon have poor signal strength
assuming they are not block entirely). In any case, within the geographical
confines of a endurance race, the difference between geoid and reference
altitude will be very vary within a very small number. Most units also have
at least horizontal conversions from one datum to another built in - I have
never questioned whether the vertical datums are built in (450 feet in error
vertically to me is a lot more critical than 300 feet horizontally, so I
don't rely on the altitude which I would use only as means of finding
horizontal location from topos anyway).

But this whole discussion raises another question. How does AERC define
mileage - horizontially (which is what you get from a map if you do not do
some trig calculations involving altitude changes) or surface distance
covered as would be measure by a wheel?

Sampling time can be altered on most units (certainly is on my Garmin GPS
12). However, if coverage is lost it may make some assumptions that may or
may not be valid. It assumes straight line motion for a period of time
before it quits tracking. If coverage is only partially lost (4 satellites
are required, but if 3 are available) the unit assumes the previous altitude
and makes horizontal calculations under that assumption). The downside of
frequent sampling is that if I am recording a track, the limited size of the
track file will make it impossible to record the whole track (although the
accumulated mileage will not be affected).

Slow movement has an effect on the mileage accumulation. Small distance
changes are recorded in the track log, but ignored in the mileage
accumulation totals on the units assumption that small changes are probably
an artifact of selective availability as opposed to real motion.

However, your conclusion about the accuracies of handheld units without DGPS
correction (which requires an additional piece of hard connected to your
GPS) is valid. Horizontal accuracy is generally +- 100 meters and vertical
+- 150 meters. DGPS correction (which for the most part corrects for SA)
would be +- 10 meters horizontally and +-15 meters vertically.

Of course if the ride is a few tenths short, you can always add a couple of
laps around the parking lot - the crowded conditions might make it
interesting.


Duncan Fletcher
dfletche@gte.net


-----Original Message-----
From: Truman Prevatt <truman.prevatt@netsrq.com>
To: BMcCrary27@aol.com <BMcCrary27@aol.com>
Cc: rides2far@juno.com <rides2far@juno.com>; ridecamp@endurance.net
<ridecamp@endurance.net>
Date: Saturday, December 05, 1998 5:15 PM
Subject: Re: GPS to calculate mileage


As Angie pointed out the altitude measurement using hand held GPS is
very bad.  The GPS is a system developed by the Defense department to
distribute time world wide to within 50 nanoseconds.  Becuase of
relativistic effects associated with the rotating earth time tranfer
world wide is a very difficult technicial problem. This is the
overriding technical issue that drove the definition of the GPS program
in the early 70's.  The location is a by produce and is calculated by
measuring the time of transmission of the "same event" from multiple
space craft.  This calculation of location in the inexpensive hand held
units is to the reference World Geoditic System Geoid (WGS84 to be exact
and soon to become WGS 96 so most of the units will be obsolete <G>).
In order to get accurate location (including altitude) the US Defense
Mapping Agency world altitude data base is required.  This takes up 100
CD's.  Sophiicated military systems use it but not hand held units.

So an altitude quantization of almost 1/5 of a mile results.  So any
altitude variation less than 1/5 of a mile will not be counted in the
total milage.

The other issue with the hand held models in the calculation of distance
is sampling time.  Most are on the order of minutes to sample the
location and then that data is used by a statistical algorithm - a
Kalman Filter in most units - to calculate the distance.  The sampling
interval is on the order of 4 minutes in the hand held units.  So every
4 minutes a location is calculate.  The unit then takes the last
location and "draws a line to the new location."  The distance of this
line is the distance that goes into the filter.  This is very sufficient
of hikers traveling at 2 mph.

Now a horse trotting at 7.5 mph.  It goes 1/2 mile in 4 minutes.  Now
assume it goes in a path of 1/2 of a circle with a circumference of one
mile.  You would go from a point to its antipodal point in 4 minutes.
So, depending on the time of sampling, you hand held GPS would say you
went 1/Pi which is about 0.3183 miles when you really went 0.5 miles.
So the error is about 15%.

Because of the way it estimates distance it will almost always measure
short.

So if somenone comes to me claiming their GPS measured a trail short,
I'd give them my mountain bike and tell them to "prove it"<G>.

Truman


>
> In a message dated 12/4/98 2:24:31 PM Pacific Standard Time,
> rides2far@juno.com writes:
>
> << What Truman said about blocked satellite visability by foliage et al
>  is true. But I work with GPS every day, in a very specialized research
>  mode not with hand held units, but I know how the system works. And
>  as a self-proclaimed GPS expert I would not trust a GPS derived mileage
>  measurement for a 50 unless it was on a straight flat road!  A handheld
>  GPS measurement is accurate to about 100 meters.  The up and down
>  accuracy is only about 300 meters. Thats pretty good for telling
>  where you are relative to a topo map, but if you are trying to get
>  accumulated distance you need to put in a "way point" every time your
>  trail bends and every way point will have 100 meters of error.  So if
>  anybody ever tries to tell me a trail distance is wrong based on GPS
>  I'll tell them to prove it with a wheel measurement.  >>
>
>
--
Truman Prevatt
Mystic “The Horse from Hell” Storm
Rocket a.k.a. Mr. Misty
Jordy a.k.a. Bridger (when he is good)
Danson Flame - hey dad I'm well now and ready to go!

Brooksville, FL