GPS Tip: Increasing accuracy for a single location
Sam Wormley
Tuesday - February 10, 2006
GPS Tip: Increasing accuracy for a single location
http://www.mobilecrossing.com/support/tips.html
The easiest way to increase the accuracy of a single location is to average the values of several GPS entries. Some GPS
units can do this with an averaging feature built into the product. PDA’s can use special programs that are designed to
do this sort of averaging. However, a short term average done in this way isn’t quite as good as taking a number of
separate readings at different times of the day and then averaging manually.
Another interesting way to average is to make sure the GPS (or a program with a tracklog) is not pinned or locked to a
road and then just let it sit for a while with the tracklog enabled. The resultant tracklog can be viewed graphically
and the center position can be determined fairly easily.
_____________________________________________________
To find a more thorough treatment of averaging look at:
http://users.erols.com/dlwilson/gps.htm
http://users.erols.com/dlwilson/gpsavg.htm
http://edu-observatory.org/gps/gps_accuracy.html
Lisa Murphy
about .15 Hz offset and goes to flicker (toward 0 frequency). Also the
variance of the data gets worse with the # averaged. Data from Globalsat
BT-338.
Joe Murphy
G
"Sam Wormley" <swor...@mchsi.com> wrote in message
news:0%qGf.753996$x96.511103@attbi_s72...
Hans-Georg Michna
nudge it towards the current position every time by a fraction
of the current distance delta, say, by 20%.
Hans-Georg
--
No mail, please.
Heinrich Pfeifer
> GPS TIP OF DAY
>
> average the values of several GPS entries.
you can mark a fix each time you meet a location, and average the
coordinates taken. But be careful: when adding a new fix to an already
existing average, the new fix must be inserted with a smaller weight than
the old average. Example:
you have a waypoint which was created by averaging 5 fixes. Now you want to
improve the accuracy by adding the 6th fix. The old coordinates must get 5
times the weight of the new coordinates, in order to have each of the 6
fixes equally weighted.
Too complicated? No. Gartrip - www.gartrip.de - remembers the correct weight
of averaged locations, and it does all this automatically when using the
menu function "new WP from average".
> Another interesting way to average is to make sure the GPS (or a program
> with a tracklog) is not pinned or locked to a road and then just let it
> sit for a while with the tracklog enabled. The resultant tracklog can be
> viewed graphically and the center position can be determined fairly
> easily.
once again, be careful. Do not leave the GPS in "auto" mode - in this mode
it may decide that it should not record track points while not moving. It is
the best to set a fixed time interval. Five minutes is a good choice.
And at the end, Gartrip does the avaraging by the menu function "new
waypoint from average of track range".
--
Heinrich
www.gartrip.de
new<at>gartrip.de
Ted
"Hans-Georg Michna" <hans-georgN...@michna.com> wrote in message
news:4j4mu1tmk114m9o9q...@4ax.com...
> Yet another method is, if you visit a waypoint several times,
> nudge it towards the current position every time by a fraction
> of the current distance delta, say, by 20%.
>
> Hans-Georg
You can also perform the averaging function visually by making many
waypoints from the same spot over days or weeks and then downloading all the
waypoints to software such as Garmin's Mapsource. The waypoints will show
up as a cluster on the map when zoomed in all the way. Take the mouse
pointer and make a new waypoint at what looks like the center of the
cluster. This technique also has the advantage of allowing your eye to
exclude any "rogue waypoints" that are far away from the main cluster.
Chris Malcolm
There is a 6 hour semi-periodic cycle in GPS location errors due to
the satellite orbital periods which ideally you want to sample
beyond. Easily seen if you graph the tracklog of a GPS stationary for
more than six hours.
--
Chris Malcolm c...@infirmatics.ed.ac.uk +44 (0)131 651 3445 DoD #205
IPAB, Informatics, JCMB, King's Buildings, Edinburgh, EH9 3JZ, UK
[http://www.dai.ed.ac.uk/homes/cam/]
John Tyson
> There is a 6 hour semi-periodic cycle in GPS location errors due to
> the satellite orbital periods which ideally you want to sample
> beyond.
When you say "6 hour semi-periodic" what would be a "full period"? My
understanding from earlier threads is that the satellite orbits are 12
hours, but with the earth's rotation you will be on the opposite side after
one orbit and will only see the same satellite configuration every 24 hours.
John
Sam Wormley
For a given location on the earth, the constellation pattern
repeats every 23h 56m.
The diurnal pattern of night and day, the sun's effect on the
earth's ionosphere will have a far greater effect on positional
accuracy.
H.W. Stockman
"Sam Wormley" <swor...@mchsi.com> wrote in message
news:0%qGf.753996$x96.511103@attbi_s72...
> GPS TIP OF DAY
>
> Tuesday - February 10, 2006
> GPS Tip: Increasing accuracy for a single location
> http://www.mobilecrossing.com/support/tips.html
>
> The easiest way to increase the accuracy of a single location is to
average the values of several GPS entries. Some GPS
> units can do this with an averaging feature built into the product. PDA’s
can use special programs that are designed to
I've found the following procedure gives incredible z accuracy. When I come
to a benchmark of known elevation, I watch as the GPS randomly varies from
higher to lower elevations. Then, just as it crosses to the benchmark
elevation, I take a waypoint. Sure enough, when I come home and download my
waypoint, I find that it closely matches the surveyed elevation. What a good
boy am I!
Sound stupid? You might be surprised how many well-meaning people use this
procedure in one form or another.
Sam Wormley
Actually I don't think you can rely on that technique, as there
is little correlation.
Rotareneg
> H.W. Stockman wrote:
> > I've found the following procedure gives incredible z accuracy. When I come
> > to a benchmark of known elevation, I watch as the GPS randomly varies from
> > higher to lower elevations. Then, just as it crosses to the benchmark
> > elevation, I take a waypoint. Sure enough, when I come home and download my
> > waypoint, I find that it closely matches the surveyed elevation. What a good
> > boy am I!
> >
> > Sound stupid? You might be surprised how many well-meaning people use this
> > procedure in one form or another.
> >
> >
>
> Actually I don't think you can rely on that technique, as there
> is little correlation.
What do you mean? I've tried it and it's accurate to within the
precision of my GPS, plus it also works horizontally as well as
vertically if you're willing to wait long enough for all three axes to
match the surveyed location. :D
Hans-Georg Michna
>"Hans-Georg Michna" <hans-georgN...@michna.com> wrote in message
>news:4j4mu1tmk114m9o9q...@4ax.com...
>> Yet another method is, if you visit a waypoint several times,
>> nudge it towards the current position every time by a fraction
>> of the current distance delta, say, by 20%.
>You can also perform the averaging function visually by making many
>waypoints from the same spot over days or weeks and then downloading all the
>waypoints to software such as Garmin's Mapsource. The waypoints will show
>up as a cluster on the map when zoomed in all the way. Take the mouse
>pointer and make a new waypoint at what looks like the center of the
>cluster. This technique also has the advantage of allowing your eye to
>exclude any "rogue waypoints" that are far away from the main cluster.
Ted,
yes, good method. Not quite so simple and can't be done
casually. Also requires storing a number of waypoints, but
ultimately it is the more precise method.
ted
I just notice that Sam suggested the same idea a couple of days earlier than
me using the track log instead of individual waypoints.
Hans-Georg Michna
>I just notice that Sam suggested the same idea a couple of days earlier than
>me using the track log instead of individual waypoints.
Ted,
the track log won't cut it quite so nicely, because the whole
point is to take measurements at widely different times, not
just within a few hours.
Dan Anderson
> On Fri, 10 Feb 2006 23:11:26 GMT, ted wrote:
>
>>I just notice that Sam suggested the same idea a couple of days earlier than
>>me using the track log instead of individual waypoints.
>
> point is to take measurements at widely different times, not
> just within a few hours.
>
The use of the track log doesn't preclude the possibility of
taking measurements at widely different times. The receiver
could be set to record a track point every so many hours
and/or the track log can be transferred to another storage
medium, cleared from the receiver and restarted.
I was doing some experiments where I would fill the track log
at one sample per second, sometimes save it to a laptop, and
repeat; then return on another day and do it again, etc.
--
Dan
(email change 2001 to 2004)
Dale DePriest
Dan Anderson wrote:
As the original person that suggested the tracklog (the tip was mine,
posted by Sam) I agree. A tracklog is a very easy method to collect
data. One way is to turn the unit on for a few minutes, just to get a
clear fix, and then turn it off again. On a Garmin this will create a
disconnected log but a log none the less. You can visually look at the
log or save it. Saving it will connect all the pieces together which may
make the visualization easier. One time, on a cruise, I just took
samples every now and then and when I saved the route I have the entire
cruise.
Dale
--
_ _ Dale DePriest
/`) _ // http://users.cwnet.com/dalede
o/_/ (_(_X_(` For GPS and GPS/PDAs
Heinrich Pfeifer
"Dale DePriest" <Da...@gpsinformation.het> schrieb im Newsbeitrag
news:11v9nrr...@corp.supernews.com...
> As the original person that suggested the tracklog (the tip was mine,
> posted by Sam) I agree. A tracklog is a very easy method to collect data.
> One way is to turn the unit on for a few minutes, just to get a clear fix,
> and then turn it off again. On a Garmin this will create a disconnected
> log but a log none the less.
you should note that Garmins use to record some very inaccurate track points
after being switched on, mostly in 2D mode. This means you will have to
delete those start points from each track segment before averaging.
For this reason, I recommend to keep the receiver on all the time, and set
the record interval to five minutes.
--
Heinrich
http://www.gartrip.de
mail: new<at>gartrip.de
Dale DePriest
Heinrich Pfeifer wrote:
Hard to do on a cruise ship since I can't stay beside the unit for long
periods of time.