#GPS Error
31 messages in this thread
Hi, Lanny
I found the following in the 4X4s, Trucks and Vans section of the automobile
forum tonight. Item 2 might be a possible explanation for the two mile error
your Scout gave you in the field last summer…
=====================================================================
Alistair,
WRT the Coast Guard undoing the error the DOD introduces into GPS signals:
1. Even without the injected error, GPS accuracy would not be good enough for
blind harbor entries, which the differential GPS system would permit.
2. The differential GPS system also tells the user when a satellite is
broadcasting bad data. It can be up to several hours before a satellite can
be "turned off", thus causing positional errors to unknowing users for that
time
There are some other reasons differential GPS would be used even without the
DOD injected error ("Selective Availability"), but you're right we would be
much better off without this error.
Chip
=====================================================================
Are you planning to return to site of the error any time soon? (Well, after
things thaw out anyway?) I'd be interested to see how the next fix compares
with the first.
Happy Holidays, Guy
Guy,
I am an engineering student, and I just finished a couple of courses in
GPS, so I will try to give you and explanation of your error.
First, since the GPS became officially operational a short time ago,
selective availibility has been in effect. With SA on, the DOD "guarantees"
that your calculated 2D position will be correct within +- 100 metres, about
95% of the time. Your error may be a result of that 5% that does not occur
very often.
Second, your reception of the GPS signal is subject to the effects of
multipath or in laymans terms, refletions. This means that instead of the
signal reaching your reciever directly, it is reflected off of some object and
then to your reciever, thus giving an incorrect range to the satellite. I have
heard of errors up to 20 metres or so due to multipath, but anything is
possible.
Third, the accuracy of your position is very much dependent on the
satellite geometry. You can check how good the satellite geometry is by
looking at the PDOP (positional dilution of precision) value that the reciever
reports. The smaller the value, the better the geometry (should be lower than
3).
Hope that sheds some light on your problem. If you are interested in GPS, see
"Satellite Geodesy", 1993, by Seeber, or GPS World magazine.
Rob
Hi, Rob
Thanks for the info, but it was Lanny that got a two mile error on a field
trip.
Happy Holidays, Guy
Hi Robert,
Thanks for the info. Wish I could believe your figures, but having spent a
field season with a Trimble Scout and occassionally with two other geologists
with a Garmin (don't know the model) and a Magellan Meridian I find that the
system tends to typically produce readings that vary more like 50-200 m. Such
variations can be within an hour or two or several days apart. By this I mean
taking a reading and at the same spot taking a reading an hour later or days
later and having the two readings differ by that much in lat and/or long.
Or, taking readings with my Scout and one of the other units side by side and
having the two readings vary by up to 200 m. We either all have faulty
machines, or the system does not give the accuracy often stated.
The error Guy referred to was a reading in the west desert of Utah I made last
October. Out of several readings taken several miles apart, they generally
plotted on a 15 min topo map within the area I thought I was standing. But this
one reading plotted about 2 miles off. The actual spot was easy to determine
because of the terrain, which was low hills and I was standing on the top at
the edge of a particular drainage. I didn't notice the error until I got home
and checked it out of curiousity.
Did you really mean to say "errors up to 20 metres or so due to multipath"?
From my experience in canyons I would say that error could more than likely be
200-2000 m!
Lanny
Lanny,
Like I said in my message, the precision of your readings is very much
dependent on the satellite geometry. For example, if at a particular time all
of the satellites you are observing are relatively close together in the sky,
your position is very weakly determined. On the other hand, if the satellites
are "well" spaced, your position will be known to a greater degree of
certainty. If you can, you should record the PDOP value reported by the
reciever. This will give you an indication of the "goodness" of the satellite
geometry. I don't know how familiar you are with the GPS, so forgive me if am
telling you something you already know. The satellite geometry is constanly
chaging, so the precision of your readings is also constantly changing.
The differences in readings between recievers does not make alot of sense, but
it is probably a function of the method of computation or quality of the
reciever. The computations involved are rather lengtly, and there are several
corrections that must be made for atomospheric effects, and the methods used by
different recievers is very likely different. Some navigational recievers may
not even account for these sources of error. Also, the satellite ephemeris
being used by one reciever may be a newer or older one than that used by the
other (the ephemeris is the data sent by the satellites that allows the
reciever to calculate the satellites' position, and it is updated periodically)
With regards to making measurements from 15 min quad maps: Keep in mind that
the map standards dictate that positions plotted at that scale only have to be
correct to within about +- 100 feet, and elevations to within +- 1/2 of the
contour interval.
And with regards to errors in canyons: While multipath may be a problem there,
keep in mind that the canyon walls are probably blocking your view of several
satellites, thus reducing your precision.
It is very difficult to explain errors of several thousand feet. The figures
for DOD "guaranteed" precision of +- 100 metres is from their publications.
The 20 metres due to multipath is from the book referenced in the previous
message.
FYI: Using GPS for surveying work using geodetic recievers, we can routinely
get precisions in our measured baselines of +- a few centimeters (repeatable).
This is a somewhat more involved process requiring longer observation times and
post processing of the observations, but it does illustrate the precison
possible with the system.
Let me know if I can try to answer any questions–it really helps me to
reinforce what I have learned!
Happy Holidays!
Rob
Thanks Rob,
Actually, by now I have a pretty good handle on the GPS system (thanks to
people like yourself on various forums on CIS and a few publications). But I am
still learning.
I might change some of my methods of use though. When I finally decided to buy
a unit the thought was to use it–take readings and record them. Now, I am not
as impressed about the accurracy and will probably make note of the PDOP value
(or what similar data the Trimble actually gives me–don't recall at the
moment). The way I have been using it is to just record the lat/lon in the
field and put that in my notes.
From what I've seen of the readings though, they are nowhere near as accurate
as your stated +-100 feet of a topo map. Doing geology work, many readings are
in canyons or on steep mountainsides where at least one direction of the view
to the sky is obscured, but those readings taken in more moderate terrain
haven't been reproduceable to within 100-200 feet either. As an example, last
summer, a friend impressed me with his 3 year old Garmin as we drove to one
site. He read off the lat/lon from his notes and as we drove up the Garmin came
up with exactly that reading (to the thousandth of a minute I believe it was).
He turned it off, but I asked him to turn it back on and I got out of his truck
and took a reading with my Trimble, which gave a reading about 0.2 minute
different from his (which I believe is roughly 1,200 feet lat), he turned his
on and it now read about 0.12 minute greater (closer to my reading). And that
was within just a few minutes of the original reading. This was in a broad
canyon, but there still was enough rock walls and hillsides to mess things up I
suppose.
Lanny
<< Also, the satellite ephemeris being used by one reciever may be a newer or
older one than that used by the other (the ephemeris is the data sent by the
satellites that allows the reciever to calculate the satellites' position, and
it is updated periodically)>>
This is not very likely to be the case since the ephemeris data for the whole
configuration is downloaded from the satellites themselves. Receivers should
therefore be using the same ephemeris at the same time.
Rob
Lanny —
I wouldn't think a 2 mi error would be caused by a high PDOP number as Robert
suggested. If the PDOP was bad enough to cause an error of that magnitude, your
Scout should have said "bad satellite geometry" and not given you a reading at
all.
BTW – I think it was you who was looking for an internet provider. I signed up
with one last Friday and it is great. It is worlds apart from AOL or even
Delphi in ease of use, flexibility and speed.
— Bill
Hi, Bill
PMJI–
> If the PDOP was bad enough to cause an error of that magnitude, your Scout
> should have said "bad satellite geometry" and not given you a reading at
> all.
I agree – that fits with my observations.
> … internet provider. I signed up with one last Friday …
What a coincidence! I just signed up with one last Tuesday. Of course that was
the day CI$ announced full Internet access. Unless CI$ puts "GO INTERNET" in
the Basic Service area, my provider will be _much_ cheaper though, and faster
to boot. I regularly get "CONNECT/38400" from work.
Happy Holidays, Guy
Guy —
>> Unless CI$ puts "GO INTERNET" in the Basic Service area <<
FAT CHANCE!!
>> I just signed up with one last Tuesday. <<
Do you use a SLIP or PPP type connection? If so, Netscape Navigator is the best
program I have found to move around. It's magic!
— Bill
Hi, Bill – hope your Christmas was a good one.
re: "GO INTERNET" in Basic Services…
> FAT CHANCE!!
Hmmm, I'm not surprized. I wonder why not? <g> Lessee here, CI$ is $$+ an
hour @14.4kbps, Teleport is $15 for 30 hours a month @38.4kbps. I don't think
I'll "GO INTERNET" much! <G>
> Do you use a SLIP or PPP type connection? If so, Netscape Navigator is the
> best program I have found to move around.
I initially signed up for simple dial-up, but I haven't paid for it yet (still
in free trial mode). When I pay it will be for a SLIP connection. It's 5 bucks
more a month, but looks like it will be well worth it. That Netscape name came
up in an earlier conversation today – I'll have to look into it. I have the
Trumpet Winsock, and Mosaic. What else will I need? FTP? Gopher?
Happy New Year, Guy
Hi, Guy–
A recent PC Magazine reviewed the various Internet comm packages; Internet in
a Box was one of two editor's choices – I am visiting in CA and cannot look up
the other. I seem to recall Internet in a Box supported the widest variety of
things, including FTP, Gopher, etc., reportedly doing a good all-around job.
–Doug
Bill,
Yeah, the Scout does tell me when there is bad geometry, and depending on just
how bad it is it will sometimes give a reading anyway, this goes in the notes
with the reading.
Yeah, I am still looking for an Internet provider; should have one soon. There
is a direct call up access coming to Coeur d'Alene, supposed to be this month.
I haven't called them yet to find out just what they offer, but the fee isnt'
bad–$30/mo. with sound and graphics for 100 hours ($.50/hr over that). Guess I
better call them right after Christmas and see just what they offer.
Lanny
Lanny
Is there any chance that your topo map is just plain wrong? When the State of
Vermont did it's own Geological mapping in the late 19th and early 20th
Century, the state geologists commented repeatedly on the gross inaccuracies in
the USGS topo maps they were trying to work from.
Of course, Western mapping was done later and better than Eastern mapping. But
I once had major problems trying to find some fossil sites East of SilverPeak,
Nevada in a jumbled area of low hills. It appears that the Western side of the
area was mapped in from the West, and the Eastern Side was mapped from the East
(the Lake Wobegone effect) resulting in some "Townships" that are 8 sided
geometric shapes roughly 6 miles by 7 miles —
implying an error of a mile or so in the mapping.
Don
Don,
That is an interesting thought. The topo is a 15 minute (Dugway Range, Utah),
so maybe I should buy the 7 1/2 minute quad and see how it plots out. The 15
minute topo is based on a 1927 datum, mapped with 1951 air photos and field
checked in 1953. With the air photo base I wouldn't think it has errors that
are too great. But it will be interesting to see what readings the GPS gives me
next time.
The early surveys out west have a lot of errors too, many of them were done in
the late 1800s, don't recall anything as far off as causing "8 sided"
townships, but we do have a few extras like Township 45 1/2 North (usually
shown as a wedge, and many that intersect as much as a mile off, or a row of
sections that are about 1 1/2 miles high. Those are shown on the topos, but
with the air photo base they should at least be fairly accurate at locating
major points in conjunction with the topo lines.
I hope.
Lanny
<<Thanks for the info. Wish I could believe your figures, but having spent a
field season with a Trimble Scout and occassionally with two other geologists
with a Garmin (don't know the model) and a Magellan Meridian I find that the
system tends to typically produce readings that vary more like 50-200 m. Such
variations can be within an hour or two or several days apart. By this I mean
taking a reading and at the same spot taking a reading an hour later or days
later and having the two readings differ by that much in lat and/or long.>>
Lanny, I think what you are seeing here is more an effect of your use of
"budget" receivers, rather than the "design accuracies" built into the control
and space segment elements of the system. A Trimble 4000 series receiver or
equivalent would not show this level of drift.
Not surprised at the level of multipath effects seen in canyons… you'd be
better off with more conventional survey tools in those locations.
Rob
Rob,
Thanks, but you just gave me an answer that I didn't want to receive. Before
buying a Trimble Scout ($800, certainly a few thousand less than a better one,
but definitely not trivial), I asked a lot of people, in this forum, and in two
other forums on CIS, and read messages in the Internet sci.geo.satellite-nav
Newsgroup, as well as a couple individuals in the Bu of Mines who have used
them for a few years and was basically told that these "cheap" units gave the
same "accuracy" as any more expensive unit except for DGPS or expensive
decoders. Now you tell me they don't, which is what I was beginning to believe
from experience.
The canyon problem probably is telling me to just plot the point on a map as I
best can do standing there and use a ruler and the "Over and Up" feature of the
Trimble to tell me what the lat/lon reading is! Wow, what an expensive
calculator!
Lanny
Sorry about spoiling your day, Lanny, but just look t it this way:
Why would I, as a surveyor, spend L 16,000 plus on a GPS receiver if I can get
the same accuracy from a $800 unit?
Rob
Rob,
That wasn't quite what I had in mind. I had gotten the idea that the under
$1,000 units would give one a reading within about 50-100m accuracy, whereas
yours would do it down to the cm or so without DGPS…
Amazing how before I bought one, all the info I could gather said one thing,
now what I find says another. One would think I had asked the government for
info. :-).
Lanny
Lanny,
You probably just got info from either vested interests or GPS
"newbies". Until about 2 years ago the number of people in the whole
world who could speak authoritatively on GPS was very small. Those
working in each specialism had generally known each other for years.
Now the salesmen, and the "I heard it from the salesman" brigade are
out in force, exploiting new markets for this wonderful tool. Of
course it's going to be oversold for a while, until people catch on
that. like anything, you get what you pay for.
If I operate a Trimble 4000 in non-differential mode, my results will
fall inside the 100m ( 95% confidence level) quoted by the system
owners. These results will also be reasonably repeatable, but I won't
get centimetre level accuracies without resorting to some special
techniques. The cheaper receivers have fewer channels, so have to
sequence through satellites. Their tracking loop circuitry is also
less robust than on the more expensive models. And the antennae! For
precision geodetic work I use large ground plane extensions on the
antenna, which, unlike that of the cheaper models, already has a built
in ground plane. In fact, an antenna for the more expensive units can
cost several times the price of a whole "budget" rx!
I'm sorry you feel that you were given bum info before purchase, but
you're in the unfortunate position of entering the market at a
transition stage between the technology being used only by
specialists, and it filtering throught to general use. People in that
position "always" feel that they've had bum info, whatever they've
bought ( I still feel that I was "had" over my first CD-Rom drive!
Rob,
Will thanks. The "bum information" is no surprise, considering I new at the
time that most of it was from the type of sources you describe. Some of it was
from this forum, but much of it was from other sources who probably were new
"experts." If we could just get DoD to let us use the system the way it could
be, then us non-professionals probably could get readings that were a lot more
useful. Some day.
Right now, for many mineral collecting localities, I am not sure the readings I
have recorded are really that useful. For some, anything within a 1,000 feet
would make it possible to find them, but for others it needs to be a 200-300
hundred feet at least. As the decades parade buy these things do improve.
Lanny
As a surveyor, I would always make a site plan of any location that I planned
to return to: even if I was locating myself with GPS. This need only be a
sketch with major features: maybe for you, 3 rough compass bearings to
prominent outcrops would be enough for you to say " my GPS antenna was about
here". Presumably you'd then have more accurate measurements from your GPS
antenna position to the actual work site.
Rob
Rob,
That's a good idea, actually, a method that could have been used at many
localities all along before GPS was priced down where we could almost afford
it.
Lanny
Lanny
I've been out of touch with the satellite business for decades, but I did work
with satellite tracking in the 1960s. I know that enormous advances have been
made, and they need to have been, because in the 1960s we didn't know where
ANYTHING was to 50 meter accuracy. At the couple of meter level, the earth is
not an oblate spheroid, and satellite orbits are not exactly ellipses. While
it is possible under certain circumstances to measure some distances with great
precision that doesn't mean that it's easy, or even possible, to convert those
distance measurements into relative distances on the surface of the Earth with
anything like the precision you can measure the distances to.
Don
Advances sure have been made, Don. Gravitational models are available now as
series up to order 180, with over 32,000 coefficients. You can even model in
time terms, these days, to take care of tidal distortions in the crust. We can
easily do relative positioning to a couple of millimetres plus 1-2 parts per
million of the baseline distance up to continental distances. With the right
GPS receiver and SA switched off or compensated for ( eg, military receivers),
we can do single point positioning withing the define WGS 84 reference frame to
a couple of metres, no problem. The nice thing about GPS, or any satellite
system, is that the ellipsoid is not used in the position computation.
Positions are computed in terms of a gravitationally geocentric cartesian
coordinate system, and only trandsormed to lat/long or grids for those who
still need to work in terms of these coordinates.
I retriangulated part of Northern Brazil in 1986 using GPS. The original trig
had been put in using Hiran and Transit Doppler. We found a 56 m bust in the
original net, so I would agree with your comment about the '60's <g>.
Rob
Don, sorry for any confusion, the "Denise Greenwood" msg is from me. Fatal
log-on cock-up!
Rob
Hi, Rob.
I've been using GPS and DGPS in oil-exploration activities since 1985. I've
seen a few changes since waiting up all night in Brazil to catch a 25 minute
period of good Block I geometry!! It's a while since I spent any time in this
Section, but I see a lot of GPS "misinformation" all over CIS. I've applied, on
behalf of the Hydrographic Society ( 2000 individual and corporate members in
70 countries) to organise a "Spatial Information" forum. This would include
Land and Hydrographic Surveying, as well as LIS/GIS, Geophysics, TMS etc. It
is intended to have a whole section devoted to GPS. Would you ( or anyone else
here) like to see such a beast ( I need all the ammunition I can get, since CIS
Management doesn't seem to see the same need that we do!). Let me know whether
you would visit the forum, if it were to be established.
Rob
Rob,
Fantastic idea!!! I would love to see such a forum established. My
opinion is that GPS and GIS especially will experience a great deal of growth
in the next decade or so. The possible uses of GIS is especially exciting, and
I think there is a lot of room for growth in the technology. Almost all
government agencies are now working on GIS projects, and a public forum would
be extremely helpful in getting and sharing information about individual
projects. If we can get everyone talking, maybe some loose standards can be
agreed upon.
The problem is that most of the public does not know anything about GPS
or GIS and how important it will be in their lives in the future, so I can
understand the hesitation of CompuServe management. For example, does the
layperson realize that detailed information about their real property is
probably being put into a GIS by the local county assessor? Not to say that
this information is not already available, but never before has a person had
the ability to access this information with a computer. Who should have access
to this information? What consequences are there to having this information so
easily available? How accurate is the information? For what purposes should
this information be used?
These are real questions that are only now beginning to be asked as
people begin to see the potential uses and abuses of such a massive quantity of
personal information available so easily.
There should be a forum where people can discuss technical issues as
well as societal issues.
Rob
Guy,
That's a new one. I guess that implies a satellite can just go bonkers and send
bad data until they shut it down and fix it (if possible). That could explain a
2 mile error.
I hope to get back down there next spring. I'll take some readings when down
there, hope to spend at least a day on that spot of ground.
The locality is really interesting, if one is interested in unusual mineralogy
and micros-durangite, hematite, topaz, cassiterite and reportedly another 55 or
so minerals. Wish I was there now instead of here (although camping in the
Thomas Range in the winter can often be a challenge!). Try spending News Years
Eve out there!
Lanny
Hi, Lanny
> I guess that implies a satellite can just go bonkers and send bad data until
> they shut it down and fix it …
My guess is that a stray cosmic ray or two could flip a few bits at random and
cause, as IBM would say, "unpredictable results". I suppose they have a way to
order the satellite to re-boot and then upload new data to it.
> … (although camping in the Thomas Range in the winter can often be a
> challenge!). Try spending News Years Eve out there!
Uuh – no thanks. Back when I was in high school, a group of friends and I
tried a winter camping trip in the high Cascades (near Clear Lake just South
of the Santiam Pass road). Had a dickens of a time getting the tent pegs to
hold in the snow. <G> Besides, It looks like I may be spending part of New
Years Eve at work this year. 🙁
Speaking of work, I've been asked to put together my vacation schedule for
'95. I'd like to take a few days off either before or after the MM meeting in
May and go mineral exploring/collecting. Any thoughts?
Regards, Guy
Guy,
Apparently you are like I feel about winter camping now. My son is a Boy Scout
and each winter they have a thing called the Klondike Derby (aka Freezeriee!)
and he just loves it. It is Friday night through Sunday in Mid January, they
race sleds and do a bunch of stuff to show their camping skills. Even the
winter when the lows were near zero he came back with nothing but good stories!
(I stayed home.)
Next May, too early to tell. The chances are good though. The best thing is for
you to just set your vacation either side of the meeting and I will try to be
able to take some time to spend a couple days over there too.
Tucson is only a month and a half away. I am giving a talk at the Topaz
Symposium on Saturday so have too go this year (Yay!). Unfortunately due to the
cost I don't go every year, but it will be good to get down there and get that
break from winter.
Lanny