What affects GPSr Accuracy?

Discussion about software such as GSAK, OziExplorer etc, as well as all things hardware, GPSrs, laptops, PDAs, paperless caching, cables etc
User avatar
Grank
Posts: 483
Joined: 15 January 05 1:26 pm
Location: ....

What affects GPSr Accuracy?

Post by Grank » 18 December 05 1:32 pm

When looking for a cache and finding that the coordinates vary from where I have been looking, I have put this down to "different day, different satellite configuration". But I am now wondering if this is the case. <br><br>

I recently went to an event and teamed up with two others to look for a local cache in the area. I had a Magellan Explorist 100, they each had a Garmin - an Etrex Legend and a 76CS. As I had the coordinates in my GPSr they copied them into their GPSr manually. When we arrived close to the GZ each wanted to go in a separate direction. And guess who was closest - the Garmin 76CS, then the Etrex Legend and finally me. And it was a mater of 20 meters - not close at all. The cache was placed by a Garmin owner (not sure what model). <br><br>

We talked about this and they revealed that even they have this problem when searching together. Sometimes due to the inevitable incorrect entry of coordinates, but quite regularly without this error. <br><br>

Today I have found a cache that had me in the same predicament. Placed with a Garmin, FTFed with a Garmin, but I was looking about 20 meters away again. <br><br>

I had thought (perhaps foolishly) that point X would be in the same place for all GPSrs on the same day at the same time. Maybe some variance due to local factors like the exact position of where the person was standing at the time - e.g. facing a direction that blocked a satellite that another personÂ’s stance wasn't blocking. <br><br>

It would actually seem that not only does the "exact location" vary from maker to maker but also from model to model ... or does it actually vary from GPSr to GPSr? Wasn't it Mark Twain that commented something like "When I had one watch I knew exactly what time it was. Now that I have two I am never quite sure." <br><br>

I have heard that lightening and overhead electrical wires do not affect the GPSr, but to be wary of mobile towers. I had a brief discussion at an event with the Loon about signal reflectance form rocks, which was news to me. <br><br>
So just what factors do affect the accuracy of your GPS?

User avatar
CraigRat
850 or more found!!!
850 or more found!!!
Posts: 7015
Joined: 23 August 04 3:17 pm
Twitter: CraigRat
Facebook: http://facebook.com/CraigRat
Location: Launceston, TAS
Contact:

Post by CraigRat » 18 December 05 2:04 pm

There's hundreds of things that can have an effect.

If you left the GPSr's in the same spot for a LONG time, they'd probably get a similar fix, but all GPSr's have different aerials, algorithms, ways of calculating how far you've moved in any instance... and you will get varying results.

Some people insist that overcast weather will have an effect too, but most manufacturers say that this is not so (so do I). Wet foilage on the other hand is probably the WORST thing I've ever seen to change accuracy.... it sets up lots of reflections that can cause havoc on your unit....

I've never seen 2 GPSr's give the same reading in the same spot myself.

Damo.
Posts: 2183
Joined: 04 April 04 5:01 pm
Location: Jannali

Post by Damo. » 18 December 05 4:25 pm

It's astounding that they work at all really when you consider the work the GPSr actually has to do to figure out where it is. The Satellites don't really tell them much at all.
I have read on the forum about an event they had in SA where all the attendees marked the same coordinates in a field and they were all over the place. They did observe a 'grouping' effect from the different brands though.

I was trying to get coordinates for a Water Reserviour with a mobile tower on top once and at one stage the GPSr though it was 15km away.

Kerry
Posts: 224
Joined: 22 April 03 6:45 pm
Location: Australia

Post by Kerry » 18 December 05 5:32 pm

Lots of things can affect accuracy and even though the basic principles used are based on the same specification different manufacturers use different propriety algorithms.

Even 2 identical models will most likely not give similar results, well not the type of results one might expect.

System SPS (SIS) accuracy is based over a 24 hour period so at any given time 20 metres is quite possible but then unless you are sure that the coordinates you have are absolute (correct) then 20 metres really doesn't indicate accuracy. The system accuracy specs does not take into account how the user influences the position solution.

Yes the way a person faces can affect results and with a handheld always try and face the equator, there's no sats over either of the poles.

Cheers, Kerry.

User avatar
Grank
Posts: 483
Joined: 15 January 05 1:26 pm
Location: ....

Post by Grank » 19 December 05 10:14 am

So in summary:-<br><br>
Environment<br>
- Wet foliage<br>
- Mobile phone towers<br>
- Direction faced (i.e. face the equator for best results) <br><br>
Technical<br>
- Algorithm used by manufacturer<br>
- Aerial type on GPSr

Kerry
Posts: 224
Joined: 22 April 03 6:45 pm
Location: Australia

Post by Kerry » 19 December 05 3:07 pm

Grank,

Also

- Satellite geometry (much more important than satellite numbers)
- Signal-In-Space (SIS) errors. Ionosphere, Tropsphere etc
- Satellite errors. Clocks, Ephemeris etc
- Obstructions (including the person)
- Multipath. Extremely hard to detect and a complete unknown at the time.
- Receiver. Additional issues than simply alogrith and antenna. Receiver noise etc
- Processor speeds (not a big drama these days)

Much of the SPS errors can be attribued mainly to clocks, ephemeris and ionosphere.

Also been in a period where there are a mix of old and new generation satellites. The hardware of the newer satellites contribute to improving things considerably but does depend on the constellation and any given point in time.

Cheers, Kerry.

User avatar
EcoTeam
200 or more found
200 or more found
Posts: 1267
Joined: 03 April 03 7:57 pm
Twitter: EEVblog
Location: Crestwood, NSW
Contact:

Post by EcoTeam » 19 December 05 6:03 pm

In practice, on caches with good signal strength you should be getting within 5m *after* some averaging. Anything more than this is generally considered "a bit out". >10m is not very common and would be viewed as "the co-ords are out".

As someone said, satellite geometry has a very big effect. If you get three birds in a line you won't get very good accuracy at all. So you want a good spread of birds around you.
Generally speaking, signals from satellites on the horizon will introduce greater error into the final calculation, so you can often gain a bit more accuracy by deliberately blocking it out (with your body or a tree etc).

The quoted RMS accuracy of a typical GPSr is 15m, but averaging and other tricks gets us much better than this.

EcoDave :)

User avatar
EcoTeam
200 or more found
200 or more found
Posts: 1267
Joined: 03 April 03 7:57 pm
Twitter: EEVblog
Location: Crestwood, NSW
Contact:

Post by EcoTeam » 19 December 05 6:06 pm

Grank wrote:So in summary:-<br><br>
Environment<br>
- Wet foliage<br>
- Mobile phone towers
Mobile phone towers don't matter.
The US military poured billions of dollars into the development of the system to ensure that this is so!

Not just wet foliage, but any foliage or any other obstruction.

EocDave :)

User avatar
Chwiliwr
10000 or more caches found
10000 or more caches found
Posts: 900
Joined: 10 April 05 10:39 pm
Location: Leeming Western Australia

Post by Chwiliwr » 19 December 05 6:29 pm

Phone towers are a great source of multi path signals and shouldn't be ignored completely no matter what the authorities tell you.

In handheld units the single most common source of inaccurate readings is the timing device of the unit.

You also have to remember that commercial handheld GPS don't have to be that accurate. I personally think having a sub metre accurate unit would take a lot of the fun out of geocaching.

User avatar
caughtatwork
Posts: 17015
Joined: 17 May 04 12:11 pm
Location: Melbourne
Contact:

Post by caughtatwork » 19 December 05 7:04 pm

Chwiliwr wrote:You also have to remember that commercial handheld GPS don't have to be that accurate. I personally think having a sub metre accurate unit would take a lot of the fun out of geocaching.
Not if I hide one and take a crappy reading when I hid it :-) Your sub-meter accurate unit could still be more than 10m out. Not that I would do this, of course.

Kerry
Posts: 224
Joined: 22 April 03 6:45 pm
Location: Australia

Post by Kerry » 19 December 05 7:09 pm

GPS signals are extremely robust and the IIR-M has a little bit more power as will the next generation, it all helps. Obstructions are quite different to interference, one is fairly obvious and the other can be quite misleading.

As for commercial handheld GPs units the accuracy specification is no different from any SPS unit. The system specifies a guaranteed accuracy specification and it's the manufacturer that might let the accuracy down but these days that's not a good commercial decision to deliberately reduce accuracy capability.

SPS system accuracy is 13 metres @ 95% SIS but remember that is with a specific satellite constellation and for quite some time now users have been a little spoilt with an over populated constellation. The thing is to not expect the general accuracy as experienced at present.

The low horizon satellites are always an interesting topic as which manufacturer allows a user configured mask angle. yes all recievers should have a user configurable elevation mask but there's this silly thinking that the more satellites in view the better and this is simply not the case but a 12 channel receiver will probbaly sell more over a 8 channel model but in affect the 8 channel can have better accuracy than a 12 but sales people and PR people are only there to sell units not to sell the facts.

Now averaging this a completely misunderstood subject and in most cases averaging especially with SA set to zero can actually degrade the position solution, not improve it. GPS positions are in no way random, continuous yes, but certainly not random hence averaging does not apply as the theorists would expect. Averaging? gives you a 50/50 change, 50% of the time it could be better and 50% of the time it could be worse, which 50% one is working in at time is totally an unknown at the time.

Cheers, Kerry.

c.j.b
Posts: 39
Joined: 04 January 05 8:30 pm
Location: Newcastle
Contact:

Post by c.j.b » 19 December 05 10:35 pm

Also:
- waypoint downloading software incorrectly rounding the least significant digit and losing accuracy when converting from wholly decimal coordinates (nn.nnnnn) into decimal-minutes format (nn nn.nnn) and probably back again into whole-decimal.

A lot of the time I'm finding the LSB of the longitude is plus-one on what it should be according to a printout of the cache's GC page.

User avatar
Chwiliwr
10000 or more caches found
10000 or more caches found
Posts: 900
Joined: 10 April 05 10:39 pm
Location: Leeming Western Australia

Post by Chwiliwr » 19 December 05 11:18 pm

c.j.b wrote:Also:
- waypoint downloading software incorrectly rounding the least significant digit and losing accuracy when converting from wholly decimal coordinates (nn.nnnnn) into decimal-minutes format (nn nn.nnn) and probably back again into whole-decimal.

A lot of the time I'm finding the LSB of the longitude is plus-one on what it should be according to a printout of the cache's GC page.
I have noticed that too with the software I am using. It sometimes gives the wrong figure on converting between the two.

I have also noticed that the UTM conversion is not very accurate for Western Australia both on the GPS units and on the GC site. Whatever algorithm they are using is not designed for WA UTM conversion. It may be the same in the other parts of Australia but I don't have known coordinate values of points to test with other than WA.

The algorithm is OK for relative positioning between two points using the GPS as it produces the same error for each if they are not too far apart but if you use it for finding a particular point on a map you can be tens of metres out, especially on the 1:25000 or better scale maps. It also doesn't help that the most popular street directory in WA (StreetSmart) uses the old datum (AGD84) on its directory UTM grid and not GDA94. It does have the GDA94 UTM numbers there as well but you have calculate them for points not on the edge of the page.

Kerry
Posts: 224
Joined: 22 April 03 6:45 pm
Location: Australia

Post by Kerry » 20 December 05 9:16 am

UTM on most handhelds is a little misleading as just because the coordinates display to the 1 metre does not mean the coordinates are good to 1 metre. I would have thought any rounding issues are a lot less critical than the real influences on accuracy. Relative to when GDA94 was fixed Australia has moved almost a metre so everything needs to be kept in perspective.

Also PRN17 should now be showing up, which became fully operational a few days ago. IIR-14(M) the first of the next generation civil dual frequency sats with increased power apart from other tthings. Might be only the first but every modernized sat improves things even for the current generation. Won't be too far away now that the first civil dual freq receivers start to appear. Civil dual freq accuracy in time will get to around the 95% 1.5 metre mark.

Cheers, kerry.

User avatar
caughtatwork
Posts: 17015
Joined: 17 May 04 12:11 pm
Location: Melbourne
Contact:

Post by caughtatwork » 20 December 05 3:32 pm

This is an interesting discussion and one that will assist others (including me) in understanding more.

This information has been distilled (possibly inaccurately) into the following wiki page.
http://wiki.geocaching.com.au/wiki/Accuracy

I encourage you to review it and update it as you see fit.

Post Reply