Take the 2-minute tour ×
Geographic Information Systems Stack Exchange is a question and answer site for cartographers, geographers and GIS professionals. It's 100% free, no registration required.

I have a Garmin Forerunner 305 and a Google Nexus One Android phone. From the GPS on both these devices I can get a "Accuracy" value that is usually between 6-20 meters. How much can I trust this value? Is it close to correct or may it be very incorrect? I think it must be derived from the NMEA strings received.

I think that both my devices is using standalone-GPS and Assisted GPS from the mobile phone network, if that matters.

share|improve this question
    
"How much can I trust this value?" In what measurement units would you like the answer? –  MLowry Apr 19 '12 at 14:32

6 Answers 6

up vote 16 down vote accepted

The "accuracy" figure displayed by your GPS receiver will most likely be fairly reliable, but it can also be 'way off'.

It isn't "accuracy" that the GPS is displaying - it is the EPE, which is the Estimated Position Error. In other words, it is the probability that the location the GPS is displaying is within the "accuracy" distance from the true location.

Keep in mind that a GPS receiver doesn't actually know its true location. It calculates a location, based on the data received from the satellites (GPS receivers don't receive 'NMEA strings' from the satellites, but many can output NMEA sentences). While most GPS manuafacturers aren't going to tell you how they calculate "accuracy", you can consider it a figure that says "most of the time, the displayed location coordinates are within within X distance of the GPS receiver" (where X is the "accuracy" figure).

share|improve this answer
1  
What is this probability you speak of? No GPS unit I've ever seen ever presents a probability (although I've never read a manual). Is it just some simplified <95% probability type thing? –  naught101 Dec 9 '12 at 5:23

Accuracy with GPS is multidimensional. In other words if you can locate yourself accurately to within 2 meters horizontally there is also a vertical +-.

When you throw in other technology (GPS assisted) I think the formula for determining acccuracy would need a supercomputer to run it. When using phone location first you would require a well designed/ well built municipal network. in other words the phone company has to have plenty of subscribers to pay for a solid infrastructure in which to use for geolocating suscribers (note: with the prescribed technology on the phone).

Then you have to have situations such as no overloading on the network bandwidth, interior location within the network, no overpowering magnetic events. I already mentioned that the technology has to match for the network and the phone so if you are traveling outside your subscribed area you may be on someone elses technology.

Then you have to consider that each carrier would inevitably implement the same type of technology with differing nuances, security, bandwidth, protocol hierarchy, etc.

All this being said it is still possible to measure the accuracy and provide a reasonable std dev formula that in most cases is going to fall within the +-.

Do some research of your own and find some known points on a map, go to them and take a reading (be sure to correlate the stated accuracy estimate to each point). spread about 5-8 points across your network (city coverage area). Then go outside the coverage area and take 2-3 readings.

I would imagine that outside the coverage area the carrieer would rely more heavily on gps, while determining location within the network is done by tower triangulation.

The answer is yes, as long as you take into consideration factors that can have a bearing on the accuracy and then weigh that against what the reading is.

share|improve this answer

The more satellite fixes/locks the better the accuracy - though local conditions apply (urban built up towers, dense woodland and even cloud base cover has negative effects on accuracy).

The more expensive the GPS chip (plus peripherals antenna size & includes battery strength) the better the accuracy.

Remember Land can move: so what was recorded 10 years ago might not be the exact location now - caused by natural events (position x,y,z correct but features seemed have moved).

share|improve this answer
1  
Yes, I know that "the more satellite fixes/locks the better the accuracy". But my question was if I can trust the accuracy value I can get from these devices, and not how to get a better value. –  Jonas Nov 11 '10 at 18:29

New to the GIS industry, but I have spent a long time in the GPS industry. Enable WAAS on your GPS devices for increased accuracy!

share|improve this answer
    
I don't think that WAAS is available in Europe where I live. And I don't think I can enable it on my devices if it isn't enabled already. But maybe EGNOS work, I don't know how I can check that. –  Jonas Nov 17 '10 at 18:00
2  
In europe WAAS is called EGNOS, you should check if you get satellite ID 33. Some countries offer DGPS on the FM-broadband via RDS, however that service costs money. –  Ton Plomp Mar 7 '11 at 21:00

Topic is old. But Another thing you can do it to use aerial photography to "check" how accurate your gps locations are.

some may say the the aerials could be off, and thats true, but it will give you a general idea of how far off the GPS is. Just stand on something that is visible in the photos, and take your points, and then compare them.

Probably the easiest way to do this is with arcgisonline.com since they have the Bing 1 foot aerial imagery for most locations. and you can just drag and drop an excel file with lat and lon into the browser.

share|improve this answer

You can figure out the accuracy for yourself by calculating the RMS (root mean square) error by taking 10-20 readings of x-y coordinates over a known point (such as a survey monument) and applying a formula. THAT will tell you the true accuracy of your device BUT it will only pertain to that specific point in time that the coordiantes were collected. If you did the same test another time you may get a different answer due to cloud cover, tree cover (leaves vs. no leaves, or a tree grew or was cut down), and satellite position / number of satellites visible.

share|improve this answer
    
Excellent suggestion, and the only answer that actually addresses and gives some kind of solid result to the fairly simple question. Check out gis.stackexchange.com/questions/25798/… for some example data. –  naught101 Dec 9 '12 at 6:33

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.