# Need help in understanding latitude, longitude, and GPS April 23, 2013 1:48 PM   Subscribe

I'm working on a new piece of music and need to understand a little better how longitude, latitude, and GPS works. Specifically with regard to the resolution and what can be reasonably expected from current technology.

I do not have a GPS capable device. If I did then I might not need as much help with this. So here goes. From what I've been able to figure out the resolution for GPS is like 3 meters, but with certain augmentation techniques this can be tightened up a lot, down to centimeters even. How common is this type of augmentation to average people? What I mean is that if the average person were to go outside and use their smart phone to figure out their geographical coordinates would most people be stuck at 3 meters or might it be 1 meter, less, more? Related, if two people were to stand within 1 meter of each other how likely would it be that they would generate the exact same coordinates?

Also, what about elevation? Is that information normally available? Or let's say someone get's the basic information about the spot they've chosen, would they then be able to go online and figure out the elevation or maybe even narrow down their specific coordinates?

And then how standardized is all this information? I would like to think that every device would generate the exact same kind of information using the same standards but are there idiosyncrasies that might require conversions in order to get all the data to "look the same"? Is it possible to automate that?
posted by bfootdav to Technology (13 answers total) 1 user marked this as a favorite

Not sure about accuracy values, but here are some basic concepts to start you off, beginning with enhancements to the basic GPS signal.

- The switchoff of "selective availability" in 2000, in which before that, intentional degradation of the signal was done due to national security concerns. This improved consumer accuracy from 150-200 feet to about 20-30 feet. This was big stuff back then!

- Differential GPS (DGPS), where a receiver at a known, fixed survey marker (with exact lat/long) receives the GPS signal. It takes note of the drift due to atmospheric conditions. Meanwhile you carry a portable unit around and do your surveys. After the fact, it is possible to reconcile the time/error logs at the fixed location to correct your own survey. It is possible to get survey-grade accuracy this way. When you see surveyors out and there's an antenna with a tripod, that's probably what it is.

- NOAA broadcasts differential GPS signal corrections on longwave radio (at least I think they still do) at a number of regions around the US, which means you can get instant correction on your GPS unit. It's only a few hundred dollars, and is used widely by marine recreation & small businesses. I have an old Garmin III + backpack receiver that does this type of DGPS and it brings the error down to within 10-15 ft, though you really have to be within 50-100 miles of a station for best results.

- WAAS, which is a more reliable, consistent way of providing DGPS centrally-computed corrections. They are transmitted with the GPS signal down to your unit, which uses this to improve on the data. Since it doesn't involve a lot of extra gear and is fairly foolproof, this is the most economical way to improve on the signal.

- There's also cellular phone triangulation, which a lot of smart devices use, but I don't know a whole lot about that.

My knowledge is about 5 years old so I may be a bit off, I'm sure others will correct me if I erred somewhere.
posted by crapmatic at 2:02 PM on April 23, 2013

Response by poster: Thanks for the information, but what I'm actually looking for is what the average person will already have and be able to do. I see I did not make this clear up front, but I'm going to be recruiting random people to help with the project and need to know what can be reasonably expected from the technology that most people doing this would have, i.e., what kind of resolution does an iPhone or Android phone have by default and what easy and free improvements are available for people to implement.
posted by bfootdav at 2:07 PM on April 23, 2013

There is unaugmented GPS, which allegedly has about 10 meter accuracy at 95% of samples, but local terrain, reflections off buildings, trees, etc, can seriously degrade even that.

There are space based augmentation systems. These work with a network of ground stations at known locations, and an additional broadcast that sends error correction from those ground stations from satellites. Most modern receivers I've seen implement one or more SBAS techniques (WAAS in the US), and this can pull you down to 95% of samples within 2 or 3 meters.

There are ground-based augmentation systems. These require expensive carrier phase capable receivers (my gut says starting at about a thousand bucks and going way way up), and fall off with distance from a known reference receiver. There are mechanisms for doing this in real-time (either with your own reference receiver at a known location, or with a subscription service from someone who has a network of reference receivers, these networks seem to be most comprehensive in large farming areas), or as a post-process (NOAA has a bunch of reference receivers you can get data from later and re-apply to your GPS data). This can be down to... I think a few mm if you're relatively close to the reference receiver.

I believe that much of the error comes from the same high atmosphere sorts of signal propagation delays, so it is likely that two people standing next to each other with identical receivers would show up as in different places, and the error would apply equally to both of them. But I wouldn't build any life-critical systems based on that assumption without lots of testing.

Because of basic geometry (satellites are usually not directly overhead, they're out by the horizon) GPS altitude measurement is about 10x less precise than horizontal measurement.

I believe that NMEA output is available from pretty much any device, and I think (I've only worked with cheaper consumer-ish devices for the most part) that NMEA also has messages for specifying carrier phase. But the basic NMEA message gives you time, lat and lon (usually in WGS84 datum, which you can assume unless you want to go slowly insane), and elevation, usually all WAAS or other SBAS augmentation corrected.
posted by straw at 2:08 PM on April 23, 2013 [1 favorite]

Oh, additional source of error in urban, forested or canyon areas is multi-path signals. In this case you could get two people standing close to each other showing up well apart from each other, or positioned completely relatively different (ie: one person is in the shadow of a building, and getting a reflection off the other side of the street, the other person has a straight shot to one ore more satellites, so the first person has a signal from one or more satellites that travels twice the width of the street further).

So that "95% of samples within ..." number assumes a pretty good view of the sky.
posted by straw at 2:13 PM on April 23, 2013

Ah, the neurons are coming back: The GPS L1 carrier wavelength is about 19 cm, the carrier phase receivers can get about 1% of of the wavelength, so even with the expensive gear you're not going to get better than ¾" right next to your reference station, and if you don't have enough "dwell time" for the receiver to figure out where the carrier phase is relative to the actual signal (or start from a very known location), your readings can be off by increments of 19cm.

So going back and trying to answer your original question: My gut says that modern smart phone outside is probably, yes, about 3m. Two smart phones with identical receiver hardware 1m apart will probably show up as different places, 3m is more likely to show up. However, I was just looking through a bunch of geocoded Tweets this morning, and an amazing number of those were coming from exactly the same place, so don't bet on this differentiation. Though those could have been located using something other than GPS.
posted by straw at 2:27 PM on April 23, 2013

Best answer: However, I was just looking through a bunch of geocoded Tweets this morning, and an amazing number of those were coming from exactly the same place, so don't bet on this differentiation. Though those could have been located using something other than GPS.

When you turn on location services on a smartphone, it knows which cell towers are visible to it, and where those are, which gives it a rough sense of where you might be. Depending on the density of coverage that could give you something like a couple hundred foot precision. Then it starts doing GPS stuff, and that can take some time to get a fix. Once you've got a fix on enough satellites, then you'll the location to 3m. If you are seeing a lot of geotagged tweets from the same locations, that's probably because the twitter client didn't wait for an accurate GPS fix - it used that initial approximation, which gives the same location for everyone who can see those same cell towers.

As for elevation: GPS does tell you your elevation, but you need more satellites for that (4 instead of 3), and due to the geometry, it won't be as accurate as the lat/lon.

What I mean is that if the average person were to go outside and use their smart phone to figure out their geographical coordinates would most people be stuck at 3 meters or might it be 1 meter, less, more?

GPS is actually incredibly precise. It can report very small changes in position. But it's not nearly as accurate as it is precise. A good way to think of it is to imagine that the world is blanketed by an ever-shifting displacement field. So my GPS will say I'm 10 feet northeast of where I am. A second later, it will tell me again I'm 9.93 feet northeast of where I am. Over the course of a few minutes that might change to say I'm 7 feet south-south-east of my real location. But if I move 3 feet, it will report a new location 3-ish feet from where it did before. But it won't say I moved exactly 3 feet, because in that new location, and at that new time, the error will be a little different than it was.

So if my reported position changes by 5 feet, the device doesn't know whether I've actually moved at all, or if I'm stationary and the error has changed.

IIRC, there's no way to know for sure how accurate the information is, but devices can do a pretty good job of predicting the margin of error based on how many satellites are visible and how good the signal is.

Related, if two people were to stand within 1 meter of each other how likely would it be that they would generate the exact same coordinates?

I think this is extremely unlikely. First of all, the same receiver in the same location won't even give you the exact same coordinates from second to second. But I think that the error ("displacement field" above) is inherent to the system, and two nearby receivers will both be offset by roughly the same amount. (I am not completely sure of this). I am sure that when I was making a GPS device (for catching "ghosts") it was able to notice when I walked a few feet towards a waypoint, though again, it couldn't really distinguish that from the inaccuracy just wobbling around.
posted by aubilenon at 4:18 PM on April 23, 2013 [1 favorite]

As other's have pointed out there is no exact answer in regards to accuracy. depends a lot on where you are and what the visibility of the sky is like.

Also different devices have different antennas, which are more or less sensitive.

In regards to the "extra" services that allow you to get more accuracy, they are very localised - i.e. there is no WAAS outside North America and Hawaii.

In regards to elevation, yes a GPS receiver will output elevation. In my experience, on the same device, it's fairly consistent for time to time (e.g. visiting the same spot on different occasions) - but not within a metre - more like 5~10m. I haven't tried multiple devices at the same location.

You can use lat and long to get an elevation based on various models. Some services I use do this (websites that analyses jogging sessions). Typically I find that the GPS recorded elevation "seems" more accurate, but I don't have any surveyors data to compare.
posted by trialex at 6:41 PM on April 23, 2013

GPS vertical error is about 2x the horizontal error.

There are web services that given a latitude and longitude will return an elevation value, but the source elevation data is rather coarse and could be less accurate than the elevation calculated by the GPS. The elevation is also the elevation of the antenna, not the ground.

Two adjacent GPS receivers should calculate pretty similar positions if they are using the same satellite for their calculations. Difference in antenna designs and positions of the units might cause them to use different satellites and the difference could be larger.
posted by drwelby at 9:05 PM on April 23, 2013

A way to look at how GPS receivers see the world is to consider heat shimmer you see on roads on hot days, or from the heat pouring out of a window on a cold day. That same effect affects how the signals from the satellites propagate to the receivers. So any sample it takes might be a little off from the previous ones.

A "smart" GPS device probably does some averaging to give you a smoother looking location dot.

Accuracy versus precision is a hard concept to understand. GPS has the math to get to 1 meter of precision, I believe, but because each measurement depends on the speed of light through the ever-changing atmosphere, it's not necessarily going to be the same 1 meter spot every time. Precision is how finely something can be measured. (A ruler with mm markings is going to be more precise than one with just cm markings.) Accuracy is how close to correct an individual measurement actually is.

If you were to get a giant balance scale and try to measure the weight of a car in nickels, you can get very precise answers, since a nickel weighs something like 5 grams. Versus the car that weighs 1000-ish kilograms. But ten people measuring that car the exact same way are almost assuredly going to get different answers every time. While precise, it is not accurate.

The opposite would be trying to weigh the car in units of boulders. Just about every time, you are going to get the same answer- the car probably weighs between 5 and 6 boulders. But the precision (or resolution) of a boulder isn't all that useful.

So anyway, if you are trying to do something like match up the beat of different instruments by GPS, you are going to have to keep the receivers stationary and take a lot of samples to get a correct-enough offset. If it works at all.
posted by gjc at 10:27 PM on April 23, 2013

real world Feline GPS data
via

It didn't look like someone else had mentioned it.
posted by sebastienbailard at 11:57 PM on April 23, 2013

Ok, so I am an "average" person with no real technological knowledge of how my GPS works. Here is my experience. I have a smartphone (Iphone5) and a handheld Garmin GPS. My phone is way less accurate than the handheld Garmin GPS. We frequently use the Garmin on hikes to track our trail and do some Geocaching. (http://www.geocaching.com/). Although the GPS has some errors in location, usually when it is still finding satellites, I find it very accurate. On our way back, traveling on the same path, the trail looks pretty much the same, usually not varying more than 1 or 2m from the trail it traced on the way in. Buildings and mountains can affect it, but from my experience in the woods I haven't noticed that. When we look for geocaches, (which are essentially mini treasure hunts using GPS), the marker is usually within 3m or so. Sometimes it is within half a meter, especially the newer ones. This depends on the accuracy of the person who placed the marker. Older models are generally less accurate, as I have noticed from comparing my Garmin to my parents' Garmin which is about 8 years older. And yes, it tells you elevation.
As for my smartphone, well it just isn't as accurate, and often when I'm driving it will take some time to catch up with where I really am. This was really obvious with my 3Gs but doesn't seem noticeable with my Iphone 5.
posted by photoexplorer at 8:23 AM on April 24, 2013

Second, it's easy to throw a little JavaScript together using that W3C Geolocation API and do a little testing of your own. Note that part of the returned value is `position.coords.accuracy`, which is supposed to be to a 95% accuracy, and I believe that my phone has given numbers as high as 6000. Those are meters. Usually that comes down fairly quickly, but right now my phone (HTC MyTouch Slide running Android 2.something) is reporting 30 meters. Try it out on a couple of devices...