1
\$\begingroup\$

I am curious about the accuracy of gps data from a regular iphone 7. I am conducting field research and looking for an accuracy of about 10 meters. If that is not possible with the regular photos taken on an iphone could someone recommend an app? I need there to be no visible annotations to the photos. The annotations need to be stored in exif tags.The image size ratio is 16:9 or 16:10 but that's not as important as having no visible annotations. The photos also need bearing/heading and GPS in the exif tags.

\$\endgroup\$
4
  • 1
    \$\begingroup\$ I'm voting to close this question as off-topic because this is about global positioning for field research, not photography. \$\endgroup\$
    – mattdm
    Oct 5, 2017 at 17:19
  • \$\begingroup\$ @mattdm To me, geolocation and EXIF data have become part of photography and the iPhone has become an important photographic system. Use of photography for research and other technical endeavors goes back at least 140 years to Muybridge. None of which is to say it is a good question, but if it's off topic, it seems like it would be as 'looking for an app' shopping advice. Not for the specifics of the photographic task (but I am not saying it is a good question in that regard either). \$\endgroup\$
    – user50888
    Oct 5, 2017 at 18:09
  • \$\begingroup\$ so the research I'm conducting has to do with photography as a tool for scientist,I am just doing the ground truthing part for a program that utilizes a deep-learning framework to automatically detect and estimate the spatial distribution of plants from geotagged static imagery. \$\endgroup\$ Oct 5, 2017 at 19:02
  • \$\begingroup\$ The problem is not how exact are the stored values, but how exact they were delivered from the hardware when the picture was taken. The answer to the literal question would be: they are stored exact to less than a meter - in other words, the storing in EXIF does not lose any data. \$\endgroup\$
    – Aganju
    Oct 6, 2017 at 10:21

1 Answer 1

4
\$\begingroup\$

GPS data varies wildly in accuracy depending on location.

In big cities, GPS can be very inaccurate because of the urban valley effect and the resulting multipath interference. Often, the best location data comes from proximity to Wi-Fi at that point, which can get you moderately close to the right place, but it still won't be all that accurate.

Similarly, if you're next to a vertical cliff (under it), GPS can be off by tens of feet because of reflections off of the cliff, and in that case, Wi-Fi doesn't help you at all.

The accuracy also depends on whether a given app asks for precise coordinates or not. The Camera app probably asks, but whether it waits for sufficient accuracy or not is another question. (It probably doesn't.) You might find that taking a shot and then taking another shot a minute or two later gives you better GPS accuracy on the second shot.

Either way, it sounds like what's critical for you is knowing the current accuracy radius before you take the shot. I don't know if the iOS camera app embeds the radius information or not, but I'm pretty sure there's no way to make it show it. However, it is possible to obtain the GPS accuracy radius from CoreLocation on iOS, so it should be trivial to write an app that shows it on your screen and warns you if you take a picture when the accuracy is less than your desired accuracy. (This, of course, falls more on the programming side of Stack Overflow.)

It's also worth mentioning at this point that CoreLocation will provide a location even if it does not have a GPS fix. The error radius of a Wi-Fi-only fix is bad. The error radius of a cell-tower-only fix is even worse. So unless you're regularly switching back and forth between a mapping app (that shows the radius) and the camera app, you'll have a hard time guaranteeing an accuracy better than the few miles offered by cell tower triangulation without writing a custom app.

\$\endgroup\$
7
  • \$\begingroup\$ It can also be affected by atmospheric conditions (weather) that can influence the number of GPS satellites received if it is a true GPS system (as opposed to a cellular triangulation system). \$\endgroup\$
    – Michael C
    Oct 5, 2017 at 21:01
  • \$\begingroup\$ In practice, there's < 2dB loss in signal from atmospheric causes (gpsinformation.net/gpsclouds.htm). However attenuation can make it more problematic to lock on initially. This is where a cell phone's aGPS is a huge win, because you can get the almanac and ephemeris data over the cellular network. \$\endgroup\$
    – dgatwood
    Oct 5, 2017 at 21:10
  • \$\begingroup\$ I would think there is more than 2dB loss when under a lightning storm (that isn't just blocking the GPS signal but adding electromagnetic radiation on a wide spectrum) or a hurricane. I've been in tropical storm conditions when a large, external GPS antenna/receiver (for satellite communications) could not find enough signal to find location at all. It may have been water on the surface of the antenna cover that was the problem or it may have been the lightning. \$\endgroup\$
    – Michael C
    Oct 6, 2017 at 5:11
  • \$\begingroup\$ It could be water on the antenna, or perhaps there were enough momentary periods of weak signal to prevent decoding the ephemeris data. To get a lock, the GPS receiver needs to know what satellites to listen for data from, and where those satellites are at a given point in time. Once it has a lock, it can use even a dodgy satellite signal, but it needs to get data continuously for an extended period of time to reach that point (30 seconds for ephemeris data, 12.5 minutes for almanac data). Enough water on the antenna could prevent that, but a cell phone (with service) would have no trouble. \$\endgroup\$
    – dgatwood
    Oct 6, 2017 at 17:42
  • \$\begingroup\$ I like the term “urban valley effect” and intuitively I understand what it means but on Google I can’t find any references. Do you have a source for this term? \$\endgroup\$
    – agtoever
    Oct 7, 2017 at 6:56

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