I’m reviewing photos taken on an Android. The photos have GPS metadata, but the location information doesn’t make much sense when other details of the case are considered. Anyone have thoughts on the reliability of the GPS information? What could cause the phone to insert inaccurate GPS data into photos taken with its camera? The suspected inaccuracy is on the order of miles, not a few feet.
In my experience, GPS is generally very accurate, +/- 10 feet. The caveat is that the device is receiving signals from GPS satellites, and has a lock on at least 3 satellites. However, if a signal lock is poor, or non-existent, then the accuracy can be off quite a bit. The worst case I personally dealt with on a GPS device was literally a half a world away. Several miles is not unheard of.
Also, I seen some phones triangulate their position off cellular towers when GPS signals aren’t available. In that case, the accuracy may be anywhere from several hundred feet to several miles from the actual location.
I’ve also seen a case where one figure was accurate, but the other was off. e.g. the Latitude was accurate, but the Longitude was 30 miles away. That particular issue may have been from only getting a lock on one or two satellites.
A number of factors can effect a GPS signal, including terrain and being inside steel and concrete structures. The combination of terrain and buildings making “concrete canyons” in large cities is an issue, even when outside. On a smaller scale, signal-blocking can occur from metallic containers, including Faraday bags.
If you plot out on a map the places that the GPS shows as a given location in the photos’ metadata, and then plot out the actual location where the photo was taken, you may gain some insight into what’s gone wrong.
Thanks for sharing your knowledge, Frank. Some great information here!
Here is another idea: Could it be that the end-user intended to misrepresent their location? This is very easy to accomplish on an Android using a mock location provider. One could take a photo of their couch cushions and make the phone GPS mark it as if they were on an exotic beach somewhere
Doing this for a fixed location for a photo takes just a few clicks on the device–no need to connect the phone to a computer, etc. On the other hand, making the phone think it is following a certain path at a certain velocity (e.g., navigating a road in a remote location) would be a bit more challenging.
If this is a concern, I would review the phone data for evidence of mock location use.
How are you reviewing the data? Do you have the json data? Metadata itself is generally not as reliable as json data. If you have the json data, we are able to plot the path of the phone. With this data any anomalies are obvious and the real data can be observed.
One other issue that pops up is user error on the part of the examiner. Two problems that initially goofed me up until I figured out what was going on:
In the US we use positive Latitude (North), but negative Longitude (West). If you are manually mapping the location, and neglect to use the correct sign on the map program, then your results will literally be 180° from where you expect.
Swapping the Long/Lat values on the input to a given map program. It’s easy to do if you don’t know how the map program expects input.