Reverse Geocoding
Reverse-geocoding is the act of taking a pair of coordinates, such as:
37.33693, -121.889464
and interpolating the nearest possible address that would equate to that location, such as:
4 N Second St, San Jose, CA 95113
Reverse-geocoding is used when you wish to take a latitude and longitude point, usually gathered in batches from Global Position Systems (GPS), and make them by human understandable.
Reverse-geocoding, like geocoding, is more of an art than a science. The reasons behind this are:
- The inaccuracy in the coordinate calculation of the GPS. Inaccuracies can be anywhere from 15m to over 100m (think the length of a small street).
- Unless the velocity of travel for a specific GPS point is known, multiple streets might be candidates for the closest street to the point.
- The underlying road contains only address ranges, not specific address points. This forces the reverse geocoding engine to guess at the address.
All of the above can lead to an example like the following:
- A user of a GPS device is standing at the entrance to the office building that houses the deCarta corporate headquarters.
- The building number is 4 N Second St. The building itself is physical located about 3/5 up the first block of N Second St.
- The GPS device, including the inherent errors, calculates that the user is standing at 37.337, -121.89.
- The user of the device pushes a button to request a reverse-geocode.
- The reverse-geocoding engine notices that N Second St, San Jose, CA 95113 is the closest street.
- The reverse-geocoding engine notices that the point is on the even numbered side of the 100s block of this street. However, the point is so far up the street, that the address gets interpolated to building number 56.
- The GPS device tells the user they are standing at 56 N Second St, San Jose, CA 95113. Precise, but inaccurate.
When integrating reverse-geocoding into an application, users who read the output should be informed that the addresses returned by the reverse-geocoding operation are best-guesses and might be extremely precise, but not entirely accurate.
Back to the glossary
|