What is Reverse Geocoding Anyway?

Happy 2017! In the last 2 posts, I've discussed geocoding basics - turning an address, neighborhood, city, etc. into a geometry, and geocoding accuracy. In this post, I will cover the basics of reverse geocoding. At its most simple, reverse geocoding is the process of obtaining an address given a coordinate (a lat/long or x,y pair of values)… an address is much easier for a user to understand and find. For example, a utility crew tapping on a map on a handheld device can provide the x,y coordinate which is used to perform the reverse geocoding. If a polygon is specified, the centroid (an x,y coordinate) could be determined and the reverse geocoding performed on that. However, reverse geocoding a polygon based on its centroid can be prone to error - take the centroid of a parcel of a corner lot - which street should be returned as a result of the reverse geocoding operation?

In my blogs, you've seen points represented by two decimal numbers separated by a comma - this is called the Decimal Degrees representation. There are many spatial reference systems (over 3800), but one of the most prevalent ones is WGS 84 Spheroidal (used by the GPS system). Each spatial reference system (SRS) has a unique identifier called a SRID (Spatial Reference ID) or WKID (Well Known ID). The SRID for this particular SRS, which Google Maps uses, is 4326 (see here). Look for an upcoming blog post for a primer on Spatial Reference Systems.

In regular geocoding, an address is typically specified and only one result is returned - the x,y coordinate. However, more than one value is returned when an x,y coordinate is reverse geocoded. Why? Because that coordinate may not only specify an address, but it is within nested political boundaries. For example, consider a coordinate located somewhere in Washoe County, Nevada in the United States (the reverse geocoding webpage is accessible here).

This represents the closest address to the specified coordinates. Clicking on one of the other returned centroids (in this case the ZIP Code (point B)) shows the following. It does warn that the coordinates are approximate. Which is correct because I know areas to the east of I-580 have a ZIP Code of 89521.

What's interesting is regardless of whether you click on the entry for point C or D, the centroid is the same (40.560839, -119.603549) even though the bounds are different - which is possible. But I'd expect the centroid for point D to be much closer to Reno-Sparks (i.e. south of point D).

And, here is Reno-Sparks, NV:

Hopefully, after my first 3 posts, you have an appreciation for geocoding, geocoding accuracy and reverse geocoding. More importantly, I hope you've thought about how geocoding your enterprise data can bring your organization insight maybe it hasn't had before and in a way that's much easier to understand. I'd love to hear your feedback!