How Accurate is Google Maps

For this year's AVC I decided to focus on writing better support tools for programming my car's path.  Last year my approach had been to physically drive the car to each waypoint and let the car's GPS take a record a measurement.  However, once the competition started, I found that I had to resort to Google maps to tweak the coordinates as I no longer had access to the track to take new measurements.  So, this time around I've designed a Java application that uses maps supplied by Google Static Maps to create a tool that lets me place moveable markers (to record the location of obstacles, such as barrels) and waypoints on the map.  The tool is coming along quite well and I'll probably write about it more in a future post but during development I began to wonder just how closely GPS coordinates taken from a receiver would match up with the same position displayed on Google Maps.

The National Geodetic Survey supposedly has physical markers placed in various locations and many of these markers have been subsequently surveyed with precision GPS receivers and the locations logged and recorded.  After some research, I was able to locate a marker designated as "SAN DIEGO GPS 17" a few miles from the office where I work and I drove by after work one day to check it out.  Here's a photo of what I found:

The actual marker is a brass disc in the upper right corner of the concrete pad on which it's placed.  However, the disc itself does not record the GPS location.  To get that you have to consult the full NGS Datasheet for this marker, which describes it's position and condition, and give many more details about it.  Here's the first part of what that datasheet describes:

PROGRAM = datasheet95, VERSION = 8.1
1        National Geodetic Survey,   Retrieval Date = MAY  8, 2013
 DC2129 ***********************************************************************
 DC2129  CBN         -  This is a Cooperative Base Network Control Station.
 DC2129  PID         -  DC2129
 DC2129  COUNTRY     -  US
 DC2129  USGS QUAD   -  LA JOLLA (1975)
 DC2129                         *CURRENT SURVEY CONTROL
 DC2129  ______________________________________________________________________
 DC2129* NAD 83(2011) POSITION- 32 49 06.34554(N) 117 08 52.19963(W)   ADJUSTED  
 DC2129* NAD 83(2011) ELLIP HT-    90.953 (meters)        (06/27/12)   ADJUSTED
 DC2129* NAD 83(2011) EPOCH   -  2010.00
 DC2129* NAVD 88 ORTHO HEIGHT -   125.5    (meters)     412.    (feet) GPS OBS   
 DC2129  ______________________________________________________________________
 DC2129  NAVD 88 orthometric height was determined with geoid model    GEOID99
 DC2129  GEOID HEIGHT    -        -34.52  (meters)                     GEOID99
 DC2129  GEOID HEIGHT    -        -34.59  (meters)                     GEOID12A
 DC2129  NAD 83(2011) X  - -2,448,209.147 (meters)                     COMP
 DC2129  NAD 83(2011) Y  - -4,774,387.181 (meters)                     COMP
 DC2129  NAD 83(2011) Z  -  3,437,102.525 (meters)                     COMP
 DC2129  LAPLACE CORR    -          6.86  (seconds)                    DEFLEC12A
 DC2129  FGDC Geospatial Positioning Accuracy Standards (95% confidence, cm)
 DC2129  Type                                         Horiz  Ellip  Dist(km)
 DC2129  -------------------------------------------------------------------
 DC2129  NETWORK                                       0.75   1.86
 DC2129  -------------------------------------------------------------------
 DC2129  MEDIAN LOCAL ACCURACY AND DIST (060 points)   0.95   2.44     76.19
 DC2129  -------------------------------------------------------------------

The GPS location, converted to decimal degrees is 32.818429, -117.147833.  Punching this location into Google Maps produces the following display when viewed with Google Map's highest zoom level:

As you can see, the position indicated on Google Maps is not exact, but it quite close to the actual position in the upper right corner of the pad (indicated by the white, L-shaped stripe on two sides of the concrete pad.)  To my eye, the indicated position is within a few inches of the marker's actual location.  To see for yourself, click this link to view in Google Maps.  Note: deselect the default 45 degree view (move mouse to "Map" in upper right corner to reveal option) to get a more accurate view of the location.

How about Google Earth?

I entered the same coordinates into Google Earth and found I could not zoom is as close as I could when using Google Maps in my browser and the images are less sharp than on Google Maps.  However, the results, as you can see below, seem to be about the same as with Google Maps:

Let's try Bing, Too!

Bing seems to a little closer to the mark than GMaps

Google Maps on iPhone is not so Accurate

Curiously entering the same GPS coordinates into the latest version of Google Maps on my iPhone produces a very different result, placing the marker out in the middle of the nearby street, Kearny Village Road, as shown here:

Click to see larger image

Out of curiosity, I tried the same thing using Apple's much maligned Maps software on my iPhone and got this result:

Click to see larger image

Here's the result is not as close to the marker as when using Google Maps in my browser, but it's much more accurate than Google Maps on my iPhone.  It's hard to see why this would be the case, as I would imagine the two versions of Google Maps use the exact same dataset.  But, perhaps there's a bug in the iPhone app that causes an offset?