GPS Accuracy–It’s all guesswork!

GPS receiver accuracy is a moving target. I’ve had GPSes get me within 10 feet of waypoints I stored on my old Lowrance. I’ve had my usually very accurate Garmin 60CSx get quite confused about my location on switchbacks in a rock chute area and show wildly inaccurate location points.

GPSes are darn good for giving you a decent estimate of how far you’ve gone and where you are. Consumer-grade GPS receivers are not accurate enough for road-race course certification, and I’ll bet no one is running with a multi-thousand dollar survey-grade GPS tied to a base reference station. When I took my 60CSx over to the University of Arizona’s track and walked the track holding the GPS a foot from the inner rail, it claimed a 400 meter track was 405 meters around. If that percentage error carries to longer distances, that’s significant.

On the other hand, GPS technology helps me create decently accurate maps of trail runs that give runners estimates of how far they’ll run and how much they’ll climb. Especially on routes that aren’t marked on good maps with those estimates, GPS support is invaluable combined with modern computer mapping software and color printers.

So how does a GPS receiver like yours or mine make its guesses? According to a rocket scientist friend, an author of GPS mapping software, and reported conversations with a manufacturer’s technical support, the standard mode of operation of pretty much all GPSes whether they cost $100 (in your hand) or $40,000 (in the rockets) is to take a reading once a second. If you look at the specifications for your unit, it probably says something like “within X feet 95% of the time”. In the case of my high-end model, that’s 10 feet. Every reading is a little bit different, whether you’re moving or not. On a recent run at a local park, I set my GPS to record track log points every second; as often as the GPS receiver took readings from the satellites. Here’s a screen shot of a tiny section of that run courtesy of TopoFusion (a Windows program)–

Coming from the north, I set my GPS down the a low wall while I went into the restroom (my friend guarded the GPS). The strange series of point you see is the GPS drifting in its readings as it sits on the wall, then me coming out and picking up the GPS and trotting off. Most mapping software will treat the point drift as movement. The GPS itself decides after awhile that its probably not moving and stops adding distance to its tripmeter as its readings drift.

Now here’s another screen shot of the same run, this time from MacGPS Pro (a Mac product; equal time)–

I ran west on a dirt median between the road and a wall, then back on the same median, about 3 feet from my trip out. Do these look like straight, parallel lines? This situation was a little worse because I was under trees. My GPS gets good signal under trees, but still doesn’t do as well as in open spaces.

So the GPS can make it look you’re moving when you’re standing still, and even when you’re running a straight line, it may look like you’re a bit tipsy. Calculations of the distance based on summations of distances from one location point to the next will ALWAYS overestimate the distance traveled.

Most GPS receivers try to make good guesses about when you’re moving and when you’re standing still. They try to correct for drift or scatter in the location data points. GPS manufacturers use sophisticated software running filtering algorithms to improve the GPS’s guesses about where you are and what you’re doing. But it’s all guesswork. Likewise, for my Garmin GPS, the track logs it collects for you are it’s best guess about when to record a track point in the track log, unless you override that like I have for my experiments.

So what’s the message here? GPSes tend to make decent, but not perfect guesses about where and how far you’ve traveled. For most purposes, that’s close enough. If you’re comparing a GPS distance estimate to a carefully measured, certified road race course, don’t argue with the race director that your GPS is right and his or her course is wrong. On trail runs, we’re hoping to get within one or two tenths of mile of reality, but my fingers are crossed.

In the months to come, I’m going to lay out some accurately measured calibration courses, both straight and zig-zag. I’m going to try out different GPSes and see how close they are to reality. I’m considering using a measuring laser since they’re more accurate than tape measures. If I get the laser, I’ll experiment on some trails, too. I may need a volunteer to run and stop, mark a line in the dirt, get shot in the back with a laser, and run again. I’ll follow from line to line. Sound fun?