GPS in the woods.

"Why don't you post a GPS track of the trails?"

Because GPS is not as reliably accurate in the woods as is commonly believed. GPS can be useful for locating a trailhead or other large feature, but for finding one's way on a system of trails a simple map and basic navigation skills are more useful. A GPS track is useful for indicating a general route across a large expanse of country, but should not be depended on for following faint trails in the woods.

(If your interest in GPS is to avoid getting lost, please see the "How NOT to get lost" page for tips that will probably be more helpful than a GPS. And be sure to carry a compass--one that doesn't require batteries!--and a map.)

There are two basic problems to using GPS in the woods. First, it is often difficult to impossible to get a GPS signal in the woods, and without a signal the whole exercise is pointless. Second, consumer-grade GPS receivers simply do not provide accuracy nor even precision at a scale useful for trails.

For sure, geophysicists use GPS to determine earth movements on the order of a few millimeters (tenths of an inch). But they do this with fixed stations using top of the line survey-grade instruments, continuous signal monitoring, and heavy post-processing. Such "heavy wizardry" is not available when you stand at a trail junction trying to decide which branch your track takes.

It is not unknown to replicate a waypoint to within a meter. While GPS accuracy is highly variable, the several factors that affect the accuracy of GPS vary at different rates, and within a period of a couple of hours there is generally little change. So the position you obtained after lunch may be within a meter or two of what you got before lunch, but the position you get tomorrow might be 50 meters away. Even if you get some repeatability (precision), you need to consider: Is it reliable (i.e., will not suddenly take a wild swing)? And, is it accurate (actually close the the true value)? The only way to tell is to take set of positions (at least a dozen) over an extended period and compare them. This is seldom done. (But read about the GPS test.)

Take a look at this plot of five waypoints (positions) taken over several months (using a Garmin 12XL) at the center of the bridge where the John Wayne Trail crosses Boxley Creek. (Should have been eight waypoints, but on three occasions I couldn't get a signal!) The dashed circles represent the GPS receiver's estimated accuracy (or probable error range) for each waypoint. While two waypoints actually coincide, overall the waypoints vary by some 13 meters (over 40 feet); the estimated error ranges largely don't even overlap. This is good enough to find the parking lot where your car is parked, but please do not rely on it for driving across bridges. Nor even for finding the exit of the parking lot.

And that was a fairly tight cluster of waypoints. Take a look at this plot of eight waypoints taken about half a kilometer to the west, where the JWT crosses the section line. In trying to locate this position I have encountered a variation of about 200 feet. In trying to return to that point you are likley to encounter similar variation. If you are lucky we might have offsetting errors and you could end up right where I was (even if the reported coordinates are not quite correct). Or you might end up 400 feet away. Within this range is another trail and two old roads--one of which will catch you a trespassing citation if you get caught on it.

(By the way, I am collecting more waypoints for these locations. Read up on the GPS test for details.)

It should be noted for those not familiar with this area that the foregoing results were affected by being under partial forest canopy and having portions of the sky obscured by nearby ridges--which are very typical real-world trail conditions.

So how is this view to be reconciled with the common statement (such as found in the manufacturers' documentation) that consumer grade GPS receivers can be accurate within 10 meters (about 30 feet) or thereabouts? Partly because they can be that accurate, though they often fail to do so. The "fine print" that is often overlooked is that such accuracy is obtained only 60 or 90 per cent of the time; some fraction of the time you get a joker, but you don't necessarily know it. Also, these kinds of measurements are usually obtained under somewhat idealized conditions.

(For a more rigorous treatment see DLWilson's GPS Accuracy Web Page. For some information overload see Sam Wormley's GPS Errors & Estimating Your Receiver's Accuracy. For a lighter excerpt with a nice summary of GPS error sources and their magnitudes see the GPS General FAQ at

Consumer-grade GPS receivers can be useful in the woods--provided that due regard is given to their limitations. The primary problem is that under forest canopy, or where the topography blocks parts of the sky, the signal might be blocked, or only intermittently available (due to how the satellites shift). A repeatable and fairly accurate position might be reliably obtained only on peaks and ridges, and broad open areas; you should not depend on having a good signal elsewhere.

The secondary problem is proper scale. As shown above, typical GPS accuracy is good enough to put you in a parking lot. But once you have gotten that close you would certainly use your own eyes to find your car. GPS can show you the general location of a trail junction, but don't expect it to reliably guide you through a patch of Devil's Club.

By the way, GPS signals are notoriousy unreliable in regards of elevation, often off by 200 or 300 feet. If you need fairly accurate elevations, use an aneroid barometer (which some GPS receivers have built in).

My recommendations: GPS locations and routes should be regarded as broad spots or zones of fuzziness, typically 200 to 400 feet across. Within that zone you should expect GPS derived positions to be somewhat variable, and use good judgment (and your finely honed trail sense) to guide your steps. If you really need a more accurate position (perhaps for geocaching?), specify a landmark within the zone of fuzziness (perhaps a certain tree), and then specify a compass direction and distance from the landmark.

Back to Cedar Butte.