GPS Sport, Running and Personal Trainer Devices Proven to be Unreliable

Many people run and train with various devices that use GPS to track locations in time and so to calculate distance run, time taken, average speed or pace and instantaneous pace. There are many watches, iPhone and Android apps and high-tech electronic devices made by various companies.

These devices include Polar, Nike, Garmin and Timex and many others. Some of these devices can be set to beep at set intervals, such as every mile, or every 5 kilometres. After you have finished training or a race, the device stores the information in a built-in database and you can download all your data onto your computer. These tools have become very popular and many people use and rely on them. But how accurate are these devices given that GPS is known to be unreliable and can have degraded performance caused by interference from trees, buildings and mountains.

Step-counting watches and gps trackers are very popular during training
Step-counting watches and gps trackers are very popular during training. Source: Public Domain CC0
Runner with eight watches for the tests
Runner with eight watches for the tests. Source: https://runninglifestyle.blogspot.com.au/2011/07/great-gps-test.html
Track Run - shows straight lines between position readings
Track Run - shows straight lines between position readings. Source: https://runninglifestyle.blogspot.com.au/2011/07/great-gps-test.html
Trail Run
Trail Run. Source: Public Domain CC0

Many runners who have checked the results of their devices have been very disappointed. The problem is that people expect them to be extremely accurate and not affected by interference. All the devices are dependent on the limitations of the GPS technology which is improving, but some inaccuracy remains in certain circumstances.

Trees, mountains, deep ravines or clouds or surrounding tall buildings can block or interfere with the satellite signals required for the devices to calculate positions and work out distances. Tracks with lots of turns cause errors as well. So if you run around a circular training track for 5-10 km, the results will be inaccurate. The device will draw a straight line between successive positions at one second intervals, estimating a series of these straight lines around the curve. So the distance around the curve will not be tracked accurately and the errors add up if many circuits are made - say for the 1500 or 5,000m events.

A global positioning system (GPS) device is a very complex receiver that gets signals from and array of 24 satellites orbiting the earth at an altitude of 20,000km. Each satellite transmits its identity and the time a signal was sent. The GPS receiver works recording the time the GPS signal sent by a satellite is received. From the time the signal is received and sent the GPS receiver then works out how long the signal took to get to the receiver. Distance from the satellite to the receiver is then calculated (for all three satellites). The device then calculates the geographic position or the receiver using simple triangulation using the distances.

Usually the signals from three (minimum) to five satellites are monitored. From the information the GPS device works out the longitude and latitude (position) of the GPS receiver on the earth’s surface.

Runner with eight watches for the tests | SourceTrack Run - shows straight lines between position readings | SourceTrail Run | Source

The best study of the accuracy of GPS watches under various condition was a simultaneous study of eight GPS watches. The results from this study are summarised below:

Three conditions were tested on three runs:

The results are shown in the tables below that have been adapted from those in the original study.

For the Circular track the average error for the distance was about 3% with a maximum error of 6%. The average pace error was about 2% with a maximum of 5%.

For the Trail Run the average error for the distance was about 9% with a maximum error of 17%. The average pace error was about 11% with a maximum of 22%.

For the Road Run the average error for the distance was about 3% with a maximum error of 10%. The average pace error was about 2% with a maximum of 10%.

The study showed that the GPS watches were quite accurate for the Track and the Road when there are no obstacles such as trees or buildings.

The most likely reason for poor performance on the Trail Run was probably the tree cover which average 75% cover. The trees block the signal temporarily and the watch then assumes a straight line to the next location it can pick up. This obvious leads to large errors approaching 20%. Similar errors could be expected in urban setting when tall buildings may block the signals.

Track workout consisting of 6 x 800m

WatchDistanceMileage error% Distance errorPace% Pace ErrorTime
ACTUAL 4.5 6:32 29:32
Forerunner 405 4.45 0.05 -1.1 6:38 1.5
Nike+ Sportband 4.44 0.06 -1.3 6:39 1.8
Forerunner 610 4.66 0.16 3.6 6:20 -3.1
Nike+ Sportwatch 4.51 0.01 0.2 6:33 0.3
Forerunner 210 4.77 0.27 6.0 6:11 -5.4
Forerunner 310XT 4.75 0.25 5.6 6:13 -4.8
Forerunner 405 4.64 0.14 3.1 6:22 -2.6
TimexGPS 4.44 0.06 -1.3 6:39 1.8
AVERAGE ERROR 2.7% 1.3%

Trail run consisting of a 2.2 mile loop 3 times

WatchDistanceMileage error% Distance errorPace% Pace ErrorTime
ACTUAL 6.6 7:37 50:12
Forerunner 405 6.12 0.48 -7.3 8:12 9.58
Nike+ Sportband 7 0.4 6.1 7:10 -4.23
Forerunner 610 6.23 0.37 -5.6 8:03 7.57
Nike+ Sportwatch 5.65 0.95 -14.4 8:53 18.71
Forerunner 210 6.01 0.59 -8.9 8:21 11.58
Forerunner 310XT 6.45 0.15 -2.3 7:47 4.01
Forerunner 405 5.93 0.67 -10.2 8:28 13.14
TimexGPS 5.5 1.1 -16.7 9:08 22.05
AVERAGE -8.9% 11.3%

Road run of 9.45 miles around White Rock Lake

WatchDistanceMileage error% Distance errorPace% Pace ErrorTime
ACTUAL 9.45 7:07 1:07:19
Forerunner 405 9.04 0.05 -4.3 7:10 0.7
Nike+ Sportband 8.56 0.89 -5.3 7:52 10.5
Forerunner 610 9.45 0 10.4 7:07 0.0
Nike+ Sportwatch 9.31 0.14 -1.5 7:14 1.6
Forerunner 210 9.48 0.03 1.8 7:06 -0.2
Forerunner 310XT 9.45 0 -0.3 7:07 0.0
Forerunner 405 9.41 0.04 -0.4 7:09 0.5
TimexGPS 9.29 0.16 -1.3 7:15 1.9
AVERAGE -3.2% 1.9%