It has been said that using a garmin device or proper GPS unit for recording strava activities is more accurate than using a mobile phone. Recently some anomalies cropped up on my rides where some activities were recorded very accurately and others not so much. I use a garmin e-trex10 handheld device I keep in my pack for the mtb and other things. I had recently re-calibrated my cateye which is now as accurate as I can get it and my GPS unit’s readings were corresponding very closely with strava and the cateye. I’m a bit pedantic about this sort of thing so I even went and measured in comparison using satellite data to ensure all was as it should be. Everything was in order except for when I rode at Northern Farm. Every NF ride I do, my strava under-reads by 1km or so when I do the red/black route. This sort of inconsistency bugs me in a way that I just need to get to the bottom of it to try and understand. The folks I ride with were getting closer strava distances to my cateye using the iPhone and android app on their phones than I was using my garmin. The garmin e-trex 10 has a setting where you can determine the frequency at which the device records a point in time and mine is set to its highest setting or “most often” so I know I am getting the most accurate readout the device offers. As it turns out, “most often” may just not be often enough. I have figured out what is happening. On a path with relatively straight lines or minimal curves, the recording goes without a hitch but as soon as you throw in intricate curves and bends such as the windy sections near the river at Northern Farm, it becomes apparent that the garmin is either a.) not recording enough points per x amount of time to create an accurate curve or, b.) the conversion process to GPX, whether it be on the device itself or between the device and strava is causing a loss in accuracy. My findings: Garmin e-trex 10 - NF track: http://img.photobucket.com/albums/v648/flannahs/nf_garmin_eTrex_zps8297495e.jpg As you can see, the recorded path is quite jumpy and cuts out actual distance travelled. This all adds up, resulting in under-read. I suspect the faster one goes, the worse the result will become – and I sure am no racing snake. I thought that maybe this particular unit was possibly not best suited to this type of activity (although why wouldn’t it be? It’s a purpose-built GPS unit) so I compared with the same section done by someone using a garmin edge 800. Result:http://img.photobucket.com/albums/v648/flannahs/nf_garmin_edge800_zps5c1e83a1.jpg Better? Debatable, but here’s where the kicker comes in. The next image is a track recorded on a phone using the android strava app:http://img.photobucket.com/albums/v648/flannahs/nf_android_zpsace998e3.jpg Note how the curve follows the actual singletrack a lot closer than either of the garmin devices. iPhone strava app:http://img.photobucket.com/albums/v648/flannahs/nf_iphone_zps3d4ad531.jpg It seems to me that where strava and distance is concerned, the mobile phone apps may actually be better than a purpose-built garmin device – depending on the complexity of the track recorded. As far as average speed and elevation goes, I believe that the garmin devices are far more accurate. The garmin e-trex 10 does not use a barometer to calculate elevation but using tracks I have compared with satellite imagery, it seems to accurately compare. Average speed and general speed readouts appear most accurate on the GPS devices. I have tested in a car and it’s even possible to see the built-in safety margin the car manufacturers give car speedometers by comparing with the GPS unit. From what I can see, the mobile phone apps are most inaccurate with average speed, followed by elevation gain in comparison with the garmin units.