Wednesday, November 25, 2009

Garmin innacurate on hills

For some time now, I have suspected that some models of Garmin running watches are not accurate when running on hills (even moderate hills). This summer, during a certified 8K road race with some decent hills, I confirmed this. The 8K road race should have been about 4.97 miles, but the Garmin underestimated it at about 4.93 miles. Here's the explanation.

Garmin watches (at least the lower end models), calculate distance run by measuring only latitude and longitude at various points. There is no correction for elevation. In other words, picture a right triangle, in which the hypotenuse is the hill you are running up, and the base is simply the map distance traveled (subtracting the latitude and longitude of two points along the base). The hypotenuse (the hill) is the longest side of the triangle, but the Garmin actually measures the base distance, so it underestimates distance when running up or down hills, and thus overestimates your running pace (your pace is shown as too slow on the watch).

Where I live and train, elevation changes on runs are so minor (maybe 5 to 10 feet at most), that there is essentially no lost distance. However, in an area like Charlottesville, unless you are running only on the track, if you have a few decent hills in your run, you will be consistently underestimating your distance run. The good news is that you are actually running a faster pace than shown on your watch.

Feel free to comment on this. I am hoping that the newer models will have more accurate vertical GPS capabilities (the 205 shows elevation, but it is extremely innacurate (my runs show 100 foot elevation changes, when 5 feet is more like it). Then they could correct for elevation (use trigonometry calcs to calculate the hypotenuse). However, I doubt this is going to happen for some time. It's funny Garmin never mentions anything about this. When I ran that race, I was a bit discouraged with my average pace throughout the race, but when I finished, and calculated my actual pace, it was 3 or 4 seconds per mile faster.

Mark

No comments: