Discussion: Measuring height gain with GPS

I've been trying to work out the best way to measure height gain on a run. I recently ran a 10m race in Abernethy, and according to the map crossed about 13 contours, suggesting 130m of climb. But judging from my aching legs and the fact that mile 8 contained a steep uphill section not easily seen on the map, I was sure it was a bit more than that. My Garmin (Forerunner 205) registered a very generous 500m. Can't believe that either. I think the Garmin, while generally quite accurate on elevation, wobbles about quite a bit - it showed a significant hill of 100m on one of my local flat runs (never above 70m elevation!), and these wobbles accumulate. I tried Google pedometer, and once I've logged in and saved my route, it also calculates a believable height gain. (About 200m at Abernethy). Anyone else tackled this?

Best solution for accumulated climb is a gps setup with a pressure sensor, either a dedicated device or a suitable phone and app. Next best solution is to post process the data doing an online lookup to re-evaluate the altitude thus ignore the gps data which can be very bad for altitude. This is probably what the Google pedometer is doing. I am not sure what Attackpoint is doing in this regard if you upload from a Garmin.

Attackpoint can correct the altitude for you when uploading from a GPS. Like Ifor says, the way this works is that the GPS-recorded altitude is replaced with a known/surveyed elevation at the lat/lon of each track point. It does assume you were on the ground. There is also a special mode for devices with barometric altimeters, where only the end-points of the track are fixed to known elevations, and the rest of the presumably-more-accurate barometric readings are skewed to fit the end points.

Actually calculating a total number for height gain is another question. Generally you want to smooth over height fluctuations that are smaller than the expected accuracy of the elevation profile. AP does something like this in order to avoid (for example) accumulating lots of non-existent 1-meter hills. On the other hand, that might mean you don't get credit for a real 1m hill, so it is a bit of a trade-off. In practice, it seems to give mostly reasonable results.

Thanks Ifor & Ken. I hadn't noticed the donate-for-extra-features message and have duly donated. The correction applied by AP appears spot-on: easily accurate enough for what I want.

Ken, are you still replacing all uploaded climb data with database elevations? Seems like my kayak paddles shouldn't have any climb if this was the case, but they typically do. I'm using a Forerunner 920xt which has a barometric altimeter. From what I understand it isn't accurate for the first 60 mins of an activity.

While consumer-level GPS data is not great for elevation, doing a map lookup also has its fish-hooks, and there's some discussion on this in GPS forums. Any error in the horizontal position, or the map, will produce spurious climb. Take a typical hill-country track, which winds in and out of the gullies. Displace your GPS file a bit, and it will think you're going up and down instead of in and out.

Replacing depends on your AP settings (Settings > gps elevation correction). The modes still act as described above, and AP is up-to-date knowing about barometers in currently-released Garmins. My general recommendation is to use "auto select mode". You can try other modes to deal with weird behavior on particular sessions.

As gruver said, errors can come from the elevation databases, the horizontal track positions, and even recorded barometric data.

I took a look at the recorded altitude data in your June 19th paddle file. In this case it actually looks better for the first hour...the second hour fluctuates quite a bit (+/- 4m) from even a constant slope (the whole session is tilted by a general weather pressure trend...start -11m, end -20m). I don't know how to explain the fluctuations. If you don't trust your altimeter data, then maybe better results forcing AP to the non-baro mode.

For kayaking, how about other factors: tides, waves, waving your wrist up and down? Will the garmin pick up when you raise and lower your arm by a meter?

The elevation profile of a recent run shows a sudden drop of about 5 extra meters at the exact moment that I took a downhill fall, about one minute into the run.

I wonder if getting gunk in the barometer port could also cause funky readings in some cases. I haven't personally noticed that to be a problem though.

Fixing the altimeter on Garmin 910xt. In short: soap, water, and a toothbrush.

If you raise and lower your arm by a metre when kayaking, then perhaps I don't have the worst technique on earth.

My Garmin 76 CSx had excellent barometer performance, but wokred quite badly when on an airplane since it recorded a steady 2500-3000 m altitude (depending only on the airplane type). :-)

2500-3000m?

What did you expect? That sounds like good cabin pressure control.

I note a smiley at the end of Terje's post. ..

Best method is to lay the gps track on top of a 1m contour map produced from lidar and to count manually. Last night Garmin said 90 metres but map said 240.