Discussion: Correct climb with Forerunner 305

When I download the data from a run from my Garmin Forerunner 305, I get markedly different results for the climb depending on which program I use. For example, Sporttracks said I climbed 82 m on a run, and the Garmin Training Center indicated I had an ascent of 287 m! Quite a difference. Sporttracks always seems less than I think it should be, but TC always overexaggerates the climb by a considerable margin. How do the 2 programs calculate the climb, and which is more correct? (I think Sporttracks is probably closer to reality, but I think it undercalculates the climb.) Are there other programs that give a more accurate calculation of climb?

Based upon 201 experience, I think accurate calculation of climb is unlikely due to the much greater vertical error of the gps. A long sustained climb of 500 metres will result in a reasonably accurate figure, but a run with many small clilmbs and descents will be difficult as the size of each small climb will not be much different from the error margin of the gps.

I have a Garmin 305 and I look at the data on the Garmin Training Center. I think the distance is very accurate (even when running on trails on in the woods), but I never bother looking at the climb. Last time I ran on the track, Garmin claimed I did over 300 feet of ascending during 4km...

I use Motionbased which calculates climb using something it calls Gravity Elevation Correction, it seems to be pretty accurate.

I think that TC does not make any attempt at correcting the climb (or a very crude one): it just sums the differences between consecutives points.

Sporttracks, as many others, try to smooth the errors in GPS altitude by making a moving average. If I remember well, Sporttracks allows you to configure this moving average, decreasing or increasing the smoothing. Trackrunner on mac does a pretty good job about it also.

But I use MB Gravity whenever I want a clean evaluation of my climb though. It's not perfect, some minor hills/depression are not registered in database because hidden in the forest, and if you ran above a big cliff and your trace is slightly off the terrain model, it could appear as if you ran at the cliff foot, thus making extra climb down-up. If really it makes a big difference, I correct my trace but usually dont care.

I have maps with 5m contours of all the areas I usually run and I generally use these to calculate climb. I find the sporttracks figures are usually pretty close but tend to overstate the climb slightly most of the time and sometimes vastly overstate it when there are some obvious errors in the GPS elevation. I sometimes use the sporttracks figures when they are close to what I expect and I can't be bothered counting contours. I tried using MB gravity once but it didn't seem any more accurate so I didn't bother trying again.

I did a mesurement of elevation on a bike ride today using two devices simultaneously (GPS Garmin Forerunner 305 and Polar S 720i). See comparison.
Curves look much the same, but ascent given by GPS analysis SW is far too high... Polar, using barometric measurement, gives a much better estimate.
True ascent, counting contour lines, was approx. 520 m.

You get better result with your Polar if you set the start and the stop elevation at the same level. There should be tool for that in Polar software. As you can see the elevation was steadily sliding down, by fixing that we might get pretty close to 520m.

Hi Kurthu,
while the curves of garmin and polar look similar, the calculated ascent is totally different between both devices. this suggests, that the garmin training center has a bad calculation algorithm, but the measured elevation is somewhat ok (within a range of 10m or so). therefore, the improvement lies in the data analysis. it is said, that the sporttracks gives a much better calculation of the ascent, have you tried this?

Anyway, thank you very much for your comparison, it was really, really helpful!

While orienteering with a 305, I've noticed similar to simon that the Sporttracks smoothing algorithm seems to give a pretty good result. The Sporttracks calculation often comes a little under what course setters list.

I can confirm good results for Sporttracks elevation calculation. For the comparison above:
Real ascent: 520 m
Polar:490 m
Sporttracks: 567 m
MotionBased (Gravity Web Service): 717 m
Garmin TC: 821 m
MotionBased (device): 1017 m

There is a plug-in for SportTracks (Elevation Correction plugin) that calculates the climb from your workout data based on the NASA's SRTM data. I have recently downloaded it, so I can't vouch for its correctness. Based on the reviews and the download frequency (the most used plugin), I guess it will be hard to beat that.

So I downloaded the plugin. Now how do I install it so it works with the program?

Installation description here.

I used to use Training Center for climb, but Garmin Connect gives a much more accurate total climb.

...great thread necromancy there, Nick. :)

This is old news to anybody uploading tracks here, but Attackpoint has integrated SRTM-based elevation correction as well.

I think 'thread necromancy' is a wonderful phrase, and possibly also an enjoyable pastime on AP. But it would only be worth reading if it followed a basic rule... new information to inform unresolved discussions. So, as long as it follows that rule, bring it on.