Altimeter inaccuracy

7dfcd1938c0146f1804397d33ddc9c14 said:

I tried manual auto calibration uaing GPS and then the elevation reading reading says I am at 200 m, 199 m, 198 m, ... and it never stops dropping. At one point it says "calibration failed"

Garmin should do a better job of explaining the limitations of GPS altimeters. Even in ideal conditions they're not very accurate, and it's normal for GPS elevations to bounce around within +/- 25 m or so. This isn't Garmin's fault; it's an inherent limitation of the GPS system. It can be much worse if there are only a few satellites in view (there have to be at least four to get an elevation reading), or if GPS signals are bouncing off steep cliffs or the sides of building before they reach your device.

achilles_t said:

The issues appear when you do not run along a predefined route  and by default the watch estimates the altitude from scratch.

Yes, I agree. As I understand it, Garmin's auto-calibration algorithm attempts to correct barometric elevations based on GPS elevations, but the corrections are applied quite gradually, because GPS elevations are inherently noisy.  If there's a large discrepancy between barometric and GPS elevations when you start recording then it may be a while before the adjusted elevation converges on an accurate value, and in the meantime your recorded elevations could be quite inaccurate.

For those that care about such things, there's a "GPS Elevation" data field that you can add to a data screen. If it's consistently different from the regular "Elevation" field that's a good indication that your elevation readings may not be accurate.

A good explanation is always good. But I'm getting my "climbed 10 floors",  after walking my kids to school for 1,5km every morning. Thats not measuring. 

I know about GPS limitations. We are not talking about +-25 m deviation. It started with 200 m and dropped to, I guess, 100 m, and then it failed. My real elevation should be at around 150 or 160 m

When you wrote that you did calibration "using GPS", did you choose GPS or DEM for calibration? GPS takes the elevation directly from GPS altitude, which is really noisy and can be incorrect. DEM uses GPS only to get longitude and latitude, and then watch's internal DEM (digital elevation map) to look up altitude. DEM usually gives more correct altitude, since in most places altitude differs only a little even if the location accuracy is not perfect.

as written several times. When the altimetry is incorrect even though the sensor hole is clean, it is a hardware problem. I run with a buddy who also has the 955, his recordings are always perfect and mine suck. I'll have the 955 changed before the warranty expires. 

I have had an inaccuracy caused by temperature change, but really huge one. I have tried all tips and tricks mentioned here. My watch is replaced now for new unit and now it is super precise. Question is for how long...

you can be happy that garmin replaced your watch. The Slovak Garmin representative wrote to me by email that a deviation of 15-20 meters is fine...

As I have written here on the forum many times, most of my activities have different ascended and descended metres for each activity. 

Also today, a 14km run from point A back to point A, the difference in gained height was 18 meters. I calibrated the watch twice before the start using GPS... I thought until now that it was because of the cold thats affects the barometer, but today it was quite warm and the difference between the gained heights  is huge. There is something wrong with the barometer in the watch and I can't do anything about it. Here on the forum, no one from Garmin will respond to your problem.

left, data from watch, right data after corections on garmin connect web...

first number is ascent, second descent

Hi,

there seems to be a big difference between individual devices. I used an instinct solar for some years, where I always noticed a significant drift in altitude (quite certainly caused by temperature change) - like 30 to 50 m difference when returning to my starting point of a run of maybe 10 k. Since about half a year I´ve been using the FR 955 solar and am positively surprised by the accuracy of the altimeter, which is by far more realistic than the one of the instinct solar. As an example: I had a long run two weeks ago (about 2 h, 20 km and about 360 m ascent / descent) - when I returned to the starting point it showed a difference of just 3 m, which for me is a perfect result (after 2 h!). And I have never experienced a really "bad" result since using the FR955.

It would be interesting to find out, why some devices don´t seem to work properly, while others of the same model do.

By the way, I have auto-calibration off and do a manual calibration every time I start off for a run (as I am mostly starting from my home, I know the exact value).

I have instant replacement on Alza. So nobody discussed with me. Garmin support in Czech is bad. It is my fourth Garmin watch and all have some HW problems, so instant replacement will be must have option, if I will buy another Garmin in future, what I am not sure right now. But I had much worse measurements, than you. It was like 50m wrong in 10 min run 

Throwing my own experiences in here.

I've had similar issues with the 955 Solar--when I first started using it, the altimeter worked fine, then it went haywire a few months ago. I hadn't put 2 and 2 together until I read this forum, but it does seem to coincide with the colder weather.

The issue is definitely a problem with the watch itself. I know that others have mentioned that GPS-based altitude will be off by +/-25m or so, but I'm seeing numbers far outside this range. As an example, I had a 10-mile run yesterday that was flat other than some slight undulation and running over a couple (short) bridges, and my watch says my overall elevation gain was nearly 25,000 feet (~760m). There's no way that's just a GPS problem--I live in the Midwest; we don't have hills.

The big issue here is that, at least for outdoor running, Garmin calculates running power using elevation gain (and other metrics), so it says my average running power is 612W and my max is 2,169W--not remotely realistic. (For comparison, on a recent treadmill run--GPS/altimeter inactive--my average power was 240 and max was 470. Still a lot of weird up-and-down in the graphs, but the overall averages made much more sense.) As others have also insinuated, because my watch thinks I'm running at such a high power, all of my other metrics are thrown off as well. My TSS for yesterday's run was 565 (compared to just 87 when calculated using heart rate), and my Training Readiness is perpetually in the single digits.

My previous watch (Forerunner 935) had similar problems with grossly overestimating elevation gain, although it didn't calculate run power. I'm thinking about upgrading to a 965, but I'm doing some research first; I don't want to make that investment if it's going to have the same issue by the end of the year & leave me at square one all over again. I agree with what others have said: there needs to be an option to turn the altimeter off entirely to avoid feeding junk data into my training plan.