Altimeter continuous calibration via GPS

Heavy filtering and smoothing. Something like: if the ten-minute average of GPS elevation, filtering out any values more than 20m away from those before and after, is in line with the ten-minute average of barometric elevation, no change; if it's 25m out, start gradually adjusting the barometric calibration.

It's not completely clear to me that the F3/5 watches aren't still doing this, mainly because my regular routes all have calibration points at the start now. I'll have to go for a long walk when there's a storm blowing in and see - that, and large elevation/temperature changes, are the things which throw the calibration off significantly.

Barometric altimeter may suffer from some drift but usually I find it quite minimal with a good quality altimeter like the one in Stryd for example. Perhaps 1 or 2 m per hour or sometimes none at all.

It makes me wonder how GPS could actually improve this though when the values I usually get from GPS elevation are much more over the place and random than altimeter even after drift.

It's actually not so much about barometer drift, which is usually not that high, but temperature compensation and associated difference in altitude.
The watch always assumes changes happen in the standard atmosphere. If it's warmer or colder than standard atmosphere, then the actual change in altitude is different. That's why in winter time when you reach the mountain summit, wour watch can indicate a few hundred meters more, and when you get back to the start, it's correct again.

Suunto's FusedAlti actually works remarkably well. It uses barometer readings for small changes, but compares it to GPS. It obviously uses some filtering and averaging on GPS, but if there's a drift, it will correct the baro readings. So even in the winter when you get to the top of the mountain, it's correct.

Exactly
Coming from a Suunto Traverse, the fused alti is indeed very accurate
I was disappointed to see that the F5 did not have something similar
Last weekend a friend of mine had his A3P with him and we were comparing our altimeters and the Suunto was much more accurate (with the fusedalti)

Suunto's FusedAlti actually works remarkably well. It uses barometer readings for small changes, but compares it to GPS. It obviously uses some filtering and averaging on GPS, but if there's a drift, it will correct the baro readings. So even in the winter when you get to the top of the mountain, it's correct.

It's actually not so much about barometer drift, which is usually not that high, but temperature compensation and associated difference in altitude.
The watch always assumes changes happen in the standard atmosphere. If it's warmer or colder than standard atmosphere, then the actual change in altitude is different. That's why in winter time when you reach the mountain summit, wour watch can indicate a few hundred meters more, and when you get back to the start, it's correct again.

Suunto's FusedAlti actually works remarkably well. It uses barometer readings for small changes, but compares it to GPS. It obviously uses some filtering and averaging on GPS, but if there's a drift, it will correct the baro readings. So even in the winter when you get to the top of the mountain, it's correct.

I see. So probably if the drift is hundreds of meters it's better to have 10 or so meters off than that.

I had a chat yesterday with Garmin UK support and they opened up a ticket
Below is the reply

Dear Thomas,

I am writing to you in regards to your case 1345682.

If you turn Auto Cal on, the altitude will be continuously calibrated on track that use GPS.

Kind regards,

I also emailed the Fenix5 Beta team about the same issue
They also replied :)
The automatic calibration of the altimeter by the GPS is only at the start of an activity. Since a GPS calibration can be within 400 feet (120 meters), continual re-calibration by the GPS is not necessary. For a constant elevation accuracy of within 10 feet (3 meters), the altimeter would need to be manually calibrated every hour with your current elevation.

So, it is apparent that the 2 teams are not inline
I emailed them again and added both teams. Waiting for an answer

Unfortunately, F5 series does not support continuous altimeter calibration via GPS (similar to Suunto's FusedAlti)
Such a feature is very useful, especially for a high end ABC watch, since the barometric altimeter is bound to drift in long running
activities (such as hiking, trekking, mountaineering etc)

I emailed Garmin and they confirmed that the F5 does not have this particular feature
I asked them whether they accept feature requests and told me I could submit one at the below URL
http://www8.garmin.com/contactUs/ideas/

If any of you believe such a feature to be useful, please spend some time and submit it
The more requests they receive for the particular feature, the more possible it is to add it

Thanks

Done!

Also done

Garmin has always had problems with the altimeter, both barometric and GPS. The edge units, which have generally always had BP elevation, often have had bugs where mid ride you'd have the elevation suddenly jump by 50 or more feet, but not necessarily making it accurate. My first FR310 I believe (with just GPS elevation) was constantly getting messed up for total ascent every time Garmin would come out with a new FW. One FW release it'd be pretty good, the next it'd be off by a factor of 4 if on a relatively flat trail.

The Epix is a terrible example as it had so many bugs that never were even attempted to be fixed, especially around elevation.

Why going on 11 years after my first Garmin Forerunner they haven't been able to figure this out is a mystery to me. They've made many attempts along the way to get elevation and ascent right and seem to fail every time.

Those that think it's fine not to do on going calibration, try hiking a mountain with 5000' of vertical and finding out when you think you should be at the top, you still have another 300' to go. Happens pretty much every time I do this with a Garmin with a BP altimeter. For these situations, the GPS altimeter on my 630 works much better as it's never off by 300'. Usually stays within 50'.

Those that think it's fine not to do on going calibration, try hiking a mountain with 5000' of vertical and finding out when you think you should be at the top, you still have another 300' to go. Happens pretty much every time I do this with a Garmin with a BP altimeter. For these situations, the GPS altimeter on my 630 works much better as it's never off by 300'. Usually stays within 50'.

Exactly. My experience as well. The issue is more prevalent in activities with high elevation gain and long lasting ones
Above the tree line, the GPS is usually pretty accurate. I am not saying to ditch the barometric one, but they should use the GPS for reconsiliation