Last year I bought a Garmin Edge 200 and joined Strava.Several of the people I ride with also have Garmins and are on Strava. Their Garmins are more expensive than mine and when we post to Strava they show more elevation than mine. Last night was a good example. We did a group ride of about 26 mi. @ 15 mph. When I posted to Strava mine showed 1133 ft. of elevation while my friend, Ray, showed 1522 ft. of elevation. Just wondering if anyone knows why that is so.

Edge 200 doesn't have a barometric altimeter, so it uses the public elevation mapping data. Higher end Edge's have an altimeter, so all things equal, they are generally more accurate assuming the weather and temperature don't fluctuate significantly while on the ride. Even among devices with altimeters on the same route, it isn't uncommon for a ride to differ by a couple hundred feet. I wouldn't get too hung up on it.

Thanks. That's clear and concise.

It makes sense, though I don't understand where the "public mapping data" comes from, if not originally obtained from using altimeters?

It's probably topographical map data or provided by google earth, something like that.

Honestly -- I'm convinced that a lot of bike paths are just given an altitude that matches the street near where they are.

We have a bike path nearby that is a ton of climbing...up and down...up and down...

Think like what they build to stop noise from a highway from getting to a local town.

I know I don't get the proper numbers when I go on it with my Edge 200.

--

Elevation databases
For devices without barometric altimeters, we consult elevation databases to determine the elevation at each point in the activity. The resolution of these databases can vary based on location. For example, the USGS National Elevation Dataset (NED), which we consult for activities located in the US, generally has a 10 meter resolution though some small areas have a 3 meter resolution. For activities located outside of the US, we consult the ASTER and SRTM databases which have a resolution of anywhere from 30 to 90 meters.

https://support.strava.com/hc/en-us/...-Your-Activity

They probably stand more while climbing.

If you buy a more expensive Garmin, they boost your elevation so you can feel better about your ride. The more expensive of a computer you buy, the higher it tells you (and your friends) you've climbed. Just one of the perks.

If it also boosts MPH I might have to get one.

Put one of these on your bike and then you will get more accurate readings.

My impression is that a graded road with bridges is not the sort of elevation change that you would see on a contour map that the 'public mapping data' might be using. It's funny to look at the elevation profile from strava for the local rail trail vs what comes out of the garmin's altimeter. Strava looks like a saw tooth, which isn't correct if you can imagine how a railway grade really looks. heh

In other words, elevation changes are smoothed out when they build roads, and if the public data isn't based on how the road was built, then it will instead show lots of ups/downs that don't really exist.

Strava elevation was inflated by 30-40% in a lot of segments (Newport Coast, Pelican Hills, Ridge Park, Vista Ridge, Santiago Canyon...) here in Orange County, South Cal. For example, Strava will give you two very different readings for the same loop clock- or counterclockwise. On the other hand, my Garmin 800 (with barometer) has been quite accurate. In the end, I don't mind taking the inflated strava reading though

I understand this is a joke but it's essentially correct.

BTW it annoys me that Strava "trusts" Garmins to give good elevation data, but still overrides an iPhone 6 which has a built in barometric altimeter.

The fact is that when I ride with my friends who own Garmins, their climbing numbers can vary by up to +/- 500ft, so it's not like Garmin is magically accurate or anything. If Strava are going to correct my elevation without asking, they should do so for everyone else as well!

Elevation totals are always an approximation.
I just assume it's within 10% or so, and that's good enough.

Barometers
These can be affected by changes in the weather. I got 200 feet of elevation gain at a rest stop when a tiny thunderstorm rolled through (lowering the air pressure). And it can be affected by the bike wind speed.

But I can walk upstairs with my Garmin and watch it count off elevation feet as I climb. Just think of the tiny difference in air pressure in a couple of stair steps! That's amazing to me.

EDIT -- I've been on the Blue Ridge Parkway, and the overlooks have signs showing the elevation there. I'm often within 25 feet or less of the number. That's very good.

GPS elevation
It's fast for GPS devices to locate your spot on the ground, since the satellites are located at lots of different directions. It calculates the intersection of circles (spheres) from each satellite, and the satelllite position is known to high precision.
But elevation takes longer to get an accurate number. On my old Garmin 705, it has an "averaging" option when saving a location, and I can see the elevation number converge to an accurate value, but it takes 30 seconds to a minute, averaging in new readings ever second. On my newer phone, the GPS elevation calculation is faster, more like 10 to 20 seconds. But these are too slow to do continuously while riding.

Using mapped elevations
There are known elevation points on the maps, and the roads nearby are calculated from these points. If the road is heading up on the side of a mountain, the software needs to estimate exactly where the road is, and can be off a few feet. Bridges often show as a dip down to the river level and back up again.

Rounding and smoothing
It's a complicated problem. Here's the ridewithgps programmer, Cullen King's blog, on elevation calculations. Lots of math to correct the raw data.

I posted a comment about smoothing my Garmin 705 recordings in this post. It includes some graphs showing the smoothing changes.

Years ago, back when I still used mapmyride, they ignored any small elevation of 10 meters/33 feet. I can see the argument for this, but it makes a lot of rides with small rolling hills show up very flat.

Yes, I have a barometer app on my iphone and if I run it, it tells me changes in altitude, I can literally lift the phone 2 feet of my desk and it will measure a 2 feet elevation change. Pretty amazing. Obviously changes in atmospheric pressure screw them up, but other than that they should be accurate for deltas, if not absolute measurement (apps will rely on either GPS for an absolute value of elevation, or perhaps map data, neither of which are hugely accurate).

But given that I'm utterly mystified as to why multiple Garmin devices will almost always read completely different elevation changes on group rides, where everyone is riding the same speed and the same route. It seems that although individual units appear to be accurate, when you look at the data from multiple devices, what you really see are large errors in the measurement of climbing distance.

I can only go by my own experience but: when I did Washington Pass from Mazama the actual elevation gain according to USGS topo maps is 3,374 feet and my Garmin recorded 3,343. The next day I did Cutthroat Pass which is officially 2,300 feet and my Garmin recorded it as 2,313. I can keep going with examples, it's remarkably accurate. The displayed elevation changes as I walk up and down the stairs. Of course this is with a watch, not with an Edge. I have no idea what my Edge recorded, but in general it tends to be pretty similar.

I'm guessing the the different Garmins use different methods to smooth. It would be interesting to compare the same models on the same ride to check the accuracy.

Lots of riders complain about Garmin's buggy software. For instance, on my 705, if I stop at the side of the road while climbing, and using my wheel sensor so the 705 knows I've stopped, it starts counting up the grade: 5%...6%...7%...8%... What kind of strange grade calculation would work this way!

No idea about watches, but look at these numbers reported by Strava for a recent 24 mile group ride I did:

Me: 1,308ft (Strava) / 1431ft (Cyclemeter using iPhone Barometer)
Rider 1: 1,663ft (Garmin Edge 510)
Rider 2: 1,677ft (Garmin Edge 810)
Rider 3: 1,903ft (Garmin Edge 800)

So you can see that although 2 of the Garmin guys were close, the third got over 230ft of climbing extra recorded. That and all the Garmin riders got over 300ft more climbing than I did on Strava, by benefit of using a Garmin! I suppose if you were to take a positive here, 2 of the Garmin users were pretty close, so it's possible only one Garmin was anomalous, but Strava trusts it anyway!

My conclusion is that if I want to improve my climbing stats, I need to buy a Garmin.

Now I'm curious. I always ride with two Garmins except when I rent a bike on vacation, let's see how they compare. I keep the watch's altimeter calibrated but haven't done that to the Edge in years.

Washington Pass ride
Fenix 3 = 3,435 feet
Edge 800 = 3,399 feet

Renton loop
Fenix 3 = 1,821 feet
Edge 800 = 1,732 feet

Suiattle River Road
Fenix 3 = 2,562 feet
Edge 800 = 1,696 feet

Overall it's pretty good, but there's a massive discrepancy on the last one. Which was on a day that started off sunny but got more and more cloudy until it rained. I wasn't using the continuous re-calibration from GPS option, but the watch probably has better software in other ways to correct for atmospheric changes.

Which value is correct? That's pretty important to know. You're assuming the Garmins are wrong, but without any better information I'd leave it up to a vote and say the true elevation from that ride was most likely around 1,650 feet and Strava routinely cheats people out of the pride of their honest work.

I'm not assuming Garmin is wrong at all, but without an accurate measurement of the real elevation change, I have no evidence to suggest that the two Garmins that are consistent are right, only that they are consistent. That may be random chance. If I pick a different ride (this was just my most recent group ride), I'll find often no consistent values.

About the only thing I can conclude, and I'll admit this is still somewhat anecdotal, is that Strava's corrected elevations seem to consistently be lower than Garmin elevation calculations.

I've also found that if you're riding at sea level the elevation is absolutely worthless. I'm in south Florida and my rides are way off, giving me 1500 feet of climbing on absolutely pan flat roads and even when temp and pressure stay constant I get several hundred feet of climbing. I'll be interested to see how it works in the mountains later this month.

It would be great to have lots of data on this and be able to tease patterns out. Ironically, you know who was all that data? Garmin and Strava. They have that flyover thing, who did the same ride as you, it would be great to not only tell you your elevation gain for the ride but the average among everyone else who did the same ride and then how that breaks out by device type. Of course, most people wouldn't care and it would not be worth the cost to develop.

Hmm.. Seems Strava themselves would have the best, or at least the most extensive database of reported elevations for any given segment of road, than any other source. I find it curious they revert to the USGS for the data.