Yesterday did a 109 miles ride and when reviewing the ride info from the Garmin 705 computer I do see discrepancies in elevation gain.

WKO+ : 9266 ft.
Traningpeaks.com : 8999 ft.
Garmin Connect : 9745 ft.

Everything else is pretty much correct, distance, time etc.

Is Garmin website correct? Why would trainingpeaks.com or WKO+ was so much different?

Did you try enabling the elevation corrections on GarminConnect? (bottom left under "additional information")

It supposedly references known elevation points on or near your route instead of basing it solely on barometric pressure. It usually changes quite a bit when I enable/disable.

There is no universally correct way to determine elevation gain. It's a little like measuring the length of the coastline. The finer the resolution use the longer the coast or the bigger the elevation gain. I suspect each of the programs applies a different amount of smoothing.

Just tried the elevation correction and got 9,433 ft. Little bit lower than before but still about 5% off the trainingpeaks.com.

BTW I'm not complaining at all, I was just surprised to see those differences as I never really paid any attention to that

I posted this photo and some comments in the thread Is there a problem with my Garmin 705. The green graph is my 705's recorded data, and the purple line is the calculated elevation at each point, using the SRTM satellite survey data. The calculated elevation depends on exactly where the road bed falls between known elevation points on the side of the mountain.

Most differences here are small, less than 50 feet, but all the tiny bumps add up to a substantial elevation difference. The maps tend to have a somewhat higher total ride elevation for me, the opposite of your example ride. A large change in air pressure during the ride would affect the elevation, too.

I wonder if WKO or Training Peaks use the actual data or mapped data, and if they apply any smoothing to it.

IMO, your best number is the one recorded on your Garmin (unless you're using 1 sec recording - then it's too high). Unless the weather is variable with dramatic changes in barometric pressure. All the data sites online use different methods of data smoothing which results in different results.

why do you say that?

j

At some point you just have to embrace the ambiguity and take the numbers for what they are. For me I just pay attention to the same reading each time - normally the barometric altimeter on my 305. That having been said, at the end of a ride I did with a friend yesterday the barometric altimeter on my 305 had registered 4200ft elevation while my friends' 705 (also with a barometric altimeter) read 3800ft, I'd looped back once for maybe 50ft more gain, but other than that we'd ridden the exact same route.

Measuring vertical distance gained is kinda like measuring a coastline. It's fractal. The closer in you get, the bigger the numbers. For cycling, 1-second recording has too fine a granularity to get a reasonable number.

It's really not possible to get an exact number. You can get a solid estimate.

FWIW, I did a route today that I've done a number of times. I get elevation numbers on my 500 anywhere from 840 ft to 990 ft. The slower I ride up the hill, the bigger the number I get.

I use one sec recording on my 705 and I consistently get 10-15% lower ascent nos. than what Strava or Garmin Connect estimate. I start most rides from my home so I correct for the change in barometric pressure when I get home. I believe the 705 far more than I believe the online estimates.

I just did a ride today - I have an Edge 705 and my friend has the 800. Her elevation number was 10% higher than mine.

I just use the same source to compare between rides, and figure it's not exactly right, but it does give me a way to compare different rides that I've done. I look at the number in SportTracks, since that's where I've been downloading my data for the last few years.