I generally keep track of my rides using a cycling app. I have noticed that the reporting of the elevation gain statistic by the apps is wildly variable. For example, here's what the app told me for the last 10 times I did my frequent 18-mile ride:

Elevation gain (in feet):
866
1287
1476
1297
1296
1141
1460
1134
1251
1224

These are the same ride along the same path on different days. These were tracked with the Samsung Health app in my Android phone, but I've gotten similar inconsistency with other apps (including Strava and Map My Ride) and other phones (including an iPhone.) How do the apps keep track of elevation? Do they use GPS data (which I think would be very accurate) or barometric pressure (probably less accurate)? Has anyone else noticed anything like this?

That’s interesting that it’s so far off.

I use a Garmin 235 watch, it has occasionally shown me double the gain. I know it will do this this so I don’t pay much attention to it. It works entirely from gps data.

After the ride, I sync to Strava and the data seems to get averaged based on their databases. I’ve always accepted this as accurate enough for my purposes.

I think that using wheel magnets for speed and barometer for altitude has been done since the early 90’s (Avocet 50). The technology still exists and is probably available now wirelessly.

I personally like having a single watch since I have 4 bikes. To me that’s worth a slight decrease in accuracy.

I use the mapping feature on route planner on ridewithgps and find it consistent over Garman Connect which IMO overinflates elevation gained on my Vivoactive watch.

Remove the outliers and take a mean if you want a generalized measure of your elevation gained.

I have similar issues between my garmin watch and lezyne GPS. A quick Google search says that phone are only accurate to 4.9 meters. If that's true maybe a bouncing GPS signal would account for the difference.

Your average is 1274 feet ± 174 feet. From what I can see, the accuracy of phone elevation gain is ±30 m (98 feet). Your variance is a bit higher than that but not significantly more. I suspect your phone is using a barometer which would cause more error.

Phones typically use GPS to determine elevation, while many bike-specific devices (like the Garmin Edge series) use barometric pressure.

GPS is terribly inaccurate at measuring elevation. If you are using a phone to record a ride, that easily explains the wide variation in elevation gain.

Barometric pressure is much more accurate. But errors can creep in there, also. If the pressure rises or falls during a ride, the device will record it as phantom elevation loss or gain.

There’s “yabut” in there. GPS is slower in measuring elevation so it varies more when moving. Barometric pressure altitude measurements are quicker but they are influenced by more factors…temperature and pressure, as you pointed out…which can lead to errors. Both have about the same amount of error in measurement. If the phone is depending on the GPS signal for altitude, the errors might propagate faster because of the slow measurement

Re: GPS is slower in measuring elevation so it varies more when moving

I am not familiar with this "slowness in measurement" error with GPS.

Re: Both have about the same amount of error in measurement.

That's not my understanding. GPS errors are typically considerably larger, because they exist and are roughly the same on every measurement. Barometric errors creep in as the pressure drifts over several hours, which is why it's a good idea to re-calibrate elevation periodically at a known position.

Atmospheric pressure isn't constant. It's always changing. A change of 1" or Hg (mercury) will equal 1000 feet of elevation. So even on a short ride, it's reasonable that the barometer might have changed a few tenths of an inch, so 200, 300, 400 feet of difference will be in the ballpark.

Elevation gain is just a gee-whiz number, don't get caught up in it. If you want to climb good, then climb a lot.

If that's the case, I wonder if the accuracy would vary on different days as the barometric pressure varies. I also wonder if any apps or devices use topographical maps along with GPS data -- I expect that would be significantly more accurate.

The Garmin Connect website (connect.garmin.com) has an "elevations correction" feature you can use after the fact to correct the errors in recorded elevation data.

From my most recent ride (82.23 mi), elevation gain:

Without elevation corrections (barometric elevation): 6,726 ft
With elevation corrections (elevation from "professional surveys"): 8,089 ft

It's hard to know which of these is "correct". I was riding in hilly terrain, so any GPS inaccuracies could have put me up a steep slope or down a ravine.

I trust the barometric data more than these "professional surveys" corrections, and both of them over any GPS elevation data.

From Sunnto

A Google search of accuracy of the two measurements give a raw error of 10m to 20 m for GPS (35 to 70 feet) while barometers are accurate to ±15 meters. The error is almost the same per measurement. GPS may suffer more from a refresh rate that a barometer doesn’t experience. GPS is averaged over several seconds while barometric pressure may be closer to instantaneous as it is an on-board instrument. If you are stationary, this may no make much difference but when mobile, that averaging can lead to larger errors. Most phones have both GPS and a barometer to increase accuracy.

Airplanes use barometric pressure deltas for several different instruments. It’s more accurate than some here will have you believe.

It’s probably a combination of factors. Errors aren’t additive but generally multiplicative. Daily variances in satellite communication, refresh rate, where the measurement is made, etc can lead to errors propagating more in one measurement than another. For example, the 886 foot measurement could have been a case of the GPS missing measuring at the high points. The 1400 foot measurement may over poled the satellite at the high points more.

All that said, it’s not the big of a deal. Since you have multiple points of data, average, do a standard deviation, and accept the average.

An accuracy of ±15 meters seems too high to me for a barometric pressure sensor. Garmin reports ±10 feet accuracy for their barometric altimeter (exclusive of barometric drift).

Someone already said it earlier in this thread and it’s worth repeating. Unencrypted GPS data — in other words, GPS available to the public through commercial products — was initially only accurate to +\- 30 meters. GPS has gotten better over the years. Now, in many parts of the world you can expect much better GPS accuracy depending on how many satellites you are hitting and the GPS unit you are using. However, in some places there is no accuracy at all due to intervening terrain. GPS is line of sight. I’ve been in mountainous spots, tunnels, etc. with no GPS signal update. And rate of travel also affects received GPS data updates. Some apps are better than others, and some GPS units have better receivers and antennas than others. I use Strava on a series 3 Apple Watch matched to a 6S+ iPhone. My rides track pretty well with where I physically know I’m at. Elevation occasionally varies a bit on repeated rides but only by a few feet. If you are constantly getting wildly varying elevation deviations there is probably something in the locale or equipment or software or rate of travel that is the cause. For an example of how inaccurate otherwise seemingly reliable GPS can sometimes be, my iPhone “Find Me” app often places me at my neighbors house when I’m actually in my living room. Go figure😉 Go ride, have fun, and be okay with the GPS limitations you experience!

Pilots of aircraft also depend on getting the altimeter setting, which is a correction factor to allow their barometric instruments to show them an accurate altitude or clearance to terrain. Each air traffic controller will routinely give the altimeter setting for that area so the pilots know what to set on their altimeters.

There are various methods for GPS devices with a barometric sensor use to calibrate their reading to the actual elevation, but that initially takes effort from the user and only works for predetermined locations. And the GPS device AFAIK only calibrates itself at the beginning of a ride. So normal changes in barometric pressure will produce errors in what is actually done.

Almost any ride I do, the elevation I left my house at is not the shown elevation when I return.

I have always been curious about the Garmin elevation correction feature, which is switched off by default. My Edge 530 is very consistent with elevation gain. On my local loop it’s always within about 20m. On specific event routes where the elevation gain is known it always reports very close to the advertised figure. So I have never felt the need to switch on the correction feature.

This is my observation when using Garmin's elevation correction for mountainous switchback rides. On each turn the GPS drops me off the cliff and then climbs me back up adding 1000s of feet to the ride. It makes me look good but i switch this feature off. Regrettably it seems permanently activated when using the Connect route planning function, so I defer to RWGPS when planning a ride.

In the old days ... when we carried barometric altimeters (I had a pricey Thommen - actually, I still have it somewhere in the basement) when hiking and climbing you would always reset when you knew precisely where you were - trail intersection, lake, pass or summit. If on an overnight trip, you would check the "altitude' when arriving and then check it the next morning, resetting to match that prior reading. I miss my Avocet 50!!!!

You can manually calibrate Garmin units like this too. Not that I ever have!

If it bothers you use the elevation correction button on Strava to default to their map data and it should be consistent (if not on target). Then even if wrong it will be the same.

^^^This. I and many of my riding friends use Wahoo to record and our elevations can vary by hundreds of feet on a ride. The Strava elevation correction reduces the variance substantially. It doesn't matter to me which is correct as long as I have consistency. To further muddy the water, we use RWGPS to plan routes. We rarely end up with the predicted elevation. Sometimes more, sometimes less. I can't see how a few hundred feet would matter either way.

FWIW, GPS measurement of elevation is VERY poor compared to horizontal position. If your GPS device could read satellite data from satellites on the opposite side of the planet, then elevation measurements would be as good as horizontal ones. Barometric elevation measurements in aggregate are more accurate than GPS elevation measurements.

dave

The correction feature is disabled by default for devices that use a barometer. It's enabled for devices that only use GPS.

Garmin rates barometric data as better than GPS for elevation.

The manual calibration is useful for determining altitude (which few cyclists care about). It should not matter for elevation gain.