The calibration is necessary for determining altitude. But cyclists don't care too often about that.

How different is the elevation at the start and the end?

Drift in the reading (if it's slow) probably won't change the gain by much.

Small gains over large distances likely have a higher percentage of noise.

Having an accurate elevation is useful in one situation: when you know the elevation of the summit of your climb, and you want to know how much more climbing you have remaining.

​​​​​​Sure. But that's not a common interest. (I didn't say it wasn't ever useful.) It also wouldn't let you know how much more if there were inflection points.

You can also get that information from the relative elevation in the track file (if you are using one), which is how Garmin's "climbpro" works.

Accurate gain doesn't require accurate altitude. If the altitude is off by a constant amount over the course of a ride, the gain will still be the same.

GPS makes geometric assumptions (geodetic model) that contribute additional errors to altitude measurements. GPS wasn't really designed for elevation measurements.

Altitude Accuracy

If you are interested in gain, the measurements "on different days" won't really matter.

Strava and some other activity apps reference existing databases for elevation, based on map data and data submitted by users who post data based on their computers and phones. Presumably these are reasonably accurate. At least it offers a fairly consistent baseline.

My XOSS G+ bike computer has a barometric pressure sensor and the elevation estimates vary wildly depending on the weather. I never even look at it during rides.

And because my computer elevation data is unreliable, after syncing my ride to Strava, I then use Strava's options to correct elevation and distance. This at least makes the other data relevant in comparisons with rides on the same routes.

Unless it's hella windy, an accurate power meter, an accurate speedometer, and a pretty good bathroom scale work very well for calculating altitude gained.

My 1700m ride on Saturday turned into a 2400m ride in real life 😂

Those extra 700m are way harder when you don't know they're coming

Thanks, that explains why the correction is disabled. The barometric data seems very good on my Edge 530 and certainly not seeing the issue the OP is with his phone.

Yeah I like ClimbPro and as you say it doesn't need an altitude calibration to track your progress up a climb.

The weather here has been pretty stable. No fronts of areas of high and low pressure moving through. Just sometimes rain and a thunderstorm that develop and dissipate in pretty much the same general area. So my last 5 rides over 2 hours had a start finish difference averaging 48 feet. The biggest 62 feet and the smallest 38 feet. Not a lot. However some where back in my history I've seen on the order of 100 feet of change or more during a ride.

I agree the calibration is only needed for those that want to know their elevation. I've never understood the need to know actual elevation for cycling. My house is the normal start point for my rides so I have the driveway as a calibration waypoint. On the topo maps of my area a contour line goes right through it. But even that might be incorrect as the topo maps data set includes data from times before this house or even the subdivision was built. So who knows and why should I really care? I don't!

I'm not suggesting this is the reason for the OP. But this and other things can add up to always make the relative or actual elevation numbers be different from one ride to the next.

There is even an overpass I travel routinely and on days with the wind blowing steady from the soutwest, I'll show a higher elevation going over it than days with no wind. I wonder if that is a localized reduction in air pressure because the wind is being funneled along that area at a higher velocity than air elsewhere. So Bernoulli's Principle might be another reason for errors in elevation or gain/loss calcs. And it can occur in many places.

For me, elevation gain/loss is a lot like Calories. The actual exact number is not important. If someone wants to ride hilly terrain, then ride the routes that show the most gain loss per distance. Doesn't matter if they don't match up from day to day.

Map data isn't correct for total elevation gain, nor are our barometric sensors in our devices and certainly GPS calculate altitude is less trustworthy. So people need to learn to judge by the trends and not the actual exact number they will never have.

They are not. Strava still suffers from the inaccuracy of data that users upload.

Like this Strava segment, based on some very noisy user-uploaded data.

The actual grade is about 7%, but Strava says it's 12.3%. Strava says the KOM rider climbed it with a VAM of 2703 meters per hour, which is ridiculous.



Bad Strava segment

Here's what that segment would look like with "good" GPS data:



"good" Strava segment

Yes, any wind blowing past the sensor reduces the static pressure, and the elevation shown will be high.

I've noticed this on climbs with lots of switchbacks. When I'm climbing a portion with a tailwind, the climbing rate number is steady. But when I round a hairpin and turn into the wind, I'll briefly see a burst of increased climbing rate.

Alas, it's only a brief and false moment of glory.

2 hours is about 30 miles. I suspect the percent of error is larger for for smaller elevation gains (let's say < 1000 feet) than it is for larger numbers. a 100 foot difference is only 10% of 1000 feet. I wonder what you get for gain and loss.

There also isn't a "right" elevation gain number. It has to smooth the data (otherwise, it would count every little jiggle).

The OP's numbers are on the low side. And, except for two outliers, the numbers are pretty close. Is a 50 foot difference even meaningful?

Since elevation gain is not considering grade at all, it's a weak indicator of effort anyway.
Yes. Some people appear to expect the same level of accuracy for elevation gain as for horizontal distance. That doesn't seem realistic.

Hmmm. I think there's a "right" elevation gain number.
Some estimates of elevation gain take grade into account.

How long is a coastline?

There isn't one standard for smoothing the data and what changes to include or drop.

I had a Bontrager computer that was kinda close to my Edge 800.

Which ones and how? Is it any of the common ones?

One might be able to take the gain and divide it by distance. But that wouldn't be gain (it would be a gain rate or ratio). Or, I suppose, one could ignore gain below some sort of grade threshold.

Yes, one time Garmin had me riding up a climb underneath a freeway bridge that 's WAY overhead, then magically CROSSING that bridge before falling back down onto the road beneath.

Luckily, the fall didn't kill me.

I tried "starring" OLH so that I could track my progress relative to my previous best time. TWICE I started up the climb and less than half a mile into it my Garmin said I was "off segment" and stopped giving me those data. Similarly, one time when I rode up Kings Mountain,. between Garmin and Strava, I was off the road most of the time, so it didn't find segments for half the climb - so, also not for the whole climb.

Sounds like a good place for an "Everesting" record attempt.

i'll bet the power pod could accurately measure elevation if it only knew what your elevation was when you started.

my understanding is that it uses both grade and speed to help compute power. speed and grade are all that are needed to calculate elevation gain.

https://velocomp.com/powerpod-lite/

I was told there would be no math problems.

You're mixing up a distance with the measurement of that distance (or, in this case, a height with the measurement of that height). This isn't a fractal problem. Distances and heights are actual things, so there is a "right" height. The measurement of that height can and does have an error, but your claim was "There also isn't a 'right' elevation gain number." There is a right elevation gain number.

Not one of the common ones, but it's one that is built on the effort it takes to climb a hill. It uses power and speed data (among other things) to build up road gradient and then integrates gradient to get elevation gain. It can reliably and repeatably reproduce small changes in road grade. Better still, we can use the Central Limit Theorem to reduce error in the elevation gained when integrating the gradient.

My apologies. I once got a student course evaluation with: "You lied when you said we wouldn't need calculus for the class." So I didn't stop using calculus, but I did stop saying they didn't need it.

Will there be a refresher tutoring session for those of us who haven't used calculus in a few decades? Or do we need to process a withdrawal from BF?

I disagree, total elevation gain depends on how much resolution you want.

How do you define the "right" length of a coastline?