chadtrent

We did a ride yesterday on a rail-to-trial. There were four of us, and we all got four different elevation numbers when we were done. Here are the results:

1. Wahoo Element Roam (mine) - 548 feet
2. Garmin watch - 460 feet
3. Garmin Edge (510 I think) - 991 feet
4. Strava on android phone - 840 feet

Also when we map the ride on Ride with GPS it says 2048 feet. When I uploaded the route to Ride with GPS from my Wahoo app it says 742 feet.

What is causing such inconsistent readings?

Iride01 

Comparing your device to another is the main issue. <grin>

Do you know for certain your device has a barometric sensor? Even if it does, different ones may have a different resolution and might not pick up a change of only a couple feet where others might. Also, different models/makes might uses differing algorithims as to when and what to add to the accumulations. And also for barometric sensors, most need a static reference and blockage of the static port or wind currents hitting it at times can create "noise" that may or may not cause a value to be wrongly accumulated.

Deriving elevation gain from map info is fraught with problems too. The resolution of available elevation data in that area is probably a lot more than we'd like to believe and ones gps coordinate might put them off the side of a cliff where another is on the up-slope side.

chadtrent

I have no idea how it determines elevation. So given that they do it in different ways, which is right? This ride wasn't a climbing route so not really a big deal. But it would be nice to know.

Tourist in MSN

In general, GPS units are not very accurate at determining elevation, elevation error often is double what horizontal error is. And when you are accumulating elevation data over time during your ride, as your elevation error fluctuates, that is cumulative and adds up.

I do not know if those devices have WAAS or not, that probably cuts the error in half or better. On my GPS that I use for my bike, I have WAAS correction turned off to save battery life, it is not important for me to know if I am 40 feet from something vs 15 feet from it, so I put battery life at a premium.

When you first turn on a GPS unit, it takes some time to download the latest orbit information from the satellites. Not sure how long that takes if you are in a clear view of the sky, maybe as fast as 10 minutes but I generally assume under most conditions it might take as long as 30 minutes.

And I have no clue how the math works for using the GPS data to correct and calibrate the atmospheric pressure data if you have that sensor.

Bottom line, I am not really surprised at all about your differences in readings. You are not using surveyor grade instruments with differential correction.

79pmooney

I had fun last summer riding past the 1272' elevation sign and my Elemnt Bolt showing an elevation several hundred feet different. (No radical dropoffs there. Should have been within 50')

I'm waiting for the day when reslution gets really good. THen we will see the pro riders doing thousands of feet of climbing riding the dead flat Paris-Roubaix with its nasty cobbles. An inch of bounce hundreds/thousands of times.

Ben

adamrice 

There are a number of apps and services that will correct your elevation after the fact: you upload your GPX file, the app looks up the known elevation for points on your route, and updates the GPX file. It looks like RideWithGPS and Strava will both do this. I used to have a janky Java app that did it, but can't find it now.

Seattle Forrest

Rail to trail means railroad grade which means not very steep. It will take a lot of miles to accumulate much elevation gain. Typically, there are always exceptions to everything.

You get so many different answers because there isn't a perfect way to measure. Everything you checked is an estimate, and they're all estimated different ways.
​​​​​
​​​​​​Ride with GPS is significant more than any of the rest and probably wrong. Did you go over many bridges? One thing that can happen when you use a digital elevation model (kind of map) is the database doesn't always know there's a bridge where you rode, so in those cases it will think you went all the way down the ravine and back up.

In general barometric altimeters are the best way to measure to in a bike. But changes in air pressure (which often proceed weather changes) are their weakness.

GPS isn't great at elevation, it's always close but almost never right, and jumps up and down a lot because of the way it works. It's especially bad under a forest canopy or with buildings or cliffs nearby.

I have a Garmin watch with a barometric altimeter, and it's mostly been excellent for elevation data. A lot of our better roads here go over mountain passes or to summit lookouts, all of these things with known elevation. When I hike I use topographic maps. Both of these offer a way to check the device and get an idea how trustworthy it tends to be, and the results tend to be excellent.

Some watches and computers use both a barometer and GPS, with GPS recalibrating the barometer regularly because of drift. I think newer Garmin watches default to this. It might be more accurate over long activities (backpacking trip, trail ultra marathon) but can introduce jitter which is more pronounced over shorter ones.

It's interesting the way there are two sets of fairly similar numbers. Did everybody start and finish the ride together?

chadtrent

Yes, there were probably 20 bridges on the route. Some close to 100 feet high. We did all start and finish together. The only time we were separate was for a few minutes in the middle of the ride when I went about 1/4 mile up to a store to get water and everyone else stayed together. Very little elevation change going to the store though.

canklecat

Strava doesn't share much info about how it handles data but it's pretty clear they have their own methods for evaluating and changing data from other apps, phones and computers. There's always some minor discrepancy between the data I get from my phones with Wahoo Fitness, Cyclemeter, etc., or my bike computer, and what Strava shows.

That includes when I run Strava from my phone and my bike computer simultaneously. Direct-to-Strava logs always show me a bit slower, and with minor differences in elevation, distance, etc. Not enough to worry about, unless a KOM is at stake -- which, with my mediocre speed, is rarely an issue. At most I'm shooting for a PR or top ten, but I'm never in serious contention for any KOM worth having.

My 28 mile ride Sunday shows:
XOSS G+ computer/app: 1,445 feet elevation gain; 27.5 miles; moving time 1:40:54; 16.3 mph
Strava interpretation of that data: 1,346 feet; 27.36 miles; moving time 1:41:14; 16.2 mph

Direct to Strava log showed I averaged 15.4 mph. I deleted that backup log and don't remember the rest, but there were differences in every category.

One primary source of differences is how each app or device handles the auto-pause/resume threshold. This makes a difference in average speed, moving time, distance, etc.

My computer has a barometric altimeter and I haven't compared data enough to see any significant differences, but there are always minor differences between Strava, devices and apps.

tomtomtom123

Locus map uses both barometric sensor and elevation maps to optimize the GPS. It doesn't explain how it calculates though. It says the live tracking will display the raw GPS and then the adjustment calculation is made after the recording ends.
https://www.locusmap.eu/everything-y...afraid-to-ask/
It also doesn't display whether it's reading the barometer at the moment or not so you don't know whether it's helping to optimize the GPS reading. It does warn that some phones especially Samsung will turn off the barometer on battery saving mode, I guess when the screen turns off. On my phone from a different manufacture, after the Android 9 upgrade the GPS would switch off when the screen turned off so I had to add location services to battery saver exceptions in the settings.

Amt0571

My Etrex Touch 35 (and previously my Dakota 20) are quite accurate. They consistently show the same altitude at the same places, and elevation gain is usually within 25m on rides I usually do.

To get a decent accuracy you need (on a Garmin device) to do a couple of things:

- Enable the automatic continuous calibration of the altimeter: this periodically recalibrates de altimeter using the GPS signal when it determines the signal is strong enough, negating any atmospheric pressure changes that may affect the elevation calculation. IMHO, it works great, as it allows to use the barometric altimeter for precise calculations, while periodically recalibrating with the GPS signal to avoid large errors.

- Manually calibrate the altimeter before starting to record a ride. You can input the elevation manually (if you know it), or you can tell it to accept the GPS data, which should be good if you have good signal and wait some time to let it stabilize.

I suppose Wahoo devices have similar settings, but have never used them.

pdlamb

To quote Mark Twain, "A man with one watch knows what time it is. A man with two watches is never sure." Given five guesses from five different devices, I'd discard the high and the low and average the others.

Steve B.

GPS alone is notoriously bad at determining elevation. Has to do with the method of calculating on a sphere that is not perfectly round (the Earth) as well as huge inconsistencies in the digital maps used to determine where you are located on that sphere. Device manufacturers try to compensate with a barometric reading, but that can have errors due to temperature variations, weather patterns, etc....
Some folks find their device is consistent at generating the same elevation and gains/losses over a regular route, but is it accurate to begin with ?.

Add in 4 different devices with 4 different measuring systems and no wonder you get the results you've seen. I pay no attention to what my 1030 tells me is my climbing for the ride.

If you have any interest in what other folks find with the elevation problem, there are pages and pages of comments.

https://www.google.com/search?q=garm...hrome&ie=UTF-8

DrIsotope

Wahoo Elemnt Bolt, internal barometric altimeter, vs. the Strava estimate of total route elevation for my past five rides:

Bolt: 1,434ft; Strava 1,445ft (-0.7%)
Bolt: 2,238ft; Strava 2,306ft (-3.0%)
Bolt: 1,129ft; Strava 1,171ft (-3.6%)
Bolt: 2,057ft; Strava 2,056ft (-0.0%)
Bolt: 1,427ft; Strava 1,400ft (+2.0%)

I can't find recent files where I covered the exact route multiple times-- I routinely make course changes based on everything from traffic to stop lights to the prevailing wind.

I did look up some very short neighborhood rides with my wife, to compare the Bolt to a few-years-aged Edge 520, and our respective elevations have never differed by more than 2-3%. The stinger is that she always gets more. Always.

I think most consumer devices equipped with barometric altimeters trend toward high accuracy/low precision-- two units can agree while another two will disagree, but we will never know which one is actually correct. So long as results are repeatable, good enough for me.

GPS alone for elevation is most certainly low accuracy/low precision.

Iride01 

If you really have to know the exact number of feet you gained or lost, then you will be disappointed. It really can't be done, Take for instance those guys with a watch, do they get another foot or so added and subtracted to their gain/loss every time the wave at someone?<grin> (This is humor I know that you probably don't wave enough to make a big difference). Logging data is problematic too, if you are going 20 mph then you are moving almost 30 feet per second and if you speed at 30 mph, then you are 44 fps. Even with one second recording your log data might miss the top and bottom of a hill but another's device might have gotten the top and bottom logged. But is what's in the tog the same as the accumulated totals on the device. I don't know.... the mfr doesn't tell us this. But on my edge 500 I suspect not.

Also when you look at any site that's deriving the gain loss by summing the data points, it may look quite a bit off from what your devices accumulation is. How those websites use and smooth our data as well as the accumulated totals is not easy to find out.

However for using the number your device gave you to decide about riding this route or that route, then you'll know what route will give you more climbing than another you rode. So over time you can make some valid use of the info. So if you are wanting to work on climbing, then do the routes that your device says gives you more. You really shouldn't care that another's device gives them a different number. It's not what any device can do when you compare them, it's what YOU can do when compared to the riders of those other devices.

JohnJ80

Trying to get accurate elevation gain out of a bike computer, cell phone or fitness watch is an exercise in futility. I've even had gradient changes jump or drop when a semi truck drove past at highway speed. Grade/gradient also uses the barometric sensor.

GPS, even running at it's best accuracy, is most inaccurate in vertical altitude measurements as a function of the overall GPS system design.

Treat vertical ascent as a figure of merit and recognize you'll never get the stuff to match exactly.

J.

ChrisAlbertson

GPS always has a lot of trouble with the altitude. It is never very accurate (except on marine GPS units where yu fix the elevation to sea level.) The reason is just the way the satellite geometry works out.

If you really want to know elevation you have to look at a USGS topographic map. These were all checked by ground survey ad are accurate. The web site "Ride with GPS" will lot your rout on a USGS topo map and you can figure out where you were. You can also download a digital (PDF) version of these maps for free from USGS.

THose built-in altimeters are very sensitive. I have some engineering sample chips here and I can read raw data from them and notice when I move a foot up or down but they have ZERO ideas of their absolute altitude above sea level.

A good computer should combine GPS and barometric data.

Using the topo map is not so bad because most of us tend to ride the same routes over and over so you only have to look it up once.

Seattle Forrest

The last time I rode Washington Pass my barometer watch had a total error of about 25 feet (vs USGS topos) out of more than 3,400. This is pretty typical except when the pressure changes either due to a weather system or crossing a significant divide. I don't agree that the situation is hopeless, at least in good conditions.

njkayaker

Over what distance?

1000 feet over 100 miles is likely close to noise.

I suspect the numbers are more accurate with greater elevation changes over shorter distances.

It's an estimate. Being within +- 20% might be what's reasonable to expect. Or more for small numbers (~600 feet over any real distance is small).

RWGPS is using surface elevation not road elevation. It seems one should expect measured elevation gain for a rail trail would be less than the gain determined from topo data. In other cases, topo data might miss smaller bumps, producing a lower value than what is measured with a barometer.

Some people expect the same level of relative accuracy/precision for gain as one can horizontal distance but it's almost certainly much worse.

Part of the problem is that people care about small differences in gain that they wouldn't notice for distance. A 1000 foot difference in gain is significant. No one would really care about a 1000 foot difference in ride distance.

​​​As people mentioned, GPS is poor for determining gain. Barometric data is considered to be the best.

There also is no one correct way of determining gain. Different devices likely have differences in what they consider a real change (differences in the methods they use to calculate the estimate). That's not the case with measuring absolute elevation.

It's an estimate.

Gain is a poor estimate of effort (what people are generally interested in) anyway since it doesn't take grade into account.

njkayaker

​​​​​​This would only be useful (for gain) you were also using the automatic continuous calibration (which might make doing the manual calibration superfluous).

Gain is measuring a difference and the error manual calibration removes won't change that.

The calibration matters if you are interested in absolute elevation (which most cyclists don't generally care about).

njkayaker

This situation is much more favorable for getting accurate numbers.

Larger gain over shorter distances is going to be more accurate. Even moreso if the gain is continuous. It's also the case where differences in the methods used by different devices matters less. It moves closer to being the simple (and accurate) difference in elevation between the start and finish.

500-1000 feet over "long" distances is going to be less accurate.

njkayaker

Garmin provides this correction by default for devices that don't use barometers. That is, Garmin believes the barometric data is better.