Bike Forums

Bike Forums (https://www.bikeforums.net/forum.php)
-   General Cycling Discussion (https://www.bikeforums.net/general-cycling-discussion/)
-   -   Cellular elevation gain accuracy (https://www.bikeforums.net/general-cycling-discussion/1249588-cellular-elevation-gain-accuracy.html)

Moe Zhoost 04-04-22 05:36 PM

Cellular elevation gain accuracy
 
I've been using the MapMyRide app to monitor my rides and find that its distance stats match well with my cyclometer. I've never paid much attention to the elevation stats, though. Well today I encountered something most curious that has me wondering whether they are at all accurate. I took advantage of a nice spring day today to go for a row in the new skiff . I wanted to get an idea of my distance and speed so I started the app. After a rather nice row I took a look at the stats and found that I did a total elevation gain of 152 feet with the highest elevation 60 feet above the lowest. What's the deal with this? I figure I surely would have noticed that 50 foot rise and drop in less than a mile. The lake seemed dead flat to me, but maybe I'm missing something.

BkSaGo 04-04-22 06:29 PM

MapMyRide uses your cellphone GPS for navigation.

GPS vertical position determination is less accurate than horizontal positions, so you are are going to have more error in the vertical dimension and therefore in altitude gain/loss than in position and velocity. Also, GPS measures altitude relative to a “reference ellipsoid” rather than the actual earth surface - although over a mile than won’t make much of a difference in vertical gain/loss.

gpburdell 04-05-22 07:17 AM

As mentioned, GPS elevation measurements tend to be less accurate - with the satellites generally overhead the Z accuracy is less than the X & Y accuracy, and thus change-in-elevation figures can be wonky. Some apps will "correct" the elevation using your position (X &Y) on the basemap. See https://support.strava.com/hc/en-us/...on-Strava-FAQs

Cycling computers these days typically have a barometric altimeter which does a far better job of measuring changes in elevation.

njkayaker 04-05-22 08:31 AM


Originally Posted by Moe Zhoost (Post 22461857)
After a rather nice row I took a look at the stats and found that I did a total elevation gain of 152 feet with the highest elevation 60 feet above the lowest. What's the deal with this? I figure I surely would have noticed that 50 foot rise and drop in less than a mile. The lake seemed dead flat to me, but maybe I'm missing something.

What phone?

These numbers are likely close to noise. You really need steeper and longer slopes to get out of the noise.

GPS is not appropriate for elevation gain. Garmin treats barometric data as better (but even that's not perfect). Your phone might have a barometer and the gain might be using that.

People often think that elevation data is better than it is in reality. It could easily be +20% at larger numbers (even more when the ride is flat). That is, it's going to be more accurate for hillier rides. It's best to treat gain as an estimate.

Altair 4 04-05-22 09:01 AM

For a real giggle, ride through a downtown section of a city with tall buildings. I ride through downtown Pittsburgh and MapMyRide will show crazy elevation changes. Obviously I must be riding on the rooftops of the highrises!

70sSanO 04-05-22 09:37 AM

I’ve been using an iwatch to measure my elevation gain on rides. My wife and I will sometimes use it on short hikes.

Basically it shows elevation gain and also shows the min and max elevation. Subtracting the min from max never gives the same elevation gain, it is always greater.

I’m not certain of the algorithm used, but I always figured there has to be some real time recording and not just absolutes.

For instance if you ride on a flat stretch with 10 underpasses, or even overpasses, how are they recorded. I’ve never checked for these? Technically the elevation gain is zero for the overpass/underpass.

John

_ForceD_ 04-05-22 11:00 AM

I think that irregularities like this are due to glitches in the proprietary software of the fitness app. Here’s why I think that. Attached is the screenshot of the elevation profile from a run I did in 2019 along the beach at a resort in Punta Cana, D.R. And I mean I jogged right along the water’s edge so there were no hills. But, for the 4-mile jog along the water there’s a total of 1355 feet of elevation. You’ll notice that at a couple places on the profile it goes from ZERO feet elevation way down to more than negative 900 feet in the space of just a couple hundred yards. I did this particular run to see if it would mimic the same irregularity from the previous year…that unfortunately I didn’t notice until we’d returned home from the vacation. Sure enough…it made the same exact elevation plunge in the same exact place as it did the previous year. And I can assure you there wasn’t a “Grand Canyon” there on the beaches of the D.R. that I fell into. This activity was recorded with my iPhone using the Strava app. When I made this investigative run the second year, I also wore a Garmin GPS watch. When I looked at the elevation profile on the Garmin…there were no erratic elevation plunges, just normal elevation for a jog along the beach. So either there was/is some glitch in the Strava software, or software that Strava uses…or there is some sort of electromagnetic disturbance right at that particular spot that causes the cell phone’s GPS reception to go haywire. But then again it’s probably NOT that because ‘GPS reception is GPS reception’ regardless of the device. So it must be in a software algorithm.
FWIW…I have a friend currently living in the Pensacola, FL area. On one of his regular cycling routes he’s experiencing a similar occurrence. He’s riding along road adjacent to the waterfront. No bridges or overpasses. But at one point (same place every time) it shows him going up/down a hill of about 250 feet in the space of less than 100 yards.— Dan

https://cimg9.ibsrv.net/gimg/bikefor...9fefdda2d.jpeg

gpburdell 04-05-22 11:10 AM


Originally Posted by 70sSanO (Post 22462418)
I’ve been using an iwatch to measure my elevation gain on rides. My wife and I will sometimes use it on short hikes.

Basically it shows elevation gain and also shows the min and max elevation. Subtracting the min from max never gives the same elevation gain, it is always greater.

I’m not certain of the algorithm used, but I always figured there has to be some real time recording and not just absolutes.

For instance if you ride on a flat stretch with 10 underpasses, or even overpasses, how are they recorded. I’ve never checked for these? Technically the elevation gain is zero for the overpass/underpass.

John

It's collecting the info in real time, though likely smoothed.

The elevation gain represents the sum total of how much elevation you've climbed during the ride. Without subtracting descents.

So if you ride a flat route and climb over a 50' tall bridge three times it'd report 150ft elevation gain (assuming 100% accurate readings, which doesn't occur)

70sSanO 04-05-22 11:18 AM


Originally Posted by gpburdell (Post 22462532)
It's collecting the info in real time, though likely smoothed.

The elevation gain represents the sum total of how much elevation you've climbed during the ride. Without subtracting descents.

So if you ride a flat route and climb over a 50' tall bridge three times it'd report 150ft elevation gain (assuming 100% accurate readings, which doesn't occur)

Thanks. Then I guess the min and max would show 50’ difference, assuming a zero elevation gain on the flats.

John

njkayaker 04-05-22 11:22 AM


Originally Posted by 70sSanO (Post 22462418)
Basically it shows elevation gain and also shows the min and max elevation. Subtracting the min from max never gives the same elevation gain, it is always greater.

The difference between the lowest and highest absolute elevation is almost always going to be less than the elevation gain.

You could, for example, be up-and-down the same hill multiple times.


Originally Posted by 70sSanO (Post 22462418)
For instance if you ride on a flat stretch with 10 underpasses, or even overpasses, how are they recorded. I’ve never checked for these? Technically the elevation gain is zero for the overpass/underpass.

There are three ways to determine the elevation along your path: barometer, GPS, using a database (DEM).

Barometer is the best, GPS is inaccurate (and doesn't work inside/undercover), and DEM often doesn't include the elevation of bridges or tunnels and has some other issues as well.

Your iWatch should have a barometer.

https://slate.com/technology/2015/03...r-station.html



Originally Posted by 70sSanO (Post 22462541)
Thanks. Then I guess the min and max would show 50’ difference, assuming a zero elevation gain on the flats.

The difference between the lowest and highest elevation just uses two points. The elevation gain is looking at the elevation of every point.

zandoval 04-05-22 11:32 AM

Here's another one. Up in Northern Alaska the elevation points are accurate but the distances are off... Go Figure.

All in all, for a free app, Map My Ride has been great...

livedarklions 04-05-22 11:37 AM


Originally Posted by Moe Zhoost (Post 22461857)
I've been using the MapMyRide app to monitor my rides and find that its distance stats match well with my cyclometer. I've never paid much attention to the elevation stats, though. Well today I encountered something most curious that has me wondering whether they are at all accurate. I took advantage of a nice spring day today to go for a row in the new skiff . I wanted to get an idea of my distance and speed so I started the app. After a rather nice row I took a look at the stats and found that I did a total elevation gain of 152 feet with the highest elevation 60 feet above the lowest. What's the deal with this? I figure I surely would have noticed that 50 foot rise and drop in less than a mile. The lake seemed dead flat to me, but maybe I'm missing something.


Were you row, row, rowing your boat, gently up the lake?

livedarklions 04-05-22 11:41 AM


Originally Posted by _ForceD_ (Post 22462521)
I think that irregularities like this are due to glitches in the proprietary software of the fitness app. Here’s why I think that. Attached is the screenshot of the elevation profile from a run I did in 2019 along the beach at a resort in Punta Cana, D.R. And I mean I jogged right along the water’s edge so there were no hills. But, for the 4-mile jog along the water there’s a total of 1355 feet of elevation. You’ll notice that at a couple places on the profile it goes from ZERO feet elevation way down to more than negative 900 feet in the space of just a couple hundred yards. I did this particular run to see if it would mimic the same irregularity from the previous year…that unfortunately I didn’t notice until we’d returned home from the vacation. Sure enough…it made the same exact elevation plunge in the same exact place as it did the previous year. And I can assure you there wasn’t a “Grand Canyon” there on the beaches of the D.R. that I fell into. This activity was recorded with my iPhone using the Strava app. When I made this investigative run the second year, I also wore a Garmin GPS watch. When I looked at the elevation profile on the Garmin…there were no erratic elevation plunges, just normal elevation for a jog along the beach. So either there was/is some glitch in the Strava software, or software that Strava uses…or there is some sort of electromagnetic disturbance right at that particular spot that causes the cell phone’s GPS reception to go haywire. But then again it’s probably NOT that because ‘GPS reception is GPS reception’ regardless of the device. So it must be in a software algorithm.
FWIW…I have a friend currently living in the Pensacola, FL area. On one of his regular cycling routes he’s experiencing a similar occurrence. He’s riding along road adjacent to the waterfront. No bridges or overpasses. But at one point (same place every time) it shows him going up/down a hill of about 250 feet in the space of less than 100 yards.— Dan

https://cimg9.ibsrv.net/gimg/bikefor...9fefdda2d.jpeg


It's the holes where they store the Pokemon.

gpburdell 04-05-22 12:06 PM


Originally Posted by 70sSanO (Post 22462541)
Thanks. Then I guess the min and max would show 50’ difference, assuming a zero elevation gain on the flats.

John

Bingo. I did a ride yesterday with a reported minimum elevation of 708ft, max of 764 ft, and an total ascent (elevation gain) of 505ft. Overall fairly flat but enough 10 and 20ft rises to add up over the 20 miles.

And to emphasize that these figures are approximate - my reported total descent was 479ft, yet I didn't end up hovering 26 ft above my parked car. :foo:

Moe Zhoost 04-05-22 05:22 PM


Originally Posted by livedarklions (Post 22462559)
Were you row, row, rowing your boat, gently up the lake?

Apparently. Those aqueous hills are a bear, but the worst part is there is that I can't seem to coast going down.

Moe Zhoost 04-06-22 06:50 AM


Originally Posted by njkayaker (Post 22462347)
These numbers are likely close to noise. You really need steeper and longer slopes to get out of the noise.

Well, it turns out that it's not noise at all. I reviewed my rowing track to check where the elevation changes were occurring and discovered that the highest elevations were close to shore. Apparently the elevation data are referenced to the lake bottom, not the actual surface.

njkayaker 04-06-22 10:42 AM


Originally Posted by Moe Zhoost (Post 22463318)
Well, it turns out that it's not noise at all. I reviewed my rowing track to check where the elevation changes were occurring and discovered that the highest elevations were close to shore. Apparently the elevation data are referenced to the lake bottom, not the actual surface.

It's not noise because the device isn't measuring your elevation. It's using a database (DEM), which, as I said, has issues and not ideal for elevation gain.

The issue might have been more apparent if the complete data (horizontal location and elevation) was provided.

livedarklions 04-06-22 01:22 PM


Originally Posted by Moe Zhoost (Post 22463318)
Well, it turns out that it's not noise at all. I reviewed my rowing track to check where the elevation changes were occurring and discovered that the highest elevations were close to shore. Apparently the elevation data are referenced to the lake bottom, not the actual surface.


I was wondering about that this morning. I'd be curious to know whether the app is using the actual lake bottom numbers or just extrapolating from the elevation of the nearest shoreline.

Any chance you could get on an ocean liner anytime soon?

rm -rf 04-06-22 06:43 PM


Originally Posted by Moe Zhoost (Post 22463318)
Well, it turns out that it's not noise at all. I reviewed my rowing track to check where the elevation changes were occurring and discovered that the highest elevations were close to shore. Apparently the elevation data are referenced to the lake bottom, not the actual surface.

I suppose that's possible, but unlikely. Is this at one end of your recording? It could be barometer related.

Here's an extreme example: on my ride up to Mt Mitchell in NC, I stopped at the Craggy Gardens visitor center for about 3-4 minutes. A dark summer cloud passed by, and the wind picked up. The barometric pressure went up, making my Garmin 705 think the elevation had dropped by 350 feet. (Those Garmins are stupid. It could easily get new vertical GPS readings while I'm stopped, but it never did. It just did differentials off the initial elevation, and even that could be way off. I think the newer ones are better.)

Visitor center stop at mile 18.4:
https://cimg8.ibsrv.net/gimg/bikefor...f3f4b28fdc.jpg

rm -rf 04-06-22 06:51 PM

Ha, here's my post from 2015 in another elevation accuracy thread. It's a perennial subject.

Every device and site uses it's own methods to smooth out the raw data. And different devices may use GPS, barometer, or post-event map data to determine elevations.

I expect or accept maybe a 10% inaccuracy in general. That's good enough.

rm -rf 04-06-22 06:56 PM


Originally Posted by njkayaker (Post 22463628)
It's not noise because the device isn't measuring your elevation. It's using a database (DEM), which, as I said, has issues and not ideal for elevation gain.

The issue might have been more apparent if the complete data (horizontal location and elevation) was provided.

Oh, that's likely. A post-event method, like strava has as an option, uses known elevation points. But it has to estimate the exact elevation at spots in between, and that's difficult on roads along a hill slope. Where exactly is the road surface?

And on ridwithgps, for planning routes, sometimes a fairly recent road cut isn't known, so it shows a steep climb. Or a creek crossing bridge drops down to the creek level and back up again.
The most obvious is the tunnels on the Blue Ridge Parkway. The local elevations go up and over the ridge line, making very steep grades on the route there.

njkayaker 04-07-22 07:59 AM


Originally Posted by rm -rf (Post 22464099)
(Those Garmins are stupid. It could easily get new vertical GPS readings while I'm stopped, but it never did. It just did differentials off the initial elevation, and even that could be way off. I think the newer ones are better.)

Again, GPS is poor for elevation.

You really wouldn't want it to keep fussing with the elevation every time you stopped.

Anyway, cyclists generally are interested in elevation gain (not absolute elevation).

The drop in atmospheric pressure when you were stopped should not have effected the gain much. (It would have effected the elevation loss but riders don't tend to care much about that.)


Originally Posted by rm -rf (Post 22464110)
Oh, that's likely. A post-event method, like strava has as an option, uses known elevation points. But it has to estimate the exact elevation at spots in between, and that's difficult on roads along a hill slope. Where exactly is the road surface?

And on ridwithgps, for planning routes, sometimes a fairly recent road cut isn't known, so it shows a steep climb. Or a creek crossing bridge drops down to the creek level and back up again.
The most obvious is the tunnels on the Blue Ridge Parkway. The local elevations go up and over the ridge line, making very steep grades on the route there.

The elevation data was measured using satellites and the shuttle a while ago. It doesn't include detail like roads (bridges and tunnels are a typical issue).

This is why you want your device to measure it. For tunnels, you need a barometer (GPS doesn't work in tunnels and GPS is poor for elevation anyway).

Strava can use the elevation other people recorded for your recorded ride.

livedarklions 04-07-22 10:44 AM


Originally Posted by rm -rf (Post 22464110)
Oh, that's likely. A post-event method, like strava has as an option, uses known elevation points. But it has to estimate the exact elevation at spots in between, and that's difficult on roads along a hill slope. Where exactly is the road surface?

And on ridwithgps, for planning routes, sometimes a fairly recent road cut isn't known, so it shows a steep climb. Or a creek crossing bridge drops down to the creek level and back up again.
The most obvious is the tunnels on the Blue Ridge Parkway. The local elevations go up and over the ridge line, making very steep grades on the route there.


According to the MapMy page, they use USGS data, so who knows what that's indicating for a lake's surface?

Moe Zhoost 04-07-22 11:32 AM


Originally Posted by njkayaker (Post 22463628)
It's not noise because the device isn't measuring your elevation. It's using a database (DEM), which, as I said, has issues and not ideal for elevation gain.

The issue might have been more apparent if the complete data (horizontal location and elevation) was provided.

Yes, just so. I only mentioned noise because you brought it up initially. Coordinates and elevation are a reasonably qualitative match when compared to the lake depth chart. I could not find a way to download the actual data from MMR, though.


Originally Posted by livedarklions (Post 22463780)
I was wondering about that this morning. I'd be curious to know whether the app is using the actual lake bottom numbers or just extrapolating from the elevation of the nearest shoreline.

Any chance you could get on an ocean liner anytime soon?

I think there is a definite relationship to the underwater topography. I'll do some testing on my next visit to the lake.


Originally Posted by rm -rf (Post 22464099)
I suppose that's possible, but unlikely. Is this at one end of your recording? It could be barometer related.

No barometer, gps only. And yes it is likely (see above)


njkayaker 04-07-22 11:45 AM


Originally Posted by Moe Zhoost (Post 22464684)
Yes, just so. I only mentioned noise because you brought it up initially. Coordinates and elevation are a reasonably qualitative match when compared to the lake depth chart. I could not find a way to download the actual data from MMR, though.

Even the page from MMR. There's some way of exporting the ride.

https://support.strava.com/hc/en-us/...%20to%20Strava.

It's common in "debugging" situations that people only report the information they think is important. I like to see "everything" because what people think is important often isn't.

Your original post didn't even provide the elevation profile. My comment about noise was based on the very limited information you provided.


Originally Posted by Moe Zhoost (Post 22464684)
I think there is a definite relationship to the underwater topography. I'll do some testing on my next visit to the lake.

It seems like that's the most likely explanation. Maybe, the lake was a reservoir? I don't know if the orbital scans of topography read water surface or not.

No one uses these to determine elevation gain on flat water. It's not too surprising that this use-case doesn't work very well.


All times are GMT -6. The time now is 07:43 PM.


Copyright © 2021 MH Sub I, LLC dba Internet Brands. All rights reserved. Use of this site indicates your consent to the Terms of Use.