Altitude readings - GPS versus Barometric

[limbot]

After a ride yesterday with a mate found out the real difference between GPS and Barometric Altitude readings.


Bit of a difference here The top one is via a Galaxy S3 running Strava, the bottom using a Garmin 1000


Here's how the S3 records the altitude


Here's how the Garmin records the altitude.

Interestingly enough I've done this ride about 3 times and the S3 records as around 1300 M for the 3 rides. It's worth noting that the ride is on a fairly heavily forested area on a hillside ( it's the Pipeline Track Mount Wellington).

Distance is pretty close on both and I did do some backwards/forwards riding that the other rider didn't while waiting/checking on other riders

Looking at other rides recorded with the S3 however this problem doesn't occur, it's obviously just the location of this ride.

Worth considering when deciding on what device to record with

[SheikYerbouti]

Is that difference due to the device, or Strava messing with the GPX generated by the device?

[sogood]

Heavily wooded area would also affect GPS signal reception. Barometer when calibrated is far more accurate for altitude readouts. Typically you'll need to allow a little time at the beginning of a ride for it to calibrate and stabilise.

[limbot]

Weirdly I have 3 rides on this track that show 1338, 1351 and 1311 m elevation and 26.3, 25.6 and 27.2 km distance. Weirdly consistent readings for an inconsistency ?

[silentC]

My understanding is that Strava will give preference to barometric altimeter readings if present, otherwise they use elevation maps. Because the granularity of those is not particularly fine they have to extrapolate a lot of in-between elevations based on the adjacent contours. Given that their segment matching is far from perfect, I'd say they have similar issues with the extrapolation algorithm.

So if you were 3/4 of the distance between two contours but it was relatively flat up to that point, they might read you higher than you actually are. I'm sure it's more complicated than that. I've written to them about a couple of issues along those lines but you basically get a shrug of the shoulders.

[limbot]

Mmm, from https://strava.zendesk.com/entries/2096 ... r-Activity" onclick="window.open(this.href);return false;

. For devices without barometric altimeters, we consult elevation databases to determine the elevation at each point in the activity. The resolution of these databases can vary based on location. For example, the USGS National Elevation Dataset (NED), which we consult for activities located in the US, generally has a 10 meter resolution though some small areas have a 3 meter resolution. For activities located outside of the US, we consult the ASTER and SRTM databases which have a resolution of 30 meters and 90 meters respectively.

[silentC]

Yep so if you are between two contours on the map, then they have to try and calculate your elevation. If you are riding along the top of a ridge or along the base of a sharp slope you might get weird readings as you weave along getting further away and closer to the contour, even though your actual elevation change is much smoother.

I was looking at one yesterday which shows about a 2 metre change in elevation as you cross the road.

[lobstermash]

I find that when it's raining/wet, my 510's barometer goes well out of whack, giving all sorts of strange altitude profiles and usually under-reading by a long way - not quite as extreme as your example though!

[MattyK]

I did 20km on the Great Ocean Road once, Strava/iPhone logged 1000m+ of climbing... Must have thought I was jumping up and down the cliffs or something.

http://www.strava.com/activities/118880736" onclick="window.open(this.href);return false;

[Great Scott]

I find that when it's raining/wet, my 510's barometer goes well out of whack, giving all sorts of strange altitude profiles and usually under-reading by a long way - not quite as extreme as your example though!

That's to do with atmospheric changes. I use a Garmin 510.
Riding down Beach Road (Melb), if I have cold cloudy front passed over, my elevation will drop -10 to -20m below sea level and returns to normal when weather front has passed over.

[silentC]

I did 20km on the Great Ocean Road once, Strava/iPhone logged 1000m+ of climbing... Must have thought I was jumping up and down the cliffs or something.

http://www.strava.com/activities/118880736" onclick="window.open(this.href);return false;

If you look at the contours, you can see that the zig-zagging sections in the elevation chart correspond to where you are riding parallel to the steepest parts of the terrain, so the change from one contour to the next is compressed. Every time your route crosses the contour line, they are putting you proportionally higher or lower, even though your elevation is not changing that much because you're basically straddling the contour line all the way. They would be assuming a straight line between contours, which is the only thing they can really do.

[biker jk]

I find that when it's raining/wet, my 510's barometer goes well out of whack, giving all sorts of strange altitude profiles and usually under-reading by a long way - not quite as extreme as your example though!

That's to do with atmospheric changes. I use a Garmin 510.
Riding down Beach Road (Melb), if I have cold cloudy front passed over, my elevation will drop -10 to -20m below sea level and returns to normal when weather front has passed over.

Actually, it's due to water getting into the barometer sensor hole in the Garmin. I've had the same experience as lobstermash in the rain where the Garmin has stopped recording any altitude change and/or giving a -2% slope when climbing. If you use a cover for the Garmin it's less likely to happen.

[Great Scott]

I did 20km on the Great Ocean Road once, Strava/iPhone logged 1000m+ of climbing... Must have thought I was jumping up and down the cliffs or something.
http://www.strava.com/activities/118880736" onclick="window.open(this.href);return false;

If you look at the contours, you can see that the zig-zagging sections in the elevation chart correspond to where you are riding parallel to the steepest parts of the terrain, so the change from one contour to the next is compressed. Every time your route crosses the contour line, they are putting you proportionally higher or lower, even though your elevation is not changing that much because you're basically straddling the contour line all the way. They would be assuming a straight line between contours, which is the only thing they can really do.

I've never heard of "contours" being used when discussing GPS, that's an old paper mapping concept isn't it?.
GPS has poor accuracy at calculating vertical height, triangulation of satellites simply struggle because there being asked to estimate a distance which is near parallel to their line-of-sight. Vertical accuracy can be as bad as +/-20m up/down (varies).

Your device tracks your position by logging GPS data points every 0.1-2.0 secs depending on your device settings. Garmins can be adjusted, phones generally cant. Higher number of data points translates to better the accuracy.
At each GPS data point, your device attempts to calculate your position XYZ from scratch. With the vertical estimate being poor, your elevation points are logged up/down eg. -10m to +10m even if you may be riding on near level ground.

Barometric data is far more accurate because it records changes in pressure (converted to elevation.) It take the previous data point recorded and adjusts the elevation based on the change. Even if you don't calibrate your starting altitude, it will still more accurately record overall climbing and ascending for a ride.

Strava has elevation correction as an option for those using GPS altitude data. I have mates record ridiculous climbs on their Iphones, eg. 4500m going up Mt. Buffalo (1400m?). They dont correct it because it looks good on their stats.

[silentC]

They use elevation databases which provide elevations at spot points. It's the resolution that affects the accuracy. There can be up to 90 metres between points, according to their FAQ. The GPS provides the XY coordinates and then they look up the elevation at that point in the database. If the location is between points (what I called contours, sorry) then they have to extrapolate an estimated elevation based on adjacent points.

Now if you look at the contours on the actual overlay map on Strava, you will see that when the route meanders either side of a contour line, there is a lot of fluctuation in the elevation plot. I am drawing the conclusion that this is because their extrapolation is placing you somewhere along a line between point A and point B and then estimating an elevation based on the difference between the two and how far along it you are.

From what I can gather, they don't use GPS elevation data at all if they can avoid it. They prefer altimeter, but use elevation data cross referenced to the GPS lat and long in preference to the GPS calculated elevation.

[silentC]

Actual terrain:

c1
__\
___\
____\_____c2
___________\

What Strava thinks it looks like:

c1
____\
______\
________\
__________c2
___________\

So you are riding along an elevation at C2. If your route takes you closer to the slope, then Strava will plot you higher up than you actually are because it can't know that you're on a plateau since the elevation database doesn't have fine enough resolution.

[wombatK]

Barometric data is far more accurate because it records changes in pressure (converted to elevation.) It take the previous data point recorded and adjusts the elevation based on the change. Even if you don't calibrate your starting altitude, it will still more accurately record overall climbing and ascending for a ride.

Have you actually done that ? I've owned three Garmin devices (705 and two 500's) and a Bryton Rider 35 which also has a barometric altimeter.

None of them produce anything but rubbish for their altitude plots ... you can calibrate them all you like, and then ride around a 5 km loop like Sydney Olympic Park and find elevation increases by 5m or more per 5km lap. Every time you do it (i.e. irrespective of weather conditions). And I've been at -20 to -100m altitude so many times it's a wonder I'm still breathing.

On the other hand, I've checked the height accuracy of my various GPS devices (car, phone, pocket PC etc.,.) against a number of different trig stations and found them to be within a few meters on height and latitude and longitdue.

Provided you can see the sky, the GPS data or even the interpolated data that Strava and other elevation lookup tools produce is better than
the barometric sensors readings any of my devices gives.

Cheers

[biker jk]

Barometric data is far more accurate because it records changes in pressure (converted to elevation.) It take the previous data point recorded and adjusts the elevation based on the change. Even if you don't calibrate your starting altitude, it will still more accurately record overall climbing and ascending for a ride.

Have you actually done that ? I've owned three Garmin devices (705 and two 500's) and a Bryton Rider 35 which also has a barometric altimeter.

None of them produce anything but rubbish for their altitude plots ... you can calibrate them all you like, and then ride around a 5 km loop like Sydney Olympic Park and find elevation increases by 5m or more per 5km lap. Every time you do it (i.e. irrespective of weather conditions). And I've been at -20 to -100m altitude so many times it's a wonder I'm still breathing.

On the other hand, I've checked the height accuracy of my various GPS devices (car, phone, pocket PC etc.,.) against a number of different trig stations and found them to be within a few meters on height and latitude and longitdue.

Provided you can see the sky, the GPS data or even the interpolated data that Strava and other elevation lookup tools produce is better than
the barometric sensors readings any of my devices gives.

Cheers

Just to back up this point my Garmin recorded my altitude as -100m when I was on the Pacific Hwy, 1km north of the Hawkesbury river. I had to set the altitude to 30m manually for the readings to make any sense on the ascent of Mt White.

[sogood]

There are various settings associated with the barometric function on the Garmin. User really need to get the settings right as well as to ensure both GPS and the altimeter had time to stabilise before use. Otherwise there's bound to be complaints. Then there's the potential that certain firmware version may have bugs. If all else fails, then consider HW problems.

[Tim]

I find that when it's raining/wet, my 510's barometer goes well out of whack, giving all sorts of strange altitude profiles and usually under-reading by a long way - not quite as extreme as your example though!

That's to do with atmospheric changes. I use a Garmin 510.
Riding down Beach Road (Melb), if I have cold cloudy front passed over, my elevation will drop -10 to -20m below sea level and returns to normal when weather front has passed over.

Actually, it's due to water getting into the barometer sensor hole in the Garmin. I've had the same experience as lobstermash in the rain where the Garmin has stopped recording any altitude change and/or giving a -2% slope when climbing. If you use a cover for the Garmin it's less likely to happen.

Water in the sensor will definitely upset the readings, but, when a front is approaching this is a different situation.
As a cold front approaches the barometric pressure will gradually decrease. When the front is very close and as it passes overhead the air pressure will suddenly rise until it has passed through and returns back to the approximate pre-frontal lower pressure . The Garmin is fooled into thinking you have rapidly descended in altitude ie. descending altitude leads to higher pressure.
Having said that, my Edge 800 produces some very whacky elevation readings. Very rarely the same on identical consecutive rides. It all depends on the constantly changing air pressure/s on the day/s.

[sogood]

...when a front is approaching this is a different situation.
As a cold front approaches the barometric pressure will gradually decrease. When the front is very close and as it passes overhead the air pressure will suddenly rise until it has passed through and returns back to the approximate pre-frontal lower pressure . The Garmin is fooled into thinking you have rapidly descended in altitude ie. descending altitude leads to higher pressure.

True scenario. That's also why one needs to activate continuous calibration in the settings, so that the unit will make a decision by incorporating both the baro and GPS data input. If one wants to record the barometric reading for weather prediction, then one needs to set 'fixed elevation' in the settings. Otherwise and if there's good GPS signal, 'Continuous' calibration is best. With the crossing of a front and depending on the rate of change, there may still be a dip or blip, but it'll soon be normalised. That's just a fact of this technology. In any case, aircrafts also use barometer for altitude readings.

[biker jk]

...when a front is approaching this is a different situation.
As a cold front approaches the barometric pressure will gradually decrease. When the front is very close and as it passes overhead the air pressure will suddenly rise until it has passed through and returns back to the approximate pre-frontal lower pressure . The Garmin is fooled into thinking you have rapidly descended in altitude ie. descending altitude leads to higher pressure.

True scenario. That's also why one needs to activate continuous calibration in the settings, so that the unit will make a decision by incorporating both the baro and GPS data input. If one wants to record the barometric reading for weather prediction, then one needs to set 'fixed elevation' in the settings. Otherwise and if there's good GPS signal, 'Continuous' calibration is best. With the crossing of a front and depending on the rate of change, there may still be a dip or blip, but it'll soon be normalised. That's just a fact of this technology. In any case, aircrafts also use barometer for altitude readings.

Where is the "continuous calibration" setting on the Garmin 800?

[sogood]

Where is the "continuous calibration" setting on the Garmin 800?

Good question. It's on my Garmin Oregon 650.

Given that the thread is not device specific and I occasionally use my O650 or iPhone on rides, I was just suggesting that there are various settings that may come into play on this. My new iPhone6 now has a barometric sensor but it seemed to be buried and controlled by iOS presets. I am still testing its performance characteristics.

[BaldPatch]

I find that when it's raining/wet, my 510's barometer goes well out of whack, giving all sorts of strange altitude profiles and usually under-reading by a long way - not quite as extreme as your example though!

That's to do with atmospheric changes. I use a Garmin 510.
Riding down Beach Road (Melb), if I have cold cloudy front passed over, my elevation will drop -10 to -20m below sea level and returns to normal when weather front has passed over.

Erroneous readings of gradient and elevation can also occur when riding in the rain.
There is a small barometric pressure sampling port on the front right of the Garmin, when water enters this port it affects readings.

[Tim]

Where is the "continuous calibration" setting on the Garmin 800?

If it's on the 800 I've never been able to find it and I know the 800 menu items back to front.
Weirdly though a very few time a message has popped up along the lines of "Automatic Elevation Calibration" or words to such effect on startup, but I've only seen it once or twice. I thought maybe the barometric elevation reading may have been being calibrated to the installed mapping software contours.

[sogood]

I'd think that for the sports orientated Garmin units, continuous calibration of its barometric data feed would be the built-in default. But no direct experience with it. I'd agree that port interference would be a potential issue. Otherwise I've found the barometric elevation reading to be reliable once properly calibrated and settled prior to each recording session.