RWGPS vs Garmin Edge 810 elevation discrepancies

[LucF]

Hi all,

I plan my rides using Ride With GPS and use a Garmin 810 to record my rides. I am seeing large discrepancies between RWGPS's predictions and actual ride data, the ride elevation (and grades) always turning out to be higher than what RWGPS predicted.

For example, on this ride RWGPS is predicting 1852 meters of elevation. After riding it and uploading my Edge 810 ride in Training Peaks and Garmin Connect I get 2739 meters of elevation (which is also what the Garmin gives me in the ride summary), and 3111 meters after uploading the exact same ride in RWGPS.

1. I'd like to know if others are seeing the same discrepancies.

2. I understand that Garmin's method of using barometric pressure to calculate elevation on the 810 is not reliable (why they didn't decide to use the satellite data is a mystery), but is it unreliable enough to explain the 1852 vs 2739 meter delta?

[Eric Keller]

both GPS and barometric measurements can produce large altitude errors. My understanding is that GPS is usually worse.  I don't know if combining both would actually do anything to improve the altitude estimate.

 I would go with the RWGPS totals, they are more accurate.  This is data I uploaded after a ride that featured a heavy thunderstorm at mile 42.  Note that the elevation of my house is about 300 feet higher at the end of the ride than the beginning
http://ridewithgps.com/trips/1885037

I suspect this error is a combination of lowered temperatures and lower barometric pressure due to the front coming through.

I thought about pulling that ride into my routes to see what the RWGPS elevation estimate would be, but you get the idea

--
You received this message because you are subscribed to the Google Groups "randon" group.
To unsubscribe from this group and stop receiving emails from it, send an email to randon+un...@googlegroups.com.
To post to this group, send email to ran...@googlegroups.com.
Visit this group at http://groups.google.com/group/randon.
For more options, visit https://groups.google.com/d/optout.

[Larry Parker]

Garmin Connect does have the option of switching to a "corrected" altitude, down in the bottom left (of the old interface, anyway!) This is supposed to be based on other data, maps, (surveys?)  They say the barometric Altimeter can be more accurate, but it is interesting to see the differences when you switch between them.

 

Larry

[Dan Shinnick]

If I had to guess, I would say that the Garmin elevation data is based on NED DEM for the U.S. anyway.  They probably extract elevation from points along a route and viola!  Profile!  The resolution of the NED DEM varies by location.  Since RWGPS' capabilities appear global in scope... it's possible they use the SRTM DEM which is even more coarse.  

--

Daniel Shinnick

[Hamish Moffatt]

Garmin Connect shows the ascent measured by your device unless you turn on elevation correction. The correction looks at the actual maps and should always be more accurate. Strava has a similar function.

GPS elevation is pretty inaccurate hence better devices include a barometric sensor. I'm not sure how they combine the data from the two as obviously the barometric sensor is somewhat affected by the weather. And on the bike computers prone to filling up with water during a wet ride.


Hamish

[LucF]

Thanks for the replies. What I gather is that, once again, my Garmin 810 cannot be trusted and that RWGPS's courses should provide more reliable elevation data. All those big total elevations in my training summary pages in Training Peaks should then be taken with a grain of salt. Geez, thought I was doing great :)

Eric, I converted your ride to a course here and we can clearly see the loss of resolution created by the thunderstorm when comparing its elevation profile to your ride data at mile 42. In this case however, the elevation of the RWGPS course is superior to your ride data (1736 vs 1479 meters).

[NickBull]

What is your evidence that RWGPS is "more accurate"?

What device did you use to record your track?  Whatever it was, it looks like you did not have it set to use GPS observations to correct the barimetric data.

[NickBull]

No.  RWGPS elevation is not a "gold standard".  When you design a route, it uses digital elevation mapping (DEM) data to determine how much climbing there is.  There are known problems with these data, for instance the fact that the digital elevation follows the contours of the land, not the road.  So when there is a bridge, the DEM data "walk across the water", even if it means going down a 1000 foot canyon instead of giving you the elevation of the bridge.  To try to adjust for that, RWGPS "filters" the data.  But in the process, they also throw out a lot of actual climbing.  Basically, it is garbage in, garbage out.

I would trust your Garmin 810 more than I would trust RWGPS.  Set it so that the GPS autocorrects the barometer.  Then you can compare your rides with each other based on the climbing that _you_ have measured.

Nick

[Craig McGregor]

The most accurate would be a barometer only without any GPS recalibration. However, manually adjusting for known elevation at the start of a ride may not be as user friendly as a GPS adjusted barometer.

GPS elevation data is far, far, far less accurate than barometric estimates, and even more so in hilly terrain due to the need to reference more satellites for accurate elevation, than required for position. That is why the aviation industry still relies on barometers for this purpose. Likewise when I am tramping (hiking), I make use of a barometric altimeter to assist navigation.

The rises and falls of inaccurate GPS altitude values can a lot of height to a ride, by adding many additional rises and falls.

Any pressure changes during a ride are likely to affect a barometer's climbing estimate far less than the totals achieved via GPS inaccuracy and approximated use of map data that lacks sufficient granularity.

I get the closest matches to barometric calculated totals with RWGPS which shows vast improvements in their data smoothing algorithms that they have applied to map data over time.

Q. Does anyone know how Garmin adjust the barometric altitude estimate based on the (less accurate) GPS readings? Or does it use known points in map data for that purpose?

Craig.

--
Craig McGregor
cr...@mcgregor.gen.nz
http://craig.mcgregor.gen.nz
http://twitter.com/craigofnz

[littlecirclesvt.com :: mike beganyi]

If you are on a PC (or a Mac running parallels or have a bootcamp setup) check out TopoFusion. Scott has built some algorithms into the software when analyzing elevation, and when you do a climbing analysis it will give you results based on the track and several popular websites and software packages. And you can analyze the track by checking it against time (you might be standing still and the GPS and altimeter data are bouncing all over the place), or you might have moved very slowly or very quickly over some terrain and have some anomalies to filter out.

I've been using the software for its sat and Topo maps access for years, and typically after creating a route on RWGPs will bring it into TopoFusion to study the various terrain it covers (more for bikepacking / mountain biking / etc).

Mike

[Nicholas Bull]

Pilots adjust their altimeters frequently to known altitudes, because barometers are affected by changing external air pressure. 

Setting your barometer to a known altitude at the start of a ride is fine if you are only riding for an hour or so.  But temperature and pressure changes through the day on any randonneuring event will mean that barometric elevation change is inaccurate.

I have been using a GPS with a barometer since 2006.  It is rare that on a ride where I finish at the same location as I started, that the total climbing is different than the total descending by more than twenty or thirty feet.

Of course, there is the issue that all GPS's and barometers have to measure altitude relative to a hypothetical reference sea level.  In principle, this should be the geoid, but in practice they only approximate the geoid, so the altitude that the GPS reports may not perfectly match a DEM, even if the GPS were perfectly accurate. 

RWGPS's goal is for its climbing to match what will be reported on a given route by a GPS with a barometer: "With extensive testing, accuracy has been greatly improved so that planned routes now have elevation gain/loss much closer to what a GPS unit with a barometric altimeter reports."  For measuring actual climbing on a route, a GPS with a barometer is going to be more accurate than a barometer alone.

Nick

[Nicholas Bull]

In answer to your question about how Garmin is doing its error correction:  According to Cullen at RWGPS, Garmin uses a Kalman filter to fuse the two sensors (barometer and GPS).

[Hamish Moffatt]

On 14/07/14 07:01, Craig McGregor wrote:
>
> Q. Does anyone know how Garmin adjust the barometric altitude estimate
> based on the (less accurate) GPS readings? Or does it use known points
> in map data for that purpose?
>
>

http://www.dcrainmaker.com/2010/05/understanding-sport-device-gps.html
suggests that GPS is used for the initial reading and the barometer used
for tracking after that. I've noticed that if I start riding before the
GPS locks, the elevation on my Edge 800 is offset for the whole ride, so
I suppose that adds up.

Also you can add waypoints to the Garmin and specify your own elevation,
and if you are close to the waypoint when starting the ride the computer
will use that elevation as the baseline.


Hamish

[Bill Gobie]

Some interesting information/conjectures here:

http://regex.info/blog/2006-04-16/179

Points I found interesting:

* Some (all?) Garmins w/barometers automatically recalibrate the barometric altimeter to match the GPS altitude. I read this somewhere else, where it was recommended to start the GPS a half hour or so before an event to let the altimeter calibrate.

* Wind velocity can alter the pressure the barometer sees enough to cause a significant change in derived altitude. 

My experience with a GPS on a sailboat is that GPS altitude is often significantly wrong. According to the gps I was usually aboard a submarine or airplane.

Bill

[David Buzzee]

I was a member of AUKs for several years. At that time (and perhaps now as well) official climbing measures were calculated by counting the contour lines on a topographic map along the route of interest, then dividing by the contour interval and again by two (to account for elevation loss going downhill). Perhaps not terribly sophisticated but reproducible and probably beyond misinterpretation - and undeniably easy.

db

--
Craig McGregor
cr...@mcgregor.gen.nz
http://craig.mcgregor.gen.nz/
http://twitter.com/craigofnz

[LucF]

Thanks to all for the additional, very useful information.

Looking at my ride data more closely, I notice that I left home at -87 meters (reality is somewhere around 17 meters according to Google Earth, which is what I had set my home location to) and came back at +117. So something was wrong from the beginning, and it rained for the last third of the ride. I will define a location+altitude somewhere away from home where I'll pass by when starting rides so that the 810 can calibrate from there. I had defined my home's altitude but since I live on the fourth floor, I realize that starting the Garmin on the fourth floor or at ground level could be a problem if it's calibrating both positions to the same altitude.

@mike: I tried TopoFusion. Thanks.

@Craig: I found this reassuring: Any pressure changes during a ride are likely to affect a barometer's climbing estimate far less than the totals achieved via GPS inaccuracy and approximated use of map data that lacks sufficient granularity. So my 810's elevation measurements probably aren't that bad after all, if it's given the means to calibrate properly.

I will try to gather more meaningful data and report back on my findings.