GPS elevation changing during the day!

[Bianchi Boy]

https://www.strava.com/activities/454205477

Is a link to one of Kurt's days in the year long challenge. If you notice the elevation starts the day at above 50m and finishes the day at about 13m. The measuring device was a Garmin Edge 500 and this does not have a barometer. This appears on a number of days. Does Florida sink in the afternoon? If it was inaccuracies caused by measurement I would have expected a random effect, but it is consistent and has happened on other days.

Does anyone know why this is?

 

BB 

[T42]

He maybe starts the GPS when he starts to ride.  It'll get his position right pretty quickly, but takes a while to suss out the altitude.  I live at 170m, but my Etrex 30 usually starts at 40 or 50 metres above or below.  Even if I switch it on beforehand, when I reset the current track it flips to some default altitude, or possibly starts zeroing in all over again.

Just had a look. The other day it recorded 358m for the first point and dropped to 168 a few seconds later.

K's profile's a bit different, though.

[rr]

The Edge 500 does have a barometer.
You can tell it the height of a location and it will correct itself when it detects that it is there otherwise it will drift with pressure.
I imagine that Kurt moves around to much to have many set for each day.

[nikki]

https://www.strava.com/activities/454205477

Is a link to one of Kurt's days in the year long challenge. If you notice the elevation starts the day at above 50m and finishes the day at about 13m. The measuring device was a Garmin Edge 500 and this does not have a barometer. This appears on a number of days. Does Florida sink in the afternoon? If it was inaccuracies caused by measurement I would have expected a random effect, but it is consistent and has happened on other days.

Does anyone know why this is?

 

BB

Interesting!

As I understand it, variations in GPS-measured altitude (Vertical Dilution of Precision, VDOP) caused by the shifting geometries of the satellites would give more erratic-looking variations and over a shorter timescale. (See the stepped shape of the calculated VDOP line in Figure 3 here: http://www.nrem.iastate.edu/class/assets/nrem446_546/week3/Dilution_of_Precision.pdf).

Agreed that the smoothness in drift for the Strava track you linked to doesn't smell like the same sort of thing. If there's a barometric altimeter involved, and if you're seeing similar behaviour repeated over different days, then I'd maybe look more towards meteorological effects - the temperature chart for example show's a ~15°C increase!

[Ningishzidda]

Atmospheric pressure has increased.

https://en.wikipedia.org/wiki/File:Atmospheric_Pressure_vs._Altitude.png

[TimC]

https://www.strava.com/activities/454205477

Is a link to one of Kurt's days in the year long challenge. If you notice the elevation starts the day at above 50m and finishes the day at about 13m. The measuring device was a Garmin Edge 500 and this does not have a barometer. This appears on a number of days. Does Florida sink in the afternoon? If it was inaccuracies caused by measurement I would have expected a random effect, but it is consistent and has happened on other days.

Does anyone know why this is?

 

BB 

The Edge 500 most certainly does have a barometric altitude sensor, and the altitude drift is caused by changes in atmospheric pressure.

[Bianchi Boy]

Hi - so this looks like drift in atmospheric pressure. Hmm I thought the barometers in Garmins were self calibrating. Am I wrong about this and does it just use the barometer from the elevation last recorded? This appears odd as the device knows the accuracy and it's position within a few meters, why would it record the elevation say 30 meters out when the accuracy was say 3 meters?

BB

[Ningishzidda]

A barometer can never be 'self calibrating'.

A barometer can be a mercury or water column, or a straingauge diagphram device.

To calibrate a barometer, you either have to expose it to a known pressure with a very accurate calibration machine, or adjust the 'zero' to agree with what the ambient pressure is by observing it with a calibrated master device.

On a Garmin, the barometer is a teeny-weeny cheap mass-produced straingauge device. The Garmin firmware knows the curve of p vs elev, and the user adjusts the 'zero' by telling the Garmin what the elevation is where the device is.

'Elevation' readouts on hand held GPS units is handy for sea level hikers or cyclists who go up mountains where O2 density is 75% of Std Atmosphere. Aspen CO.
Seeings as the summit of Ben Nevis is just over half the elevation of Aspen, Colorado, an altimeter on a bicycle in the UK is surplus to demand. Where Kurt is riding, it is also.

[Pingu]

The eTrex 30 generally does a pretty good job of correcting for ambient pressure itself, though there have been a couple of occasions where it's taken a while (maybe half an hour or so) to sort itself out. It's much more useful in this regard than my VDO bike computer which often needs manual adjustments during the day if the pressure is changing.

[frankly frankie]

A barometer can never be 'self calibrating'.

Barometric Garmins (being a combination of barometer and GPS) are self-calibrating, and this is the default setting.  It's possible the self-calibration has been switched off on Kurt's device (this would happen if he calibrates manually, to 5m s.l.m for example - but I find it hard to believe he would have this as part of his daily routine).
Self-calibration works very well IME, and especially given the alternative in the UK, which would be to re-calibrate to a known height every time you ride under a weather front - there ain't nobody gonna do that.  I would only manually calibrate if I were using the GPS to survey a hill or something like that.

[mcshroom]

He had some issues with one getting soaked and not working properly a few weeks back IIRC. I wonder if this is the device and the calibration mechaism got damaged, or turned off when they were sorting the unit out?

[rr]

But the edge 500 doesn't have any map data to calibrate against. You just set elevation points and it adjusts its self at those.
My rides generally start at home or work so I have points set up for there.
If you have a unit with mapping I can see how it could compare map height and barometer height and make adjustments to get them to agree, but the edge 500 lacks one half of the needed data.

[Ningishzidda]

There might be a mix-up in terminology here as I am referring to the strain gauge pressure transducer ( barometer ) as one measuring instrument. The GPS derived elevation is a second measuring instrument within the same housing enclosure. The Garmin hand unit contains two methods of assessing elevation. The user can ask the unit to use one method to verify the other, thus the unit’s firmware performs all the button pressing automatically.
Puzzle.
Garmin are using a less accurate device to calibrate a more accurate device

As Pingu says, a lenghty timescale is sometimes required.

[Kim]

But the edge 500 doesn't have any map data to calibrate against.

It doesn't need any, it gets elevation from the GPS receiver.

[Kim]

Garmin are using a less accurate device to calibrate a more accurate device

We're back to the difference between precision and accuracy.  GPS elevation is noisy, but over the long term fairly consistent.  The barometric altimeter is more precise for small changes in elevation, but susceptible to drift due to weather.  So it takes a long-term average of the GPS elevation and uses it to bias the altimeter.

Last time I looked, you could configure this behaviour - either the self-calibration described above, manually setting the offset, or a mode where the elevation is assumed constant and it uses the sensor to measure changes in pressure, should you feel the need for a graphing barometer (perhaps at sea, or in a base camp type scenario).

[Diver300]

GPS altitude is less accurate in metres than GPS position, but is usually similar magnitude. However, GPS altitude needs satellites in view that are near the horizon for accuracy, so can be very badly compromised in urban areas.

The GPS receivers don't have accurate clocks in them, so can only actually measure timing differences between the signals from different satellites, they can't directly work out how long a signal took to arrive. If all the satellites in view are nearly overhead, being higher up would make all the signals arrive sooner, but would hardly change the difference between them, so the receiver can't work out the altitude very well.

[TimC]

As far as I'm aware, the Garmin 'calibration' only works at a point of known elevation. So, on my Edges 800 and 1000, I am able to specify the 'home' location and elevation for instance, and when the unit is turned on and recognises it's at that point, it will start its elevation trace at the known value, then use the barometer to track changes. If it doesn't know any other values for other locations (and I'm not sure it can), the trace will be distorted over time by changes in atmospheric pressure. This is one of the reasons why barometer-equipped GPS unit cumulative elevation calculations aren't that accurate, and why Strava and Connect and other online data-gatherers offer the option to correct the cumulative elevation total by using whatever elevation grid data they have available. Of course, that grid may be quite coarse and thus be even less accurate.

GPS elevation is another subject, and in cycling units is very inaccurate, which is why the grid correction above is automatically applied to non-barometric GPS tracks recorded in Strava and Garmin Connect.

[frankly frankie]

Probably different for different devices then.  The barometric Etrexes self-calibrate from the GPS elevation, at intervals of approximately 20 minutes.

The GPS receivers don't have accurate clocks in them, so can only actually measure timing differences between the signals from different satellites, they can't directly work out how long a signal took to arrive.

I don't think it's quite like that.  The GPS receiver clock gets synchronised to the sats, and then is accurate enough as long as the unit remains switched on.  So it's measuring delay (and therefore distance) from each visible satellite.  I don't really see why the shape of the visible constellation should have any bearing on accuracy - but I accept that it does, because there is lots of data published which shows this effect.

[Bianchi Boy]

Hi, Thanks for all the replies. To come back to the original point - how come the same spot changed elevation from 50m to 13m over the length of a day? It has happened on other tracks that Kurt has done and it does not look like noise to me. It is a consistent drift. I must admit I still have not seen a clear definition of how Garmin devices use the barometric pressure. I thought they used this in conjunction with the satellite to eliminate noise.

I have read quite a lot about how the GPS system works and I am confused by some of the statements made. The GPS signals are time signals and the position is determined from the time difference between the signals when they arrive at the GPS unit. The clocks in the satellites are ran on synchronised atomic clocks of mind bending accuracy. Also the position derived by GPS is a 3D co-ordinate. The positional measurements know nothing about the topography of the earth. Any assumptions that are made by Strava or any other site are based on the assumption that you are on the surface of the earth and not sky diving or caving.

BB

[Pingu]

GPS for caving 

[Feanor]

The answer is in the previous posts, but may have become buried in noise.

Basically, the GPS system is designed to give much higher accuracy in XY than in  elevation Z. The reasons are complex, but a simple analogy may help..

Imagine you are looking down directly onto the centre og a giant chess board.
You can see the pieces move in the X Y plane easily.
But for a piece directly below you, you cannot see if it elevates up off through chessboard towards you.
Pieces at the edge of the board which you are viewing slightly sideways on,you can see some vertical movement, but not as clearly as X Y movement.

Because GPS derived elevation is not too good, we often substitute barometric elevation which has much better vertical precision.

That would be the end of it if the atmosphere would just stay put.
But it doesn't,  and atmospheric pres sure rather anoyingly drifts.
This causes s alow drift in elevations that use atmospheric pressure.

That is what you are seeing in the OP.

The rest of the thread is discussing various ways of calibrating out this baseline drift.

[Kim]

GPS for caving 

Someone needs to sort that out, so we can have a Netherton Tunnel Strava segment.

[JStone]

GPS for caving 

Someone needs to sort that out, so we can have a Netherton Tunnel Strava segment.

a few pseudolites > problem solved

[JonBuoy]

GPS for caving 

Someone needs to sort that out, so we can have a Netherton Tunnel Strava segment.

Netherton Tunnel Northbound 

Not sure how Strava will handle this but it isn't as if you can take a short cut.  Whether it is a sensible segment is another matter altogether.


ETA (27/12/15): The segment wasn't populated with anyone other than me after a couple of days so I deleted it.

[JonBuoy]

The pressure sensor on my 705 always seemed to have thermal drift problems.  The first fix after a winter cafe stop usually showed a significant change in elevation which came back to 'normal' as the unit returned to ambient temperature.