Garmin 60CSx records elevation change when stationary

[JoeCedar]

I borrowed a Garmin GPSMAP 60 CSx with barometric altimeter from a friend, intending to do some hiking comparisons of iPhone apps (Gaia GPS and Motion X) which run on GPS altitude (no barometric sensor). With the Garmin on my desk under USB power with GPS off, I noticed that the ascent/descent values were increasing with time. Both ascent and descent were increasing. If this increase was due to ambient changes in barometric pressure, one should increase but not both equally. I have relied for years on a separate barometric altimeter, Suunto X6M, for recording ascent/descent while hiking and have found it accurate, reliable, and reproducible. So I did a test over three hours with both units running simultaneously. Results are shown below. Just by luck, there was essentially no change in ambient barometric pressure today during the test.


Stationary Garmin GPSMAP 60CSx versus Suunto X6M barometric altimeter

CSx was set in 'variable elevation' and autocalibration on, attached to USB power while stationary.
GPS was off.
X6M was running simultaneously on my desk (recording a point every 10 seconds)

Garmin CSx Feet In Hg
Time Asc Dsc Barometer Comments
2:00 0 0 29.84 autocalibration ON, GPS off
2:30 225 228 29.84
3:00 463 458 29.84
3:30 704 699 29.84
4:00 913 907 29.83
Reset 4:01 0 0 29.83 autocalibration OFF
4:30 243 249 29.84
5:00 502 512 29.84
Ascent/Descent values increased 450-500 feet per hour with no change in local barometric pressure
Autocalibration had no effect.
Barometer readings were from a separate barometer as well as hourly checks from a nearby NWS station

Suunto X6M
2:00 0 0
5:00 20 19
Essentially no change in Ascent/Descent while recording for three hours.


I am surprised that the Garmin accumulates significant ascent/descent values when the unit is stationary. I don't know if this is a problem with this particular unit, or it is common to all. I would greatly appreciate if some other users could try this simple test and help me understand what is going on.

 

[bikehikeskifish]

FWIW, I got my 76CSx out and it has gone 60+ minutes with at most +/- 10 feet from the correct elevation. GPS is off. It's saying total ascent of 55 feet.

Tim

 

[DougPaul]

Autocalibration requires the GPS to be turned on. It uses the GPS altitude to continuously calibrate the barometric altimeter. If I leave my GPS on and autocalibrate for my drive 2+ hour drive to the Whites, the indicated altitude at a summit is generally within 10 feet of the published altitude. Garmin's accuracy spec for the 60CSx is within 10 feet if properly calibrated.

With the GPS turned off, you are essentially running a pure barometric altimeter.

Note: autocalibration is a long-term process. While it starts immediately and runs continuously, it can take a while to remove an initial error. So its accuracy will improve as time passes.

Doug

 

[TJsName]

Autocalibration requires the GPS to be turned on. It uses the GPS altitude to continuously calibrate the barometric altimeter. If I leave my GPS on and autocalibrate for my drive 2+ hour drive to the Whites, the indicated altitude at a summit is generally within 10 feet of the published altitude. Garmin's accuracy spec for the 60CSx is within 10 feet if properly calibrated.

With the GPS turned off, you are essentially running a pure barometric altimeter.

Note: autocalibration is a long-term process. While it starts immediately and runs continuously, it can take a while to remove an initial error. So its accuracy will improve as time passes.

Doug

Occasionally when I turn my phone's GPS on right at the start of a hike, my elevation is way off. After about 10-20 minutes it'll catch up, but it's generally enough to make it look like I've gained a lot of elevation in a short time. Sometimes when I'm taking a break I can see the calibration adjusting, so my elevation steadily changes while my location has not. It's not significantly off once it's been on for a while, and it certainly doesn't impact it's usefulness.

 

[DougPaul]

Occasionally when I turn my phone's GPS on right at the start of a hike, my elevation is way off. After about 10-20 minutes it'll catch up, but it's generally enough to make it look like I've gained a lot of elevation in a short time. Sometimes when I'm taking a break I can see the calibration adjusting, so my elevation steadily changes while my location has not. It's not significantly off once it's been on for a while, and it certainly doesn't impact it's usefulness.

The total climb (or descent) may still be off after autocalibration. (Of course, the altitudes in the recorded track will also be off for the early part of the hike.)

By having my GPS on during the drive to the trailhead, I can ensure that it is calibrated at the beginning of the hike. I just have to remember to reset the trip odometer at the trailhead and read it before driving off after returning...

Doug

 

[nartreb]

JoeCedar, can your Garmin display the numeric (mmHg) barometer value?

DougPaul is assuming that the barometer inside your Garmin is giving a fluctuating reading. (Back and forth around 0.1mmHg per minute in each direction, if my math is right.)

If nobody else is seeing what your'e seeing, you could just have a defective barometer. But maybe you're not alone here:

FWIW, I got my 76CSx out and it has gone 60+ minutes with at most +/- 10 feet from the correct elevation.

So what does the cumulative ascent/descent show? If it's off by ~3 feet in a random direction and changing every minute, then in one hour you're looking at ~200 feet of ascent *and* descent. If it fluctuates every half minute, you'll match what JoeCedar saw.

 

[TJsName]

The total climb (or descent) may still be off after autocalibration. (Of course, the altitudes in the recorded track will also be off for the early part of the hike.)

By having my GPS on during the drive to the trailhead, I can ensure that it is calibrated at the beginning of the hike. I just have to remember to reset the trip odometer at the trailhead and read it before driving off after returning...

Doug

My issue tends to be not turning it off after the hike. Lots of GPS tracks with a top speed of 50/60/70 mph. It really boosts the average speed.

 

[DougPaul]

DougPaul is assuming that the barometer inside your Garmin is giving a fluctuating reading. (Back and forth around 0.1mmHg per minute in each direction, if my math is right.)

Actually DougPaul misread the original question and answered a different one.


When the GPS is off, autocalibration is disabled--therefore only the barometric sensor is active. There are small variations (likely just noise) in the measured altitude. The ascent and descent numbers are just adding up the upward and downward fluctuations. If you watch the actual altitude, you will likely find the numbers to vary slightly around a fixed value (if barometric pressure is constant) or a slowly varying value (if barometric pressure is changing).

This is normal behavior. Noisy data fed to accumulators often results in steadily increasing values. (The same will happen to the accumulated distance estimate for a stationary GPS (with the GPS on, of course).)

The Suunto may include some form of filtering to hide the noise from the user. This, however, may decrease the accuracy of the ascent (and descent) estimates under hiking conditions.

If nobody else is seeing what your'e seeing, you could just have a defective barometer. But maybe you're not alone here:

Yes, the barometer could be defective, but the effect will still occur if it is operating properly.

Doug

 

[DougPaul]

My issue tends to be not turning it off after the hike. Lots of GPS tracks with a top speed of 50/60/70 mph. It really boosts the average speed.

You can download the track to a GPX file, edit it so it only contains the hike, and then compute the stats. These stats will be less accurate than those computed by the GPS because the GPS has more info available to it than is in the track files.

Also, the top speed is a vary iffy parameter--it only takes one inaccurate trackpoint to give you a high speed. (One of my older GPSes would occasionally give me max speeds of hundreds of mph due to a bad point or two.)

Doug

 

[JoeCedar]

JoeCedar, can your Garmin display the numeric (mmHg) barometer value?

DougPaul is assuming that the barometer inside your Garmin is giving a fluctuating reading. (Back and forth around 0.1mmHg per minute in each direction, if my math is right.)

If nobody else is seeing what your'e seeing, you could just have a defective barometer. But maybe you're not alone here:



So what does the cumulative ascent/descent show? If it's off by ~3 feet in a random direction and changing every minute, then in one hour you're looking at ~200 feet of ascent *and* descent. If it fluctuates every half minute, you'll match what JoeCedar saw.

The choices of pressure units are In Hg (inches of mercury, often used in the U.S.), millibars, or Hectopascals.

Electronic barometers can measure very small changes in pressure and convert the pressure to altitude by a standard calculation. See
https://en.wikipedia.org/wiki/International_Standard_Atmosphere
The calculations make several simplifying assumptions about the change in atmospheric pressure with altitude (lapse rate), but it is an important international standard, especially in aviation.

I agree with Doug here that the specific unit I am using seems to have unusually high noise (in relation to signal) and this was confirmed by Tim's measurement (at least confirmed with one other GPS unit). This unit is interpreting/ noise as real changes in altitude (pressure) and adding it up (and down) inappropriately. I don't know how common this erroneous reporting is and I would still be interested in other users trying the simple test and reporting it. Note that this is not drift where one or the other of ascent/descent values would be larger.

A more important question, but not part of this discussion, is how all this affects the functional use of the unit. In comparing results from this 60CSx with iPhone apps and my altimeter simultaneously during hikes, I haven't seen anomalously high ascents/descents.

 

[DougPaul]

Electronic barometers can measure very small changes in pressure and convert the pressure to altitude by a standard calculation. See
https://en.wikipedia.org/wiki/International_Standard_Atmosphere
The calculations make several simplifying assumptions about the change in atmospheric pressure with altitude (lapse rate), but it is an important international standard, especially in aviation.

One of the effects of continuous automatic calibration from the GPS is that differences between the standard lapse rate and the current conditions are calibrated out. (This is good for hiking, but has pros and cons for aviation.)

I agree with Doug here that the specific unit I am using seems to have unusually high noise (in relation to signal) and this was confirmed by Tim's measurement (at least confirmed with one other GPS unit). This unit is interpreting/ noise as real changes in altitude (pressure) and adding it up (and down) inappropriately. I don't know how common this erroneous reporting is and I would still be interested in other users trying the simple test and reporting it. Note that this is not drift where one or the other of ascent/descent values would be larger.

I have no idea whether the barometric altimeter (including the electronics) in your unit is any noisier than any other unit. However the accumulation of noise in climb or descent for a stationary unit is normal.

A more important question, but not part of this discussion, is how all this affects the functional use of the unit. In comparing results from this 60CSx with iPhone apps and my altimeter simultaneously during hikes, I haven't seen anomalously high ascents/descents.

In normal hiking use, the (actual) rate of climb or descent is usually high enough to exceed the noise and the effect will disappear. At slow climbing or descent rates there may be some mixing of signal (true changes) and noise resulting in somewhat inflated values.

For example:
Assume that the noise is +1, 0, or -1 units added to the correct altitude. If the GPS is stationary and the pressure is constant, the climb estimator will add up all of the increases and the descend estimator will add up all the decreases. Now if the GPS is climbing at a steady 5 units/sec, the output will be 4, 5, or 6 units which will be all increase with a long-term average rate of 5/sec. (The GPS takes one measurement per second.)

FWIW, I have a 60CSx running (with calibration off) as I type. The altitude reading is fluctuating by a foot or two around a fixed value and my desk has "accumulated" 6 feet of climb and 5 feet of descent.

Doug

 

[iAmKrzys]

Ascent & descent is not the only casualty of cumulative measurement error. Here is an experiment anyone who owns a gps can try: reset the counters / current track and leave your gps on for an hour in one spot. Most likely at the end of the exercise you will be able to admire an nice snowflake pattern on your gps screen and you will discover that your gps did quite a bit of hiking on its own when you check the counters or download the gpx to your computer.

 

[DougPaul]

Ascent & descent is not the only casualty of cumulative measurement error. Here is an experiment anyone who owns a gps can try: reset the counters / current track and leave your gps on for an hour in one spot. Most likely at the end of the exercise you will be able to admire an nice snowflake pattern on your gps screen and you will discover that your gps did quite a bit of hiking on its own when you check the counters or download the gpx to your computer.

All real-world measurements of continuous parameters (such as location and altitude) have noise (errors). You have to learn how to deal with them.

Note that long-term averaging of the location or altitude will generally give you a more accurate value than a single measurement.

<technical>
FWIW, I have recorded 24hr tracks from a stationary GPS to do such averaging. I also computed the autocorrelation (dot product of a vector with a time-shifted copy of itself) of the difference from the average--it was positive for shifts up to 90 minutes and then swung negative. (This is probably due to satellite ephemeris errors and the orbital paths.)

FWIW2, observed signals (measurements) are often modeled as a combination of signal and noise. There are many kinds of noise and modeling them as well as signal estimation in the presence of noise can be a complex topics.
</technical>

Doug

 

[JoeCedar]

Ascent & descent is not the only casualty of cumulative measurement error. Here is an experiment anyone who owns a gps can try: reset the counters / current track and leave your gps on for an hour in one spot. Most likely at the end of the exercise you will be able to admire an nice snowflake pattern on your gps screen and you will discover that your gps did quite a bit of hiking on its own when you check the counters or download the gpx to your computer.

This is a good example of how smartphone technology has surpassed purpose-built GPS receivers (maybe newer GPS models do this; I don't know). You didn't say what make/model you were using. I repeated your experiment with my iPhone 5S (2 years old) with GPS (location services) on and cellular off (airplane mode). I started two GPS apps (Motion X and Gaia GPS) and left them running 15 minutes in a stationary position. Motion X recorded an empty gpx file and Gaia recorded the start and finish points (which were identical). I have found that these two apps, which are the most commonly used for hiking applications, are programmed quite differently in how they acquire data.

The app programmers are apparently using the inertial sensors in the phone to prevent recording of waypoints when the device is not moving, eliminating the annoying hairballs with the older GPS devices.

 

[DougPaul]

The app programmers are apparently using the inertial sensors in the phone to prevent recording of waypoints when the device is not moving, eliminating the annoying hairballs with the older GPS devices.

That is quite possible, however, such strategies (often used to give the appearance of better accuracy) can decrease accuracy under certain conditions and/or cause other odd behaviors.

The inertial sensors also have noise, and like the barometric sensor, it can be difficult-to-impossible to distinguish between slow movement and noise.

An example of an odd behavior caused by such a strategy can be found in my digital scale: if the measured weight changes by less than one pound, it just repeats the previous weight. (Which reduces accuracy for small changes.) Presumably, the manufacturer thinks customers are incapable of understanding that multiple weighings may give slightly different values. (There is a simple workaround--just weigh oneself twice, the first time with enough extra weight to break the one pound limit, the second time to get the proper weight.)

Doug

 

[iAmKrzys]

I imagine that a moving average filter ( https://en.wikipedia.org/wiki/Moving_average ) could take care of some of this type of noise albeit at the "cost" of making the unit appear to be less sensitive to instantaneous changes.

 

[DougPaul]

I imagine that a moving average filter ( https://en.wikipedia.org/wiki/Moving_average ) could take care of some of this type of noise albeit at the "cost" of making the unit appear to be less sensitive to instantaneous changes.

A moving average filter is a form of low-pass (smoothing) filter. They are a specific use of windowing. https://en.wikipedia.org/wiki/Window_function. Windows are a well-studied tool in signal processing. (My degree specialty is classical signal processing.)

I have done some experiments using median filters https://en.wikipedia.org/wiki/Median_filter to see if I could reduce the effects of the noise on the altitude portion of a 1 per second track. They didn't work very well. I think this is a harder problem than it appears at first glance...

Both median filters and moving averages would work well for a steady climb or descent, but they both clip the peaks and valleys.


Note also that the GPS has more information available to it than is recorded in the track. The GPS outputs the position (3d), the velocity (3d), and the time (1d) once per second (for consumer GPSes). The track only records the position and time and usually only saves a subset of the track-points. (Garmin GPS tracklogs are limited to 10K points so you can only save ~3hrs at 1/second.)

Doug

 

[JoeCedar]

I borrowed a Garmin GPSMAP 60 CSx with barometric altimeter from a friend, intending to do some hiking comparisons of iPhone apps (Gaia GPS and Motion X) which run on GPS altitude (no barometric sensor). With the Garmin on my desk under USB power with GPS off, I noticed that the ascent/descent values were increasing with time. Both ascent and descent were increasing. If this increase was due to ambient changes in barometric pressure, one should increase but not both equally. I have relied for years on a separate barometric altimeter, Suunto X6M, for recording ascent/descent while hiking and have found it accurate, reliable, and reproducible. So I did a test over three hours with both units running simultaneously. Results are shown below. Just by luck, there was essentially no change in ambient barometric pressure today during the test.


Stationary Garmin GPSMAP 60CSx versus Suunto X6M barometric altimeter

CSx was set in 'variable elevation' and autocalibration on, attached to USB power while stationary.
GPS was off.
X6M was running simultaneously on my desk (recording a point every 10 seconds)

Garmin CSx Feet In Hg
Time Asc Dsc Barometer Comments
2:00 0 0 29.84 autocalibration ON, GPS off
2:30 225 228 29.84
3:00 463 458 29.84
3:30 704 699 29.84
4:00 913 907 29.83
Reset 4:01 0 0 29.83 autocalibration OFF
4:30 243 249 29.84
5:00 502 512 29.84
Ascent/Descent values increased 450-500 feet per hour with no change in local barometric pressure
Autocalibration had no effect.
Barometer readings were from a separate barometer as well as hourly checks from a nearby NWS station

Suunto X6M
2:00 0 0
5:00 20 19
Essentially no change in Ascent/Descent while recording for three hours.


I am surprised that the Garmin accumulates significant ascent/descent values when the unit is stationary. I don't know if this is a problem with this particular unit, or it is common to all. I would greatly appreciate if some other users could try this simple test and help me understand what is going on.

Thanks all, for your comments and discussion. I noticed last night and confirmed today that the cause of the noise and concomitant increase in ascent/descent was having the unit plugged into USB power. It does not occur when under battery power. I also connected the USB cord to the iPhone charger in a wall outlet, located more than 6' from the computer, and the noise occurs there too. So it is not specific to the computer USB power source but just being connected to external power.

 

[CaptCaper]

My Garmins with Glonass/GPS with WAAS ON has the lowest noise out of any of the previous units I've owned since 1996... And the most accurate in tracking,elevation data,etc. Also there is ton's of data on the web telling you exactly why elevation data in these units will never be accurate say as over the ground gps use.. which is more accurate per say in fix then elevation data is. Just the nature of the beast... just a slight change in the air pressure while hiking ( a front coming in) will change the accuracy. Happy reading and searching... why this is.