GPS height accuracy?

Hi all,

I carry a basic Garmin ETREX 10, it doesnt include any maps, but gives me an accurate grid ref in poor visibility. (I have also downloaded a track from SotaMaps, just to prove that I could, which worked very well)!

Anyway, on Black Mountain GW/SW041 a couple of days ago, I walked past the trig point at Hay Bluff, and onwards along the ridge until I was somewhere near the spot height of 703 metres shown on the map. Stupidly, I didn’t think to check the GR on the GPS, but I did notice, with it on the ground beside me, that it was showing a height of 708 metres.
It claimed to be using a number of satellites, and showed a positional resolution of 3 metres.

So, I’m just curious - is the displayed height a reasonable tolerance for the device? What would be the tolerance of an OS spot height, and would this be greater or smaller than the height tolerance of a trig point? Would the spot height necessarily be the highest point?

I almost always visit the summit point before activating, but occasionally I have relied on the GPS to confirm that I am within the AZ. The above suggests that it could be +/- 5 metres (or possibly more), though…

Your thoughts would be appreciated,


1 Like

In reply to G4AZS:

GPS vertical accuracy is nearly always worse than horizontal accuracy. It’s to do with the geometry of the satellites. I believe the rule of thumb is that vertical accuracy is 1.5x worse than horizontal accuracy.

If you have EGNOS enabled then +/-3m is typical horizontal accuracy ISTR. Typically that gives 4.5m vertical accuracy. On a summit with a clear view to the horizon, more sats can be seen, so vertical accuracy should be better.

But that is giving you the height on a WGS84 sphere and the Earth and UK isn’t so there will be further variations.

I can’t find the link now, but ISTR the OS height data is more accurate the nearer to Newlyn you are. I believe that it’s of the order +/- 7m by the time you get up to Northern Scotland.

In my case, if the summit was 800m ASL, then the AZ starts at 775m ASL. With my GPS I would assume I’m in the AZ when the GPS says 780m ASL. I wouldn’t argue that someone with a GPS saying 775m was not in the AZ but as an engineer I know there is some error and that extra 5m gives me greater comfort I was in the AZ. Normally I try to get to the top but that isn’t always possible.


In reply to MM0FMF:
Thanks Andy, that figures.
In the horizontal plain, I just use the 6 digit Grid ref, so any inaccuracy is not obvious. It hadn’t occurred to me to question the vertical accuracy - one of the pitfalls when presented with a specific number, the tendency is to accept it…


In reply to G4AZS:
It’s generally accepted that for the reasons given by Andy, Altitude accuracy for consumer GPS receivers should be considered +/-23meters with a DOP of 1.
Things only get worse with higher DOP figures.
That’s without differential or augmentation systems enabled.

I own a couple of older Etrex Vistas, the reason being that they have barometric altimeters. If these are set at a known elevation, and especially if the pressure is steady, then combined with the GPS altimeter they give a very accurate reading of altitude.
On the all the summits I’ve activated the trig point readings are always within +/- 1 metre.

There’s a very good reason aircraft do not use GPS to measure height above ground!

In reply to G4AZS:
A similar tale, I was was working in Norfolk last month and had to mark a location in a field with my GPS which gave the height as 65m, a week later a colleague was at the same location and gave the height from his GPS as 55m. The OS map shows a 60m contour running through the (very flat) field very close to the point we were identifying, the OS map I’d hazard a guess as being a tad more accurate than either of our GPS’s.
Iain, MM3WJZ

In reply to G4ISJ:

I own a couple of older Etrex Vistas, the reason being that they have
barometric altimeters. If these are set at a known elevation, and
especially if the pressure is steady, then combined with the GPS
altimeter they give a very accurate reading of altitude.

When using the barometric altimeter setting the correct height before starting the hike (or somewhere in between) is essential. This cancels out the air-pressure fluctuations. For the same reason QNH is part of the VOLMET information for aviation.
Newer Garmin models like the Oregon I am using do some tricky “autocalibration” which averages GPS height to track the altimeter setting. I do not know the exact algorithm behind but it works quite well (at least most of the time).

On the all the summits I’ve activated the trig point readings are
always within +/- 1 metre.
With the altimeter set correctly at a starting point I also reached this accuracy. Of course changing air-preasure (stromy days!) easily can lead to errors of several 10m in a few hours.

BTW: In German maps I consider contours only well educated guesses. For recalibration of the altimeter I only rely on trig points or height points.

73 de Michael, DB7MM

In reply to DB7MM:

I don’t know if it is there now, but in the bar of the Clachaig Inn in Glencoe there used to be a copy of one of the original Ordnance Survey maps of part of Glencoe hanging on the wall. There were no contours at all - I think they were invented in 1774 by Charles Hutton so I don’t know why they were missing, but the map had a large number of level traverses on it and some indication of crags. Given the sort of data on that old map I would agree with Michael - well educated guesses!


Brian G8ADD

In reply to MM0FMF:

But that is giving you the height on a WGS84 sphere and the Earth and
UK isn’t so there will be further variations.

Indeed so, but the distance between the GPS reference ellipsoid and the Geoid is something in the order of 45-50m in Great Britain. I dare say it is worse elsewhere. So even consumer grade GPS receivers tend to correct for this by having a built-in Geoid model. Newer devices are likely to have a more detailed model.

There’s a further correction to get from Geoid height to Ordnance Datum Newlyn, but I think this is less than a metre.

Measurement of height is a lot trickier than most people think. I would strongly recommend anybody interested in the topic to read this:

(especially the “myths” section at the beginning!)

In reply to M1MAJ:

FWIW: The model in my very old Garmin GPS-III+ is about 15m above actual sea level here on the western coast of North America. State Hwy 37 parallels the shoreline of San Francisco/San Pablo Bays * and the GPS reads 19m when the actual road elevation is almost exactly 4m.

The other cause for higher VDOP than HDOP is that for HDOP, the satellites are scattered all over the sky which improves determination of the intersections of the “spheres of position.” For VDOP, all the satellites are on one side *.

Fred “Skip” K6DGW
Auburn CA USA

In reply to K6DGW:
Hi Skip and all,

It’s a very good idea to compare your GPS readings with an accurately determined position and/or height. Do it frequently and record the readings.

This is a long post and I apologize for that. However GPS and barometer accuracy discussions could fill a book. The discussion of different globe shapes and what is best is another book.

Consumer GPS devices are less accurate than we would like to think. We have been seduced by digital displays and apparently stable readings.

Professional GPS surveying instruments use different signals from the GPS satellites cf consumer ones and if you pay enough money you can get one that will give a reading with an uncertainty of a few mm if left to do the number crunching for a day or more.

The resolution on the typical consumer GPS display, 1 m, should not be confused with the real resolution of the measurement. Any kind of resolution should not be confused with accuracy, now referred to as measurement uncertainty.

The internal number crunching has limited resolution, usually worse than a metre. More resolution costs more electricity and dollars, and using different satellite data.

The measured results calculated in the GPS typically vary randomly by a metre or more from one measurement to the next and readings often follow a rough cyclic pattern of maybe 10 m. You can reduce the random errors significantly by averaging the readings over say half an hour - if you have half an hour. However averaging over only 5 minutes helps.

The systematic errors are dependent on your device, the satellite constellation and the ionosphere. And of course there are temperature, altitude, battery voltage and local signal reflection effects to consider.

Oh and the Circle of Probability represents a 68% probability the true value lies within that circle. Double the COP to get a 95% probability. Multiply the horizontal error by at least 1.5 for the vertical error.

Cheap hand held altimeters are not necessarily better and their “accuracy” might be a 68% probability value, or it might have come from the Sales Dept. Use them only for relative measurements unless independently calibrated more than once.

I use both a GPS and a small altimeter both of which have claimed accuracies of 5 m vertical. I measure the AZ by walking down from the trig point or cairn until both indicate 30 m change. This is the most optimistic determination I can make. When I have the opportunity I recheck these measurements usually by immediately walking back to the peak.

Can you use spot heights or paper maps to check your GPS? Yes, but … Geodetic reference peaks will certainly be determined to within 1 m of the defined sea level, other peaks are less well determined. Official Australian government survey paper maps generally have a contour line accuracy of 70% of the distance between the contour lines and spot heights may be in error by up to 3 m to 5 m. I understand this meets international standards. Contour lines are often only at 20 m intervals except in some limited areas where 10 m or even 5 m might be available. Hence there is still plenty of work for surveyors to accurately determine local terrain for a variety of reasons.

If you use an instrument carefully they are a wonderful asset but if you rely on them without accounting for some uncertainty then you may be fooling yourself. As a rule of thumb I would take a reading of 30 m below the summit as maybe beyond the extent of the AZ and a reading of 20 m as being maybe within the AZ but I would really want to do a difference measurement between the known summit (cairn or similar marker) and the operating position.

Differential measurements always help reduce the systematic (offset) errors.

Good luck and stay on the tracks.


In reply to G4AZS:

I almost always visit the summit point before activating, but
occasionally I have relied on the GPS to confirm that I am within the
AZ. The above suggests that it could be +/- 5 metres (or possibly
more), though…

Hi Adrian,

I wouldn’t rely on GPS vertical accuracy. It is notoriously unreliable for all the reasons several others have set out.

If you want to check whether you are in the AZ I suggest you use a good old fashioned map and use the horizontal (grid ref) position to find a place which is above the relevant contour line.

I do use a good quality altimeter (I use a wrist worn Suunto Vector - other brands are available) which if regularly “spotted” at a known height can be incredibly accurate. A combination of altimeter and traditional map and compass work has proved to be efficient and accurate even in Scottish Winter whiteout conditions.

The key is to “respot” the altimeter at a definite known height as often as you can. You then get an idea of how the air pressure is generally trending as if you are constantly revising your height upwards the general trend is for air pressure increase and vice versa. Anyway, navigation by barometer is a whole subject in its own right.

In summary - GPS Height, take it with a pinch of salt.


In reply to AX3AFW:

Personally I rely on a “Jimmy”. If my GPS and Jimmy disagree by more than 1m I would throw the GPS in the nearest loch or stamp on it. In my experience the accuracy provided by Jimmy, even in the dark, far supercedes technology :o))

This, of course, explains why Tom, M1EYP, is navigationally challenged on his solo outings and has to go up The Cloud or Gun so often - his boots now have a memory of those routes and he doesn’t need a navigator!

Guess who has cabin fever!!


Barry GM4TOE

PS - a GPS is cheaper to run

In reply to GM4TOE:
Heh heh, made me grin.

When I first got a Garmin 12XL in the late 1990’s it was very revolutionary!

One of my good friends referred to it as “Clever Fred” and the name stuck. So we would play the “What does Fred say?” game…

This was in the days of “Selective Availability”… what a lousy idea that was!

Deliberately degraded accuracy meant using Fred could be “interesting”. Altitudes were regularly over 100m out!

Happy memories.

Wish I had a Jimmy, I always seem to be the one navigating if I am in a group.


In reply to DB7MM:
I am still none the wiser about finding position on a Deutscher Alpen Verein map as there are lines of longitude and latitude for a ‘grid’ and then regions of diamonds and triangles to make , I assume, a spherical correction. I was using map, compass and observation in the DL/MF region at Christmas 2012 becaus e i didn’t know how to make the link between GPS and the map. I should have got the Garmin Etrex altimeter out as an aid at least when we took a wrong path [20+ tons of felled trees sat on the path junction HI].

David M0YDH

The last thing that the GPS calculates is altitude and first is location. Makes sense of course one would rather know location than altitude, unless you are maybe in a hang glider … smile.

Rule of thumb is the altitude error is 1.5 times the circle of error. Of course each time you lock onto another bird, the accuracy gets better. As someone mentioned landing a plane with it could be hazardous to ones health.