And if you try importing a GPX from a Garmin then you can get some very odd figures (mate had one where it looked like it was reporting the ascent in feet as if they were metres - resulting in an ascent three times higher than it should have been.

Reported it to Endo Support and they just said they knew there were "some inaccuracies" :roll:

Everytrail iirc was one that simply added the +/- differences between every consecutive data point in the file. Data points were typically at 1s intervals, but several may be missing due to bad lock, etc.
Endomondo iirc smoothed it a bit.
Strava invented some extra relative height differences from somewhere else.
edit: not sure what the eTrex does - think it might be just the sum like Everytrail uses.

The exact same file was used for all 3.

Those spot checks seem pretty accurate, certainly accurate enough. Mine gives a reasonably consistent +40m offset.

Can anyone explain why my last endomondo trip decided my altitude dropped to 281m below sea level? The only descent I did was off a bridge crossing the railway about halfway thru a ride. (can sort pic later).

Incidentally, does anyone know if there is some standard way to work out Total Ascent/Descent?

Obviously it is the total of every up/down but at what sort of distance/time resolution?

i.e. the ground is never totally flat, but at what resolution does/should it count towards total ascent/descent?

For example, crossing a road could add 20cm to ascent and descent for the kerbs, plus another say 50cm for going up and down the road camber, but presumably it normally doesn't (partly because the GPS isn't precise enough to detect it).

http://i1015.photobucket.com/albums/af280/AdrianH2010/3a6b90bdd7170f16c4d2f1297dd51444.jpg
http://i1015.photobucket.com/albums/af280/AdrianH2010/400d7e5e42f42f28ee01ce878ad8b1fa.jpg

The trig point is 97m, or 318.24ft, and my phone shows it at 315ft, after it had settled down and aquired enough satellites. Accurate enough for most people, I would think.

To get an accurate fix, atmospheric conditions must be taken into account. With my Leica Smart Rover equipment, ( RRP £20,000) I also need to pay for a smart net licence that is IRO £2,000 PA.

This dials into the smart net network via a mobile phone to a series of base stations to enable it to gain accuracy of between 10-15mm. If you have no GPRS phone signal, forget it.

I have other GPS equipment that is only 4m accurate unless its sited over known stations with GPS coordinates. This is essentially your car or bike sat nav but I have 2 that interact by radio with one another to gain fix. Their E an N are still only accurate to 3 or 4m and elevation to 5 or 6 m ( due to atmospheric conditions ) unless told where the base is, either by correction, eg set up, survey, manipulate on drawings like CAD, apply adjustments or as above with a smartnest licence. Or with a previous survey E.g, set up base, apply coordinated, survey knowing you are effectively inaccurate, apply adjustments on the computer later to gain true OS coordinates of base station location and carry on with survey.

My base and rover kit costs over £20,000 new. My Leice kit, similar plus the licence, paid annually.

What do you expect for a few hundred quid from your bike GPS?

Elevation is the hardest to get accurate BTW, E and N, 3-4m generally.

http://i1015.photobucket.com/albums/af280/AdrianH2010/5fdef34807d0030b28d2bf2f6f230017.jpg

I used to suffer a similar thing with my 705 on wet rides, turned out it was water blocking the barometric sensor holes, putting a plastic bag over the GPS cured the problem but made the screen hard to see. :roll:
the rubber case thingy for the 800 seems to stop the problem but makes the alarms hard to hear.