Bonjour,
Je me permets de reproduire ici une conversation que j’ai eue sur le sujet avec Lindsay Griffin, de l’American Alpine Journal. Nous discutions des relevés altimétriques que j’avais effectués en expé (Antarctique, je précise cela pour le fait que l’on parle de latitudes assez élevées, entre 60 et 65°S), où j’avais vu de grosses différences entre ma montre alti Suunto Observer recalée chaque matin au niveau de la mer et mon GPS Garmin Map 62.
[quote]I believe for reliable, ie small error, reading, you have to carry a survey GPS. If you can wade through the below, it’s actually quite interesting
Camilo Rada is THE MAN on GPS, and was sent by the Omega Foundation to measure the altitudes of all the Vinson Massif peaks. He is Chilean, so his English is not perfect (I read it three times and I’m not sure I understand 100 percent!). The party he is referring to is Slovenian, connected with a peak in India.
So, you are saying your altitude readings are primarily from your watch?
Cheers
Lindsay
[quote]The errors reported by this kind of GPS are estimated 95% of the time. Therefore there is chances of a measure to lay in the 5% of the samples over the reported error. But taking samples for several minutes as they did, basically makes sure that the reported errors is close to the real maximum error.
With my own Garmin GPS, I’ve done several tests with 1 minute averages against survey grade GPS, and based on that experience, I think that in HORIZONTAL POSITION their measurement has an error very likely to be under 3 meters. Also in the summits the GPS reception is exceptionally good so 2-3 meter accuracy with user grade GPSs are achievable.
But the elevation is a whole different thing. The garmin GPS I have, does average the elevation when set in the position averaging mode, but I know other models only work on the horizontal coordinates. I don’t know what’s the case with the GPSMAP 60CSx. He can do some simple tests at home to check it.
That GPS have an integrated barometer, so all elevation readings are a mix between GPS and barometric measurements. This is intended to be more accurate the just GPS but it has some problems.
It works like this:
GPS elevations samples are very noisy, jumps up and down, but the average is right, this means there is no bias. Barometric elevations samples are very smooth and stable, but the average is wrong, due to the bias produced by regional pressure changes.
So Garmin GPS units show always Barometric readings, and the bias is calculated and removed by using the average of the difference with GPS readings over a period of time. Therefore, if you keep the GPS on during the whole ascent I can spect the barometric bias estimation to be pretty good when I reach the summit, and the elevation at the summit after averaging for few minutes could have accuracies as good as 10 meters.
But if I turn the GPS on at the summit after one or two days off, the barometric bias calculation can be completely outdated and off, leading to errors easily in the range of 100 meters or more.
Each GPS unit deal with this a bit different, some time when the bias calculation is outdated they give more weight to the GPS reading or use other approaches, so it is difficult to say how accurate is the elevation reading. But if they kept the GPS on during the ascent, I would guess an accuracy between 20-30 meters could easily be achieved by averaging.
What I do, is to have a look of the pure GPS elevation reading at the summits, but each model needs a different trick to access that figure.
I hope this clarify a bit the issue.
Cheers
Camilo[/quote]
[/quote]
En résumé extrêmement rapide, les mesures d’altitude par GPS sont très bruitées, avec beaucoup de précision, mais en moyenne elles sont exactes (pas de biais). Par contraste, les mesures barométriques sont très stables et très lisses, mais peuvent être biaisées (écart avec la vraie valeur). Un bon appareil GPS va alors combiner le meilleur des deux types de mesures pour calculer l’altitude, c’est-à-dire évaluer le biais par comparaison des mesures satellites et barométriques, et corriger ainsi les mesures barométriques.
Bref, il faut vraiment savoir dans le détail comment l’appareil fait sa cuisine avant d’afficher une altitude. Et en mouvement, rien ne vaut la précision d’une mesure barométrique, si tant est qu’elle soit bien calée, bien sûr.
En espérant que cela ait fait avancer un peu le débat. 
Amicalement,
A.
[/center]