Good post Mark, thank you. My Garmin ETrex seemed notoriously inaccurate when I used it in the boats - on the Google Earth trace I posted some time ago it showed a maximum elevation of 8 metres and a couple of other peaks around 5 metres. The track was recorded on a windless day with a pretty much flat sea. Five and 8 metre peaks? I'm not that brave!
Yambo,
It's just as bad when you're on land, but you aren't as aware of sea level and so the error isn't as obvious.
The US FAA has measured actual vertical accuracy in GPS to be 4.7 meters, which is right in line with your errors. If you had access to WAAS, which only works in the Western Hemisphere, it would be 1.3 meters.
There are also errors caused by the difference between the actual shape of the Earth and the mathematical model of the shape used in the calculations. The WGS84 model currently the most used in civilian receivers is supposed to accurate overall to +/- 1 m.
Measurements taken on the water are also affected by how seal level is defined in the mapping application. It's almost guaranteed to never match up with the actual water level due to tidal variation.
So your vertical errors are pretty reasonable for how the system works and not the fault of your receiver.
Laszlo
An addendum about the Wood Duck 12 sprint speed in light of Mark's excellent discussion of GPS inaccuracies.
I did indeed get that value using the max speed function of my GPS unit. However, it was a 50-satellite receiver tracking 18 satellites at the time, WAAS was enabled and the correction had been applied to all the satellites and it was a 5Hz unit (though the display and NMEA string output were 1 Hz).
There was no wind, it was low slack tide, the course was a straight line orthogonal to the direction of tidal flow. There was an unobstructed view of the entire sky from the water and the run was on the order of a minute. Just before I started, I observed the receiver to verify that it was stable (no sudden changes in position or velocity) and I reset the trip computer function to clear all previous data.
I was seriously trying to get an accurate reading and, like Mark, I've had some experience with GPS and its vagaries. That's why all the elaborate setup and noting of the conditions. So, while I can't give an exact statement of error for the reading, I'm pretty confident that it's in the right ballpark.
Laszlo
Thanks Laszlo. the GPS is no longer with us and isn't really missed. I'm very much a map/chart and compass guy although not a complete Luddite - I use Navionics quite a lot in my fishing boat.
I totally agree that "maxium speed" as measured by a phone GPS is worthless. But I do think that a sprint speed measured over at least 30 seconds -- (preferably more but who can "sprint" for much longer?) -- is worth measuring. I've found it to be fairly consistent in my own testing and I'm glad that Mark agrees: "Once stabilized for 30+ seconds on a constant course at a constant speed, the speed displayed should stabilize and be reasonably accurate."
As an avid cyclist, the most distressing part of such measurement errors is the great uncertainty about elevation gain data. Strava basically gives up on this and defaults to elevation data from maps unless the cyclist is using a barometric altimeter. I'd appreciate further insight on this, but I should think that the margin of error for tiny barometric altimeters is quite substantial, leading to very suspect data in recording the short hill climbs of most cycling routes. . . . I realize this it way out of bounds for a forum on boats. But I doubt I'm the only peddler/paddler on the Forum.
As an alternative, I'll be spending my winter months trying to wire a Garmin paddle-wheel sensor and some accelerometers to an arduino board. In theory this should resolve speed fluctuations at sub-second intervals. In practice... we'll see!
If you get that done, please do report back. There are some interesting challenges with that approach and it will be fun to see how you solve them.
Laszlo
Compass and GPS might be right up there with the dressing table mirror and the bathroom scale in terms of being devices which make no concession to human vanity. <;-)
.....Michael