GPS Position Accuracy

https://apps.garmin.com/en-US/apps/e3f85a7a-7241-4fd1-b6b3-39eddf26ed43

looks like u don't even have to code it. if you're using that way

See the API doc for a description of  what the levels mean.

QUALITY_USABLE = 3

Thanks for that, well that answers what the DOP is at the readings, but probably now not much use since we cannot tell if the GPS is using doppler speed, or a position calculated DOP effected speed.  Thanks for the help.!

FWIW I have done extensive testing of the data returned from the GPS on actual devices - F5 and VA-HR and am fairly certain that the speed and heading are sourced from doppler and not from two-position filtering.

My approach has been to compare a dead reckoning (DR) track created from speed, heading and time with the lat/lon positions and have found that whilst there is a not unexpected drift over time, speed and heading values were significantly smoother than would have been obtained from two-position filtering.

My application is yacht racing, so the speeds were in the 3-5 m/s range.

well, I spun my watch in a circle, so that it would record a doppler shift, should have recorded about 4 to 5 kts, and instead it displayed speed of zero....

I'm fairly sure, if it was a doppler shift measurement, then spinning your watch in a circle would give a speed, where as positional fixing would not because of the location relatively not changing

I wouldn't expect it to report speed and heading when working in such a small region at such a high angular velocity: they wouldn't be experienced on a 5-ton yacht, or a 200 Kg dinghy!

If you repeated the same experiment cycling around 200 m circle I think you would get some interesting results.

[EDIT]

And don't forget that whilst the satellites are broadcasting the PRN at 1 Khz (from which the watch's GPS chip is probably doing the doppler calculation) the chip is itself doing a bunch of filtering and only reporting at 1Hz - so spinning the watch is never going to produce a meaningful result.

What we'd all like to know is a bit more about what filtering is actually occurring on the chip!

The other thing I noticed, is the delay in reporting speeds when accelerating, and decelerating.

Now if it was using doppler shift, the errors would be tiny and reading instant... but it seems to be filtering it....which is what makes me think it's a positional based calculation. I was comparing it to another GPS that does use doppler shift, and mine was in about 3 second lag.

that and moving in a cirlce gives nothing...... 

One thing you need to considred here is what a watch is designed for.  It's not just pure GPS data all the time, and may even vary by specific sport.

Take running.  (There was a really good presentation about this at the last summit).  While running, your arm is moving up/down/forwards and back in relation to your body while you run.  While your body is moving in a specifc direction, it's different than your arm.  Then you need to look at GPS drift.  It's not down to the cm - not really even close to that, and second to second (that's how often GPS is read) you could be in the same range of accuracy or could be up to double that.

Start something like a run activity, and just stay in the same spot for a few minutes, then save the activity and look at the GPS track.  You'll likely see something that looks like a starburst, with short tracks in various directions, even though you're stationary.  It's the GPS drift.

Open Water Swimming is another example, as the watch is under water a good bit of the time and loses GPS, so there's "smarts" to handle that.

Nick said:

that and moving in a cirlce gives nothing...... 

How do you mean, "nothing"? Did you compare the GPS speed and heading with the two-position filtering?

I mean, the speed was reading 0.0 kts the entire time.

The speed I am displaying is the activity info speed with continuous GPS enabled, it's not a derivative of a sport or activity type in particular....

Can you think of any other way to categorically figure out if it's using one or the other?