Hi all,
New to this gps recording game, and I have been playing with a few analysis apps. Using the same gpx file, all three come up with different results!

Recorded on a Suunto Spartan watch.

On ka72:

On Windsport Tracker:

And on Windsurf Tracks:


Any tips or info on why this is happening, and how I should interpret this ?

Many thanks,
Simon

It's just a rough guess, I've zero familiarity with the device or windsport tracker, but I have a bit of experience with KA72, sometimes we see a small difference between KA72, and GPSResults, GPSactionreplay and GPSSpeedreader. But nothing like you've got there. Without seeing the file myself I'm only guessing, but I'd put my money on KA72. You can down load GPSResults, that's a good analysis tool, you can check the file for abnormalitys with that. There's a chance that KA72 is not filtering suspect data as hard as windtracker. GPSResults will be able to give you some idea of what is going on.

Watches can produce strange results if you use the underhand method, this may be part of the problem. If you have more errors on one tack than the other, that's probably the cause.

There's also 2 different sorts of data, track points and doppler. KA72 will use doppler if it's available, I've no idea what the others do, or if your watch includes doppler in the downloaded file. Again GPSResults will tell you that.

I use the GW60 and was surprised to see exactly the same numbers on KA72, GPS results V6 and Windsport Tracker.

I suspect your issue could be to do with the Suunto "Best" fix rate is 1/second.

I remember my results on Windsport Tracker were a little different until Matthias updated it to handle the GW60 faster recording rate.

Thanks for the info, I will try the GPSResults and see.

I'm sure the data from the watch is not as good as it could be, but I hadn't heard of the underhand issue. If I've only got the watch, what's the best spot for it? I could strap it to the mast I guess...

Cheers,
Simon

best spot for GPS reception is on your head pointing at the sky, failing that the upper arm facing slightly out ward. The upper forearm is the next alternative, it's better than the wrist as it doesn't swivel when you change grip.
Remery may have a good point if it's only recording at 1s intervals.

There is a strong suggestion from an SUUNTO watch analyst in Finland that what it produces in not Doppler data as is reported on some analysis programs, but is actually 'smoothed' or filtered positional points (Trackpoints).

But that does not really account for your discrepancies.

The GPX files (and of course those from a Suunto) dont contain the Doppler error data that many analysis programs use to find and filter or flag errors in the data. So analysis programs that do this can't correct or flag all types of errors with GPX files.

+ 1 for Decrepits recommendation of GPS-Results. It handles GPX files and should be consistent and help to find problems.

Sample rate of device and how the software interprets it.

...and how the speed numbers are determined. Doppler is the method of choice. Trackpoint dx/dt is not the method of choice.

Hi All,

Thanks for all your feedback - I've had a chance to run it through GPS Results and interestingly, it is almost identical to ka72 when I have the Display Traces checkbox marked off, but when it is on, the numbers come way down - for example the 2sec peak is 28kts without the checkbox, and 8kts with.


www.seabreeze.com.au/img/photos/windsurfing/16723303.jpg' />

At the end of the day I'll work with the Suunto I have, however, flawed the data - but it's great to have a better understanding of the mechanics of the data!

Cheers,

Simon

I think the checkbox you are talking about is actually controlling filters, not the display of traces. Without filters, you get the higher numbers. With filters, GPSResults recognizes that some of the data points are probably artifacts, and does not use them, hence the lower numbers. Unfortunately, every program out there has a different implementation of filters.

In your case, there are two likely causes for the discrepancies. One is missing points - the watch did not record a point every second. Since ka72 supports data from all kinds of watches, regardless of how often the log data points, it is more tolerant towards that.

The second possible cause is that there is a huge jump in speeds, which is then caught by the acceleration filter in GPSResults. That can happen in trackpoint data somewhat randomly, for example when the satellites used change. It also happens all the time when watches loose GPS reception in crashes because they are under water. When they regain reception, the initial coordinates are often very different from the last point, resulting in a spike in speeds. Doppler speeds are much less prone from such errors (but not totally safe), but most watches either don't use doppler speed at all, or use some kind of mix that also includes trackpoint-based speeds.

You can usually get an idea what's going on by looking at the data points in GPSResults. Clicking on the top speed will result in the points being selected, so you can check the time difference and acceleration.

Select to expand quote

BakerSailor said..
Hi All,

Thanks for all your feedback - I've had a chance to run it through GPS Results and interestingly, it is almost identical to ka72 when I have the Display Traces checkbox marked off, but when it is on, the numbers come way down - for example the 2sec peak is 28kts without the checkbox, and 8kts with.


www.seabreeze.com.au/img/photos/windsurfing/16723303.jpg' />

At the end of the day I'll work with the Suunto I have, however, flawed the data - but it's great to have a better understanding of the mechanics of the data!

Cheers,

Simon

I find the sunnto records random very high peaks which would throw the average off if not smoothed out.

Typically I see this kind of mixture of results when a GPS is not set to record data every second (or more frequently.) By default most GPS watches are configured to record data less frequently than every second because that takes up less space I guess, and is sufficient for most activities. Opening the file in GPSResults would tell you immediately whether this was the case. Look at the recorded trackpoints on the left side if the screen and verify they are all 1s apart.

I'd have to say that 95% of the time, this is the problem.