I had a fin session at the same place where i supposedly hit 36 knots on foil. Even gps speed surfing said i hit a 36knot 2sec that day




Anyway on fin my watch said i hit a 63knot 2 sec. I didn't have the balls to submit to gps site to see if it would of approved the 2sec

There are high tension power lines right in the area where i hit that speed. Maybe throws off the gps

Not according to the following paper I found in a Google search. Although the testing was done on static a GPS. It may be different for a moving GPS.

www.tandfonline.com/doi/abs/10.1179/sre.2001.36.279.54

I doubt high tension power lines would have any effect under normal conditions. But I guess a high energy arc over would create a very broadband transmission over the electromagnetic spectrum. This could certainly corrupt the satellites signals. But decent filters should remove such spikes. What analysis program are you using?

Select to expand quote

decrepit said..
I doubt high tension power lines would have any effect under normal conditions. But I guess a high energy arc over would create a very broadband transmission over the electromagnetic spectrum. This could certainly corrupt the satellites signals. But decent filters should remove such spikes. What analysis program are you using?

I didn't run the file through any analysis, I should do that.

I put it in GPSresults V6 and it said I had a 2sec of 34.7 knots. Pretty certain I did not go that fast.

Not sure what gps-speedsurfing does but they said it was a 63knot 2sec. I did not leave it there for very long so maybe it would have been reviewed by their software.

very dodgy looking data, I wouldn't trust it to be within a couple of knots. Those holes in the speed graph are where the data falls outside GPSResults's filters.

Ahhh I see it's a coros, that sort of explains it.
GPSS>3 isn't going to say it's bad data, as they have a financial interest in that watch.

Select to expand quote

decrepit said..
Ahhh I see it's a coros, that sort of explains it.
GPSS>3 isn't going to say it's bad data, as they have a financial interest in that watch.

Seriously?

Select to expand quote

decrepit said..
very dodgy looking data, I wouldn't trust it to be within a couple of knots. Those holes in the speed graph are where the data falls outside GPSResults's filters.

Just an example below that not all data is filtered correctly no matter what software you use, example from GPSspeedreader which did not filter a spike. Not that I'm qustioning the 234knot speed, seems valid to me to post And no this wasn't posted by the sailor but goes to show we shouldn't be jumping to conclusions.

Select to expand quote

Aus501 Boz said..
Just an example below that not all data is filtered correctly no matter what software you use, example from GPSspeedreader which did not filter a spike. Not that I'm qustioning the 234knot speed, seems valid to me to post And no this wasn't posted by the sailor but goes to show we shouldn't be jumping to conclusions.

How well any software filters works depends entirely on the quality of the GPS used. The data you show seem to be from a Locosys unit, since the error estimates for 2 sec and alpha are way above the thresholds for u-blox based units like the Motion and the ESP32 DIY loggers. Looks like GT-31 data to me. That's 20-year old technology that only worked reasonably well because of pretty heavy filtering in the firmware (which is somewhat similar to the Coros in this respect).

Select to expand quote

thedoor said..
Anyway on fin my watch said i hit a 63knot 2 sec. I didn't have the balls to submit to gps site to see if it would of approved the 2sec

Your 63 knot top speed is during a swim near the end of your session. Artificial peaks like that are very common when GPS units loose reception because they are pushed under water, and then regain reception. The first fix will always be with a small number of satellites and therefore low accuracy. It is very common that the position for this first fix is wrong by dozens to hundreds of meters, as can also be seen in your second image. That happens even when miles away from the closest power lines.
The accuracy then improves quickly as more satellites are added, and the position is adjusted accordingly. These adjustments lead to large spikes in positional speeds, and quite often, similar spikes are seen in "doppler" speeds (a possible indication that "doppler" speeds are not only based on doppler measurements).
Note that some GPS units do not seem to report satellites used in "real time" during crashes. Your watch seems to fall into this category, since the number of satellites never changes during the swim, even after missing 28 seconds and then 5 seconds worth of data points.

With GPS units based on the u-blox chip and the current version of GPS Speedreader (which has stricter filter settings than older versions and GPSResults for 5Hz+ u-blox data), such "swim spikes" are almost always filtered out, since it it extremely rare to see artifact speeds with a low error estimate. The Coros watch does not provide any error estimates, to the only filters that can be used are time errors (the 68 knot peak is after 5 missed peaks) and acceleration. The acceleration filter often fails in cases like this, since GPS chips tend to keep artifact (i.e. guessed) speeds at the same level. So your 2 second speed estimate from GPSResults ends up about 4-5 knots too high. Better than the unfiltered speed of 63 knots, but still wrong.

If you have two boom-mounted u-blox based loggers, and use GPSSpeedreader to calculate "intelligent averages" from both units, the chances that you get any wrong speeds during crashes or swims are very close to zero. That's because even if one of the two units is under water and starts to make up artificial speeds, the other unit on the other side of the boom is above water, and still produces accurate data. Speedreader can easily identify which of the two units has good data from the accuracy estimates.

If you have just one GPS, then mounting the GPS to the top of the helmet will avoid most swimming artifacts. The one exception is when your head goes under water in a good crash - but that much less frequent, and usually shorter, than submerging a GPS worn on the upper arm or the wrist.

Select to expand quote

thedoor said..
I had a fin session at the same place where i supposedly hit 36 knots on foil....
Anyway on fin my watch said i hit a 63knot 2 sec.

Thanks for submitting two excellent examples of why the Coros watch would be problematic for use in competitions. While the speed is overstated by so much in your examples that it is quite obviously wrong, the bigger concern for competitions would smaller artifacts that are less obvious. Spikes liked to swimming or crashing could probably be identified quite well by better software filters. That does not really help the GPS Team Challenge, though, since sessions can be posted from a variety of different programs and web sites.

Select to expand quote

boardsurfr said..

Aus501 Boz said..
Just an example below that not all data is filtered correctly no matter what software you use, example from GPSspeedreader which did not filter a spike. Not that I'm qustioning the 234knot speed, seems valid to me to post And no this wasn't posted by the sailor but goes to show we shouldn't be jumping to conclusions.

How well any software filters works depends entirely on the quality of the GPS used. The data you show seem to be from a Locosys unit, since the error estimates for 2 sec and alpha are way above the thresholds for u-blox based units like the Motion and the ESP32 DIY loggers. Looks like GT-31 data to me. That's 20-year old technology that only worked reasonably well because of pretty heavy filtering in the firmware (which is somewhat similar to the Coros in this respect).

thedoor said..
Anyway on fin my watch said i hit a 63knot 2 sec. I didn't have the balls to submit to gps site to see if it would of approved the 2sec

Your 63 knot top speed is during a swim near the end of your session. Artificial peaks like that are very common when GPS units loose reception because they are pushed under water, and then regain reception. The first fix will always be with a small number of satellites and therefore low accuracy. It is very common that the position for this first fix is wrong by dozens to hundreds of meters, as can also be seen in your second image. That happens even when miles away from the closest power lines.
The accuracy then improves quickly as more satellites are added, and the position is adjusted accordingly. These adjustments lead to large spikes in positional speeds, and quite often, similar spikes are seen in "doppler" speeds (a possible indication that "doppler" speeds are not only based on doppler measurements).
Note that some GPS units do not seem to report satellites used in "real time" during crashes. Your watch seems to fall into this category, since the number of satellites never changes during the swim, even after missing 28 seconds and then 5 seconds worth of data points.

With GPS units based on the u-blox chip and the current version of GPS Speedreader (which has stricter filter settings than older versions and GPSResults for 5Hz+ u-blox data), such "swim spikes" are almost always filtered out, since it it extremely rare to see artifact speeds with a low error estimate. The Coros watch does not provide any error estimates, to the only filters that can be used are time errors (the 68 knot peak is after 5 missed peaks) and acceleration. The acceleration filter often fails in cases like this, since GPS chips tend to keep artifact (i.e. guessed) speeds at the same level. So your 2 second speed estimate from GPSResults ends up about 4-5 knots too high. Better than the unfiltered speed of 63 knots, but still wrong.

If you have two boom-mounted u-blox based loggers, and use GPSSpeedreader to calculate "intelligent averages" from both units, the chances that you get any wrong speeds during crashes or swims are very close to zero. That's because even if one of the two units is under water and starts to make up artificial speeds, the other unit on the other side of the boom is above water, and still produces accurate data. Speedreader can easily identify which of the two units has good data from the accuracy estimates.

If you have just one GPS, then mounting the GPS to the top of the helmet will avoid most swimming artifacts. The one exception is when your head goes under water in a good crash - but that much less frequent, and usually shorter, than submerging a GPS worn on the upper arm or the wrist.

thedoor said..
I had a fin session at the same place where i supposedly hit 36 knots on foil....
Anyway on fin my watch said i hit a 63knot 2 sec.

Thanks for submitting two excellent examples of why the Coros watch would be problematic for use in competitions. While the speed is overstated by so much in your examples that it is quite obviously wrong, the bigger concern for competitions would smaller artifacts that are less obvious. Spikes liked to swimming or crashing could probably be identified quite well by better software filters. That does not really help the GPS Team Challenge, though, since sessions can be posted from a variety of different programs and web sites.

I swim a lot :) and I have never had this sort of 2 sec before, so I am not sure I agree. The only time I got it was exactly this location on that river. The last one being the 63knot 2 sec came up live on my watch while riding. The watch sends buzzes when you increase a stat.

Select to expand quote

dedekam said..

decrepit said..
Ahhh I see it's a coros, that sort of explains it.
GPSS>3 isn't going to say it's bad data, as they have a financial interest in that watch.

Seriously?

No financial relationship between GPSSpeedsurfing and coros as far as I understand. I think they have sold devices before, but more than just coros.

I guess if there was a financial conspiracy they would ban all other devices from their site

Just blame the war in Ukraine, GPS's are being played with by the yanks to stuff up the reds

Just a thought, power lines near salt water, will probably get a buildup of salt and dust on the insulators. Combine that with any moisture and you'll get an ark over. This could definitely affect the gps signal. In some places they use helicopters to clean the insulators with a water spray, before that starts happening.

Select to expand quote

thedoor said..
I swim a lot :) and I have never had this sort of 2 sec before, so I am not sure I agree. The only time I got it was exactly this location on that river. The last one being the 63knot 2 sec came up live on my watch while riding. The watch sends buzzes when you increase a stat.

Have a closer look at your data, and you'll see more of the artifacts. There is one region earlier in the same run that's likely to also be an artifact during a swim:


Typically, the speed in such artifacts depends on how long the GPS is or was under water, with higher speeds after longer times under water. With a wrist-mounted GPS, that's usually just a short time before the hand is close enough to the water surface again for the GPS to get a fix. Your 63 knot artifact was the exception, since the GPS had lost reception for 28 seconds.

With 2 boom mounted units, it's more common that one unit remains under water for a longer time, so larger artifacts are more common. Here is an example with a top speed of almost 600 knots:

The big artifact is at the start of the session, where the sail was in the water for a while while I was adjusting things. The position error was more than 2 miles, after no GPS fix for 24 seconds. The GPS got 3 satellites when it reported the 597 knots, so it only had a 2-D fix and very high errors. But there's another error visible near the end of the track visible in the other (blue) GPS track. The speed went only up to 74 knots for 3 points (0.6 seconds), but the GPS got a 3D fix this time.

Similar artifacts can be seen after almost every crash with 2 boom units when one boom remains in the water for a few seconds. Most of the time, the artifacts are smaller, and often, they are easy to identify as artifacts by looking at the GPS fix, acceleration, or missing points. But sometimes, these filters (and additional filters Coros uses) fail, while filters using accuracy estimates still work.

With a decent quality GPS, most speed artifacts are linked to crashes and/or loss of reception. Those artifacts are not always easy to spot in a single track. The dual boom-mounted GPS units are great here because the problem generally affect only one of the GPS units, and the differences are easy to see.

Been using gps on bikes for years and noticed there's some areas where the signal always gets messed up, I can guarantee that one certain sections of road it will loose signal or show some stupid high speed, never worked out why.

Select to expand quote

Ben1973 said..
Been using gps on bikes for years and noticed there's some areas where the signal always gets messed up, I can guarantee that one certain sections of road it will loose signal or show some stupid high speed, never worked out why.

Yeah. If it was device or water related I would expect a more random stretch of overshoots. I have only sailed that spot about three times this year and these had my only too overshoots at that exact location

Nice thing about GPS on water is that you generally have the absolute best conditions for grabbing satellites. "Clear view of the sky." No canyons or trees or buildings to block the signals. Now, power lines over water could be an issue. I have never noticed it.

People talk about having the GPS watch on the underside of the arm, or signal blockage by boom or sail or mast. I have never noticed that, either. But then I use GPS only for recreation and archiving. I don't need the absolute ultimate in accuracy or HDOP.

I'm fairly sure 50hz or 60hz radiation from power lines will have no effect on GPS reception.
However a small arc over on dirty moist insulators probably will.

Select to expand quote

Ben1973 said..
Been using gps on bikes for years and noticed there's some areas where the signal always gets messed up, I can guarantee that one certain sections of road it will loose signal or show some stupid high speed, never worked out why.

One thing that will mess up the GPS signal when biking or driving are trees. Here's an example from a test drive:
In this region, the trees next to the street are dense, and the leaves mostly cover the road so you can barely see it. It's a narrow 2 lane road, so the distance between the two tracks shows that the position had a relatively high error, too. All of the speeds are quite noisy, and one of the three GPS units was off by about 3 knots for 3 successive points (0.3 seconds).

For comparison, here's data from a bit earlier in the same drive where there were very few trees next to the street:

Error estimates are much lower, positional accuracy higher, and the differences between the GPS units are much smaller.
These results are very typical and can be seen on just about any test drive where trees get close to the street.

Dense tree canopies block the GPS signal, since leaves are mostly water, so having a GPS under trees is somewhat similar to pushing a GPS under water. Since leaves are thin, the tree canopy has to be pretty dense before you loose reception completely. Accuracy can be affected even if trees are only at the sides and don't cover the street (or bike path), since they can weaken or block the signal from all satellites that are not directly above the GPS. Having only satellites above, but not to the sides, affects both triangulation (position accuracy) and the doppler signal (which is highest when moving directly towards or away from a satellite).

If you are biking (or driving) in cities, houses, bridges, etc. will also cause problems by blocking GPS signals. In addition, they can reflect the GPS signal, which can cause multipath problems, which cause noisier signals.