Select to expand quote

decrepit said..

I think there's a need to think if the SDoP data is really necessary, for non record breaking results.

As a matter of fact, they are not. The only software that definitely uses SDoP, AFAIK, is GPSResults. GPS Action Replay does not seem to even read SDoP data - it does not show them in the track point table, nor does it calculated accuracy (+- numbers) as GPSResults does. ka72.com does not show accuracy, either, so I'd assume it does not use SDoP. Even in GPSResults, it only takes one mouse click to ignore the SDoP values completely.

It's probably safe to assume that the vast majority of speed results posted on GPSTC was calculated without SDoP data.

SDoP for records? Absolutely, a must. For testing devices? Nice to have. For routine GPSTC postings? See above.

Amen

Select to expand quote

boardsurfr said..

decrepit said..

I think there's a need to think if the SDoP data is really necessary, for non record breaking results.

As a matter of fact, they are not. The only software that definitely uses SDoP, AFAIK, is GPSResults. GPS Action Replay does not seem to even read SDoP data - it does not show them in the track point table, nor does it calculated accuracy (+- numbers) as GPSResults does. ka72.com does not show accuracy, either, so I'd assume it does not use SDoP. Even in GPSResults, it only takes one mouse click to ignore the SDoP values completely.

It's probably safe to assume that the vast majority of speed results posted on GPSTC was calculated without SDoP data.

SDoP for records? Absolutely, a must. For testing devices? Nice to have. For routine GPSTC postings? See above.

So we now have the question at the heart of all this, the next question is, "How do we decide the answer????"

The GTC isn't exactly a democratic organization, and the administration is a bit undefined, now that Hardie and Nebs have pulled out.

My vote is for a bit of patience to see if any of the alternatives in the pipeline are viable.
I agree that SDoP is a great ideal to strive for, but although I enjoyed using my GW52 test unit i was disappointed in it's life span, (Admittedly the pouch got a little damp just before it died, which could have been my fault.) I still find myself reluctant to buy a replacement. So I'm sort of straddling both camps.

Yes , we should be using the concept of 'claimed speed' by deducting the error from the raw result. GPS Results can that now. Even RealSpeed can do it! Everyone else should catch up!

That aside, the 'SDOP' data is vital for us to tell if a questionable run is valid or not. Of course, we won't know that unless we run it in software that tell us that like GPS-Results, but when someone queries it, we have a definitive tool to make a judgement.

But more importantly, it tells us directly the actual accuracy confidence of any run and the GPS device. When analysed in GPS-Results, it immediately signals if a run is invalid, a 'spike' or just junk data.

I can tell you unequivocally that the 99% of the time, a 10 second run from a GT-31 worn correctly is accurate to less than 0.25 knots, and that a GW-52 is accurate to less than 0.05 knots. Not only that, I can tell you that from hundreds of actual, on the water side by side tests of both, that the difference between multiple units is, 99% of the time, far less than the error value and commonly in the hundredths of a Knot range. Rarely is it in the tenths of a knots range. Tom Chalko gathered hundreds of KM of side by side tests with 6 GW-52's and calculated that the average difference was just over 0.02 knots. I am yet to see even one similar side by side test of multiple non 'SDOP' capable GPS.

If we use 'claimed speed', when a new GPS device comes along that includes 'SDOP' data, all we need to do is verify that the error data is valid and then just use it! If it has high error, we just deduct it. If it has really high error we just invalidate the run. Simple!

If a new GPS device comes along that does not have 'SDOP' data we can get a rough picture of how accurate it is by side by side testing with a GPS that does produce 'SDOP' data. We can improve that picture by multiple side by side speed testing of half a dozen or more of the new device for a few million points and comparing the variance. We might develop a certain level of confidence in it in that situation. But when we use that GPS on it's own, we have no actual quantifiable data on how accurate any individual run is. It could be a PB, or it just might be a GPS glitch. Sure, a lot of the time, or even most of the time, it might be correct. But if it was your new PB of 41 knots when your last PB was 38 Knots, wouldn't you want to know for sure? Or are you happy to kid yourself?

Select to expand quote

sick_em_rex said..
Don't forget, GPSSS exists for record chasers, GPSTC is our fun comp meant to encourage participants.

The rules for use of GPS devices for GPSSS and GPSTC, as stated on their respective web sites, are almost identical and don't reflect this differing philosophy

I always knew my Canmore was robbing me of speed lolololololol, but I went to buy a GW52 6 months ago but was told that they are not Mac compatible - is this true ? That is why I got the Canmore.

Select to expand quote

Tinlyds said..
I always knew my Canmore was robbing me of speed lolololololol, but I went to buy a GW52 6 months ago but was told that they are not Mac compatible - is this true ? That is why I got the Canmore.

You need a windows emulator of some description to run them, Sailquick is doing this.
That's another of it's big draw backs. If you're a mac person, not so bad for linux users because they can dual boot windows more easily, (but it's still a pain in the bum)

Select to expand quote

decrepit said..
>>>>>>
I still find myself reluctant to buy a replacement. So I'm sort of straddling both camps.

@#%^^%$##, must have cursed myself, because this arvo I sent my GT31 into the next dimension somehow. I was sure it was in my jeans pocket, but when i got home, I changed out of then because they were wet. After a while when I wanted to crunch the numbers the jeans pocket was empty! Trouble is I have no memory of exactly what I did after removing the gps from it's pouch.
So now I'm going to buy myself a GW52 and keep it very dry!

I have a Mac and run parallels desktop to run Windows 7 from that I can use navi link & GW52 utility with no worries.

Select to expand quote

Windzup said..
I have a Mac and run parallels desktop to run Windows 7 from that I can use navi link & GW52 utility with no worries.

Same here. Have a Macbook Pro but I run Windows XP in VMware.

Wouldn't it be easier to setup fields which actually work to a GPS's strengths? at present there are too many divisions is 100m 250m 500m NM alpha's etc why not have fields which are a lot more accurate? There is no way of having exact accuracy I find when I ride my bike there is always .4km difference on my garmin 510 compared to my mate I always ride with and when using strava on my phone it can be out by almost 2km compared to the garmin.

It really wouldn't take much effort to build a new website, where people can join and simply enter their data manually. The only rule would be that you need to have the file for checking, but other than that any devices is allowable. Plausible? I think so.

you mean something like ka72 Nigel ??

files are available and all devices are accepted, doesn't do rankings like gpstc though.

But I'm quite happy with my New Cal group, no point in comparing our data with another team from somewhere else in the planet.

Just waiting for Dylan to come with a yearly/gender ranking and we'll be fine.

Select to expand quote

seanhogan said..
you mean something like ka72 Nigel ??

files are available and all devices are accepted, doesn't do rankings like gpstc though.

But I'm quite happy with my New Cal group, no point in comparing our data with another team from somewhere else in the planet.

Just waiting for Dylan to come with a yearly/gender ranking and we'll be fine.

Yeah mate.... KA72 works for me too. I'm in the Budgie Smugglers group. I like that it's really just a bit of fun with others who've sailed in the same conditions from the same lake and generally on the same day. Like you say, I only ever compare myself with other local (NSW) sailors, not the other states, so KA72 groups work great for that.

We need a Botany Bay group when summer kicks in

agree with Andrew, unless they allow watches then I won't be posting on GPSTC either when my GT31 dies... I'll just post on KA72. It's a bit sad, but if GPSTC rules aren't modified, it'll be dead in 5 years and everyone will be using KA72. That's just progress though.

It would be interesting to see some usage statistics on the GPSTC, if the number of people posting is increasing or decreasing over the last few year.

That's fairly easy, just filter the overall rankings by year, and for this purpose I've also selected Australia.
2016 434, but it's not a full year yet, there may be a few more yet.
2015 462
2014 463
2013 451
2012 398
2011 325

So that looks like a rapid rise to 2014, a small plateau then the start of a drop off.

I started sailing GPS in 2004 when Ian Fox and Mal Wright saw the advantage of what small gps's could do.
They ran one of the first comps that just used gps to get a result, we used the Garmin Fortrex.
Great little unit, sat on your wrist, easy to read and a good price. But let water in on crashes, horrible to charge and down load, poor accuracy and a dumb speed filtering system. On the wrist it produced great speed spikes, I have a spike that went 1000s of ks away after bouncing of the water surface. You couldn't have the ranking system that we have now using this device. Too many people and not enough accuracy.

GT11/31 The Roll Royce and we have been spoilt. Thanks to Tom Chalko who did the work and the testing for FREE we had a device that would allow us to explore the limits of our sport and prove it. NOT "I'm faster than you cos" Together with Mal's software it gave us accuracy that we could prove over and over again

Despite the limitations of the GT32 its still the best we have ATM and we can prove it.

As for fun ranking that's even worse.......it belittles the work that many have done behind the scenes to improve our sport and results of the sailors that participate.
For all those who do what some change and be able to use "Any" device......DO THE WORK and prove your claims that XYZ is better not just cos its a watch.
Don't be a lazy #$#$% do the work.

A new website ,with a new comp that can use any device!!!! I'll join that but I'm going to use my custom Garmin/Trimble watch with its 3kt filter to give me extra speed. But its only for fun against mates
Want to sail for fun with any device.........DONT POST.........ever


No one who is improving and testing our devices is getting payed , they just want the best for our sport.

A big thanks to
Ian Fox, Mal Wright,Tom Chalko, Hardie, Nebs, Andrew Daff, GPS3 ,GPSTC and Ka72
they have all contributed to our sport in a huge way for the benefit of US and I'm very grateful

Select to expand quote

boardsurfr said..

decrepit said..

I think there's a need to think if the SDoP data is really necessary, for non record breaking results.

As a matter of fact, they are not. The only software that definitely uses SDoP, AFAIK, is GPSResults. GPS Action Replay does not seem to even read SDoP data - it does not show them in the track point table, nor does it calculated accuracy (+- numbers) as GPSResults does. ka72.com does not show accuracy, either, so I'd assume it does not use SDoP. Even in GPSResults, it only takes one mouse click to ignore the SDoP values completely.

It's probably safe to assume that the vast majority of speed results posted on GPSTC was calculated without SDoP data.

SDoP for records? Absolutely, a must. For testing devices? Nice to have. For routine GPSTC postings? See above.

ka72 does use SDoP data when it is available.

Are you going to allow the GW 60 watch?

It takes a lot of time and effort for the tech heads to test the devises and they do this with no reward but cop plenty of crap, very unfair!

Any way, this may be stupid but I'll put it out there.

Why couldn't we have unapproved devises included in the GPSTC in the following way.

Where we have the check box for "T" for Track-point or "D" for Doppler we have the "U" for Unapproved.
Any scores with a "U" would not contribute to the teams real score, the one that shows up in bold.
This way everyone is included but only the approved devises actually score for the challenge.
Tracks still have to available for checking if there is a record or suspected glitch or posting T or D instead of U.

If & when a device is approved then the scores from these devises would be included.

This would keep the serious guys happy & all the others allowed in the game to a limited extent.

This is from yesterday's session ..gt31 and fenix 3 . .

The acuracey of these watches looks really good Im not sure if every one realises but all the modern gps chips actually average and filter the data in side thats why they give the choice when setting them up of different modes like airplane, car, boat so it can make the best guess at what the data should be.
This is partly what makes them all look so acurate. If i set one to boat mode it will give a smoother output but also a slightly more lag before sending the data I asume it also eliminates spikes due to altitude change which it knows wont be happening in a boat. If set to airplane mode the data is much more jittery and quicker to respond.From how good the data looks coming from these garmin watches they probably have a system like this like the ublox and skytraq chips.

www.facebook.com/GPSSpeedsurfing/photos/a.262379773851804.60653.204388169650965/1082223505200756/?type=3&theater

After intensive testing we are very pleased to announce the official approval of the Gyro1 for all categories onGPS-Speedsurfing.com. The Gyro1has been develeoped by Raymond Wortel and we want to thank him for all his efforts and dedication in the last year. We also want to thank Manfred Fuchs for all his efforts, with his advice the Gyro1 transformed in one of the most accurate GPS Devices for our speedsurfing-sports ever !!! The Gyro1 is based on the ublox 8 chipset withthe required binary sentences (instead of NMEA) including speed-error measures as requested, . The chipset is the best affordable at the moment and 18Hz without missing points is more than enough (10Hz might be sufficient). Due to the speed-errors (sort of SDoP-values, similar to what GT31 (1Hz) and GW52 (5Hz) record) the integrity of the Doppler-data can be verified and the units can even be used for record attempts (two units would be required though).
<div class="text_exposed_show">For more information and/or pre-orders please check out gearloose.nl GP3S Crew


Not that I would want one.... too uggly !!!

Can't disagree ;)
But its a blue box to protect against a broken display...
It's a tech proto type

For GPSTC: What might be helpful is some GPS categories. =<30kts, 30<>40kts, 40<>50kts, 50kts+. I reckon 90+% of participants will fall into the first two categories. I don't think it would be too hard to set equipment restrictions based on category. I'm guessing 2 sec peak is the biggie but each cat might have limits. If ka72 was the equaliser for GPSTC... it might be doable. We should remember Hardie's objective, keep it fun and social. There are other sites for elite records.

Select to expand quote

sailquik said..
Yes , we should be using the concept of 'claimed speed' by deducting the error from the raw result. GPS Results can that now. Even RealSpeed can do it! Everyone else should catch up!

With all due respect, that's just bloody nonsense for something like the GPSTC. Subtracting the error margins calculated from SDoP values is a way to make sure that the speeds are wrong. This kind of approach only makes sense when you want to compare the speed you have reached with a measurement made by a completely different technology, and you want to claim that this speed has been reached with a probability of at least 99%. That may make sense for records. But when comparing speeds in the team challenge, we want to compare the most likely speed.

Side-by-side comparisons of identical units are all nice and good, but they actually can not prove accuracy of GPS units. They can only prove the absence of a subtype of random measurement errors. This gets a bit technical, so I won't go into it.

Instead, let me address the claim that the accuracy of approved GPS devices was "scientifically evaluated". Sorry, mate, but that is a false claim. "Evaluated by scientists", maybe; "evaluated with scientific methods", maybe. But there are some hard, indisputable requirements for science, and they include publication of methods and results in a manner that allows reproduction by other interested scientists. Tom Chalko has at least made some attempts back in the GT-11/31 days, but there is pretty much nothing available for the GW-52. If you cannot take the time to publish your methods and results, don't call it science. A close-knit group of guys making decisions that affect a lot of others? That's called politics, not science.

Select to expand quote

boardsurfr said..
With all due respect, that's just bloody nonsense for something like the GPSTC. Subtracting the error margins calculated from SDoP values is a way to make sure that the speeds are wrong. This kind of approach only makes sense when you want to compare the speed you have reached with a measurement made by a completely different technology, and you want to claim that this speed has been reached with a probability of at least 99%. That may make sense for records. But when comparing speeds in the team challenge, we want to compare the most likely speed.

To beat a record, obviously you need to beat the posted time -> but you also have to beat the measurement error of the device that was used to measure the *previous* record. Thus it becomes _necessary_ to keep/store the record of the accuracy of the measurement equipment.

"Claimed speed" takes that same premise and turns it on its head, building the measurement-error into the record-value itself. Claimed speed is far easier to understand as the record is only a single value [ vs. a value and many-page document describing the error of the previous equipment ].

So ... given that people want to use whatever GPS they like -> how can there be a sensible competition amongst friends, unless there is at least some chance that the speeds being read out, are vaguely comparable ? Can I go make my own GPS which makes sure to always choose an "optimistic" speed ?

Just because we have historically used value printed on the screen as a meaningful value forever and a time... and simply ignored the error, doesn't mean we shouldn't move to a better technique. Claimed speed isn't wrong ... it is a smarter way of keeping records.

Select to expand quote

boardsurfr said..
Side-by-side comparisons of identical units are all nice and good, but they actually can not prove accuracy of GPS units. They can only prove the absence of a subtype of random measurement errors. This gets a bit technical, so I won't go into it.

You are correct - side by side comparisons cannot. Which is why we use the value calculated by the GNSS engine.

Please dont use an "appeal to authority" - your own subsequent paragraph asks for others to not do it.... Please do go into it.

Select to expand quote

mathew said..

boardsurfr said..
Side-by-side comparisons of identical units are all nice and good, but they actually can not prove accuracy of GPS units. They can only prove the absence of a subtype of random measurement errors. This gets a bit technical, so I won't go into it.

You are correct - side by side comparisons cannot. Which is why we use the value calculated by the GNSS engine.

Please dont use an "appeal to authority" - your own subsequent paragraph asks for others to not do it.... Please do go into it.

I did not go into details because (a) it probably bores most readers, and (b) those who have participated in the evaluation (like Manfred Fuchs) most likely are quite aware what side-by-side comparisons can and cannot do. But since you asked, I'll expand a bit. After all, nobody is forced to read the rest of this post.

We want to measure the true speed. At least when evaluating a unit, we would also like to have an estimate how accurate our measurement is, so that we get a speed range. If we know the true speed, we can then check if the speed range we get includes the true speed. Depending on what exactly the accuracy estimate means, we expect the true speed to be within the range a certain percentage of time - somewhere between 68% of the time (if the error is 1 standard deviation) and 99.7% of the time (3 std dev). The most common use is 2 standard deviations, where we would expect the true speed to be within the indicated range 19 out of 20 times (95%). That's what GPSResults uses when applying Gaussian error propagation.

The problem is that we generally do not know the true speed. In some rare circumstances, we have a very good approximation of it from highly accurate speed measurements (e. g. Lüderitz). Even if we do not know the true speed, however, we can examine the ranges given from two side-by-side units. Again, we have the expectation that these ranges should overlap most (but not all!) of the time. This can be done for single data points or for any range like 10 seconds or 500 meters. If the frequency of higher-than-expected non-overlaps in the speed ranges given by two identical units is higher than expected, then we know we have a problem: at least one of the two units must be wrong.

This is not necessarily a "fatal" problem. Apparently, this can happen with GT-31 data if you use Gaussian error propagation, since the errors are not sufficiently independent. This can be addressed by using a more conservative error propagation; GPSResults uses averaging instead of Gaussian propagation for 1 Hz data for this very reason (I'm reporting what Manfred Fuchs told me in emails; I did not even have a GPS when the GT-11/GT-31 work was done originally).

Now let's look at the other scenario, where the two units usually give overlapping speed range estimates. That's a very good start - but does it mean that the true speed must be within the given range (or, more exactly, must be within the given range 95% of the time)? The answer is no. Any factors that distort the measurements and affect both units the same way will introduce the same error on both units. There are many potential sources of such errors, for example filters in the GPS firmware; ionospheric distortions; problems with GPS satellites; and electromagnetic interference. Even if we are lucky and the accuracy estimates reflect all these problems, we can still have incorrect measurements if the errors are not sufficiently random (like they were in the GT-31).

Select to expand quote

boardsurfr said..With all due respect, that's just bloody nonsense for something like the GPSTC. Subtracting the error margins calculated from SDoP values is a way to make sure that the speeds are wrong. This kind of approach only makes sense when you want to compare the speed you have reached with a measurement made by a completely different technology, and you want to claim that this speed has been reached with a probability of at least 99%. That may make sense for records. But when comparing speeds in the team challenge, we want to compare the most likely speed.

With all due respect, that is just your 'bloody nonsense' opinion. No one has ever claimed that GPS speeds are 'right'. The very essence of the discussion is that there is error, therefore all speeds are wrong! All devices and systems have error. Manfred calculated the error of the Luderitz gate system (which he helped build) and it was not that much better than what he calculated for the GPS we tested! (ask him). The WSSRC just use a different way to compensate for that. It makes perfect sense in any competition to compensate for known error. 'Most likely' speed is meaningless. Claimed speed using calculated error sets a standard of confidence and makes perfectly good sense. Since our fun competition is not based on start and finish lines like most racing, but on individual timing, evaluating error, and compensation for error becomes essential. Using calculated Claimed Speeds could actually allow the use of a wide range of GPS devices as long as they produce the Doppler speed error evaluation and we use that for correction. In this way one could use various devices that commonly produced higher or lower error and it would automatically be compensated for. Of course, sailors would also be automatically motivated to seek out the devices with the lowest error values. Or they could use less than optimised, maybe cheaper devices, and just accept the higher error values and the higher compensation that goes with that 'just for fun'.

Select to expand quote

boardsurfr said..Side-by-side comparisons of identical units are all nice and good, but they actually can not prove accuracy of GPS units. They can only prove the absence of a subtype of random measurement errors. This gets a bit technical, so I won't go into it.

Of course to both statements. Side by side testing evaluates the consistency of devices and gives us an evaluation of their likely error range. That includes the way they react to things that can cause random measurement errors. Actual accuracy (assuming perfect GNSS receiver device implementation) relies on the physical implementation of the GNSS system and the underlying mathematics, both of which are very well known and studied.

Select to expand quote

boardsurfr said..Instead, let me address the claim that the accuracy of approved GPS devices was "scientifically evaluated". Sorry, mate, but that is a false claim. "Evaluated by scientists", maybe; "evaluated with scientific methods", maybe. But there are some hard, indisputable requirements for science, and they include publication of methods and results in a manner that allows reproduction by other interested scientists. Tom Chalko has at least made some attempts back in the GT-11/31 days, but there is pretty much nothing available for the GW-52. If you cannot take the time to publish your methods and results, don't call it science. A close-knit group of guys making decisions that affect a lot of others? That's called politics, not science.

That is just BS. It is wrong in so many ways and is disingenuous and extremely disrespectful. Like politics!