GPS Precision vs GPSTC ?

So we had today nice result one of our team member quite amazing that brought me to some relaxions.
You may experienced similar already in your own results.
GSPTC is using precision up to 3 numbers after "."
But that question is what is precision of out GPS loggers ? Even the best 10HZ and Doppler corrected?

What is the point or 29.997 or 49.999 Ktn if rounded looks nicely ?


citation
" ......
If we had GPS precision correctly at 2 not three after . then rounding 29.997 possibly comes to 30 ktn TonyK. I doubt that using precision to thousands of seconds not hunders or even tens is correct. Our GPS 10Hz even corrected by Doppler are not that precise. I did such experiment and type your result in excel. Both are same numbers . So technically you did 30 ktn.
That should be follow be question to GPSTC why even they using such extent of precision which is not compatible with our GPS devices? ...."

What is interesting on the picture above, the fact that when the rest of our pack were fining TonyK was foiling.



What would you say if you spent all your life savings , go to Lauderitz and do 49.999 knt? !

I knew that

Select to expand quote

sailquik said..
The precision of any device is only as good as it's error range.

If you want to be pedantic, you should really subtract the reported error value for the number to be 99% sure you have achieve 'at least' the reported value.

Error values in the range of 0.02 to 0.07Kt are quite often seen for 10 second runs, and differences between devices tested side by side in controlled conditions, definitly confirm that, and that their results are better than 95% within the error range. A lot of the time, the difference in reported results is less than 0.01 knots.

But since the devices do output the raw numbers to 3 decimal places, we decided many years ago to simply report what they output.

The actual resolution of the reported number can't be better than 0.003Kt anyhow, in the best devices.

Then you can draw your own conclusion. And of course, you are correct that anyone with results within the error values of each other is effectively on the same ranking.

Yep. I promised myself to get one day to Luna park and get a ride on biggest carousel available with my GPS logger fully loaded. We could assume that rotating speed is very stable and circle perfect.
Then we could see what our GPS logger register. We could take a bunch of different GPSes and see if all the same> Also test what is the difference between device with Doppler and without Doplper.

Select to expand quote

Macroscien said.. >>>Also test what is the difference between device with Doppler and without Doplper.

No point we already know, without doppler you're down to positional accuracy.

I resort to placing them all against the windscreen , or tying them on the roof rack.
The thing with the carousel, is making sure the antennas are all at identical radiuses otherwise they will be doing different speeds

Select to expand quote

Macroscien said..
This time GPS travel vertically and could be very interesting to see what our GPS actually record?Third option is to hang GPS on a long line and use motor to rotate. That should also produce valid experimental data.

Depends if the GPS is set to record 2d or 3d. In some cases 2d is better as it eliminates chop distortion, but the UBX chip changes back to 3d after 40knots.

Select to expand quote

Macroscien said..

Theoretically the circle recorded by GPS on our carousel should be perfectly round and increments same , speed constant. So my experiment will show. I did check and the biggest carousel at Gold Coast is under renovation till 9 Feb.
So in two weeks I should have some results for further analysis.Carousel is running at 47km/h which should be quite adequate to our needs.Another interesting experiment could be made on huge wheel in Brisbane. This time GPS travel vertically and could be very interesting to see what our GPS actually record?Third option is to hang GPS on a long line and use motor to rotate. That should also produce valid experimental data.

That test wont be representative - if your carousel doesn't keep the device un-rotated, then you end up with this:
gge.ext.unb.ca/Resources/gpsworld.september06.pdf

A better test is just to place them on a roof-rack and drive a long distance. Since this thread is about accuracy and precision, it would be important to also independently measure the kinematics of the mounting-place (say using a 12 axis IMU). Edit: as noted, this test was done many years ago. It is important to understand the state-of-the-art before repeating previous work (unless it is to verify that previous work).

Select to expand quote

Macroscien said..
Good question. So let's back to my initial statement. Why even bother to set precision to 3 numbers after .?
Suppose we want to measure our weight eith silimilar accuracy.
So your current weight will be 85 kg 387 grams. Oooops 389 grams because you just took a breath . Nope this 386 becouse you just drop a sweat or sneeze.
In fact I want to test and see what our gps really records.

Whoop de doo about the other stuff - eat a pie, drink beer, fart. We are measuring speed - you are welcome to hold onto that fart for the 60s of the run.

Select to expand quote

There is urban legend that your gps placed on your head record better speed that one on your arm .

There is no urban legend - it will record better because signal-to-noise is well understood science.

The only question that is worth discussing of head vs arm vs down-your-wetsuit, is the practically of each method vs the validity of the collected data.

Select to expand quote

Accuracy and precision of our gps system recently my hobby unrelated to sailing.
I sm designing and building robotic mower for na farm. Using civilian gps accuracy to 5 or even 1 meter is not enough. Because you want mower to run straight lines with 10cm accuracy at least. I know that John Deere solved problem possibly using local reference points.
Anyway I expect to have first test of my gpseses on carousel soon and we could see if we coukd draw sny conclusion from there.
The common risk for scientist making assumption is such that will try to get data to fit his thesis.

That last sentence is just dogma. The Rosetta spacecraft landed on a comet - it has to precisely pinpoint a position in space, months ahead of time -> its the same type of Math to understand GPS accuracy and precision. But somehow you dont trust scientists. Is the world flat ?

Select to expand quote

Macroscien said..
The common risk for scientist making assumption is such that will try to get data to fit his thesis.

No, a scientist will design a double-blind experiment to test the null hypothesis. Publish with materials and methods so that other scientists can replicate the study and verify the findings.

I suppose its similar to the spurious argument that the new BoM temperature probes are recording warmer temperatures than the old mercury thermomsters. Parallel experiments showed that there was an insignificant difference of 0.02 of a degree. However, the tolerance of the new instruments is 0.2 of a degree compared with 0.5 for the older thermometers.

I am still pleasently surprised how our GPS accurate are. Let assume following scenario.
We are going 40 ktn.
so 74.08 km/h
or 20.577 m/s !!
so during 1 second we do already 20 meters !!!

Let assume that our GPS has 5 meters precision to locate us on the globe. so for two measurements the possible margin of error could be 10 meters.
Translate it to speed now.

So our speed could be as high as 59ktn or as low as 20 ktn !!!

OK our GPS is running now 10Hz so distance shrink to 2.057 meters during one tick 1/10 of second but location precision remains the same - 5 meters !!

I don't even want to think what error margin could be.

OK lets claim that our GPS has 1 meter accuracy not 5.

So our speed measured 40 ktn may vary now
from 43.887 knots to 36.112 knots

This is the truth and the rest is up smart software and statistics.

The point I am trying to make that our measurements could be as bad as 3 full knots in worst case scenario but we fight here for every 0.001 knots !

BTW I love to be corrected because it mean i learned something new, so be kind and teach me something.
...
I know.
We trust our devices manufactures but the same may apply to Space and rocket science.
It you multiply error and software cleaning procedure to such cosmic distances it is not surprise that
rockets and landers on celestial bodies are crashing more often then success. Small error like ours here mean that you just landed your craft 1 meter below Moon surface successfully.

Yep.
we are proud to have 1 cm accuracy on our $100 GPS devices
but at same time our military claim precision of their 10 mln dollars GPS guided missiles no better then 5 meters







Oooops. Precision of our GPSTC results is actually
0.5 millimeter
like a hair makes you winner or looser on 50knt quest ( for me it is still 40 ktn quest ) .

Select to expand quote

Macroscien said..
Yep.
we are proud to have 1 cm accuracy on our $100 GPS devices
but at same time our military claim precision of their 10 mln dollars GPS guided missiles no better then 5 meters







Oooops. Precision of our GPSTC results is actually
0.5 millimeter
like a hair makes you winner or looser on 50knt quest ( for me it is still 40 ktn quest ) .

Unfortunately, we do not have 1mm accuracy on our speed devices.

Actually, as said above, we don't really care much about the positional accuracy.

Our best Doppler speed accuracy is in the order of 0.1 knots to 0.01 knots, depending on the devices used (Those with accuracy calculations in the data)

Even the very best RTK devices (as commonly used by Farmers to guide their machinery) are only claiming a couple of cm accuracy, and they cost thousands of dollars, are too large for a windsurfer to carry, and require a (mostly paid) subscription to a local Differential data broadcasting station.

Read the paper! :-)

Select to expand quote

Macroscien said..
ok this is why we are rather more interested in 2s not max.
10s or 500 meters are even more convicting and make errors negligible.

True, the longer the measurement period, the lower the error range.

But obviously, human nature comes into it. We all like to see what out peak speed is, even if we know it is a bit less reliable.

Select to expand quote

Macroscien said..
Apollo mission landed on the Moon perfectly with Armstrong at the helm overriding all computers on board.
50 years later the most recent most advanced Moon landers keep crashing because unable calculate distance correctly.
Have you see recent Japanese Moon landender landing on his head?

This is just wrong - the distance calculation was perfect. The actual reason because the engine fell off. Armstrong couldn't have made a difference to the outcome.

A good summary is here:

If you are going to make a statement about something, you really should at least be vaguely accurate.

Select to expand quote

mathew said..

Macroscien said..
Apollo mission landed on the Moon perfectly with Armstrong at the helm overriding all computers on board.
50 years later the most recent most advanced Moon landers keep crashing because unable calculate distance correctly.
Have you see recent Japanese Moon landender landing on his head?

This is just wrong - the distance calculation was perfect. The actual reason because the engine fell off. Armstrong couldn't have made a difference to the outcome.

A good summary is here:

If you are going to make a statement about something, you really should at least be vaguely accurate.

Oooo. I heard they run off fuel prematurely!
So my apology because that is engine fall off. Nobody 's fault. Just happen.
Door fall off of airplanes or rocket engines, who cares.
The point is mission is not accomplish.
As I said. Thank you again. I have been corrected, so I learn something.
Japanese also will learn something we hope. Tighten bolts on rocket engines or Toyota Ute. Oooop I need to check my Hilux if engine is not falling off too !

Select to expand quote

sailquik said..

Macroscien said..
I am still pleasently surprised how our GPS accurate are. Let assume following scenario.
We are going 40 ktn.
so 74.08 km/h
or 20.577 m/s !!
so during 1 second we do already 20 meters !!!

Let assume that our GPS has 5 meters precision to locate us on the globe. so for two measurements the possible margin of error could be 10 meters.
Translate it to speed now.

So our speed could be as high as 59ktn or as low as 20 ktn !!!

OK our GPS is running now 10Hz so distance shrink to 2.057 meters during one tick 1/10 of second but location precision remains the same - 5 meters !!

I don't even want to think what error margin could be.

OK lets claim that our GPS has 1 meter accuracy not 5.

So our speed measured 40 ktn may vary now
from 43.887 knots to 36.112 knots

This is the truth and the rest is up smart software and statistics.

The point I am trying to make that our measurements could be as bad as 3 full knots in worst case scenario but we fight here for every 0.001 knots !

BTW I love to be corrected because it mean i learned something new, so be kind and teach me something.
...
I know.
We trust our devices manufactures but the same may apply to Space and rocket science.
It you multiply error and software cleaning procedure to such cosmic distances it is not surprise that
rockets and landers on celestial bodies are crashing more often then success. Small error like ours here mean that you just landed your craft 1 meter below Moon surface successfully.

Well done Macro. You just very clearly described why it was such a huge improvement when we switched from using positional data to calculate our speeds, to Doppler derived data. We went from a potential error range of knots, to an error range of very small fractions of a Knot.

Positional accuracy is very important for the GPS guidance of lawnmowers and tractors.

We use the Doppler derived speeds which are many orders of magnitude better for measuring velocity. Speed Over Ground (SOG)

I recommend that you get up to speed on speed measurement and start with reading this paper:

nujournal.net/wp-content/uploads/sites/2/2021/02/SDOP4aa1.pdf

Thank you
Because I finally realized how our GPS works.
( Correct me if I am wrong)
We have indeed two different devices all in one box. One system is designed to establish our geographical position
and the second is simply speedometer.

Now to work out our current speed we could use both.

1. But the first will employ our primary school knowledge.
We learn our positions , work out distance and divide distance by time.
Done.

2.The second require our high school knowledge, and has nothing to do with our position on the map at all.
We just measure shift in frequency signal from satellites and we could work out our speed perfectly.
It seems that second method is more accurate for our needs.

So first will apply to all basic GPS and second to GPS with Doppler.

Now it would not be Marco post if I could not come with some invention or improvement !

I imagine that in the near future we will be using our GPS mostly at night !!!
Why?!
Because at night we could stream from our satellite not radio signal but laser generated light.
Light of passing satellite at night will be clearly visible to our optical sensors.
( Yep- astronomers already complain on night sky pollution by Elons Starlink satellites that are not even emitting any lights itself)
Due to much higher frequency of light versus radio waves accuracy of our measurements will be now thousands or million times better !
Done !

But Macro possibly will not last so long to see his invention implemented for our watersports GPS and I don't like personally to sail at night