I recently tried to use NRCAN for the first time to do PPP and compare it with NTRIP accuracy.

I put my RS2 on the top of my roof and logged for 4 hours at 1HZ, exported as Rinex.

I created several survey points in RTK with NTRIP, all lay within ±2/3 cm.

I received the NRCAN files after a few days and the distance between the points was 30 kilometers.

Any idea what could be at fault?

The NTRIP corrections come in ETRS89 but that should explain a difference of less than a meter compared to ITRS2014…

raw_202007011344.pdf (1.1 MB)

This is the PPP result in pdf (233.7 KB)

This is the solution file from RS2 with NTRIP corrections…

Update: I managed to get a solution from OPUS (RTCONV in Rinex2.1 and GPS only did the trick) and now both points lie 6cm apart from each other … still 30 km away from my survey points…

Ok, so I somehow got the results now, with one question remaining:

I used the UTM32 (EPSG:32632) coordinates in the OPUS and NRCAN solution and transformed it to ETRS89 (EPSG:4258).

NRCAN solution was 12.5 cm from what I would call the truest NTRIP solution and the OPUS was 25 cm further away.

I am looking forward to receiving the more exact solution. Both solution were roughly 15-20 cm off.

What I am puzzled by , still is: where on earth did that 30 km shift in the LLH ITRF2014 data come from?

Here: … among others, it is written that ITRF and WGS84 generally coincide within 1m…

I have no idea where that 30km shift came, but when I plug the solution from NRCAN in google maps, it’s smack on top of your apartment building, mere centimeters apart from the first line of your LLH file.

PPP: 51° 29’ 59.52933" 7° 29’ 10.39871"
LLH: 51° 29’ 59.5" N 7° 29’ 10.4"E (rounded by gmaps)

Thank you Gabriel, the solution file is the one from my RS2 with the NTRIP corrections… (1.6 MB)

In here you find the complete result from NRCAN. But the LLH values are the same as in the pdf…

I misunderstood, and checked the coordinates of NRCAN in Earth, too… and I got the same result as you.

I have to check where my error was. I read the coordinates into QGIS as ITRF2014… maybe there was a mistake…

1 Like

But anyhow, would you expect that sort of difference between NTRIP fixed solution and 4 hour logging and PPP (12 cm NRCAN/ 24cm OPUS) ?

Maybe, if the datums are not the same. Have you tried transforming the PPP result to ETRS89 (I’m assuming that’s what your NTRIP service uses)?

I converted to ETRS89 using a local survey software and I think that if that was the reason, the shift should be larger

I also did the conversion twice to be sure…

Hi @kasanmascheff,

@Gabriel_C, thanks for your help.

I’ve also pasted your results and observed the same as @Gabriel_C did. Such a difference is most likely happen due to the coordinate transformation process. I’d recommend cross-checking the convertation sequence again.

Also, had you compared the coordinate transformation results in QGIS and your own local survey software?

The 30 km distance was indeed a transforamtion error, but the 12 cm seem not to be…

I will try another time and see where I get…

QGIS has no way of transforming ETRS89 to WGS84 in cm level accuracy without doing extra work. The QGIS transformation between the two is a zero operation because it sees both CRS as the same.

Hi @kasanmascheff,

Sorry for the delay.

I saw you mentioned in another thread that you’ve transformed your coordinates with the Transdat software which does not have the ITRF2014 CRS with the corresponding datum. I suppose the results from this thread are having the shift because of the same reason and its worth cross-checking them too.

Hello Artoem,

I don’t think so because in the other project I did not need to do any transformation because both data (UAV and GCP) are in ETRS89 by default. And even if there would have been a chance to mistake WGS84 and ETRS89 --> it makes a difference of around 70 cms at the moment…

But again, no transformation happenend…

Hi @kasanmascheff,

I am sorry for the delay.

I’ve converted your NRCAN results in ITRF2014 (EPSG:9000) to ETRF2000 (EPSG:9067) with GDAL CLI. ETRF2000 is the basis for the ETRS89 realization in your country. As the transformation is time-dependant, it is necessary to check if the correct datum is applied. The difference between the coordinates from your NTRIP caster and NRCAN after the conversion is in the range of 1-2 centimeters. I believe such a distinction is pretty tolerable considering the time-dependent complexity of the transformation.

I can PM you the transformation parameters to avoid publishing your location data.

Hello @artem.fomenko,

thanks for getting back.

Do you know of a step-by-step tutorial to make this transformation in GDAL?

Please send me the transformation parameters.

Thank you!

Hi @kasanmascheff,

Please, check your PM. I’ve sent you the transformation parameters there.