RTPK Long Range, Short Observation Test

Shawn Billings

Shawn Billings
5PLS
Saturday evening I took a few minutes to scratch an itch I've had for some time now. I wanted to test RTK and RTPK precision at long distances. Several years ago I tested RTK at distances of 10+ miles and found some very interesting results using GPS and Glonass which convinced me that our stated precisions were overly pessimistic. With multiple constellations it seemed that we could probably achieve even better results today. With the old GPS and Glonass only v6 engines, the furthest I ever used RTK was about 25 miles. Precision started to open up and time to fix also opened up. The fixes were correct, but precision was so loose as to not really have a use for most of my surveying work.

HOWEVER....
I setup a Triumph-3 at my office, using a Javad issued Verizon SIM card with static IP, transmitting RTCM3 MSM4 Short, GPS+Glonass+Galileo+Beidou corrections to my Triumph-LS Plus with J-Field 3.0.10.401 and GNSS 4.0.0.200918 firmware. I used DPOS derived coordinates for the base (at present I'm not sure what the statistics are on the coordinates I used). I drove 30+ miles to a HARN monument at a county airport DATASHEETS (noaa.gov)

I occupied the monument for 30 seconds. Due to default settings, I did not immediately have an RTK fix, but I logged the data and processed using RTPK. The baseline distance was 170255 feet (or 32.2 miles or 52km). I was shocked that RTPK reported a fixed solution. Moreover I was blown away by the repeatability of the following points collected and the low residuals compared to the published coordinate of the mark.

Published:
N 6898032.82 sFT
E 3259192.75 sFt
U 348.9 sFt
Texas North Central Zone
NAD83, 2011 Adjustment, Epoch 2010, Orthometric Height Geoid18

Residuals (sFt):
Point 1 - 30 seconds
N +0.084 E -0.009 U +0.3

Point 2 - 30 seconds
N +0.076 E -0.042 U +0.2

Point 3 - 180 seconds
N +0.023 E -0.038 U +0.25

Point 4 - 60 seconds
N +0.067 E -0.038 U +0.25

Point 5 - 60 seconds
N +0.062 E -0.034 U +0.1

Point 6 - 30 seconds
N +0.031 E -0.016 U +0.25

I would not consider this a thorough test. While I do trust the coordinates of the base and of the NGS monument, I cannot say that their precision is better than the precision RTPK appears to produce (it could be that there is an error between them that happens to be in favor of the RTPK result). Also, I cannot say that through a 24 hour cycle that the precision would continue to be as good as what I saw here. The sample size is just two small and the test site too unknown to make definitive assertions. Having said all of that, I've been using GPS for more than 20 years now. There was a time when L1 only GPS was ubiquitous. The general useful range was limited to about 20km (12 miles) and precision was noticeably poorer at those ranges than at 10km or less. My dad and a few colleagues across the country experimented with long range L1 solutions with great success, but the required procedures to achieve those results was beyond practicality for most surveying applications. Later as we invested in dual frequency GPS equipment I found that I could acquire fixed positions, post processed, with only five minutes of data within 10 miles of the base. I never really pushed the limits much further, except to say that even with OPUS and then DPOS, I would see how short an occupation I could use to get a reliable fix with the CORS spacing I had (about 30 miles). I found that 15 minutes was the bare minimum.

So you can imagine my surprise when I consistently acquired tight fixes using RTPK at 32 miles with only 30 second observations. Eventually I made some changes to a few settings in the advanced settings of the RTK engines which allowed me to get reliable RTK fixes at that range as well, but I would prefer to wait to discuss this until I better understand these settings and can more consistently achieve good results.

I do not know how important the addition of Galileo and Beidou were to getting a fix at this distance with these short observations, but I suspect it was critical. This definitely makes an office base station more practical for many surveyors. I'm not sure I would consider 32 miles "practical". I suspect fix times in canopy will be impossibly slow. But perhaps 15 miles would be an acceptable radius. Furthermore, with an office base, a user could establish a base position on the project site and then setup a secondary base for the duration of the project.

I've said it a lot lately, but I feel I must say it again, well done to @Alexey Razumovsky and his team. These results are truly spectacular.
 

Bryan Enfinger

Active Member
I'm seeing basically the same results using Justin and Justin 3 office software which I suspect is the same engine used in the LS. This is using surrounding CORS for 6 minute observation on known position in multi path area. See results here posted in another thread

 

Bryan Enfinger

Active Member
I need to look in that T2 observation file to see if I enabled all constellations in the PP. Very seldom I see all 4 systems during observations. Mostly GPS, GLONASS and maybe Galileo. I think I maybe have seen Beidou once with maybe 2 sats. From reading some of Alexey's posts, Galileo and Beidou are only single freq in the T2 and are more of a hinderence than any benefit for the T2 receiver if all 4 are used in PP.
 

James Suttles

Active Member
So would a fixed base at the office, give a range of 20 miles, comfortably in canopy, with LS Plus rovers? Or is that too optimistic?
 

Shawn Billings

Shawn Billings
5PLS
I've been satisfied with RTK performance at 10-12 miles in canopy. I haven't tried 20 miles yet to know what to expect. I can say that at the moment 30 miles seems a bit to far for production work with RTK, although I think with some settings changes, 30 miles might be ok, at least in the open. But these settings are too sophisticated for users, including myself and would need to be made by the developers.
 

Steve Douty

Well-Known Member
Caveat - I am not qualified to respond. Now for my response.
I am comfortable at twelve miles and less, and VERY comfortable at six miles and less. I personally think that at more than twelve miles, and certainly at more than twenty miles, we are getting into a discussion of absolute accuracy opposed relative precision. There have been thousands of pages written on this topic. At "great" distances we start talking about speed of light in the atmosphere, length of signal travel, plate tectonics, quality of the base coordinates, errors in the assumed geoid model, and etc. (This starts to be a search for "Truth". Diogenes teaches us that this may not be a rewarding career choice.) This doesn't even consider the merits of bourbon Vs. gin and which bourbon or which gin.

I believe at more than six miles we start talking about repeatability and relative precision between positions on our local project. At great distances the discussion of accuracy gets too esoteric for mere mortals. The questions start to center on time, money, profit, ethics and the ability to stay out of court. An additional discussion about what our state laws allow, what does our licensing board allow (we won't talk about what they might be able to understand), what are trying to accomplish and what is our E&O premium starts to muddy the water. In the end, it is a question that requires us to exercised our professional judgement.

"42 is the answer to the Ultimate Question of Life, the Universe and Everything."
 

Shawn Billings

Shawn Billings
5PLS
With GPS and Glonass I generally kept within seven miles as I could see precision open up a bit, not horribly but noticeably. Time to fix in canopy was a bit longer too. Not horribly, but noticeably. With 4 constellations this seems to be a little conservative now, but I can only speak anecdotally at present.

I am always surprised when I get support calls from customers who are using single base RTN corrections who wonder about time to fix issues or perceived precision only to find that the base they're using is 15 miles away. These were always GPS and Glonass. Baseline distance definitely matters.
 
Caveat - I am not qualified to respond. Now for my response.
I am comfortable at twelve miles and less, and VERY comfortable at six miles and less. I personally think that at more than twelve miles, and certainly at more than twenty miles, we are getting into a discussion of absolute accuracy opposed relative precision. There have been thousands of pages written on this topic. At "great" distances we start talking about speed of light in the atmosphere, length of signal travel, plate tectonics, quality of the base coordinates, errors in the assumed geoid model, and etc. (This starts to be a search for "Truth". Diogenes teaches us that this may not be a rewarding career choice.) This doesn't even consider the merits of bourbon Vs. gin and which bourbon or which gin.

I believe at more than six miles we start talking about repeatability and relative precision between positions on our local project. At great distances the discussion of accuracy gets too esoteric for mere mortals. The questions start to center on time, money, profit, ethics and the ability to stay out of court. An additional discussion about what our state laws allow, what does our licensing board allow (we won't talk about what they might be able to understand), what are trying to accomplish and what is our E&O premium starts to muddy the water. In the end, it is a question that requires us to exercised our professional judgement.

"42 is the answer to the Ultimate Question of Life, the Universe and Everything."
This post is an example of why I use JAVAD. I'll bet the Trimble forum doesn't have a post like this. I'll bet they don't mention Greek philosophers much.
 

Sdrake14

Active Member
I have been testing this same thing for some years now Shawn and have pretty much same results. I was using my Triumph 2 Tcp base with Triumph LS and jetpack getting results at 20-25 miles that I felt acceptable, although I have seen lag fixes. We are splitting hairs here and pretty dang spoiled given limits of closures in the past with ground survey and usually retracing records not as precise. I currently work in remote coastal mountains and can find corners out on a 1960s survey in the range of 5 to 15 off the record position. I do not think I need to sweat a potential residual of even 0.15' relative to that kind of retracement.
 

Sdrake14

Active Member
So I was able to finally get a Triumph-LS plus with which I repurposed my old LS along with a GrAnt to be the TCP base when appropriate. With this I was also running some range tests just days ago and I had the same results (residuals 0.02'-0.07') on some Caltrans control monuments with published coordinates at up to 31.5 miles away. And 3 days ago at about 12.5 miles out with this new set-up I took a shot in about the worst canopy I have seen and was blown away by 3 fixed RTK shots. It was on a steep hillside, under dense Doug Fir branches double and tripled up at roughly 100' tall, I would have expected to not get it RTK with my old setup. Granted I had to let it sit for 15-20 minutes but it took it. Now I have been working inthese same conditions daily for several years now with my old Triumph-LS and T2 base and know about what is possible and I am pretty certain no other system would have taken that position. (wish I had taken a photo, was the last shot of day and I was a little dumbfounded walking away)
 
Top