While I was flagging a 40 line yesterday; I intended to leave my Topcon GR-3 collecting data on a corner in the woods for which I already have good coordinates. My plan was to collect several hours of static data and see how it compares. The canopy was quite heavy, but, to my surprise; I got a fix before I had completed my set-up. So, I took a shot. Then I collected raw data for 95 minutes (intending to check with and OPUS-RS solution). I am running Survey Pro and when I returned the rover was fixed and averaging 384 epochs; so I saved that shot.
Now, back at the office; both of my fixed shots are within 0.12' of record coordinates for this point. The two fixed shots vary by 0.075' from one another. My OPUS-RS solution, however, is off by 0.3'.
Do you think a8+ hour OPUS-S solution would have been better. Or is this simply because Survey Pro uses Glonass in the solution and OPUS does not?
I'm trying to refine my "lie detector" for RTK under canopy.
As an academic exercise:
If you had a mechanim, to set it in the woods, and it could gain a number of fixed epochs, then store it, then automatically reset, gain fix, number of epochs, save, reset, and you could do this all day long, then youd find that for each location, you set it on, that you will get a certian number of fixed shots, within a few tenths or so of each other, and a certian number of outliers.
From this, you can develop a ratio, of outliers, to good shots. It is able to attain.
So that each place gets a file, that looks like this:
7 hrs time
90% good, 10% outliers
Total number of fixed shots 20
Size and shape of the good data (error elipse) .31' ew and 0.39' ns.
Then, you will find, that by looking at the data, that it has: no shots 8:00 to 8:30 am.
Then, a few outliers 8:30 to 8:45 then 8:45 to 10:45 is a large number of good shots.
Then a few outliers, then nothing, until 12:30, when a few outliers, show up, the 3 hrs of pretty good data, and so forth.
Now, we all know you cannot run a business like this.
But, we can learn stuff from this.
There are bad times of the day, and good times. And, most of the outliers are collected during transition, from good to bad, or bad to good.
And, that it requires time, to sort, good and bad.
And, that the error elipses generated, may be larger than your error budget will allow.
So, you need more data, to refine the location.
Thats an example of how to be sure.
This is why mfr's say "gps dont work in the woods".
I was retracing another surveyor today. He's a "one fix assumer" Gets one fix, and assumes its right. Found a 6.5' bust.
I'm finding those, sort of busts occasionally these days. We all are going to have to deal with the mess left by these kinds of practices.
As an academic exercise:
If you had a mechanim, to set it in the woods, and it could gain a number of fixed epochs, then store it, then automatically reset, gain fix, number of epochs, save, reset, and you could do this all day long, then youd find that for each location, you set it on, that you will get a certian number of fixed shots, within a few tenths or so of each other, and a certian number of outliers.
From this, you can develop a ratio, of outliers, to good shots. It is able to attain.
So that each place gets a file, that looks like this:
7 hrs time
90% good, 10% outliers
Total number of fixed shots 20
Size and shape of the good data (error elipse) .31' ew and 0.39' ns.
Then, you will find, that by looking at the data, that it has: no shots 8:00 to 8:30 am.
Then, a few outliers 8:30 to 8:45 then 8:45 to 10:45 is a large number of good shots.
Then a few outliers, then nothing, until 12:30, when a few outliers, show up, the 3 hrs of pretty good data, and so forth.
Now, we all know you cannot run a business like this.
But, we can learn stuff from this.
There are bad times of the day, and good times. And, most of the outliers are collected during transition, from good to bad, or bad to good.
And, that it requires time, to sort, good and bad.
And, that the error elipses generated, may be larger than your error budget will allow.
So, you need more data, to refine the location.
Thats an example of how to be sure.
This is why mfr's say "gps dont work in the woods".
I was retracing another surveyor today. He's a "one fix assumer" Gets one fix, and assumes its right. Found a 6.5' bust.
I'm finding those, sort of busts occasionally these days. We all are going to have to deal with the mess left by these kinds of practices.
Thanks for chiming in Nate.
My good fixes could have been just coincidence. If I understand correctly; my 384 epochs means that I was fixed for roughly 6.5 minutes. Since I wasn't at the rover for the entire observation, each fix could have been 30 seconds or less as the constellation changed. There may be no substitute for long, static observations as an rtk check ( short of buying a Javad system, of course).
Users of GNSS frequently forget that they are not collecting coordinates that fall from the sky. They are measuring vectors between one or more fixed bases and a rover. Those vectors have error components that are, in part, proportional to the length of the line being measured.
Your OPUS position is computed using vectors from at least 3 different bases, using data collected during changing satellite geometry, and thus changing multipath conditions, and applying weighted averaging to compute a coordinate. So it has the advantage of greater redundancy. A long duration observation from your base also has that changing satellite geometry component while being a shorter vector, and thus perhaps having less error (assuming that your base is in as clear a setup as CORS antenna are) in that single vector. But you don't have the redundancy. And it is relative to your base position. How did you position your base?
The other thing to remember is that when you take 2 shots under the same (marginal) conditions the sources of error are going to have the same biases. You repeat your errors and get the same answer. That doesn't make it a good answer. Repeatability is generally good, but it's not proof of perfection. Repeating vectors measured a couple hours apart is a good idea because the satellite geometry changes, and this you are less likely to be repeating your errors.
Mark Mayer, post: 446020, member: 424 wrote: they are not collecting coordinates that fall from the sky
He he he!
Misconception number 3082.... 🙂
Actually, Mark, my OPUS-RS used 7 CORS stations. That's why I generally prefer them. You're right, though. I've had the best results using two 1.5 hour observations separated by 6-8 hours and processed through OPUS-RS. My base is ideally located in this case. It's on top of a bare hilltop.
My original question was whether or not Survey Pro's use of Glonass in it's fixed solution was the reason it was closer to record than my OPUS-RS. I think it was probably coincidence. Unless I continue to see similar results...
KevinFoshee, post: 446051, member: 8314 wrote: My original question was whether or not Survey Pro's use of Glonass in it's fixed solution was the reason it was closer to record than my OPUS-RS.
More satellites is a great advantage in marginal conditions so I'd say yes, the GLONASS added solution is the better.
Here in Oregon the DOT operates a GPS only RTN and offers it to private users for free. Meanwhile, the Trimble dealership offers an RTN, for a fee, which utilizes GLONASS. The selling point of the for fee service is that the positions in marginal conditions are better because of the additional satellites.
Thanks for your input Mark.
KevinFoshee, post: 446051, member: 8314 wrote: Actually, Mark, my OPUS-RS used 7 CORS stations. That's why I generally prefer them
Because of the shorter duration of observation in an OPUS-RS session you will get less change in satellite geometry. Collecting for 95 minutes is probably enough time to cover that, but nevertheless the longer duration is probably a more important factor than the 7 CORS in such a case.
