Average base point and adjustment

[James Suttles]

If we are picking ideas. Would it be possible to have the LS take the base points that are collected, each day, average each day, and then adjust all the points to the newly averaged base point.

Work flow.....
Day 1 collect a 7 hour base point, dpos the point, and adjust the collected points from that day to the dpos position. (We do that now)
Day 2 collect a 7 hour base point, dpos the base point, average with Day 1 Base point, and adjust all the previous coords and the current corrds to the new average location.
Day 3 collect a 7 hour base point, dpos the base point, average with Day 1 and Day 2, adjust previous days points along with the current coords, to the 3 day average.
Day 4, repeat as before....

Just a thought. Currently a project that we have been on for about 7 days. The only way currently is to average the 7 days of base points, then translate all the points collected to the 7 day average point. Just thought the process could be a little more automated with the LS.

Curious to know the pitfalls with this type of adjusting.

Thanks

 

[Nate The Surveyor]

One of the pitfalls would be that if you upload each days work, as you go, then the coords in the LS are continually changing. So, if you go to stake corners on day 3, then day 4 shifts them again.
N

 

[James Suttles]

Agreed, but would they not be relative, as everything is adjusting by a few hundredths one way or another.

Just a trying to think of how to utilize all the data we are collecting. We are gathering huge amounts of data, that should average out to a very good coordinate.

It's just a thought, like you said there are probably pitfalls that I have not considered.

 

[James Suttles]

Here is some real life data, after using the same base point for about 10 days, each day having 6 to 8 hour static sessions. then sent each of those 8 days to DPOS, used cluster average, to get a mean coordinate. After the 8 day averaged base point was collected check to see how the averaged point compares to the next 2 days of DPOS. What I am seeing is a difference of roughly about 0.01' horizontally and about 0.02' vertically from day to day. So with this in mind, would there be a way, without having to pick each days base point, compute the difference from the averaged point, and then shift/adjust the points collected from each days base point to the new coordinate of the cluster averaged base point.

I would think as long as each day is handled and adjusted correctly, everything you are doing should be relative to itself.

What do you see as the pitfalls in this solution.

 

[Nate The Surveyor]

I see 2 things.
1.) You'd have to pay attention, to do this.
2.) There is an incomplete part of the software. Once you use m-local, you cannot change between rtk, and ppk. (For side-shots) I've complained about this, assuming that I simply wasn't doing it right. But, it's something that needs fixed.
If you go back to autonomous, you can make these changes, then re-do m-local. But that's cumbersome. I had a bunch of trajectories, and those take forever, (I thought my LS was locked up).
Nate

 

[Phillip Lancaster]

James. You worried about 0.01'?

 

[Nate The Surveyor]

Its a simple exercise, to wring all the accuracy possible, from equipment, that so far, has proven itself, to do impossible things. He wants to get it as close as humanly possible. If it ever does matter, then it would be handy to get it solved now, (proper steps) not later. More of a personal training exercise.
I do the same thing, sometimes.
For example, when corner setting, and trying to achieve total station accuracy, set a nail, within a few tenths, of the corner. Shoot nail 3x. Avg it. Set corner, with compass and tape, off of avg nail.
Or, clean off the spot, stake it 3 times, and make a dimple in the dirt, at each point. Then, do your own average, physicly in the dirt.
Some of us had pedantic fathers, that drilled us on the idea that "adjustment software is for people without proper disciplines". I say it with a grin today, but not as a kid. I developed habits of that sort.
Nate

 

[Nate The Surveyor]

One more thing. Occasionally, we do short setups. Yields poor DPOS results. I've had .25 h and .56 vertically, differences, in DPOS solution. It's in cases like this, that we NEED fluency, in manipulating our base solution.
There are even times when we work off coords, that were built before we got there. They are SPC, but our solution is a foot or so different. We just need total fluency, and control, with our processes.
Carry on...
N

 

[James Suttles]

I just was thinking if we are collecting all this data, why can we not use it to its fullest. I am not worried about 0.01' but like anything you it can only get better as you average all those 6,7 8 hours static sessions that you are collecting each day you are working on a large project.

Nate, you 100% correct, just trying to make it as accurate as possible. So the next man will not have to scratch his head as much as I have had to, in order to replicate the same results.

 

[Nate The Surveyor]

One more thing.
Reverse shift. Unless you fix m-local, so it does not later restrict your ability to choose rtk/ppk, then it restricts the use of m-local/reverse shift.
Rendering it less useful/functional.
Nate

 

[James Suttles]

Let me say, the LS is hands down an incredible machine.