Here's the issue, we ran a 4300' traverse south between two sets of GPS pairs. One pair to the north about 500 feet apart, and a pair on the south about 450 feet apart. All GPS points were set up over multiple times and static sessions lasted over hour each time. There is a considerable amount of cover and the GPS data had to be worked to get the acceptable 95% confidence level. That being said, coordinates were established for those points. The reduction of the GPS data was not done in our office but in an affiliated office. I have 100% faith that it was done correctly yielding the best possible results. All grid factors are correct according to my calculations.
Once the GPS points had been established, traditional surveying methods were used to traverse between these points. Angles doubled, with hard setups for the foresights and backsights, distances shot between both at all the setups. The first BS and IP were the northerly GPS points. We had 12 setups to close back into the southern pair. The stations were fairly balanced out for distance.
The problem: After coming off the northern pair and traversing into the southern pair there was an angular error of 46”. We were told to hold the coordinates of the first setup and to throw the angular error into the first angle turned (not to hold the BS as 00-00-00). After this was done the linear error was about 0.08’.
The explanation I received was to hold the coordinates of the GPS stations because they are far better than traditional methods of surveying could render. I was just wondering what others thought about this? I have my reservations about doing this!
Thanks, Pat
> Here's the issue, we ran a 4300' traverse south between two sets of GPS pairs. One pair to the north about 500 feet apart, and a pair on the south about 450 feet apart. All GPS points were set up over multiple times and static sessions lasted over hour each time. There is a considerable amount of cover and the GPS data had to be worked to get the acceptable 95% confidence level. That being said, coordinates were established for those points. The reduction of the GPS data was not done in our office but in an affiliated office. I have 100% faith that it was done correctly yielding the best possible results. All grid factors are correct according to my calculations.
>
> Once the GPS points had been established, traditional surveying methods were used to traverse between these points. Angles doubled, with hard setups for the foresights and backsights, distances shot between both at all the setups. The first BS and IP were the northerly GPS points. We had 12 setups to close back into the southern pair. The stations were fairly balanced out for distance.
>
> The problem: After coming off the northern pair and traversing into the southern pair there was an angular error of 46”. We were told to hold the coordinates of the first setup and to throw the angular error into the first angle turned (not to hold the BS as 00-00-00). After this was done the linear error was about 0.08’.
>
> The explanation I received was to hold the coordinates of the GPS stations because they are far better than traditional methods of surveying could render. I was just wondering what others thought about this? I have my reservations about doing this!
>
> Thanks, Pat
4300' by 46 arc seconds is almost a foot. to just put the angular error into the first angle is arbitrary and not at all mathematically sound. what about compass rule, or better yet LSA? either one would produce results much more reliably. that explanation souds like someone in the office needs to reevaluate their method(s).
the gps coordinates were to establish location AND orientation. they are telling you to traverse between three stations, not four. that sounds wrong to me. the errors should be distributed elsewhere (throughout the network)
If I understand your post correctly, you have roughly 4" of angular error per traverse point if it was to be distributed equally throughout the traverse to adjust the angular closure, correct? Seems to me, that unless it could be shown that all of that error was amassed in the first set-up off of your northerly GPS pair, I don't understand the rationale for applying all of the correction (adjustment) to a single set-up as you have been advised to do. It has nothing to do with the GPS being better (more accurate)than your total station, IMO.
> The explanation I received was to hold the coordinates of the GPS stations because they are far better than traditional methods of surveying could render. I was just wondering what others thought about this? I have my reservations about doing this!
this person is having your traverse run through only three station, NOT FOUR! sounds wrong to me. distribute the errors throughout the network. drop it in LSA and see the differences
I do a LSA on most of my traverses. It was just a little over a foot with the actual numbers. I'm not a fan of throwing it all into the first angle either. Personally I would adjust it out correctly using LS but it's not my call. And yes it should have been run through 3 separate points but these points(pairs) had already been establist by the time I got on the job!
Less see, You occupied one point, back-sighted one about 500 feet away, turned to and traversed through an intermediate point to a third GPS point then occupied it and turned the closing angle to the remaining point about 450 feet away. 43 seconds in 450 feet is about, did not compute it, a little over an inch. Not bad when the GPS data was rotated from one point. Think I would distribute the angular difference between the last two angles, inversely weighed by distance.
jud
The 1st thing that came to mind was How steep is this country? Oblique angles to prisms, and targets, sometimes produce funny results, with total stations.
Also, steep country can contribute multipath.
So, maybe a combination of that?
Nate
One of the central tenants of adjusting data is you never adjust data that contains blunder.
Whoever is telling you to throw all the angular error into that first setup must be contending that there is significant blunder at that location. I would want to know the basis for that assumption. If there is no real evidence for that, then Least Squares is the only way to go.
Error of 4 seconds per set up isn't unreasonable depending on the type of equipment and methodology used.
You might also use some of the blunder detection routines from your LS software to see if anything interesting comes up.
Finally, I noticed you mentioned some heavy cover and considerable massaging of the GPS data to get it to meet specs. That would lead me to not lock down the control points too much. There might be more slop in their location than you know.
Larry P
Given the linear nature of the work least squares should not be used
ric
When traversing between GPS pairs, if the traverse is (more or less) straight, you can adjust the traverse graphically using CAD. I will sometimes rotate the entire link traverse to meet the closing GPS point on the other end. Distances are usually remarkably close using this method, especially if using the correct combined scale factor. You can include your initial backsite in the rotation. Not sure if this is the best method, but it works for me.
> Given the linear nature of the work least squares should not be used
>
> ric
Ok Ric. I'll bite. Why?
Larry P
How well did your traverse distance compare to the GPS distance between each control pair?
eh? Why not...? It may be true that there is no "network", but it would still be the most mathematically sound way to distribute random error...
> How well did your traverse distance compare to the GPS distance between each control pair?
Good question Dave.
Hope things are well in your world. Long time no talk.
Larry P
Since you had a hard time with the GPS to start with, you might want to let each GPS point float a tenth (maybe more), then run it through StarNet. You may be surprised at the tightness of the residuals on the final fit if your conventional traverse was tight.
KS
I tried to edit but Larry jumped in! I also should have offered what I would try:
Say your two north GPS points are #1 & #2 and your two south GPS points are #3 & #4. How well do your conventional traverse inverses between 1-3 & 1-4 and 2-3 and 2-4 compare to the GPS calcs between the same points? I'd find the best fitting distance, hold those two points and then see how you much you miss the other two control points.
Ya, Larry, we've not talked much lately! If there are anyones posts that I try not to miss - I don't want to miss yours! Always well thought out, courteous replies!
Hope all is well with you and still hope to make one of your cruises!
Bullcrap rational and a poor way to adjust. Your rational of the GPS pairs were better is in conflict with the fact that you put all the error in it. Also, it's about 4"/hub, and after angle adjustment, it's 1:54,000. What freaking traverses are you running that this is not acceptable? After about 1:30,000, it's just a number, under that, it's indicative of the quality of work.
Compass rule this and move on down the road.
If at all possible, and since you traversed conventionally between the GPS pairs, I would perform a solar observation on each end just to check the GPS azimuths anyway.
:good:
> The explanation I received was to hold the coordinates of the GPS stations because they are far better than traditional methods of surveying could render.
Well, that is a load of hooey. But this is a situation that calls for Least Squares analysis. Anything less is just a guessing game. A least squares package capable of handling this problem is included in Carlson Survey. It's really not cutting edge anymore.
I'm assuming you don't have access to LS, so let me say that 46" in 10 or so traverse points isn't really all that much. Probably half of those 46" are in the GPS points and maybe more. But just hold the GPS, distribute the angular error, and compass rule adjust. You are talking about adjusting 0.08' out of 4300'. For most any work, it's going to be okay.
But better would be to give all the GPS points a standard error of a couple of hundreths in an LS adjustment and let them move a little.
The "best possible" GPS positions in this case isn't going to be as good as GPS positions can be. And even when they are as good as they can be they aren't perfect. If the conditions in the field are less than ideal, the positions are going to be less than ideal. There is no avoiding that. As Scotty said, we canna break the laws of physics, Cap'n.
