My company does a lot of road work. As a result we run a lot of long linear traverses(1 to 2 miles) between GPS control pairs. We have noticed that when we run a traverse in the East to West direction our closures are not what we would like them to be. Maybe just a coincidence but the mis-closure always seems to be to the north when running to the east and to the south when running to the west. We have had our instruments calibrated as well as the tri-bracks. All GPS pairs were part of a static network with 1 hour observations with multiple redundancies. Our raw field closures are around 1:16,000 unadjusted and 1:50,000 adjusted by compass rule. We are expecting better closures based on the fact that we are running 1" Total Stations and turning 5 sets of angles.
Strangely enough the few traverses we have run with control in the north to south direction have closed very well(1:50,000+ Raw).
Has anybody else seen anything like this?
There seems to be more going on than just an anomaly with the coordinates from static GPS. A 1 in 16 closure with tribracts indicates either a field procedure that's not very good or the static derived coordinate are in a reference that not relative to the field ground work. Tribracts and a good field procedure should always result in 1 in 50 or more closure. Next time close the traverse on itself and that will show you the problem is in your field procedure or in your coordinates.
Sounds like a geometric/gravimetric correction issue.
Are your coordinates transverse or lambert?
Is there a significant deflection of the vertical in your locale?
You must understand that once you involve GPS into your survey, you have moved outside the parameters of plane surveying and will need to adjust your bearings as you traverse in east or west directions. The lines are actually arcs.
You could apply an angle adjustment to the setups between your two beginning points and your two ending points that will set you on your way.
Probably the best adjustment would be made using StarNet.
I've encountered a couple of things over the years that can cause this:
1. Particularly since you mention East/West traverses, sharp shadows caused by a low sun (such as during late fall, winter and early spring) have given me considerable problems in the past. I keep wanting to call this phasing, but I don't believe that's the correct term. Using targets in which the visibility is not affected by shadows is good, or simple care to make sure the target is properly collimated is key. The eye is naturally drawn to center the bright side of the target (which will be the South side of the target if in the Northern Hemisphere). This error doesn't compensate as the backsight and foresight centering are both drawn to the South, so as you run West, your angles to the right will always be too small and your angles turned right while running to the East will be too big. If I'm thinking of this correctly then, to the West your traverse coordinate at the end of the line will be too far to the South and running to the East, your traverse coordinate at the end of the line will be too far to the North.
2. Since you mention that your North/South traverses are good, this probably isn't the reason, but I'll throw it out there. We did a traverse several years ago with tribrach for backsights and pole/bipod for foresights. The closure was poor. We checked centering, we checked plumb. We determined the error must be angular. Turned out the center we were using for the prism was not perpendicular to the axis of the tribrach. It had a slight cant. I was sighting a "candy cane" on top of the prism and because it was elevated above the tribrach a bit, it induced a constant, non-compensating error. We stopped using that center after that. This would affect any traverse in any direction.
Because of your direction, I'd think No. 1 above is a likely candidate.
Obvious issue.
Probably it's a State Plane Coord System Scale Factor problem.
Where I'm at, we use a Lambert System so E-W traverses are not affected.
N-S traverses have a constantly changing scale factor.
Transverse systems are typically (depending on orientation) the opposite as in your case.
I doubt shadows, wind speed, phase of the Moon or whatever have much to do with it.
Thanks for the responses.
A few things I should have added to begin with.
1. We are using a 1" robotic total station
2. I am in Georgia around Atlanta
3. We have looped these traverses back and found really good closures
Manually sighting or letting the robot Auto Track?
If manually, what is the instrument man sighting on?
Are you talking about convergence? I have researched this and I can not figure out how to apply the convergence or what to apply it to.
Autotrack
The residuals for each set was under 1"
This could be attributed to many factors. What may be interesting is to evaluate the static network. How many control points were held? What was the quality of the control points? Were they all HARN points? Were the values held as fixed or according to the Monument Order? A comparison of the minimally constrained and fully constrained adjustment may be enlightening. You may try using the minimally constrained coordinates in your traverse just as a test. Of course, the issue could still be in the conventional traverse. You have to look at everything.
Transverse with not much elevation change. Maybe 15' to 20'
Agreed.
In a word, it's convergence your dealing with. As others pointed out, your no longer surveying on a plane surface. This was the original impetus behind the development of State Plane projections by DOT in the 1930's. Same exact problem they had. If you ran your traverse using a controller/software that allows your data to be computed within the appropriate projection, with a scale factor, mapping angle and earth curvature being applied to your measurements you should be able to minimize the distortion that your seeing. Remember as your traveling East to West, meridians are not parallel but converging.
Just because I'm paranoid, doesn't mean they aren't out to get me.
This article will show you a good example of applying convergence to a linear traverse.
http://archives.profsurv.com/magazine/article.aspx?i=71419
I don't know how much you have considered, but IF you shoot a backsite, to the east, with the sunlight coming from a particular angle, and then the foresite with the sunlight from another angle, it can affect the final outcome. I go with the sunlight shadow theory to answer your enigmatic question.
I often move my eye closer, and farther from the eyepiece of the total station, and it helps me identify the shadow, and the object, I'm trying to site.
N
I agree with Bob Krohn. That is the first thing I thought of, the scale factor.
The way I did it in the old days (before computers and least squares, with a calculator), is that I would run through the traverse once, just using raw distances. That gave me preliminary coordinates. Then I would compute scale and elevation factor for each point using the preliminary SP coordinates. I would use the mean values for each line between points to compute a grid distance for each line. Then I would re-run the traverse using the grid distances.
After all that was done I would start computing closures, adjusting by compass rule, etc.
Curved lines, arcs, convergence, etc have nothing to do with it, in my opinion.
But, the real answer is use a program like star*net or other least squares program. No need to fool with scale factors and elevation factors, they use the ground distance.
Auto track will be impervious to parallax due to shadows. You can probably scratch that theory.
Shadow induced parallax can have low residuals in the set if the operator sights wrong consistently. Since you're using auto track, it's irrelevant.