I like that idea. I'll do that first. If the angle varies from what I'm only assuming it is now (and in the right direction, i.e. the interior angle is somewhat bigger than shown now. It'd take about 24' error though to make up all the closure though, and that's quite a bit, but you never know until you measure.
If it turns out that way, I've saved myself a lot of work.
Here is the Star*Net .dat file for the case I ran above, disregarding some data so there was only one check measurement. You can change this as needed to run your actual data. Everything else I tried had vastly poorer fit.
# Traverse by "rfc" on forum 2014 July 28
# Disregarding distance 2-3 and angle at 4
C 1 1000 1000 ! !
B 1-2 216-56-16 !
D 1-2 384.65 0.02
D 2-3 43.2 * # was this 37.7 horiz and 43.2 slope?
D 3-4 96.09 0.02
D 4-5 203.95 0.02
D 5-6 75.222 0.02
D 6-7 137.282 0.02
D 7-8 141.665 0.02
D 8-9 329.635 0.02
D 9-1 376.83 0.02
A 9-1-2 141-05-10 10
A 1-2-3 85-04-07 10
A 2-3-4 242-29-37 10 # no angle measured
A 3-4-5 93-02-07 *
A 4-5-6 136-24-50 10
A 5-6-7 129-29-40 10
A 6-7-8 215-48-20 10
A 7-8-9 179-06-18 10
A 8-9-1 36-45-18 10
I'm curious about your field procedures. Are you shooting a reflector on a hand-held rod? If so, what is being done to stabilize the rod?
> I'm curious about your field procedures. Are you shooting a reflector on a hand-held rod? If so, what is being done to stabilize the rod?
Well, the simple answer would be "my wife"! She's pretty stable, I'd say! 😀
A few of the longest back sights were set up on a tripod (1 to POB, 2 to 1, 8 to POB).
The others, she used the prism pole with a circular vial at the 5' level of the pole.
When shooting, If I observed the prism moving about some, I'd call "ready?", then carefully sight the middle of any "departures" the prism made back and forth. I'd guess that at an average of 150', there was never more than perhaps 1/2" of "woga-woga", which translates to less than a minute.
It could be a contributing factor, and I'll pay attention next time round, but I'm looking for nearly a degree here (47' if it was all in one place). That's huge.
The lack of repetitive measurements really puts a kabosh on trying to find errors after the fact. Big mistake.
It's true that rod movement couldn't be the source of a 47 minute error. I'd suspect something else, such as turning the instrument with the horizontal motion not fully locked. On an older instrument that has not been cleaned and lubricated recently, the motion lock can sometimes be a little uncertain, and can move when it appears to be locked. Doubling the angles would of course enable you to spot that right away.
To double angles with a single rod person, it's better to set a fixed backsight so the rod person doesn't have to move back and forth. I generally use a 6-foot piece of 1 x 2, setting it a couple of inches ahead of the backsight point and sighting a plumb bob string to get it plumb and on line between the point and the gun.
On handheld rod shots, I've had good results with giving the rod person two 4-foot lath to help stabilize the rod. He/she plants the lath out to the sides and slightly in back of the rod, and uses one or both hands to clamp the rod and lath together, allowing the lath to slide up and down a bit until the bubble is centered and then holding them firmly. That would get rid of the 1/2" movement you're seeing, although it isn't the immediate problem.
Some progress
Well, I'm down to 2.22'...a thrilling 1:809 closure ratio! But I've cut my error in half at least.
I set up on station 3, doubled angles twice (Face Left and Right). Sure enough, that angle measured 243°25' 19" compared to the previously assumed 242°29' 37". This is more than the 47' I was looking for.
Unfortunately, all of the remaining calls (5,6,7,8, etc.) remained what they were, so I had to go back and put the angles I'd turned back in, as if I had traversed it in reverse.
Finally I put back in a short dog leg I'd taken out in my previous posted file (probably because I was too embarrassed to admit I calculated the angle at point 8 directly back to the POB because I made the lame decision to originally shoot to the top of a rock next to the driveway at point 8 rather than just stay in the middle of the stinkin' road.
The problem was, the rock wasn't big enough for me and the instrument and tripod, so I had hop up carefully without banging anything, try to shoot, hop down and record...it's a long story, but I learned (as some advised in another thread about setting up on difficult monuments): Set up where the ground is stable, and accessible, and take side shots.
As a result, I suspect my readings between there and the POB, so before I proceed with Compass Rule, Least Squares, etc., I'm going to re-do the sight from POB directly to 8, and also set up on 8 and shoot back to POB.
After all, it's not the destination that counts, it's how you get there, isn't it? I like doing things the hard way. Stay tuned.
Proper way to close with differential angle confidence
I've re-measured the angles and distances at points 5,6,7,8, using double, triple or quadruple angles. High degree of confidence that they're right. I'm also very confident of my measured distances.
I've moved the POB to where 8 was (The infamous rock). It looks like this will close by adjusting angles 1,2,3,and 4. I could do that "manually".
Or, I could just use the Compass rule and make it all close. But is it proper to differentially adjust angles that you've measured in the field if your confidence is lower on those measurements? In other words, what role does subjectivity play in closing a traverse?
My gut feeling tells me that if I'm still this far out, I should go back out and re-measure 1,2,3, and 4, using more careful redundant techniques, and this I will do, but raise the question wondering what an LS would do if say he had to travel a gazillion miles back to the site?
I give up
Well I pretty much did the whole thing again. Still sitting on a .3' disclosure on an 1800' traverse. My internal angles are off by 7'35"...still more than (I think) should be capable of a 10" instrument. Can that much angular misclosure be due to sloppy leveling?
More than once I looked at the leveling after having completed a set of angles, and noticed it was off. (Instrument has only a single axis of compensation...don't know if that affects things, but it must.
Anyway, it's at a closure/perimeter ratio of 1:5961, before any mucking about with the Compass rule or Least squares. I'll try Starnet next, and I'll consider this one a good first step with a long way to go.
Looking better
I suspect a lot of the remaining error is centering on the points. For instance, if your target leans off 0.05 ft (5/8 inch) for your shortest sight that is 4 minutes of angle right there. It wouldn't do much good to measure that angle to an arc second.
If you get into least squares, I would recommend adding any angle and/or distance sights you can across the figure, such as POB-6, 1-6, 2-6, 3-6, 2-5, or 3-5, and angle 3-2-6 or 3-1-6. The more you add, the better it can do at checking for you and averaging out the errors.
I'm curious how you got lat-lon values, and whether you took the elevation into account in converting distances. You may not be at a high enough elevation that it matters at your level of accuracy, but at high altitudes or in precision work elsewhere, it is necessary to account for the difference in the radius of the earth at your location versus the ellipsoid model.
Looking better
> I'm curious how you got lat-lon values, and whether you took the elevation into account in converting distances. You may not be at a high enough elevation that it matters at your level of accuracy, but at high altitudes or in precision work elsewhere, it is necessary to account for the difference in the radius of the earth at your location versus the ellipsoid model.
Thanks for the input.
It would be hard to get across the traverse (dense woods and brambles), but I might be able to get one or two.
Elevation is around 1000' I put the Temp and Pressure in the TS before taking shots.
The Lat/Long comes from Google Earthing the starting point from the corner of the property.
Looking better
I was putting your coordinates into GE while you posted. I see the difficulty of sight lines. However, 2-7 is a high priority, along with whatever angle you can get there.
https://dl.dropboxusercontent.com/u/25124076/rfc.kml
Are points 1 and 2 really in the middle of the road where GE shows them? Sounds like a dangerous place to take angles.
I think your longitude listed for POB is wrong, or else the vector POB-1, since all the others make a reasonable figure to match the picture you posted and POB doesn't fit the distances to 1 and 7.
Looking better
> I was putting your coordinates into GE while you posted. I see the difficulty of sight lines. However, 2-7 is a high priority, along with whatever angle you can get there.
> https://dl.dropboxusercontent.com/u/25124076/rfc.kml
>
> Are points 1 and 2 really in the middle of the road where GE shows them? Sounds like a dangerous place to take angles.
>
> I think your longitude listed for POB is wrong, or else the vector POB-1, since all the others make a reasonable figure to match the picture you posted and POB doesn't fit the distances to 1 and 7.
Oh, crap! In all my changes, the POB moved from where point 4 was!
The Lat/Long of point 4 is the Lat/Long of the POB. That is, the Lat/Long of POB shown in the table is really where point 4 is. Everything is west of there.
I guess my software is not smart enough to know that when the POB moves to a different point on the traverse, so does the Lat/Long!
You're sure right about trying to take angles in the middle of a State Highway, lol! I'll try to fix this and re-post.
Looking better
I moved the POB back to where it should be. This should put the plat in the right place on GE.
Still not quite
The POB on this version is not at the right bearing or distance from point 1 or point 7. Holding those other points I find the longitude of POB to be near 72°34'59.7"W, which is about 95 ft east from that listed, and latitude about as listed.
I don't think it is all the other points that are wrong because they plot to show lines of sight along driveways.
Still not quite...How about now?
You are absolutely right, Sir. One thing I'm learning is that you can never be too careful, too methodical with the information you're putting down.
In this case, my only link between my traverse and real coordinates is Google Earth. I manually zoomed in on a visible location, read the coordinates, and manually entered them into my plotting program. But that was long ago, and so much has changed that I never went back to see if everything was still correct.
Thanks for your sharp eye (and for taking the time to point out the errors of my ways!
The Final Chapter
Well, I think I've beat this dead horse enough. I've pretty much re-done the whole thing, start to finish; eliminated the "bad corner" at the southwest end. But everything else is exactly as I shot it...no adjustments yet. The closure ratio is still shy of 5000, maybe good enough for a rural survey (but not necessarily good enough for me. I'm still over 2' 15" high on the Angle Sum, which seems high, given my expected standard errors on angles.
Making educated guesses on standard errors in Starnet gives me a 1.185 error factor, which passes; the traverse will close easily of course using either least squares or the compass rule.
Looking back on it, other than the relatively "construction grade" instrument, as someone put it, I did think of one source of error on the distances is my inconsistency setting both the instrument height and Reflector heights precisely. I did a little trig and noted that if one is .5' out between the two, that alone can throw a distance off by a tenth or more. I'm more than3 tenths from closing, so I'm sure that could have been a factor.
Finally, I've been reading with interest the thread on prism centering standard errors. I'm not convinced the vial on prism poles is worth a hill of beans. I'm not sure what the angular error is if the bubble touches the centering circle, but I bet it's enough to throw a tenth of error in distance (or angular), if it's not dead nuts.
On to the next challenge.
The Final Chapter
It is easy to check a prism pole...
1. Assuming you don't have a push button tripod...steady the pole against the inside corner of a desk with return or tailgate in carpet or soft ground (to secure the point), such that the bubble is centered. Rotate the prism pole 180 degrees, the bubble should stay in the center or adjust out HALF the error.
Or
2. Set it up and look at it with your leveled total station. Rotate 90 degrees and do the same.