I've been contemplating the process of local "network development"...planning stations and measurements to achieve the highest degree of local accuracy (no grid, SPC etc. involved).
In the course of doing that, I've been thinking about whether measuring distances is inherently better than measuring angles.
For example, let's say you're establishing a line between two distant points (property line, pipeline, whatever). If you can't do both due to terrain or whatever, are you better off:
A: Shoot precisely along the line, establishing points at various distances, or
B: Getting to a point offset from the line, and measuring the points in a fan shape along the line.
"A" seems that it would be better at making all the points in one straight line, but is it the "right" straight line?
"B" seems better in that the precision at many distances close to perpendicular to the line would be better than angles would be.
I get that multiple triangles--a mesh--with redundant measurements in every direction from as many points as you can get to, to as many points as you can measure to (redundantly if possible), is the best, but is there a "distance threshold" or other aspect to consider, that pushes the "balance of errors" between angles and distances, one way or another?
As always, accept apologies in advance if the question falls into either the "dumb", or "easily answered with 25 years of experience" categories.
> I've been contemplating the process of local "network development"...planning stations and measurements to achieve the highest degree of local accuracy (no grid, SPC etc. involved).
>
> In the course of doing that, I've been thinking about whether measuring distances is inherently better than measuring angles.
Well, presumably you're going to adjust your network by least squares and will be using LSA software to do the number crunching. Does your software have a pre-planning feature similar to Star*Net's that allows you to answer these questions in advance of actually making the measurements?
Remember that distance * tangent(angle error) tells you how far off sideways you might expect to be. That's a good start for comparing angle versus distance error. Always consider the centering errors in addition to the instrument errors.
And as Kent says, pre-planning with approximate values in a least squares program will give you a much better feel for the expected results. If the software has a specific pre-planning feature, use it. Otherwise, just make up values that are somewhat realistic and run the simulation.
> Well, presumably you're going to adjust your network by least squares and will be using LSA software to do the number crunching. Does your software have a pre-planning feature similar to Star*Net's that allows you to answer these questions in advance of actually making the measurements?
I can use Star*Net, and have read (from you) that the feature is in there, but admit I haven't really looked at it or understand what it does. If it's the "Pre-analysis" button, if I recall correctly, I thought that had more to do with GPS adjustments. I'll spend some more time in that part of the manual (If that's the "pre-planning" feature you speak of).
> I can use Star*Net, and have read (from you) that the feature is in there, but admit I haven't really looked at it or understand what it does. If it's the "Pre-analysis" button, if I recall correctly, I thought that had more to do with GPS adjustments.
Pre-Analysis is what will enable you to simulate various combinations of conventional measurements in a traverse or network. You just have to know the standard errors that will apply to those measurements and have to specify some coordinates to show the approximate spacing between stations and the general shape of the traverse or network.
Then, it is just a matter of trying different numbers and combinations of measurements to see which most easily hits your uncertainty targets for the points positioned.
> > I can use Star*Net, and have read (from you) that the feature is in there, but admit I haven't really looked at it or understand what it does. If it's the "Pre-analysis" button, if I recall correctly, I thought that had more to do with GPS adjustments.
>
> Pre-Analysis is what will enable you to simulate various combinations of conventional measurements in a traverse or network. You just have to know the standard errors that will apply to those measurements and have to specify some coordinates to show the approximate spacing between stations and the general shape of the traverse or network.
>
> Then, it is just a matter of trying different numbers and combinations of measurements to see which most easily hits your uncertainty targets for the points positioned.
So, taken the theoretical E/W line between points 2 and 6, and the offset point 1, which is visible to one or more of the other points, you'd just put either angles or distances where you think you could see from one station to another?
Does the output of the pre-analysis look just like a regular analysis?
[pre]
C 1 0 2000
C 2 2000 0
C 3 2000 1000
C 4 2000 2000
C 5 2000 3000
C 6 2000 4000
[/pre]
> Does the output of the pre-analysis look just like a regular analysis?
Well, here's one ready to run that was taken from an example another poster provided a few months ago.
[pre]
C 101 4982.03900 5037.07700 ! !
C 109 4743.02935 5056.34458
C 103 4979.02416 5524.66468
C 102 4951.35739 5181.78878
C 105 5024.97915 5000.99832
C 106 4966.74447 5207.68883
C 107 4973.26955 5042.13586
C 108 4975.03713 4998.98416
C 177 4706.67948 5040.78253
C 178 4737.58253 5178.40789
C 219 4705.93120 5206.21916
B 101-109
A 102-101-103
D 101-103
D 101-102
A 102-101-105
D 101-105
A 102-101-106
D 101-106
A 102-101-107
D 101-107
A 102-101-108
D 101-108
A 102-101-109
#A 103-101-109
D 101-109
A 101-109-177
D 109-177
A 101-109-178
D 109-178
A 109-178-102
D 178-102
A 109-178-102
D 178-102
A 109-178-219
D 178-219
#A 178-102-101
[/pre]
These were the settings for that one:
[pre]
Project Option Settings
STAR*NET Run Mode : Preanalysis
Type of Adjustment : 2D
Project Units : FeetUS; DMS
Coordinate System : LOCAL
Default Project Elevation : 0.0000 FeetUS
Apply Average Scale Factor : 1.0000000000
Input/Output Coordinate Order : North-East
Angle Data Station Order : From-At-To
Distance/Vertical Data Type : Slope/Zenith
Convergence Limit; Max Iterations : 0.010000; 10
Default Coefficient of Refraction : 0.070000
Earth Radius : 6372000.00 Meters
Create Coordinate File : Yes
Create Ground Scale Coordinate File : No
Create Dump File : No
Instrument Standard Error Settings
Project Default Instrument
Distances (Constant) : 0.005000 FeetUS
Distances (PPM) : 2.000000
Angles : 4.000000 Seconds
Directions : 2.800000 Seconds
Azimuths & Bearings : 1.000000 Seconds
Centering Error Instrument : 0.001000 FeetUS
Centering Error Target : 0.003000 FeetUS
[/pre]
> > Does the output of the pre-analysis look just like a regular analysis?
>
> Well, here's one ready to run that was taken from an example another poster provided a few months ago.
>
> [pre]
> C 101 4982.03900 5037.07700 ! !
> C 109 4743.02935 5056.34458
> C 103 4979.02416 5524.66468
> C 102 4951.35739 5181.78878
> C 105 5024.97915 5000.99832
> C 106 4966.74447 5207.68883
> C 107 4973.26955 5042.13586
> C 108 4975.03713 4998.98416
> C 177 4706.67948 5040.78253
> C 178 4737.58253 5178.40789
> C 219 4705.93120 5206.21916
>
>
>> [/pre]
Thank you!
I'm looking for that thread to read more, but for now...
Were the coordinates of the stations put in prior to the analysis, or did the analysis come up with them?
> Were the coordinates of the stations put in prior to the analysis, or did the analysis come up with them?
You need to enter approximate coordinates for the stations on the network or traverse. You don't enter any measurements at all in the pre-analysis, just which angles and which distances will be measured and how to form their standard errors.
In that case, the poster had generated precise coordinates, so I just copied them. The coordinates could have as easily been rounded to the nearest foot without any likely ill effect.
For large-scale work, a surveyor could generate the approximate station coordinates by various means, including:
- prior surveys,
- topo maps, or
- recon with a handheld GPS receiver.
The idea is to get somewhat realistic spacings on stations and a rough picture of the geometry.