12/29/2014
Network Least Squares
Adjustments Demystified
By Bruce Carlson and Dean Goodman
Carlson Software
Course Outline
Measurement Errors
Types of Errors
Error propagation
Adjustments
Averaging
Traditional Adjustments
Least Squares Adjustment
Advantages of LSA
Creating a Project in SurvNET
Sample traverse project in Carlson SurvNET
Processing unique datasets
Traverse networks
Traverses with triangulation/resection data
GPS Vectors
Combination of Traverse/GPS data
ALTA Surveys
Certifications
Relative error ellipses
1
12/29/2014
Surveyors make measurements
One aspect of a land surveyor’s job is to make
measurements.
There are no perfect measurements, all
measurements have errors.
Errors are inherent in the instrumentation we
use.
A 5” instrument measures angles to +- 5”.
An EDM may measures distances to +- 0.01’ and
3PPM.
Types of Errors
There are three classifications of
errors:
Blunders
Systematic
Errors
Random Errors
2
12/29/2014
Blunders
Blunders are mistakes and cannot be adjusted.
Blunders MUST be removed from the data set
prior to adjustment.
Examples of Blunders:
Failure to level the instrument.
Failure to put the instrument or target over the
point.
Assigning the wrong point number to a point.
Instrumentation out of adjustment.
Recording the wrong measurement.
Systematic Errors
Systematic Errors are predictable errors and
should be removed from the dataset prior to
adjustment.
Examples of systematic errors:
Atmospheric corrections – temperature, and
pressure.
Curvature and Refraction corrections.
Wrong prism constant
3
12/29/2014
Random Errors
Random errors are errors inherent in the
instrumentation used to make the measurement.
A theodolite may measure to a +- five second
precision.
An EDM may measure to 0.01 feet plus 3 PPM
You can only set your instrument over a point to a
+- 0.01 feet accuracy.
You can only measure from the point to the height
of instrument to a +- 0.01’ accuracy.
Random errors can be adjusted out of the
traverse.
Error Propagation
Random errors propagate throughout the
traverse.
Errors can accumulate, cancel or decrease.
Prior to a least squares adjustments, each
measurement is assigned an initial standard error
based on the equipment specifications.
Error propagation can then be predicted using
statistical, mathematical models.
4
12/29/2014
Error Propagation
As you traverse further from known control
points, the less you know your position.
The next slide shows the error ellipses of a
traverse with one control point.
5
12/29/2014
Error Propagation
Error ellipses increase as you get further from a
known point.
The next slide shows the error ellipses of the
same traverse with two known points,
6
12/29/2014
Error Propagation
Additional control points and redundant
measurements add strength to the traverse and
make the statistics more meaningful.
GPS can be used to add additional control.
Traverse densification and measurements to
points from different locations can add
redundancy to the traverse.
Triangulation can also be useful in adding
redundancy.
Adjustments
Why do we adjust traverses?
All traverses have errors; they do not close exactly
on the terminal point.
If we do not adjust the traverse, all the error is
placed in the last leg of the traverse which is not a
valid assumption.
Error adjustments are important to future work on a
project. Placing all the error in one measurement can
prove problematic for both project design and
layout.
7
12/29/2014
Averaging
Averaging is a type of adjustment.
We average SETS of angles measured in both direct and reverse
faces:
For additional precision
(Sm = S / √N) Standard deviation of the mean equals the standard deviation
of the single measurement divided by the square root of the number of
measurements.
As a check against blunders
To help remove instrument errors
We average distances and zenith angles measured in both faces
and in both directions:
For additional precision
To help remove instrument errors
To mitigate curvature and refraction errors
To improve vertical closures.
Traditional Adjustments
Prior to the advent of high powered computers, when
only calculators or hand calculations were used, there
were three popular adjustments:
Transit Rule
Crandall’s Rule
Compass Rule
Typically angles were balanced prior to adjusting the
traverse with these methods.
8
12/29/2014
Crandall’s Rule
Crandall’ Rule is a “special case” least squares
adjustment.
Crandall’s Rule assumes there is NO error in the angles
– angles/directions are assigned an infinite weight.
Therefore, adjustments are made only to distances.
This adjustment was typically used to match bearings
with previous surveys but under certain conditions can
give unexpected results.
The assumption that angles contain no error is not a
valid assumption.
Transit Rule
Transit Rule adjusts both angles and distances
but makes the assumption that angles are
measured with a higher precision than distances.
This assumption may have been valid in the past
when using a 10” theodolite for angles while
pulling a chain for distances, but it is not valid
for today’s measuring equipment.
9
12/29/2014
Compass Rule
The Compass Rule assumes both angles and
distances are measured with equal precision.
Of the traditional adjustments, this assumption
is most valid for today’s measuring equipment.
The Compass Rule remains a very popular form
of adjustment by surveyors but has distinct
disadvantages when compared to Least Squares
Adjustments.
What is Least Squares?
Least squares is a statistical method used to compute a
best-fit solution for a mathematical model when there
are excess measurements of certain variables making up
the mathematical model.
Least squares requires a mathematical model, a system
of equations. It requires redundant measurements of
one or more variables (the known variables). Lastly it
requires variables that are unknown that are being
solved for.
The least squares criteria is reached when the sum of
the squares of the residuals have been minimized.
Least Squares Adjustment applies the least possible
amount of correction when adjusting the measurements
which arguably makes it the best adjustment.
10
12/29/2014
Advantages of LSA
It allows the simultaneous adjustment of a
network of traverses. Traditional adjustments
can only adjust one traverse at a time.
It allows the combined adjustment of traverse
data, GPS data and level data – 1D, 2D or 3D
adjustments.
It allows complete control of the adjustment
process. Measurements are weighted based on
the equipment used and the number of
measurements made.
Advantages of LSA
It allows processing data of different precision –
weights can be applied to individual measurements or
groups of measurements.
It allows flexible control. The control points can be
anywhere in the traverse, you don’t need to start on a
known point. Control points do not have to be
contiguous and they can be side-shots.
It can handle resection data (measurements from an
unknown point to known points), triangulation (angle-
only measurements) and trilateration (distance-only
measurements).
11
12/29/2014
Advantages of LSA
Automatic reduction to State Plane Coordinate
systems if desired.
Extensive data analysis provides more
information for evaluation of traverse networks.
Enhanced blunder detection tools.
Allows flexible field procedures; the data does
not have to be in any specific order.
Provides tools for ALTA or State survey
certifications.
Common misconceptions of LSA
Least Squares Adjustments are just too
complicated.
Reports are intuitive and easily understood
Flexibility of LSA makes processing of difficult
datasets easy.
Least Squares Adjustments are only necessary
for very precise surveys.
LSA can and should be used for any type survey.
LSA can be used for simple loop traverses as well as
complex traverse networks.
12
12/29/2014
Creating a Project in SurvNET
Choose a coordinate system
Select data files to be processed
Enter pre-processing settings
Enter preliminary standard errors
Select adjustment and output options
Define control points (known points)
Process data
Analyze report
13
12/29/2014
Coordinate System
Select the adjustment model
Choose the type coordinate system for your
project
Local (assumed)
NAD83 State Plane
NAD27 State Plane
UTM
User defined
14
12/29/2014
Input Data Files
Select the data files to be processed
Multiple files of each type can be selected
Traverse data files
GPS Vector data files
Differential/Trig Level files
15
12/29/2014
Preprocessing
Select the preprocessing settings
Curvature and Refraction corrections
Point Substitution feature
Measurement tolerances
Closure computation
16
12/29/2014
Standard Errors
Select preliminary standard errors for
measurement weighting
Standard errors should reflect the realistic error
expectations for the equipment used
Distance errors
Angle errors
Setup errors
Control errors
Leveling and GPS errors
17
12/29/2014
Adjustment Settings
Number of iterations to perform prior to
showing convergence error
Confidence Interval for statistical analysis
ALTA certifications
Select connections for relative error ellipse
calculations
18
12/29/2014
Output Options
Output rounding options
Units
Output coordinate file
Optional coordinate file for ground coordinates
Sample Traverse Network
Sample Least Squares Adjustment project
presented using Carlson SurvNET software
See project PARK.PRJ
19
12/29/2014
Processing Unique Datasets
Least Squares adjustments can process data that
the traditional methods of adjustment cannot
Traverses with non-contiguous control points
Networks of Traverses
Triangulation
Total Station Traverse with GPS vectors
20
12/29/2014
Non-contiguous
control points
See SEP1.PRJ and SEP2.PRJ
21
12/29/2014
Traverse Networks
See N_DRUIDH.PRJ and
DEANPROP.PRJ
22
12/29/2014
23
12/29/2014
Traverse with
Triangulation
See NORTHLN6.PRJ
24
12/29/2014
GPS Vectors
See “CC&V Control.prj”
25
12/29/2014
Traverse combined
with GPS vectors
See M07052.PRJ
26
12/29/2014
ALTA and Kentucky Survey
Standards
ALTA Title Surveys require that the surveyor
make the following Certification:
The relative positional precision between any two
property corners shown on the survey shall be
within 0.07 of a foot plus 50 parts per million at a
95% confidence level
This certification requires the statistical model
produced by a least squares adjustment.
27
12/29/2014
What is Relative Positional Precision?
2011 ALTA Standards
28
12/29/2014
Kentucky Survey Standards
The Kentucky Survey Standards as stated in
KAR 18:150 are similar to the ALTA standards
we just discussed.
KAR 18:150 Standards of Practice
29
12/29/2014
KAR 18:150 Standards of Practice
KAR 18:150 Standards of Practice
30
12/29/2014
Relative Error Ellipses
The determination of the Relative Positional Precision
between any two points requires the calculation of the
relative error ellipse between the points
Relative error ellipses are confidence regions
established for the coordinate difference between any
two points.
This is not the same thing as an error ellipse. Error
ellipses are the measure of the positional error of a
point. It is defined by an ellipse which is defined by a
semi-major axis and semi-minor axis., and the direction
of the maximum error.
Sample ALTA Survey Project
Sample ALTA Survey project presented using
Carlson SurvNET
See project ALTA.PRJ
31
12/29/2014
Thank you for your
attention
32
