Adjustments

Created by:
Khiel S. Yumul
BSGE-3A
Adjustment
of Level
Nets

2
Introduction
Adjustment of Level Nets
▸ Differential leveling observations
are used to determine differences in
elevation between stations.
▸ As with all observations, these
measurements are subject to
random errors that can be adjusted
using the method of least squares.
4
▸ The observation equation
method for adjusting differential
leveling observations by least
squares is developed, and several
sample problems are given to
illustrate the adjustment
procedures.
5
OBSERVATION EQUATION
To apply the method of least squares in leveling adjustments,
a prototype observation equation is first written for any
elevation difference. The figure in the next slide illustrates
the functional relationship for an observed elevation
difference between two stations I and J .
The equation is expressed as:

Ej − Ei = ∆Elevij + ν∆Elevij

6
OBSERVATION EQUATION
This prototype observation equation relates the unknown elevations of
any two stations, I and J , with the differential leveling observation ∆Elevij
and its residual ∆νElevij . This equation is fundamental in performing least
squares adjustments of differential level nets.

Differential leveling 7
UNWEIGHTED EXAMPLE
In the figure below, a leveling network and its survey data are shown.
Assume that all observations are equal in weight. In this figure, arrows indicate
the direction of leveling and thus, for line 1, leveling proceeds from benchmark
X to A with an observed elevation difference of +5.10 ft.

Interlocking leveling 8
UNWEIGHTED EXAMPLE
By substituting into prototype (Ej − Ei = ∆Elevij + ν∆Elevij) an
observation equation is written for each observation in the figure. The
resulting equations are
A −BM X = 5.10 + ν1
−A +BM Y = 2.34 + ν2
C −BM Y = −1.25 + ν3
−C +BM X = −6.13 + ν4
−A B = −0.68 + ν5
B −BM Y = −3.00 + ν6
−B C = 1.70 + ν7
9
UNWEIGHTED EXAMPLE
Rearranging so that the known benchmarks are on the right-hand side
of the equations and substituting in their appropriate elevations yields

A = + 105.10 + ν1
−A = − 105.16 + ν2
C= + 106.25 + ν3
−C = − 106.13 + ν4
−A +B = − 0.68 + ν5
B = + 104.50 + ν6
−B C= + 1.70 + ν7
10
UNWEIGHTED EXAMPLE
In this example there are three unknowns, A, B, and C. In matrix form
Ej − Ei = ∆Elevij + ν∆Elevij , are written as
AX + B = L + V
In Equation AX + B = L + V, the B matrix is
Where,
a vector of the constants (benchmarks)
collected from the left side of the equation
and L a collection of leveling observations.
The right side of equation is equal to L − B.
It is a collection of the constants in the
observation equations and is often referred to
as the constant matrix, L. Since the
benchmarks can also be thought of as
observations, this combination of
benchmarks and leveling observation is
referred to as L in this slide and equation AX
+ B = L + V is simplified as AX = L + V 11
UNWEIGHTED EXAMPLE
Also note in the A matrix that when an unknown does not appear in an
equation, its coefficient is zero. Since this is an unweighted example,
according to equation ATAX = ATL the normal equations are

12
UNWEIGHTED EXAMPLE
The rearranged form of is used to compute the residuals as

The most probable elevations for A, B, and C are 105.14, 104.48,

and 106.19, respectively.
13
UNWEIGHTED EXAMPLE
The rearranged form of equation AX = L + V is used to compute the
residuals as
V = AX -- L

14
WEIGHTED EXAMPLE
▸  It was shown that relative weights for adjusting level lines are inversely
proportional to the lengths of the lines:
W=

The application of weights to the least squares adjustment of the level

circuit is illustrated by including the variable line lengths for the
unweighted example of slide no. 11. These line lengths for the leveling
network in the figure and their corresponding relative weights are given in
the figure. For convenience, each length is divided into the constant 12, so
that integer “relative weights” were obtained. (Note that this is an
unnecessary step in the adjustment.)

15
WEIGHTED EXAMPLE
The observation equations are now formed as the equation in slide #11,
except that in the weighted case, each equation is multiplied by its
weight.

16
WEIGHTED EXAMPLE
After dropping the residual terms in slide #17, they can be written
in terms of matrices as

17
WEIGHTED EXAMPLE
Weights for example in slide #10

LENGTH RELATIVE
LINE
(miles) WEIGHTS
1 4 12/4 = 3
2 3 12/3 = 4
3 2 12/2 = 6
4 3 12/3 = 4
5 2 12/2 = 6
6 2 12/2 = 6
7 2 12/2 = 6

18
WEIGHTED EXAMPLE
Applying the equation (ATWAX = AT WL), we find that the
normal equations are
(ATWA) X = NX = ATWL

19
WEIGHTED EXAMPLE
By the equation X = (AT WA)−1 AT WL = N−1AT WL, the solution
for the X matrix is

20
WEIGHTED EXAMPLE
The residual equation (V = AX − L) is now applied to compute the
residuals as
It should be noted that these adjusted
values (X matrix) and residuals (V
matrix) differ slightly from those
obtained in the unweighted adjustment of
equation in
Slide #11 .

This illustrates the effect of weights on an

adjustment. Although the differences in
this example are small, for precise-level
circuits it is both logical and wise to use a
weighted adjustment since a correct
stochastic model will place the errors
back in the observations that probably
produced the errors. 21
REFERENCE STANDARD
DEVIATION
The standard deviation for a weighted set of observations as
 
𝜮 𝒘𝒗 𝟐
𝑺 𝟎=
√𝒏− 𝟏
However, in the equation above, it applies to a multiple set of
observations for a single quantity where each observation has a
different weight. Often, observations are obtained that involve
several unknown parameters that are related functionally like those
in equations in slide #11 and slide #17 . For these types of
observations, the standard deviation in the unweighted case is

22
REFERENCE STANDARD
DEVIATION
▸ the  equation in previous slide, ν2 is expressed in matrix form as VTV, m is the
In
number of observations, and n is the number of unknowns. There are r = m − n
redundant measurements or degrees of freedom.
The standard deviation for the weighted case is

 Where wν2 in matrix form is VT WV . Since these standard deviations relate to

the overall adjustment and not a single quantity, they are referred to as
reference standard deviations. Computation of the reference standard
deviations for both unweighted and weighted examples is illustrated in the nest
slides
23
Unweighted Example
In the example in slide #11, there are 7 – 3, or 4 degrees of freedom. Using the
  𝜮𝒗𝟐
unweighted example is
𝑺 𝟎=
residuals V = AX − L and using
√ the reference standard deviation in the
𝒓

24
Unweighted Example
  𝑽 𝑻𝑽
This can be computed using the matrix expression
√
𝑺 𝟎=
𝒓
as

  𝑽 𝑻𝑽
𝑺 𝟎=
√ 𝒓

= 

25
 Weighted Example
Notice that the weights are used when computing the reference
standard deviation in   𝜮𝒗𝟐
𝑺 𝟎=
√ 𝒓
That is, each residual is squared and multiplied by its weight, and thus the
reference standard deviation computed using nonmatrix methods is
 

= 

26
Sample Problems
and Computations
 Weighted Adjustment
The level net shown below is observed with the following results (the elevation
differences and standard deviations are given in meters, and the elevation of A is
437.596 m):

What are the most probable values

for the elevations of B, C, and D?

28
Weighted Adjustment (Solutions)
1. Write the observation equations without their weights:

2. Rewrite observation equations in matrix form AX = L + V as:

29
Weighted Adjustment (Solutions)
 
3. In accord with equations σ02 −1
and wi form the weight matrix as

Where,

30
Weighted Adjustment (Solutions)
4. Compute the normal equations using the equation

(ATWA) X = NX = AT WL
Where,

31
Weighted Adjustment (Solutions)
5. Solve for the X matrix using the equation X = (AT WA)−1 AT WL = N−1 AT WL
yields

6. Compute the residuals using the matrix expression V = AX − L

32
Weighted Adjustment (Solutions)
7. Calculate the reference standard deviation for the adjustment using
  𝑽 𝑻 𝑾𝑽
the matrix expression which is the equation𝑺 𝟎=
√ 𝒓

33
Weighted Adjustment (Solutions)
Since the number of system redundancies is the number of observations
minus the number of unknowns, r = 6 − 3 = 3, and thus
= 

8. Tabulate the results showing the adjusted elevation differences, their

residuals, and final adjusted elevations.
Adjusted Adjusted
From To Residual Station
∆Elev. Elevation
A B 10.513 0.0004 A 437.596
B C 5.36 0 B 448.109
C D -8.525 -0.002 C 453.468
D A -7.348 0 D 444.944
B D -3.165 0.002
A C 15.872 -0.009
34
Triangulatio
n
Introduction
Triangulation
▸ Prior to the development of electronic
distance measuring equipment and the global
navigation satellite systems, triangulation was
the preferred method for extending horizontal
control over long distances. The positions of
widely spaced stations were computed from
observed angles and a minimal number of
observed distances called baselines. This
method was used extensively by the National
Geodetic Survey in extending much of the
national network.
37
▸ A least squares triangulation adjustment can
use condition equations or observation
equations written in terms of either
azimuths or angles. In this chapter the
observation equation method is presented.
The procedure involves a parametric
adjustment where the parameters are
coordinates in a plane rectangular system
such as state plane coordinates. In the
examples, specific types of triangulations
known as intersections, resections, and
quadrilaterals are adjusted. 38
ADJUSTMENT OF
INTERSECTIONS
When an unknown station is visible from two or more
existing control stations, the angle intersection method is one
of the simplest and sometimes most practical ways of
determining the horizontal position of a station. For a unique
computation, the method requires the observation of at least
two horizontal angles from two control points.

39
ADJUSTMENT OF
INTERSECTIONS
For example, angles θ1 and θ2 observed from control stations R and S in
the figure below will enable a unique computation for the position of
station U. If additional control is available, computations for the unknown
station’s position can be strengthened by observing redundant angles such
as angles θ3 and θ4 in the figure and applying the method of least squares.
In that case, for each of the four independent angles, a linearized
observation equation is written in terms of the two unknown coordinates
Xu and Yu .

Intersection
example

40
ADJUSTMENT OF RESECTIONS
Resection is a method used for determining the unknown
horizontal position of an occupied station by observing a
minimum of two horizontal angles to a minimum of three
stations whose horizontal coordinates are known. If more than
three stations are available, redundant observations are
obtained and the position of the unknown occupied station can
be computed using the least squares method. Like intersection,
resection is suitable for locating an occasional station and is
especially well adapted over inaccessible terrain.

41
ADJUSTMENT OF RESECTIONS
Consider the resection position computation for the occupied
station U of the figure below having observed the three horizontal
angles shown between stations P, Q, R, and S whose positions are
known. To determine the position of station U, two angles could be
observed. The third angle provides a check and allows a least squares
solution for computing the coordinates of station U.

Resection
example

42
ADJUSTMENT OF TRIANGULATED
QUADRILATERALS
The quadrilateral is the basic figure for triangulation.
Procedures like those used for adjusting intersections and
resections are also used to adjust this figure. In fact, the
parametric adjustment using the observation equation method
can be applied to any triangulated geometric figure, regardless
of its shape.

43
ADJUSTMENT OF TRIANGULATED
QUADRILATERALS
The procedure for adjusting a quadrilateral consists of
first using a minimum number of observed angles to solve
triangles, and computing initial values for the coordinates.
Corrections to these initial coordinates are then calculated by
applying the method of least squares. The procedure is iterated
until the solution converges. This yields the most probable
coordinate values. A statistical analysis of the results is then
made

44
Sample Problems
and Computations
ADJUSTMENT OF
INTERSECTIONS
Using the method of least squares, compute the most probable
coordinates of station U in the figure. The following equally weighted
horizontal angles were observed from control stations R, S, and T:

θ1 = 50o 06’ 50” θ2 = 101o 30’ 47” θ3 = 98o 41’ 17” θ4 = 59o 17’ 01”

The coordinates for control stations R, S, and T are

xr = 865.40 xs = 2432.55 xt = 2865.22
yr = 4527.15 ys = 2047.25 yt = 27.15

46
ADJUSTMENT OF INTERSECTIONS
(Solutions)
1. Determine initial approximations for the coordinates of station U.
(a) Using the coordinates of stations R and S, the distance RS is computed
as  

= 2933.58 m

(b) From the coordinates of stations R and S, the azimuth of the line between R
and S is determined using the Azimuth Observation Equation. Then the initial
azimuth of line RU is computed by subtracting θ1 from the azimuth of line RS.

47
ADJUSTMENT OF INTERSECTIONS
(Solutions)
1. Determine initial approximations for the coordinates of station U.
(a) Using the coordinates of stations R and S, the distance RS is computed
as  

= 2933.58 ft

(b) From the coordinates of stations R and S, the azimuth of the line between R
and S is determined using the Azimuth Observation Equation. Then the initial
azimuth of line RU is computed by subtracting θ1 from the azimuth of line RS.

48
ADJUSTMENT OF INTERSECTIONS
(Solutions)
(c) Using the law of sines with triangle RUS, an initial length for RU 0 can be
calculated as

(d) Using the azimuth and distance for RU0 computed in steps 1(b) and (c),
initial coordinates for station U are computed as

49
ADJUSTMENT OF INTERSECTIONS
(Solutions)
2. Formulate the linearized equations. As in the trilateration adjustment, control
station coordinates are held fixed during the adjustment by assigning zeros to
their dx and dy values. In forming the observation equations, b, i, and f are
assigned to the backsight, instrument, and foresight stations, respectively, for
each angle. For example, with angle θ 1, B, I, and F are replaced by U, R, and S,
respectively. By combining the station substitutions shown in the table with
prototype equation, the following observation equations are written for the four
observed angles:

50
ADJUSTMENT OF INTERSECTIONS
(Solutions)

51
ADJUSTMENT OF INTERSECTIONS
(Solutions)

Notice that the initial coordinates for Xu0 and Yu0 were calculated using θ1
and θ2, and thus their K-matrix values are zero for the first iteration. These
values will change in subsequent iterations. 52
ADJUSTMENT OF INTERSECTIONS
(Solutions)
3. Matrix solution. The least squares solution is found by applying the equation
X = (JTJ )−1 JTK =
N−1JT K

53
ADJUSTMENT OF INTERSECTIONS
(Solutions)
4. Add the corrections to the initial coordinates for station U

5. Repeat steps 2 through 4 until negligible corrections occur. The next iteration
produced negligible corrections, and thus equations in #4 produced the final
adjusted coordinates for station U.
6. Compute post-adjustment statistics. The residuals for the angles are

54
ADJUSTMENT OF INTERSECTIONS
(Solutions)
The reference standard deviation for the adjustment is computed using equation
  𝑽 𝑻𝑽
𝑺 𝟎=
√ 𝒓

55
ADJUSTMENT OF INTERSECTIONS
(Solutions)
The estimated errors for the adjusted coordinates of station U, given by
equation, are

The estimated error in the position of station U is given

by

56
ADJUSTMENT OF RESECTION
The following data are obtained for Figure 15.4. Control stations P, Q, R, and S
have the following (x, y) coordinates: P (1303.599, 1458.615), Q (1636.436,
1310.468), R (1503.395, 888.362), and S (1506.262, 785.061). The observed
values for angles 1, 2, and 3 with standard deviations are as follows:

What are the most probable coordinates of station U ?

57
ADJUSTMENT OF RESECTION
(SOLUTIONS)
1. From equation

2. Substituting the appropriate angular values into ∠QPU + ∠URQ = G =

360◦ − (∠1 + ∠2 + ∠RQP) gives

3. Substituting the appropriate station coordinates into

yields
PQ = 364.318 and QR = 442.576

58
ADJUSTMENT OF RESECTION
4. Substituting the appropriate values into yields H as

5. Substituting previously determined G and H into equation

∠URQ is computed as

6. Substituting the value of ∠URQ into equation ∠RQU = 180o − (∠2 +

∠URQ), ∠RQU is determined to be

59
ADJUSTMENT OF RESECTION
4. Substituting the appropriate values into yields H as

5. Substituting previously determined G and H into equation

∠URQ is computed as

6. Substituting the value of ∠URQ into equation ∠RQU = 180o − (∠2 +

∠URQ), ∠RQU is determined to be

60
ADJUSTMENT OF RESECTION
4. Substituting the appropriate values into yields H as

5. Substituting previously determined G and H into equation

∠URQ is computed as

6. Substituting the value of ∠URQ into equation ∠RQU = 180o − (∠2 +

∠URQ), ∠RQU is determined to be

61
ADJUSTMENT OF RESECTION
7. From the equation , RU is

8. Using the equation azimuth = α + C, the azimuth of RQ is

9. From the figure, AzRU is computed as

62
ADJUSTMENT OF RESECTION
10. Using the equation, the coordinates for

station U are

For this problem, using prototype equation, the J and K matrices are

63
Trilateration
Introduction
Trilateration
▸ Horizontal surveys are performed for the purpose
of determining precise relative horizontal positions
of points. They have traditionally been
accomplished by trilateration, triangulation, and
traverse. These traditional types of surveys involve
making distance, direction, and angle observations.
As with all types of surveys, errors will occur in
making these observations, and thus they must be
analyzed and, if acceptable, adjusted. In the
following three chapters procedures are described
for adjusting trilateration, triangulation, and
traverse surveys, in that order.
66
▸ Horizontal surveys, especially those covering a
large extent, must account for the systematic
effects of Earth curvature. One way this can be
accomplished is to do the computations using
coordinates from a mathematically rigorous map
projection system such as the state plane
coordinate system (SPCS), universal transverse
Mercator (UTM) system, or a local plane
coordinate system that accounts rigorously for
Earth curvature

67
▸ Recently, global navigation satellite systems
(GNSS), such as GPS have gradually been
replacing these traditional procedures for
conducting precise surveys. In fact, GNSS
surveys not only yield horizontal positions, but
give ellipsoidal heights as well. Thus, GNSS
provides three-dimensional surveys. Again as
with all observations, GNSS surveys contain
errors and must be adjusted.

68
DISTANCE OBSERVATION
EQUATION
In adjusting trilateration surveys using the parametric least squares
method, observation equations are written that relate the observed
quantities and their inherent random errors to the most probable
values for the x and y coordinates (the parameters) of the stations
involved. Referring to figure in the next slide, the following distance
equation can be written for any observation lij :

lij  + vl

69
DISTANCE OBSERVATION
EQUATION
In the equation I the previous slide, lij is the observed distance
of a line between stations I and J, v1 the residual in the
observation lij , Xi and Yi the most probable coordinate values
for station I, and Xj and Yj the most probable coordinate
values for station J . The distance equation is a nonlinear
function involving the unknown parameters Xi , Yi , Xj , and
Yj , which can be rewritten as

70
DISTANCE OBSERVATION
EQUATION
and dxi , dyi , dxj , and dyj the corrections to the approximate coordinate values
such that

The evaluation of partial derivatives is straightforward and will be illustrated with

∂F/∂xi . The equation can be rewritten as

Taking the derivative of the equation with respect to Xi yields

71
DISTANCE OBSERVATION
EQUATION
Simplifying the equation yields

Employing the same procedure, the remaining partial derivatives are

 If the simplified equation and the equation with partial derivatives remained
are substituted into linearized form of equation and the results substituted into
equation lij + vl , the following prototype linearized distance observation
equation obtained is

72
TRILATERATION ADJUSTMENT
Even though the geometric figures used in trilateration are many and
varied, they are equally adaptable to the observation equation method
in a parametric adjustment. Consider the example shown in the figure ,
where the distances are observed from three stations with known
coordinates to a common unknown station U . Since the unknown
station has two unknown coordinates and there are three observations,
this yields one redundant observation. That is, the coordinates of
station U could be determined using any two of the three observations.
But all three observations can be used simultaneously and adjusted by
the method of least squares to determine the most probable value for
the coordinates of the station.

73
TRILATERATION ADJUSTMENT
The observation equations are developed by substituting into prototype
equation. For example, the equation for distance AU is formed by interchanging
subscript I with A and subscript J with U in the prototype linearized distance
observation equation . In a similar fashion, an equation can be created for each
observed distance using the following subscript substitutions:

Trilateration
example 74
TRILATERATION ADJUSTMENT
Using the appropriate substitutions, the following three linearized observation
equations result:

75
TRILATERATION ADJUSTMENT
The Jacobian matrix can systematically be formed using the following steps:
Step 1: Head each column with an unknown value.
Step 2: Create a row for every observation.
Step 3: Substitute in the appropriate coefficient corresponding to the column
into each row.
If this procedure is followed for this problem, the Jacobian matrix is

76
Sample Problems
and Computations
TRILATERATION ADJUSTMENT
To clarify the computational procedure, a numerical example for Figure 14.2 is
presented. Suppose that the observed distances l AU , lBU , and lCU are 6049.00,
4736.83, and 5446.49 ft, respectively, and the control stations have coordinates
in units of feet of
xa = 865.40 xb = 2432.55 xc = 2865.22
ya = 4527.15 yb = 2047.25 yc = 27.15

78
TRILATERATION ADJUSTMENT
(SOLUTION)
1. Calculate approximate coordinates for station U .
A. Calculate azimuth AB from the coordinate values of stations A and B.

B. Calculate the distance between stations A and B from their coordinate values.

79
TRILATERATION ADJUSTMENT
(SOLUTION)
C. Calculate azimuth AU0 using the law of cosines in triangle AUB:

D. Calculate the coordinates for station U .

80
TRILATERATION ADJUSTMENT
(SOLUTION)
2. Calculate AU0, BU0, and CU0. For this first iteration, AU0 and BU0 are
exactly equal to their respective observed distances since Xu 0 and Yu0 were
calculated using these quantities. Thus,

3. Formulate the matrices. The elements of the Jacobian matrix in the equation
are

81
TRILATERATION ADJUSTMENT
(SOLUTION)
2. Calculate AU0, BU0, and CU0. For this first iteration, AU0 and BU0 are
exactly equal to their respective observed distances since Xu 0 and Yu0 were
calculated using these quantities. Thus,

3. Formulate the matrices. The elements of the Jacobian matrix in the equation
are

82
TRILATERATION ADJUSTMENT
(SOLUTION)
The elements of the K matrix in the equation are
K1 = 6049.00 − 6049.00 = 0.000
K2 = 4736.83 − 4736.83 = 0.000
K3 = 5446.49 − 5446.298 = 0.192

4. The matrix solution using the equation X = (J TJ ) JTK is

83
TRILATERATION ADJUSTMENT
(SOLUTION)
The revised coordinates of U are

Now perform the second iteration.

Step 1: Calculate AU0, BU0, and CU0.

84
TRILATERATION ADJUSTMENT
(SOLUTION)
Step 2: Formulate the matrices. With these minor changes in the lengths, the J
matrix (to three places) does not change, and thus (J TJ )–1 does not change
either. However, the K matrix does change, as shown by the following
computations.

Step 3: Matrix solution

The revised coordinates of U are

85
Traverse
Adjustment
using Least-
Squares
Method
Introduction
Traverse Adjustment
using Least Square
▸ The method of least squares is based on the theory
of probability, which models the occurrence of
random errors. This results in adjusted values
having the highest probability. Thus the least-
squares method provides the best and most
rigorous traverse adjustment, but until recently the
method has not been widely used because of the
lengthy computations required. The availability of
computers has now made these calculations
routine, and consequently the least-squares method
has gained popularity. 88
▸ In applying the least-squares method to traverses,
angle and distance observations are adjusted
simultaneously. Thus no preliminary angle
adjustment is made, as is done when using the
compass rule. The least-squares method is valid
for any type of traverse, and has the advantage
that observations of varying precisions can be
weighted appropriately in the computations.
89
OBSERVATION EQUATION
When adjusting a traverse using parametric equations, an
observation equation is written for each distance, direction, or angle.
The necessary linearized observation equations developed
previously are recalled in the following equations.

Distance observation equation:

90
OBSERVATION EQUATION
Angle observation equation

Azimuth observation equation:

91
REDUNDANT EQUATIONS
As noted earlier, one observation equation can be written for each observed
angle, distance, or direction in a closed traverse. Thus, if there are n sides in the
traverse, there are n distances and n + 1 angles, assuming that one angle exists
for orientation of the traverse.

For example, each closed traverse in the figure in the next slide has four sides,
four distances, and five angles. Each traverse also has three points whose
positions are unknown, and each point introduces two unknown coordinates
into the solution. Thus, there are a maximum of 2(n – 1) unknowns for any
simple closed traverse. From the foregoing, no matter the number of sides, there
will always be a minimum of r = (n + n + 1) – 2(n – 1) = 3 redundant equations
for any simple closed traverse. That is, every simple closed traverse that is fixed
both positionally and rotationally in space has a minimum of three redundant
equations.
92
REDUNDANT EQUATIONS

(a) Polygon and (b) link traverses

.
93
MINIMUM AMOUNT OF
CONTROL
All adjustments require some form of control, and failure to
supply a sufficient amount will result in an indeterminate solution. A
traverse requires a minimum of one control station to fix it in
position and one line of known direction to fix it rotationally. When
a traverse has the minimum amount of control, it is said to be
minimally constrained. It is not possible to adjust a traverse without
this minimum. If minimal constraint is not available, necessary
control values can be assumed and the computational process carried
out in arbitrary space. This enables the observed data to be tested for
blunders and errors.

.
94
ADJUSTMENT OF NETWORKS
With the introduction of the EDM instrument, and particularly the total
station, the speed and reliability of making angle and distance observations
has increased greatly. This has led to observational systems that do not
conform to the basic systems of trilateration, triangulation, or traverse.
For example, it is common to collect more than the minimum
observations at a station during a horizontal control survey. This creates what
is called a complex network, referred to more commonly as a network. The
least squares solution of a network is similar to that of a traverse. That is,
observation equations are written for each observation using the prototype
equations given in Angle Observation Equation . Coordinate corrections are
found using the equation
X = J TWJ −1 J TWK = N −1J TWK
and a posteriori error analysis is carried out.

.
95
DISTANCE OBSERVATION
EQUATION
In the equation I the previous slide, lij is the observed distance
of a line between stations I and J, v1 the residual in the
observation lij , Xi and Yi the most probable coordinate values
for station I, and Xj and Yj the most probable coordinate
values for station J . The distance equation is a nonlinear
function involving the unknown parameters Xi , Yi , Xj , and
Yj , which can be rewritten as

96
DISTANCE OBSERVATION
EQUATION
and dxi , dyi , dxj , and dyj the corrections to the approximate coordinate values
such that

The evaluation of partial derivatives is straightforward and will be illustrated with

∂F/∂xi . The equation can be rewritten as

Taking the derivative of the equation with respect to Xi yields

97
DISTANCE OBSERVATION
EQUATION
Simplifying the equation yields

Employing the same procedure, the remaining partial derivatives are

 If the simplified equation and the equation with partial derivatives remained
are substituted into linearized form of equation and the results substituted into
equation lij + vl , the following prototype linearized distance observation
equation obtained is

98
Sample Problems
and Computations
NUMERICAL EXAMPLE
To illustrate a least squares traverse adjustment, the simple link
traverse shown in the figure will be used.

Simple link 100

NUMERICAL EXAMPLE
(SOLUTION)
1. Calculate initial approximations for the unknown station coordinates.

2. Formulate the X and K matrices. The traverse in this problem contains only one
unknown station with two unknown coordinates. The elements of the X matrix thus
consist of the dxu and dyu terms. They are the unknown corrections to be applied to the
initial approximations for the coordinates of station U . The values in the K matrix are
derived by subtracting computed quantities, based on initial coordinates, from their
respective observed quantities.
Note that since the first and third observations were used to compute initial
approximations for station U , their K-matrix values will be zero in the first iteration.

101
NUMERICAL EXAMPLE
(SOLUTION)
3. Calculate the Jacobian matrix. Since the observation equations are nonlinear, the
Jacobian matrix must be formed to obtain the solution. The J matrix is formed using
prototype equations (16.1) for distances and (16.2) for angles.
As explained, since the units of the K matrix that relate to the angles are in seconds, the
angle coefficients of the J matrix must be multiplied by ρ to obtain units of seconds. In
developing the J matrix using prototype Equations (16.1) and (16.2), subscript
substitutions were as shown in table. Substitutions of numerical values and computation
of the J matrix follow.

102
NUMERICAL EXAMPLE
(SOLUTION)
4. Formulate the W matrix. The fact that distance and angle observations have
differing observational units and are combined in an adjustment is resolved by
using relative weights that are based on observational variances in accordance
with equation. This weighting makes the observation equations consistent
dimensionally. If weights are not used in traverse adjustments (i.e., equal
weights are assumed), the least squares problem will generally either give
unreliable results, or result in a system of equations that has no solution. Since
weights influence the correction size that each observation will receive, it is
extremely important to use variances that correspond closely to the
observational errors. The error propagation procedures discussed in Chapter 7
aid in the determination of reasonable estimated errors for each observation.
Repeating Equation (10.6), the distance and angle weights for this problem are

103
NUMERICAL EXAMPLE
(SOLUTION)
Again the units of the weight matrix must match those of the J and K matrices.
That is, the angular weights must be in the same units of measure (seconds) as
the counterparts in the other two matrices. Based on the estimated errors in the
observations, the W matrix, which is diagonal, is

104
NUMERICAL EXAMPLE
(SOLUTION)
χ2 test was used as discussed to see if the a posteriori reference
variance differed significantly from its a priori value of 12. 1
The test revealed that there was no statistically significant
difference between the a posteriori value of 1.822 and its a
priori value of 12 at a 99% confidence level, and thus the a
priori value was used for the reference variance when
computing the standard deviations of the coordinates. By
applying the equation , the estimated errors
in the adjusted coordinates are

105