Faculty of Natural Resources and Spatial Sciences

Geomatics Theory of Errors (GTE710S)

Michael Mutale
Office: 105 Poly Heights
Telephone: 061 207 2677
Date: Thursday, 06 December 2018
 Geomatics Theory of Errors (GTE710S)

Lecture Outline
• Propagation of Errors

• Error Propagation Derivation

• General Form of Error Propagation

• Error Propagation for Inter-dependent Observations

• Error Propagation for Independent Observations

2
 Geomatics Theory of Errors (GTE710S)

Propagation of Errors
• Errors in direct observations have been looked at thus far
• Often times other quantities can be determined based on
other direct measurements (e.g. angles and distances to
compute coordinates)
• By error propagation, the impact of observations on
computed quantities can be investigated
• Error Propagation is the evaluation of errors in computed
quantities
• It involves identification of errors in the final quantities
which result from errors in the original observations

3
 Geomatics Theory of Errors (GTE710S)

Propagation of Errors – Example

Area of 100x40 rectangle is 4000
A one unit error gives an area of 101x41=4141
Error in area is 141, which is approximately 100x1 + 40x1

4
 Geomatics Theory of Errors (GTE710S)

Error Propagation - General Form

For a given function y of inter-dependent measurements x1, x2, x3
… xm, such that y = ax1 + bx2 + x3 … + cxn
The Generalized Law of the Propagation of Variances (GLOPOV) is
given as:
2 2 2 2
2
δy 2
δy 2
δy 2
δy
σy = σx1 + σx2 + σx3 + ⋯ + σ2xn
δx1 δx2 δx3 δxn

δy δy δy δy
+2 σ + ⋯ All Combinations … + 2 σ
δx1 δx2 x1x2 δxn−1 δxn xn−1xn
Where:
σ2x1 , σ2x2 , σ2x3 , etc. are the variances of each variable

σx1 x2 , σx2 x3 , etc. are the covariance factors (relationship between variables)
5
 Geomatics Theory of Errors (GTE710S)

Error Propagation – Example 1

For the following equation and the given statistical data calculate
the following :
Y = 3x1 + 2x2

σ2x1 = 0.02, σ2x2 = 0.04, σx1 x2 = −0.0003

(a) Variance of y:  y2
(b) Standard deviation of y:  y

6
 Geomatics Theory of Errors (GTE710S)

Error Propagation – Example 1

Solution: Equation: Y = 3x1 + 2x2
δy δy
= 3, =2 σ2x1 = 0.02, σ2x2 = 0.04, σx1x2 = −0.0003
δx1 δx2

2 2 2 2
2
δy 2
δy 2
δy 2
δy
σy = σx1 + 𝜎x2 + σx3 + ⋯ + σ2xn
δx1 δx2 δx3 δxn

δy δy δy δy
+2 σ + ⋯ All Combinations … + 2 σ
δx1 δx2 x1x2 δxn−1 δxn xn−1xn

Variance = σ2y = 32 0.02 + 22 0.04 + 2 3 2 −0.0003 = 0.3364

Standard Deviation = σ2y = ± 0.3364 = ±0.58

7
 Geomatics Theory of Errors (GTE710S)

Error Propagation – Example 2

The following equation and statistical data are given:
Y = 2x1 - 5x2 - 7x3 and

σ2x1 = 0.01, σ2x2 = 0.03, σ2x3 = 0.02, σx1 x3 = 0.002, σx1 x2 = σx2 x3 = 0

Calculate the following:

a) Variance σy2 and
b) b) Standard error σy

8
 Geomatics Theory of Errors (GTE710S)

Error Propagation – Example 2

Solution: Equation: Y = 2x1 - 5x2 - 7x3
σ2x1 = 0.01, σ2x2 = 0.03, σ2x3 = 0.02, σx1x3 = 0.002, σx1x2 = σx2x3 = 0
δy δy δy
= 2, = −5, = −7
δx1 δx2 δx3
2 2 2 2
2
δy 2
δy 2
δy 2
δy
σy = σx1 + 𝜎x2 + σx3 + ⋯ + σ2xn
δx1 δx2 δx3 δxn
δy δy δy δy
+2 σ + ⋯ All Combinations … + 2 σ
δx1 δx3 x1x3 δxn−1 δxn xn−1xn
Note: With σx1x2 and σx2x3 = 0, it is unnecessary to calculate all combinations (as prescribed by
the formula) as these terms would also be = 0

Variance = σ2y = 22 0.01 + (−5)2 0.03 + (−7)2 0.02 + 2 2 −7 0.002 = 1.714

Standard Deviation = σ2y = ± 1.714 = ±1.31

9
 Geomatics Theory of Errors (GTE710S)

Error Propagation - Independent Variables

Variables are said to be independent when the variables in an
equation or calculation have no covariance or relationship
between them

Given y a function of independent measurements x1, x2, x3 … xn,

such that y = ax1 + bx2 + x3 …+cxn

If the n unknown variables are independent, then their

covariance elements are 0 by definition and hence the
Simplified Law of Propagation of Variances (SLOPOV) is given as:
2 2 2 2
δy δy δy δy
σ2y = σ2x1 + 𝜎x22 + σ2x3 +⋯+ σ2xn
δx1 δx2 δx3 δxn
10
 Geomatics Theory of Errors (GTE710S)

Error Propagation - Sum of Observations

If we assume a function such that:

A = B1 + B2 + B3 + B3 + … + Bn
Then, applying the Simplified Law of Propagation of Variances
(SLOPOV), the variance in the computed value A will be:
2 2 2 2
2 δy 2 δy δy δy
σA = σB1 + 𝜎B22 + 2
σB3 + ⋯ + σ2Bn
δB1 δB2 δB3 δBn

δy δy δy δy
And since = = = = 1, we get:
δB1 δB2 δB3 δBn

σ2A = σ2B1 + 𝜎B22 + σ2B3 + ⋯ + σ2Bn

11
 Geomatics Theory of Errors (GTE710S)

Error Propagation - Mean of Observations

If we assume a function such that:
B1 + B2 + B3 + B3 + … +Bn
A =
n
Then, applying SLOPOV, the variance in computed value A will be:
2 2 2 2
2 1 δy 2 1 δy 2 1 δy 2 1 δy
σA = σB1 + 𝜎B2 + σB3 + ⋯ + σ2Bn
𝑛 δB1 𝑛 δB2 𝑛 δB3 𝑛 δBn
δy δy δy δy
And since = = = = 1, we get:
δB1 δB2 δB3 δBn
2 2 2 2
1 1 1 1
σ2A = σ2B1 + 𝜎B22 + σ2B3 +⋯+ σ2Bn
𝑛 𝑛 𝑛 𝑛
1 2
σ𝐴2 = 2 σB1 + 𝜎B22 + σ2B3 + ⋯ + σ2Bn
𝑛
And if σ2B1 = σ2B2 = σ2B3 = σ2Bn = σ2 , then
Which is the
n 2 1 2 1 2 σ standard deviation
σ2A = 2 σ = σ Implying that σA = σ =
n n n n of the mean‼! 12
 Geomatics Theory of Errors (GTE710S)

Error Propagation – Example 3

Given that d1 and d2 are measured distances and given that σd1 =
±0.30m and σd2 = ±0.25m, calculate the standard deviation of the
sum & difference of these measurements, and comment on the
results
Solution: Standard deviation of Sum:
Equation: D = d1 + d2
2 2
δD 2 δD
σ2sum = σd1 + σ2d2
δd1 δd2
δD δD
= 1, =1
δd1 δd2

σ2sum = 1 2
0.30 2
+ 1 2
0.25 2
= 0.1525m2

σsum = 0.1525 = ±0.391m

13
 Geomatics Theory of Errors (GTE710S)

Error Propagation – Example 3

Solution: Standard deviation of Difference:
Equation: D = d1 - d2
2 2
δD δD
σ2diff = 2
σd1 + σ2d2
δd1 δd2

δD δD
= 1, =1
δd1 δd2

σ2diff = 1 2 0.30 2 + 1 2 0.25 2 = 0.1525m2

σdiff = 0.1525 = ±0.391m

Comment: standard deviation of sum and difference of independent variables is the same
14
 Geomatics Theory of Errors (GTE710S)

Error Propagation – Example 4

A theodolite is set up at point A. In order to determine the height of a
remote point (B), the slope distance as well as vertical angle from point
A is measured Given that:
 HI : Height of Instrument
 HS : Height of Target
 HD : Horizontal distance
 SD : Slope distance
 HB : Height of Point B above datum
 HA : Height of Point A above datum
 Alpha : Vertical angle from Point A to Point B
 The following data is available:
 σAlpha = ± 30”, Alpha = 20°, σSD = ± 0.015m,
σHI = σHS = σHA = ±0.01m
 Calculate the standard deviation in
calculated height at point B (σHB), if the
slope distance of 100m was measured 15
 Geomatics Theory of Errors (GTE710S)

Error Propagation – Example 4

Solution : Formulate the equation to calculate the height of point B
(HB), using the relative independent variables, i.e.:
HB = HA + HI – HS + SD[Sin(Alpha)]
Given: σHA = ±0.01m, σHI = ±0.01m, σHS = ±0.01m, σSD = ± 0.015m,
σAlpha = ± 30”, Alpha = 20° and SD = 100m
2 2 2 2 2
δHB δHB δHB δHB δHB
σ2HB = σ2HA + σ2HI + σ2HS + σ2SD + σ2Alpha
δHA δHI δHS δSD δAlpha

δHB δHB δHB δHB δHB

= 1, = 1, = −1, = Sin Alpha , = SD[Cos Alpha ]
δHA δHI δHS δSD δAlpha

σ2HB = 1 2
0.01 2
+ 1 2
0.01 2
+ −1 2
0.01 2
+ 0.015 2
0.01 2

2
2 2 2
30"
+ Sin(20) 0.015 + 100Cos(20) = 0.000513m2
206265

σHB = ± 0.000513 = ±0.023m

16
 Geomatics Theory of Errors (GTE710S)

Error Propagation – Example 5

An angle between points B and C is measured from Point A using a 01”
theodolite and the following parameters are given:

 c = ± 03”
 p = ± 05”

Given that:
σc : instrument centering standard deviation
σp : target pointing standard deviation

For two and four arcs of observations respectively, calculate:

(a) Standard deviation in directions d1 and d2
(b) Standard deviation in Angle alpha and comment on results 17
 Geomatics Theory of Errors (GTE710S)

Error Propagation – Example 5

Solution :
 
2 2
 c2   p2
  
2
d1
2
d2
c p
 alpha
2

2n n

Where n = The number of arcs of observations

For two (n=2) arcs of observations:

9  25
Direction Variance =  d2   d2   8.5"
1 2
4
Direction Standard Deviation =  d1   d2   8.5"  2.9"
9  25
Angle Variance =  2
alpha   17"
2
Angle Standard Deviation =  alpha   17"  4.1" 18
 Geomatics Theory of Errors (GTE710S)

Error Propagation – Example 5

For four (n=4) arcs of observations:

9  25
Direction Variance =    2
d1
2
d2   4.3"
8

Direction Standard Deviation =  d1   d2   4.3  2.1"

9  25
Angle Variance =  2
alpha   8.5"
4
Angle Standard Deviation =  alpha   8.5"  2.9"

Comment: Better results are obtained by taking more arcs of

observations, reducing both σd and σalpha

19
 Geomatics Theory of Errors (GTE710S)

Error Propagation – Example 6

What is the standard deviation in the horizontal distance (HD) between
A and B if the slope distance (SD) is 1345.67m and the vertical angle α is
4° 13ʹ 56ʺ ± 16.8ʺ?
Particulars for the instrument setup:
Assume an instrument with setup error of ±0.005m, target setup error
of ±0.01m and EDM precision of 5mm + 5ppm

20
 Geomatics Theory of Errors (GTE710S)

Error Propagation – Example 6

Solution: First compute the error in Slope Distance:
2
 HD x ppm 
 SD =  Inst
2
+  2 +  2 +
Target EDM 
 1000000 
2
 1345.67x5 
 SD = 0.0052 + 0.012 + 0.0052 +   = ±0.014m
 1000000 
Horizontal Distance is given by formular:
HD = SCos
HD = 1345.67Cos  4° 13' 56" = 1342.00m

Then compute the error in Horizontal Distance (Non-linear)

2 2
 δHD   δHD 
H =  σS  +  σα 
 δS   δα  21
 Geomatics Theory of Errors (GTE710S)

Error Propagation – Example 6

Error in the Horizontal Distance:
δHD
σS = σSCosα = 0.014Cos  4° 13' 56" = 0.014m
δS

δHD σ   16.8" 
σ α = -S  α  Sinα = -1345.67   Sin  4° 13' 56" = -0.008m
δα  ρ   206265 

2 2
 δHD   δHD 
 HD =  σS  +  σα 
 δS   δα 

 HD = 0.0142 + 0.0082 =  0.016m

So the distance precision is 1342.00  0.016m

22
 Geomatics Theory of Errors (GTE710S)

Derivatives of Trigonometric Functions

d
(cos x)   sin x
dx

d
(sin x)  cos x
dx

d
(tan x)  sec 2 x
dx

1
[ Note : sec x  ]
cos x
23
 Geomatics Theory of Errors (GTE710S)

Derivative Rules
Derivatives of Sums and Scalar Multiples:
Sums: The derivative of a sum of two functions is equal to the sum of
d df dg
the individual derivatives  f  g  
dx dx dx
Scalars: the derivative of a scalar multiple of a function is equal to the
derivative of the function multiplied by the scalar multiple
d df
 af   a
dx dx dy dv du
Product Rule of Differentiation: u v
dx dx dx
The product rule states that for two functions, u and v If y = uv, then:

dy dy du
Chain Rule of Differentiation:  x
dx du dx

24
 Geomatics Theory of Errors (GTE710S)

Product Rule of Differentiation

Given that y = 2x(x2 − 1)5
Let u = 2x and v = (x2 − 1)5
Then by the product rule of differentiation:
dy dv du
=u +v
dx dx dx

=  2x 10x  x  1 = 20x  x  1 and v = 2  x  1

dv 2 4 2 2 4 du 2 5
u
dx dx
= 20 x 2 ( x 2  1) 4  2  x 2  1
dy 5
Therefore
dx
Expanding and factorising we get:
= 20 x ( x  1)  2  x  1  x 2  1
dy 2 2 4 2 4

dx

=  20x +2x  2  x -1 = 2 11x  1 x  1

dy 2 2 2 4 2 2 4

dx 25
 Geomatics Theory of Errors (GTE710S)

Chain Rule of Differentiation

Given that y = (3x − 2)4

Then let u = 3x − 2 giving y = u4

dy dy du dy 3 du
= x = 4u and =3
dx du dx du dx

dy
Therefore = 4u 3 x 3 = 12u 3
dx

dy
But u = 3x  2  = 4  3 x  2  x 3 = 12  3 x  2 
3 3

dx

26
 Geomatics Theory of Errors (GTE710S)

Next Topics – Unit 6

Introduction to Matrices

Types of Matrices

Matrix Methodology

Simultaneous Equation Solving

27
 Geomatics Theory of Errors (GTE710S)

The End!!!!

QUESTIONS/DISCUSSION

28