Understanding

Errors
In Measurements

James A. Coan, Sr., P.L.S.

 A Friendly Pre-Test
n 1) Precision and Accuracy are the same thing T F
n 2) All errors can be avoided T F
n 3) All errors can be eliminated T F
n 4) How many significant figures will an answer
n have if a measured distance is multiplied by Pi?
n 5) How many significant figures is 0.032 ?
n 6) Individual random errors are added together
n to find the total random error of a measurement T F
n 7) If a manufacturer states that your EDM can
n measure to + (3mm+2ppm) what confidence
n level are they referring to ?
n 8) An instrument centering error refers to
n how well a surveyor can set up over a point T F
n 9) In Surveying, all angle measurements
n are about the same precision T F
10) Weighting measurements allows the
n surveyor to put the error where it belongs T F
Understanding Errors In
Measurements
(WHAT WE WILL COVER)
Errors and Mistakes
Precision and Accuracy
Sources of Errors in Surveying
Significant Figures
Random Error Propagation
Statistics for Land Surveyors
Random Errors in Angles and Distance
Practical Weights of Observations
Errors And Mistakes
 Errors And Mistakes
Errors
The difference between the true and
measured value of a measurement
The error equals the measured
distance minus the true distance
E = M-T
They are unavoidable
 Errors And Mistakes

Mistakes (Blunder)
A mistake is a blunder caused by
carelessness

This type of problem has nothing to

do with errors
Systematic And

Random Errors
 Systematic And Random
Errors
Systematic Errors
Their magnitude and direction can be
determined
They are predictable
Systematic errors can be eliminated
This separates them from random errors
 Systematic And Random
Errors
Random Errors
Are unavoidable
Can be minimized but never eliminated
Have a tendency to cancel but never
completely do so.
Can be dealt with by the science of
propagation
Precision And Accuracy
 Precision And Accuracy
Precision
The agreement of readings of the same
quantity
The better the precision the smaller the
random error
Good precision exposes random error

Crude precision hides random errors

Precision And Accuracy

Accuracy
The agreement of readings with the
true value

Accuracy is to systematic errors as

precision is to random errors
 Precision And Accuracy

Precision

Accuracy
Sources Of Errors In
Surveying
Sources Of Errors In Surveying

Natural Errors
Caused by nature, wind, temperature, earth
curvature, etc.
If the error is subject to known physical laws,
it is systematic.
If the error does not follow known physical
laws it is probably random
Sources Of Errors In Surveying

Instrumental Errors
Caused by manufacture, wear and tear, or
maladjustment of instruments.

This type of error can also be thought of as a

blunder

Most instrumental errors are random in nature

Sources Of Errors In Surveying
Personal Errors
Caused by the inability of a person perceive anything
exactly

Can be controlled by good training, good motivation,

and good technique

Personal errors are random and each person has

their own
Significant Figures
 Significant Figures

Significant figures is a part of mathematics that

deals with the relevancy of digits in a number,
and are mainly used in the surveying, scientific,
and engineering community.
 Significant Figures

Significant figures can be separated into

two areas:

1) Significant figures in measurements

2) Significant figures in computations

 Significant Figures
Measurements
Relates to the way measurements are
made and recorded

No measurement is exact

When measuring, record only the digits that

have meaning
 Significant Figures
Computations

Concerns itself with round-off error

The surveyor must use established rules of

significant figures
 Significant Figures
Rules

The following rules apply to both recording

data and interpreting recorded data
 Significant Figures

Zeroes used merely to indicate the position of

a decimal point are not significant

Example:

0.056 has two significant figures

 Significant Figures
Rules
Zeroes recorded at the end of measurement
are significant

Example:

1.30 has three significant figures

 Significant Figures
Example:

Zeroes between non zero digits are

significant

Example:

1.04 has three significant figures

 Significant Figures

Rules

Numbers ending with one or more

zeroes to the left of the decimal
should have a special indication
 Significant Figures

Example:

175,000 has three significant figures

375,000 can have six significant figures

 Significant Figures

Rules
When adding or subtracting measured
distances the number with the fewest
decimal places will dictate the number of
significant figures
 Significant Figures
Rules
Example: Adding or Subtracting
15.495
10.21
12.2 control, fewest number of
decimals
37.905
The proper answer is 37.9, having three
significant figures
 Significant Figures
Rules: Multiplication or Division

The product or quotient is

determined by the fewest number
of significant figures in the values
used, if both are measured values
 Significant Figures

Example:

5.29 x 0.052 = 0.28 two significant figures

 Significant Figures

Rules
Conversion factors do not
determine significant figures

Example:

1534.5 in / 12in per ft = 127.88 ft.

 Significant Figures
Rules
With large or infinite number conversion
factors; use one extra digit

Example:
174.35 x 3.14 = 547.46 incorrect
174.35 X 3.14159 = 547.74 correct
 Significant Figures
Rules
With intermediate calculations use one
extra digit and round off your answer

Intermediate
Example:
Calculation
43.56 x 23.43 1020.6
= = 48.55
21.02 21.02
Random Error

Propagation
 Random Error Propagation

All measurements have random errors

Random errors have a tendency to cancel

but never completely do so
 Random Error Propagation

How random errors accumulate, cancel,

decrease or behave through the process
of computing the final value is termed
“Propagation of Random Errors”
 Random Error Propagation
Errors in a sum
Used when all errors are different

It is the square root of the sum of the squares

of the errors

E = + e12 + e22 + e32 +…+ en2

 Random Error Propagation
Errors in a Series

Used when the same error happens several

times

Derived from the formula of errors in a sum

E=+e n
 Random Error Propagation
Errors in a Product
Used in determining the random error of area
calculations
Is also the square root of the sum of the
squares of the errors

E = + (Lew)2 + (WeL)2
Statistics For

Surveyors
 Statistics For Surveyors
Definitions

Direct Measurements

A measurement made directly

between two or more points
 Statistics For Surveyors
Definitions

Indirect Measurements

A computed measurement
between points
 Statistics For Surveyors
Definitions

Sample Size

The number of observations or measurements in

a sample

n = Sample Size
 Statistics For Surveyors

Mean
The sum of observations of a sample divided
by the sample size

Σ xi
X=
n

Xi = The value of the

X =Mean
sample
Sample Size = 25

Σn = 779.4
Mean = 779.4 / 25 = 31.2
 Statistics For Surveyors
Definitions

Median

The middle value of the sample when the data

is arranged in ascending or descending order
 Statistics For Surveyors

Definitions

Mode

The value which occurs most frequently in a

sample
Mode

Median

Mode
 Statistics For Surveyors
Definitions
Residual

The difference between an individual value

in a sample and the mean of the sample

Vi = Xi - X
Vi = Residual
 Statistics For Surveyors
Definitions
Standard Deviation

68% probability of an occurrence

Σ vi2
σ=+
n-1

σ = Standard Deviation
 Statistics For Surveyors
Levels Of Certainty
NAME OF SYMBOL VALUE %
ERROR CERTAINTY
PROBABLE E50 0.6745σ 50
STANDARD
DEVIATION σ 1σ 68.3
90%
ERROR E90 1.6447σ 90

TWO SIGMA E95 2σ 95

99%
ERROR E99 2.5σ 99
THREE
SIGMA E99.7 3σ 99.7
 Statistics For Surveyors
Definitions
Standard Error of the Mean

The interval of uncertainty around the true

value.

σ
Σx = +
n
 Statistics For Surveyors
Definitions

Standard Error of the Mean

σ 0.65
σx =± =± = ±0.13
n 25
Pre-analysis

Formulas
 Pre-analysis Formulas
Reading Error Directional Theodolite

σr 2
σα = ±r
n
σαr = The total reading error
σr = The individual reading error
n = The number of angles turned
 Pre-analysis Formulas

Reading Error Repetition Theodolite

σr 2
σα = ±r
n
σαr = The total reading error
σr = The individual reading error
n = The number of angles turned
 Reading Error Example
Directional Theodolite
σr = individual reading error ( 0.65)

n = number of angles turned (4)

σr 2 0.65 2
σα = ±
r
σ αr =± = 0.46
n 4
 Reading Error Example
Repetition Theodolite

σr = individual reading error ( 0.65)

n = number of angles turned (4)

σ r 2 σ = ± 0.65 2 = 0.23
σα =± α r
4
r
n
Reading Error Example Topcon IS
 Pre-analysis Formulas
Pointing Error
Epb = Error in pointing to the backsight
Epf = Error in pointing to the foresight
dc = Estimate of how closely the observer
can center the cross-hairs on the target
D = The distance to the backsight or the
foresight
 Pre-analysis Formulas
Pointing Error

dc
= Angle in radians
D

dc
(206,265) = Angle in Seconds
D
 Pre-analysis Formulas
Pointing Error

db
E pb =± (206,265) = Pointing Error, Backsite
Db

df
E pf = ± (206,265) = Pointing Error, Foresite
D f
 Pointing Error Example
Pointing Error

db
E pb = ± (206265) db = 0.01’ ; Db = 350.25
Db

0.01
E pb = ± (206,265) = 589
. "
350.25
 Pointing Error Example
Pointing Error

df
E pf = ± (206,265) df = 0.01’ ; Df = 425.36
Df

0.01
E pf =± (206,265) = 4.85"
425.36
 Pre-analysis Formulas
Pointing Error

Error in one angle turned

E p = ± E pb + E pf
2 2
 Pointing Error Example
Pointing Error

Error in one angle turned

Ep = ± E 2
pb +E 2
pf

E p = ± 589
. + 4.85 = 7.63"
2 2
 Pointing Error Example
Pointing Error

Total error in multiple sets turned

Ep 2
σ αp = ±
n
Total Pointing Error

7.63 2
σα p
=± = 5.40"
4
 Pre-analysis Formulas
Instrument centering error
d c D3
σ αi =± (206,265) = Angle In Seconds
D f Db 2

dc= Estimate of how well you can set up over

a point
Db= Distance to backsight
Df= Distance to foresight
D3= Distance from the backsight to the foresight
 Instrument Centering Error
Example: d c D3
σα i
=± (206,265)
D f Db 2

dc = 0.005’ Df =425.36’
Db = 350.25’ D3 = 372.11’

(0.005' )(372.11' )
σα i
=± (206,265) = 182
. "
(425.36' )(350.25' ) 2
 Pre-analysis Formulas
Target Centering Error
db+ df = How well you can set up a target
over a point
Db= Distance to the backsight

Df= Distance to the foresight

Etb & Etf = Angular error to the backsight and
foresight in seconds
σαt= Total angular error
 Pre-analysis Formulas
Target Centering Error

db
Etb = ± (206,265)
Db

df
Etf = ± (206,265)
Df
 Pre-analysis Formulas

Target Centering Error

σ αt = ± Etb + Etf 2 2
Target Centering Error Example

db db = 0.005'
E tb = (206,265)
Db Db = 350.25

0.005'
E tb = (206,265) = 2.94"
350.25'
 Target Centering Error Example

df d f = 0.005'
Et f = (206,265)
Df D f = 425.36'

0.005'
Et f = (206,265) = 2.42"
425.36'
 Target Centering Error Example

E tb = 2.94"
σα = ± E 2
tb +E 2
tf
t E t f = 2.42"

σαt = + 2.942 + 2.422 = 3.81”

 Pre-analysis Formulas
Bubble Centering Error

( f d µTanγ b ) + ( f d µTanγ f )
2 2

σ αb = ±
n
γ= Vertical angle to backsight and foresight (not
zenith angles)

µ= Bubble sensitivity of the spirit level in seconds

 Pre-analysis Formula
Bubble Centering Error

( f d µTanγ b ) + ( f d µTanγ f )
2 2

σ αb = ±
n
fd= Estimate of how closely the instrument can be
leveled during a set of angles in
fractions of divisions of the spirit level

n = Number of angles turned

Bubble Centering Error Example

( f d µTanγ b ) + ( f d µTanγ f )
2 2

σ αb = ±
n

fd = 0.5 µ = 10” γ b= 25°30’45”

γ f = 32°15’09” n=4
 Bubble Centering Error Example

(0.5 • 10 • Tan25 30'45") + (0.5 • 10 • Tan32 15'09")

o 2 o 2
σ αb =± = 0.99"
4
 Total Random Error
σα = + σr2 + σp2 + σi2 + σt2 + σb2

σr = Reading Error
σp = Pointing Error
σi = Instrument Centering Error
σt = Target Centering Error
σb = Bubble Centering Error
 Total Random Error
σα = + 0.462 + 5.402 + 1.822 + 3.812 + 0.992 = 6.93”
Reading Error directional = 0.46”
Pointing Error = 5.40”
Instrument Centering Error = 1.82”
Target Centering Error = 3.81”
Bubble Centering Error = 0.99”
Total Angular Error = 6.93”
 Pre-analysis Formulas
EDM Error

σEDM = + Em2 + Et2 + Ei2

Em= (Standard manufacturer error) (distance)

Et= Target centering error in feet

Ei= Instrument centering error in feet

 Pre-analysis Formulas
EDM Error

σEDM = + 0.012 + 0.0052 + 0.0052 = 0.01’

Em = 0.01 + [ 3mm + (3ppm)(425.36’)]

Et = 0.005’ Target Centering
Ei = 0.005’ Instrument Centering
Pre-analysis Formulas
 In Summary

Random errors are part of every measurement

taken by a surveyor.
In order to control random errors a surveyor
must first understand them.
 In Summary

Once the surveyor understands random errors

they can place them where they belong.

This is critical because some measurements are

better than others.

 Weights Of
Observations
 General

Some measurements are better than others. This

is common knowledge in the surveying
profession.
 General

Measurements made in good conditions,

using good equipment, with proper survey
procedures will produce good results.
 General

Measurements made in adverse conditions,

using poor equipment, and bad surveying
procedures will produce bad results.
 General

As a rule, surveyors use good equipment, with

good procedures, but can work in poor conditions
 General

Because of this, the quality of measurements

can vary from setup to setup in the same job.
 General

Weighting measurements allows the surveyor

to distribute errors of the measurements
where the error should go.
 General

If one angle in a survey is turned under good

conditions and another angle, in the same survey,
is turned under bad conditions, weighting
measurements allows for putting more of the
error in the bad angle than in the good angle.
 General

According to statistical theory, the weight of a

measurement is inversely proportional to the
variance.
 General

The higher the precision of the measurement,

the smaller the variance.
 General

The smaller the variance, the larger the weight.

OR

The better the measurement, the larger the

weight.
 The Variance

What is the variance?

Simply put, the variance is the square of the

standard deviation.
 The Variance

The standard deviation of a set of measurements

can be determined by statistical analysis.
 The Variance

This can get complicated in a hurry, and

many practicing surveyors tend not to use
weights
 The Variance

If weights are not applied to the surveyors

measurements, all the measurements are
treated the same. This is false !
 Weights Of Observations

Weighting measurements does not need to be

this complicated

Surveyor can assign weights to their

measurements
 Weights Of Observations

The worse the measurement, the smaller the

weight.

The more precise the measurement, the

larger the weight.
 Example One

A distance between two points were measured

four times.

The first distance was found to be 396.57 feet

and was measured with a cloth tape
 Example One

The second and third distances were found to

be 396.61 and 396.62 feet, and were
measured with a steel calibrated tape
 Example One

The fourth distance was found to be 396.64

feet and was measured with an EDM.
 Example One
The party chief assigned a weight of one (1) to
the first measurement, a weight of two (2) the
second and third measurement, and a weight of
four (4) to the final measurement.

These values were not based on statistical

theory, but based on field judgment.

Low precision, large variance, low weight

High precision, small variance, large weight

 Example One
The formula to calculate the weighted mean of
the distance is.
ΣWM
= MW
ΣW

ΣWM = The sum of the measurements times

their weights
ΣW = The sum of the weights
MW = The weighted mean distance
 Example One
Using the weighted mean formula in our example
we have.
396.57(1) + 396.61(2) + 396.62(2) + 396.64(4)
= 396.62
1+ 2 + 2 + 4

While the above example shows how to weight

measurements, it is not very practical as most
distance measurements are made with EDM’s,
and the need to weight measurements are rare.
 Example Two
A five sided traverse was run with the following
results
Station Angle Weight
A 76°46’35” W=1
B 87°15’20” W=2
C 122°10’45” W=2
D 165°58’25” W=3
E 87°48’50” W=4
Total 359°59’55” 12
Error 00°00’05”
 Example Two

Historically, one second would be put into each

angle and the error would be eliminated.
 Example Two

If this is done, than all of the angles are

treated the same, that is, all of the angles
are given the same weight.
 Example Two

In this case it is the party chief that

determines the weights of the angles from
the conditions in the field.
 Example Two
The party chief assigns the weights of the angles
as follows.
Angle “A” is the worst angle
Angles “B” and “C” are better than “a” but still
not real good.
Angle “D” was better than the others but not the
best.
Angle “E” was the best angle in the traverse
 Example Two
The weight of angle “A” = 1,
The weights of angles“B” and “C” = 2
The weight of angle “D” = 3
The weight of angle “E” = 4

.
 Example Two

Angle adjustments are made inversely

proportional to their weights. The larger the

weight of the angle, the smaller the

adjustment
 Example Two
The table illustrates how weights are used to
distribute errors
Station Measured Weight Correction Numerical Rounded Adjusted
Correction Correction Angles
A 76°46’35” 1 12X 1.94” 2” 76°46’37”

B 87°15’20” 2 6X 0.97” 1” 87°15’21”

C 122°10’45” 2 6X 0.97” 1” 122°10’46”

D 165°58’25” 3 4X 0.65” 1” 165°58’26”

E 87°48’50” 4 3X 0.47” 0” 87°48’50”

Sum 539°59’55” 12 31X 5.00” 5” 540°00’00”

31X=5” X=0.16”
 Example Two
To find the correction factor , divide the individual
weights into the sum of the weights.

To find the value of “x”, place the sum of the

correction factors equal to the error and solve for
“x”
To find the numerical correction, multiply the
individual correction factor times the value of “x”
 Example Two
As seen in the chart, the worst angle received the
largest correction (2”), and the best angle had no
correction at all.

This is more realistic with what the party chief

observed in the field when the angles were turned
 How Weights Can Be
Determined
A surveyor can make their own system. One
example will be to divide the angles into four
categories, as follows.
 How Weights Can Be
Determined
Category 1 might be one where the setup is on

soft ground and the surveyor is having a hard

time keeping the instrument level, and their

backsite or foresite (or both) is short.

This could receive a weight of one (1)

 How Weights Can Be
Determined

Category 2: the second example might be where

the surveyor is on firm ground but still has a
shore backsite and is having a hard time seeing
his foresite. This type of angle could receive a
weight of two (2)
 How Weights Can Be
Determined
Category 3: the third example might be where the
surveyor is on firm ground, with good sites both
front and back, but the conditions are not very
good. It could be foggy or there might be a lot of
heat waves. This type of angle could receive a
weight of three (3)
 How Weights Can Be
Determined
Category 4: the last example is the best angle.
The surveyor is on firm ground with good sites,
turning an angle with strong strength of figure.
The weather is high overcast skies and about
65°f This type of angle could receive a weight of
four (4)
 Conclusion

Many programs, such as star*net, allow you to

weight your measurements very easily and with
very little practice the surveyor can use weights
and put the error where it belongs.
 Conclusion

Using weights helps take control of

measurements and helps distribute errors in
a more realistic manner that can result in a
better, more accurate survey.