# 1

Bivariate Normal Distribution
and Error Ellipses
Bivariate Normal Distribution
• Joint distribution of two random
variables
• Very useful when dealing with
planimetric (x,y) positions in surveying
• Density function is a bell-shaped surface
centered at x = µ
x
and y = µ
y
Bivariate Density Function Bivariate Density Function
• The joint density function of two random variables (X
and Y) which have a bivariate normal distribution is:
where:
µ
x
and σ
x
= mean and standard deviation of X
µ
y
and σ
y
= mean and standard deviation of Y
ρ = correlation coefficient of X and Y
2 2
2
2
1 1
( , ) exp 2
2(1 )
2 1
¦ ¹
| | | | − − | | | | − − − ¦ ¦
= − + | | ´ ` | |
| |

− \ ¹ \ ¹ ¦ ¦ \ ¹ \ ¹
¹ )
y y x x
x x y y
x y
y y x x
f x y
u u u u
ρ
ρ σ σ σ σ
πσ σ ρ
=
xy
x y
σ
ρ
σ σ
Marginal Density Functions
• Density functions for X and Y
• Components of the bivariate normal distribution at the
X and Y axes
• The same as the usual density functions for individual
normally distributed random variables
2
2
1 1
( ) exp
2 2
1 1
( ) exp
2 2
¦ ¹
| | − ¦ ¦
= −
´ ` |
\ ¹ ¦ ¦
¹ )
¦ ¹
| | −
¦ ¦
= − | ´ `
|
¦ ¦ \ ¹
¹ )
x
x x
y
y y
x
f x
y
f y
u
σ σ π
u
σ σ π
Cutting Ellipse/Ellipse of Intersection
• When a plane parallel to the x,y coordinate plane cuts
the bivariate density surface at a height K, an ellipse is
formed
• The equation of this ellipse is: (obtained by making
f(x,y) = K)
where:
2 2
2 2
2 (1 )
| | | | − − | | | | − −
− + = − | | | |
| |
\ ¹ \ ¹\ ¹ \ ¹
y y x x
x x y y
y y x x
c
u u u u
ρ ρ
σ σ σ σ
1
2 2 2 2 2 2
ln 4 (1 ) a constant

= − =
x y
c K π σ σ ρ
2
Example
The parameters of a bivariate normal distribution
are µ
x
= 4, µ
y
= 5, σ
x
= 1, σ
y
= 0.5, and ρ
xy
= 0.5.
A plane intersects the density function at K = 0.1
above the x,y coordinate plane. Evaluate the
ellipse of intersection.
Solution
The equation of the ellipse is:
Simplifying,
1
2 2 2 2 2 2
2 2
ln 4 (0.1) (1) (0.5) (1 (0.5) ) 2.60
(1 ) (1 0.25)(2.60) 1.95

= − =

− = − =
c
c
π
ρ
2 2
4 4 5 5
2(0.5) 1.95
1 1 0.5 0.5
− − − − | | | || | | |
− + =
| | | |
\ ¹ \ ¹\ ¹ \ ¹
x x y y
2 2
( 4) 2( 4)( 5) 4( 5) 1.95 − − − − + − = x x y y
Error Ellipse
• Produced when the bivariate probability
distribution is centered at the origin (µ
x
= µ
y
= 0)
This equation is used if we want to represent the
random errors only
2 2
2
2
1 1
( , ) exp 2
2(1 )
2 1
¦ ¹
| | | | | | | | − ¦ ¦

= − + | | ´ ` | |
| |

− \ ¹ \ ¹ ¦ ¦ \ ¹ \ ¹
¹ )
x x y y
x y
x x y y
f x y ρ
ρ σ σ σ σ
πσ σ ρ
Error Ellipse
• The corresponding equation for the cutting
ellipse in this case would be:
This equation represents a family of error
ellipses centered on the origin
2
2
2 2
2 (1 )
| | | |
| | | |
− + = − | |
| |
| |
\ ¹ \ ¹
\ ¹ \ ¹
x x y y
x x y y
c ρ ρ
σ σ σ σ
Standard Error Ellipse
• When c=1, we get the equation of the standard
error ellipse:
• Represents the area of uncertainty for the location
of a control point
• Size, shape, and orientation of a standard error
ellipse are governed by the parameters σ
x
, σ
y
, and
ρ.
2 2
2
2 (1 )
| | | |
| | | |
− + = − | | | |
| |
\ ¹ \ ¹\ ¹ \ ¹
x x y y
x x y y
ρ ρ
σ σ σ σ
The Standard Error Ellipse
3
Sample variants of the standard error
ellipse (by varying the parameters)
Standard Error Ellipse
• In general, the principal axes (x’ and y’)
do not coincide with the coordinate axes
(x and y)
• The major axis of the ellipse (x’) makes
an angle θ with respect to the x-axis
Positional Errors
• A positional error is expressed in the
x,y coordinate system by the random
vector
• The same positional error is expressed
in the x’,y’ coordinate system by the
random vector

X
Y

X
Y
Orthogonal (Rotational) Transformation
• The two vectors can be related by the equation:
• θ is the angle of rotation
• Transformations from one coordinate system to
the other can be made using the above equation
correlated errors may be transformed to
uncorrelated errors using the equation
cos sin
sin cos

=

′ −

X X
Y Y
θ θ
θ θ
Covariance Matrices
• The covariance matrices for the random
vectors are:
X’ and Y’ are uncorrelated (they are the
principal axes of the ellipse)

X
Y

X
Y
2
2

x xy
xy y
σ σ
σ σ
2
2
0
0

x
y
σ
σ
Covariance Matrices
Recall:
Applying this to the vector relationship, we get:
Solving the matrix:
T
yy xx
A A Σ = Σ
2 2
2 2
0 cos sin cos sin
0 sin cos sin cos
x x xy
y xy y
σ σ σ θ θ θ θ
σ σ σ θ θ θ θ

=

2 2 2 2 2
2 2 2 2 2
2 2 2 2
cos 2 sin cos sin
sin 2 sin cos cos
0 ( ) sin cos (cos sin )
x x xy y
y x xy y
y x xy
σ σ θ σ θ θ σ θ
σ σ θ σ θ θ σ θ
σ σ θ θ σ θ θ

= + +
= + +
= − + −
Eq. 1
Eq. 2
Eq. 3
4
Orientation of Error Ellipse
But and
The third equation (Eq. 3) therefore becomes:
Which can be further simplified as:
sin 2
sin cos
2
θ
θ θ =
2 2
cos sin cos 2 θ θ θ − =
2 2
2
tan 2
xy
x y
σ
θ
σ σ
=

2 2
1
( ) sin 2 cos 2 0
2
y x xy
σ σ θ σ θ − + =
Semimajor and Semiminor Axes
The first two equations (Eq. 1 and Eq. 2) become:
σ
x’
= semimajor axis
σ
y’
= semiminor axis
1/ 2
2 2 2 2 2
2 2
1/ 2
2 2 2 2 2
2 2
( )
2 4
( )
2 4
x y x y
x xy
x y x y
y xy
σ σ σ σ
σ σ
σ σ σ σ
σ σ

+ −
= + +

+ −
= − +

Standard Error Ellipse
Example:
The random error in the position of a survey
station is expressed by a bivariate normal
distribution with parameters µ
x
= µ
y
= 0, σ
x
= 0.22 m, σ
y
= 0.14 m, and ρ = 0.80.
Evaluate the semimajor axis, semiminor axis,
and the orientation of the standard error
ellipse associated with this position error.
Standard Error Ellipse
Solution:
( ) ( ) ( )
( ) ( )
( )
( ) ( ) ( )
( )
( )
2
2 2 2 2
2
1 2
2 1 2
2 2 2
2 2
2 2 2
1 2
2
2 2 2 2
2 2
2 2
0 8 0 22 0 14 0 0246 m
0 22 0 14
0 0340 m
2 2
0 22 0 14
0 0246 0 0285 m
4 4
0 0340 0 0285
2 4
0 0625 m
xy x y
x y
/
/
x y
xy
/
x y x y
x' xy
x'
. . . .
. .
.
. .
. .
. .
.
σ ρσ σ
σ σ
σ σ
σ
σ σ σ σ
σ σ
σ
= = =
+ +
= =

− −

+ = + =

− +

= + + = +

=
Standard Error Ellipse
( )
( )
( ) ( )
1 2
2
2 2 2 2
2 2
2 2
2 2 2 2
0 0340 0 0285
2 4
0 0055 m
0 0625 0 25 m
0 0055 0 074 m
2 2 0 0246
2
0 22 0 14
2 1 711
/
x y x y
y' xy
y'
x'
y'
xy
x y
. .
.
. .
. .
.
tan
. .
tan .
σ σ σ σ
σ σ
σ
σ
σ
σ
θ
σ σ
θ

− +

= − + = −

=
= =
= =
= =
− −
=
2θ lies in the 1
st
xy
and σ
x
2
- σ
y
2
are both
positive. Therefore, 2θ = tan
-1
(1.711) = 59.7° and the
orientation of the error ellipse is θ = 29.8°
Probability of Error Ellipse
Assumption: Independent (Uncorrelated) random errors X
and Y.
Given a point whose position is defined by the random
errors (X and Y), the point will lie within the error
ellipse if:
2 2
2 2
2 (1 )
| | | | | | | |
− + = − | | | |
| |
\ ¹ \ ¹\ ¹ \ ¹ x x y y
x x y y
c ρ ρ
σ σ σ σ
2 2
2
x y
X Y
c
σ σ
| | | |
+ = | |
|
\ ¹ \ ¹
2 2
2
x y
X Y
c
σ σ
| | | |
+ ≤ | |
|
\ ¹ \ ¹
5
Probability of Error Ellipse
The random variable U such that:
has a chi-square distribution with two degrees of freedom.
(Recall from the past lecture that Y has a chi-square
distribution:
where Z
n
are independent standard normal random
variables)
2 2
x y
X Y
U
σ σ
| | | |
= + | |
|
\ ¹ \ ¹
2 2 2 2
1 2 n
Y Z Z Z = + + + ⋯
Probability of Error Ellipse
For two degrees of freedom, the probability density
function of U can be derived:
( / 2) 1 / 2
/ 2
(2/ 2) 1 / 2
2/ 2
1
( )
2 ( / 2)
1
( )
2 (2 / 2)
n y
n
u
f y y e
n
f u y e
− −
− −
=
Γ
=
Γ
/ 2
1
( ) for u>0
2
u
f u e

=
Probability Density Function of U
Probability of Error Ellipse
Therefore, the probability that the position given by X and
Y lies on/within the error ellipse is:
2
2 2
2 2
/ 2
0
1

2
x y
c
u
X Y
P c P U c
e du
σ σ

| | | |
+ ≤ = ≤ | |

|

\ ¹ \ ¹

=

2
2 2
2 / 2
1
c
x y
X Y
P c e
σ σ

| | | |

+ ≤ = − | |
|

\ ¹ \ ¹

Probability
Probability of Error Ellipse
Example:
The random error in the position of a
survey station is expressed by a bivariate
normal distribution with parameters µ
x
=
µ
y
= 0, σ
x
= 0.22 m, σ
y
= 0.14 m, and ρ =
0.80. Evaluate the semimajor and
semiminor axes of the error ellipse
within which it is 0.90 probable that the
error in position will lie.
Probability of Error Ellipse
Solution:
The semimajor and semiminor
axes are:
| |
2
2
2
2
2
2
2
0 90
1 0 90
0 10
2 3026
2
4 6052
2 146
c /
c /
P U c .
e .
ln . ln e
c
.
. c
c .

≤ =

− =

=

− =
=
=
( )
( )
2 146 0 25 0 54 m
2 146 0 074 0 16 m
x'
y'
c . . .
c . . .
σ
σ
= =
= =