1 Spherical Coordinates
The purpose of spherical coordinates is to "describe" a 3dimensional space in terms of angles and the distance to
a "target", instead of rectangular (cartesian) coordinates.
Consider an orthogonal reference system Orj. or the "unit" vectors
i ,
, ,
i + j
, + .
/
r =
_
r
2
+ j
2
+ .
2
=
_
_
_
(r, j, .)
_
_
_
Then project the point (r, j, .) onto the "horizontal" plane rOj and onto the O. axis.
r sin
z = r cos
(x,y,z)
(0,0,0)
x
y
z
r
Imagine the origin (0, 0, 0) is at the "center" of the Earth, the O. axis is the North - South pole axis, the
horizontal plane rOj be the "ecuatorial" plane.
Consider two angles,
"0" the angle between the vector
(r, j, .)
onto the plane rOj.
It is easy enough to gure out that . = r cos 0 and the projection onto rOj is r sin0 .
Now we use polar coordinates in the rOj plane and we get r = r cos ,sin0 , j = r sin,sin0 .
1
y = r sin sin
x = r sin cos
r sin
(0,0,0)
x
y
The numbers (r, 0, ,) are called spherical coordinates. They are connected to cartesian coordinates by the
relations
r = r(r, 0, ,) = r cos ,sin0
j = j(r, 0, ,) = r sin,sin0
. = .(r, 0, ,) = r cos 0
To cover the 3-dimensional space we need
r (0, +) ,
0 (0, ] that is from the "North pole to the South pole"
, [0, 2) that is "all around the O. axis" ,
x
y
z
2
Therefore we get a function, a "one to one" correspondence between the 3-dimensional space R
3
and (0, +)
(0, ) [0, 2)
Not exactly the entire space.
We miss the origin (0, 0, 0) , for which r = 0 but there are no corresponding values for the angles , and 0
We also miss the O. axis, for which 0 = 0 (or ) but there is no corresponding value for the angle ,
However, when using spherical coordinates to compute triple integrals, missing just the O. axis is irrelevant.
The Jacobi matrix corresponding to spherical coordinates is
J =
_
_
_
@x
@r
@x
@
@x
@'
@y
@r
@y
@
@y
@'
@z
@r
@z
@
@z
@'
_
_
_ =
_
_
cos ,sin0 r cos ,cos 0 r sin,sin0
sin,sin0 r sin,cos 0 r cos ,sin0
cos 0 r sin0 0
_
_
and the corresponding jacobian is
det J =
=
= cos 0
+ (1)(r sin0)
=
= cos 0 r
2
[cos
2
, (1) sin
2
,
. .
1
] cos 0 sin0 + r sin0 r sin
2
0[cos
2
, (1) sin
2
,
. .
]
1
=
= r
2
sin0 (cos
2
0 + sin
2
0) = r
2
sin0
Examples.
To cover
1) a sphere centered at (0, 0, 0) with radius 1 we need r [0, 1] , 0 [0, ] , , [0, 2]
2) the upper hemisphere r
2
+ j
2
+ .
2
_ 1
2
, . _ 0 we need r [0, 1] , 0 [0, ,2] , , [0, 2]
3) the lower hemisphere r
2
+ j
2
+ .
2
_ 1
2
, . _ 0 we need r [0, 1] , 0 [,2, ] , , [0, 2]
Just as for polar coordinates, there is also an algebraic trick that leads to spherical coordinates.
0 _ r
2
= r
2
+ j
2
+ .
2
= 1 =
_
r
r
_
2
+
_
j
r
_
2
. .
A
2
+
_
.
r
_
2
We may replace
2
,
_
z
r
_
2
by sin
2
, cos
2
. Choose
_
r
r
_
2
+
_
j
r
_
2
. .
A
2
= sin
2
0 and
_
.
r
_
2
= cos
2
0
next
_
r
r
_
2
+
_
j
r
_
2
= sin
2
0 =
_
r
r sin0
_
2
+
_
j
r sin0
_
2
= 1
and again replace
_
x
r sin
_
2
,
_
y
r sin
_
2
by sin
2
, cos
2
. Choose
r
r sin0
= cos c and
j
r sin0
= sinc
Which nally leads to the spherical coordinates
r = r cos ,sin0 , j = r sin,sin0 , . = r cos 0
4) Consider now an ellipsoid centered at the origin
(r, j, .) R
3
,
r
2
a
2
+
j
2
/
2
+
.
2
c
2
_ 1
3
x
y
z
a
c
b
We use the same algebraics trick as before.
r
2
=
r
2
a
2
+
j
2
/
2
+
.
2
c
2
_ 1 =
clearly r runs from 0 to 1 , r [0, 1]
= 1 =
r
2
r
2
a
2
+
j
2
r
2
/
2
+
.
2
r
2
c
2
_ 1 = 1 =
_
r
ar
_
2
+
_
j
/r
_
2
. .
A
2
+
_
.
cr
_
2
We may replace
2
,
_
z
rc
_
2
by sin
2
, cos
2
. Choose
_
r
ar
_
2
+
_
j
/r
_
2
. .
A
2
= sin
2
0 and
_
.
cr
_
2
= cos
2
0
next
_
r
ar
_
2
+
_
j
/r
_
2
= sin
2
0 =
_
r
ar sin0
_
2
+
_
j
/r sin0
_
2
= 1
and again replace
_
x
ar sin
_
2
,
_
y
br sin
_
2
by sin
2
, cos
2
. Choose
r
ar sin0
= cos c and
j
/r sin0
= sinc
Which nally leads to the "generalised" spherical coordinates (or "elliptical" coordinates)
r = r(r, 0, ,) = ar cos ,sin0
j = j(r, 0, ,) = /r sin,sin0
. = .(r, 0, ,) = cr cos 0
Remark. In this case "r" no longer represents the distance from (0, 0, 0) to (r, j, .) , "0" and "c" are no
longer the angles dening spherical coordinates. However, it is a convenient way to cover an ellipsoid, we need
r [0, 1] , 0 [0, ) , c [0, 2)
4
Just as for spherical coordinates, we "miss" the O. axis (which is irrelevant when computing triple integrals).
The Jacobi matrix corresponding to "elliptical" coordinates is
J =
_
_
_
@x
@r
@x
@
@x
@'
@y
@r
@y
@
@y
@'
@z
@r
@z
@
@z
@'
_
_
_ =
_
_
a cos ,sin0 ar cos ,cos 0 ar sin,sin0
/ sin,sin0 /r sin,cos 0 /r cos ,sin0
c cos 0 cr sin0 0
_
_
and the corresponding jacobian is
det J =
= a/c
= a/cr
2
sin0
Example
Compute the volume of an ellipsoid
= (r, j, .) R
3
,
r
2
a
2
+
j
2
/
2
+
.
2
c
2
_ 1
o|() =
___
1drdjd. =
1
_
0
_
_
_
0
_
_
2
_
0
1 a/cr
2
sin0
. .
jdet Jj
dc
_
_
d0
_
_
dr =
=
1
_
0
_
_
_
0
2 a/cr
2
sin0
. .
jdet Jj
d0
_
_
dr = 2a/c
1
_
0
_
r
2
cos 0
_
0
[ dr = 2a/c
1
_
0
2r
2
dr = 4a/c
r
3
3
1
0
=
4a/c
3
For a = / = c = 1 we get a sphere, and the volume is
4R
3
3
.
6) Compute the volume of the domain
= (r, j, .) R
3
, r
2
+ j
2
+ .
2
_ 9 , .
2
_ r
2
+ j
2
, . _ 0
Clearly r
2
+ j
2
+ .
2
_ 9 describes the interior of a sphere centered at (0, 0, 0) with radius 3
The equation .
2
= r
2
+ j
2
represents a cone, since
for r = 0 we get .
2
= j
2
= . = j or . = j two straight lines in the jO. plane (this a vertical
"cut" )
for . =constant= ct we get r
2
+ j
2
= ct
2
and these are circles. (these are horizontal "cuts")
5
x
y
z
3
B A
y
z
z = y
z = -y
O
B
A
In the triangle O1 we have O = 3 , the angle ]O1 = ,4 .
Finally we use spherical coordinates to describe the domain
r [0, 3] , 0 [0, ,4] , c [0, 2]
Therefore the volume is
o|() =
___
1drdjd. =
3
_
0
_
_
_
=4
_
0
_
_
2
_
0
1 r
2
sin0dc
_
_
d0
_
_
_dr =
3
_
0
_
_
_
=4
_
0
2r
2
sin0d0
_
_
_dr =
6
= 2
3
_
0
_
_
_
=4
_
0
r
2
(cos 0)[
=4
0
d0
_
_
_dr = 2
3
_
0
r
2
(cos ,4 + cos 0)dr =
= 2(1
1
_
2
)
3
_
0
r
2
dr = 2(1
1
_
2
)
r
3
3
3
0
= 2
_
2 1
2
27
3
= 9(
_
2 1)
We may consider as well the domain
' = (r, j, .) R
3
, r
2
+ j
2
+ .
2
_ 9 , .
2
_ r
2
+ j
2
, . _ 0
In this case the domain lays inside the sphere, but outside the cone, and we have
o|(') =
___
1drdjd. =
3
_
0
_
_
_
=2
_
=4
_
_
2
_
0
1 r
2
sin0dc
_
_
d0
_
_
_dr = ....
7) Consider the domain
= (r, j, .) R
3
, r
2
+ j
2
+ .
2
_ 2 , . _ r
2
+ j
2
,
In this case the equation . = r
2
+ j
2
represents a (circular) paraboloid , since
for r = 0 we get . = j
2
which is clearly a parabola in the jO. plane (this a vertical "cut" )
for . =constant= ct we get r
2
+ j
2
= ct and these are circles. (these are horizontal "cuts")
y
z
x
C
A
B
Spherical coordinates are no longer useful in this case.
Clearly r
2
+ j
2
+ .
2
_ 2 describes the interior of a sphere centered at (0, 0, 0) with radius
_
2
Next . _ r
2
+ j
2
describes the interior of the paraboloid.
Consequently this domain is projectable on the rOj plane, and the projection is a disk centered at (0, 0, 0) , let
1 be this disk.
The intersection of the sphere with the paraboloid is a circle centered at 1 with radius = 1
7
x
y
A
B
(0,0,0) C
To get the intersection proceed as follows
_
r
2
+ j
2
+ .
2
= 2
. = r
2
+ j
2
= . + .
2
= 2 = .
2
+ . 2 = 0 = .
1;2
=
1 3
2
Only . = 1 is acceptable since . _ 0 . So O =
_
2 , O1 = 1 = 1 =
_
2 1 = 1
The disk projection disk 1 is actually 1 = (r, j, 0) R
3
, r
2
+ j
2
_ 1
x
y
(0,0,0)
C
D
To compute the volume of the domain "." runs from the paraboloid (. = r
2
+ j
2
) to the sphere (. =
_
2 r
2
j
2
)
o|() =
___
N
1drdjd. =
__
D
_
_
_
_
z=
_
2x
2
y
2
_
z=x
2
+y
2
1d.
_
_
_
_
drdj =
__
D
_
_
2 r
2
j
2
(r
2
+ j
2
)
_
drdj =
8
next we need polar coordinates r = r cos t , j = r sint , to cover the disk 1 r [0, 1] , t [0, 2]
=
1
_
0
_
_
2
_
0
_
_
2 r
2
r
2
_
rdt
_
_
dr =
1
_
0
2
_
_
2 r
2
r
2
_
rdr =
= 2
1
_
0
_
_
2 r
2
r r
3
_
dr = 2
_
_
_
2 r
2
_
3
3
2
r
4
4
_
1
0
= 2
_
3
2
1
4
+
_
_
2
_
3
3
2
_
=
_
6
_
2
7
2
_
Remark.
Many text books call this "cylindrical" cordinates (r, t, .)
r = r(r, t) = r cos t
j = j(r, t) = r sint
. = .
It seems to be a "hystorical" denomination, no longer useful since practically we project the domain onto the
rOj plane and then we use polar coordinates in that plane.
9