5 Vector calculus in spherical coordinates z

In studies of radiation from compact antennas it is more convenient to use rsinθ y

spherical coordinates instead of the Cartesian coordinates that we are
familiar with. In this lecture we will learn r

r cos θ
θ
1. how to represent vectors and vector fields in spherical coordinates,

2. how to perform div, grad, curl, and Laplacian operations in spherical φ r sin θ sin φ
coordinates. r sin θ cos φ x

• A 3D position vector
r = (x, y, z)
with Cartesian coordinates (x, y, z) is said to have spherical coordinates
(r, θ, φ) where
✬ ✩
✬ ✩
In terms of spherical coordinates, Cartesian coordi-
! nates can be expressed as
length r ≡ |r| = x2 + y 2 + z 2
!
−1 x2 + y 2 x = r sin θ cos φ
zenith angle θ = tan
z y = r sin θ sin φ
y
azimuth angle φ = tan−1 . ✫ z = r cos θ. ✪
✫ x ✪

Ratios x/r = sin θ cos φ, y/r = sin θ sin φ, and z/r = cos θ are referred
to as direction cosines as they represent the cosine of the angle between
vector r = (x, y, z) and the x-, y-, and z-axes, respectively.

1
• In Cartesian coordinates we have mutually orthogonal unit vectors z r̂

x̂, ŷ, ẑ φ̂
y
pointing in the direction of increasing Cartesian coordinates x, y, z, θ̂
r
respectively.
θ

r cos θ
• Likewise, in spherical coordinates we have mutually orthogonal unit
vectors rsinθ r sin θ sin φ
φ
r̂, θ̂, φ̂ r sin θ cos φ x
pointing in the direction of increasing coordinates r, θ, φ, respectively. Unit-vectors r̂, θ̂,and φ̂ shown
in red, green, and blue point
• However, unlike x̂, ŷ, ẑ, the unit vectors r̂, θ̂, φ̂ are not global — in mutually orthogonal direc-
tions of increasing spherical
rather they are local in the sense that their directions depend on the coordinates r, θ, and φ, re-
local coordinates. spectively, such that θ̂× φ̂ = r̂.
Note that r̂, θ̂,and φ̂ are local
unit vectors (i.e., coordinate
– The local nature of r̂, θ̂, φ̂ becomes clear when they are expressed dependent) unlike the global
unit vectors x̂, ŷ, and ẑ of the
in terms of the global unit vectors x̂, ŷ, ẑ as follows: Cartesian coordinate system.

r (x, y, z)
r̂ = = = x̂ sin θ cos φ + ŷ sin θ sin φ + ẑ cos θ
r r
(−y, x, 0)
φ̂ = ! = −x̂ sin φ + ŷ cos φ
x2 + y 2
θ̂ = φ̂ × r̂ = x̂ cos θ cos φ + ŷ cos θ sin φ − ẑ sin θ

Make sure you understand each of the terms above with reference
to the figure shown in the margin.

2
• In Cartesian coordinates we have an infinitesimal volume element z
(x, y, z + ∆z)
dV = dxdydz y (x, y + ∆y, z)
(x, y, z)
which is used in 3D volume integrals and often denoted as “d3r”. (x + ∆x, y, z)

– Note that dV is the volume of a rectangular box formed by the

intersection of constant coordinate surfaces of two infinitesi- x
mally close points having a separation vector Any vector

A(r) = Ax x̂ + Ay ŷ + Az ẑ,
dr = x̂dx + ŷdy + ẑdz.
where Ax , Ay , and Az are the projec-
tions of A(r) on red, green, and blue
• Infinitesimal volume element d3r expressed in terms of spherical coor- arrows aligned with x̂,ŷ, ẑ, respectively.
z dr
dinates and their increments is y
r sin θdφ
dV = (dr) (rdθ) (r sin θdφ) = r2 sin θdrdθdφ.
r rdθ

– Once again dV is the volume of a rectangular box formed by the θ θ

intersection of constant coordinate surfaces of two infinitesi- in
rs
mally close points having a separation vector φ x
Any vector
dr = r̂dr + θ̂rdθ + φ̂r sin θdφ.
A(r) = Ar r̂ + Aθ θ̂ + Aφ φ̂,

– Note that in this case constant coordinate surfaces are no longer where Ar , Aθ , and Aφ are the projec-
tions of A(r) on red, green, and blue
planar globally, but over infinitesimal dimensions of dV the sur- arrows aligned with r̂,θ̂, φ̂, respectively.

faces will appear locally planar.

3
✬ ✩
✬ ✩
In Cartesian coordinates div, curl, In spherical coordinates div, curl, and grad
and grad
1 ∂(r2Ar ) 1 ∂(sin θAθ ) 1 ∂Aφ
∂Ax ∂Ay ∂Az ∇·A = 2 + +
∇·A = + + r ∂r r sin θ ∂θ r sin θ ∂φ
∂x ∂y ∂z
" "
" r̂ θ̂ φ̂ "
" " " r2 sin θ r sin θ r "
" x̂ ŷ ẑ " " "
" " " "
" ∂ ∂ ∂ " ∂ ∂ ∂
∇ × A = " ∂x ∇ × A = " ∂r
" "
∂y ∂z " ∂θ ∂φ
"
" Ax Ay Az
" " " "
" " "
" "
" Ar rAθ r sin θAφ "
∂V ∂V ∂V
∇V = x̂ + ŷ + ẑ
∂x ∂y ∂z ∂V 1 ∂V 1 ∂V
∇V = r̂ + θ̂ + φ̂
are obtained by applying the del operator ∂r r ∂θ r sin θ ∂φ

∂ ∂ ∂ are obtained for vectors

∇≡( , , )
∂x ∂y ∂z A = Ar r̂ + Aθ θ̂ + Aφ φ̂
“algebraically” to vectors
and scalars
A = Ax x̂ + Ay ŷ + Az ẑ V (r, θ, φ)
as indicated above. Note that there is no del operator
and scalars
that “works algebraically” in spherical coordinates.
V (x, y, z) ✫ ✪

as indicated above.
✫ 4 ✪
 z
Example 1: Verify the r̂ component of ∇ × A formula in spherical coordinates by
showing that it corresponds to
# dr
A · dl y
lim C
AC →0 AC r sin θdφ
where AC is the enclosed area of contour C orthogonal to r̂ marked in the margin r rdθ
by blue and green edges.

Solution: In spherical coordinates θ θ

in
"
" r̂ θ̂ φ̂
"
" rs
" r2 sin θ r sin θ r x
"
"
"
"
" φ
" ∂ ∂ ∂
∇ × A = " ∂r
"
∂θ ∂φ
"
" "
" "
" "
" Ar rAθ r sin θAφ "

and, therefore, r̂ component of ∇ × A is

1 ∂ ∂ 1 ∂ ∂
(∇ × A) · r̂ = ( r sin θA φ − rA θ ) = ( sin θA φ − Aθ ).
r2 sin θ ∂θ ∂φ r sin θ ∂θ ∂φ

To show that this expression corresponds (as it should by definition) to

#
A · dl
lim C
AC →0 AC
where circulation path C and enclosed area AC are as described in the question
statement, we first note that
AC ≈ (r sin θdφ)(rdθ)

5
 to second order in increments dθ and dφ. Also,
$
A · dl = Aθ (r, θ, φ)rdθ + Aφ (r, θ + dθ, φ)r sin(θ + dθ)dφ
C
−Aθ (r, θ, φ + dφ)rdθ − Aφ (r, θ, φ)r sin θdφ

starting on the green edge. Thus

#
C A · dl Aθ (r, θ, φ) − Aθ (r, θ, φ + dφ)
=
AC r sin θdφ
Aφ (r, θ + dθ, φ) sin(θ + dθ) − Aφ (r, θ, φ) sin θ
+
r sin θdθ
which yields in the limit of vanishing dθ and dφ
1 ∂ 1 ∂
− Aθ + sin θAφ ≡ (∇ × A) · r̂
r sin θ ∂φ r sin θ ∂θ
as requested.

See Appendix A and B in Rao for a complete coverage of the

derivation of div, grad, curl in spherical coordinates.

6
Example 2: Verify the gradient procedure
∂V 1 ∂V 1 ∂V
∇V = r̂ + θ̂ + φ̂
∂r r ∂θ r sin θ ∂φ
in spherical coordinates.

Solution: Independent of the coordinate employed, the total differential dV and the
gradient ∇V of a scalar field V (r) are related by

dV = ∇V · dr.

In the Cartesian coordinate system where V = V (x, y, z), this relation expands as
∂V ∂V ∂V
dV = dx + dy + dz = ∇V · (x̂dx + ŷdy + ẑdz)
∂x ∂y ∂z
and implies
∂V ∂V ∂V
∇V = x̂ + ŷ + ẑ.
∂x ∂y ∂z
Likewise, for spherical coordinates where V = V (r, θ, φ), we have
∂V ∂V ∂V
dV = dr + dθ + dφ = ∇V · (r̂dr + θ̂rdθ + φ̂r sin θdφ)
∂r ∂θ ∂φ
implying that
∂V 1 ∂V 1 ∂V
∇V = r̂ + θ̂ + φ̂.
∂r r ∂θ r sin θ ∂φ

7
Example 3: Show that the Laplacian of a scalar field V (r, θ, φ) is specified as

2 1 ∂ 2 ∂V 1 ∂ ∂V 1 ∂ 2V
∇ V = 2 (r )+ 2 (sin θ )+ 2 2 .
r ∂r ∂r r sin θ ∂θ ∂θ r sin θ ∂φ2

Solution: Since the Laplacian is the divergence of a gradient, we start by noting that
∂V 1 ∂V 1 ∂V
∇V = r̂ + θ̂ + φ̂.
∂r r ∂θ r sin θ ∂φ
Applying to this vector the divergence formula
1 ∂(r2(∇V )r ) 1 ∂(sin θ(∇V )θ ) 1 ∂(∇V )φ
∇ · ∇V = 2 + +
r ∂r r sin θ ∂θ r sin θ ∂φ
1 ∂V
1 ∂(r2 ∂V
∂r ) 1 ∂(sin θ 1r ∂V
∂θ ) 1 ∂( r sin θ ∂φ )
= 2 + +
r ∂r r sin θ ∂θ r sin θ ∂φ
the above result for the Laplacian is readily obtained.