Surface integrals are similar to line integrals: line integrals are integrals on 1D curves, and surface integrals are integrals on 2D "curves" (i.e., on 2D surfaces). The steps for calculating a surface integral are basically the same, except things are harder just because surfaces live in 3D space.
A vector surface integral is called a flux integral and looks like
∬SF⋅dS
where S is an oriented surface, and the steps to calculate one are:
Parametrize S via a function G(u,v) with domain D.
Calculate the normal vector N(u,v), which is either
Gu×GvorGv×Gu
depending on the orientation of S. (Note that Gu×Gv=−Gv×Gu, so based on the orientation, you may or may not need to multiply your cross product by −1.)
Plug in everything to get a 2D integral:
∬SF⋅dS=∬DF(G(u,v))⋅N(u,v)dudv.
Integrate like normal.
Example 1.
Calculate ∬SF⋅dS, with F(x,y,z)=⟨0,3,x⟩ and S is the part of the sphere x2+y2+z2=9 where x≥0, y≥0, and z≥0, with outward pointing normal.
Solution.
First, we need to parametrize this region. Since it's a part of a sphere, we can base our parametrization on spherical coordinates:
G(θ,φ)=⟨ρsinφcosθ,ρsinφsinθ,ρcosφ⟩
The sphere has radius 3, so ρ=3. To get x≥0 and y≥0, we need 0≤θ≤2π, and to get z≥0, we need 0≤φ≤2π, so the parametrization is
To figure out what N is, we need to figure out whether Gθ×Gφ points inwards or outwards, which we can do by just testing it at a point. We can test it at the point (x,y,z)=(3,0,0), which corresponds to (θ,φ)=(0,2π), so
Gθ(0,2π)×Gφ(0,2π)=⟨−9,0,0⟩,
which points inwards. This means we need to reverse it to get an outward pointing normal, i.e.,