A surface needs two coordinates
A parametrization describes a surface patch by a map . Two parameters are needed because a small patch has two independent tangent directions. If is continuously differentiable and
then it is regular at that parameter value. The cross product is perpendicular to both tangent vectors. Its length is the area of the parallelogram generated by those vectors, and therefore measures how parameter area is stretched in the tangent approximation.
Definition 4.5.1
(Scalar surface integral)
Suppose the parameter domain is compact and Jordan measurable, with piecewise boundary; the map is on a neighborhood of , regular on , and one-to-one on its interior. Suppose boundary identifications occur only along finitely many piecewise curve images, and is continuous on the surface . Then the scalar surface integral is
For area, put . This formula concerns an unoriented scalar density. A flux integral instead uses a chosen normal and is sensitive to orientation; that distinction matters when the order of the parameters is reversed.
The tangent explanation also shows why the norm belongs in the formula. A parameter rectangle of side lengths is approximately mapped to the parallelogram spanned by and . Its area is approximately the cross-product norm times . Summing and taking a limit yields the surface-area formula under the regularity and integration hypotheses above. The tangent argument motivates the formula; it does not on its own prove a surface-area theorem for arbitrary rough maps.
Graph charts
If is the graph of over a planar region , the natural chart is
Thus . The final coordinate of the cross product is 1, so any continuously differentiable graph is regular in this chart. The square root is the stretch factor, not the height of the surface. For a horizontal plane it equals 1, as it should.
Concept lensGeometric
The stretch factor as a tilt
The unit normal of the graph is . If is its angle with the vertical unit vector , then , so
Surface area is projected area divided by the cosine of the tilt. For , and . A vertical tangent plane has , which is where a graph chart must fail.
Quick check
(Find the area element for the graph .)
Compute and first.
Solution · Answer
and , so .
A paraboloid and its apparent singularity
Worked example 4.5.2
(Area of a paraboloid cap)
Find the area of the part of over the disk , where .
Using the graph formula and then polar coordinates in the parameter plane,
The substitution gives
We can reach the same integral through the surface parametrization . Its cross-product norm is . This map has a vanishing cross product at , and all angles there represent the same point. That is a failure of polar coordinates at the apex, not a singularity of the paraboloid: the graph chart has normal , which never vanishes. Apply the polar formula first on an annulus , with a seam along an angular edge, then let ; the omitted cap area tends to zero. This also explains why the polar factor must not be inserted a second time after computing the cross product.
Common mistake
(Inserting the polar factor twice)
The norm of the cross product of already contains the factor of polar coordinates. Writing gives a different integral. For small the correct area is close to , whereas the doubled factor gives an area of order .
A cylinder in two sets of coordinates
Worked example 4.5.3
(A half-cylinder in two parametrizations)
Consider the half-cylinder , , , and calculate both its area and .
Use
The two derivatives are and , so their cross-product norm is . The chart is regular and one-to-one on this rectangle, including its boundary. Therefore
Set , . Now and
The integral is unchanged: the new parameter rectangle has twice the width and the area factor is halved. This is not a coincidence, but an instance of the chain rule and change of variables.
Why changing coordinates preserves scalar area
Theorem 4.5.4
(Invariance of scalar area under reparametrization)
Let be regular on a patch and let be a bijective change of parameters with a inverse and nonzero Jacobian. Then is regular, and for continuous
Proof
(Chain rule and the Jacobian determinant)
Let be a bijective change of parameters with a inverse and nonzero Jacobian on the relevant patch. The chain rule and bilinearity of the cross product yield
Taking lengths introduces , exactly the factor required in the planar change-of-variables theorem.
Proof X-ray
Why the determinant appears exactly once
The chain rule writes and . In the cross product the terms and vanish and the two mixed terms combine to . That is the key move: a bilinear, alternating product turns the matrix into its determinant. Nonzero keeps the new chart regular; the absolute value appears only because we take lengths; injectivity of is what lets the planar change-of-variables theorem cancel without a multiplicity. Without it the integral picks up the multiplicity, as when an angle runs through two turns of a cylinder.
A negative determinant reverses the normal but does not change scalar area. One must transform the domain too. Conversely, if a parametrization traverses the interior twice, integrating over its whole domain counts the surface twice; regularity alone does not prevent this. For example, allowing a full-cylinder angular parameter to run through two complete turns doubles the calculated area.
Quick check
(Does a negative Jacobian determinant of a change of parameters change the scalar surface area? Explain.)
Look at how enters the transformed area element.
Solution · Answer
No. The area element involves , so only the orientation of the normal flips; the scalar area, and every scalar surface integral, is unchanged.
A sphere cannot be represented globally by one nonsingular longitude-latitude rectangle: the longitude coordinate collapses at the poles, and a seam is required. Use overlapping regular charts or remove the seam and poles temporarily, checking that the removed pieces contribute zero area. A chart's coordinate failure should not automatically be called a geometric singularity.
A graph density
Worked example 4.5.5
(Integrating the height over a plane graph)
On the graph over , calculate . Here , so the area factor is . Consequently
The integrand is the height composed with the chart; replacing it with 1 would calculate area instead. Integrating height against without the stretch factor would calculate a planar integral, not this surface integral.
Common mistake
(Forgetting the stretch factor)
On the graph the planar integral equals , whereas the surface integral equals . The factor is the ratio between surface area and planar area for this tilted plane and cannot be omitted.
Checklist
Choose a domain that covers the intended surface exactly once away from boundary seams. Check differentiability and regularity, compute the cross product, and distinguish scalar area from oriented flux. Include a density evaluated on the surface when one is given. At degenerate chart points, choose another chart or justify an improper limiting argument. More general rough surfaces and a full construction of surface measure are outside the scope of this note.
The first fundamental form as a computational check
Sometimes expanding a cross product is cumbersome. Introduce the scalar products
Proposition 4.5.6
(Area element from the first fundamental form)
Let be and let be as above. Then , and is regular at a point exactly when there.
Proof
(Lagrange's identity)
The vector identity yields the same area element . It also shows regularity is equivalent to : the two tangent vectors are linearly independent precisely when their parallelogram has positive area. A nonzero and nonzero alone are insufficient, because the two vectors could be parallel.
Quick check
(For the cone with , find and the area element.)
Compute and first.
Solution · Answer
and , so , , . Then , and the area element is .
For the cylinder above, , , , so . For the graph , the values are , , ; cancellation gives . These checks produce the formulas already used without introducing a second Jacobian.
Local injectivity versus global coverage
Regularity is a local statement about tangent directions. It does not show that two distant parameter values cannot land on the same surface point.
Counterexample mode
Regular but not one-to-one
One might claim that a regular chart on a domain covers each surface point once. Take on . Here has norm , so the chart is regular everywhere, yet . The area integral is , whereas the lateral surface of the unit cylinder of height has area . Restricting to avoids interior duplication; the two boundary edges form a seam.
When a surface is described by several charts, one must also prevent overlap from being counted twice. One practical method is to divide the surface into pieces whose interiors are disjoint and sum their integrals; shared regular boundary curves have zero surface area. The latter conclusion relies on the standard piecewise smooth setting, not on an assertion that every possible boundary is negligible. A chart can fail because of a poor choice of coordinates even while another chart remains regular, as at the paraboloid apex.