Surface parametrizations and flux integrals

Let r(u,v) be a C¹ parametrization of a surface patch on a plane parameter region. Regularity means its tangent vectors ru and rv are linearly independent, so their cross product is nonzero. A small coordinate rectangle produces a tangent parallelogram spanned by these vectors times its side lengths. Its area scale is the norm of their cross product. Multiplying a scalar surface density by that norm and integrating over the parameter region gives the corresponding accumulation on the patch.

Coverage must be checked as carefully as local scale. A parametrization that covers a sheet twice generally counts its density twice. Many familiar surfaces need several patches, and overlaps must be handled without double-counting; boundaries of well-behaved patches can often be assigned to one side because they have surface area zero. The regularity condition may fail at coordinate singularities even when the surface itself is smooth, as at the poles of spherical coordinates. Another patch can describe those points regularly.

For a vector field F and an oriented surface, flux integrates the normal component F dotted with the chosen unit normal. In parameters, the oriented area vector is ru cross rv when that order matches the selected orientation. Its magnitude already includes the area scale, so F(r) dotted with the cross product is integrated directly. Multiplying by the cross-product norm again would count the area scale twice. Reversing the normal reverses flux but does not change a scalar surface-density integral.

Orientation is a consistent choice over the surface, not a separate arbitrary sign at every point. Closed surfaces bounding ordinary solid regions have a natural outward orientation. For an open sheet, the problem must specify a side or a compatible boundary orientation. Some surfaces, such as a Möbius strip, admit no globally consistent unit normal, so the ordinary globally oriented flux setup does not apply to them. The local area formula still makes sense because taking a norm removes the sign choice.

For a graph z=g(x,y), take r(x,y)=(x,y,g(x,y)). The upward oriented area vector is (−gx,−gy,1) dx dy, and its norm is the square root of 1+gx²+gy² times projected area. A steep sheet therefore has more actual area than its horizontal shadow. Yet a vertical unit field dotted with that upward area vector gives exactly dx dy. Its flux measures the horizontal projection, while a constant scalar density measures the larger slanted area.

Before integrating, state whether the task asks for mass, unsigned area, or oriented flux. Check units as well: a velocity field integrated over area produces volume per time, whereas a mass density on the sheet produces mass. For flux, examine simple limiting cases. A tangential field contributes zero, reversing a normal changes the sign, and an outward field on a closed surface should have a plausible outward total. These checks catch errors that a mechanically correct antiderivative cannot reveal.