Why the Choice of Coordinates Matters
A double integral can often be written in more than one coordinate system. The choice affects both parts of the calculation: how complicated the region's bounds are and how complicated the integrand becomes after substitution.
Polar coordinates are often effective when a region is built from circles, circular sectors, or radial boundaries, or when the integrand depends on the distance from the origin. Cartesian coordinates are often more direct for rectangles and regions bounded by horizontal or vertical lines, especially when the integrand is already simple in \(x\) and \(y\). Neither system is automatically best just because the region or integrand has one familiar feature.
What Changes in Polar Coordinates?
In polar coordinates, a point is described by its distance \(r\) from the origin and its angle \(\theta\) from the positive \(x\)-axis. The conversion formulas are \(x=r\cos\theta\) and \(y=r\sin\theta\), with \(r\geq 0\). A double integral over a region described by \(a\leq\theta\leq b\) and \(g(\theta)\leq r\leq h(\theta)\) takes the form below, provided these bounds describe the region without unintended repetitions and the integrand is integrable there.
The extra factor \(r\) is essential. It comes from how areas change under the coordinate transformation: a small radial change and a small angular change span an area approximately \(r\,dr\,d\theta\). More precisely, the absolute value of the Jacobian determinant of \((r,\theta)\mapsto(x,y)\) is \(r\). The change-of-variables formula therefore gives \(dA=r\,dr\,d\theta\).
At the origin, polar coordinates do not assign a unique angle, and an angular interval can also represent the same ray at its endpoints. These coordinate duplications affect only boundaries or individual points in the usual setups; bounds should still be chosen so that the interior of the region is covered once.
A Practical Comparison
Before setting up an integral, make two comparisons. First, ask which coordinates describe the boundary with fewer pieces and simpler formulas. A circle centered at the origin is the simple radial bound \(r=c\), while a rectangle may require a radial bound that changes with angle. Second, substitute the coordinates into the integrand. A function of \(x^2+y^2\) becomes a function of \(r^2\), but a simple Cartesian function can become an angular expression with an inconvenient radial bound.
| Feature | Often favors polar coordinates | Often favors Cartesian coordinates |
|---|---|---|
| Region boundary | Circles centered at the origin, sectors, or radial curves | Rectangles or boundaries given simply as \(y\) in terms of \(x\), or \(x\) in terms of \(y\) |
| Integrand | Dependence on \(x^2+y^2\), or on distance from the origin | A simple expression in \(x\) and \(y\) that remains simple under Cartesian bounds |
| Overall setup | Few angular intervals and straightforward radial limits | Simple nested bounds without a need to split the region |
This table gives clues, not automatic rules. A good choice is the one that keeps the complete integral manageable. In particular, check whether either choice requires splitting the region into pieces.
Worked Example: A Radial Integrand on a Disk
Worked Example: A Radial Integrand on a Disk
Evaluate \(\iint_D (x^2+y^2)\,dA\), where \(D\) is the disk \(x^2+y^2\leq 4\). The integrand and the region both have circular structure, so polar coordinates are a natural choice.
In polar coordinates, \(x^2+y^2=r^2\). The disk is described by \(0\leq r\leq 2\) and \(0\leq\theta\leq 2\pi\). Including the area factor gives
Integrate with respect to \(r\). Since \(\frac{d}{dr}(r^4/4)=r^3\),
Thus the remaining integral is
As a check, the radial antiderivative differentiates back to the required radial integrand, \(r^3\), and the angular factor is the full angular range, \(2\pi\). Both the radius limit and the angle range describe the entire disk once.
In Cartesian coordinates, one possible description is \(-2\leq x\leq2\) and \(-\sqrt{4-x^2}\leq y\leq\sqrt{4-x^2}\). The integrand remains \(x^2+y^2\), but the curved \(y\)-limits make the setup less direct. Here polar coordinates simplify the region and turn the integrand into a power of \(r\).
Worked Example: A Square Where Cartesian Bounds Are Simpler
Worked Example: A Square Where Cartesian Bounds Are Simpler
Evaluate \(\iint_S(x+y)\,dA\), where \(S\) is the unit square \(0\leq x\leq1\), \(0\leq y\leq1\). The region is already a rectangle, and the integrand is simple in Cartesian coordinates.
Using \(x\) as the outer variable, the Cartesian setup is
For fixed \(x\), integrate the two terms with respect to \(y\):
Then integrate with respect to \(x\):
As a check, differentiating the displayed antiderivative with respect to \(x\) gives \(x+\frac12\), the integrand of the outer integral. The Cartesian bounds also describe the square directly, with no split.
Polar coordinates are possible, but the square's radial boundary changes direction at \(\theta=\pi/4\). On \(0\leq\theta\leq\pi/4\), the ray meets \(x=1\) first, so the outer radial limit is \(\sec\theta\). On \(\pi/4\leq\theta\leq\pi/2\), it meets \(y=1\) first, giving \(\csc\theta\). The resulting polar integral has two pieces:
For completeness, evaluate these pieces. In the first piece, integrating in \(r\) gives \(\frac13(\cos\theta+\sin\theta)\sec^3\theta\). Since \((\cos\theta+\sin\theta)\sec^3\theta=\sec^2\theta+\sin\theta/\cos^3\theta\),
Interchanging \(x\) and \(y\) maps the square onto itself and leaves \(x+y\) unchanged. Equivalently, the second polar piece has the same value as the first after replacing \(\theta\) by \(\frac{\pi}{2}-\theta\). Thus \(I_2=\frac12\), and the total is \(I_1+I_2=1\), agreeing with the Cartesian evaluation. This agreement checks the calculation, while the single Cartesian integral shows why Cartesian coordinates are the more efficient choice here.
Worked Example: A Quarter-Disk with a Distance-Based Integrand
Worked Example: A Quarter-Disk with a Distance-Based Integrand
Evaluate \(\iint_Q \frac{x}{\sqrt{x^2+y^2}}\,dA\), where \(Q\) is the part of the unit disk in the first quadrant. At the origin the displayed integrand is undefined; assign it any value there, such as zero. A single point does not change the integral. The region is a quarter-disk, and the integrand involves distance from the origin, so polar coordinates simplify both.
The bounds are \(0\leq\theta\leq\pi/2\) and \(0\leq r\leq1\). For \(r>0\), the integrand becomes \(r\cos\theta/r=\cos\theta\). Including the area factor,
For the check, differentiating \(r^2/2\) gives \(r\), and differentiating \(\sin\theta\) gives \(\cos\theta\). The quarter-disk bounds include each interior point once. In Cartesian coordinates, the corresponding integral is \(\int_0^1\int_0^{\sqrt{1-x^2}}\frac{x}{\sqrt{x^2+y^2}}\,dy\,dx\). Integrating in \(y\) introduces an inverse hyperbolic function, so the polar form is notably more direct.
A Common Misjudgment: Circular-looking Does Not Mean Polar Bounds Are Automatic
A coordinate choice still has to describe the actual region. For example, the unit square in the first quadrant is not the quarter-disk \(0\leq r\leq1\), \(0\leq\theta\leq\pi/2\). Treating it as that quarter-disk gives the wrong area:
whereas the square's area is \(1\). The incorrect calculation has used a circular radial boundary instead of the square's boundary. The correct polar description of the square needs two angular intervals: \(0\leq r\leq\sec\theta\) up to \(\theta=\pi/4\), followed by \(0\leq r\leq\csc\theta\) from \(\pi/4\) to \(\pi/2\). The lesson is not to avoid polar coordinates, but to verify that their bounds reconstruct the intended region.
A Short Justification for the Polar Area Factor
For a small change \(dr\) in radius and a small change \(d\theta\) in angle, a polar coordinate patch has radial side approximately \(dr\) and angular side approximately \(r\,d\theta\). Its area is therefore approximately \(r\,dr\,d\theta\). The Jacobian makes this local-area argument exact in the change-of-variables formula. Indeed,
Thus the absolute Jacobian is \(r\) for \(r\geq0\), which gives \(dA=r\,dr\,d\theta\). This derivation also explains why the factor depends on distance from the origin: equal angular changes sweep out longer arcs at larger radii.
How to Decide Efficiently
A useful choice can usually be made by comparing a few concrete features before doing any integration.
Identify whether its boundaries are simpler as Cartesian bounds or as radial and angular limits.
Substitute \(x=r\cos\theta\), \(y=r\sin\theta\) and include the factor \(r\). Check whether the new expression is actually simpler.
Prefer the setup with manageable bounds and integrands, while accounting for any necessary splits.
Reconstruct the region from the limits and confirm that it is covered once, without omitted parts or extra area.
Polar coordinates are particularly valuable when they simplify the region and the integrand at the same time. Cartesian coordinates remain preferable when the region is already rectangular or has simple nested bounds and the integrand is easy to integrate there. The decision is a comparison of complete setups, not a rule that one coordinate system is generally superior.
Check Your Understanding
For each question, consider both the description of the region and the form of the integrand.
- What are the two main features to compare when choosing between Cartesian and polar coordinates?
- Why does a polar double integral include a factor of \(r\) in its area element?
- For the unit square in the first quadrant, why does the polar radial bound change at \(\theta=\pi/4\)?
- What do \(x^2+y^2\) and the area element become after converting to polar coordinates?
- Why can a circular boundary alone be insufficient reason to choose polar coordinates?