Stereographic Projection

Why stereographic projection matters

Stereographic projection has been used since Hipparchus in the 2nd century BCE for celestial cartography. Its modern uses span complex analysis, where the Riemann sphere is the canonical compactification of ℂ; differential geometry, where it provides smooth coordinate atlases on spheres; physics, where the Bloch sphere of qubits embeds in the extended complex plane; and applied statistics, where directional data on a sphere are routinely pulled back to Euclidean space for analysis.

• Riemann sphere ℂ̂ = ℂ ∪ {∞}. Adding a single point at infinity to ℂ produces the simplest Riemann surface — topologically a 2-sphere, made conformal-equivalent to it by stereographic projection. Meromorphic functions on ℂ are exactly holomorphic maps to ℂ̂. The Möbius group PSL(2, ℂ) acts on ℂ̂ as the full group of conformal automorphisms.
• Cartography of polar regions. The Universal Polar Stereographic (UPS) grid covers latitudes above 84°N and below 80°S, where Mercator's projection blows up. UPS is conformal, so navigation directions are exact, but it distorts areas — a feature shared with all conformal maps of curved surfaces to the plane.
• Crystallography and pole figures. Crystal orientations are visualized on Wulff stereographic nets: each crystallographic direction is plotted as the image of its corresponding sphere point under stereographic projection. The conformal property means inter-axis angles are exactly preserved in the diagram.
• Directional statistics. The von Mises-Fisher and Watson distributions on S² are non-trivial to sample directly. Pulling back to ℝ² via stereographic projection gives planar distributions whose density transforms by the Jacobian; many estimators are easier in the planar domain.
• Bloch sphere of qubits. A pure qubit state ψ = α|0⟩ + β|1⟩ (up to phase) corresponds to a point on S². Stereographic projection from the north pole maps this to ℂ̂ via β/α; the resulting Möbius action describes single-qubit gates as conformal transformations of the extended complex plane.
• Conformal compactification in physics. Penrose's conformal compactification of Minkowski spacetime, the chordal metric on ℂ̂, and the Reissner-Nordström spatial sections all use stereographic-type maps. The Atiyah-Ward correspondence in twistor theory uses ℂP³ and ℂP¹ ≅ S² with stereographic structure.
• Hopf fibration and topology. The Hopf map S³ → S² (the simplest non-trivial sphere bundle) is best visualized by stereographic projection of S³ ⊂ ℝ⁴ into ℝ³: the fibers (great circles in S³) become a beautifully nested family of circles in ℝ³, demonstrating π₃(S²) ≠ 0 visually.
• Computational geometry. Mesh processing algorithms often need a single-chart parameterization of a 2-sphere; stereographic projection is the standard tool when the sphere is the target manifold, although it generally requires two charts (one from each pole) to cover everything.

Formulas and inverse

The 2-sphere case. Let S² = {(x, y, z) ∈ ℝ³ : x² + y² + z² = 1} and N = (0, 0, 1) the north pole. For P = (x, y, z) ∈ S² \ {N}, the line from N through P hits the equatorial plane z = 0 at

σ(P) = ( x / (1 − z), y / (1 − z) ) ∈ ℝ².

Inverse. Given (u, v) ∈ ℝ², the inverse map back to the sphere is

σ⁻¹(u, v) = ( 2u / (1 + u² + v²),  2v / (1 + u² + v²),  (u² + v² − 1) / (1 + u² + v²) ).

The north pole N corresponds to the point at infinity in ℝ². The unit circle in ℝ² is the image of the equator of S². Points inside the unit disc come from the southern hemisphere; points outside, from the northern hemisphere minus the north pole.

Pulled-back metric. Express the round metric on S² in stereographic coordinates (u, v):

ds²_{S²}  =  ( 4 / (1 + u² + v²)² ) · (du² + dv²).

The conformal factor 4/(1 + u² + v²)² is positive everywhere and varies pointwise — that is exactly what makes the map angle-preserving but not isometry. The Euclidean angle structure du² + dv² is preserved up to the scalar factor.

n-dimensional case. For S^n ⊂ ℝ^{n+1} with north pole N = (0, …, 0, 1), the projection from N is σ(x₁, …, x_n, t) = (x₁/(1 − t), …, x_n/(1 − t)) ∈ ℝ^n. The inverse and pulled-back metric have the same form as in the 2D case, replacing u² + v² with |u|² = u₁² + … + u_n².

Why conformality is exact

Conformality is the property that the differential dσ at every point is a scalar multiple of an isometry. Equivalently, the pull-back of the Euclidean metric to S² is a scalar multiple of the round metric — what we computed above.

Geometric proof. Consider two intersecting curves α, β on S² meeting at P with tangent vectors u, v. The angle between them is cos θ = (u · v)/(|u||v|). Under σ, the tangent vectors transform by the differential dσ, and we want to show the angle is preserved.

The key geometric fact: the differential dσ at P is conformal because the line from N to P meets S² at right angles to a small "compass" centered at P, and the projection from N preserves the relative angles between vectors tangent to the sphere at P. More concretely, the inversion in the sphere of radius √2 centered at N maps S² to its tangent plane at the south pole; inversions are conformal (a classical result), and stereographic projection differs from this inversion only by an affine map.

Analytic proof. Compute the Jacobian of σ directly: at a point (x, y, z) ∈ S², the tangent space is spanned by ∂_x and ∂_y (where ∂_y on the sphere means moving in the y-direction while staying on the sphere; this differs from the ambient ∂_y by a normal component). Computing dσ · ∂_x and dσ · ∂_y and taking inner products in ℝ² reveals the differential matrix is a scalar multiple of the identity rotation. The scalar is 1/(1 − z), whose square gives the conformal factor squared.

Costed claim. Stereographic angle preservation is exact at all points of S² \ {N} — there is no approximation involved. Unlike many cartographic projections (Mercator, Lambert) which preserve angles up to controlled distortion, stereographic projection literally pulls the round metric back to a positive scalar multiple of the Euclidean metric at every single point.

The circle property

Theorem. Stereographic projection maps circles on S² to circles or lines in ℝ². Specifically: circles on S² not passing through the projection pole map to circles in the plane; circles passing through the pole map to straight lines.

Proof sketch. A circle C on S² is the intersection of S² with a plane Π. The cone of lines through N meeting C is a quadric — specifically, a circular cone if N ∈ Π and a (generally) tilted cone if N ∉ Π. The cone intersects the equatorial plane in a conic section.

• If N ∈ Π: the cone degenerates because all lines through N lie in Π; the "cone" is just the plane Π itself, which meets the equatorial plane in a line.
• If N ∉ Π: the cone is non-degenerate. Its cross-section on the equatorial plane is a conic. Symmetry (the cone has a circular cross-section through C, parallel to which lies the equator's cross-section) forces this conic to be a circle.

The combination — circles through N → lines, circles not through N → circles — is exactly the statement "circles map to circles" in the extended sense where lines are "circles through ∞". This is the same equivalence used to define generalized circles in inversive geometry; in fact, stereographic projection conjugates Möbius transformations of ℂ̂ (which permute generalized circles) to rotations and inversions of S² (which also permute circles).

Projection from north vs south pole

Aspect	From north pole N = (0,0,1)	From south pole S = (0,0,−1)	Identification
Formula	(x, y, z) → (x/(1−z), y/(1−z))	(x, y, z) → (x/(1+z), y/(1+z))	Antipodal map composes them
Image of equator	Unit circle in ℝ²	Unit circle in ℝ²	Same circle
Southern hemisphere → disc	Open unit disc	Complement of closed disc (outside)	Inverted around unit circle
Northern hemisphere → ?	Outside unit circle	Inside open disc	Swapped roles
Excluded point → ∞	North pole N	South pole S	Different infinity points
Transition map (atlas)	z → 1/z̄ (orientation-reversing)	or z → 1/z (orientation-preserving, with chart flip)	Defines smooth structure on S²
Conformal factor	4/(1 + |u|²)²	4/(1 + |u|²)² (after pole swap)	Same form, different chart

The Riemann sphere

The Riemann sphere ℂ̂ = ℂ ∪ {∞} is the conformal-equivalent of S² under stereographic projection. Identify ℂ with the equatorial plane and S² \ {N} ↔ ℂ via σ. Send N to ∞.

Möbius transformations f(z) = (az + b)/(cz + d), with ad − bc ≠ 0, are the conformal automorphisms of ℂ̂. They form the group PSL(2, ℂ) = SL(2, ℂ) / {±I}, which acts triply transitively on ℂ̂. Composed with stereographic projection, Möbius transformations correspond to orientation-preserving conformal transformations of S² — specifically, rotations of S² for those Möbius transformations in PSU(2) = SU(2)/{±I}.

Holomorphic maps to ℂ̂ are exactly meromorphic functions on the source domain — a single object unifying poles, zeros, and ordinary values. The fundamental theorem of algebra has a slick proof using ℂ̂: a non-constant polynomial extends to a holomorphic map ℂ̂ → ℂ̂ of positive degree, and any such map is surjective; in particular, 0 is hit, giving a root.

Applications in statistics

• Directional data on a sphere. Wind directions, geological-fault orientations, antenna polarizations, animal-migration tracks — many real-world data live on S² or higher spheres. Pulling back to ℝ² via stereographic projection turns these into 2D distributions amenable to standard estimation. The Jacobian of the inverse map is 4/(1 + u² + v²)², which weights the density appropriately.
• von Mises-Fisher distribution. f(x) ∝ exp(κ μ · x) on S². Under stereographic projection, the density transforms to a complicated rational expression on ℝ²; maximum-likelihood estimation can still be done numerically, sometimes more easily in the planar than the spherical formulation.
• Watson distribution. f(x) ∝ exp(κ (μ · x)²), an antipodal-symmetric distribution on S². Useful for axial data (orientations modulo direction reversal). Stereographic projection makes the analysis algorithmically convenient when κ is moderate.
• Geodesic and Fréchet means. Computing the mean direction of spherical data requires either iterative algorithms on S² or pull-back to a tangent plane via stereographic/exponential maps. Newton iterations converge faster in the planar setting due to the conformal coordinate structure.

Common pitfalls

• "Stereographic projection preserves areas." It does not. Areas are scaled by the square of the conformal factor: (4/(1 + |u|²)²)². The southern hemisphere fits inside the unit disc but expands as it stretches toward the equator; small caps near the south pole map to small discs near the origin, while caps near the equator map to nearly-equal-area annuli. Only angles are preserved, not areas.
• "It maps the sphere onto the plane." Not the whole sphere — the pole used for projection is excluded. The sphere needs two charts (one from each pole) for a full atlas. The pole maps "to infinity," which is a way of saying it does not map anywhere in the plane.
• "Only works for the unit sphere." Works for any 2-sphere; for a sphere of radius R, replace 1 by R in the denominator and rescale appropriately. The conformal property is intrinsic and survives rescaling.
• "Conformal means distance-preserving." No — conformal means angle-preserving. Distances are scaled by the conformal factor and vary point-to-point. An isometry would preserve both angles and distances; conformal maps preserve only angles.
• "Lines in ℝ² come from great circles." The lines in ℝ² come from circles on S² passing through the projection pole; these are not in general great circles. The great-circle through the south pole and any other latitude does map to a straight line, but circles passing through the north pole at any inclination also map to lines.
• "Stereographic projection is a homeomorphism S² → ℝ²." No — it is a homeomorphism S² \ {N} → ℝ². To get a homeomorphism, you compactify ℝ² with a point at infinity, recovering S² as ℝ² ∪ {∞}. This is the topological content of "Riemann sphere = one-point compactification of ℂ".

History

The earliest known use of stereographic projection is by Hipparchus of Nicaea (c. 150 BCE) in his star catalog, where he projected the celestial sphere onto a planar disc to make navigational and astronomical tables. The Almagest of Ptolemy (c. 150 CE) systematized the construction. The Arabic astrolabe (al-Bīrūnī, al-Tūsī) is a physical instrument implementing stereographic projection of celestial coordinates.

The modern mathematical treatment dates to the 17th century. The conformal property was identified by Edmond Halley (1696) and rigorously proved by Bernhard Riemann in the 1850s, who introduced the Riemann sphere as a fundamental object of complex analysis. The connection to Möbius transformations and the conformal-equivalence with ℂ̂ was developed by August Möbius and Felix Klein in the late 19th century.

In the 20th century, stereographic projection became a standard tool of differential geometry (Cartan, Klingenberg), complex analysis (the modern theory of meromorphic functions and Riemann surfaces), and physics (Penrose's conformal compactification, the Bloch sphere of two-level quantum systems). Cartographic applications continue: UPS for polar regions, USGS Astronomy uses stereographic projection for planetary maps, and IPCC climate datasets use stereographic grids for polar sea-ice analysis.