Orbital Elements and Parameterization

Somewhere above your head right now, tens of thousands of objects — working satellites, spent rocket stages, flecks of debris — are being tracked, and every single one of them is filed in the world's catalogues as just six numbers. Not a position and not a velocity: six orbital elements, a description so compact and so physically transparent that a two-line string of text is enough to tell a ground station where to point its antenna next Tuesday. Those six numbers are the subject of this article. Why six? The two-body problem solved in the Kepler problem is a system of six first-order equations (three positions, three velocities of the reduced particle), so its general solution carries six constants of integration. The raw initial conditions $(\mathbf{r}_0,\mathbf{v}_0)$ are a legal choice but a miserable one — no component of $\mathbf{v}_0$ tells you at a glance whether the orbit is big or small, tilted or flat, circular or a long cigar. The classical elements repackage the same information along physical seams: two numbers $(a,e)$ for the size and shape of the conic, three angles $(i,\Omega,\omega)$ that orient it in space like the Euler angles of a rigid body, and one anomaly ( $\nu$ , or its uniform-clock cousin $M$ ) that says where on the ellipse the body is now. The deeper beauty is that each element is a conserved quantity wearing geometric clothing: $a$ is the energy, $e$ and the perihelion direction are the Laplace–Runge–Lenz vector, and $(i,\Omega)$ are the direction of $\mathbf{L}$ . Master this dictionary and you can read a satellite's whole life from one line of a catalogue — and see immediately what a perturbation will and will not touch.

Key takeaways

A three-dimensional Kepler orbit needs exactly six constants; the classical elements split them as $(a,e)$ = size and shape, $(i,\Omega,\omega)$ = orientation, and one anomaly = position along the orbit.
The size is pure energy, $a=-k/(2E)$ ; the shape needs both conserved scalars, $e=\sqrt{1+2EL^2/(\mu k^2)}$ .
The orientation angles are read off conserved vectors: $\cos i=L_z/L$ , the node vector $\mathbf{n}=\hat{\mathbf{z}}\times\mathbf{L}$ fixes $\Omega$ , and the eccentricity vector $\hat{\mathbf{e}}$ (the Laplace–Runge–Lenz direction) fixes $\omega$ .
Position on the orbit is quoted as the mean anomaly $M$ , which grows linearly in time; the true anomaly $\nu$ is recovered through Kepler's equation.
The elements degenerate gracefully: circular orbits lose $\omega$ , equatorial orbits lose $\Omega$ , and standard replacement angles exist for each case.
Key idea: six constants of integration of the two-body problem are repackaged as $(a,e)$ for geometry, $(i,\Omega,\omega)$ for orientation, and one anomaly for position — turning the abstract phase-space state into a geometrically meaningful parameterization tied directly to $E$ , $L$ , and the Runge–Lenz vector.

Where this lives today

Every satellite you have ever used — GPS, weather, broadband — is bookkept as a set of these six numbers. The public catalogue maintained by the US Space Force publishes each tracked object as a two-line element set (TLE): a compact text format carrying $i$ , $\Omega$ , $e$ , $\omega$ , $M$ and the mean motion $n$ (from which $a$ follows via Kepler's third law) at a reference epoch. With tens of thousands of catalogued objects and mega-constellations still growing through 2026, conjunction screening — the daily, automated search for close approaches between every pair of objects — runs by propagating exactly these elements forward and flagging encounters, and operators receive warnings phrased in element language: which node, what relative inclination, how the along-track (mean-anomaly) uncertainty dominates the miss distance. The same dictionary powers discovery science. When astronomers announce that an exoplanet has $e=0.3$ or that an interstellar comet arrived on a hyperbolic orbit with $e>1$ , they are quoting fitted orbital elements; when a mission is designed to rendezvous with an asteroid, the target is a six-element row in the JPL small-body database, and the transfer is a controlled rewriting of the spacecraft's own elements, one burn at a time.

The physical picture

Build an orbit the way you would stage a photograph. First draw an ellipse on a flat sheet of paper: pick how big ( $a$ ) and how stretched ( $e$ ). Then hold the sheet in space: tilt it out of the reference plane by an angle $i$ , and swivel the tilted sheet about the vertical so that the line where the two planes cross — the line of nodes — points in a chosen direction $\Omega$ . The paper can still spin within its own plane, so rotate it until the ellipse's near point, the periapsis, sits at angle $\omega$ from the node line. Finally place the planet somewhere on the curve and start the clock: that position is the anomaly. Six decisions, six numbers, one completely specified orbit:

Figure

The six orbital elements — two planes, one node line, three angles, and a clock

Read the figure as three nested pointing instructions. The grey sheet is the agreed reference plane — Earth's equator for satellites, the ecliptic for planets — with one agreed direction in it (toward the vernal equinox). The blue sheet is the orbital plane; it meets the grey one along the dashed line of nodes, and the ascending node (green) is the crossing where the body climbs from below the reference plane to above it. Then $\Omega$ swings you from the reference direction to that node within the grey plane, $i$ tips you up onto the blue plane, and $\omega$ swings you from the node to periapsis within the blue plane. From periapsis, the true anomaly $\nu$ finds the body itself. Notice which arrows are secretly conservation laws. The blue arrow $\mathbf{L}$ is the conserved angular momentum: its direction is the whole content of $(i,\Omega)$ , because a vector normal to the orbital plane needs exactly two angles to specify. The red periapsis marker is the conserved Laplace–Runge–Lenz direction — the frozen arrow studied in detail by the sibling article — and $\omega$ just records where that arrow points relative to the node. In the pure Kepler problem every one of these arrows is nailed in place, which is precisely why five of the six elements are constants and only the anomaly moves. Within the orbital plane itself, the geometry looks like this:

Figure

In-plane anatomy — the ellipse the elements describe, viewed face-on

Both angle markers pivot on the focus, not the centre: $\omega$ opens from the node line to periapsis, and $\nu$ opens from periapsis to the body. Everything an orbit "does" — where it comes closest, where it crosses the reference plane, where the body is — is measured from the point where the physics lives, the centre of attraction.

Size and shape from the conserved scalars

Start inside the orbital plane, where the Kepler problem has already done the heavy lifting. The bound orbit is a focal conic, and quoting it in terms of the true anomaly $\nu$ (the polar angle from periapsis) makes the first two elements appear as the coefficients of the curve itself:

r=\frac{p}{1+e\cos\nu},\qquad p=a(1-e^2)=\frac{L^2}{\mu k} \qquad (1)

where: $r$ — distance from the focus (primary) to the body; $p$ — the semi-latus rectum, the radius at $\nu=90°$ ; $e$ — eccentricity; $\nu$ — true anomaly; $a$ — semi-major axis; $L$ — angular momentum of the reduced particle of mass $\mu$ ; $k$ — force constant of $V=-k/r$ , equal to $Gm_1m_2$ for gravity. The two purely geometric elements are sitting in this one formula: $a$ sets the overall scale, $e$ the departure from a circle, and periapsis and apoapsis fall out as $r_{\min}=a(1-e)$ and $r_{\max}=a(1+e)$ . But geometry is bookkeeping; the elements earn their keep because they are the conserved quantities in disguise. The dictionary between $(a,e)$ and the dynamical invariants $(E,L)$ reads:

E=-\frac{k}{2a},\qquad e=\sqrt{1+\frac{2EL^2}{\mu k^2}} \qquad (2)

where: $E$ — total mechanical energy of the relative motion (negative for bound orbits); the other symbols as in (1). Read it asymmetrically, because it is asymmetric: the size $a$ is fixed by the energy alone — pump energy into an orbit and it grows, no matter what happens to the angular momentum — while the shape $e$ needs both scalars. Two orbits with equal $a$ but different $L$ have identical energies and identical periods and look completely different: one nearly circular, one a thin cigar. The remaining, orientation-flavoured piece of the in-plane geometry is which way the ellipse points, and that is owned by a conserved vector rather than a scalar. The Laplace–Runge–Lenz vector — built and proven constant in its own article — points from the focus to periapsis, and normalizing it gives the dimensionless eccentricity vector:

\hat{\mathbf{e}}=\frac{\mathbf{A}}{\mu k e},\qquad \mathbf{A}=\mathbf{p}\times\mathbf{L}-\mu k\,\hat{\mathbf{r}},\qquad |\mathbf{A}|=\mu k e \qquad (3)

where: $\mathbf{A}$ — the Laplace–Runge–Lenz vector; $\mathbf{p}$ — momentum of the reduced particle; $\hat{\mathbf{r}}$ — radial unit vector; $\hat{\mathbf{e}}$ — unit vector toward periapsis. For the elements, $\mathbf{A}$ is the anchor of the angle $\omega$ : an orbit "points" where its periapsis is, and $\omega$ simply measures that direction from the node line.

Derivation ASize and Shape from Energy and Angular Momentum

The goal is the dictionary (2): the semi-major axis from the energy alone, and the eccentricity from energy plus angular momentum. The strategy is to stand at the two turning points of the radial motion, periapsis and apoapsis, where the radial velocity vanishes and the conservation-of-energy statement degenerates into pure algebra — a quadratic equation whose two roots are the two apsidal distances. Vieta's formulas then hand us the sum and product of the roots without ever solving for them individually, and matching against the geometric statements $r_{\min}+r_{\max}=2a$ and $r_{\min}r_{\max}=a^2(1-e^2)$ closes the loop. Watch the signs: for a bound orbit $E<0$ , so expressions like $-k/E$ are positive exactly as a length must be.

Step 1. Write the radial energy equation at a turning point. Where $\dot r=0$ the kinetic energy is purely angular, so $E=\dfrac{L^2}{2\mu r^2}-\dfrac{k}{r}$ . Why: periapsis and apoapsis are by definition the extremes of $r$ , so the radial velocity passes through zero there — the energy is all centrifugal plus potential.

Step 2. Multiply through by $r^2/E$ and rearrange into a quadratic in $r$ : $r^2+\dfrac{k}{E}\,r-\dfrac{L^2}{2\mu E}=0$ . Why: the turning radii are now the two roots of a polynomial — the calculus is finished and only algebra remains.

Step 3. Apply Vieta's formulas. The sum of the roots is $r_{\min}+r_{\max}=-k/E$ and their product is $r_{\min}r_{\max}=-L^2/(2\mu E)$ . Why: sum and product of a quadratic's roots are read directly off its coefficients, no square roots needed.

Step 4. Match the sum against geometry. An ellipse's two apsidal distances always add to the full major axis, $r_{\min}+r_{\max}=2a$ , so $2a=-k/E$ , giving $a=-\dfrac{k}{2E}$ . Why: this is the cleanest possible statement that size is energy — angular momentum never entered the sum.

Step 5. Match the product against geometry. With $r_{\min}=a(1-e)$ and $r_{\max}=a(1+e)$ , the product is $a^2(1-e^2)=-L^2/(2\mu E)$ ; substituting $a=-k/(2E)$ and solving for $e$ gives $e^2=1+\dfrac{2EL^2}{\mu k^2}$ . Why: the product carries the shape information because it compares how deep the orbit dives against how far it swings out — that balance is what $L$ controls. *Verified: identical to the turning-point analysis in Goldstein §3.7 and Landau–Lifshitz §15; at the circular-orbit energy $E=-\mu k^2/(2L^2)$ the formula gives $e=0$ exactly, as it must.*

Orientation: three angles from two conserved vectors

Now leave the plane. An orbit in three dimensions is the in-plane ellipse plus a rigid-body orientation, and orientations in three dimensions take exactly three angles — the same count Euler established for spinning tops. The elements choose those three angles so that each one is directly readable from a conserved vector. The tilt comes first. The angular momentum $\mathbf{L}$ is normal to the orbital plane, so the angle between $\mathbf{L}$ and the reference pole $\hat{\mathbf{z}}$ is the angle between the two planes:

\cos i=\frac{L_z}{L},\qquad \mathbf{n}=\hat{\mathbf{z}}\times\mathbf{L} \qquad (4)

where: $i$ — inclination, ranging from $0°$ (orbit in the reference plane, prograde) through $90°$ (polar) to $180°$ (equatorial but retrograde); $L_z$ — component of $\mathbf{L}$ along the reference pole; $\mathbf{n}$ — the node vector, which by construction is perpendicular to both $\hat{\mathbf{z}}$ and $\mathbf{L}$ and therefore lies along the intersection of the two planes, pointing at the ascending node. Note the built-in convention: $i>90°$ means $L_z<0$ , a body circling the "wrong way" — no separate flag is needed for retrograde motion. The node vector then anchors the two swivel angles. The longitude of the ascending node $\Omega$ locates $\mathbf{n}$ within the reference plane, measured from the reference direction $\hat{\mathbf{x}}$ ; the argument of periapsis $\omega$ locates the eccentricity vector within the orbital plane, measured from $\mathbf{n}$ :

\cos\Omega=\frac{n_x}{|\mathbf{n}|},\qquad \cos\omega=\frac{\mathbf{n}\cdot\hat{\mathbf{e}}}{|\mathbf{n}|} \qquad (5)

where: $n_x$ — component of the node vector along the reference direction; $\hat{\mathbf{e}}$ — unit eccentricity vector of (3). Each arccosine is ambiguous between an angle and its mirror, so a quadrant check rides along: if $n_y<0$ the node sits in the far half-plane and $\Omega\to360°-\Omega$ ; if $e_z<0$ periapsis lies below the reference plane and $\omega\to360°-\omega$ . Forgetting these two checks is the single most common bug in home-rolled orbit code. Assembling the pieces runs the three rotations in sequence — first swing to the node, then tilt, then swing to periapsis:

\mathbf{r}=R_z(\Omega)\,R_x(i)\,R_z(\omega)\,\mathbf{r}_{\text{PQW}} \qquad (6)

where: $R_z,R_x$ — rotation matrices about the $z$ - and $x$ -axes; $\mathbf{r}_{\text{PQW}}$ — the position in the perifocal frame, the natural in-plane frame whose first axis points at periapsis and whose third axis is $\hat{\mathbf{L}}$ . This is exactly a 3-1-3 Euler-angle sequence: the orbit's orientation in space is a rigid-body attitude, and $(\Omega, i, \omega)$ are its Euler angles.

Figure

Three rotations carry the reference frame onto the orbit — a 3-1-3 Euler sequence

The clock: locating the body in time

Five elements freeze the geometry; the sixth must move. The honest position angle is the true anomaly $\nu$ of equation (1), but $\nu$ advances non-uniformly — Kepler's second law makes the body sprint through periapsis and dawdle at apoapsis, so $\nu(t)$ is a lousy thing to tabulate. Celestial mechanics therefore invents a fictitious uniform angle, the mean anomaly, which is nothing but a rescaled clock reading:

M=n\,(t-t_p),\qquad n=\frac{2\pi}{T}=\sqrt{\frac{k}{\mu a^3}} \qquad (7)

where: $M$ — mean anomaly, advancing at a perfectly constant rate; $n$ — the mean motion, the orbit's average angular speed; $t_p$ — the epoch of periapsis passage; $T$ — the period. Because $M$ is linear in time, it is the element catalogues actually store: to move an orbit forward a week, add $n\times(\text{one week})$ to $M$ and you are done. The price is a translation step. Converting the uniform clock $M$ into the physical angle $\nu$ passes through an intermediate, the eccentric anomaly $E_a$ , and requires solving the transcendental relation $M=E_a-e\sin E_a$ — Kepler's equation. That machinery (the geometry of $E_a$ , the derivation, and how to actually solve it numerically) is the entire subject of Kepler's equation and time of flight; here we only need to know which angle to reach for when:

you are asked for	reach for
the body's actual direction from the focus	true anomaly $\nu$ , via (1)
the radius at a known position angle	$r=p/(1+e\cos\nu)$
position at a given clock time	mean anomaly $M=n(t-t_p)$ , then Kepler's equation
an angle that is exactly linear in time	$M$ — and only $M$
the geometric middleman between the two	eccentric anomaly $E_a$

Go deeper · Derivation B · 2 minReading the Orientation Angles off the Conserved Vectors

Derivation BReading the Orientation Angles off the Conserved Vectors

The claim in equations (4) and (5) is that the three orientation elements are pure vector geometry: given the conserved $\mathbf{L}$ and $\mathbf{A}$ , all of $(i,\Omega,\omega)$ follow from dot and cross products, with no reference to the trajectory at all. We prove this by construction, building each angle from the definition of the planes involved. The strategy is to characterize each plane by its normal vector, get the intersection line as a cross product, and then use the dot-product definition of angle. The care point throughout is quadrant ambiguity — $\arccos$ returns values in the upper half-range only, so each angle needs one extra sign to be pinned down.

Step 1. Establish that $\mathbf{L}$ is normal to the orbital plane. Both $\mathbf{r}$ and $\mathbf{p}$ lie in the plane of motion, and $\mathbf{L}=\mathbf{r}\times\mathbf{p}$ is perpendicular to each, so $\hat{\mathbf{L}}$ is the plane's normal. Why: a plane is completely characterized by its unit normal — two angles of information, which is exactly what $(i,\Omega)$ will encode.

Step 2. Read the inclination as the angle between normals. The reference plane has normal $\hat{\mathbf{z}}$ ; the dihedral angle between two planes equals the angle between their normals, so $\cos i=\hat{\mathbf{z}}\cdot\hat{\mathbf{L}}=L_z/L$ . Why: no quadrant check is needed here — $i$ is defined on $[0°,180°]$ , precisely the range of $\arccos$ , with $i>90°$ meaning retrograde.

Step 3. Construct the line of nodes. The intersection of the two planes is perpendicular to both normals, so $\mathbf{n}=\hat{\mathbf{z}}\times\mathbf{L}$ lies along it. Why: the cross product of the normals is the one direction contained in both planes — and the sign convention of the cross product makes $\mathbf{n}$ point at the ascending node specifically, where the body crosses moving upward.

Step 4. Measure $\Omega$ in the reference plane. Since $\mathbf{n}$ lies in the reference plane, its polar angle from $\hat{\mathbf{x}}$ is $\cos\Omega=n_x/|\mathbf{n}|$ , and the ambiguity is resolved by the other in-plane component: $n_y<0$ places the node in the lower half-plane, so $\Omega\to360°-\Omega$ . Why: an angle in a plane needs two coordinates to fix; the arccosine only consumed one.

Step 5. Measure $\omega$ in the orbital plane. Both $\mathbf{n}$ and the eccentricity vector $\hat{\mathbf{e}}$ lie in the orbital plane, so the angle between them is $\cos\omega=\mathbf{n}\cdot\hat{\mathbf{e}}/|\mathbf{n}|$ , with the quadrant fixed by whether periapsis sits above or below the reference plane: $e_z<0$ means $\omega\to360°-\omega$ . Why: $\omega$ must be measured from the node in the direction of motion, and the sign of $e_z$ is what distinguishes the first half of the lap from the second. *Verified: this is the standard construction of Goldstein §3.7–3.9 and Murray & Dermott Ch. 2; for an orbit lying exactly in the reference plane ( $\mathbf{L}\parallel\hat{\mathbf{z}}$ ) the construction correctly degenerates — $\mathbf{n}=\mathbf{0}$ and $\Omega$ becomes undefined, as the limiting-cases table demands.*

Go deeper · Derivation C · 2 minFrom a State Vector to the Six Elements

Derivation CFrom a State Vector to the Six Elements

This is the workhorse algorithm of astrodynamics: a tracking station hands you one simultaneous position $\mathbf{r}$ and velocity $\mathbf{v}$ , and you must return the six elements. The strategy is to spend the state vector on the three conserved objects — energy, angular momentum, and eccentricity vector — and then harvest the elements from them using the two derivations above. It is worth internalizing the flow: state $\to$ invariants $\to$ elements; nothing is ever integrated. Two conventions to fix before starting: work per unit mass (so define the gravitational parameter $\kappa\equiv k/\mu=G(m_1+m_2)$ and the specific angular momentum $\mathbf{h}=\mathbf{r}\times\mathbf{v}$ ), and keep the quadrant checks from the previous derivation in force.

Step 1. Form the specific angular momentum $\mathbf{h}=\mathbf{r}\times\mathbf{v}$ . Its direction is the orbit normal and its magnitude gives the semi-latus rectum $p=h^2/\kappa$ . Why: one cross product captures both the plane and the "width" of the conic.

Step 2. Get the size from the vis-viva energy. The specific energy is $\varepsilon=\tfrac12 v^2-\kappa/r$ , and $a=-\kappa/(2\varepsilon)$ — equivalently, solve $v^2=\kappa(2/r-1/a)$ for $a$ . Why: this is equation (2) per unit mass; a negative computed $a$ is not an error but the signature of a hyperbolic orbit.

Step 3. Build the eccentricity vector, $\mathbf{e}=\dfrac{\mathbf{v}\times\mathbf{h}}{\kappa}-\hat{\mathbf{r}}$ . Its magnitude is the eccentricity $e$ and its direction marks periapsis. Why: this is the Laplace–Runge–Lenz vector of (3) divided by $\mu k$ — the per-unit-mass disguise of the same conserved arrow.

Step 4. Read the inclination from $\cos i=h_z/h$ . Why: Step 2 of the previous derivation, verbatim.

Step 5. Form the node vector $\mathbf{n}=\hat{\mathbf{z}}\times\mathbf{h}$ and extract $\Omega$ from $\cos\Omega=n_x/|\mathbf{n}|$ (flip to $360°-\Omega$ if $n_y<0$ ), then $\omega$ from $\cos\omega=\mathbf{n}\cdot\mathbf{e}/(|\mathbf{n}|\,e)$ (flip if $e_z<0$ ). Why: the orientation harvest of the previous derivation, now with concrete numbers flowing through it.

Step 6. Locate the body: $\cos\nu=\mathbf{e}\cdot\mathbf{r}/(e\,r)$ , flipping to $360°-\nu$ if $\mathbf{r}\cdot\mathbf{v}<0$ . Why: the sign of $\mathbf{r}\cdot\mathbf{v}$ says whether the radius is currently growing (outbound from periapsis, $\nu<180°$ ) or shrinking (inbound, $\nu>180°$ ) — the cleanest possible quadrant test. *Verified: reproduces the standard conversion algorithm of Murray & Dermott Ch. 2 and Goldstein §3.9; the worked example below pushes real numbers through every step and lands on a self-consistent element set.*

Limiting cases

The elements are a coordinate chart on the space of orbits, and like most charts it has coordinate singularities — places where perfectly good orbits make individual elements meaningless:

Case	Result	Meaning
Circular orbit, $e\to0$	$\omega$ and $\nu$ individually undefined	No periapsis to measure from; use the argument of latitude $u=\omega+\nu$ , the angle from node to body, which stays finite.
Equatorial orbit, $i\to0$ or $180°$	$\mathbf{n}\to\mathbf{0}$ , $\Omega$ undefined	No node line when the planes coincide; use the longitude of periapsis $\varpi=\Omega+\omega$ .
Circular and equatorial	only $a$ and the position survive	Both degeneracies at once; use the true longitude $l=\Omega+\omega+\nu$ , measured straight from the reference direction.
Parabolic limit, $e\to1$	$a\to\infty$	Energy is zero, so $a=-k/(2E)$ diverges; the periapsis distance $q=a(1-e)=p/2$ takes over as the size element.
Hyperbolic orbit, $e>1$	$a<0$ , formulas carry over	The bound-orbit algebra survives with negative $a$ ; the anomaly machinery goes hyperbolic, $M=e\sinh F - F$ .

Common trapThe mean anomaly points at the satellite

It does not — and this misconception quietly wrecks more first orbit-propagation codes than any other. Of the three anomalies, only the true anomaly $\nu$ is a physical direction you could point a telescope along. The eccentric anomaly $E_a$ is measured at the centre of the ellipse (not the focus) and refers to a projected point on the auxiliary circle, not to the body. The mean anomaly $M$ is purest fiction: a uniformly spinning clock hand defined so that bookkeeping is linear in time, corresponding to no geometric point at all. The three agree only at periapsis and apoapsis. If you compute $M$ at some epoch and feed it into $r=p/(1+e\cos M)$ as if it were $\nu$ , you get positions that are wrong by up to $2e$ radians of phase — for a Molniya-class orbit ( $e\approx0.74$ ), roughly $85°$ of "where is my satellite". Always run the chain $M\to E_a\to\nu$ before touching geometry.

Worked examples

1 · A GTO in elements (confidence builder)

A launcher releases a spacecraft into a geostationary transfer orbit with perigee altitude $250\ \mathrm{km}$ and apogee altitude $35{,}786\ \mathrm{km}$ . Using $R_\oplus=6378\ \mathrm{km}$ and $\kappa=GM_\oplus=3.986\times10^{14}\ \mathrm{m^3/s^2}$ , find the in-plane elements $a$ , $e$ , $p$ and the period.

Step 1. Convert altitudes to radii from Earth's centre: $r_p=6378+250=6628\ \mathrm{km}$ and $r_a=6378+35{,}786=42{,}164\ \mathrm{km}$ .

Step 2. The apsidal distances always add to the major axis: $a=\tfrac12(r_p+r_a)=\tfrac12(6628+42{,}164)=\boxed{24{,}396\ \mathrm{km}}$ .

Step 3. The eccentricity measures the asymmetry of the two apsides: $e=\dfrac{r_a-r_p}{r_a+r_p}=\dfrac{42{,}164-6628}{42{,}164+6628}=\dfrac{35{,}536}{48{,}792}=\boxed{0.728}$ .

Step 4. The semi-latus rectum follows from the shape pair: $p=a(1-e^2)=24{,}396\times(1-0.728^2)=\boxed{1.146\times10^{4}\ \mathrm{km}}$ .

Step 5. Kepler's third law gives the clock rate: $T=2\pi\sqrt{a^3/\kappa}=2\pi\sqrt{(2.4396\times10^{7})^3/3.986\times10^{14}}=3.79\times10^{4}\ \mathrm{s}=\boxed{10.5\ \mathrm{hours}}$ — the familiar half-day-ish GTO coast. Note what was not needed: nothing about tilt or node. The in-plane elements decouple completely from the orientation ones — that separation is the whole design of the element set.

2 · State vector to elements, end to end (exam level)

A tracking station reports, in an Earth-centred inertial frame, $\mathbf{r}=(6524.8,\ 6862.9,\ 6448.3)\ \mathrm{km}$ and $\mathbf{v}=(4.9013,\ 5.5338,\ -1.9763)\ \mathrm{km/s}$ . Find all six classical elements ( $\kappa=3.986\times10^{14}\ \mathrm{m^3/s^2}$ ).

Step 1. Magnitudes first: $r=\sqrt{6524.8^2+6862.9^2+6448.3^2}=11{,}456.6\ \mathrm{km}$ and $v=7.652\ \mathrm{km/s}$ .

Step 2. Specific energy for the size: $\varepsilon=\tfrac12v^2-\kappa/r=\tfrac12(7652)^2-\dfrac{3.986\times10^{14}}{1.14566\times10^{7}}=-5.52\times10^{6}\ \mathrm{J/kg}$ , so $a=-\kappa/(2\varepsilon)=\boxed{36{,}127\ \mathrm{km}}$ . Negative energy: the orbit is bound.

Step 3. Specific angular momentum: $\mathbf{h}=\mathbf{r}\times\mathbf{v}=(-4.925,\ 4.450,\ 0.247)\times10^{4}\ \mathrm{km^2/s}$ , with magnitude $h=6.642\times10^{4}\ \mathrm{km^2/s}$ .

Step 4. Eccentricity vector: $\mathbf{e}=(\mathbf{v}\times\mathbf{h})/\kappa-\hat{\mathbf{r}}=(-0.3146,\ -0.3852,\ 0.6680)$ , so $e=|\mathbf{e}|=\boxed{0.8329}$ — a long, thin ellipse.

Step 5. Inclination from the normal: $\cos i=h_z/h=0.247/6.642=0.0372$ , so $i=\boxed{87.9°}$ — a nearly polar orbit, its plane standing almost perpendicular to the equator.

Step 6. Node vector $\mathbf{n}=\hat{\mathbf{z}}\times\mathbf{h}=(-4.450,\ -4.925,\ 0)\times10^{4}$ . Then $\cos\Omega=n_x/|\mathbf{n}|=-0.6703$ gives $132.1°$ ; but $n_y<0$ puts the node in the lower half-plane, so $\Omega=360°-132.1°=\boxed{227.9°}$ .

Step 7. Argument of periapsis: $\cos\omega=\mathbf{n}\cdot\mathbf{e}/(|\mathbf{n}|\,e)=0.5966$ gives $53.4°$ , and since $e_z=0.668>0$ periapsis is above the equator, so no flip: $\omega=\boxed{53.4°}$ .

Step 8. True anomaly: $\cos\nu=\mathbf{e}\cdot\mathbf{r}/(e\,r)=-0.0408$ gives $92.3°$ ; the quadrant test $\mathbf{r}\cdot\mathbf{v}=+5.72\times10^{10}\ \mathrm{m^2/s}>0$ confirms the body is outbound, so $\nu=\boxed{92.3°}$ — just past periapsis, radius still growing. A final sanity check knits the set together: $p=h^2/\kappa=11{,}068\ \mathrm{km}$ should equal $a(1-e^2)=36{,}127\times(1-0.8329^2)=11{,}068\ \mathrm{km}$ ✓.

Origins · 1 minLeonhard Euler and how this was discovered

Leonhard Euler

1707–1783 · St Petersburg and Berlin

Euler owns the mathematics of this article twice over. He created the theory of rotations — the theorem that any orientation of a rigid body is reached by three successive angles bears his name, and the sequence $R_z(\Omega)R_x(i)R_z(\omega)$ that assembles an orbit in space is a textbook 3-1-3 Euler rotation. And he spent decades on celestial mechanics itself: his lunar theories were the first grand assault on the Moon's wandering elements, his methods drove the eighteenth-century tables that navigators used to find longitude at sea, and his perturbation techniques — letting the "constants" of an orbit drift slowly — grew into the osculating-element machinery that still runs inside every modern orbit propagator. He did much of this while losing his sight, computing, it was said, as other men breathe.

History

1609

— Kepler's Astronomia Nova establishes the ellipse-with-focus geometry that the elements parameterize.

1687

— Newton's Principia derives conic orbits from the inverse-square law, making the six-constant description exact.

1748

— Euler's lunar and planetary work pioneers treating orbital constants as slowly varying — the germ of osculating elements.

1801

— Gauss recovers the lost asteroid Ceres from a handful of observations, inventing modern orbit determination in element space.

1957

— Sputnik opens the satellite era; orbital elements become the operational currency of tracking catalogues worldwide.

Going further

References. Goldstein, Poole & Safko, Classical Mechanics, 3rd ed., §3.7–3.9 · Morin, Introduction to Classical Mechanics, Ch. 7 · Landau & Lifshitz, Mechanics, 3rd ed., §15 · Taylor, Classical Mechanics, Ch. 8 · Murray & Dermott, Solar System Dynamics, Ch. 2

Provenance. Element-to-conserved-quantity relations ( $a=-k/2E$ , $e=\sqrt{1+2EL^2/\mu k^2}$ , $p=L^2/\mu k$ ) verifiable from Goldstein §3.7–3.8 and Landau-Lifshitz §15. Kepler's equation derivation and anomaly relations from Murray & Dermott Ch. 2 and Taylor Ch. 8. Mercury's perihelion excess ( $\approx 43''$ /century) and orbital eccentricities are standard ephemeris values. A verifier should recompute the Vieta-formula derivation of $a$ and $e$ , confirm $M=E_a-e\sin E_a$ from the equal-area law, and check the half-angle $\nu$ – $E_a$ relation. Worked-example numbers recomputed from $\kappa=3.986\times10^{14}\ \mathrm{m^3/s^2}$ : GTO $a=24{,}396$ km, $e=0.728$ , $T=10.5$ h; state-vector case $a=36{,}127$ km, $e=0.8329$ , $i=87.9°$ , $\Omega=227.9°$ , $\omega=53.4°$ , $\nu=92.3°$ .

Builds on: The Kepler Problem · Conic-Section Orbits · Angular Momentum Conservation in Central Fields · Classification of Orbits by Energy Leads to: Kepler's Third Law: Period–Semimajor Relation · Kepler's Second Law: Equal Areas · Kepler's First Law: Elliptical Orbits

Why exactly six elements — why not three, or seven?

Count the degrees of freedom. The reduced two-body problem is one particle in three dimensions: three position components and three velocity components, so the state at any instant is six numbers, and any complete set of constants that labels a solution must also be six. Three is too few — it could fix the ellipse's size, shape and tilt but never both the node direction and the periapsis direction and the phase. Seven would be redundant: you might think $E$ , the three components of $\mathbf{L}$ and the three of $\mathbf{A}$ give seven constants, but two identities eat the excess ( $\mathbf{A}\cdot\mathbf{L}=0$ , and $|\mathbf{A}|$ is fixed by $E$ and $L$ ), leaving five independent constants plus one phase angle — six again, from the other direction.

If position and velocity also take six numbers, what do the elements actually buy?

Constancy and legibility. The state vector $(\mathbf{r},\mathbf{v})$ changes every instant, so tabulating it means tabulating a time series; five of the six elements are constants of the unperturbed motion, so one row of six numbers describes the orbit forever, with only $M$ advancing — and it advances linearly, which is trivial bookkeeping. Legibility matters just as much: $e=0.83$ tells you at a glance the orbit is a long ellipse, while no glance at $(6524.8, 6862.9, 6448.3)$ km tells you anything. And when perturbations act, each element responds to a specific physical cause — Earth's oblateness walks $\Omega$ and $\omega$ but barely touches $a$ — so element language is diagnostic in a way Cartesian coordinates never are.

Ascending node — ascending through what, exactly?

Through the reference plane, in the +z sense. An inclined orbit spends half its lap below the reference plane (negative $z$ ) and half above. It crosses the plane at exactly two points — the nodes — and the ascending one is where the body pierces the plane moving upward, from $z<0$ to $z>0$ . The convention is baked into the node vector: $\mathbf{n}=\hat{\mathbf{z}}\times\mathbf{L}$ points at the ascending node specifically, because the cross product's right-hand rule encodes the direction of motion. For Earth satellites the reference plane is the equator, so the ascending node is the equator crossing headed north — the quantity ground-track plots mark with a little upward arrow.

Why is the argument of periapsis measured from the node rather than from some fixed star?

Because the node is the only marked direction that lives in the orbital plane. An angle is only meaningful measured between two directions in the same plane; periapsis lies in the orbital plane, but a fixed celestial direction like the vernal equinox lies in the reference plane, and for a tilted orbit the two planes share exactly one line — the line of nodes. So the node is the natural hinge: $\Omega$ carries you along the reference plane to the hinge, $\omega$ carries you along the orbital plane from the hinge to periapsis. When the hinge itself disappears (equatorial orbit, planes coincide), the convention gracefully collapses to the sum $\varpi=\Omega+\omega$ , which is then measured from the equinox after all.

Do the elements of a real satellite actually stay constant?

No — and their drift is a feature, not a failure. The elements are exact constants only for the pure two-body problem. Real orbits feel Earth's oblateness, lunar and solar tugs, drag, radiation pressure; the fix is to define osculating elements, the elements of the ideal ellipse the body would follow if all perturbations vanished this instant. These drift slowly, and the drift pattern is a fingerprint of the perturbation: Earth's $J_2$ oblateness makes $\Omega$ regress steadily (exploited by sun-synchronous orbits, whose node deliberately precesses $360°$ per year), drag shrinks $a$ , and the pure-Kepler elements $e$ and $i$ wobble but hold secularly. That is why a TLE carries an epoch date — it is a snapshot of a slowly evolving element set, stale after days to weeks.

Can I recover the six elements from a single position measurement?

Not from position alone — a single $\mathbf{r}$ is three numbers, and six constants cannot hide in three measurements. The state-vector algorithm needs the simultaneous velocity too: it is $(\mathbf{r},\mathbf{v})$ together, six numbers, that map one-to-one onto the six elements. Historically this was the crux of the Ceres problem: telescopes measure only angles (two per night, no distance, no velocity), so Gauss needed observations on several nights plus the constraint that the path is a Kepler conic to pin all six constants — a harder inverse problem, and the birth of orbit determination. Radar and laser ranging changed the game precisely because they deliver range and range-rate, completing the state.

Is the inclination measured from the equator or the ecliptic, and does it matter?

It matters completely, and the answer is a convention you must check every time: the elements are defined relative to a declared reference plane, and different communities declare differently. Earth-satellite catalogues use the Earth's equator (so the ISS at $i=51.6°$ means $51.6°$ to the equator); solar-system ephemerides use the ecliptic (so Mercury's $i=7.0°$ is measured to Earth's orbital plane); galactic and exoplanet work uses the sky plane. The same physical orbit has different $(i,\Omega)$ in different frames — only the in-plane elements $a$ , $e$ and the anomalies are frame-independent. Quoting elements without naming the frame is like quoting coordinates without naming the map.

What is actually stored in a TLE, and why the mean anomaly instead of the true one?

A two-line element set carries the epoch, $i$ , $\Omega$ , $e$ , $\omega$ , $M$ and the mean motion $n$ (plus drag terms) — note that $a$ never appears explicitly, because $n$ and Kepler's third law encode it. The mean anomaly is stored precisely because it is the boring one: it advances at the constant rate $n$ , so propagating the catalogue forward by $\Delta t$ is a multiplication and an addition per object, done millions of times a day in conjunction screening. The interesting-but-nonlinear angles are reconstructed on demand: $M\to E_a$ through Kepler's equation, then $E_a\to\nu$ through the half-angle relation — the pipeline that the next article builds in full.

Orbital Elements and Parameterization

Where this lives today

The physical picture

Size and shape from the conserved scalars

Orientation: three angles from two conserved vectors

The clock: locating the body in time

Limiting cases

Worked examples

1 · A GTO in elements (confidence builder)

2 · State vector to elements, end to end (exam level)

Going further

Before this

Where this leads