---
title: Keplerian Orbits
module: Gravitation and Matter
moduleNumber: 6
lessonNumber: 1
order: 601
summary: >
  Why do the planets trace ellipses rather than any other curve? Newton's
  inverse-square law collapses the two-body problem onto a single conic section,
  and the answer falls out of two conserved quantities: a central force can exert
  no torque, so angular momentum is fixed, and gravity is conservative, so energy
  is fixed. We read an orbit's size and shape straight off those invariants,
  recover all three of Kepler's laws, and derive escape speed, the vis-viva
  relation, and the timing of a pass. We also mark where the ideal ellipse breaks
  down — drag, oblateness, and a third body slowly move a real orbit.
topics: [Gravitation and Matter]
draft: false
sources:
  - book: Tipler & Mosca
    ref: "Ch. 11 — Gravity; §§11-1–11-3"
---

## Kepler's laws and Newtonian orbits

Planetary motion about a dominant spherical mass obeys three Kepler laws:

- **Orbit law:** the orbit is an ellipse with the central body at one focus.
- **Area law:** the radius line sweeps equal areas in equal times.
- **Period law:** $T^2/a^3$ has one value for every orbit around the same central
  mass, where $T$ is period and $a$ is semimajor axis.

A circular orbit is the special ellipse with eccentricity $e=0$ and $a=r$.

An ellipse has nearest and farthest radii

$$
r_p=a(1-e),\qquad r_a=a(1+e),\qquad 0\le e<1.
$$

The body travels fastest at periapsis and slowest at apoapsis. Equal-area motion
is the geometric signature of angular-momentum conservation, not an independent
force law. A planet sweeps the same area in a fixed time by moving more rapidly
when its radius is smaller.

$$
% caption: The occupied focus holds the central mass $M$. The two shaded sectors have equal area, so the planet sweeps them in equal times: a wide fast sweep near periapsis and a narrow slow sweep near apoapsis.
\begin{tikzpicture}[>=stealth,font=\footnotesize]
\definecolor{acc}{HTML}{4A6FA5}
\draw[thick] (0,0) ellipse (2.8 and 1.5);
\coordinate (F) at (1.96,0);
\fill[acc!16] (F)--(2.63,0.513)--(2.8,0)--(2.63,-0.513)--cycle;
\fill[acc!16] (F)--(-2.757,0.260)--(-2.8,0)--(-2.757,-0.260)--cycle;
\draw[black] (F)--(2.63,0.513); \draw[black] (F)--(2.63,-0.513);
\draw[black] (F)--(-2.757,0.260); \draw[black] (F)--(-2.757,-0.260);
\draw[fill=white,draw=black] (F) circle (2.4pt) node[below left] {$M$};
\node[acc] at (2.40,0.62) {fast};
\node[acc] at (-2.05,0.60) {slow};
\end{tikzpicture}
$$

**Newtonian derivation.**

Newton's law gives the force on a satellite of mass $m$ due to central mass $M$:

$$
\vec F=-\frac{GMm}{r^2}\hat r.
$$

The force is central, so its torque about $M$ vanishes. Therefore

$$
\vec L=\vec r\times m\vec v=\text{constant},
\qquad \frac{\d A}{\d t}=\frac{L}{2m}.
$$

The second equality follows because a swept sector has differential area
$\d A=r^2\d\theta/2$ and $L=mr^2\dot\theta$. It proves Kepler's area law for any
central force. The inverse-square force further produces conic-section paths; a
negative-energy orbit is an ellipse, zero energy is a parabola, and positive
energy is a hyperbola.

In a circular orbit, gravity is the radial force:

$$
\frac{mv^2}{r}=\frac{GMm}{r^2},\qquad
v=\sqrt{\frac{GM}{r}},\qquad
T=2\pi\sqrt{\frac{r^3}{GM}}.
$$

Squaring the period establishes Kepler's third law. The satellite mass cancels,
as it must for bodies in the same gravitational field. The formula uses distance
from the central mass centre, not altitude above its surface.

$$
% caption: In a circular orbit the velocity $v$ is tangent to the path while the gravitational force $F_g$ points inward along the radius, supplying the centripetal acceleration.
\begin{tikzpicture}[>=stealth,font=\footnotesize]
\definecolor{acc}{HTML}{4A6FA5}
\draw[thick] (0,0) circle (1.85);
\draw[fill=white,draw=black] (0,0) circle (2.4pt) node[below left] {$M$};
\draw[black,dashed] (0,0)--(1.31,1.31) node[midway,above left] {$r$};
\draw[fill=white,draw=black] (1.31,1.31) circle (2.4pt);
\draw[->,acc,very thick] (1.31,1.31)--(0.40,2.22) node[above right] {$v$};
\draw[->,black,thick] (1.31,1.31)--(0.55,0.55);
\node[black] at (1.02,0.52) {$F_g$};
\end{tikzpicture}
$$

## Orbital energy and escape

Gravitational potential energy, with zero at infinite separation, is

$$
U(r)=-\frac{GMm}{r}.
$$

In a circular orbit, substitution of $v^2=GM/r$ gives

$$
E=\frac12mv^2+U=-\frac{GMm}{2r}.
$$

The negative sign denotes a bound system. More generally, an ellipse has
$E=-GMm/(2a)$, independent of eccentricity. Reaching infinity with zero terminal
speed requires $E=0$, hence

$$
v_{\rm esc}=\sqrt{\frac{2GM}{r}}=\sqrt2v_{\rm circ}.
$$

Escape speed is an energy threshold in the ideal two-body vacuum problem. It is
not a statement that propulsion ends at one altitude or that atmosphere can be
ignored in a launch calculation. An object initially above escape speed can still
collide with the central body if its angular momentum is small enough.

The radial motion can be represented with effective potential

$$
U_{\rm eff}(r)=\frac{L^2}{2mr^2}-\frac{GMm}{r}.
$$

The first term is the angular-motion contribution. Its minimum is the stable
circular orbit; turning points occur where $E=U_{\rm eff}$. This one-dimensional
form combines radial kinetic energy and angular momentum without discarding the
vector origin of the latter.

$$
% caption: The effective potential $U_{\mathrm{eff}}(r)=L^2/2mr^2-GMm/r$ has a single minimum at the stable circular orbit. A bound energy line $E<0$ meets it at two radial turning points, between which the orbit oscillates in radius.
\begin{tikzpicture}[>=stealth,font=\footnotesize]
\definecolor{acc}{HTML}{4A6FA5}
\draw[->,black] (0,0)--(5.9,0) node[right] {$r$};
\draw[->,black] (0,-1.9)--(0,2.6) node[above] {$U$};
\draw[thick] plot[domain=0.72:5.5,samples=200] (\x,{4.28/(\x*\x)-4.76/\x});
\draw[black,dashed] (0.9,-0.78)--(5.2,-0.78);
\node[black] at (5.35,-0.58) {$E$};
\draw[fill=acc!12,draw=acc] (1.095,-0.78) circle (1.9pt);
\draw[fill=acc!12,draw=acc] (5.007,-0.78) circle (1.9pt);
\draw[fill=white,draw=black] (1.80,-1.324) circle (1.9pt);
\draw[black,dotted] (1.80,-1.324)--(1.80,0.02);
\node[black,above] at (1.80,0.05) {circular};
\end{tikzpicture}
$$

> **Worked example.** A satellite circles Earth at radius $r=7.00\times10^6\ \mathrm m$
> from the centre, where $GM_\oplus=3.986\times10^{14}\ \mathrm{m^3\,s^{-2}}$.
> Find its circular speed, period, and the escape speed at that radius.
>
> Circular speed comes from balancing gravity against the centripetal requirement:
>
> $$
> v_{\rm circ}=\sqrt{\frac{GM_\oplus}{r}}
> =\sqrt{\frac{3.986\times10^{14}}{7.00\times10^6}}
> =7.55\times10^3\ \mathrm{m\,s^{-1}}=7.55\ \mathrm{km\,s^{-1}}.
> $$
>
> The period is one circumference at that speed, or equivalently
>
> $$
> T=2\pi\sqrt{\frac{r^3}{GM_\oplus}}
> =2\pi\sqrt{\frac{(7.00\times10^6)^3}{3.986\times10^{14}}}
> =5.83\times10^3\ \mathrm s=97.1\ \mathrm{min}.
> $$
>
> Escape speed is set by $E=0$, giving $v_{\rm esc}=\sqrt2\,v_{\rm circ}=10.7\ \mathrm{km\,s^{-1}}$.
> The factor $\sqrt2$ multiplies the speed, not the energy: escape needs twice the
> kinetic energy of the circular orbit, and the speed ratio is its square root.

## Geometry, state variables, and reference frames

Keplerian motion assumes a point or spherical central mass, an isolated two-body
system, and Newtonian gravity. Atmospheric drag lowers a low-orbit satellite's
energy and angular momentum. Other planets perturb an orbit. Oblateness changes
the field from exact central symmetry and causes orbital-plane precession. These
effects are small in many introductory applications but determine long-term
spacecraft trajectories and precision satellite navigation.

**Geometry of conic orbits.**

The polar equation of a Newtonian two-body orbit is

$$
r(\theta)=\frac{p}{1+e\cos\theta},
$$

where $p=L^2/(GMm^2)$ is the semilatus rectum. At $\theta=0$, the denominator is
largest and the radius is periapsis; at $\theta=\pi$, the radius is apoapsis.
For $0\le e<1$ the curve closes as an ellipse. At $e=1$ it is parabolic, and for
$e>1$ it is hyperbolic. The eccentricity therefore summarizes the balance between
energy and angular momentum, not a separate force parameter.

The ellipse has two foci, but only one is dynamically distinguished by the central
mass. Its semiminor axis is $b=a\sqrt{1-e^2}$. The area is $\pi ab$, so Kepler's
second law and full-orbit period are consistent: the constant areal speed is
$\pi ab/T$. The geometric description does not replace the dynamical one. The
central-force equation is needed to predict how $a$, $e$, and the orbit orientation
change when an impulse or perturbing force occurs.

**Vis-viva equation.**

Combining the total-energy expression $E=-GMm/(2a)$ with
$E=mv^2/2-GMm/r$ gives the vis-viva equation:

$$
v^2=GM\left(\frac2r-\frac1a\right).
$$

At periapsis, $r$ is smallest and speed is largest. At apoapsis, the reverse is
true. Circular motion follows immediately when $r=a$, yielding $v^2=GM/r$. The
equation determines speed at a known point of an ellipse without separately
calculating angular momentum. It does not determine velocity direction; orbit
geometry sets that information.

An impulsive tangential speed increase at one point changes the orbit into an
ellipse with that point usually as periapsis. The spacecraft then climbs outward,
trading kinetic energy for gravitational potential energy. A tangential decrease
places the impulse point near apoapsis and lowers the opposite side of the orbit.
This behavior follows from energy and angular momentum simultaneously. Applying
only one conservation law gives incomplete conclusions about the new path.

**Circular-orbit energy changes.**

In a circular orbit, $E=-GMm/(2r)$ becomes less negative as $r$ increases. A
spacecraft transferred to a higher circular orbit ultimately has greater total
energy, but the initial burn at the lower orbit increases speed only temporarily.
After coasting to the higher radius, its circular speed is smaller because
$v_{\rm circ}\propto r^{-1/2}$. The higher orbit has greater potential energy,
which exceeds the reduction in its circular kinetic energy.

The simplest ideal transfer between two coplanar circular orbits uses an ellipse
tangent to both circles. A first burn changes the lower circular orbit into the
transfer ellipse. A second burn at the other end circularizes. The calculation
uses vis-viva at both endpoints and assumes instantaneous burns, no atmosphere,
and no perturbations. Mission design adds finite thrust, inclination changes,
eclipse constraints, and navigation uncertainty.

**Centre of mass and reduced mass.**

Neither body is literally stationary. Both masses orbit their common centre of
mass. The relative coordinate reduces the two-body problem to one effective mass
$\mu=mM/(m+M)$. For a planet much lighter than its star, the centre of mass lies
near the star's centre and treating the star as fixed is accurate. For binary
stars, both orbital radii can be measured and used to infer the mass ratio from
$m_1r_1=m_2r_2$.

The gravitational parameter $GM$ is usually measured more precisely than $G$ and
$M$ separately. Orbital observations therefore determine $GM$ directly; satellite
formulas use it rather than separately rounded values of the universal constant and
planetary mass.

**Reference frames and observational quantities.**

An inertial frame centred on the two-body centre of mass is the natural frame for
the conservation laws. A rotating orbital frame introduces apparent forces and
requires care. Apparent weightlessness in orbit means that spacecraft and contents
share nearly the same gravitational acceleration; it does not mean gravity has
vanished. Tidal differences remain because gravitational field strength changes
with position.

Observed radial velocity changes shift spectral lines through the Doppler effect.
Together with period and inclination information, such data constrain unseen orbiting
companions. The inference is limited by unknown viewing angle: a face-on orbit can
have substantial true speed while producing little radial velocity. Orbital
mechanics therefore connects conservation laws to astronomical mass measurement,
but every inferred mass carries geometric assumptions.

**Checks on orbital calculations.**

The dimensions of $GM$ are $\mathrm{m^3\,s^{-2}}$. Thus $GM/r$ has dimensions of
speed squared, and $r^3/(GM)$ has dimensions of time squared. A period formula
with $r^2$ rather than $r^3$ fails this check. Bound gravitational energy should be
negative when zero is defined at infinite separation. The distinction between
radius from centre and altitude above a surface is equally important: use
$r=R+h$ for an Earth satellite, not $h$ alone.

> **Worked example.** An orbit about Earth has semimajor axis $a=10.0\times10^6\ \mathrm m$
> and eccentricity $e=0.20$, with $GM=3.986\times10^{14}\ \mathrm{m^3\,s^{-2}}$.
> Find the periapsis and apoapsis speeds and check angular-momentum conservation.
>
> The apsidal radii are $r_p=a(1-e)=8.0\times10^6\ \mathrm m$ and
> $r_a=a(1+e)=12.0\times10^6\ \mathrm m$. Vis-viva at periapsis gives
>
> $$
> v_p=\sqrt{GM\left(\frac{2}{r_p}-\frac1a\right)}
> =\sqrt{3.986\times10^{14}\left(\frac{2}{8.0\times10^6}-\frac{1}{10.0\times10^6}\right)}
> =7.73\ \mathrm{km\,s^{-1}},
> $$
>
> and the same relation at apoapsis gives $v_a=5.15\ \mathrm{km\,s^{-1}}$. Both apses
> are points where the velocity is purely tangential, so the specific angular momentum
> is $rv$ there:
>
> $$
> r_pv_p=(8.0\times10^6)(7.73\times10^3)=6.19\times10^{10}\ \mathrm{m^2\,s^{-1}}
> =r_av_a=(12.0\times10^6)(5.15\times10^3).
> $$
>
> The products agree, as $r_pv_p=r_av_a$ requires.

## Escape trajectories and launch energy

An unbound orbit has hyperbolic excess speed $v_\infty$, the limiting speed far from
the central body. Energy conservation
gives

$$
\frac12mv^2-\frac{GMm}{r}=\frac12mv_\infty^2,
\qquad v^2=v_{\rm esc}^2+v_\infty^2.
$$

The escape case has $v_\infty=0$. A body with positive energy follows a hyperbola,
but its path can still bend substantially near the central mass. Gravitational
assist maneuvers use this deflection in a moving planet's frame to exchange energy
and angular momentum with the planet's orbit about the Sun. The spacecraft does
not obtain energy from gravity in isolation; it obtains it from the planet's
orbital motion through the multi-body interaction.

**Inclination and orbital planes.**

Under a purely central force, the constant angular-momentum vector is perpendicular
to one fixed orbital plane. Its direction identifies inclination and the sense of
motion. An out-of-plane force component produces torque and changes the plane.
Plane changes require substantial velocity change because the velocity vector must
be rotated; they are least costly where orbital speed is low. This is one reason
why high-altitude manoeuvres can be advantageous for missions requiring large
inclination changes.

An orbit may be described by several elements beyond $a$ and $e$: inclination,
the line of nodes, and an angle locating periapsis. In the ideal two-body problem
these elements remain fixed. Perturbations cause secular changes or periodic
oscillations. The simplified Kepler elements organize observations, but they are
not permanent material properties of the orbiting body.

**Energy bookkeeping in launches.**

A launch from a rotating planet already has an inertial tangential speed due to
planetary rotation. Its contribution depends on latitude and launch direction.
Atmospheric drag, gravity losses during finite burns, and nonhorizontal ascent
mean the required rocket energy exceeds the simple circular-orbit or escape energy
change. The ideal formulas remain the baseline for estimating the minimum orbital
mechanical-energy change, not a complete vehicle performance model.

Orbital energy is often quoted per unit mass, $\varepsilon=E/m$. For a bound
ellipse, $\varepsilon=-GM/(2a)$. This form removes the test-mass factor and
makes comparison between spacecraft direct. Specific angular momentum,
$h=L/m=r^2\dot\theta$, similarly describes orbit geometry independently of
satellite mass.

The Newtonian treatment is accurate to parts in $10^7$ for ordinary planetary and
satellite motion. Relativistic corrections become measurable only for compact
objects, high-precision timing, and the perihelion precession of Mercury
($43''$ per century). Those effects deform the Kepler ellipse gradually; they do
not overturn the conservation laws used for routine orbital work.

## Ellipse geometry and orbit classification

A bound Kepler orbit is an ellipse with the attracting mass at one focus, not at the
geometric centre. Semimajor axis a sets its scale and eccentricity e sets its shape.
Periapsis is closest, with $r_p=a(1-e)$; apoapsis is farthest, with
$r_a=a(1+e)$. A circle is the e equal to zero limit. The focus location is essential:
measuring distance from the ellipse centre gives the wrong gravitational radius and
misidentifies the points of greatest and least speed.

$$
% caption: Bound-orbit geometry. The central mass $M$ sits at a focus, not the centre. Periapsis and apoapsis are the near and far ends of the major axis, with focal distances $r_p=a(1-e)$ and $r_a=a(1+e)$.
\begin{tikzpicture}[>=stealth,font=\footnotesize]
\definecolor{acc}{HTML}{4A6FA5}
\draw[thick] (0,0) ellipse (2.9 and 1.6);
\draw[black] (-2.9,0)--(2.9,0);
\draw[black] (-0.09,0.09)--(0.09,-0.09); \draw[black] (-0.09,-0.09)--(0.09,0.09);
\node[black] at (-0.42,0.30) {center};
\draw[fill=white,draw=black] (1.9,0) circle (2.4pt) node[above] {$M$};
\draw[fill=acc!12,draw=acc] (2.9,0) circle (2.2pt);
\draw[fill=acc!12,draw=acc] (-2.9,0) circle (2.2pt);
\node[acc,right] at (2.98,0) {periapsis};
\node[acc,left] at (-2.98,0) {apoapsis};
\draw[->,black] (1.9,0.18)--(2.82,0.18); \node[black,above] at (2.36,0.16) {$r_p$};
\draw[->,black] (1.9,-0.18)--(-2.82,-0.18); \node[black,below] at (-0.5,-0.20) {$r_a$};
\end{tikzpicture}
$$

Kepler's equal-area law follows from conservation of angular momentum under a
central force. Equal times sweep equal sectors from the focus. Near periapsis the
radius is short, so the object covers a larger angle and moves faster. Near apoapsis
the radius is long, so the same sector area requires a smaller angular advance and
slower motion. Equal area is not equal arc length; it is a geometric timing law.

Orbital elements add orientation to the shape. Besides a and e, periapsis direction,
inclination, and a reference-plane node specify how the conic sits in space. These
are coordinate descriptors, so an orbit report requires a reference plane and epoch.
Specific energy and angular momentum distinguish the conics: negative energy gives
an ellipse, zero the parabolic escape limit, and positive energy a hyperbola. The
vis-viva relation $v^2=\mu(2/r-1/a)$ connects speed with focal distance and explains
why a bound object is fastest at periapsis.


The major axis is a geometric line through both foci, whereas the instantaneous
radius vector joins the occupied focus to the orbiting body. The semilatus rectum
sets the local curvature scale in the polar conic equation
$r=p/(1+e\cos\nu)$, where nu is the true anomaly measured from periapsis. This
angle is not proportional to time except for circular motion. It advances rapidly
near periapsis and slowly near apoapsis because equal areas, rather than equal
angles, are swept in equal times.

The area law can be written quantitatively as one-half r squared times angular rate
being constant. This is the specific angular momentum divided by two. For an
observational test, compute focal sector areas over equal time intervals from
successive position measurements. A constant result supports central motion
within measurement uncertainty. A systematic drift can indicate perturbations,
incorrect focus placement, perspective error, or unequal timestamps.

Periapsis and apoapsis are the locations where radial velocity changes sign. At both
points the velocity is purely tangential, but its magnitude differs. Conservation of
angular momentum requires
$r_pv_p=r_av_a$ in the planar two-body model. Combined with energy conservation,
this relation determines the speed ratio and shows why an eccentric orbit spends
most of its period far from the focus even though it travels fastest near it.

An ellipse drawn with the central mass at its centre can still look plausible, but
it fails every timing test. Equal arc segments would then be associated with equal
times, contrary to measured orbital speed variation. The focus construction is
therefore a physical statement about the inverse-square problem, not a decorative
choice of drawing convention. When fitting observations, focus position, scale, and
orientation should be solved together rather than assuming that the apparent centre
of an image is the attracting mass.


**Angular momentum and energy classification.**

Specific angular momentum is $\vec h=\vec r\times\vec v$. A central
gravity force has zero torque about the focus, so h remains constant. In planar
motion this gives constant $r^2\dot\theta$ and therefore equal focal areas in equal
times. The changing speed around an ellipse is required by this conservation law:
a short radius near periapsis must be accompanied by a larger tangential speed than
a long radius near apoapsis.

Specific energy $\epsilon=v^2/2-\mu/r$ classifies conics. Negative energy gives a
bound ellipse, zero energy is the parabolic escape threshold, and positive energy
gives an unbound hyperbola. For an ellipse, $\epsilon=-\mu/(2a)$; combined with
angular momentum, this fixes eccentricity. The vis-viva equation
$v^2=\mu(2/r-1/a)$ then relates speed to focal distance and confirms that speed is
largest at periapsis.

$$
% caption: Trajectories sharing one focus $M$ and one periapsis, classified by specific energy. The bound ellipse has $E<0$, the parabola is the escape threshold $E=0$, and the hyperbola is unbound with $E>0$.
\begin{tikzpicture}[>=stealth,font=\footnotesize,scale=0.66]
\definecolor{acc}{HTML}{4A6FA5}
\begin{scope}
\clip (-3.4,-2.85) rectangle (1.6,2.85);
\draw[thick] plot[domain=0:360,samples=180] ({1.5/(1+0.5*cos(\x))*cos(\x)},{1.5/(1+0.5*cos(\x))*sin(\x)});
\draw[acc,thick] plot[domain=-118:118,samples=170] ({2.0/(1+cos(\x))*cos(\x)},{2.0/(1+cos(\x))*sin(\x)});
\draw[thick,dashed] plot[domain=-112:112,samples=180] ({2.4/(1+1.4*cos(\x))*cos(\x)},{2.4/(1+1.4*cos(\x))*sin(\x)});
\end{scope}
\draw[fill=white,draw=black] (0,0) circle (2.8pt) node[below right] {$M$};
\draw[fill=white,draw=black] (1.0,0) circle (2.4pt);
\node[right] at (0.62,1.35) {bound};
\node[acc,right] at (0.62,1.95) {escape};
\node[right] at (0.62,2.55) {unbound};
\end{tikzpicture}
$$

Energy alone does not determine an orbit. Equal-energy trajectories can have
different closest approaches because their angular momenta differ. A position and
velocity state vector at one epoch contains both invariants: energy sets size,
angular momentum sets plane and areal speed, and the eccentricity vector identifies
periapsis direction. These checks are central to orbit reconstruction and numerical
propagation.


The vis-viva relation locally checks the speed of every fitted bound orbit. At a known focal
distance, the combination of measured speed and gravitational parameter determines
semimajor axis. Repeating the calculation near periapsis and apoapsis should give
the same a within uncertainty, even though the speeds differ substantially. A
mismatch can indicate that distance was measured from the ellipse centre instead of
the focus, that a velocity component was omitted, or that the two-body model is not
adequate over the observed arc.

Period inference follows once semimajor axis is known. In the ideal two-body model,
the period grows as a to the three-halves power. Eccentricity affects where the body
spends time and how its speed varies, but not the period associated with a. Timing
successive passages through the same reference direction provides an observational
period that can be compared with the inferred value. A short arc gives a weak period
constraint because energy and along-track position remain correlated; later
observations or range measurements reduce that ambiguity.

Perturbation limits define the validity of a Kepler prediction. Third-body gravity,
oblateness, atmospheric drag, radiation pressure, and thrust change the osculating
state. The fitted ellipse then remains an instantaneous summary but not a
permanent path. Residuals should be resolved into radial, along-track, and
cross-track components. Their trend can distinguish a period error from a plane
error or a physical perturbing acceleration. A reported orbit therefore needs an
epoch, reference frame, gravitational parameter, force model, and stated prediction
interval. These details specify a reproducible numerical propagation and prevent a
short-term ellipse fit from being interpreted as an exact long-term trajectory.

## State reconstruction and perturbations

Orbit determination begins from a state vector: position and velocity measured at a
stated epoch in an inertial frame. The specific angular momentum vector fixes the
orbital plane. Specific energy fixes semimajor axis when the orbit is bound. The
eccentricity vector fixes both eccentricity magnitude and periapsis direction.
These invariants convert one instantaneous state into a Kepler conic, but every
input must use the same origin, time scale, and gravitational parameter.

Newtonian perturbations alter the elements because they add acceleration beyond the
central inverse-square term. A distant body can produce periodic changes and secular
precession. Oblateness produces predictable node and periapsis drift. Drag removes
energy and angular momentum, shrinking an orbit. Thrust changes the state by design.
The instantaneous Kepler ellipse is an osculating orbit, but it must
be updated as the state evolves.

Apsidal precession is rotation of the periapsis direction between successive orbits.
In Newtonian celestial mechanics it can arise from other bodies or a nonspherical
central mass. It should be measured relative to a stated inertial reference, since
an apparent rotation can also come from a rotating coordinate system. A small
precession accumulates slowly, so long timing baselines and repeated periapsis
observations are more informative than one fitted ellipse.

$$
% caption: Apsidal precession. The orbit keeps its focus at the central mass $M$ while its periapsis direction advances a small angle each revolution; the effect accumulates over many cycles.
\begin{tikzpicture}[>=stealth,font=\footnotesize,scale=0.72]
\definecolor{acc}{HTML}{4A6FA5}
\draw[black,thick] plot[domain=0:360,samples=160] ({1.4/(1+0.6*cos(\x))*cos(\x)},{1.4/(1+0.6*cos(\x))*sin(\x)});
\draw[thick,dashed,rotate=32] plot[domain=0:360,samples=160] ({1.4/(1+0.6*cos(\x))*cos(\x)},{1.4/(1+0.6*cos(\x))*sin(\x)});
\draw[fill=white,draw=black] (0,0) circle (2.6pt) node[below left] {$M$};
\draw[->,black,thick] (0,0)--(0.85,0);
\draw[->,acc,thick] (0,0)--(0.721,0.451);
\draw[fill=white,draw=black] (0.875,0) circle (1.9pt);
\draw[fill=acc,draw=acc] (0.742,0.464) circle (1.9pt);
\draw[->,acc] (1.25,0) arc (0:32:1.25);
\node[acc,right] at (1.28,0.35) {advance};
\end{tikzpicture}
$$

The two-body model has clear limits. It neglects finite light speed, relativistic
corrections, mass loss, tides, and all other gravitating bodies. It is often highly
accurate over a limited interval, but accuracy is a quantitative claim that depends
on required position and timing precision. A model should be extended when residuals
exceed measurement uncertainty in a repeatable pattern. One later point displaced
from an earlier fitted ellipse does not establish a perturbation signature.

Observational reconstruction should retain covariance. A short arc can fit many
nearby state vectors whose future positions diverge. Range, range rate, and later
angles reduce this uncertainty differently. Propagating an ensemble of plausible
states gives a prediction region rather than one unjustifiably precise path. The
reported epoch, frame, force model, residual statistics, and validity interval are
therefore part of the orbit solution itself.

**State reconstruction, anomaly time, perturbation signatures, and uncertainty.**

A state vector determines an osculating conic only after coordinates are expressed
in one inertial frame and at one epoch. Position gives focal distance and velocity
determines energy and angular momentum. The eccentricity vector identifies periapsis
direction. From these quantities, a reconstruction predicts the instantaneous
ellipse and its future two-body phase. A small error in velocity can dominate the
semimajor-axis error because energy is the difference of kinetic and potential terms.

True anomaly measures angle from periapsis at the focus, but it does not advance
uniformly with time. Eccentric anomaly provides an auxiliary circular construction,
and mean anomaly advances uniformly as $M=n(t-t_p)$ with mean motion n. Kepler's
equation connects them through $M=E-e\sin E$. Solving it numerically converts a
uniform time coordinate into the nonuniform orbital angle required by the area law.

$$
% caption: The auxiliary-circle construction behind Kepler's equation. Point $P$ on the ellipse projects vertically to $Q$ on the circumscribing circle of radius $a$. The angle $QOx$ at the centre is the eccentric anomaly; the angle $PMx$ at the focus is the true anomaly.
\begin{tikzpicture}[>=stealth,font=\footnotesize]
\definecolor{acc}{HTML}{4A6FA5}
\draw[thick] (0,0) circle (2.6);
\draw[black,thick] (0,0) ellipse (2.6 and 2.0);
\draw[black] (-0.09,0.09)--(0.09,-0.09); \draw[black] (-0.09,-0.09)--(0.09,0.09);
\node[black,below left] at (0,0) {$O$};
\draw[fill=white,draw=black] (1.66,0) circle (2.4pt) node[below right] {$M$};
\draw[fill=white,draw=black] (2.6,0) circle (2pt);
\draw[fill=black!55,draw=black] (1.3,1.732) circle (1.8pt) node[below right] {$P$};
\draw[fill=acc,draw=acc] (1.3,2.252) circle (1.8pt) node[above right] {$Q$};
\draw[black,dashed] (1.3,1.732)--(1.3,2.252);
\draw[acc,thick] (0,0)--(1.3,2.252);
\draw[black,thick] (1.66,0)--(1.3,1.732);
\draw[acc] (0.55,0) arc (0:60:0.55);
\draw[black] (1.66,0) ++(0.5,0) arc (0:101.7:0.5);
\end{tikzpicture}
$$

Perturbations leave characteristic residual patterns. A persistent along-track
residual often indicates period or energy error. Cross-track residual can indicate
plane orientation error, nodal precession, or a third-body acceleration. A radial
residual can indicate wrong focus geometry, range bias, or a force-model error.
Periodic signatures often trace a perturbing body; secular signatures can trace
oblateness, drag, or continuous thrust. Classification requires calibrated times and
reference frames before assigning a physical cause.

Orbital-element uncertainty is correlated. A short angle-only arc may constrain the
apparent plane while leaving range, energy, and period poorly constrained. Range
rate or later observations reduce different directions of uncertainty. Covariance
propagation or an ensemble of plausible state vectors gives a prediction region,
not one falsely precise future position. An orbit report includes epoch,
reference frame, gravitational parameter, force model, element covariance, and the
interval over which the propagated uncertainty remains acceptable.

**Vis-viva inference and perturbation limits.**

The vis-viva equation relates position and speed at one epoch. Given r and v,
rearranging
$v^2=\mu(2/r-1/a)$ gives the semimajor axis and therefore the specific energy. A
negative resulting energy identifies a bound ellipse. If the measured speed equals
$\sqrt{2\mu/r}$ within uncertainty, the orbit is near the parabolic escape limit
and small measurement errors can change the inferred class. This sensitivity should
be reported rather than hidden by rounding the energy sign.

In a bound two-body orbit, semimajor axis determines the period through
$T=2\pi\sqrt{a^3/\mu}$. This relation permits period inference from one state vector,
but its uncertainty can be substantial when the observed arc is short. Semimajor
axis is correlated with the unobserved along-track position and velocity. Later
observations, range information, or a longer timing baseline reduce that correlation.
A period inferred from a visually fitted ellipse without focal geometry has no
comparable physical basis.

Orbit prediction requires an epoch. The elements computed from a state vector are
osculating elements: the Kepler conic tangent to the actual trajectory at that time.
They are exact only for an isolated two-body force. Third-body gravity, nonspherical
mass distributions, atmospheric drag, radiation pressure, and thrust make the state
and therefore the osculating elements change. A fitted ellipse can still summarize
a short arc accurately while being a poor long-term prediction.

Perturbation limits are diagnosed through residuals after propagation. A steady
periapsis rotation suggests a central-force correction or oblateness; a secular
change in semimajor axis can indicate drag or thrust; periodic residuals can indicate
a third-body perturbation. Timing, reference-frame, and station-location errors can
produce similar patterns, so physical interpretation requires calibrated observations
and an uncertainty model. Report an estimated perturbing acceleration or element
drift rate rather than claiming that Kepler's laws failed.

A reconstruction report should state the gravitational parameter, reference frame,
epoch, measurement types, force model, and covariance of the fitted state. It should
also state whether quoted elements are mean, osculating, or averaged over an interval.
These distinctions define a reproducible Kepler solution and the range over
which its period, periapsis distance, and future position are physically defensible.


## Observation design and validation

Multi-epoch sampling is more informative than a dense cluster of observations from
one short interval. Measurements separated in orbital phase constrain different
parts of the state vector. Periapsis observations constrain high-speed geometry and
timing; apoapsis observations constrain the long-period portion of the ellipse;
out-of-plane observations constrain inclination and nodes. Secular perturbations
appear as trends across a sequence spanning many periods; a single apparent ellipse
cannot separate them from initial-state uncertainty.

Residual covariance must be retained during fitting. Angles measured by one camera
share pointing and timing errors, while range and range-rate data can share a clock
or station-location bias. Treating all observations as independent makes a fit
appear more certain than the apparatus permits. Covariance propagation turns a
state estimate into a prediction region. Its elongation often follows the orbital
track because small period uncertainty accumulates into large along-track error.

| observable | directly constrained state information | principal ambiguity | calibration record |
| --- | --- | --- | --- |
| astrometric direction | line of sight and apparent orbital plane | distance and radial speed | plate scale, pointing, reference frame, epoch |
| range | station-to-target separation | transverse velocity | station coordinates, delay convention, clock |
| Doppler range rate | line-of-sight velocity | tangential velocity | transmitter frequency, sign convention, timing |
| transit or occultation time | orbital phase and recurrence | three-dimensional orientation | event definition, time scale, light-time treatment |

Sampling these observables at separated orbital phases constrains different
combinations of the state vector. A repeated measurement of one quantity at nearly
one geometry mainly reduces its local random scatter.

Time standards are part of orbital dynamics. A timestamp must specify the scale and
reference used by the ephemeris and station data. Mixing a civil clock, a receiver
clock with unknown delay, and a dynamical time scale can create residuals that look
like period drift or apsidal motion. Light-time correction also matters when angular
observations refer to photons emitted earlier than their reception time. The required
correction depends on accuracy and distance, but the assumption to neglect it should
be stated rather than implicit.

Force-model selection should begin with the precision target. A short educational
calculation may use a fixed two-body gravitational parameter. Precision navigation
can require nonspherical gravity, third bodies, radiation pressure, tides, drag,
and relativistic terms. Adding every possible correction is not automatically better:
poorly calibrated parameters can degrade a fit. A model extension is justified when
it reduces structured residuals consistently across independent data and has a
physically constrained parameterization.

| validation operation | comparison made | conclusion supported |
| --- | --- | --- |
| withhold selected epochs | propagated state against unused observations | predictive performance outside the fitted arc |
| change observation geometry | residuals across orbital phase or station | state components constrained by the new geometry |
| add one force term | residual structure before and after propagation | physical need for a perturbation model |
| shift timing or frame convention | residual sign and secular trend | clock or reference-frame consistency |
| propagate an uncertainty ensemble | prediction corridor at the target epoch | defensible prediction horizon |

Each comparison must preserve the same calibrated measurement record. A lower
residual obtained by changing both the force model and the data convention has no
unique physical interpretation.

The restricted three-body limit is a warning boundary. When a secondary
body's gravity is comparable to the primary's over the required trajectory segment,
a primary-centred Kepler ellipse is not a sufficient propagation model. The state
can be transformed to a rotating frame for qualitative analysis, but apparent terms
and the chosen frame must then be included consistently. A two-body element set may
still describe the instantaneous osculating orbit without predicting passage through
the secondary's region of influence.

Extrapolation should be reported separately from interpolation. A model can match
an observed arc while accumulating unacceptable error outside it. The interval of
validity depends on state covariance, unmodeled forces, and sensitivity of the future
trajectory to those forces. Propagating an ensemble of plausible states and force
parameters gives a corridor rather than one unrealistically exact path. A prediction
that remains stable under these variations is more actionable than a nominal trajectory
with many digits but no uncertainty context.

Model validation closes the loop. Compare residuals by component and epoch, test
whether their distribution matches the stated covariance, and inspect for periodic
or secular structure. Refit after withholding some observations to test predictive
performance. If a correction improves only the fitted arc but not withheld data, it
may be overfitting. A defensible Kepler-orbit result therefore records the state,
reference frame, time standard, force assumptions, residual diagnostics, covariance,
and prediction interval as one connected physical statement.


**Timing, range, and radial-velocity measurements.**

Astrometry measures direction on the sky, usually as angles relative to a calibrated
reference frame. Repeated directions trace an apparent orbit, but an angular arc
alone does not directly give physical distance, mass parameter, or period. Those are
inferred elements obtained only after a geometric and dynamical model is fitted.
Parallax, a known observer baseline, range data, or later observations can break the
scale ambiguity. The accuracy of an angular orbit depends on plate scale, pointing,
reference-star calibration, and the time span over which curvature becomes visible.

Transit or occultation timing measures a different observable: the epoch at which a
body crosses a specified line of sight or passes in front of another object. A
sequence of transit times constrains period and changes in mean motion. It does not,
by itself, determine three-dimensional orientation or eccentricity, since distinct
orbits can share similar recurrence times. Differences between predicted and
observed transit epochs are diagnostic residuals, but they must be compared with the
clock standard, light-time convention, and definition of the geometric event before
they are assigned to a perturbation.

Range is a direct distance measurement from an observing station to the target.
Range rate is its time derivative, the line-of-sight component of relative velocity.
Together they constrain the otherwise weakly observed radial direction. Radar delay,
laser ranging, and two-way radio tracking can provide range; Doppler shift provides
range rate after the transmitter and receiver frequency standards are accounted for.
For nonrelativistic motion, a small fractional Doppler shift is approximately the
negative radial speed divided by signal speed. The sign convention must be stated:
receding and approaching targets produce opposite shifts.

Doppler data do not measure full orbital speed. They measure only the component
along the instantaneous line of sight. A large tangential velocity can remain
nearly invisible when the viewing geometry is unfavorable. Astrometry complements
Doppler measurements by constraining angular motion on the sky, while range places
the target along the line of sight. Combining the three observables over multiple
epochs is substantially more informative than repeating one observable at one
orbital phase.

Cadence and baseline set what can be inferred. Closely spaced observations resolve
rapid periapsis motion and short-period variations, whereas a long time baseline
constrains mean motion and secular element drift. Gaps can alias a period or confuse
one orbital revolution with another. Measurements concentrated near one phase may
fit position accurately while leaving semimajor axis and eccentricity correlated.
An observing plan should therefore sample distinct orbital phases and retain precise
timestamps, station coordinates, and uncertainty estimates.

A final orbit solution should identify which quantities were observed directly and
which were inferred through the force model. It should list the timing standard,
observer ephemeris, range and Doppler calibration, data cadence, and covariance of
the fitted elements. This separation prevents an apparent precision in semimajor
axis, inclination, or periapsis direction from being mistaken for a direct
measurement when it is actually a model-dependent consequence of a limited data
set.


## Prediction horizons and model limits

Osculating elements describe the Kepler conic that matches a body's position and
velocity at one instant under a chosen gravitational parameter. They compress a
six-component state vector into geometric quantities such as
semimajor axis, eccentricity, inclination, and periapsis direction. They are not,
however, permanent labels attached to the body. When an additional acceleration acts,
the instantaneous state changes, and the osculating elements change with it. A
sequence of osculating ellipses can therefore represent a smooth perturbed trajectory
without implying that the body jumps from one physical path to another.

| perturbation class | common orbital signature | model extension |
| --- | --- | --- |
| nonspherical primary gravity | nodal or periapsis precession | gravity harmonics and a stated reference frame |
| third-body gravity | periodic residuals or encounter-scale trajectory change | multi-body propagation |
| atmospheric drag | decreasing energy, semimajor axis, and period | density and ballistic-coefficient model |
| radiation pressure or small thrust | orientation-dependent element drift | attitude and acceleration model |
| rotating analysis frame | apparent Coriolis and centrifugal terms | rotating-frame equations with stated rate |

Residual structure selects the extension. A force term should be added only when its
predicted signature improves independent observations within their calibration
uncertainty.

Non-spherical gravity is one important source of element evolution. A rapidly
rotating or oblate primary has a gravitational potential that differs from that of a
point mass. Its equatorial bulge produces secular rotation of the orbital node and
periapsis, with rates that depend on semimajor axis, eccentricity, and inclination.
For low-altitude satellites these corrections can dominate long-term prediction even
though the two-body ellipse remains an excellent description over a single short
arc. The correction is Newtonian: it reflects the primary's mass distribution, not a
failure of gravity or a numerical artifact in element fitting.

Third-body gravity provides another perturbation class. A distant body can add small
periodic terms that average nearly to zero over one orbit, while a nearby body can
produce large changes in energy and angular momentum. The strength of the effect
must be compared with the primary's gravity across the relevant portion of the
trajectory. Near a secondary's region of influence, a primary-centred two-body orbit
is often a poor forecast even if it fitted earlier observations. Restricted
three-body or full multi-body integration then replaces the single fixed ellipse.

Atmospheric drag removes orbital energy and angular momentum from low-altitude
objects. The resulting semimajor axis and period typically decrease, and the effect
is sensitive to density variation, attitude, and area-to-mass ratio. Radiation
pressure acts in the opposite direction for some geometries and can be significant
for light, large-area spacecraft. Unlike central gravity, drag and radiation depend
on velocity or orientation relative to an environment, so they cannot generally be
absorbed into a constant gravitational parameter. Their coefficients must be
estimated from data or supplied by an independently validated physical model.

Reference-frame choice changes the appearance of motion but not the underlying
trajectory. In an inertial barycentric or primary-centred frame, Newton's laws use
real forces directly. In a frame rotating with an orbit or with a planet, apparent
centrifugal and Coriolis terms are required for an equivalent equation of motion.
A rotating-frame equilibrium or zero-velocity construction is meaningful only after
those terms and the rotation rate have been stated. Mixing an inertial element set
with rotating-frame accelerations produces residuals that can be mistaken for a
physical perturbation.

| forecast quantity | leading uncertainty source | observation or model input that reduces it |
| --- | --- | --- |
| along-track arrival position | mean motion and period | separated timing, range-rate, and post-burn tracking |
| orbital-plane position | inclination, node, and normal acceleration | out-of-plane angular observations |
| low-altitude lifetime | density, attitude, and area-to-mass ratio | drag calibration over varied atmospheric conditions |
| encounter geometry | third-body ephemeris and initial-state covariance | multi-epoch range and multi-body propagation |

The relevant horizon is reached when the uncertainty in the reported quantity exceeds
its stated tolerance. A single position uncertainty is insufficient for forecasts
that depend on time of arrival, plane crossing, or close-approach geometry.

Forecast precision declines because uncertainty in the measured state and force
model is propagated forward. Along-track uncertainty often grows fastest: a small
mean-motion error accumulates into a large phase error after many revolutions.
Uncertainty in drag, radiation pressure, or third-body ephemerides can introduce
additional growth. The shape of a prediction covariance is therefore as important as
its nominal centre. A narrow state uncertainty at the fitting epoch can become a
long corridor of plausible future positions rather than a single precise point.

State updates from new observations reset and reshape this uncertainty. They do not
merely correct a plotted orbit; they constrain the combinations of position,
velocity, and force parameters that previous data left correlated. An update near a
rapidly changing orbital phase may be especially valuable, while repeated
measurements at nearly identical geometry can add little new information. Data
assimilation should retain time standards, observer locations, and measurement
covariance so that apparent element changes are not created by inconsistent inputs.

A prediction should therefore be accompanied by an epoch, reference frame,
propagation model, included perturbations, numerical tolerance, and a stated horizon
or accuracy criterion. Beyond that horizon, the nominal trajectory may still be a
scenario, but it is not a precision estimate. A Kepler ellipse is a local two-body
model; reliable long-term prediction requires measured state updates and a force
model matched to the demanded accuracy.