Orbital perturbation analysis is the activity of determining why a satellite's orbit differs from the mathematical ideal orbit. A satellite's orbit in an ideal two-body system describes a conic section, usually an ellipse. In reality, there are several factors that cause the conic section to continually change. These deviations from the ideal Kepler's orbit are called perturbations.
It has long been recognized that the Moon does not follow a perfect orbit, and many theories and models have been examined over the millennia to explain it. Isaac Newton determined the primary contributing factor to orbital perturbation of the moon was that the shape of the Earth is actually an oblate spheroid due to its spin, and he used the perturbations of the lunar orbit to estimate the oblateness of the Earth.
In Newton's Philosophiæ Naturalis Principia Mathematica, he demonstrated that the gravitational force between two mass points is inversely proportional to the square of the distance between the points, and he fully solved the corresponding "two-body problem" demonstrating that the radius vector between the two points would describe an ellipse. But no exact closed analytical form could be found for the three body problem. Instead, mathematical models called "orbital perturbation analysis" have been developed. With these techniques a quite accurate mathematical description of the trajectories of all the planets could be obtained. Newton recognized that the Moon's perturbations could not entirely be accounted for using just the solution to the three body problem, as the deviations from a pure Kepler orbit around the Earth are much larger than deviations of the orbits of the planets from their own Sun-centered Kepler orbits, caused by the gravitational attraction between the planets. With the availability of digital computers and the ease with which we can now compute orbits, this problem has partly disappeared, as the motion of all celestial bodies including planets, satellites, asteroids and comets can be modeled and predicted with almost perfect accuracy using the method of the numerical propagation of the trajectories. Nevertheless several analytical closed form expressions for the effect of such additional "perturbing forces" are still very useful.
All celestial bodies of the Solar System follow in first approximation a Kepler orbit around a central body. For a satellite (artificial or natural) this central body is a planet. But both due to gravitational forces caused by the Sun and other celestial bodies and due to the flattening of its planet (caused by its rotation which makes the planet slightly oblate and therefore the result of the Shell theorem not fully applicable) the satellite will follow an orbit around the Earth that deviates more than the Kepler orbits observed for the planets.
The precise modeling of the motion of the Moon has been a difficult task. The best and most accurate modeling for the lunar orbit before the availability of digital computers was obtained with the complicated Delaunay and Brown's lunar theories.
For man-made spacecraft orbiting the Earth at comparatively low altitudes the deviations from a Kepler orbit are much larger than for the Moon. The approximation of the gravitational force of the Earth to be that of a homogeneous sphere gets worse the closer one gets to the Earth surface and the majority of the artificial Earth satellites are in orbits that are only a few hundred kilometers over the Earth surface. Furthermore they are (as opposed to the Moon) significantly affected by the solar radiation pressure because of their large cross-section to mass ratio; this applies in particular to 3-axis stabilized spacecraft with large solar arrays. In addition they are significantly affected by rarefied air below 800–1000 km. The air drag at high altitudes is also dependent on solar activity.
If the force is the sum of the "Kepler force" and an additional force (force per unit mass)
one therefore has
and that the change of in the time from to is
If now the additional force is sufficiently small that the motion will be close to that of a Kepler orbit one gets an approximate value for by evaluating this integral assuming to precisely follow this Kepler orbit.
In general one wants to find an approximate expression for the change over one orbital revolution using the true anomaly as integration variable, i.e. as
This integral is evaluated setting , the elliptical Kepler orbit in polar angles. For the transformation of integration variable from time to true anomaly it was used that the angular momentum by definition of the parameter for a Kepler orbit (see equation (13) of the Kepler orbit article).
For the special case where the Kepler orbit is circular or almost circular
Let and make up a rectangular coordinate system in the plane of the reference Kepler orbit. If is the argument of perigee relative the and coordinate system the true anomaly is given by and the approximate change of the orbital pole (defined as the unit vector in the direction of the angular momentum) is
where is the component of the perturbing force in the direction, is the velocity component of the Kepler orbit orthogonal to radius vector and is the distance to the center of the Earth.
For a circular or almost circular orbit (5) simplifies to
In a circular orbit a low-force propulsion system (Ion thruster) generates a thrust (force per unit mass) of in the direction of the orbital pole in the half of the orbit for which is positive and in the opposite direction in the other half. The resulting change of orbit pole after one orbital revolution of duration is
The average change rate is therefore
where is the orbital velocity in the circular Kepler orbit.
Rather than applying (1) and (2) on the partial derivatives of the orbital elements eccentricity and argument of perigee directly one should apply these relations for the eccentricity vector. First of all the typical application is a near-circular orbit. But there are also mathematical advantages working with the partial derivatives of the components of this vector also for orbits with a significant eccentricity.
Equations (60), (55) and (52) of the Kepler orbit article say that the eccentricity vector is
The eccentricity vector is by definition always in the osculating orbital plane spanned by and and formally there is also a derivative
corresponding to the rotation of the orbital plane
But in practice the in-plane change of the eccentricity vector is computed as
ignoring the out-of-plane force and the new eccentricity vector
is subsequently projected to the new orbital plane orthogonal to the new orbit normal
computed as described above.
The Sun is in the orbital plane of a spacecraft in a circular orbit with radius and consequently with a constant orbital velocity . If and make up a rectangular coordinate system in the orbital plane such that points to the Sun and assuming that the solar radiation pressure force per unit mass is constant one gets that
where is the polar angle of in the , system. Applying (2) one gets that
This means the eccentricity vector will gradually increase in the direction orthogonal to the Sun direction. This is true for any orbit with a small eccentricity, the direction of the small eccentricity vector does not matter. As is the orbital period this means that the average rate of this increase will be
In the article Geopotential model the modeling of the gravitational field as a sum of spherical harmonics is discussed. By far, the dominating term is the "J2-term". This is a "zonal term" and corresponding force is therefore completely in a longitudinal plane with one component in the radial direction and one component with the unit vector orthogonal to the radial direction towards north. These directions and are illustrated in Figure 1.
Figure 2: The unit vector orthogonal to in the direction of motion and the orbital pole . The force component is marked as "F"
To be able to apply relations derived in the previous section the force component must be split into two orthogonal components and as illustrated in figure 2
Let make up a rectangular coordinate system with origin in the center of the Earth (in the center of the Reference ellipsoid) such that points in the direction north and such that are in the equatorial plane of the Earth with pointing towards the ascending node, i.e. towards the blue point of Figure 2.
The components of the unit vectors
making up the local coordinate system (of which are illustrated in figure 2) relative the are
where is the polar argument of relative the orthogonal unit vectors and in the orbital plane
where is the angle between the equator plane and (between the green points of figure 2) and from equation (12) of the article Geopotential model one therefore gets that
Secondly the projection of direction north, , on the plane spanned by is
and this projection is
where is the unit vector orthogonal to the radial direction towards north illustrated in figure 1.
where is the eccentricity and is the argument of perigee of the reference Kepler orbit
As all integrals of type
are zero if not both and are even one gets from (21) that
this can be written
As is an inertially fixed vector (the direction of the spin axis of the Earth) relation (22) is the equation of motion for a unit vector describing a cone around with a precession rate (radians per orbit) of
In terms of orbital elements this is expressed as
is the inclination of the orbit to the equatorial plane of the Earth
are the components of the eccentricity vector in the coordinate system this integral (25) can be evaluated analytically, the result is
This the difference equation of motion for the eccentricity vector to form a circle, the magnitude of the eccentricity staying constant.
Translating this to orbital elements it must be remembered that the new eccentricity vector obtained by adding to the old must be projected to the new orbital plane obtained by applying (23) and (24)
Figure 3: The change in "argument of perigee" after one orbit is the sum of a contribution caused by the in-plane force components and a contribution caused by the use of the ascending node as reference
This is illustrated in figure 3:
To the change in argument of the eccentricity vector
must be added an increment due to the precession of the orbital plane (caused by the out-of-plane force component) amounting to
One therefore gets that
In terms of the components of the eccentricity vector relative the coordinate system that precesses around the polar axis of the Earth the same is expressed as follows
where the first term is the in-plane perturbation of the eccentricity vector and the second is the effect of the new position of the ascending node in the new plane
From (28) follows that is zero if . This fact is used for Molniya orbits having an inclination of 63.4 deg. An orbit with an inclination of 180 - 63.4 deg = 116.6 deg would in the same way have a constant argument of perigee.