Prev
Unit 7 · Lesson 7-1
Earth Oblateness and J₂
38m75 XP
Next

Earth Oblateness and J₂

Nodal regression, apsidal rotation, sun-synchronous and frozen orbits, critical inclination

Twenty-one kilometres
Earth is not a sphere. It has been spinning for four and a half billion years, and centrifugal force has thrown its equator outward: the equatorial radius is 6378.1 km, the polar radius is 6356.8 km. A difference of 21.3 kilometres. As a fractional error that is 0.3%. In almost any other engineering context you would round it away without a second thought. In orbital mechanics it is the single most consequential thing about the planet. That bulge takes the orbital plane of every satellite ever launched and rotates it, continuously, for as long as the satellite exists. A typical LEO orbit's ascending node sweeps around 5 to 7 degrees per day — more than a full revolution of the plane every two months. It also rotates the ellipse within its own plane, at rates of several degrees per day. And this is not a small correction bolted onto Kepler. Over a fifteen-year mission a satellite's node moves through tens of thousands of degrees. The eccentricity vector — the thing Unit 2 proved was conserved — is not conserved at all, and never was. What makes this the most beautiful part of the subject is what engineers did next. They did not fight the bulge. They used it, and built two entire classes of orbit that only work because the Earth is the wrong shape.
Watch first · presented by Shoaib Aziz
The numbers in this film are the labs' own: King-Hele and Merson's 1958 Sputnik 2 flattening of 1/298.1 against the surveyors' 1/297.1; Earth's radii 6,378 and 6,357 km; the ISS (a = 6,790 km, i = 51.6°) regresses 4.97° a day, a full circle in 72 days; a sun-synchronous orbit needs +0.9856° a day, so 96.7° at 300 km, 98.2° at 700 km and 100.4° at 1,200 km, and none exists above about 5,970 km; perigee freezes at 63.4°, where Molniya's drifts 0.0004° a day; and integrating J₂ for 5 days measures 4.99° a day against the predicted 4.97°.
The clean way to handle a non-spherical body is to expand its gravitational potential in spherical harmonics — a series of correction terms on top of the point-mass μ/r\mu/r:
U(r,ϕ)=−μr[1−∑n=2∞Jn(R⊕r)nPn(sin⁡ϕ)]U(r,\phi) = -\frac{\mu}{r}\left[1 - \sum_{n=2}^{\infty} J_{n}\left(\frac{R_\oplus}{r}\right)^n P_{n}(\sin\phi)\right]

The zonal harmonic expansion. φ is geocentric latitude and PnP_{n} are Legendre polynomials. J₂ is the oblateness term; higher terms describe progressively finer lumps.

TermValueRelative to J₂What it describes
J₂1.0826 × 10⁻³1Equatorial bulge — oblateness
J₃−2.53 × 10⁻⁶0.0023Pear shape — northern hemisphere slightly pointier
J₄−1.62 × 10⁻⁶0.0015Higher-order flattening
J₂₂1.82 × 10⁻⁶0.0017Equatorial ellipticity — the equator is not a circle either
💡Why J₂ dominates the zonal harmonics
J2J_2 is about four hundred times larger than every other term. That single number dominates so completely that most operational analysis uses a J₂-only model and calls it done. The reason is physical rather than mathematical: J2J_2 is the signature of rotation, and Earth rotates fast. Every rapidly-spinning body has a large J2J_2. Jupiter's is 1.47 × 10⁻², thirteen times Earth's, which makes the precession of its moons dramatic and was one of the first ways its interior structure was probed. The higher terms describe genuine lumpiness — mountain ranges, mantle density variations, the Hudson Bay gravity low left over from the last ice age — and they matter for precise geodesy, for GRACE-class gravity mapping, and for anything orbiting a small irregular body like an asteroid, where the "higher-order" terms are not small at all. For an Earth satellite, though: J2J_2 and everything else.
Figure 7.1
satelliteoblateness exaggerated 50×21.3 km — a 0.3% errorequatorial radius 6378.1 kmpolar radius 6356.8 kmThe extra mass at the equatorpulls a satellite slightly*toward the equatorial plane*rather than straight at thecentre.That off-axis component is atorque — and a torque on aspinning system precesses.
The bulge, exaggerated fifty-fold so it is visible at all. The extra equatorial mass pulls a satellite crossing the equator slightly toward the equatorial plane rather than straight at the centre. That small off-axis component is a torque, and a torque on a spinning system produces precession.
The mechanism is gyroscopic. The extra mass at the equator exerts a torque on the orbit, and — exactly as with a spinning top — the response is not to tip over but to precess. Averaged over one revolution, the secular drift rates are:
Ω˙=−32 μ J2R⊕2a7/2(1−e2)2 cos⁡i\dot{\Omega} = -\frac{3}{2}\,\frac{\sqrt{\mu}\,J_{2} R_\oplus^2}{a^{7/2}(1-e^2)^2}\,\cos i

Nodal regression. The minus sign and the cos i together mean prograde orbits (i < 90°) drift westward, and retrograde orbits drift east. A polar orbit does not drift at all.

ω˙=34 μ J2R⊕2a7/2(1−e2)2 (5cos⁡2i−1)\dot{\omega} = \frac{3}{4}\,\frac{\sqrt{\mu}\,J_{2} R_\oplus^2}{a^{7/2}(1-e^2)^2}\,(5\cos^2 i - 1)

Apsidal rotation — the ellipse turning within its own plane. This is the death of the eccentricity vector: e was conserved only for an exactly inverse-square force, and J₂ is not one.

Figure 7.2
poledΩ/dt < 0 — westwarddΩ/dt ∝ −cos i / a^(7/2)ISS: a = 6790 km, i = 51.6°dΩ/dt = -4.97° per daya full circuit every 72 daysThe inclination does not change.Neither does the shape or thesize. Only the node walks —which is why J₂ can rotate anorbit forever but can neverbring anything down.This 72-day cycle is the ISS beta angle.
The orbital plane precessing about the polar axis. The inclination and the shape are untouched; only the node walks. For a 400 km orbit at 51.6° that walk is about 5° per day — a complete circuit in roughly seventy days.
💡Sun-synchronous orbits
Here is the exploitation, and it is genuinely elegant. An Earth-observation satellite wants every image taken in the same lighting. Same sun angle, same shadow lengths, every pass, every year — otherwise you cannot compare one image with another, and change detection is impossible. That requires the orbital plane to keep a fixed angle to the Sun. But Earth moves around the Sun at 0.9856° per day, so a fixed plane would drift out of that relationship within weeks. Unless the plane precesses at exactly the same rate. Set Ω˙=+0.9856°\dot{\Omega} = +0.9856°/day and solve for the inclination. Because Ω˙\dot{\Omega} carries a minus sign, a positive precession requires cos⁡i<0\cos i < 0 — the orbit must be retrograde. At 700 km altitude the answer is i=98.2°i = 98.2°. That is why nearly every imaging satellite ever launched — Landsat, Sentinel, SPOT, the entire commercial constellation — flies at 98-ish degrees. It is not an aesthetic choice. It is the exact inclination at which Earth's imperfection cancels Earth's orbital motion. The orbits are named for the local solar time they hold: a "10:30 descending node" satellite crosses the equator southbound at 10:30 local time on every single orbit, for its entire mission.
Figure 7.3
Sunthe flaw, turned into a toolEarth moves 0.9856°/day around the Sun.Choose i so that J₂ precesses theplane at exactly the same rate:Ω̇ = +0.9856° / day600 km → i = 97.79°700 km → i = 98.19°800 km → i = 98.60°Positive dΩ/dt needs cos i < 0 — retrograde.the plane holds a fixed angle to the Sun all year
A sun-synchronous plane through one year. The orbit precesses at just under a degree per day, so the angle between the orbital plane and the Sun direction never changes. Ground illumination at the sub-satellite point is identical on every pass.
The second exploitation comes from the apsidal rotation equation, and it is even sharper.
Look at the factor (5cos⁡2i−1)(5\cos^2 i - 1). It vanishes when cos⁡2i=1/5\cos^2 i = 1/5, that is, at
icrit=arccos⁡(±15)=63.435° or 116.565°i_{\text{crit}} = \arccos\left(\pm\frac{1}{\sqrt{5}}\right) = 63.435° \text{ or } 116.565°

The critical inclination. At exactly this angle the argument of perigee does not drift, and an eccentric orbit keeps its apogee pointed at the same latitude indefinitely.

Figure 7.4
050100150-101inclination (degrees)dω/dt (degrees / day)63.43°116.57°5 cos²i = 1Apogee drifts one way belowthe crossing, the other wayabove it, and does not driftat all on it.Molniya sits here, exactly.Off it, dω/dt ≈ 0.8°/day wouldwalk apogee almost 300° in ayear, sliding the dwell out ofthe Arctic and into the south.
Apsidal drift rate against inclination. It is positive below 63.4°, negative above, and exactly zero at the crossing. Molniya, Tundra and the Sirius radio constellation all sit on that zero.
🛰Why Molniya is at 63.4° and not 65°
Unit 3 introduced the Molniya orbit: 12-hour period, e=0.74e = 0.74, apogee over the Arctic where the satellite crawls for eight hours per revolution. The whole design depends on apogee staying over the north. If ω\omega drifts, apogee slides toward the equator and then into the southern hemisphere, and the mission is dead — the satellite spends its dwell time loitering over Antarctica. At an arbitrary inclination, ω˙\dot{\omega} for a Molniya orbit is around 0.8° per day. Over a year that is nearly 300° — apogee would end up essentially anywhere. So the Soviets placed it at exactly 63.4°, where ω˙=0\dot{\omega} = 0 by construction. Not approximately zero. Zero, to the accuracy of the J₂ model, with only the much smaller higher harmonics left to clean up. The critical inclination is a piece of pure mathematics — the root of 5cos⁡2i=15\cos^2 i = 1 — and there are satellites flying today whose inclination was chosen because of it. Sirius XM's original constellation, the Russian Tundra orbits, and every Molniya since 1965 sit on that number.
A third exploitation is quieter but just as useful. A frozen orbit exploits the small J3J_3 term against J2J_2 to hold both ee and ω\omega effectively constant, so the satellite's altitude over a given latitude never changes across the mission. Altimetry missions — TOPEX/Poseidon, Jason, Sentinel-6 — all fly frozen orbits, because measuring sea level to centimetres requires knowing your own altitude to better than that.
📐Worked Example: the ISS plane, and why the [beta angle](https://en.wikipedia.org/wiki/Beta_angle) matters
ISS: a=6790a = 6790 km, e≈0e \approx 0, i=51.6°i = 51.6°. Nodal regression. Ω˙=−32 μ J2R⊕2a7/2cos⁡i\dot{\Omega} = -\frac{3}{2}\,\frac{\sqrt{\mu}\,J_2 R_\oplus^2}{a^{7/2}}\cos i Numerator. 32μJ2R⊕2=1.5×631.35×1.08263×10−3×4.0681×107=4.1708×107\tfrac{3}{2}\sqrt{\mu}J_2R_\oplus^2 = 1.5 \times 631.35 \times 1.08263\times10^{-3} \times 4.0681\times10^{7} = 4.1708\times10^{7}. Denominator. a7/2=67903.5=2.580×1013a^{7/2} = 6790^{3.5} = 2.580\times10^{13}. Angle. cos⁡51.6°=0.6211\cos 51.6° = 0.6211. Ω˙=−4.1708×1072.580×1013×0.6211=−1.004×10−6\dot{\Omega} = -\dfrac{4.1708\times10^{7}}{2.580\times10^{13}} \times 0.6211 = -1.004\times10^{-6} rad/s. Convert: ×86400×180/π=−4.97°\times 86400 \times 180/\pi = \mathbf{-4.97°} per day. What that means operationally. The ISS plane sweeps a full circle in 360/4.97=72360/4.97 = \mathbf{72} days. That single number governs an enormous amount of station operations. The beta angle — the angle between the orbital plane and the Sun — cycles with the same period, running from about −75° to +75°. Near high beta the station is in continuous sunlight and overheats; near zero it spends 36 minutes of every 92 in shadow and the batteries work hardest. Solar array orientation, thermal planning, EVA scheduling and visiting-vehicle windows are all planned around a cycle driven entirely by the fact that Earth is 21 km fatter at the equator.
⚠What J₂ does and does not touch
The secular J₂ effects are specific, and it is worth being precise about which elements move. Drifts secularly: Ω\Omega (the node), ω\omega (perigee), and MM (a small correction to the mean motion, which is why TLE mean motion is not μ/a3\sqrt{\mu/a^3}). Does not drift secularly: aa, ee and ii. They oscillate a little within each orbit — short-period variations — but they do not accumulate. That distinction matters. aa not drifting under J₂ means J₂ cannot deorbit anything. The orbit's orientation rotates forever, and its size and shape stay put. Which is precisely why the next lesson is necessary. Something else is removing energy, and unlike J₂, it does not stop.
↓But there is a problem
So the first lie is repaid. Earth is not a point mass, the orbital plane is not fixed, and the eccentricity vector Unit 2 proved was conserved rotates steadily for every real satellite. But notice what J₂ has not done. It has changed where the orbit points. It has not changed the orbit's size. aa is untouched, so the energy is untouched, and a satellite under J₂ alone would orbit, precessing gently, until the Sun swallowed the Earth. The next perturbation is not like that. The next one removes energy. Every orbit, every pass, without pause and without any possibility of reversal. It has brought down a space station in front of the world's press, and it is currently the only thing standing between low Earth orbit and permanent saturation with debris. And the amount of it varies by a factor of ten, depending on what the Sun happened to be doing that decade.