3.4.1 · D5Rocket Flight Mechanics
Question bank — Coordinate systems — Earth-Centered Inertial (ECI), Earth-Centered Earth-Fixed (ECEF), North-East-Down (NED), launch, bo
Before you start, hold three pictures in mind:
- ECI = axes nailed to the distant stars (they never turn).
- ECEF = axes glued to the crust (they spin once per day with the planet).
- NED = a little tripod planted on the ground at your feet: one leg North, one East, one straight Down into the dirt.
True or false — justify
TF1. In ECEF a rocket sitting on the pad has zero velocity, therefore zero acceleration.
False — in ECEF its coordinate velocity is zero, but ECEF spins, so the pad is being swung in a circle; it has real centripetal acceleration pointing at the spin axis.
TF2. Because a rotation matrix satisfies , transforming a velocity from ECEF to ECI is just .
False — that formula only moves position. Velocity picks up the extra transport term, , because the frame itself is turning while you differentiate.
TF3. NED and ECEF share the same origin.
False — ECEF's origin is Earth's center; NED's origin sits on the surface at your current point, so going between them needs a translation as well as the rotation.
TF4. Since Earth spins about , the ECI→ECEF change is a rotation about the shared axis by a single angle .
True — the two frames share origin and spin axis, so they differ only by that one Greenwich angle; nothing tilts, only and swing.
TF5. Down in NED always points to the geometric center of the Earth.
Roughly, but strictly Down points along the local vertical (opposite the geodetic normal on the reference ellipsoid), which only aims exactly at the center at the equator and poles — see Geodetic vs Geocentric Latitude.
TF6. The launch frame is inertial, so it is safe to use it for a mission of any duration.
False — it is approximately inertial only for a short flight; over hours Earth rotates out from under it and the neglected rotation errors grow, so you must fall back to true ECI.
TF7. Rotation matrices for yaw, pitch and roll can be multiplied in any order and give the same body attitude.
False — finite rotations do not commute, so ; the 3-2-1 (yaw→pitch→roll) order is a convention that fixes what each angle means, see Rotation Matrices and Euler Angles.
TF8. Every frame in the chain is right-handed.
True by construction — each is defined right-handed (e.g. NED: North×East = Down), which keeps for every transform and lets you compose them without sign surprises.
TF9. The eastward launch bonus of ~465 m/s comes from thrust, not geometry.
False — it is purely the surface speed you already have from Earth's spin, ; you inherit it for free before the engines even help.
TF10. If Earth stopped spinning, ECI and ECEF would become the same frame.
True in effect — with the Greenwich angle freezes, so the rotation between them becomes a fixed (or identity) matrix and the transport term vanishes.
Spot the error
SE1. "I'll integrate in ECEF because the pad is fixed there, then read off latitude directly."
The error is treating ECEF as inertial; it rotates, so needs added Coriolis and centrifugal terms — integrate in ECI and transform only for output.
SE2. "East is just , so at the North pole East points along ."
At the pole is parallel to , so their cross product is the zero vector — East is undefined there; NED degenerates at the poles, which is a genuine gimbal-like singularity, not a bug in your algebra.
SE3. "The ECEF→NED matrix has no longitude in the East row, so somebody forgot a term."
No error — look at Row 2 of the matrix above: the East unit vector's ECEF components depend only on longitude (which is there) and never on latitude , because moving north–south doesn't change which way "East" tilts.
SE4. "To go body→ECI I multiply the same matrices in the same order I used for ECI→body."
Wrong — reversing a chain of rotations requires transposing and reversing order: ; keeping the order gives a completely different, invalid rotation.
SE5. "I dropped the term because it's small."
At the surface it is m/s at the equator — far from negligible; dropping it silently throws away the entire rotational launch advantage and corrupts orbit insertion. See Rotating Reference Frames and Coriolis Force.
SE6. "Latitude in the NED matrix is the same as the angle from Earth's center to my position."
Not generally — the map uses geodetic latitude (angle of the geodetic normal / local vertical to the equatorial plane), which differs from the geocentric angle by up to ~0.19° because Earth is an ellipsoid, per Geodetic vs Geocentric Latitude.
SE7. " points at the Sun."
No — points to the vernal equinox, a fixed direction among the stars; the Sun moves around the ecliptic through the year, so it can't anchor an inertial axis.
SE8. "I entered a launch site 120° West as in the NED matrix and got the East vector pointing the wrong way."
Sign-convention error — on this page East longitude is positive, so 120° West must be entered as ; feeding the wrong sign flips and mirrors the East row, a classic quadrant trap.
Why questions
WHY1. Why can't a single coordinate frame do every job in rocket flight?
Different tasks want different "still" points — Newton's law needs a non-rotating frame (ECI), a fixed pad wants an Earth-fixed frame (ECEF), the horizon wants a local frame (NED), and thrust wants the rocket's own axes (body).
WHY2. Why is ECI, not ECEF, the frame we integrate the equations of motion in?
Only ECI is genuinely non-rotating and non-accelerating, so holds with the real forces alone — no fictitious Coriolis or centrifugal terms to add, see Rocket Equations of Motion.
WHY3. Why does only one angle relate ECI and ECEF?
They share the same origin and the same spin axis , so the only freedom left is how far and have swung around that axis — captured entirely by the Greenwich angle .
WHY4. Why do we launch eastward and from near the equator?
Eastward aligns the rocket with Earth's existing surface velocity so you keep that free speed; near the equator that speed is maximum because you are farthest from the spin axis, see Launch Azimuth and Orbital Inclination.
WHY5. Why does the transport term appear when transforming velocity but not position?
Position is a single snapshot, so only the axes' orientation (the matrix mapping ECI→ECEF) matters; velocity is a rate, and while you take the derivative those axes are themselves turning (), which injects the correction.
WHY6. Why does aerospace prefer the body frame for expressing thrust and IMU readings?
Those quantities are physically tied to the vehicle — thrust points out the nose, gyros/accelerometers are bolted to the structure — so their components are constant and simple in the body frame, then rotated out via attitude.
WHY7. Why might one switch from Euler angles to quaternions for attitude?
Euler angles hit singularities (gimbal lock) where two axes align and one angle becomes undefined; quaternions have no such gap and interpolate smoothly, see Quaternions for Attitude.
WHY8. Why is "Down" defined as rather than choosing "Up"?
Convention picks a right-handed North-East-Down set so that ; choosing Up would flip handedness and force sign changes throughout the transform chain.
Edge cases
EC1. What happens to the NED frame exactly at the North or South pole?
The North and East directions become ambiguous because "which way is North" is undefined at the pole — the local tangent plane loses its reference, so NED is singular there and must be handled specially.
EC2. What is the eastward rotation bonus for a launch from a pole ()?
Zero — the factor , since a point on the spin axis doesn't move; polar launches get no rotational head-start, which is one reason equatorial sites are prized.
EC3. At , what does the ECI→ECEF rotation matrix reduce to?
The identity — with , and , so ECI and ECEF momentarily coincide before Earth turns them apart.
EC4. If a spacecraft is at the exact center of the Earth (a thought experiment), why does NED break down?
The radial direction is undefined at , so Down, and hence the whole North-East-Down tripod, cannot be constructed — the frame needs a nonzero position vector to exist.
EC5. Over one full sidereal day, how does the ECI→ECEF angle behave, and what does that imply for reuse?
sweeps a full and then the frames realign, so the transform is periodic; you can't treat any single frozen matrix as valid across that span — you must recompute continuously.
EC6. For a very short flight (seconds to a couple minutes), why is treating the launch frame as inertial acceptable?
In that window Earth turns by a tiny angle ( is minuscule), so the neglected rotation and transport corrections stay below trajectory tolerances — the approximation is safe only because the elapsed time is small.
EC7. What does evaluate to for an object sitting exactly on the spin axis?
The zero vector, because is parallel to there and the cross product of parallel vectors vanishes — consistent with a point on the axis having no rotational speed.
Recall wrap-up
Recall Quick self-test
Which frame is truly inertial, and why does that matter for integration? ::: ECI — its axes neither rotate nor accelerate, so holds with real forces only, making it the correct frame to integrate the equations of motion. Name one place where NED is undefined. ::: At the poles (North/East ambiguous) and at Earth's center (radial direction undefined). One sentence: why velocity transforms differently from position between ECI and ECEF. ::: Because the frame is rotating during the derivative, adding the transport term on top of the plain rotation. What is the sign convention for and on this page? ::: positive North, positive East of Greenwich — the right-handed choice used in every formula here.