Skip to content

Vallado (2010): ICATT Astrodynamics Tutorial

The tutorial is structured around a single end-to-end problem: “Determine when you can see a satellite from a ground site.” This mirrors Astrolock’s exact pipeline:

  1. Time systems — converting between UTC, UT1, TAI, TT, and GPS time
  2. Coordinate frames — TEME, J2000, ITRF, geodetic, topocentric
  3. Propagation — Kepler’s equation, perturbation theory, SGP4
  4. Orbit determination — observations to orbital elements
  5. Observation geometry — altitude, azimuth, range, visibility constraints

Where the individual papers in this archive each address one piece of that pipeline, the tutorial shows how the pieces connect. It is a pedagogical bridge between the foundation papers and the textbook.

The 120 slides map directly to chapters in the textbook:

SlidesTopicTextbook Chapter
1—7Introduction
8—20Time and coordinate systemsCh. 3
21—49Kepler’s equation and two-bodyCh. 1—2
50—80Perturbations and propagationCh. 8—9
81—108Orbit determination and estimationCh. 10
109—120ApplicationsCh. 11

The weighting is telling: perturbations and propagation (31 slides) and orbit determination (28 slides) together consume half the tutorial. These are the two domains where SGP4 operates.

Slides 50—80 cover the same forces that SGP4 models:

  • Gravity harmonics (J2J_2 through J5J_5): zonal, sectoral, and tesseral harmonics with clear visualizations (slides 53—56)
  • Nodal regression and apsidal rotation: visual demonstrations of how oblateness rotates the orbital plane and the line of apsides (slides 57—59)
  • Atmospheric drag: force model and altitude-dependent effects
  • Third-body perturbations: Sun, Moon, and planetary gravitational influences
  • Solar radiation pressure: area-to-mass ratio effects on orbit evolution

Slide 52 presents the decomposition of perturbation effects into three categories:

δ(element)=δsecular+δlong-periodic+δshort-periodic\delta\text{(element)} = \delta_\text{secular} + \delta_\text{long-periodic} + \delta_\text{short-periodic}

This is the same three-stage structure that Brouwer (1959) derived analytically and that the SGP4 FORTRAN implements as three sequential code blocks:

Perturbation TypePeriodSGP4 FORTRAN Lines
SecularCumulative29—52
Long-periodicMonths53—72
Short-periodicOne orbit120—175

The tutorial’s visual treatment of this decomposition is the clearest available diagram of how these terms combine — more accessible than the formal derivation in the original papers.

Slide 51 shows which perturbation forces matter at which orbital altitudes. This is the physical justification for two key design decisions in the Spacetrack propagators:

  • SGP4’s drag model dominates at LEO (<2000 km) where atmospheric density is significant, but is negligible for higher orbits
  • SDP4’s third-body perturbations (Sun and Moon) become the dominant non-J2J_2 force for deep-space orbits (period \geq 225 minutes)

The 225-minute threshold that the STR#3 DRIVER uses to select between SGP4 and SDP4 sits at the crossover point on this chart.

Slides 81—108 cover the statistical orbit determination pipeline, progressing from raw observations to refined orbital elements:

SlidesTopicKey Concept
81—99Observation types and errorsBias, noise, drift in radar and optical data
100Observation modelingTransforming measurements to state-space quantities
101Statistical foundationsMultidimensional probability distributions
102Covariance matricesP=(ATWA)1\mathbf{P} = (\mathbf{A}^T \mathbf{W} \mathbf{A})^{-1}
104Least squaresStep-by-step matrix inverse approach
105Sequential batch least squaresBayesian update for incorporating new observations
106—107Extended Kalman filterFull prediction-update cycle
108Fit spans and averagingConnection to mean elements (what TLEs contain)

This OD material complements the AIAA 2008-6770 paper by providing the theoretical foundation in an accessible format. Where the 2008 paper gives the full mathematical treatment with operational results, these slides provide the conceptual framework needed to follow that paper’s derivations.

Slide 108’s treatment of fit spans and mean element averaging is particularly relevant for understanding why TLE accuracy degrades with age — the mean elements are a best-fit average over the observation span, not an instantaneous state.