Accuracy and limitations
The models here are design-grade engineering approximations. Each states its assumptions, units, reference frame, and accuracy limits. They are built for constellation trade studies and mission design.
They are not certified for operational use, and this page is the one to read before a number from this tool reaches a design review or a regulator.
Suitable for
- Sizing a constellation against coverage, revisit, and latency requirements.
- Comparing design options in a trade study where relative differences matter more than absolute truth.
- Closing a first-order link budget against a stated availability target.
- Producing CCSDS products for downstream tools that will apply their own higher-fidelity processing.
- Teaching and research where assumptions are stated and results are reproducible.
Not suitable for
Do not use these outputs for collision avoidance, conjunction assessment decisions, re-entry prediction, operational maneuver execution, regulatory filings that require certified propagation, or any decision where a wrong answer damages hardware or people.
Conjunction screening exists in the tool as a design-stage filter for assessing whether an architecture creates conjunction pressure. It is not an operational conjunction assessment service and must not be used as one.
Where the error comes from
Error accumulates from four independent places, and they do not trade off against each other. A high-fidelity propagator fed a stale two-line element set is still wrong.
| Source | Typical dominant effect | What reduces it |
|---|---|---|
| Propagation model | Unmodeled-force effects: oblateness, drag, solar radiation pressure, third bodies | Choose a higher-fidelity model; supply a real ballistic coefficient |
| Initial state | Epoch age and observation quality of the source element set | Use fresher elements; use precise ephemerides where available |
| Environment | Atmospheric density is uncertain by tens of percent and varies with solar activity | Accept the uncertainty; treat drag-dominated results as a range, not a value |
| Numerics | Integrator tolerance and step size | Tighten tolerance; verify convergence by halving the step |
Propagation error in low Earth orbit grows roughly along-track and roughly linearly with time for a fixed model error, so a position error that looks acceptable at one revolution can be unusable at fifty.
Choosing a fidelity level
| Model | Includes | Reasonable use |
|---|---|---|
| Two-body | Point-mass Earth only | First look, geometry intuition, teaching |
| Secular J2 | Nodal regression and apsidal rotation | Constellation geometry over days to weeks, sun-synchronous design |
| SGP4/SDP4 | The perturbation model matching public TLEs | Working with catalogue objects, and only with TLEs |
| Numerical Cowell | Zonal harmonics, drag, solar radiation pressure, luni-solar | Design reviews, maneuver planning, anything quantitative |
| Ephemeris interpolation | Whatever produced the supplied table | Reproducing an externally supplied trajectory |
SGP4 is only meaningful when the initial state came from a TLE. TLEs are fitted to the SGP4 model, so feeding SGP4 a state vector from another source produces a confidently wrong answer rather than an error.
Reporting a suspected error
If a result contradicts an independent tool, the useful report includes the inputs, the model selected, the epoch, the frame, and the magnitude and direction of the disagreement. “The orbit looks wrong” cannot be investigated.
Open an issue through the feedback link at the foot of any page.
main (pre-release)