Research prototype Browser-side analysis Detector method 0.4 Not operational SSA

Satellite Maneuver Screening Workbench

Retrieve or validate orbital-element history and screen for unusual catalog-level changes using SGP4 common-epoch comparison, multivariate robust baselines, orbit-specific profiles and conservative abstention rules.

A reported candidate is a cue for follow-up analysis. It does not confirm a burn, infer intent or replace orbit determination.

Run demonstration How to use

Choose an analysis path

Run the built-in demonstration, retrieve real historical data, or provide your own TLE or OMM history.

Full usage guide
1
Choose a pathDemonstration, real history, or your own file.
2
Run the analysisDemos run automatically; real data loads for the selected dates.
3
Read the resultReview the event, data quality, charts and limitations.

Built-in demonstration

Uses a synthetic history, runs immediately and does not contact any external data service.

Demonstration analysed. The result is ready below.

Paste or import orbital-element history
Drop a TLE, TXT or OMM JSON file here, or choose a file.
Analyse a real satellite using historical data

Load historical orbital data

Retrieve OMM history through the same-site endpoint, or continue using pasted and imported data. Basic SSA data is supplied by Space-Track.org, operated by USSPACECOM, and stored persistently by date interval on this server. Credentials remain on the server and are never sent to the browser.

Historical-data guide · Troubleshooting

Current CelesTrak snapshot

CelesTrak provides the current GP record as OMM JSON. One snapshot cannot support screening, but it can replace the input or be appended safely to an existing history. Responses are cached locally for two hours.

Analysis result

No analysis yet

Load an example, import a file or paste orbital-element history, then run the analysis.

Object
n/a
Period analysed
n/a
Valid snapshots
n/a
Data quality
n/a
Result
n/a
Strongest epoch
n/a
Observed Δa
n/a
Candidate events
n/a

Orbital change evidence

The two views share the same time range. Oversized sampling gaps are excluded from calibration and screening.

Accessible chart data

Tabular values underlying the two charts. Candidate observations are identified in the final column.

Per-epoch orbital-change evidence
Epoch UTCSemi-major axisΔaGapΔa/ΔtRadialAlong-trackCross-trackStatus
Run an analysis to populate the accessible chart data.
Orbit context, loaded on demand

Approximate three-dimensional comparison of element sets immediately before and after a selected candidate. This visual context is rendered locally with Plotly and is not operational orbit propagation.

Run an analysis, then open this section to load the 3D context.

Export

Export a lightweight evidence package for follow-up analysis, reports or review discussions.

Other exports
Technical details

Data quality

Results are interpreted only when the history is internally consistent and provides at least five sequential evaluation points.

Waiting for analysis
n/a

Method metadata

Detector method
0.4
State comparison
SGP4 at a common epoch
Signals
RTN residuals and element changes
Calibration
rolling robust multivariate baselines
Output
event-level screening
3D comparison
visual context only
Checking SGP4 dependency. Plotly charts and 3D context load on demand.

Method in brief

  • Accepts checksum-valid TLE pairs or validated OMM JSON records and can retrieve authenticated Space-Track GP history through a same-site server endpoint.
  • Propagates consecutive element sets to the later epoch with SGP4 and computes radial, along-track and cross-track residuals.
  • Combines state residuals with Δa/Δt, mean-motion, eccentricity-vector, inclination-vector and B* changes.
  • Uses orbit-specific signal floors and rolling robust empirical baselines fitted only to preceding eligible pairs.
  • Groups adjacent anomalous epochs into candidate events and abstains when identifiers, data quality, propagation or history are inadequate.

Limits and follow-up

  • A candidate event groups one or more eligible multivariate breaches that pass the configured element-change or corroborated state-residual gates.
  • A flag does not confirm a burn, command, intent or operational purpose.
  • Weak, Moderate and Strong are internal event-screening tiers, not calibrated probabilities.
  • Oversized epoch gaps are excluded rather than compared directly with dense sampling intervals; SGP4 and element-space signals are calibrated only from preceding eligible pairs.
  • False positives can still arise from catalog updates, drag effects, noisy element fitting or unmodelled operational context.
  • Follow-up should include source review, orbit determination, additional catalog data and mission-context checks where available.
Run and reproducibility details

The run identifier, detector settings and audit record support reproducibility and troubleshooting.

Developer validation tools

Run deterministic checks against synthetic candidate, quiet, sparse and malformed histories. These are software regression checks, not an operational performance benchmark.

Public-evidence benchmark statusAccessibility statement
RankObjectProfileClassificationScoreQuality
Run ranked demo screening to populate this table.
Run a labelled event benchmark

Import a benchmark JSON file containing orbital histories, documented event intervals and source citations. The public manifest route is a small v0.1 runner-format seed, not the frozen publication benchmark v1.0. The workbench reports event-level precision, recall, F1, false events per object-month and first-alarm detection delay. No operational performance claim is made unless you supply a suitable ground-truth dataset.