RESEARCH & METHODOLOGY
How SkyTransit studies aircraft transits
SkyTransit investigates computational methods for detecting, predicting and analysing apparent alignments between aircraft and celestial objects as seen from ground-based observing locations. The work combines aircraft trajectory observations with astronomical ephemerides, geodesy and computational geometry. The current focus is the Sun and Moon.
“Transit” here means apparent line-of-sight geometry from a specific observer location — not a physical encounter.
Research scope
The research concerns the geometry of apparent alignments: where on the ground, and at what instant, an aircraft appears to cross the disc of the Sun or Moon for an observer at that location. It spans trajectory reconstruction from imperfect telemetry, topocentric astronomy, spatiotemporal corridor calculation, and the aggregation of historical evidence for locations where such alignments have already occurred.
Status
SkyTransit is an independently developed academic research prototype with no dedicated project funding. It is under active methodological development: the prediction and analysis methods, their predictive performance, and their real-world validation are all ongoing work. Outputs should be treated as research results from a system under evaluation, not as a formally validated forecasting or operational service. It is currently non-commercial: there is no subscription, paid tier, advertising, or commercial operation.
The platform is being developed iteratively as a working software system and a testbed for methodological research. Methods and results may be used for academic or methodological dissemination as the work matures; no specific publication is promised here.
Research questions
- Can aircraft trajectory data and astronomical ephemerides be combined to identify ground locations from which aircraft–Sun/Moon alignments occur?
- How robust are predicted transit corridors to trajectory sampling, timing, positional uncertainty and observation geometry?
- Can historical aircraft trajectories reveal geographic locations where Sun/Moon alignments have occurred repeatedly?
- How can live and historical ADS-B evidence be combined without counting the same physical event twice?
- How should historical occurrence be represented without incorrectly presenting it as a probability or guarantee of a future transit?
Methodology
The pipeline runs from raw observations to cross-checked, aggregated results. Each stage is deliberately conservative: any candidate that cannot be proven rejectable falls back to the full scientific calculation. Predictive performance and real-world validation are still being evaluated (see Status and Limitations).
Aircraft trajectory observations
Bounded sets of ADS-B observations for the region and time window under analysis.
Trajectory reconstruction
Observed positions are reconstructed into continuous tracks with associated uncertainty.
Topocentric Sun / Moon ephemerides
Apparent solar and lunar positions computed for the observer location and instant.
Closest-approach geometry
Line-of-sight angular separation between aircraft and target is minimised in space and time.
Ground transit corridor
A spatiotemporal corridor: the ground band from which the alignment is visible, with its own time window.
Observed / historical evidence
Candidate past alignment events are accumulated and grouped geographically across dates and datasets.
Validation and aggregation
Results are cross-checked against reference computations; recurrence is described, never projected as probability.
Aircraft observations
Live and historical ADS-B observations are retrieved for a bounded region and time window. SkyTransit treats callsign, operator, flight number and route as separate evidence-backed fields and never a filed flight plan.
Trajectory reconstruction
Discrete position reports are reconstructed into continuous tracks. Short-term motion used for forward prediction is an explicit kinematic extrapolation from the observed state, not scheduled or planned routing.
Astronomical ephemerides
Solar and lunar positions come from the JPL DE440s ephemeris via Skyfield. Satellite geometry, where relevant, uses CCSDS GP/OMM elements propagated with SGP4 from preserved immutable snapshots.
Topocentric geometry
Apparent target directions are computed for the observer's location and instant, including parallax and local horizon. Observer-dependent results are never shared across observers.
Transit corridor calculation
The ground band from which an alignment is visible is derived as a spatiotemporal corridor with its own time window; the transit occurs at a different instant along its length.
Event qualification
A candidate is only qualified as a transit when angular separation, timing and geometry criteria are met. Method parameters are currently under evaluation and may change as validation proceeds.
Historical evidence
SkyTransit Atlas analyses bounded historical trajectory sets to identify past qualified alignment events at ground locations.
Aggregation and recurrence
Repeated evidence is grouped geographically across different dates and datasets. Recurrence is described as observed evidence, not converted into a probability or a guarantee of a future transit.
Validation
Optimised computation is checked against reference results on deterministic fixtures — agreement on opportunity count, object identity, target, event time within tolerance, corridor existence and exact-place classification. Scientific validation of the overall method is ongoing and not claimed as complete.
SkyTransit and SkyTransit Atlas
SkyTransit
Forward-looking analysis
- computes near-term candidate alignments;
- identifies relevant observing geometry;
- supports timing and location workflows;
- may alert users to candidate future events.
SkyTransit Atlas
Historical analysis
- analyses past aircraft trajectories;
- identifies qualified historical alignment events;
- accumulates evidence across datasets and dates;
- groups repeated evidence geographically.
Historical evidence is not presented as a future probability, a guarantee, a “best location”, or a prediction-confidence score.
Explore SkyTransit AtlasData sources and provenance
SkyTransit uses third-party aircraft trajectory data as an input to geometric and spatiotemporal analyses. Aircraft-data queries are bounded geographically and temporally according to the analysis being performed. Raw third-party historical datasets are not intended for public redistribution through the platform. Public outputs are derived results produced by the SkyTransit methodology: reconstructed trajectories, candidate transit events, ground transit corridors, aggregated geographic evidence, and timing and location information.
Aircraft trajectory data
Aircraft trajectory data may include observations obtained from The OpenSky Network, subject to the applicable OpenSky access conditions and terms. SkyTransit is not an OpenSky partner and this does not imply any endorsement. Live development currently uses a public ADS-B feed (adsb.fi) for non-commercial development only; any use of historical aircraft datasets is planned and subject to authorisation, and is not active today.
Live development telemetry is currently provided by adsb.fi (public API, attribution required, non-commercial development use). Production provider selection remains open. Learn more about The OpenSky Network.
Other inputs
- Astronomical ephemerides — JPL DE440s (Sun/Moon), CelesTrak GP/OMM for satellites.
- Geospatial reference data — used for places, corridors and routing feasibility.
- User-defined observing regions and saved locations, which remain private by default.
No raw redistribution
Raw third-party historical aircraft datasets are not intended to be exposed through SkyTransit as a public bulk-download service. The platform publishes derived outputs: reconstructed trajectories, qualified transit events, ground transit corridors, aggregated geographic evidence, and timing and location information.
Limitations
- Predictive performance has not been formally validated against independent real-world outcomes; this is ongoing work, and results should not be read as an operational forecast.
- ADS-B sampling is imperfect and heterogeneous in coverage and rate.
- Trajectory reconstruction carries positional and temporal uncertainty.
- Predicted aircraft motion is a short-term kinematic extrapolation and can change.
- Line-of-sight alignment depends sensitively on timing and the observer's exact location.
- Historical observations do not imply a probability of future recurrence.
- Candidate and historical locations are evidence-derived, not guaranteed observing spots.
- Aircraft operations and routing can change at any time.
Research lead / developer
Alejandro Rodríguez-González
Professor / Researcher · Universidad Politécnica de Madrid (UPM)
SkyTransit is an independent research-oriented software initiative developed by Alejandro Rodríguez-González. The university affiliation identifies the researcher's academic affiliation only; SkyTransit is not an official UPM service or an institutionally funded project.
SkyTransit is free to use while it is a research prototype.
Explore SkyTransit