Ground Track & Pass Predictor

SGP4 propagation (satellite.js) from a NORAD TLE; topocentric look angles against an elevation mask give AOS/LOS/TCA per pass. All in-browser, all times UTC. Accuracy is bounded by TLE freshness - paste a recent element set.

This satellite pass predictor propagates a NORAD TLE with Simplified General Perturbations 4 (SGP4) via satellite.js - a faithful port of the reference implementation from "Revisiting Spacetrack Report #3" (Vallado et al. 2006) - and computes every ground station pass in your window: acquisition and loss of signal (AOS/LOS), time of closest approach, maximum elevation, and the rise/set azimuths to point an antenna, plus the ground track drawn on a world grid. Everything runs in your browser; nothing phones out. Accuracy is bounded by TLE freshness above all - for real scheduling, paste a current element set from CelesTrak or Space-Track.

Everything here starts from a Two-Line Element set (TLE) - the compact NORAD element format that encodes a satellite's mean orbital elements at a reference epoch. TLEs are mean elements fitted to the SGP4 (Simplified General Perturbations 4) analytical propagator, so they only mean something when propagated with SGP4 - feeding TLE elements into a generic Keplerian propagator gives kilometres of error immediately. This tool runs the genuine article: satellite.js, a faithful port of the reference SGP4/SDP4 implementation from "Revisiting Spacetrack Report #3" (Vallado et al. 2006).

SGP4 outputs position in the TEME (True Equator, Mean Equinox) inertial frame. To place the sub-satellite point on the map, the tool rotates TEME into the Earth-fixed ECEF (Earth-Centered, Earth-Fixed) frame using Greenwich Mean Sidereal Time (GMST) at each sample, then converts to geodetic latitude/longitude on the ellipsoid. Pass geometry uses the standard topocentric transform (Vallado §4): the station's look angles - azimuth and elevation - are computed from the ECEF vector between station and satellite. A pass is a contiguous interval with elevation at or above your elevation mask; AOS (Acquisition of Signal) and LOS (Loss of Signal) are the rising and falling mask crossings, interpolated between samples, and TCA (Time of Closest Approach) - the elevation peak - is refined numerically inside the pass.

What bounds the accuracy: TLE freshness, more than anything else. A fresh element set is good to roughly a kilometre; the error grows about 1-3 km per day away from epoch, so pass predictions from a week-old TLE can drift by tens of seconds, and a month-old one is geometry practice, not a schedule. The tool flags a stale epoch but can't fix it - paste a current TLE from CelesTrak or Space-Track before committing an antenna. Drag is modelled only as far as SGP4's own B* term goes (no density weather, no storms), maneuvers are invisible until they show up in the next published element set, and the bundled example TLEs are deliberately historical - fixed epoch, fixed window - so the examples are reproducible rather than current.

The map is deliberately a grid-only equirectangular plot - a ground track, a station marker, and the station's visibility circle (the locus where the satellite clears the mask, from standard coverage geometry). No basemap tiles: nothing on this page phones out, and a graticule is all a ground track needs to be read correctly. This is the same physics family as the orbit-propagator and ground-station-pass tools the AI uses server-side.