Solar System Simulator is built to be accurate enough to learn from and to check against the real sky. This page explains what it computes, how we test it, and what its limits are.
How positions are computed
All positions are computed in your browser for the exact moment on screen. Nothing is pre-recorded, and nothing about the scene is sent to a server.
- Planets and Pluto come from Astronomy Engine by Don Cross (MIT license), which implements the VSOP87 planetary theory and a dedicated model for Pluto.
- The Moon comes from Astronomy Engine's lunar theory.
- Jupiter's Galilean moons (Io, Europa, Ganymede and Callisto) come from Astronomy Engine's implementation of the L1.2 theory.
- The other moons (Phobos and Deimos; Mimas, Enceladus, Tethys, Dione, Rhea, Titan and Iapetus; Miranda, Ariel, Umbriel, Titania and Oberon; Triton; and Charon) use JPL's published mean orbital elements with the precession of their orbits. Mean elements are not a precise ephemeris, so we fitted each moon's position along its orbit to JPL Horizons over 2016 to 2036.
- Orientation of the Sun, Moon, planets and Pluto (the tilt of each axis and how far each has turned) follows the IAU Working Group on Cartographic Coordinates and Rotational Elements (2015). Moons without a published model are drawn facing their planet, which is true for all the large moons shown.
Positions are in the J2000 ecliptic frame: the plane of Earth's orbit, with the vernal equinox of the year 2000 as the reference direction. They are geometric, so they show where a body is at that instant rather than where it appears once its light has crossed the distance to us.
How we test it
We compare the simulator's positions with JPL Horizons, NASA's reference system for solar system positions. The comparison runs automatically every time the site is built, so a change that made the simulator less accurate would stop the release.
| Bodies | Span tested | Agreement with Horizons |
|---|---|---|
| Mercury to Neptune, and Pluto | 1800 to 2100 | Within 30 arcseconds in direction and 0.01% in distance at every sample |
| The Moon | 2020 to 2030 | Within 5 arcseconds and 25 km |
| Io, Europa, Ganymede, Callisto | 2024 to 2028 | Within 700 km of their true positions around Jupiter |
| Other moons | 2021 to 2031 | Most within about 2 degrees around their planet; Mimas, whose orbit librates under Tethys's pull, within about 12 degrees |
For scale, 30 arcseconds is about one sixtieth of the width of the full Moon in the sky: far too small to notice by eye, so every planet sits in the right place among the stars on the right night.
We also check events against known dates. The simulator's eclipse search finds the total solar eclipses of August 12, 2026 and August 2, 2027, and its list of events always comes back in date order.
Earth-observed coordinates and time
The Sky inspector uses geocentric apparent coordinates in the true equator and ecliptic of the chosen date. Planet positions include light travel time and aberration; the Moon uses Astronomy Engine's geocentric lunar model. Precession and nutation are included. Ecliptic longitude starts at the equinox of date. Retrograde means that longitude is decreasing, measured over a short interval around the selected moment.
Altitude and azimuth use the observer's latitude and longitude, including parallax, at sea level without atmospheric refraction. The sky diagram shows body centres above the geometric horizon. It does not predict whether a planet can be seen through daylight, clouds or local obstructions.
We check these coordinates against 120 separately downloaded JPL Horizons samples: ten bodies at twelve dates across 1800–2099. Equatorial and ecliptic directions agree within one arcminute at every sample; distances agree within 0.02%. The largest sampled directional difference is about 20 arcseconds. This check is distinct from the geometric 3D position tests above.
For dates from 1972, UTC is converted to Terrestrial Time using the NASA/JPL leap-second table. Future dates hold the last announced offset, matching Horizons' convention; future leap seconds are not known. Earlier dates use a historical ΔT model. Earth rotation uses UTC as an approximation to UT1. The downloadable sky data identifies these conventions; neither display is intended for precision timing or navigation.
The sky and the look of each world
- Stars: 8,920 stars down to magnitude 6.5, the limit of the naked eye at a dark site, from the HYG Database version 4.1 (CC BY-SA 4.0). Their colours come from each star's measured B−V colour index.
- The Milky Way is the Solar System Scope panorama, placed by converting every direction on screen to galactic coordinates. We confirmed the alignment against the positions of the 50 brightest stars.
- Surfaces: textures from Solar System Scope (CC BY 4.0), which are based on NASA imagery. Earth has separate day, city-light and cloud maps; the clouds drift slowly as a visual cue and are not a weather forecast.
- The Sun's activity: sunspots, bright plage, prominences (dark filaments where they cross the disk), flares and the corona. How much of it there is follows the real solar cycle for the date, from WDC-SILSO's dates and sizes of each cycle's minimum and maximum: a quiet Sun in 2008, a busy one in 2014 or 2024, with spots nearer the equator late in each cycle. Where each spot and prominence sits, and when a flare goes off, is illustrative, not the Sun of that day. The colours are the false colours NASA gives its Solar Dynamics Observatory pictures: an orange surface and a golden corona. To the eye, the Sun is white.
- Physical data (diameters, masses, gravity, day lengths, tilts and temperatures) come from the NASA Planetary Fact Sheets, and moon sizes from JPL's satellite physical parameters.
- Shadows are computed for every point on each body: how much of the Sun's disc is hidden by each nearby sphere. That produces partial and total eclipses with soft edges, and the red glow of the Moon in Earth's shadow.
The other simulators use different models
| View | What it computes | Main limit |
|---|---|---|
| Solar system | Positions from published orbital theories at the chosen date | Tested date span and approximate models for some moons |
| Gravity sandbox | Newtonian motion from editable starting conditions | No relativity; collisions merge bodies |
| Black hole | Light paths around up to four non-rotating black holes; Newtonian orbits with gravitational-wave losses | No spin; approximate light paths between two close black holes; simple disc, gas and jet models; hovering camera |
| Galaxy | Two galaxy cores under Newtonian gravity, carrying 12,000 stars that have no mass of their own | Stars don't pull on each other; no dark matter halos, so the cores never merge; no gas or new stars |
The gravity checks cover reference orbits, conservation and collision behavior. The black-hole checks cover light capture, weak deflection and selected analytic limits. Passing these numerical checks does not make an arbitrary sandbox scene an astronomical prediction.
Known limits
- The date range is January 2, 1800 to December 30, 2099, the span we tested.
- Moons other than Earth's Moon and the Galilean moons are approximate: right orbit, right period, position within a few degrees.
- Surface textures of Venus, the gas giants and some regions of other worlds are artistic reconstructions of real imagery, not current photographs.
- The simulator is an educational tool. It is not meant for spacecraft navigation, precise astrometry or eclipse timing to the second.
If you find something that looks wrong, write to hello@solarsystemsimulator.com with the link from your address bar, which records the body and the moment you were looking at.