How this is computed
Where every number on this site comes from, what it was checked against, and what is approximate.
Nothing here is copied from another prediction service. Moon, Sun and planet positions come from NASA JPL's DE431. The site works out where and when each event is visible, city by city. Your browser does the same for any place you pick. Other predictions are used only to check the results.
The inputs
| What | Source |
|---|---|
| The Moon, the Sun and the planets | NASA JPL's DE431 planetary ephemeris, read with the Skyfield library |
| Jupiter's and Saturn's moons | JPL's satellite ephemerides jup365 and sat441 |
| The Moon's mountains and valleys | Topography from NASA's Lunar Reconnaissance Orbiter laser altimeter (LRO LOLA), LDEM_64: a height every 1/64° of lunar longitude, about 470 m |
| Which way the Moon faces (libration) | NAIF's lunar orientation kernels (the DE421 principal-axes frame) |
| Stars | ESA Gaia DR3 to G 9.5 along the Moon's path, Hipparcos-2 for the brightest and those Gaia cannot solve, names from the Yale Bright Star Catalogue: 30,166 stars, each moved to the event date. For asteroids: Gaia DR3 to G 12.5 over the whole sky (about five million stars), and to G 14 for the finder charts |
| Asteroids | NASA JPL's Small-Body Database: the orbits and diameters of the 4,511 asteroids of 15 km and more, and each orbit's uncertainty |
| Asteroid shapes | DAMIT, the Database of Asteroid Models from Inversion Techniques (Astronomical Institute of Charles University; Ďurech, Sidorin & Kaasalainen 2010), CC BY 4.0: asteroid shapes, spin axes and periods from light curves |
| Planets' poles and rotation | The IAU Working Group on Cartographic Coordinates and Rotational Elements |
| Earth's rotation (ΔT) | Skyfield's tables and their projection forward: ΔT ≈ 69 s in 2026 |
| The Great Red Spot's longitude | Measured, not predicted: 91° (System II) on 1 June 2026, drifting 1.75° a month. Source: Sky & Telescope, from the JUPOS drift record |
| Crater names and sizes | The IAU Gazetteer of Planetary Nomenclature (USGS), for the outlines on the Moon diagrams |
| Maps and cities | Natural Earth outlines with India's official borders; city positions from GeoNames and a curated list of Indian cities |
How each page is made
Lunar occultations
The 25 events for 2026–2029 were picked by hand from a DE431 scan of every time the Moon covers a planet or bright star. For each, the engine maps where on Earth it is seen and traces the graze limits. For each city it finds the contact times, then corrects them for the Moon's uneven edge. That moves them by a few seconds at most. Each page carries a 2 KB description of the event, from which your browser solves any place, light time and aberration included.
When the Moon rises or sets during an occultation, a place sees only half of it; the page says which half.
The Moon and the stars
Each month, September 2026 to December 2028, is one file: the Moon and Sun as daily curves, and the stars the Moon can cover. Your browser finds every occultation for your location. Each instrument has a rule of thumb for how faint a star can be at the Moon's dark and bright edges. Moonglare and twilight lower it, and the Moon must be 5° up. Each page prints the rule.
Asteroid occultations
Each month, September 2026 to August 2027, the site's own integrator moves all 4,511 asteroid orbits forward and tests every Gaia star in reach. It includes the planets and Moon from DE431, Ceres, Pallas, Vesta and Hygiea, and the Sun's relativistic term. For each shadow, the engine draws the centre line, both edges and the 1σ lines.
The 1σ combines the star's error with the orbit's. The orbit error is the largest of three: JPL's formal error, the gap between JPL's and the Minor Planet Center's orbits ÷ √2, and 10 mas. Independent orbits differ by about 10 mas, far more than JPL's formal error.
A path is listed if it passes within 100 km of an Indian city at night, with the star 10° up. It must be at least as wide as its error, with a fade you can notice. 947 so far.
Each event has a finder chart (Gaia stars to G 14 within 15′, with the asteroid's hourly track), your chord, a fade sketch, a world map, and the path as KML or GPX. It opens on its own page, with the path on a street map. Drop a pin to see what any spot gets, then share it. Or let the page suggest five stations across the path; spread-out timings reveal the asteroid's shape. The full track around the world is drawn too, bright where the star is up after dark. Press Play to watch the shadow cross. As elsewhere, your browser works out how close the path comes to you, when, and for how long.
Jupiter's and Saturn's moons
Eclipses, occultations, transits and shadow transits of Jupiter's four big moons and seven of Saturn's come from JPL's satellite ephemerides. The planet is a flattened globe casting a cone of shadow. That gives 3,882 events for Jupiter and 5,302 for Saturn, 2026–2028. Each one happens at the same instant everywhere. The page only checks whether the planet is up and the sky dark where you are.
The Great Red Spot
A storm, not a body with an orbit. The site computes when the spot's longitude reaches the middle of Jupiter's disc (System II), using DE431 and the IAU pole. The spot drifts, so its longitude is carried forward only to 31 December 2026. After that, no transits are shown until it is updated.
Calendar feeds
Each of 516 cities gets a feed with the lunar occultations it can see, binocular star occultations for twelve months, and asteroid shadows passing over or near it.
Checked against
Comparisons with independent predictions, and the checks that tie the browser's calculations to the engine's. Those marked automated test run again whenever the data is rebuilt.
| What | Compared with | Agreement |
|---|---|---|
| The Moon hides Aldebaran, 9 January 2017: disappearance at Jaipur, New Delhi and Silchar | in-the-sky.org's published times | Within 5 s at each city; through the Moon-and-stars pipeline +0.5, −1.4 and +1.2 s. Automated test. |
| Which planets the Moon hides, 2026–2029 | Fred Espenak's Sky Event Almanac | Every occultation it lists is found. The scan's few extras are within 12° of the Sun, where no one can see them. |
| Jupiter's moons, 1–3 October 2026 | Project Pluto's 2026 tables | Every Io, Europa and Ganymede event to the minute. |
| Jupiter's moons eclipsing and hiding each other | IMCCE and BAA mutual-event predictions | Io occults Europa 23 September 2026 04:29 UT; Io eclipses Ganymede 18 January 2027 04:56 UT. |
| Saturn's moons, 12 September 2026 | Astronomy.com's listing | Dione's shadow on Saturn, Dione crossing the north pole and Tethys entering eclipse, at the listed times. |
| The Moon's libration and the tilt of its pole | JPL Horizons | 0.02° and 0.3°. Automated test. |
| Jupiter's central meridian, for the Great Red Spot | JPL Horizons | 0.01°. Automated test. |
| The star-occultation solver your browser runs | Direct Skyfield searches at random places | Within 0.5 s. Automated test. |
| The any-location solver on each event page | The engine's own contact search for every listed city | The compact description sits on the Moon's edge to within 0.05″ at every contact; the build refuses to publish otherwise. |
| Asteroid occultation paths, September–October 2026: 45 events | Occult4, as published by IOTA-India | Path widths within 0.5 km; star positions within about 2 mas; the shadow reaching and leaving the Earth within a minute of the published times. The centre lines are offset sideways as a whole: a median 12 km (0.4σ) with the Minor Planet Center's orbits that Occult uses, 0.7σ with JPL's. That is how much the two orbits differ. Automated test for five of them. |
| Where the asteroids are, 2026–2027 | JPL Horizons | From JPL's own starting state, the integration stays within 0.12 km of Horizons over the 448 days the pages cover. From the Small-Body Database's published elements, which the pages use, it differs by a few km. That is under 4 mas, well within each path's error. Automated test. |
| The search that decides which asteroid and star pairs are worth solving | A plain two-minute search of the same sky | Over two test days, the fast search kept every pair the full search found. Automated test. |
| The any-location solver on each asteroid page | The engine's full model | Within 0.02 km and 0.1 s. Automated test. |
| The formula that turns a DAMIT shape to the way it faces at an event | DAMIT's own light curves | Rebuilt from the shape alone: 54 light curves of Pallas to 1.6% and 34 of the dog-bone Kleopatra to 3.4%. A reversed spin fits 2–3 times worse and a quarter-turn phase error 3–11 times worse, so the check can tell. Automated test. |
| Where that shape then sits on the sky | Textbook spherical astronomy | A test shape with a spike points where the textbook formulas say, in all four quarters of the sky. A mirrored outline or a wrong obliquity sign fails it. Automated test. |
| The path drawn on each asteroid page's map and world inset | The engine's own centre line and edges | The browser's track matches the engine's to 0.1 km, for the centre line and both edges. Automated test. |
What is approximate
- The Moon's edge. City tables and the solid graze lines use the real lunar terrain. Times for “your location”, the shaded map and the world map treat the Moon as a smooth sphere, good to about 2 seconds.
- The horizon. Altitudes are geometric, for sea level, without atmospheric refraction or hills and buildings. Near the horizon the Moon looks about half a degree higher than shown. Your skyline decides the rest.
- Earth's rotation. ΔT for future dates is an estimate, good to well under a few seconds through 2029.
- Grazes. Near a graze limit a few hundred metres north or south changes what you see. The graze profile on each event page shows how sensitive your spot is.
- Seeing a star. Visibility is a rule of thumb. Haze, light pollution, your instrument and your eyes decide. The aperture presets on the asteroid pages are estimates too. Each is about a magnitude brighter than what the aperture can detect, since a star at the limit is too faint to see fade. A camera reaches 2–3 magnitudes deeper.
- Jupiter's and Saturn's moons. The brightness drop in a mutual event assumes evenly lit discs. Saturn's rings are not modelled, so “behind Saturn” means behind the globe. The moon diagrams are illustrations, good to a tenth of the planet's radius. Shadows and eclipses are drawn from the listed times, which come from JPL.
- The Great Red Spot. Its times are only as good as its assumed longitude: each degree off moves them by 1.65 minutes.
- An asteroid's shape. Every path, time and duration is worked out for a sphere of the asteroid's catalogue diameter. Real asteroids are lumpy, so the shadow may be narrower or wider. Timings from several places measure this. Where DAMIT has a reliable model (rated 2 or better of 5, with a settled pole), the event page draws the asteroid's real shape with your chord across it: 98 of 947 events. These shapes are convex and scaled to the catalogue size. Their rotation is projected from light curves, often over decades. They are shown, not used.
- An asteroid path's position. Errors of a few milliarcseconds shift a path sideways by kilometres to tens of kilometres. The red dashed 1σ lines show how much. Each place gets a percentage chance, assuming the real path may be off by a normal 1σ error, which is how occultation software reads a prediction. The chance is only as good as the 1σ. Occult and OccultWatcher use their own orbits, so their numbers may differ. Near an edge, expect either outcome. A star Gaia flags as hard to fit (high RUWE) is often double, and may fade in steps or not at all.
- The list. The lunar occultations are a hand-picked selection, not every one that happens. The asteroid list covers asteroids 15 km and up and stars to G 12.5, on paths near Indian cities at night at least as wide as their error.
About
Occult is a static site built by Alok Mandavgane: no accounts and nothing to sign in to, with basic page analytics only. Results are checked against IOTA's Occult and in-the-sky.org, never copied. A copied time couldn't be recalculated for your location anyway. Corrections are welcome through alokm.com.