How we calculate the sky

Positions come from astronomy-engine, which derives from the VSOP87 planetary theory and the NOVAS reference implementation used by the US Naval Observatory. Planetary positions are accurate to roughly one arcminute over the years this calendar covers — far finer than anything that affects whether you can see something.

We validate it rather than trust it. The 7 September 2025 total lunar eclipse resolves to its published greatest-eclipse time, and the Sun’s altitude at the June solstice matches the geometric identity 90° − |latitude − 23.44°| to within a hundredth of a degree. These checks run as automated tests.

Calculations use your latitude, longitude and elevation — not a national average. India spans about 27° of latitude and 29° of longitude, which is enough that the Sun sets nearly an hour and a half earlier in Port Blair than in Jaisalmer despite the shared clock, and enough that an object comfortably placed in Chennai can be below the horizon in Leh.

Rise and set times include atmospheric refraction at the standard value, which is why the Sun appears to rise slightly before it geometrically does.

This is the distinction the calendar exists to make. A lunar eclipse with the Moon below your horizon is, for you, nothing at all — so it is labelled “the eclipse happens, but not in this sky” rather than being listed as an event you can watch.

Solar eclipses use local circumstances computed for your exact coordinates, never a blanket “visible in India”. If only a partial phase reaches you, the page says so and gives the fraction of the Sun covered.

• Minimum altitude 10°. Below about 10° you are looking through more than five atmospheres of haze and whatever sits on your horizon. You can change it to 5°, 15°, 20° or 30°.
• Conjunction ≤ 5°, and “close” at ≤ 1°. There is no official definition; these are ours, applied uniformly.
• Astronomical darkness is the Sun below −18°. Civil is −6° and nautical −12°; all three are shown.
• Event importance follows a rule, not taste: total and annular eclipses, showers with ZHR ≥ 100, conjunctions under 1°, and oppositions of Mars, Jupiter and Saturn are “major”.

Zenithal Hourly Rate is a normalised figure: what one observer would count under a perfect sky with the radiant overhead. Nobody ever observes under those conditions.

We correct it for the two effects we can actually compute — the radiant’s altitude from your location, and moonlight while the Moon is up — and we say so. Cloud, haze, local light pollution and your own experience all subtract further, so the number shown is an optimistic ceiling rather than a promise.

A 0–100 figure built from moon interference (45), length of astronomical darkness (25), planets above your minimum altitude (15), and notable events (15). Every component is shown with its reasoning on each day’s detail panel.

It contains no weather. Cloud is the thing most likely to ruin a night, and we have no forecast for it — folding an unmeasured factor into a measured score would make the whole number dishonest. The score describes the sky, not the weather and not the site.

Light pollution. We hold no measured sky-brightness data for these locations, so no Bortle class is shown for any of them. A plausible-looking number per city would be invention, and it would be the number people trusted most.

Weather. Not integrated. When it is, it will be reported separately from the astronomical score rather than blended into it.

Occultations. Lunar occultations of stars are not yet calculated. They need arcsecond-accurate predictions and lunar limb profiles to be worth publishing, and a rough answer would be worse than none.

All times are Indian Standard Time (UTC+5:30). India observes no daylight saving, but the engine works in UTC internally and converts on display, so it stays correct regardless.

A night is labelled by the date it begins. “Friday night” includes the small hours of Saturday, which is when most of the observing actually happens.