Moon Phase Calculator

What this tool shows

For any instant you choose — a date and a time — it reports the Moon's phase (one of the eight canonical names), the illuminated fraction of its disc as a percentage, the Moon's age in days since the last new moon, its Brown lunation number, and the dates of the next new moon and next full moon. An upcoming phases list extends that to all four principal phases — the next new moon, first quarter, full moon and last quarter after your chosen instant, each with how many days away it is. If you enter a latitude and longitude, it also gives that day's moonrise and moonset on your local clock — the one thing the page used to say it could not do. A distance and apparent size panel reports how far away the Moon actually is at that instant, in kilometres and miles, how wide its disc appears in arcminutes, how far both sit from average, and when the next perigee and apogee fall — enough to judge a “supermoon” claim yourself. It draws the phase as a shaded disc, too. Every figure recomputes the moment you change the date or time, so you can step forward day by day and watch the crescent fill and empty.

Phase, elongation and illumination

The single number that drives everything is the elongation — the angle between the Sun and the Moon as seen from Earth. At 0° the Moon sits in the same direction as the Sun and its near side is dark: new moon. At 90° it is a quarter of the way around, half lit: first quarter. At 180° it is opposite the Sun, fully lit: full moon. At 270° it is half lit again on the other side: last quarter. The illuminated fraction follows from the phase angle by the formula k = (1 + cos i) / 2, so it does not rise evenly: it barely moves near new and full moon and changes fastest around the quarters. That is why a "waxing gibbous" looks nearly full for several nights while a thin crescent thickens quickly.

The synodic month

The phase cycle repeats every synodic month — on average 29.53 days from one new moon to the next. This is noticeably longer than the 27.32-day sidereal month, the time the Moon takes to return to the same position against the background stars. The gap exists because the Earth is itself moving: during the ~27 days the Moon takes to loop its orbit, the Earth has carried it about 27° further around the Sun, so the Moon must travel roughly two extra days' worth of arc to line back up with the Sun and complete a phase cycle. The Moon's age reported here is simply how far it is into the current synodic cycle, expressed in days.

Worked example: first quarter, step by step

Enter 16 February 2024, 15:00 (the field reads your local clock; the figures below are for that instant expressed in UTC). Here is exactly what the page's own JavaScript does with it:

Walking the arithmetic: the library turns the instant into a Julian Day (2460357.125), evaluates the Meeus series to get the elongation of 90.19° — a quarter of the way around the 360° cycle, so the phase name is First Quarter. The phase angle is i = 180° − 90.19° = 89.81°, and the illuminated fraction is (1 + cos 89.81°) / 2 = 0.502, which the readout rounds to 50.2% — a quarter moon is genuinely half-lit. The age is just how far into the cycle we are: (90.19 / 360) × 29.53 ≈ 7.4 days. The lunation is a count of new moons rather than a division: the page solves for the last true new moon before your instant — 9 February 2024 at 23:07 UTC — works out that it is the 298th new moon after the reference one of 6 January 2000, 18:14 UTC (which carries Brown lunation number 953), and adds, giving 298 + 953 = 1251. The mean 29.53-day month is still in the code, but only to index which new moon to solve for; the boundary itself is the solved instant, so the lunation number changes at exactly the moment the next new moon row announces. That distinction is worth the sentence: a true new moon can fall up to about 14 hours either side of its mean position — February 2024's came 5 h 57 m after it, August 2026's 10 h before it — and a count taken from the mean month alone is a whole number out for that entire window: it still names the lunation that has ended when the true new moon arrives early, and names one that has not begun yet when the true new moon arrives late. The next full moon arrives about 7.9 days later, roughly a quarter of a synodic month, because a first-quarter moon is a quarter-cycle short of full.

Phase and when the Moon is up

Because phase measures the Moon's angle from the Sun, it also sets a rough timetable for when the Moon is visible. A new moon rises and sets with the Sun and is lost in the glare. A first-quarter moon rises around midday and sets near midnight, so it dominates the evening sky. A full moon rises at sunset and sets at sunrise, up all night long. A last-quarter moon rises near midnight and is highest at dawn. This is why a full moon feels like an all-night event and a crescent is only ever a brief companion to dusk or dawn.

Distance, apparent size and the “supermoon”

The Moon's orbit is an ellipse with an eccentricity of about 0.055, so its distance from Earth is not a constant. Over the years 2024–2030 this page's own series swings between roughly 356,700 km and 406,700 km — a 50,000 km spread, about an eighth of the distance itself. Because apparent size is just geometry, the disc follows: its angular diameter is 2·asin(1737.4 / Δ), where 1737.4 km is the Moon's mean radius and Δ the centre-to-centre distance. That works out at about 33.5 arcminutes at an extreme perigee, 31.0′ at the mean, and 29.4′ at an extreme apogee.

The closest and farthest points are called perigee and apogee. Perigee comes round on average every 27.55 days — the anomalistic month, slightly longer than the 27.32-day sidereal month because the orbit's long axis slowly rotates — but individual intervals are ragged, ranging from roughly 24.6 to 28.6 days, because the Sun's pull stretches and squeezes the orbit (the effect Ptolemy called evection). That is why perigee distances themselves vary: a "close" perigee is around 356,500 km, a shallow one nearer 370,000 km.

Worked example: the 14 November 2016 supermoon

This was the closest full moon since 1948, so it is a good calibration point. Published tables put perigee at 11:23 UTC on 14 November 2016, 356,509 km, with the full moon 2½ hours later at 13:52 UTC. Enter that instant and the page reports:

Twenty-one kilometres from the published distance, and the turning-point instant within about ten minutes of it — good enough to answer the question people are actually asking, and not good enough to publish as an almanac. Note what the honest numbers do to the headline. That record supermoon was 8% wider than a Moon at the mean distance, not twice the size. And compared with the full moon of the month before — 16 October 2016, itself near perigee at 358,471 km and 33.32′ — it was wider by 0.5%: two “supermoons” in a row, indistinguishable to the eye. The widely-repeated “30% brighter” figure is arithmetically fine but rhetorically slippery: brightness scales with the disc's area, so the 14% diameter difference between an extreme perigee and an extreme apogee becomes 1.14² ≈ 1.30. It compares the two extremes of the whole orbit, not a supermoon with a normal one.

One more thing the numbers show: a supermoon needs the full moon and perigee to coincide, and they drift against each other because the synodic month (29.53 d) and the anomalistic month (27.55 d) are different lengths. The beat period between them is 412 days, a little under 14 months, which is why supermoons arrive in small clusters roughly a year and a bit apart rather than every month.

Your horizon: the height you stand at, and the skyline you can see

Stand on a beach and your horizon is at eye level. Climb h metres and it drops below you, because your line of sight to it is tangent to a curved Earth: the observer, the planet's centre and that tangent point form a right triangle, so the horizon falls by dip = arccos(R/(R+h)). The same dense air near the ground that adds Meeus' 0.5667° of horizontal refraction also bends your sight line downward, pushing the visible horizon further away and making the dip smaller than the vacuum figure; the standard treatment is an effective radius R/(1−k) with k = 0.17, which reproduces the 1.76′ per √metre dip table printed in every nautical almanac. Fill in the optional Your elevation field (metres or feet) and the tool lowers the Moon's standard altitude from h0 to h0 − dip and re-solves, so the Moon clears your horizon earlier and drops behind it later. It is the same shared routine the sunrise calculator uses, so a summit gives both pages one identical dip.

Some concrete sizes. A 10 m coastal watchtower dips the horizon 0.09° and moves moonrise by well under a minute. A 100 m clifftop dips it 0.29° and puts the visible horizon 39 km away, which at London's latitude is typically a minute or two. Ben Nevis at 1,345 m dips it 1.07°, and 2,000 m dips it 1.31° — several minutes. How many minutes that actually buys depends on the angle at which the Moon climbs, which is why the tool prints the shift for your exact date and place (“3 min 12 s earlier than at sea level”) instead of quoting a rule of thumb. Very near the poles, where the Moon crawls almost parallel to the horizon, the same fraction of a degree can be worth an hour.

Height is the small correction. The skyline is the big one

Almost nobody watches the Moon rise out of an open sea. A ridge, a row of houses, a treeline or the far wall of a valley stands some angle above your eye level, and the Moon has to climb that far extra before you see it. The numbers are not close: at London's latitude on a late-July date, a 5° skyline holds moonrise back by about 44 minutes — while 100 m of height, the clifftop above, buys only a minute or two. For years this page modelled the small effect and simply conceded the large one.

It no longer concedes it, but it does not pretend to know it either. The two optional Moonrise skyline ° and Moonset skyline ° fields ask you for that angle, because you are the only one who can see it. They are separate on purpose: the eastern and western halves of a real skyline rarely match, and the tool solves the day twice when they differ — moonrise from the moonrise angle, moonset from the moonset one. Enter 0, or leave them blank, and nothing moves by a millisecond.

Sizes for the same London date, so you can calibrate: 2° (a low ridge a few kilometres off) delays moonrise about 18 min and brings moonset forward about 20 min; 5° about 44 min and 50 min; 10° (a steep valley side) about 1 h 32 min and 1 h 48 min. Rise and set shifts differ because the Moon meets the two skylines at different angles. Very near the poles, where the Moon crawls almost parallel to the horizon, a single degree can be worth hours — or can remove the event from the day altogether, which the tool says in words rather than printing a time it does not believe.

Why it is not "just add the ridge angle"

Moonrise means the Moon's upper limb appears over what you can actually see. With a skyline a degrees up, that limb's apparent altitude at first sighting is a, and the tool rebuilds the whole standard altitude around it rather than shifting the old one:

h0(a) = a + 0.7275·π − R(a)

Two terms carry the work. π is the Moon's horizontal parallax for that instant's distance, asin(6378.14/Δ) — about 0.91° near apogee and 1.02° near perigee, taken from the same Δ the distance panel reports, never a hardcoded 0.95°. The 0.7275 factor is Meeus' "one parallax up to the topocentric centre, one semidiameter down to the upper limb", since the lunar radius is 0.2724 of Earth's.

R(a) is refraction at the raised altitude, and this is the term a naive implementation gets wrong. Refraction is 34′ only for a ray that grazes the horizon; it collapses to 9.8′ at 5°, 5.3′ at 10° and 2.7′ at 20°. Leaving the horizontal 34′ in place when your skyline is 5° up puts the threshold 0.40° too low and understates the delay by roughly 4 minutes — a confident-looking error, in the direction that makes the tool look better. The refraction curve is Bennett's 1982 formula, normalised so that R(0) is exactly the 0.5667° this page has always used, which is what makes typing 0 bit-for-bit identical to leaving the field blank. It is the same shared module the sunrise calculator uses for its own skyline fields, so the two pages cannot drift apart.

How to get the angle without instruments

An outstretched fist at arm's length spans roughly 10°, a thumb about 2° and a little finger about 1°, and the Moon itself is only 0.5° — a quarter of a thumb-width, which is why it looks so much smaller against a skyline than people expect. Stack fists from the horizontal up to the ridgeline and you have a usable figure; the difference between 4° and 5° is about nine minutes at mid latitudes, so this is worth doing carefully but not worth agonising over. A phone clinometer or a topographic profile does better. Negative values are accepted too (down to −20°), for the case where the land falls away and your visible horizon sits below eye level; entries beyond −20° or +60° are clamped rather than rejected.

Three things the horizon settings deliberately do not change, because changing them would invent an effect that is not there:

• Phase, illumination, age and lunation. These are geocentric — angles between the Sun, the Earth's centre and the Moon. A full moon is full for everyone at the same instant, whether you are in a valley or on a summit. This is the same point the sunrise calculator makes about its twilight and golden-hour bands: they are defined by the body's own altitude, so where you stand cannot shift them.
• The altitude and azimuth readout. Those are instantaneous positions measured from the astronomical horizontal, not sightings of the visible horizon. Dip lowers the line you look along; it does not move the Moon. So the “Where to look” figures are identical at sea level and at 3,000 m, and no dip is applied to them.
• Anything you did not measure. A skyline angle is your number, not terrain data: this page holds no elevation model, no map and no building footprints, and it cannot fetch any — it is a static file with no network access. It applies the angle you type, in the direction you type it, and assumes that one angle describes the whole arc where the Moon actually crosses. A skyline that steps up sharply within a few degrees of azimuth, a lone tree, or a building you have not accounted for will still surprise you.

Below sea level, blank and unparseable entries in the height field all mean no dip correction. That is deliberate rather than lazy: in a place like the Dead Sea shore your horizon is not the curve of the Earth at all — it is the escarpment around you — so a negative dip computed from your depth would be a confident-looking guess about the wrong thing. The honest answer there is the skyline field, which asks you for the angle instead of inventing it. The same rule governs the skyline fields themselves: blank or unparseable means no correction and takes the untouched sea-level code path, never a silent zero.

Honest limits

This is a simplified, geocentric model, and it is worth being clear about what that means. The phase angle and elongation are accurate to within roughly half a degree for dates within a couple of centuries of the year 2000; the tool uses only the leading periodic terms of Meeus' series and omits the smaller ones and planetary perturbations. The illuminated fraction is good to about 1%, and a little worse in the thin crescent where a small angle error matters most. The Moon's age is derived from the mean synodic month, so it can drift up to about half a day from a true almanac age, because the Moon speeds up near perigee and slows near apogee rather than moving evenly. The upcoming phase dates — next new moon, first quarter, full moon and last quarter — are found by searching that same elongation curve; because half a degree of elongation error corresponds to roughly an hour of time, they are typically right to within an hour or two, and the upcoming-phases list rounds them to the nearest hour rather than pretending to minute precision. Because the phase is geocentric it is the same for everyone on Earth at a given instant; only the orientation differs — from the Southern Hemisphere the lit side is mirrored, which is why entering a southern latitude flips the drawn disc but leaves every number unchanged. Finally, a word on the moonrise and moonset times, which come from a second, separate calculation. Those need the Moon's actual position, not just its angle from the Sun, so the page loads an abridged version of Meeus' chapter-47 lunar series — the 30 largest terms of the longitude/distance table and the 20 largest of the latitude table, out of sixty rows each — converts the result to right ascension and declination, and then solves for the instant the Moon crosses its standard altitude h0 = 0.7275·π − 0.5667°, where π is the Moon's horizontal parallax (about 0.95°, which is why the Sun's −0.833° constant cannot be reused here). Because the Moon moves roughly 12° a day, the solver re-evaluates its position at every successive estimate until the answer stops moving by more than ten seconds. Expect agreement with a national almanac within a few minutes at mid latitudes. It gets worse where the geometry is shallow — within a few degrees of the poles, and on any day when the Moon skims the horizon for hours. Your own horizon is no longer assumed away: the optional height field lowers the standard altitude to h0 − dip, and the two optional skyline-angle fields rebuild it entirely as h0(a) = a + 0.7275·π − R(a) with refraction re-evaluated at the raised altitude, separately for the moonrise and moonset sides; the rise/set rows label how many minutes each was worth. What the page still does not have is a terrain model — no elevation data, no map, no building footprints. It applies the angle you measured and typed in, and that one angle stands in for the whole arc where the Moon crosses, so a skyline that steps up sharply over a few degrees of azimuth is still beyond it. The distance readouts come from the Σr half of that same abridged series — the term the rise/set solver was already evaluating for the parallax. Spot-checked against published perigee and apogee tables for 2016 and 2020 the distances land within about 25 km (0.007%) and the turning-point instants within about ten minutes, but the distance curve is flat at a turning point, so a small error in Δ moves the moment a great deal: the perigee and apogee times are rounded to the hour and are worth an hour or two, never a minute. The apparent diameter is pure geometry from that distance and a fixed 1737.4 km lunar radius, so it inherits the same tiny relative error — but it is a geocentric figure, measured from the centre of the Earth: an observer with the Moon directly overhead is one Earth radius closer and sees a disc about 1.7% wider than the number shown. This is a phase-and-rise/set calculator, not an eclipse predictor: an eclipse needs the full moon to fall within a fraction of a degree of Earth's shadow, which requires the Moon's orbital-node position — not modelled here.

How good are the altitude and azimuth?

The Where to look readout answers the question a moon page is actually asked — is it up, and which way do I point? — and the conventions behind those two numbers are worth stating plainly, because an unstated convention makes an alt/az figure useless.

Azimuth is a true-north compass bearing, measured clockwise through east: 0° north, 90° east, 180° south, 270° west, with the 16-point name derived straight from the angle. Note that a phone compass usually shows magnetic north, which in Britain currently differs from true north by a degree or two and by far more at high latitudes.

Altitude is topocentric. The diurnal-parallax correction is applied using the same 6378.14 km Earth radius and the same Σr distance the rise/set solver already needs — no new constants — and it matters: because the Moon is only about sixty Earth radii away, standing on the surface rather than at the planet's centre lowers it by up to a full degree near the horizon, tapering to exactly nothing at the zenith. Azimuth carries no parallax term, and that is a decision rather than an oversight: on the spherical Earth this model uses, the offset from the planet's centre to you points exactly at your zenith, so removing it slides the Moon straight down the vertical circle and cannot rotate it sideways. The real Earth is an oblate spheroid, where geodetic and geocentric latitude differ by up to 11.5′, and that does move azimuth — by under 0.01° here, which is why it is left out. Refraction is yours to choose. The Altitude convention picker switches the readout between the geometric position of the Moon's centre and its apparent one, and the note under the figure always says which is on show. Refraction defaults to OFF, and not by accident: it is the least predictable term in the whole calculation. The library uses Meeus 16.4, which assumes standard air and takes no temperature or pressure input, giving 0.48° at the horizon, 0.09° at 10° and 0.017° at 45° — but a temperature inversion over cold ground can roughly double the horizon figure, which is why a geometric number is the one worth quoting and an apparent one the one worth looking with. Below −1° the formula is out of range, so it is clamped to zero and apparent mode says the correction does not apply rather than printing a fake 0.00°.

On accuracy, a measured figure rather than an impressive-sounding one. Checked against the US Naval Observatory's own celestial-navigation almanac (its published geocentric altitude hc minus its published parallax-in-altitude pa, and its azimuth zn) at 33 instants spanning 2015–2035 and latitudes from 64°N to 35°S, the worst disagreement was 0.023° in altitude and 0.019° in azimuth — under two arcminutes, less than a twentieth of the Moon's own 0.52° disc. Ten of those comparisons are pinned as golden vectors in this site's test suite so the figure cannot quietly drift. Two arcminutes is the honest ceiling of the abridged 30-row longitude and 20-row latitude series described above, and it is not a claim of arc-second precision.

Finally, an angle is not a sightline. The altitude and azimuth are measured from the astronomical horizontal — the plane through you at right angles to your local vertical — not from the horizon you can see. That is why the optional elevation field, which moves moonrise and moonset, deliberately leaves this readout alone: climbing a mountain drops the visible horizon below that plane, but it does not move the Moon, so an instantaneous altitude of 12.4° is 12.4° at the beach and 12.4° on the summit above it. It changes when the Moon becomes visible, not where it is. Terrain still hides things this readout cannot: a ridge, a building or a line of trees can conceal a Moon that is geometrically several degrees up. That is precisely what the skyline angle fields are for — but they feed the moonrise and moonset solver only, and they take the single number you measured rather than a profile of your whole horizon. And the phase, illumination, age and lunation numbers are geocentric — entering coordinates changes where the Moon sits in your sky and which way the lit limb is tilted, but not how full it is. That is the same point the sunrise & sunset calculator makes about its twilight bands: they are defined by the Sun's own altitude, so where you stand cannot shift them.

Pair it with the rest of the sky

The Moon is only part of the night. Once you know tonight's phase, the Sunrise & Sunset Calculator tells you when the sky will actually go dark where you are — a bright full moon washes out faint stars, while a new moon leaves the sky at its darkest. To see what is up there beyond the Moon, the free star map generator plots the constellations, planets and bright stars visible from your latitude at any date and time. Between the three you get the whole picture: how lit the Moon is, when it is dark, and what you will actually see.

Frequently asked questions

How is the Moon's phase calculated?

The tool evaluates the Sun–Moon elongation — the angle between the Sun and Moon as seen from Earth — using the low-accuracy periodic series from Jean Meeus' Astronomical Algorithms (chapters 47–49). Elongation runs from 0° at new moon, through 90° at first quarter, 180° at full, to 270° at last quarter. The illuminated fraction is then k = (1 + cos i) / 2, where i is the phase angle. All of it runs in your browser with no network call.

What is the synodic month?

The synodic month is the average time from one new moon to the next — about 29.53 days. It is longer than the 27.32-day sidereal month (the Moon's orbit relative to the stars) because while the Moon circles Earth, Earth also moves along its own orbit around the Sun, so the Moon must travel a little further to line up with the Sun again. The phase cycle you see — new, waxing, full, waning — follows the synodic month.

What does the illuminated percentage mean?

It is the fraction of the Moon's disc, as seen from Earth, that is lit by the Sun: 0% at new moon, about 50% at each quarter, and 100% at full. The percentage does not climb at a steady rate — it changes slowly near new and full moon and fastest around the quarters, because it follows the cosine of the phase angle rather than a straight line.

Why does the Moon look full for several nights in a row?

Because illumination follows that cosine, it barely moves near full moon. In the roughly three-night window centred on a full moon the lit fraction stays above about 97%, so to the eye the disc looks "full" for two or three nights running. The tool shows the difference as a decimal — you might read 99.4% one night and 100.0% the next — even though your eye cannot tell them apart.

Does the Moon's phase depend on where I am on Earth?

The phase and illuminated fraction are geocentric, so they are effectively identical everywhere on Earth at a given instant — a full moon is full for everyone. What changes with location is the Moon's orientation: from the Southern Hemisphere the lit side appears mirrored compared with the Northern Hemisphere, and the times the Moon rises and sets differ. This tool flips the drawn disc for a southern latitude, but the numbers stay the same.

How does the phase relate to when the Moon rises and sets?

Phase tracks the Moon's angle from the Sun, which sets a rough rise/set rhythm. A new moon rises and sets with the Sun, so it is lost in daylight. A first-quarter moon rises around midday and sets around midnight. A full moon rises near sunset and sets near sunrise, visible all night. A last-quarter moon rises around midnight and sets around midday. Those are rules of thumb, though: the exact times depend on your latitude, longitude and the Moon's declination, so enter your coordinates in the Location box — or press the Use my location button beside it, which asks your browser for a fix, writes it into that same box and never sends it anywhere — and the Moonrise / Moonset rows will give you the real figures for the date you picked.

Can this tool tell me about a supermoon or a lunar eclipse?

Supermoon, yes; eclipse, no. A “supermoon” is press shorthand — not an official astronomical term — for a full moon that falls near perigee, the closest point of the Moon's slightly elliptical orbit. The Distance & apparent size panel gives you both halves of that: the Earth–Moon centre-to-centre distance in kilometres and miles for the instant you picked, the resulting apparent diameter of the disc in arcminutes, how far each sits from the mean, and the next perigee and next apogee with their own distances — so you can see for yourself how close a coming full moon lands to perigee, rather than take a headline's word for it. Two honest caveats. The perigee and apogee instants are rounded to the hour, because the distance curve is flat at a turning point and this abridged series cannot justify minutes. And the effect is small: about 14% in diameter between an extreme perigee and an extreme apogee, but only a few percent between a typical “supermoon” and an average full moon — real, and very hard to see without a side-by-side photograph. A lunar eclipse is still out of scope: it needs the full moon to line up with Earth's shadow to within a fraction of a degree, which requires the Moon's orbital-node position — not modelled here.

How far away is the Moon right now, and how big does it look?

Pick a date and time and the Distance & apparent size panel answers both at once. The Earth–Moon centre-to-centre distance appears in kilometres and miles beside its percentage difference from this model's mean of 385,001 km, and the apparent diameter of the disc appears in arcminutes beside its own difference from the mean 31.03′. Over 2024–2030 the distance runs between about 356,700 km and 406,700 km, and the disc between about 33.5′ and 29.4′. Both are geocentric, measured from the centre of the Earth — an observer with the Moon directly overhead is one Earth radius closer and sees a disc roughly 1.7% wider than the figure shown.

How precise are these figures?

This is a simplified geocentric model. The phase angle and elongation are within roughly half a degree of the full solution for dates within a couple of centuries of the year 2000, and the illuminated fraction within about 1% (a little more near the thin crescent). The Moon's age comes from the mean synodic month, so it can drift up to about half a day from a true almanac age because the Moon's speed varies around its orbit. The upcoming phase dates — next new moon, first quarter, full moon and last quarter — are typically within an hour or two of published almanac times, which is why they are rounded to the nearest hour.

When is the next full moon?

This page computes it live rather than quoting a stored table: the upcoming-phases readout scans the Sun–Moon elongation forward from your chosen instant and bisects to the moment it crosses 180°, then shows the date and approximate time on your device's local clock. Because full moons repeat every 29.53 days on average, the answer always falls within about a month of whatever date you pick — set the date field to any instant, past or future, and the list updates immediately. The same search also reports the next new moon, first quarter and last quarter.

Why are the upcoming phase times rounded to the nearest hour?

Honesty about precision. The low-accuracy Meeus series used here is good to roughly half a degree of elongation, and the Moon's elongation grows about 12.2° per day — close to half a degree per hour — so a half-degree model error translates into roughly an hour of timing error. Quoting these events to the minute would be false precision. Almanacs that publish new- and full-moon times to the minute evaluate hundreds of periodic terms plus planetary perturbations; this page trades that for a small calculation that runs entirely in your browser.

Why is there no moonrise on some days?

Because the lunar day is about 24 h 50 min long, not 24. The Moon rises roughly 50 minutes later each day, so its rise time creeps forward until it steps over midnight — and that calendar date then has a moonset but no moonrise. It happens at every latitude, roughly once per synodic month, and the same thing happens to moonset about two weeks later. It is not a rounding failure: when it occurs the tool says “No moonrise today” and tells you what time the Moon does rise the following morning. Separately, above about 61° of latitude the Moon can stay above (or below) the horizon for a whole day or several days running — a body whose declination exceeds 90° minus your latitude never crosses the horizon at all, and the Moon's declination reaches almost 29°.

Why does my almanac give a moonrise a few minutes different?

Three ordinary reasons, none of them a mistake. First, the model is abridged: this page evaluates the 30 largest longitude/distance terms and the 20 largest latitude terms of Meeus' chapter-47 lunar series, out of sixty rows each, while a national almanac evaluates the full theory plus planetary perturbations. Second, the horizon, which you can now describe: an optional height lowers the visible horizon by the dip angle for that height, so 100 m up dips it about 0.29° and brings moonrise forward by a minute or two at mid latitudes, and two optional skyline-angle fields raise it instead, one for the moonrise side and one for the moonset side. A 5° skyline delays moonrise by roughly 44 minutes at London's latitude, which dwarfs any height correction. Left blank, all three fields mean a flat sea-level horizon. What is still absent is terrain data: the page has no elevation model and no map, so it applies the angle you measured and nothing more. Third, refraction is a fixed number: 0.5667° at the horizon, whereas real horizontal refraction swings with air temperature and pressure by enough to move the moment a minute or so on a cold, clear morning. The solver itself converges to under ten seconds, so any disagreement you see is the model's, not the arithmetic's.

Can I account for my elevation, and for hills or buildings on my horizon?

Both, through the three optional fields in Your horizon. Height above sea level lowers your visible horizon by the dip angle — arccos(R/(R+h)), with R enlarged to account for the way low-level air bends the sight line — so the Moon clears it earlier and sets later: a 100 m clifftop dips the horizon 0.29°, 2,000 m dips it 1.31°, and at mid latitudes that is worth a minute or several. Hills and buildings are the far larger effect and they need a number only you have, so the Moonrise skyline and Moonset skyline fields ask you for the angle your skyline stands above eye level. They are separate because an eastern and a western skyline rarely match, and when they differ the day is solved twice. At London's latitude a 2° ridge delays moonrise about 18 minutes, 5° about 44 minutes and 10° about an hour and a half. The threshold is rebuilt as h0(a) = a + 0.7275 times the horizontal parallax, minus refraction evaluated at that raised altitude — refraction is 34′ only at the horizon and falls to about 9.8′ at 5°, so keeping the horizon value would understate a 5° delay by roughly 4 minutes. Every field is optional: blank, zero and unparseable all mean no correction and give the flat sea-level times bit for bit. What none of this is, is terrain data — the page has no elevation model and no map, it applies the angle you measured, and it still leaves the phase, illumination, age and lunation untouched (all geocentric) along with the altitude and azimuth readout, which is measured from the astronomical horizontal rather than from anything you can see.

Is the Moon up right now, and which way do I point?

Enter the instant and your latitude and longitude, and the Where to look readout answers both halves. Altitude is how far above your horizon the Moon's centre sits: 0° on the horizon, 90° straight overhead, and a negative figure is reported in words as below the horizon rather than as a bare minus sign. Azimuth is the bearing to point along, measured from true north clockwise through east — 0° N, 90° E, 180° S, 270° W — with the 16-point name (NNE, SW and so on) derived straight from that angle. Two conventions are worth knowing. The altitude is topocentric: corrected for your standing on the Earth's surface rather than at its centre, which lowers the Moon by up to a full degree near the horizon and by nothing at the zenith. You also pick the convention: geometric is the default and apparent adds atmospheric refraction, which lifts the Moon 0.48° at the horizon, 0.09° at 10° up and under 0.02° above 45°, with the note under the readout always naming the one that produced the figure. Refraction is the least predictable term on this page: the correction assumes standard air, the tool takes no temperature or pressure reading, and an inversion can roughly double the horizon value, which is why geometric stays the default. Checked against the US Naval Observatory's celestial-navigation almanac at 33 instants from 2015 to 2035 and latitudes 64°N to 35°S, the worst disagreement was 0.023° in altitude and 0.019° in azimuth. One honest caveat the maths cannot fix here: this readout is a position in the sky, not a sighting, so a ridge, a building or a line of trees can hide a Moon that is geometrically several degrees up. The skyline fields in Your horizon account for that in the moonrise and moonset times; they deliberately leave this altitude alone, because a hill does not move the Moon.