How Atmospheric Refraction, Weather, and Elevation Affect Moon Visibility

The Short Answer

Refraction mostly helps: near the horizon it lifts the Sun and the Moon by about 0.57 degrees (34.5 arcminutes), which slightly lengthens the observation window. Extinction, the dimming of light by the air, is what erases a low crescent. At 5 degrees altitude its light crosses about 10 air masses and loses about 1.5 magnitudes in clean desert air, about 3.1 on an average clear night and about 5.7 in haze. For a marginal crescent, a transparent sky matters more than a cloudless one.

Every crescent visibility model, Yallop, Odeh, or any other, computes the Moon's position under a significant assumption: a perfectly clear, standard atmosphere. In the real world this is almost never true. The atmosphere bends, distorts, absorbs, and sometimes completely blocks the crescent's light, creating a gap between what the mathematics predict and what the observer actually sees. This gap is the single largest source of uncertainty in visibility predictions.

This article covers the two atmospheric effects that matter most for crescent observation: atmospheric refraction, which bends the Moon's apparent position near the horizon, and atmospheric extinction, which dims its light. It also explains how Hilal Vision applies live weather data to sharpen the prediction. The underlying visibility framework these corrections feed into, the Yallop q-value and Odeh V-value, is set out in the science behind the crescent. For the geometry of where you stand, horizon dip, terrain blockage and dark adaptation, see why altitude and terrain matter. Here we stay close to the optics.

First, A Correction of Priorities: Refraction Helps, Extinction Hurts

It is tempting to treat atmospheric refraction as the villain of crescent observation. In practice refraction is mostly your friend: near the horizon it lifts the apparent altitude of both the Sun and the Moon by roughly 0.57 degrees (about 34.5 arcminutes), buying the observer a little extra altitude and a slightly longer window. Standard models (Bennett 1982; Saemundsson's refinement) handle this largely predictable nudge well under normal conditions.

The genuine destroyer of low crescents is atmospheric extinction: the dimming of light as it passes through the air. Refraction moves the crescent; extinction can erase it. For an object as faint as a young hilal, sitting low over the western horizon in bright twilight, extinction is the single most important atmospheric limiter on visibility. Getting this ordering right matters, because an observer who fusses over a fraction of a degree of refraction while ignoring two or three magnitudes of haze-driven extinction is optimising the wrong variable.

Atmospheric Refraction: The Invisible Lens

How Refraction Works

Earth's atmosphere is a stack of layers with progressively decreasing density from the surface upward. When light enters this medium at an angle, it bends toward the denser air, always making objects appear higher than they geometrically are. This atmospheric refraction is strongest near the horizon, where light traverses the maximum thickness of atmosphere. At the horizon itself, refraction lifts the apparent position of the Sun and Moon by approximately 0.57 degrees (about 34.5 arcminutes), slightly more than the full Moon's angular diameter. The practical consequence: when you watch the Sun "set", it has already set geometrically; you are seeing an atmospheric mirage. The same applies to the crescent, and it matters because sunset and moonset define the observation window.

The Standard Refraction Model

A widely used refraction correction in crescent visibility work is the Bennett (1982) formula, popularised through Jean Meeus's Astronomical Algorithms:

R = 1 / tan(h + 7.31 / (h + 4.4))   [arcminutes]

Where h is the apparent altitude in degrees. The closely related Saemundsson formula inverts this to map true altitude to apparent altitude. Both assume standard atmospheric conditions:

• Temperature: 10°C
• Pressure: 1010 hPa (approximately sea level)

For a Moon at the geometric horizon (h = 0°), this gives a refraction of about 34.5 arcminutes, a little more than the Moon's angular diameter.

Why Standard Conditions Are Rarely Standard

The critical insight is that refraction depends on air density, which depends on temperature and pressure. The first-order correction for non-standard conditions is:

R = R₀ × (P / 1010) × (283 / (273 + T))

Where P is surface pressure in hPa and T is temperature in °C. Colder, denser, higher-pressure air refracts more; warmer, thinner, lower-pressure air refracts less.

Consider the extremes:

The difference between arctic winter and desert summer is roughly 10.8 arcminutes, more than one-third of the Moon's angular diameter. For a crescent hovering just above the horizon, this difference can shift the apparent altitude enough to change a marginal call. The temperature and pressure terms are precisely what the app recalculates from live data, as described in the How Hilal Vision Applies Atmospheric Corrections section of this article.

Differential Refraction: Why a Low Crescent Looks Squashed

So far we have treated refraction as a single lift applied to the whole Moon. It is not. Refraction is a function of altitude, and it changes rapidly in the last degree or two above the horizon. The light from the bottom edge of the crescent enters the atmosphere at a slightly lower angle than the light from the top edge, so the lower limb is lifted more than the upper limb. The disc is compressed in the vertical direction while its horizontal width is left almost untouched. This is differential refraction.

The effect is the same one that makes the setting Sun look like a flattened oval rather than a circle. Near the horizon the vertical compression can reach 15 to 20 per cent: a body that is 30 arcminutes tall geometrically can appear only 24 to 25 arcminutes tall. The compression is strongest in exactly the altitude band, roughly 0 to 3 degrees, where a young crescent lives.

This distortion is not just cosmetic, and here is why it matters for the models. The Yallop criterion is built around W, the topocentric crescent width in arcminutes, measured perpendicular to the line joining the cusps. When the disc is vertically compressed, the geometry of the illuminated arc is distorted: the measured or visually estimated width no longer matches the clean geometric value the model assumes. An observer estimating W from a flattened, shimmering crescent at 1.5 degrees altitude can easily be off by a few tenths of an arcminute, and through the q formula a small error in W feeds straight into the predicted zone. A well-implemented engine therefore computes W from the underlying elongation and limb-darkening geometry at the model's "best time", rather than from a naive apparent-disc measurement that differential refraction would corrupt.

Temperature Inversions and Extreme Refraction

Under certain conditions, when a temperature inversion creates a sharp density boundary near the surface, refraction can produce anomalies that no standard model captures. An observer inside a strong inversion may see the crescent appearing several arcminutes higher than predicted, tipping a marginal call toward visible; the converse also occurs, with mirages creating the illusion of a crescent where the geometry says none should be. These events are rare but worth knowing about, because they are a genuine source of false positives and false negatives in the observational record.

Atmospheric Extinction: The Dimming of the Crescent

The Extinction Problem

Even in a "clear" sky, the atmosphere absorbs and scatters light. This atmospheric extinction is strongest at low altitudes and toward the blue end of the spectrum, and for the crescent moon, already extremely faint and sitting near the horizon, it can reduce apparent brightness by a large factor.

Extinction follows the Beer-Lambert law: dimming grows exponentially with path length, measured in air masses. The Kasten and Young (1989) air-mass formula, accurate right down to the horizon where the older sec(z) approximation fails, places a crescent at 5 degrees altitude through roughly 10 air masses and one at 2 degrees through roughly 19. That exponential climb is why the final minutes of the observation window are so punishing.

Air Mass and Why the Horizon Is Brutal

The Kasten and Young (1989) formula produces the following air-mass values:

A crescent at 5 degrees altitude is seen through roughly ten times more atmosphere than the same crescent overhead, and one right on the horizon through nearly forty times.

From Air Mass to Magnitudes Lost

Extinction follows the Bouguer / Beer-Lambert relation: the magnitude loss is the product of air mass and a zenith extinction coefficient k (magnitudes per air mass). The coefficient k is where the weather enters. At a clean, dry, high-altitude site, k in the visual band can be as low as 0.12 to 0.20 magnitudes per air mass. At a hazy, humid, dusty sea-level site it can climb to 0.4, 0.6, or worse:

A loss of three magnitudes is a brightness reduction of roughly a factor of 16. Nearly six magnitudes is a factor of around 200. The crescent's intrinsic surface brightness has not changed at all; the atmosphere alone has dimmed it into invisibility.

Aerosols, Dust, and Humidity

The coefficient k is not a constant of nature; it is a property of the air on a given night over a given place. Three things dominate it for crescent work.

Aerosols and dust. Suspended particles scatter and absorb light, concentrated in the lowest one to two kilometres of the atmosphere, precisely the layer a horizon-skimming sightline ploughs through. Saharan dust outbreaks, which routinely sweep over the Mediterranean and Arabian Peninsula, can raise k enough to push an otherwise easy crescent below the threshold of naked-eye perception. The same is true of biomass-burning smoke and industrial pollution downwind of cities.

Humidity. Water vapour both scatters light directly and, more importantly, swells hygroscopic aerosol particles so they scatter far more efficiently. As relative humidity climbs above roughly 70 to 80 per cent, aerosol extinction can rise sharply even before any visible cloud forms. This is why a clear but muggy tropical evening can be far worse for the hilal than a crisp, dry night of the same nominal cloud cover.

Seasonal and regional patterns. These effects cluster geographically, which is part of why a crescent is visible in some countries and not others on the same evening. Post-monsoon haze across South and Southeast Asia frequently coincides with calendar-critical months; tropical coasts sit under near-permanent high humidity; the Saharan fringe carries suspended dust for much of the year.

What Drives Extinction on a Given Night

Three variables dominate how severe extinction is:

• Humidity: Humid air carries more water vapour, which both scatters and absorbs, and which can condense into near-surface haze.
• Aerosols: Dust, pollen, sea salt, smoke, and industrial pollution all add scattering. Aerosol loading is the single most variable component of clear-sky extinction.
• Path length: Through the air-mass dependence above, every degree of altitude the crescent gains sharply reduces the column it must traverse.

Geographic Extinction Patterns

Certain regions are systematically disadvantaged: tropical coastal zones carry high humidity year round; the Saharan fringe is loaded with suspended dust; South and Southeast Asian monsoon zones add haze during sighting-critical months; and major urban areas layer light pollution onto aerosol scatter. Conversely, high-altitude desert sites such as the Arabian Peninsula's interior plateaus or the Iranian Plateau offer consistently excellent transparency. The reasons elevation helps so much, less air below you and a shorter total column, are treated in detail in why altitude and terrain matter.

Lag Time and the Observation Window

Refraction and extinction both bite hardest in the final minutes of the window, so it is worth defining that window precisely.

Lag time is the interval between sunset and moonset (lag = moonset - sunset). It is the entire span during which the crescent is above the horizon while the Sun is below it. A young crescent on a difficult evening may have a lag of only 20 to 30 minutes; an easy crescent may have well over an hour. Lag time tracks ARCV closely, because a larger arc of vision means the Moon is higher at sunset and therefore takes longer to reach the horizon, though lag also depends on the angle the Moon's path makes with the horizon, which varies with latitude and season.

The usable window is narrower than the raw lag: the twilight sky immediately after sunset is far too bright to reveal a thin crescent, so the practical window runs from "sky dark enough" to "Moon too low", and for a marginal crescent that intersection can last only 15 to 30 minutes.

Odeh's V-value criterion, derived from 737 observation records (Odeh 2004), gives the outcome thresholds:

• V ≥ 5.65: crescent visible to the naked eye
• 2 ≤ V < 5.65: visible with optical aid, and may then be seen with the naked eye
• -0.96 ≤ V < 2: visible only with optical aid
• V < -0.96: not visible even with optical aid

Odeh also identifies an empirical optical-aid limit of about 6.4 degrees of elongation. A short lag does not change these thresholds, but it shrinks the slice of time in which conditions sit on the favourable side, which is why a model can report a positive V while a real observer, fighting extinction and a closing window, still comes home empty-handed.

A Worked Example: Putting the Numbers Together

Abstractions only go so far, so consider a concrete evening. Take an observer near Tucson, Arizona (about 32.2°N, 110.9°W), on the evening of 17 April 2026, attempting the first crescent of the lunar month from a clear desert site. Conjunction was at 11:52 UTC that day, so the Moon is about 14 hours old at the Tucson sunset.

The astronomical geometry at the model's best time gives, at this location:

• Geocentric Moon-Sun elongation (ARCL): about 9.2 degrees
• Topocentric crescent width W: about 0.21 arcminutes
• Arc of vision (ARCV): about 7.3 degrees
• Apparent Moon altitude at best time: about 4.2 degrees
• Lag time: about 41 minutes

Now layer the atmosphere on top. The desert evening is warm and the site sits near 800 metres, so surface pressure is around 925 hPa and temperature around 24°C. Plugging those into the refraction correction, the density factor is (925/1010) × (283/297) ≈ 0.873. Refraction is therefore about 13 per cent weaker than the standard model assumes; a standard-atmosphere engine would over-lift the Moon by roughly 1 to 2 arcminutes at an altitude of 4 degrees (about 4 arcminutes right at the horizon), nudging both the apparent altitude and the predicted moonset, and so the lag, slightly too high.

Feeding the corrected geometry through the Yallop relation,

q = (ARCV - (11.8371 - 6.3226·W + 0.7319·W² - 0.1018·W³)) / 10

with W ≈ 0.21 and ARCV ≈ 7.3 gives a Yallop q of about -0.32, which falls below the Zone E band and so reads as Zone F on the table below. The Odeh V-value for the same geometry is about +1.4, in the "visible only with optical aid" region, so the two criteria disagree at this margin. (The app's map shows Zone E rather than F here, because the Moon is above the horizon.) On the canonical Yallop scale that reads as follows:

A q of -0.32 sits below Zone E: by Yallop's scale not visible even with a telescope. The same engine evaluated for the following evening, with elongation above 22 degrees and W above 1 arcminute, would climb all the way into Zone A. The key lesson is that the atmosphere does not move the crescent between zones by fiat: the warm, thin desert air slightly reduces refraction, but q and V are decided by the celestial geometry. What the atmosphere changes is the accuracy of the inputs (through the refraction correction and the distortion of W) and the brightness and contrast the observer actually has to work with.

A Note on "Effective ARCV" and Elevation

A common mistake is to say that climbing a hill adds horizon dip directly to ARCV and thereby bumps the crescent up a Yallop zone. ARCV is a geometric relationship between the Moon and the Sun at sunset; horizon dip does not change the positions of either body. What dip and elevation actually change is your local horizon clearance and your extinction path (how much air the crescent's light crosses before reaching your eye). These are real and valuable advantages, but they act on visibility through clearance and contrast, not by editing q into a different band. The geometry that produces q and V is the same wherever you stand. For the dip formula, air-mass figures by elevation, terrain blockage and dark adaptation, see why altitude and terrain matter.

Cloud Cover and the Contrast Window

Refraction and extinction degrade the crescent gradually; cloud removes it entirely. A single low cloud bank on the western horizon during the 15 to 30 minute usable window ends the attempt, since crescent observation cannot be rescheduled to 2 a.m. Low cloud (below about 2000 m) is most damaging because it sits exactly where the crescent is; high cirrus is more forgiving, often thin enough to see through with merely reduced contrast.

The usable window is governed by a tension: the sky must darken enough for the faint crescent to stand out against the twilight background, while the Moon must still be high enough to escape the worst air mass and extinction near the horizon. For a marginal crescent this intersection can last only 15 to 30 minutes. Yallop's framework formalises this by evaluating his criterion at a best time of roughly four ninths of the lag time after sunset, the practical centre of the contrast window. A useful cloud forecast must be resolved by altitude level and aimed at the western horizon, not at the sky overhead.

How Hilal Vision Applies Atmospheric Corrections

Hilal Vision treats the atmosphere as a first-class input rather than a fixed assumption. The engine connects to live data from Open-Meteo and, for each location:

• pulls current temperature and barometric pressure for the exact coordinates, and recomputes the refraction correction for the real air rather than the textbook 10 degrees C and 1010 hPa standard;
• ingests cloud cover by altitude level so low cloud on the western horizon is weighted far more heavily than high cirrus;
• folds humidity and aerosol-driven extinction into the contrast window so the recommended viewing slot reflects how much of the crescent's light will survive the trip through the air.

The Best Time to Observe algorithm combines Moon altitude, Sun depression (sky darkness, ideally past nautical twilight at minus 12 degrees), and the cloud and extinction forecast. If the ideal astronomical moment falls at 7:15 p.m. but a cloud bank is forecast then, the engine shifts the recommendation to the next clear slot, turning a theoretical window into something genuinely actionable.

Expert observers can enter measured local temperature and pressure to refine the prediction. The Pro tier adds live cloud overlays on the global visibility heatmap and access to the extended ICOP historical archive for comparing tonight's conditions against decades of recorded sightings.

Practical Recommendations for Observers

If you are new to crescent hunting, pair this with the beginner's guide to spotting the crescent moon, which covers where to look and how to prepare your eyes. Four evidence-based steps follow directly from the physics:

1. Account for local temperature and pressure. The default refraction assumptions (10°C, 1010 hPa) can be far from reality in extreme climates; the correction changes the apparent altitude and therefore the predicted moonset.
2. Expect a squashed, shimmering crescent at very low altitude. Differential refraction distorts the disc; do not try to judge visibility from its apparent shape or to measure W by eye near the horizon.
3. Treat the observation window as short. For a marginal crescent, the usable window between "sky dark enough" and "Moon too low" can be only 15 to 30 minutes inside a lag of 30 to 45.
4. Favour transparency over mere cloudlessness. Extinction near the horizon, not cloud, is what dims a marginal crescent below the threshold the model predicts.

Compute the refraction-corrected geometry, q-value and V-value for your own coordinates on moonsighting.live: open the global visibility map or inspect the geometry on the moon dashboard. The Pro tier adds live cloud overlays and an extended ICOP archive for comparing predictions against historical sightings.

Conclusion

The atmosphere is the great equaliser in crescent observation. Refraction bends and distorts the crescent's apparent position and, through differential refraction, the very width W that feeds the q-value. Extinction quietly dims it as the air mass climbs near the horizon. Neither effect rewrites the Moon's Yallop q or Odeh V: what they change is the accuracy of the inputs and the contrast you actually have to work with. The crescent does not appear in a vacuum; it appears through an ocean of air, and understanding that ocean is half the battle.

References and Further Reading

• Yallop, B.D. (1997). A Method for Predicting the First Sighting of the New Crescent Moon. HM Nautical Almanac Office, NAO Technical Note No. 69.
• Odeh, M.Sh. (2004). "New Criterion for Lunar Crescent Visibility." Experimental Astronomy.
• Bennett, G.G. (1982). "The Calculation of Astronomical Refraction in Marine Navigation." Journal of Navigation. (Saemundsson's companion true-to-apparent formula is widely tabulated alongside it.)
• Kasten, F. and Young, A.T. (1989). "Revised optical air mass tables and approximation formula." Applied Optics.
• Danjon, A. (1932, 1936). L'Astronomie. (The Danjon limit.)
• Fatoohi, L.J., Stephenson, F.R. and Al-Dargazelli, S.S. (1998). "The Danjon limit of first visibility of the lunar crescent." The Observatory.
• The Islamic Crescents' Observation Project (ICOP), under the International Astronomical Center: astronomycenter.net.
• Open-Meteo. Live weather and atmospheric data API used for real-time extinction and refraction corrections. https://open-meteo.com

Clear skies and happy sighting.