Inside the ICOP Crescent Archive: How Moon Sighting Records Are Verified

The Short Answer

The Islamic Crescents' Observation Project (ICOP), founded in 1998 by Mohammad Odeh, records negative reports as carefully as positive ones, keeps cloudy reports apart and checks each claim against the computed geometry, so a naked-eye claim below the Danjon limit is flagged. About half of the 737 records behind Odeh's 2004 criterion came from it. The Hilal Vision archive holds 1,028 of its reports, for Ramadan and Shawwal from 1440 to 1446 AH (2019 to 2025).

Science relies on data. Predictive models like the Yallop and Odeh criteria are powerful, but only as good as the real-world observations used to build and test them. Behind every clean visibility map sits a decades-long ledger of people standing on rooftops, mountain ridges and coastlines at dusk, recording whether they could, or could not, see the new crescent.

That ledger is the Islamic Crescents' Observation Project (ICOP) archive, one of the most important datasets in modern crescent science. Hilal Vision currently includes 1,028 historical ICOP reports in the platform so that you can set real reports beside computed visibility maps and see, month by month, where the models hold and where they break.

The reports in the Hilal Vision archive cover the months of Ramadan and Shawwal from 1440 to 1446 AH (2019 to 2025 CE), 14 months in all. ICOP's own collection began in 1998, its founding year.

What is the ICOP archive?

The ICOP is a global network of dedicated amateur and professional astronomers who have spent more than a quarter of a century meticulously logging their attempts to spot the new crescent, the hilal. It was founded in 1998 by Mohammad Odeh, and operates under the umbrella of the International Astronomical Center (IAC). It is one of the most sustained coordinated efforts to collect first-sighting data.

Unlike casual reports, ICOP data is highly structured. A full ICOP report does not just say "seen" or "not seen". It records:

• the observer's name and the exact location (latitude, longitude and altitude);
• the date and the local time of the attempt;
• the optical aid used (naked eye, binoculars, telescope, or CCD camera);
• the sky and atmospheric conditions (haze, cloud, transparency);
• whether the crescent was seen, not seen, or whether cloud prevented any judgement at all.

That last distinction matters more than people expect, and we will return to it. The reports shown in the Hilal Vision archive are a simplified extract: each lists the country, the city and the result (seen or not seen) for the months of Ramadan and Shawwal from 1440 to 1446 AH. Even so, they let you step back through seven years of observation history and explore how the crescent has actually behaved, month after month, across the globe.

How are ICOP sightings verified?

This is the question that separates a scientific archive from a folklore collection. A single enthusiastic claim of an impossibly young crescent proves nothing; what makes the ICOP dataset valuable is the discipline applied to filtering and contextualising every report.

Positive and negative reports are both recorded. Most people assume a sighting database only stores successes. The opposite is true: a negative report, "I looked carefully under clear skies from a good site and did NOT see it", is scientifically just as valuable as a positive one. Positive reports tell you the crescent can be seen under those conditions; negative reports tell you where the limit of visibility actually lies. Without negatives, any model would be biased towards over-prediction, because you would never learn the conditions under which the crescent fails to appear. ICOP deliberately solicits and preserves negative reports, and this is one reason the dataset is trusted.

Cloudy reports are quarantined. A "cloudy" or "not observed due to weather" entry is neither a positive nor a negative. It is uninformative about the crescent and must be excluded from any analysis of the visibility limit, otherwise it would contaminate the statistics. Keeping these as a separate category, rather than miscoding them as "not seen", is a small but crucial piece of data hygiene. (The extract in the Hilal Vision archive contains only seen and not seen results.)

Reports are cross-checked against physics. When a report arrives, it can be compared against the computed geometry for that observer's location and time: the arc of vision (ARCV), the arc of light or elongation (ARCL), the difference in azimuth (DAZ) and the topocentric crescent width (W). If someone claims a naked-eye sighting at an elongation of 5 degrees, that claim sits below the Danjon limit and is flagged as physically implausible. Mohammad Odeh and the IAC apply exactly this kind of scrutiny, and implausible naked-eye claims are treated with appropriate caution rather than taken at face value.

The dataset feeds the criteria, and the criteria police the dataset. This is a virtuous loop. Odeh's 2004 criterion was derived from 737 observation records, roughly half of them drawn from ICOP. Those records were used to define the V-value visibility regions; the resulting model is then used to sanity-check new reports. The Danjon limit acts as a hard floor on this loop: a verified naked-eye sighting below roughly 7 degrees of elongation would be extraordinary, and extraordinary claims demand extraordinary evidence.

Reading the Yallop zones in the archive

When you overlay ICOP reports on a Hilal Vision map, each location is coloured by its Yallop zone. The single most common mistake people make is to read these zones as a cumulative "greater-than" ladder. They are not. They are mutually exclusive bands, non-overlapping ranges of the q-value, and reading them correctly is the difference between understanding a map and misreading it.

The q-value itself is computed as:

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

where W is the topocentric crescent width in arcminutes, evaluated at Yallop's "best time" (about four ninths of the lag time after sunset). The resulting score drops into one of six bands:

Zone F means the crescent is below the Danjon limit and cannot be seen by any means; it does not mean "the Moon is below the horizon". This distinction trips up many readers, and it is exactly the kind of misconception the archive helps dispel: you can watch verified negative reports cluster in the very zones the model predicts they should.

The complementary Odeh V-value divides visibility into four regions: V ≥ 5.65 (naked eye); 2 ≤ V < 5.65 (optical aid first, possibly naked eye after); -0.96 ≤ V < 2 (optical aid only); and V < -0.96 (not visible even with optical aid). Odeh's empirical optical-aid limit sits at about 6.4 degrees of elongation, which is why CCD imaging has pushed assisted records lower than the naked eye could ever reach.

What does a contested month actually look like in the data?

The real value of the archive emerges around disputed months, the nights when one region declares Eid or the start of Ramadan while a neighbouring region waits another day. These disputes are rarely about bad faith; they are almost always about a crescent sitting right on the edge of visibility.

A good example of a month where the geometry matters is Shawwal 1440 AH (June 2019 CE, marking Eid al-Fitr at the end of Ramadan 1440 AH). Conjunction fell on 3 June 2019 at 10:02 UTC, so the evening of 3 June is the evening in question. When that evening is reconstructed with the same geometry that ICOP reports are checked against, the picture for the Arabian Peninsula is stark:

• Elongation (ARCL) around 4 degrees. Computed at Gulf sunsets it lies between 3.9 and 4.5 degrees, far below the naked-eye Danjon limit, where Fatoohi, Stephenson and Al-Dargazelli (1998) derived the boundary at roughly 7.5 degrees. A crescent this close to the Sun cannot be resolved by the eye, however clear the sky.
• Arc of vision (ARCV) of only about 0.5 to 1.1 degrees for the Gulf sites. A low ARCV means the Moon is barely above the Sun at sunset, drowned in twilight glare with almost no lag time between sunset and moonset. The topocentric crescent width W is only about 0.04 arcminutes.
• Moon age around 5 to 7 hours. Such a young Moon is hopeless on any criterion, but it is the geometry, not the hours, that makes it so. People hear "the Moon is 14 hours old" or "6 hours old" and assume that settles the question. It does not. Age alone is not a parameter in the Yallop or Odeh models.

The computed Yallop q-value for the Gulf sites is about -1.05 to -1.11, deep in Zone F, below the Danjon limit entirely, and the Odeh V-value is about -6. When a claimed sighting and the model disagree at this level, the ICOP dataset provides the neutral scientific record against which the religious decision can be evaluated. The data does not adjudicate the ruling; it establishes what was physically possible on that night, and that is a powerful contribution to the moon sighting versus calculation conversation.

The youngest crescents on record underline how thin this margin truly is: Mohsen Mirsaeed's binocular-aided sighting in 2002 was recorded at about 11 hours 40 minutes after conjunction, while Thierry Legault's celebrated 2013 CCD image was captured at essentially the instant of conjunction (an age near zero hours) at an elongation of only about 4.4 degrees, far below the naked-eye Danjon limit.

Why is moon age such a poor predictor?

If you take one practical lesson from the ICOP archive, let it be this: the age of the Moon, the elapsed hours since conjunction, is not a parameter in either the Yallop or the Odeh criterion, and it never should be.

The reason is geometry. Two crescents of identical age can have very different elongations depending on where the Moon is in its elliptical orbit. Near perigee the Moon races through its orbit and gains elongation quickly; near apogee it dawdles. So an 18-hour-old crescent at perigee may be far easier to see than a 22-hour-old crescent at apogee. Age is a clock; visibility is a geometry problem.

The same applies to lag time taken in isolation. A long lag between sunset and moonset is helpful, but it is not sufficient on its own; a crescent can linger above the horizon long after sunset and still be too close to the Sun in total angular distance to be seen. What actually decides visibility is the combination of four parameters working together:

• ARCV (arc of vision): how high the Moon sits above the Sun at sunset.
• ARCL (arc of light): the Moon-Sun elongation, which drives crescent width and underpins the Danjon limit.
• DAZ: the difference in azimuth between Sun and Moon.
• W: the topocentric crescent width in arcminutes, the thickness of the sliver you are actually trying to see.

Topocentric parallax adds a further complication: the Moon's horizontal parallax of about 57 arcminutes lowers its apparent topocentric altitude relative to the geocentric value by up to roughly 1 degree near the horizon. That single degree can be decisive. This is why the same conjunction yields a sighting in one country and a blank sky in another, as explored in why the crescent is visible in some countries but not others. Filter a single archived month and you will see the visibility wave sweep westward as the geometry slowly improves.

Validating the predictive models

When you open the Hilal Vision archive, you can select a historical month and see the ICOP reports for it (country, city, and whether the crescent was seen) next to the computed Yallop visibility for major cities. Where the model and the observers disagree, that disagreement is itself a finding: it points to unusually good atmospheric transparency, an over-optimistic claim, or a genuine edge case worth studying.

This is the everyday method by which the science behind the crescent advances. Each new month adds reports; each report tightens or challenges the model. Hilal Vision's triple-engine Hijri calendar draws on the same underlying physics, so the archive and the calendar are two views of one coherent system.

Testing the limits of human vision

One of the most valuable uses of historical ICOP data is probing the hard floor beneath which no crescent can be perceived: the Danjon limit. André Danjon first reported the effect in 1932 and quantified it in 1936 in L'Astronomie: below a minimum elongation, the crescent cannot be seen. The empirical threshold sits between about 5 and 7.5 degrees; Fatoohi, Stephenson and Al-Dargazelli (1998) derived roughly 7.5 degrees for the naked eye, while the optical-aid limit is closer to 6.4 degrees. The physical cause remains a hypothesis, not settled fact: Danjon attributed it to lunar topography and cusp foreshortening; Bradley Schaefer argued for photometric brightness fall-off towards the cusps; others point to atmospheric seeing and contrast thresholds against a bright twilight sky.

The archive bears the limit out empirically. Filter for the most extreme attempts and you find that verified naked-eye claims below about 7 degrees of elongation are almost non-existent. Meanwhile the rise of CCD imaging has steadily pushed assisted records lower, capturing crescents that no human eye could perceive but that a long-exposure sensor can tease out of the glare.

A legacy of data, and how to use it

Having this historical record does more than validate mathematics; it educates the community. By studying past months where the calendar was contested, you can see objectively what was scientifically possible on those nights, separate from the religious decision that followed. That is a calm, evidence-based contribution to a conversation that is too often heated.

You can put the archive to work right now: compute your own crescent prediction for the next new moon at moonsighting.live, then log what you saw after sunset. Your report joins the same scientific record that underpins the Odeh criterion. If you are just starting out, our beginner's guide to spotting the crescent will prepare you, and the moon dashboard shows the current phase and geometry at a glance. Pro members gain access to cloud-cover overlays and the extended ICOP archive, invaluable for reconstructing older contested months in full detail.

The ICOP archive is a testament to the dedication of observers around the world, watching the western horizon at dusk so that the rest of us can have a calendar grounded in both faith and physics.

See Also

• Crescent Moon Observation Log: How to Become a Master Hilal Observer

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. (Odeh founded ICOP in 1998; the criterion was derived from 737 observation records, about half from ICOP.)
• Danjon, A. (1932, 1936). L'Astronomie. (The first report and subsequent quantification of the Danjon limit.)
• Fatoohi, L.J., Stephenson, F.R. & Al-Dargazelli, S.S. (1998). "The Danjon limit of first visibility of the lunar crescent." The Observatory.
• Kasten, F. & Young, A.T. (1989). On atmospheric air-mass at low altitudes; Bennett (1982) and Saemundsson on atmospheric refraction.
• Sachs, A. & Hunger, H. Astronomical Diaries and Related Texts from Babylonia.
• Islamic Crescents' Observation Project / International Astronomical Center: astronomycenter.net.

Clear skies and happy sighting.