Human cloud reports overstate nighttime clearing: a ceilometer, GOES, and VIIRS comparison at eight Canadian airports

<jats:p>Canada’s eight busiest airports all report more clear sky at night than by day. In six years (2020–2025) of hourly METAR, the human-augmented stations report clear-or-few sky (0–2 oktas) 7–14 percentage points (pp) more often in darkness: enough to distort diurnal climatologies and nighttime sky-quality assessments. We test whether the sky or the observer is responsible. Taking nights as units, each station is compared against three references that do not share a human’s night vision: a paired automated ceilometer (four cities), the geostationary GOES ABI clear-sky mask, and the polar-orbiting VIIRS cloud mask matched to reports scene by scene, all on snow-free ground screened daily with reanalysis snow depth. The GOES product changed algorithm midrecord in 2021, and its two algorithms report day-to-night contrast differing by up to 22 pp; the current one alone is the reference, the rest reported. All three references point the same way: the human excess over GOES runs from +9 to +23 pp and excludes zero at all eight sites; the matched VIIRS test confirms seven; and the human-minus-automated difference at the four partnered cities runs +13.5 to +18.0 pp. Against a single algorithm the satellite resolves no nocturnal clearing at any station pixel, and over Alberta it reads the nights as cloudier. There is almost no real clearing to exaggerate. Residual snow mutes the VIIRS test at Edmonton; its automated partner confirms the bias independently. Nighttime human cloud fractions run optimistic and should not be taken as clear-sky truth.</jats:p>