How we score a night.
Every number here comes from public forecast data and astronomical geometry. Published in full, so you can decide how much to trust it.
Cloud
gateLayers combine as independent attenuators rather than summing. Low cloud is opaque; high cirrus is weighted at roughly half, because thin cirrus dims a sky without ending the session.
gates allTransparency
depthHow much haze and water vapour dim faint objects, estimated from relative humidity. Over bright sites, haze also scatters more existing artificial light back into the sky.
sets depthSeeing
estimateReal seeing is driven by upper-air turbulence that public point forecasts do not publish, so we use surface wind as a proxy. This is the weakest of the five factors and we label it an estimate everywhere it appears.
sets sharpnessMoonlight
scaleScaled by illuminated fraction and the moon’s altitude. The added contrast loss is smaller where artificial skyglow has already raised the background.
scales resultDarkness
scaleRamps from the end of civil twilight to full astronomical night at 18° below the horizon, computed for your coordinates rather than your time zone.
scales resultWhat the bands mean
The best window is the longest unbroken run of hours scoring 35 or better inside astronomical darkness. Every hour on a forecast page opens to show exactly where its points went.
Where the data comes from
- Cloud, humidity, wind and precipitation from MET Norway (NLOD 2.0 / CC BY 4.0).
- Place names and coordinates from GeoNames (CC BY 4.0).
- Sun and moon positions are computed locally from astronomical algorithms; no live ephemeris service is queried.
- Artificial skyglow uses Earth Observation Group VIIRS Nighttime Lights (CC BY-SA 4.0) with the Duriscoe radiance-propagation model. It is a peer site measure, not a permanent score penalty.
- Sky-darkness and visual-limit values are model ranges, not meter readings. The 2020 source can understate blue-rich LED conversions and cannot see a lamp beside your telescope.