How Dark Is Your Sky?

What the Darkness Figure Tells You

Next to every forecast, Clear Skys shows how dark the sky is at that location: a band such as Rural or City sky, an approximate Bortle range, and an SQM value.

SQM stands for sky quality meter reading, the brightness of the sky overhead in magnitudes per square arcsecond. Higher numbers are darker. A pristine site reads about 21.9; a suburb around 19 to 20; a city centre below 18.

Darkness is a property of the place, not of the night. It does not change from one night to the next, so it is shown alongside the forecast score and never changes it. A perfect night in a city still shows city sky, and that is the point: the score tells you whether to go out, the darkness tells you how much you will see when you do.

What can you see at each Bortle level? →

Where It Comes From: Satellites at Night

The starting point is NASA’s Black Marble data: a year of night-time images of the whole Earth from the VIIRS instrument on the Suomi NPP satellite, combined into a single cloud-free, moon-adjusted composite at about 500 metre resolution. We use the 2025 composite, the most recent full year.

A satellite looks down, so it measures light going up from the ground. That is not the same as how bright the sky looks to someone standing there. A dark village 20 km from a city has almost no light of its own, yet its sky is visibly washed out by the city’s glow. To estimate darkness you have to model how light from everywhere around a place reaches the sky above it.

From Night Lights to Sky Glow

For every square kilometre on Earth, we add up the light from every lit area within 250 km, weighting each by distance so that nearby towns count far more than distant cities. Light pollution really does carry that far: a large city brightens the horizon from well over 100 km away.

This is an empirical estimate, not a full simulation of how light scatters through the atmosphere. Real sky glow also depends on terrain and elevation, the direction lights shine, the colour of the light and the state of the atmosphere on the night. Distance weighting captures the main geographic relationship between where light comes from and how bright the sky is; the ground measurements described below calibrate the result to sky brightness as it is actually observed.

A few details matter. We only count light from land, so offshore oil platforms and gas flares, which are very bright to the satellite, do not brighten the coast. We ignore the faint background signal the satellite records over empty land at high latitudes, which would otherwise make the whole Arctic glow. And we work at about 1 km, because the model cannot honestly claim more precision than that.

Calibrating Against the Ground

A sum of satellite light is only a relative number. To turn it into real sky brightness, we compared it with measurements taken on the ground by volunteers in Globe at Night, a long-running citizen-science project in which people record sky brightness with handheld meters.

We used readings from 2019 to 2025 and kept only those taken under clear skies, with the sun at least 15° below the horizon and the moon down. That left about 4,000 readings at around 1,500 locations worldwide. We fitted the model to them and then tested it on readings it had not seen, repeating that five times so that every location was tested once. Those held-out results are the accuracy figures below.

How Accurate It Is

On ground readings it was not fitted to, the model is typically within about 0.4 SQM, with a median error of 0.3. That puts most places in the right band, but it is honest about three limits.

It cannot separate Bortle 1, 2 and 3. Those classes differ by only 0.1 to 0.2 SQM, finer than the model can resolve, so the darkest places share one band: Very dark.

City centres are probably brighter than shown, by about one magnitude. They still fall in the City sky band. Satellites under-count some modern LED lighting, which is richer in blue light than the satellite sensor sees well, and many city-centre readings include glare from nearby lamps.

Your exact spot can differ. A hill, a line of trees or one bright light next to you can make your own site noticeably darker or brighter than the surrounding square kilometre.

See how we measure forecast accuracy →

Why We Built Our Own

Excellent light pollution atlases already exist, but the best known are built on satellite data from 2015 and published under licences that restrict how they can be used. Street lighting has changed a great deal since then, much of it switching to LEDs.

Building our own means every location gets a darkness figure from recent data, we can rebuild it every year when NASA publishes a new composite, and we can be specific about how it was made and how accurate it is. It is a simpler model than the scientific atlases, and we say so; what it offers is that it is current, global and open about its method.

Using It

Use the forecast and the darkness together. When the forecast is good, check the darkness band for a few places near you. Moving from Suburban to Rural, often only 20 to 30 minutes’ drive, is the step where the Milky Way goes from invisible to obvious. Reaching Very dark is a different experience again.

Data credits. Night lights: NASA Black Marble VNP46A4, VIIRS/NPP Lunar BRDF-Adjusted Nighttime Lights Yearly, doi:10.5067/VIIRS/VNP46A4.002. Ground calibration: Globe at Night (globeatnight.org), licensed CC BY 4.0.