Night Sky Dimming Twice as Fast as Satellites Detected, Study Finds

A child who can identify 250 stars from their backyard tonight might manage to spot fewer than 100 from that same spot by the time they turn eighteen. The stars themselves will not have moved or dimmed. What will have changed is the veil of artificial light scattered through Earth's atmosphere, growing thicker each year, washing out the faintest points of light until they vanish from human sight.

This sobering trajectory comes from a study published in Science in 2023, built on an unusual dataset: 51,351 observations submitted by ordinary people between 2011 and 2022. These were not measurements taken by instruments or satellites. They were reports from volunteers who looked at a familiar constellation and selected the chart that best matched the faintest stars visible to their naked eye. The researchers, analyzing this ground-level data, found that sky brightness was increasing by an average of 9.6 percent each year across the locations where observations were made. Extrapolated across eighteen years, that compounds to more than a fourfold increase in the artificial glow overhead.

What makes this finding striking is how it diverges from what satellites had been telling us. A previous study using the VIIRS Day/Night Band instrument, published in 2017, had detected Earth's artificially lit outdoor area expanding by about 2.2 percent annually, with total measured radiance increasing at roughly 1.8 percent per year. The ground-based observations suggested the problem was accelerating far faster than orbital instruments could detect. The reason lies in what each method actually measures. Satellites are relatively blind to blue light, the dominant wavelength in white LED bulbs now widespread in outdoor lighting. They also struggle to detect light traveling sideways through the atmosphere, which scatters efficiently and contributes to skyglow far from its source. A shift in the color and direction of outdoor lighting can therefore transform what a person sees in the night sky without producing the same signal in a satellite's radiance record. The two measurements are related but not equivalent—one captures light leaving Earth in specific wavelengths and directions; the other captures what remains visible through the resulting haze.

The volunteers who submitted these observations were part of Globe at Night, a citizen-science program run by the National Science Foundation's NOIRLab. The dataset covered 19,262 locations, though it was not evenly distributed globally. North America and Europe supplied most of the reports, with annual brightening rates of 10.4 percent and 6.5 percent respectively. The researchers filtered out observations affected by twilight, moonlight, and snow, then compared the remaining data against a global sky-brightness model. The 9.6 percent average represents the trend that best fit the eleven-year span of observations.

But this acceleration is occurring against a baseline that was already profoundly altered. According to the 2016 World Atlas of Artificial Night Sky Brightness, more than eighty percent of humanity already lived under light-polluted skies. The Milky Way was invisible to more than one-third of the world's population—roughly sixty percent of Europeans and nearly eighty percent of North Americans. The effect is not confined to city centers. Artificial light scatters and creates domes visible hundreds of kilometers from their source, so a landscape that appears dark at ground level may still have a brightened sky above it. The loss of stars also follows a nonlinear pattern in human perception. Star brightness is measured on a logarithmic scale, and far more stars cluster at the faint end of naked-eye visibility than the bright end. A modest decline in the faintest stars visible can therefore erase a large number of them from view.

The 9.6 percent figure describes what the data showed between 2011 and 2022. It does not mean every location brightens at that rate, nor does it guarantee the trend will continue unchanged. Weather, aerosols, individual eyesight, dark adaptation, and local lighting variations all affect each observation. The large dataset and statistical modeling reveal a broad trend through that noise, but they do not eliminate uncertainty. The rate of sky brightening is also not inevitable. Skyglow depends on how much light is used, where it is directed, its color, and when it remains on. Shielding a fixture so it illuminates the ground rather than the sky addresses a different problem than simply replacing a bulb with a more efficient one. The 250 stars in the study's example are not a fixed inheritance destined to decline to 100. They are an illustration of the current measured trajectory—one that can still be altered by choices made here on Earth.