Northeast Asia at Night: A Satellite Night-Lights Map

About This Chart

Data Visualization Method

• Data: Daily nighttime radiance from NASA's Black Marble VNP46A2 product (Suomi NPP / VIIRS Day/Night Band). Only the observed radiance layer was used; NASA's gap-filled layer was not.
• Period and area: Observations from 31 May to 28 August 2026, with priority on the most recent 60 days (from 30 June) and a reach of up to 90 days only where the 60-day window held no clear observation. The map covers Northeast Asia from roughly 15°N to 55°N, drawing on 69 complete observation dates.
• Selection: For each location, up to three of its most recent clear-sky, high-quality nights were kept after removing cloud, snow, and low-quality pixels. Water pixels were removed first, so ships, fishing boats, and offshore lights are excluded by design rather than guessed at later by brightness.
• Processing: Coastal pixels were reclassified as land or water against Natural Earth 1:10m land boundaries so that coastal cities stayed while offshore points were dropped. Valid land pixels were aggregated into roughly 3 km cells; within each cell the median of valid radiance was taken, and the median of up to three recent daily values became the final value. A low-light cutoff at 0.25 nW cm⁻² sr⁻¹ trimmed noise-floor readings. No spatial interpolation and no gap-filling were applied — cells with no valid observation remain No Data.
• Chart form: An animated point map of about 80,000 land points. Each point sits at a real observed land pixel and is colored by radiance class — four base brightness levels plus three highlight classes for the brightest points. The highlight animation is a presentation effect; its timing does not represent when the satellite observed each point.

Creator's Note

This chart began as an attempt to place the whole region on a single comparable canvas — to see where recent night light actually falls across Northeast Asia, rather than freezing one, possibly cloudy, night. Because no single date passed cloud and quality checks across the entire region, each location instead carries its own most recent clear-sky reading.

What stands out is texture rather than a scoreboard: the continuous urban corridors of eastern China, the Pacific-coast belt through Japan, and the sharp north–south difference across the Korean peninsula. These are patterns of observed light — a record of where the sensor saw brightness on clear nights, not a measure of how many people live in a place or how much power it uses.

Data Sources

• NASA Black Marble (VNP46A2, Collection 2) — daily nighttime radiance from the Suomi NPP / VIIRS Day/Night Band; the observed radiance layer only (DOI 10.5067/VIIRS/VNP46A2.002).

Limitations

This is not a single-night satellite photograph. It is a recent clear-sky composite: each location shows the median of up to three of its most recent clear, high-quality nights within the window, so neighboring points may come from different dates.

Coverage reaches about 95.6% of land cells; the remaining 4.4% (37,974 cells) are left as No Data rather than filled in. Faint areas are also shaped by the 0.25 low-light cutoff and by the thinning used to keep the map light enough to animate, so a dim or empty patch reflects a mix of real darkness, how observable the location was, and these processing choices.

Each point is a cell-level summary placed at a representative land pixel, not a single building or streetlight, and the number of points reflects the observation grid and the thinning — not a count of structures, households, or people.

What This Chart Doesn't Show

Brightness here is not population, electricity use, or economic output, and differences between points or between countries cannot be read as cause and effect. No Data does not necessarily mean "no light" — it can mean persistent cloud, snow, or too few valid observations at that location.

Offshore lights — ships, fishing fleets, and platforms — were removed on purpose; no water pixels appear in the final map, so the empty seas are an intended result of land-only processing, not missing data. The twinkling of the brightest points is a visual highlight; its timing is generated for the animation and does not correspond to when each point was observed.