China's Urban Growth Over 50 Years

About This Chart

Data Visualization Method

• Data: Residential population (GHS-POP R2023A) and built-up surface (GHS-BUILT-S R2023A), combined into a single self-derived measure called Reconstructed Urban Intensity on a 0–1 scale. This is not observed night-time satellite light.
• Period and area: 1975 to 2025 in 5-year steps (11 frames), covering mainland China plus Hong Kong and Macao. Taiwan is not included.
• Selection: Cells below an intensity of 0.13 are hidden, and a connectivity filter keeps clusters of three or more cells, along with smaller clusters that already reach a strong early core. This is display sparsification for a nationwide view, not a judgment that the underlying values are wrong.
• Processing: The 1 km source grid is aggregated into 10 km blocks; population is summed and divided by land cells, and built-up area is expressed as a fraction. The two are joined through a fixed logarithmic scale, with a built-up base term and a smaller population adjustment, then clipped to 0–1. The same constants are applied to every year, with no per-year rescaling, so the frames stay directly comparable.
• Chart form: An animated timeline map on a dark base, where the index is read through point color and size across the 1975–2025 sequence. There is no twinkle effect and no radiance unit; brightness here is an encoding of the derived index.

Creator's Note

The first idea was to show fifty years of China's city lights spreading the way a night-time satellite would record them. For the sake of long-run comparability, that approach was rebuilt into Reconstructed Urban Intensity, a composite of GHSL population and built-up area held to one fixed scale across every year.

The chart is meant to be read as a process rather than a ranking. Settlements begin as isolated points, gather into clusters, and finally link into continuous urban belts around the Yangtze and Pearl River Deltas, with the corridor between Chengdu and Chongqing joining more recently. The reading is spatial. It is not a claim about which city is largest, nor about the economic causes behind the change.

Data Sources

• GHS-POP R2023A — residential population per grid cell, one of the two inputs to the index.
• GHS-BUILT-S R2023A — built-up surface per grid cell, the other input.
• Natural Earth — country boundary used to mask China, Hong Kong and Macao.

Limitations

Reconstructed Urban Intensity is a derived index built for this chart, not an official JRC classification and not a night-time radiance measurement. Its inputs are themselves modeled: GHS-POP disaggregates census and administrative population onto a grid using built-up information, so a single cell is an estimate rather than a direct head count, and GHS-BUILT-S combines several sensors and reference epochs, which means not every 5-year step is an independent direct observation.

The processing choices shape how the map looks. The 10 km aggregation merges small towns, narrow corridors and internal city structure; the combining weights and the logarithmic constants are values chosen by the author; and the faint-signal threshold and connectivity filter hide very weak or tiny scattered cells for display. Hidden cells are not errors. Coastal blocks are divided by the number of included 1 km cells rather than an exact shoreline area.

Because of this, grid counts and row counts are artifacts of the display and the processing. They should not be read as a count of distinct places, nor used to rank regions or compare countries, where observation coverage, land area and terrain all differ.

What This Chart Doesn't Show

The 2025 frame is a projection, not a present-day measurement. In R2023A both inputs treat 2025 as a future estimate produced at the 2023 release, so the final year should be read as projected rather than observed, and the earlier series should not be described as one uniform run of direct observations. The glow, color and size are visual encodings of the derived index; they do not represent electricity use, GDP, lighting output or satellite radiance.

The map also does not show official administrative city areas, exact urban population rankings, or official metropolitan and transport boundaries. The export keeps only active cells for each year, so a row appearing in a later frame means a cell has crossed the display threshold, not that a city has physically moved or newly appeared at that spot.