Executive Summary. Bangladesh's economic activity, measured from orbit, is concentrating into a single 50 km wide corridor faster than the state is governing it, and that corridor is now the country's primary public health liability. The Dhaka, Gazipur, and Narayanganj cluster leads the country simultaneously in nighttime radiance, in built-up land, and in tropospheric NO2 (170-201 micromoles per square meter). Nationally, Bangladesh ranked second-most-polluted on earth in 2025, with PM2.5 averaging 66.1 micrograms per cubic meter, 13.2 times the WHO annual guideline of 5 (IQAir, 2025). The governing thought: spatial deconcentration is no longer a regional-equity nicety, it is the lowest-cost lever Bangladesh has for cutting air-pollution mortality and resilience risk at the same time. If the next decade's economic-zone investment fails to move the nightlight map away from one bright cluster toward several, the air and the heat in the core will get worse on the current trajectory, not better.

At night, Bangladesh glows, and the glow has intensified for three decades. DMSP-OLS satellites first recorded the country's nighttime radiance in 1992 at a national mean of 1.58 units. By 2012, the last year of the DMSP era, that figure had nearly doubled to 2.87. When the more sensitive VIIRS instrument took over in 2014, national mean radiance climbed from 0.38 in 2014 to 0.92 in 2024, a 142% rise in a single decade.

Every photon in the nightlight record is a human decision: a factory turning on, a street lamp installed, a household connected to the grid. But not every photon is progress. Some mark the unplanned sprawl of cities expanding faster than any agency can regulate. Others mark industrial zones whose emissions degrade the air their workers breathe. This is Bangladesh's luminous transformation, read through satellites that never sleep.

The lights are on

The nightlight record spans two satellite eras: DMSP-OLS (1992-2012) and VIIRS-DNB (2014-2024). The two sensors use different radiance scales, so comparing absolute values across the 2013 boundary is misleading. Within each era, the trajectory is unmistakable.

During the DMSP period, national mean radiance rose from 1.58 in 1992 to a peak of 2.98 in 2010, before settling at 2.87 in 2012. The dip in 2006 (1.58, identical to the 1992 baseline) is likely a sensor-calibration artifact rather than a real regression, since surrounding years show steady growth. Total sum radiance tells the same story: from 242,282 in 1992 to 440,775 in 2012, an 82% increase.

The VIIRS era, with finer resolution and wider dynamic range, shows the acceleration continuing. Mean radiance rose from 0.38 in 2014 to 0.92 in 2024. Sum radiance more than doubled, from 58,755 to 141,462. The sharpest jump came between 2016 and 2018, when mean radiance leapt from 0.38 to 0.64, coinciding with Bangladesh's push toward universal electrification.

The 2024 city-level data shows where the light concentrates. Narayanganj, the industrial twin of Dhaka, leads the country with a mean radiance of 8.81, followed by Dhaka at 7.37 and Chittagong at 6.75. These three cities produce more light per unit area than the next seven cities combined. Gazipur (4.10) and Comilla (3.39) form a second tier. The remaining five cities, including divisional capitals such as Rangpur (2.09) and Barishal (2.52), cluster between 2 and 3.1 units.

This concentration is itself a policy signal. When three urban cores generate most of a country's nighttime light, economic activity is dangerously concentrated. A flood, cyclone, or lockdown that hit Dhaka-Narayanganj-Chittagong would register as a sudden dimming visible from space, with the GDP loss to match.

The concrete is spreading

Nightlights show where activity happens. Built-up area data shows where the earth has been permanently transformed. The Global Human Settlement Layer, derived from Landsat imagery at 30-meter resolution, provides this record from 1975 to 2020.

In 1975, Bangladesh had approximately 456 km2 of built-up land. By 1990 that had nearly tripled to 1,322 km2. By 2000 it reached 1,711 km2. By 2020 the national built-up footprint stood at 2,678 km2.

That is a 487% increase in 45 years. Roughly 2,222 km2 of land that was once paddy field, wetland, or forest is now concrete, brick, and asphalt. The entire Dhaka metropolitan area occupies about 300 km2, so Bangladesh has paved over an area equivalent to seven Dhakas since 1975.

The rate of expansion is not slowing. Between 1975 and 1985, Bangladesh added 482 km2 of built-up area (48.2 km2/year). Between 2010 and 2020, it added 535 km2 (53.5 km2/year). The pace has held near 50 km2 per year for four decades, but the denominator, the remaining unconverted land, is shrinking, so each new square kilometer of concrete now displaces a higher share of what is left.

The fastest expansion has followed transport corridors: the Dhaka-Chittagong highway, the Dhaka-Mymensingh corridor, and the ribbon development that turned Gazipur from a semi-rural district into an extension of Dhaka's urban fabric. These are not planned cities expanding to a master plan. They are accretions, factories, housing colonies, and commercial strips that grow along roads and rivers, with drainage and sewage added (if ever) as an afterthought.

The air is poisoned

The price of urbanization is measured not only in lost farmland but in degraded air. Sentinel-5P, launched in 2017, provides the first consistent satellite record of tropospheric nitrogen dioxide (NO2) over Bangladeshi cities. The 2023 data reveals a geography of pollution that mirrors the nightlight and built-up maps almost exactly.

Narayanganj records the highest mean NO2 concentration at 200.99 umol/m2, followed by Gazipur at 180.53 umol/m2 and Dhaka at 170.08 umol/m2. These three cities form the industrial heart of Bangladesh and share the same air mass. A factory in Gazipur does not pollute only Gazipur; its emissions drift south across Dhaka and Narayanganj in a continuous plume.

The gap between the top three and the rest is vast. Chittagong, the second-largest city, registers only 69.12 umol/m2, less than half of Dhaka's level. Rajshahi (57.85 umol/m2) and Comilla (52.80 umol/m2) occupy the middle tier. The cleanest cities in the dataset, Sylhet (37.79 umol/m2) and Rangpur (38.87 umol/m2), breathe air with roughly one-fifth the NO2 concentration of Narayanganj.

NO2 is both a direct health hazard (respiratory inflammation, asthma, cardiovascular damage) and a precursor to ground-level ozone and secondary fine particulate matter. The satellite cannot measure PM2.5 directly, and the NO2 map should not be read as a PM2.5 map: the two have different sources, lifetimes, and seasonal cycles. They do, however, share the corridor's dominant emitters. The same brick kilns, diesel fleets, and unfiltered industry that drive NO2 here also drive primary PM2.5, which is why the national PM2.5 average reached 66.1 micrograms per cubic meter in 2025, 13.2 times the WHO annual guideline of 5 and second-worst of any country measured (IQAir, 2025). The NO2 corridor is where that national burden is densest, not a substitute measurement for it.

The urban heat island effect compounds the air-quality problem. Satellite-derived land surface temperature data shows cities running significantly hotter than their rural surroundings.

Dhaka registers an urban land surface temperature of 30.91 degrees Celsius, against 28.86 degrees in surrounding rural areas, a UHI intensity of 2.05 degrees. Chittagong shows a similar pattern at 1.93 degrees above rural baseline. Khulna runs 1.70 degrees hotter. Only Rajshahi (0.57 degrees) shows a modest urban heat premium, likely because its surrounding landscape is already warm and dry.

Higher urban temperatures speed the photochemical reactions that convert NO2 into ozone. They raise energy demand for cooling, which raises power-plant emissions. And they directly harm human health: heat stress, dehydration, and heat stroke fall hardest on outdoor workers and the urban poor without air conditioning. UHI and air pollution are not separate problems. They are the same problem, driven by the same cause: unplanned, concrete-heavy urbanization without adequate green space, ventilation corridors, or emissions controls.

What the satellites cannot see

The nightlight record cannot distinguish a well-lit street from a diesel generator running illegally behind a garment factory. The built-up classification cannot tell you whether a building has running water, structural integrity, or fire exits. The NO2 map cannot count the children with asthma in Gazipur or the construction workers breathing brick-kiln dust in Narayanganj.

Satellites measure surfaces, not suffering. They detect radiance, not inequality. That Narayanganj glows brightest and also leads in NO2 is not a coincidence. It is a compression of economic pressure into one geography: the workers who need the factory jobs, the owners who resist emission controls, and the municipal authorities who lack the budget or mandate to enforce them.

The data also cannot capture the informal economy in the gaps: the roadside welders, the recycling yards, the leather tanneries that relocated from Hazaribagh to Savar but carried their pollution with them. These sit below the spatial resolution of any orbiting sensor, but their cumulative impact does not.

What policy must do

The satellite evidence points to three interventions, each with an owner, a mechanism, and a success signal that this analysis can re-check from the same public datasets.

First, force the economic-zone pipeline to move the map, not just the budget. Owner: BEZA (Bangladesh Economic Zones Authority), reporting to the Prime Minister's Office. The designated zones in Rajshahi, Khulna, and Rangpur do not yet register as significant brightening in VIIRS. Mechanism: tie the next tranche of zone infrastructure and tax-holiday approvals to occupancy and grid-load milestones, not groundbreakings. Success signal: by 2030, at least two cities outside the Dhaka-Narayanganj-Chittagong cluster cross a VIIRS mean radiance of 4.0 (the current Gazipur level), and the top-three cluster's share of national sum radiance falls below its 2024 level. Both are measurable in the same annual VIIRS series used here.

Second, put enforceable, automatic penalties on the Dhaka-Gazipur-Narayanganj corridor's largest emitters. Owner: Department of Environment, with NBR for the penalty-collection link. At 170-201 umol/m2 of NO2, the corridor is running an open-air experiment on millions of lungs. Brick kilns, garment-factory boilers, and diesel vehicles are the primary sources, and the abatement technology is already standard in peer economies. Mechanism: continuous emissions monitoring on the largest registered kilns and boilers, with breaches triggering a fixed per-day fine deducted at source rather than discretionary annual inspections that can be gamed. Success signal: corridor mean NO2 in the same Sentinel-5P series falls below 150 umol/m2 by 2028, and the count of monitored sites in continuous compliance rises year over year.

Third, price heat into the building-permit system. Owner: RAJUK and the city corporations, under the National Building Code. Mechanism: condition every new permit above a floor-area threshold on a minimum permeable-surface and tree-canopy ratio, verified at occupancy, not at design submission. Success signal: a measurable cut in corridor UHI intensity in the MODIS LST series, with Dhaka's 2.05-degree premium trending down rather than up over a five-year window. A one-degree reduction in peak UHI is a direct cut to peak cooling demand and to heat-related mortality among outdoor workers, the population the satellites cannot see but the morgues can count.

The strongest objection to all three is sequencing: deconcentration is slow, and the corridor's emissions are killing people now. That objection is correct and is the reason the three are ordered as they are. Enforcement (second) and heat pricing (third) are the near-term harm-reduction levers that work inside the existing geography; deconcentration (first) is the structural lever that prevents the next corridor from forming. The conclusion would change if the economic-zone data showed the secondary cities already pulling activity away from the core, or if corridor NO2 were falling on its own. Neither is visible in the satellite record as of 2024. Until one of them is, the glow keeps concentrating, and so does the cost.

Bangladesh's glow is a sign of ambition and growth. It is also a sign of a country urbanizing faster than it can sustain its cities. The satellites see the glow. They do not see the cost, but the data lets us calculate it, and it lets us check whether policy is moving the number.

This is Part 2 of the "Satellite Bangladesh" series. Part 3 will examine what is happening to the land: forest loss, vegetation health, and the battle between crops and concrete.

Sources

• DMSP-OLS nightlights: NOAA National Centers for Environmental Information, Defense Meteorological Satellite Program, 1992-2013. https://ngdc.noaa.gov/eog/dmsp.html. Internal output: bd_gis/outputs/nightlights/nightlights_timeseries.csv.
• VIIRS DNB nightlights: NASA/NOAA Suomi NPP VIIRS Day/Night Band, 2014-2024. https://www.earthdata.nasa.gov/sensors/viirs. Internal output: bd_gis/outputs/nightlights/nightlights_timeseries.csv.
• GHSL built-up area: European Commission Joint Research Centre, Global Human Settlement Layer, 1975-2020. https://ghsl.jrc.ec.europa.eu/. Internal output: bd_gis/outputs/urbanization/builtup_timeseries.csv.
• Sentinel-5P TROPOMI NO2: ESA Copernicus Sentinel-5 Precursor, 2019-2023. https://sentinel.esa.int/web/sentinel/missions/sentinel-5p. Internal output: bd_gis/outputs/airquality/no2_urban_2023.csv.
• MODIS LST / UHI: NASA Terra/Aqua MODIS MOD11/MYD11, 1 km. https://lpdaac.usgs.gov/products/mod11a2v006/. Internal output: bd_gis/outputs/climate/urban_heat_island.csv.
• National PM2.5 and global ranking (2025): IQAir, 2025 World Air Quality Report. Bangladesh annual mean PM2.5 66.1 ug/m3, second-most-polluted country; WHO annual PM2.5 guideline 5 ug/m3. https://www.iqair.com/newsroom/waqr-2025-pr
• WHO air quality guideline: World Health Organization, Global Air Quality Guidelines, 2021 (annual PM2.5 guideline 5 ug/m3). https://www.who.int/news-room/fact-sheets/detail/ambient-(outdoor)-air-quality-and-health-air-quality-and-health)
• Administrative boundaries: geoBoundaries ADM2, William and Mary geoLab. https://www.geoboundaries.org

Data Sources

• Nighttime Lights (DMSP-OLS): NOAA National Centers for Environmental Information, Defense Meteorological Satellite Program. 1992-2013. Resolution: ~2.7 km.
• Nighttime Lights (VIIRS DNB): NASA/NOAA Suomi NPP VIIRS Day/Night Band. 2014-2024. Resolution: ~750 m.
• Built-up Area (GHSL): European Commission Joint Research Centre, Global Human Settlement Layer. 1975-2020. Resolution: 30 m (Landsat-derived).
• Tropospheric NO2 (Sentinel-5P TROPOMI): European Space Agency, Copernicus Sentinel-5 Precursor. 2019-2023. Resolution: 3.5 x 7 km.
• Land Surface Temperature (MODIS): NASA Terra/Aqua MODIS LST (MOD11/MYD11). Used for urban heat island analysis. Resolution: 1 km.
• National Air Quality (PM2.5): IQAir, 2025 World Air Quality Report. Bangladesh annual mean PM2.5, country-level ranking. WHO annual PM2.5 guideline: 5 ug/m3.
• Administrative Boundaries: geoBoundaries ADM2, William & Mary geoLab.