The Heat Map Bangladesh Needs

Executive Summary. Bangladesh no longer lacks evidence that heat is dangerous. The World Bank finds that maximum temperature rose 1.1 degrees Celsius between 1980 and 2023, while the heat index rose 4.5 degrees. Dhaka warmed faster, with a 1.4 degree increase in maximum temperature. A national panel of more than 16,000 people found higher summer prevalence of persistent cough, diarrhea, depression, and anxiety. The report estimates heat-attributable economic losses of US$1.33 billion to US$1.78 billion in 2024, equal to an estimated 0.29% to 0.39% of GDP. Yet a national average cannot tell a city corporation which ward needs shade, a hospital where demand will arrive, or an employer when work must stop. Bangladesh needs a public heat-risk map that joins satellite surface temperature, humidity, population, health, work, and service access. The map should be an operating system for action, not a poster produced after a heatwave.

The national signal is already strong

The first task is to stop treating heat as an anecdote. The World Bank report uses historical weather analysis and a national household panel to establish three distinct facts. Bangladesh is warming. Humidity makes the exposure experienced by the body rise faster than temperature alone. Measurable health and productivity losses appear when hot conditions intensify.

The distinction between temperature and heat index matters. Maximum temperature is a physical reading. Heat index combines temperature with humidity to approximate how hot conditions feel to the body. A humid delta can therefore acquire a much larger physiological burden than the thermometer suggests. Between the report's endpoints, maximum temperature increased by 1.1 degrees, while heat index increased by 4.5 degrees.

Source: World Bank, An Unsustainable Life, historical analysis for 1980 to 2023.

That gap changes what a map must show. A map built only from air temperature can miss a central part of exposure. A map built only from land-surface temperature can show where roofs, roads, and bare soil radiate heat, but it cannot be read as the temperature of the air or as a clinical danger threshold. NASA's MOD11A2 product is valuable because it offers repeated land-surface temperature observations at fine spatial resolution. Its value lies in locating hot surfaces and persistent spatial patterns. It should sit beside humidity and station observations, not replace them.

This is the first design principle: preserve the meaning of every layer. Surface temperature, air temperature, heat index, and health risk are related, but they are not interchangeable. A useful public system makes those differences visible.

Dhaka proves that place changes the burden

National averages compress the very variation that local action needs. The World Bank reports that Dhaka's maximum temperature rose 1.4 degrees between 1980 and 2023, compared with 1.1 degrees nationally. Dense construction, hard surfaces, lost vegetation, traffic, and waste heat can intensify exposure around the people least able to escape it.

Source: World Bank, An Unsustainable Life, Dhaka and national historical analysis for 1980 to 2023.

The important word is exposure. A hot district is not necessarily the district with the highest health risk. Risk depends on who lives in the hot pixels, who works outdoors, whether housing sheds heat at night, whether water and power remain available, and how quickly care can be reached. A wealthy commercial area with air conditioning and reliable electricity may be hotter at the surface but safer for residents than a slightly cooler settlement with metal roofs, crowded rooms, intermittent power, and no shaded public space.

The unit of policy should therefore be smaller than the unit of weather reporting. City corporations need ward-level products. Rural agencies need upazila and crop-calendar views. Health managers need facility catchments. Labour inspectors need industrial clusters and outdoor work sites. The same heat event travels through different systems, so the map must be able to answer different operational questions from one common evidence base.

A national heat map also needs time. A warm-season average can reveal structural exposure, but it cannot trigger a response. Daily forecasts can trigger warnings, but they cannot show where tree cover, roofs, drainage, or clinic capacity should be changed. Bangladesh needs both: a long-run exposure atlas for investment and a short-run alert layer for operations.

Health evidence tells us what the map must connect

The household panel gives the map a purpose beyond meteorology. The survey covered more than 16,000 people in 2024. Persistent cough affected 3.3% of surveyed people in winter and 6.0% in summer. Diarrhea rose from 1.8% to 4.4%. Depression rose from 16.2% to 20.0%, while anxiety rose from 8.3% to 10.0%.

Source: World Bank, An Unsustainable Life, national panel survey rounds in January and May 2024.

These are seasonal associations from the report's survey, not proof that heat caused every case. Summer differs from winter in other ways. Water quality, air pollution, infection patterns, work intensity, and household behaviour can move with the season. The report's more detailed analysis addresses exposure ranges and covariates, but a public dashboard should retain the modest claim: heat is associated with a burden large enough to organize services around.

That implies a second design principle: the heat map must connect hazards to outcomes. Health facilities should be able to record syndromes consistently during hot periods. Ambulance and outpatient demand should be geocoded at a privacy-preserving level. Schools and employers should record heat-related closure or lost-time decisions. Those operational records can show whether alerts arrive before demand and whether interventions change outcomes.

The economic estimate reinforces the point. The World Bank places heat-attributable losses at an estimated US$1.33 billion to US$1.78 billion in 2024, or an estimated 0.29% to 0.39% of GDP. The range is not a precise invoice. It is a model-based estimate built from survey and productivity assumptions. Its policy use is to establish scale, not to claim that every lost dollar can be located. A risk map should narrow that uncertainty over time by collecting consistent exposure and outcome data.

The strongest counterargument is institutional, not scientific

The strongest objection is that Bangladesh already has weather agencies, satellite products, hospitals, disaster-management structures, and local governments. Another dashboard could become one more disconnected website. If no agency owns warnings, work rules, health preparation, or neighbourhood cooling, a more detailed map may only describe harm with greater elegance.

That objection is correct about the failure mode. Data products often begin with the available technology rather than the decision that needs support. A national portal can publish colourful layers while a factory manager receives no enforceable rule, a school head receives no local threshold, and a clinic sees no forecast of likely demand. Precision without authority can create the appearance of preparedness.

But the answer is not to remain spatially blind. It is to design the map around decisions and named users. Every layer should have an action attached. A forecast threshold should activate a health message, a work-rest protocol, a school instruction, or a local cooling measure. A structural hot spot should enter capital planning for trees, shade, roofs, water points, and clinics. If a layer has no user and no decision, it should not be on the public operational screen.

The map should also show uncertainty. Satellite gaps, cloud bias, station distance, model error, and sparse health records are not reasons to hide the product. They are reasons to publish confidence, data vintage, and missingness. A blank area honestly marked as unobserved is more useful than a smoothly interpolated fiction.

What would change this conclusion

A checkable test would change the argument: if existing national systems can already publish fine-scale exposure, vulnerability, action thresholds, data vintage, and response ownership in one operational view, then Bangladesh does not need a new heat map. It needs those systems opened, connected, and used. The burden of proof is a live product and documented decisions, not a project announcement.

First, build a common heat-risk data contract. Define the allowed meaning, unit, spatial resolution, update frequency, and uncertainty field for surface temperature, air temperature, humidity, heat index, population exposure, and health outcomes. Owner: Bangladesh Meteorological Department with the health directorate, statistics office, and local-government division. Success signal: a public machine-readable feed in which every layer carries its source, retrieval time, unit, and uncertainty.

Second, tie thresholds to operating rules. Publish location-specific protocols for outdoor work, schools, hospitals, water points, and public communication. The thresholds should be tested against observed outcomes and revised openly. Owner: the relevant labour, education, health, and disaster-management authorities. Success signal: every alert issued through the map records which protocol activated, when it activated, and when it closed.

Third, fund cooling where exposure and vulnerability overlap. Use the structural atlas to prioritize shade, reflective roofs, trees, drinking-water access, clinic readiness, and night-time cooling in the places with the weakest capacity to adapt. Owner: city corporations and local government, with central capital financing. Success signal: the next hot season shows lower heat-related service demand in treated areas than in comparable untreated areas, with the comparison method published in advance.

The map Bangladesh needs is not a claim that every hot pixel is dangerous. It is a disciplined bridge from observation to decision. Its success will be measured in actions taken before illness, not in layers added after it.

That standard keeps the technology subordinate to public protection and gives citizens a direct way to test whether better data produced a better response.

Sources

1. World Bank. An Unsustainable Life: The Impact of Heat on Health and the Economy of Bangladesh. Retrieved 2026-08-23: https://documents1.worldbank.org/curated/en/099111024033034518/pdf/P168901-70ed6ce2-aaea-42b2-afab-8484b7b75125.pdf
2. NASA Land, Atmosphere Near real-time Capability for EOS. MOD11A2 Land Surface Temperature and Emissivity. Retrieved 2026-08-23: https://ladsweb.modaps.eosdis.nasa.gov/missions-and-measurements/products/MOD11A2

Method note: the World Bank report contains a 65% comparison whose wording mixes Dhaka heat-index change with national temperature change. That comparison is excluded here. The chart uses only the separately reported maximum-temperature changes for Dhaka and Bangladesh.