Mapping Heat Inequality Across Neighbourhoods in Delhi

Executive Summary

Extreme heat has become a defining urban challenge in India, with Delhi illustrating how rising temperatures interact with dense built environments and unequal access to cooling. While Heat Action Plans (HAPs) set out a broad response framework, they do not yet incorporate granular evidence on how people’s ability to cope differs across neighbourhoods. This study addresses that gap by integrating high-resolution climate and remote-sensing data with a large sample of household surveys to understand the spatial, social, and behavioural dimensions of heat vulnerability in Delhi.

The distinctive contribution of this study is the spatial layering of citizen experience over micro-climate and built-form data. By combining how hot people feel with objective measures of the built environment and atmospheric conditions, the analysis identifies not only where heat is most intense but also the greatest constraints in coping. The study shows that heat exposure, coping behaviour, and health outcomes cannot be explained through temperature averages alone; they emerge from the interaction between urban form, socio-economic conditions, and household adaptation choices.

Our methodology integrates three high-resolution geospatial layers with household survey data to construct a detailed picture of micro-climatic variation and lived heat exposure across Delhi. Built-up area was mapped using the Global Built-Up Surface dataset (by the Global Human Settlement Layer – GHSL) at 100-metre resolution, where each pixel captures the square metres of constructed surface within that cell. This allowed us to identify neighbourhoods with dense, concretised forms and limited ventilation, which tend to retain heat for longer periods. Vegetation was mapped using MODIS Vegetation Continuous Fields at 250-metre resolution, with each pixel indicating the percentage of tree canopy, non-tree vegetation, and non-vegetated ground cover. This enabled us to locate areas where insufficient shade and low green cover intensify local heat stress. To capture experienced heat, we combined ERA5-Land data on surface temperature and dew point (originally at 10-kilometre resolution) with VIIRS land-surface temperature at 1 kilometre. ERA5 data were resampled to 1 kilometre, corrected for bias, and used to compute relative humidity and a heat index using the Rothfusz regression. Together, these layers provided a consistent spatial grid from which to derive neighbourhood-level micro-climate conditions.

The framework guiding this work views heat vulnerability as the interaction of three dimensions. Micro-climate conditions determine the baseline level of heat and humidity that residents experience. Built-environment characteristics, including density, tree cover, and other factors, shape how heat accumulates and dissipates. Socio-economic conditions, such as income, occupation, appliance ownership, and time spent outdoors, determine whether households can buffer themselves or are forced to absorb the effects of rising heat. Vulnerability accumulates across these dimensions: neighbourhoods with high built-up area and low vegetation face higher temperatures; households with limited cooling face greater exposure; and individuals with prolonged daily heat exposure face higher risks of both illness and productivity loss.

The findings show that even small differences in built-up form and vegetation correspond to meaningful differences in experienced heat. Increasing the built-up area from roughly 25% to 55% raises the experienced temperature by about 0.6°C. In contrast, increasing tree cover from around 3% to 11% lowers experienced heat by approximately 1°C. This asymmetry suggests that even modest improvements in green cover have a stronger cooling effect than the warming effect associated with similar increases in built-up area. The result reinforces the role of neighbourhood-scale greening as a practical lever for reducing heat exposure in dense urban settings.

Differences in coping capacity also emerge clearly. Sleep disruption rises by 5–6 percentage points with a 3°C increase in experienced heat, but households with air conditioners report substantially better sleep outcomes. AC-owning households spend nearly twice as much on electricity during extreme heat, while non-AC households face constraints in increasing cooling hours. Appliance-usage data indicates that wealthier households already cool their homes for 12–14 hours a day, leaving little scope for further adjustment as temperatures rise. Additionally, in terms of cooling hours, almost all respondents also report using the AC during the night, while 50% report using it in the afternoon. By contrast, lower-asset households lack both cooling appliances and the financial means to increase energy use, exposing them to persistent heat strain.

These inequalities extend into health outcomes. A 3°C rise in experienced heat corresponds with a 15 percentage point increase in respondents reporting illness for more than five days in the previous month. Illness prevalence is highest in the 42.5–47°C heat range, where nearly 30% report prolonged ill health, and over 80% of all respondents who reported illness, suggesting that even intermittent access to cooling reduces physiological burden. Analysis of chronic conditions shows clear clustering of hypertension, diabetes, obesity, thyroid conditions, and respiratory issues in the hotter heat-index bands.

Heat also has a measurable impact on productivity. The share of households missing work due to heat rises from around 18% to nearly 28% with a 3°C rise in experienced heat. AC-owning households report an 18% lower incidence of heat-related work loss compared to non-AC households. Workers with long commutes or significant time spent outdoors face the greatest disruptions. In the highest heat-index band, almost half of respondents work in the sun for more than two hours a day, and a third for more than six hours, compounding exposure at both home and work.

Impact on mental state shows a similarly clear pattern. The share of respondents reporting noticeable changes in mental state rises from about 15% to 30% as experienced heat moves from roughly 42°C to 45°C. These findings are consistent with global evidence linking elevated temperatures to increased stress and mental-state-related morbidity.

Taken together, the results indicate that extreme heat operates as a structural stressor that exacerbates existing inequalities. Households with limited physical, financial, and infrastructural buffers face higher levels of exposure, greater illness, and more frequent productivity losses. These disparities are likely to sharpen as temperatures continue to rise, making a strong case for policy interventions that combine spatial precision with citizen-centric data systems.

The study identifies three areas for policy action. The first is the development of micro-level heat action plans supported by routine, citizen-centred data. Institutionalising short, rapid surveys within state systems would create a low-cost, continuous feedback loop on lived experience. When combined with ward-level and settlement-scale mapping, this approach allows authorities to locate high-risk clusters and direct measures—such as cool roofs, shading, and targeted outreach—to the communities most affected. State Disaster Management Authorities, working with urban local bodies, are well placed to coordinate this model. For these state heat action plans to be effective, there is also a need to focus on developing the relevant data literacy within government bodies, including the ability to interpret charts, indicators, and spatial data to enable evidence-based planning and effective heat risk management.

The second priority is to embed heat resilience within urban planning. The study shows that modest increases in vegetation yield larger cooling effects than equivalent increases in built-up areas generate warming. Urban design, therefore, has a central role in moderating exposure. Neighbourhood-level greening, improved airflow, reflective materials, shaded pedestrian routes, and climate-responsive housing standards can reduce heat accumulation. Integrating blue–green infrastructure into zoning and redevelopment decisions strengthens longer-term resilience, while low-cost retrofits in informal settlements—such as reflective coatings and improved roofing—can reduce indoor heat without imposing high costs.

The third priority concerns energy systems and appliance design. Rising urban heat is rapidly reshaping electricity demand in India, with cooling loads emerging as a major source of peak grid stress during heatwaves. Integrating heatwave forecasting, hourly cooling demand, and neighbourhood-level exposure into energy planning is critical to avoid outages and unequal access to reliable power. Appliance efficiency (particularly for air conditioners) offers a cost-effective pathway to reduce peak demand, emissions, and long-term grid investment needs. Targeted financing, smarter energy labels, and context-specific appliance design can accelerate adoption among low-income households while strengthening urban resilience.

Acknowledgments

The authors thank colleagues from data.org, Priyank Hirani and Shivam Shukla, for their support in co-shaping this study through regular check-ins and sustained support throughout the course of the work.

We thank Radhika Khosla (Associate Professor at the Smith School of Enterprise and the Environment, University of Oxford) for her guidance on the study design and questionnaire, and Mayura Gadkari (Principal, Artha Global) for her inputs that helped shape the urban policy recommendations.

We would also like to thank the leadership team at Artha Global, Dr. Reuben Abraham, Pritika Hingorani, Dr. Niranjan Rajadhyaksha, Dr. Gyanendra Badgaiyan, and Karan Shah, for their support.

We also acknowledge the Quest data team for their work in undertaking the household data collection across Delhi.