This is the first of a two-part series on how extreme heat governance in India discounts informal workers’ experiences.
THIS ESSAY opens a series on how extreme heat is governed in India and who that governance reaches. It comes out of an independent collaboration between economists, lawyers, journalists, activists and researchers mapping heat from the ground up, tracing how far the State's promised relief travels before it breaks down in practice. The series makes the case that extreme heat should be notified and resourced as a disaster in its own right, and it documents, through sustained fieldwork, what heat means for the informal workers who cannot stop working through it.
This essay presents the first findings from an ongoing audit of the Delhi Heat Action Plan 2026 and of the order issued under it on April 29, 2026, by which the Government of the National Capital Territory of Delhi directed all thirteen District Magistrates to deploy mobile heat relief units and procure consumables for the summer. Delhi released its first draft Heat Wave Action Plan in 2023, notified it in April 2024, and updated it in 2025. The 2026 plan is the third iteration.The audit covers seven of Delhi's revenue districts through field visits, RTI applications and interviews with workers at markets, labour chowks, construction sites and transport hubs.
Between May and July 2026 we audited seven cooling centres and interviewed seventy informal workers at seventeen sites across those seven districts. Fifteen of the interviews were open-ended and fifty-five were conducted on a structured instrument, so the ranked preferences reported later are drawn from the fifty-five.
The sites were chosen because they are where the policy is supposed to function. These included markets and commercial clusters, metro and bus stands, jhuggi clusters, construction sites, and the waiting points where workers stand between jobs. In parallel, the collaboration filed eleven applications under the Right to Information Act, 2005 between June 6 and 8 with four public authorities: the Revenue Department and District Disaster Management Authority, the Labour Department of the Government of the National Capital Territory of Delhi, the Delhi Building and Other Construction Workers Welfare Board, and the Municipal Corporation of Delhi.
Replies continued to arrive until July 20. Thirty three of the seventy interviews were conducted in North district, at Model Town and Majnu ka Tila, which we treat as the audit's deep case; the other six districts are confirmatory spread.
The audit asks what kind of vulnerable body the State knows how to recognise, and which workers remain invisible even within a policy written in their name. The seventy interviews and the RTI record are set against each other to answer it.
The audit covers seven of Delhi's revenue districts through field visits, RTI applications and interviews with workers at markets, labour chowks, construction sites and transport hubs.
The Biopolitics of Counting
India's heat record has been undercounted since it began, and the undercount is institutional rather than accidental. The fatalities are drawn from the same bodies, carried across generations, the same class of workers dying in the same afternoons, decade after decade. The record treats these deaths as a fact of the weather, and that drift into the ordinary is the normalisation of climate violence on working-class bodies.
Without a reliable count, nobody outside the State can establish how severe the heat has been or whom it has killed. According to the National Disaster Management Authority’s (‘NDMA’) Beating the Heat report, casualties were concentrated among the poor and vulnerable in the unorganised sector, workers such as daily wage labourers and street hawkers who have little choice but to work outdoors to earn a living.
The India Meteorological Department (‘IMD’) has logged the country's heat wave deaths in its annual Disastrous Weather Events reports since 1978. Four studies have gone back to those numbers: De and Mukhopadhyay on the 1998 heat wave, Chaudhury, Gore and Sinha Ray on the impact of heat waves through the 1990s, Malik, Bhardwaj and Singh on fatalities from 1978 to 2014, and de Bont and colleagues on how heatwaves lift all cause mortality across multiple Indian cities.
In 1998, meteorologists U.S. De and R.K. Mukhopadhyay recorded a severe heat wave across the subcontinent that took close to thirteen hundred lives, six hundred and fifty of them in Odisha alone, in a season when Delhi reached 46.5 degrees on May 28, its highest in fifty four years. Two years later, meteorologists S.K. Chaudhury, J.M. Gore and K.C. Sinha Ray put the average annual loss of life to heat waves at 153, concentrated in Rajasthan, Bihar and Uttar Pradesh. The study by Malik, Bhardwaj and Singh, spanning across 1978 to 2014, counted 660 heat wave events and 12,273 fatalities, more than four fifths of them in five states, with Andhra Pradesh alone accounting for 42 per cent. Their table returns 1,671 deaths for 1998, against the 1,300 De and Mukhopadhyay recorded in the same year. Both are drawn from the Disastrous Weather Events series. The difference is that one counts a season and the other counts a year, and the record is unstable enough that a single heat wave can be worth 370 deaths depending on where the window closes.
The study by Malik, Bhardwaj and Singh is also the only one of the four studies to disaggregate the dead. Fatalities among women and children are negligible, together under half a per cent of the total, and the authors attribute this to men's higher participation in outdoor work, adding that those in outdoor occupations who are socially isolated and economically deprived are the most susceptible of all. Carrying the series into the current decade, de Bont and colleagues find that two consecutive days above the 97th percentile raise daily mortality by 14.7 per cent, a figure drawn from 3.5 million deaths of every cause and therefore from bodies the certified count never reaches.
48 years, 4 teams, 4 methods.
Each of these counts inherits the IMD's definition of a ‘heat wave’, which is dry-bulb temperature, the heat a thermometer reads in shade.That metric captures the labouring body that collapses in the sun. It does not capture the humid heat or the sleepless hot night, which kill indoors and kill later.
Dileep Mavalankar, director of the Indian Institute of Public Health, has described heat wave deaths as an iceberg of which 90 per cent is not visible. The indirect deaths are the part that stays invisible. The IMD and the National Crime Records Bureau (‘NCRB’) record only heat-stroke deaths during periods of extreme heat, which is to say deaths from exertion under direct sunlight, while people who die of high ambient temperature, mostly those with comorbidities, fall outside the definition altogether.
The undercounting runs through every level of the record. It shows up in the state government's tally, in the IMD's classification system, and in the district administration's certification practice. Ahmedabad's 2010 heat wave is estimated to have killed over 800 people in that city alone; the NCRB reported 58 for the whole of Gujarat that year. The IMD's heatwave death count for 2021 was zero, and that is not because nobody died of heat. Heatwaves were not a category of ‘Extreme Weather Event’ for the IMD until its ‘Statement on Climate’ in 2022, when it reported 30 deaths nationally. The NCRB recorded 730 for the same year, and 374 for 2021, the year the IMD counted none. In 2023, 68 deaths in Ballia were linked by doctors to the heat. The administration certified two, and the district magistrate said it was difficult to say if the deaths were heat-related.
In 2024, HeatWatch and Veditum India Foundation counted 733 heatstroke deaths across seventeen states between March and June, drawn from news reporting. For the same seventeen states, over a longer window running from March 1 to July 25, the Ministry of Health and Family Welfare told the Lok Sabha the figure was 360. The longer series diverges the same way.
For the two decades to 2020, the NCRB logged 20,615 heatstroke deaths in its annual Accidental Deaths and Suicides reports, the NDMA 17,767 from state-reported disaster mortality, and the IMD 10,545 in its ‘Climate Hazards and Vulnerability Atlas’. The highest of the three is nearly double the lowest, for the same deaths over the same twenty years, and the differences come from the methods and the reporting chains rather than from the dying. Taken together they suggest annual heat mortality of around 1500, against roughly 25,000 all-cause deaths a day in India.
The tree-shaded enclaves of Lutyens' Delhi, Civil Lines and Delhi Cantonment stay below the stress threshold entirely. 92 per cent of the city's construction sites and three-quarters of its informal settlements sit inside that heat.
State level figures diverge further still. For 2015, the Press Information Bureau (‘PIB’), the Ministry of Health and Family Welfare and the NCRB logged Andhra Pradesh's dead at 1,369, 1,422 and 654, and Telangana's at 541, 541 and 182 respectively. The Ministry of Health and Family Welfare recorded zero heat deaths in Delhi for every year from 2015 to 2022, though the PIB had reported five for 2015. The run has not broken.
In the surveillance figures the Ministry of Home Affairs placed before the Lok Sabha on August 4, 2026, Delhi returns 32 heatstroke cases and zero deaths for March 1 to July 26 this year. Over the same period the NDMA's reported toll fell from 2,040 deaths in 2015 to four in 2020, even as the number of heat-affected states rose from nine to twenty-three. Sixteen of the twenty-four states and union territories listed did not report at all in 2020.
The apparatus for counting now exists. National Heat-Related Illness and Death Surveillance runs across more than 51,000 reporting units, every primary health centre and above. Between March 1 and July 26, 2026, it returned 4,853 heatstroke cases and twenty confirmed heatstroke deaths for the whole country. Twenty, across thirty-six states and union territories, in a summer. The system is not failing to count. It counts heat stroke, certified as such, which is the tenth of the iceberg Mavalankar described, and it now does so at national scale with a precision that makes the number look authoritative.
48 years of data have produced better recording practice and no reduction in the deaths of the people who work outdoors.
The Geography of Heat
This June, as the southwest monsoon stalled near the Himalayan foothills and left the northern plains hot and dry deep into the month, Delhi crossed 42 degrees repeatedly . On the national heat maps the country turns a single shade of red. What that shade flattens is a city where the worst of the heat sits exactly where the poor live and work.
Geospatial analysis of a decade of satellite data shows that close to 76 per cent of Delhi was persistently heat-stressed, its concrete core cooling almost 4 degrees less at night than its edges. Land surface temperatures reach 60.77°C in parts of the city, and the industrial belts at Bawana, Mayapuri, Mundka and Anand Parbat, along with the walled city, are among the persistent heat spots. The tree-shaded enclaves of Lutyens' Delhi, Civil Lines and Delhi Cantonment stay below the stress threshold entirely. 92 per cent of the city's construction sites and three-quarters of its informal settlements sit inside that heat. Construction workers, gig workers, street vendors, the homeless and people living in informal settlements are estimated to form at least half of Delhi's population, and they do not have adequate means to adapt to it.
Mapping heat in Delhi, then, is also about mapping class. The official reading, taken by the IMD at Safdarjung as its "representative station" for Delhi, does not produce an authentic account of how the city’s population truly experiences heat.
Heat Plans Without Protection
The plans built to answer this are more than a decade old and remarkably uniform. India's template dates to 2013, when Ahmedabad, after a heat wave linked to more than 1300 excess deaths in a single month, adopted South Asia's first Heat Action Plan, later credited with averting on the order of a thousand deaths a year.
National guidelines followed in 2016 and were revised in 2017 and 2019, and by the end of the decade twenty-three states and more than a hundred cities had plans of their own. Their content barely varies across early warning tied to temperature thresholds, community outreach, and a standard set of provisions, oral rehydration solution, cooling shelters and shaded rest, drinking water and the rescheduling of outdoor work.
But the investment in heat responsiveness has been largely confined to advisory measures. Through the summer this audit covers, heat waves sat outside the list of notified natural calamities, so no state facing one could draw on the National Disaster Response Fund to meet it, and Heat Action Plans were written to be delivered on whatever a district could find.
Greenpeace India and the National Hawkers Federation have demanded notification since 2024 on the ground that the plans remain advisory and carry neither funding nor legal accountability. That demand has now been partly met. On August 4, 2026, the Ministry of Home Affairs told the Lok Sabha that, following the Sixteenth Finance Commission's recommendation for the award period 2026 to 2031, heatwaves and lightning have been included in the list of notified natural calamities under the operational guidelines for the State and National Disaster Response Funds. It also noted that heat wave mitigation projects are now eligible under the Disaster Mitigation Funds.
Heatwaves now qualify for disaster response funding under the operational guidelines. They have not been notified under the Disaster Management Act, 2005, which is the statute that would create an enforceable claim. A worker still has nothing to claim. The Delhi Heat Action Plan 2026 was issued on April 29, and the summer documented here was run before any of it.
Beneath the funding question sits a failure of design. Heat Action Plans treat vulnerability as a uniform and technical problem rather than as something produced by caste, income and the conditions of work, and they are drafted by consultants working from historical temperature data instead of from the lives of the people they cover.
Delhi's Heat Action Plan 2026 is where this series puts that apparatus to the test. The Action Plan names its object of protection precisely, outdoor workers and vulnerable persons. The IMD's seasonal forecast this year listed outdoor workers among those most at risk and instructed administrations to ready cooling shelters and drinking water. Those are the provisions this audit finds were promised and not delivered. Nor is the category shrinking. Outdoor work is the fastest-growing form of employment in the city, in construction, in street vending, and now in platform-dispatched gig work that keeps riders on the road through the worst of the afternoon, and the same urban expansion that manufactures the heat has produced more of these workers than at any point before.
Delhi's concrete core releases at night what it absorbs by day, so a worker who cannot stop through the afternoon cannot recover through the night.
From Policy to Administration
Delhi's Heat Action Plan 2026 is an advisory. It lists what a district may provide and creates no entitlement any worker could claim. But the constitutional basis for such an entitlement already exists, and it was built through cases brought by workers and the poor.
In Francis Coralie Mullin v Administrator, Union Territory of Delhi (1981) the Supreme Court held that the right to life under Article 21 includes the right to live with human dignity. Life means more than survival.
In Olga Tellis v Bombay Municipal Corporation (1985), brought by Bombay's pavement dwellers, the Court held that Article 21 also carries positive entitlements, meaning things the state must provide rather than merely refrain from taking away, and that livelihood is one of them. That opens the way to Article 39(e) of the Constitution, which directs the state to ensure that the health and strength of workers is not abused, and that no citizen is forced by economic necessity into work unsuited to their strength. It is a directive principle rather than an enforceable right, but the Court has long read Article 21 in its light.
In In Re: Ramlila Maidan Incident (2012), after police cleared a sleeping crowd by force, the Court held sleep to be a fundamental right under Article 21, as basic as breathing, eating and drinking.
Most recently, in Kaushal Kishor v State of Uttar Pradesh (2023), the Court held that the state must protect the right to life not only from its own actions but from private parties too. Applied here, that means employers. A plan that asks platform companies and contractors to protect workers, and provides nothing to make them, does not discharge that duty.
Delhi's concrete core releases at night what it absorbs by day, so a worker who cannot stop through the afternoon cannot recover through the night. The NDMA's own account describes the dead as daily wage labourers and street hawkers compelled to work outdoors out of economic necessity, which is Article 39(e)'s language handed back to the state in its own record.
This is the case for thermal dignity. This is the position that there are conditions under which work cannot be compelled, and that securing this is the State's obligation whether the compulsion comes from a contractor, a platform, or the need to earn the day's wage.
No instrument in force in Delhi this summer did that. The workers compelled to keep working through a disaster are named in every instrument that governs it, and reached by none of them. While the NDMA guidelines describe them as the casualties, the Delhi Heat Action Plan 2026 names them as the object of protection, and the order of April 29 instructs thirteen District Magistrates to procure for them. Not one of them gives a worker the right to stop, or requires an employer to allow the stoppage of work. For a worker living from one day's wage to the next, with a family to feed, an hour of shade costs an hour of earnings, and a day that pauses is a day of eating less or not at all. The apparatus built to protect them is designed so that the day remains intact, and the cost of the heat is borne by the body rather than by the employer or the state.
The second part of this series follows these documents into the districts they were written for, and asks what the administration of heat to labour actually delivered to the people it was written about.