Any serious attempt to cut Delhi’s winter pollution starts with knowing where the particulate matter in its air comes from, and a decade of studies has produced only partial answers. The emissions estimate behind IITM’s source forecasts still rests on surveys conducted in 2016, and no study has yet traced the largest share of the particles residents breathe, those that form in the atmosphere from gases, to the emitters responsible.
What is being measured?
The pollutant at issue is PM2.5, particles no wider than 2.5 thousandths of a millimetre, or roughly a thirtieth of the width of a human hair. Some are released directly as soot from exhausts, smoke from burning crop residue and wood, and dust from roads and building sites (dust in itself is larger PM10 but a component of it is also PM2.5). Scientists call these primary particles.
Then there are secondary particles, which form as gases react in the air. Sulphur dioxide, nitrogen oxides and ammonia become sulphates, nitrates and ammonium salts, collectively the secondary inorganic aerosols (aerosol being the scientific term for fine particles suspended in air), while volatile organic compounds, or VOCs, carbon-based gases given off by fuels, solvents and burning, oxidise into secondary organic aerosols.
“Unlike primary, which are emitted directly into the air, secondary aerosols form when gases undergo chemical reactions in the atmosphere and get converted into particles,” said Sachin Ghude, the Indian Institute of Tropical Meteorology (IITM) scientist who oversees its new pollution laboratory in Delhi, told HT on Saturday.
Calling a particle “secondary” only describes how it formed. The gases it formed from still come from identifiable sources: nitrogen oxides largely from power plants and vehicles, and ammonia mostly from farms. Measurements from 2013 to 2016 put nitrates, ammonium and chlorides alone at 24% to 30% of Delhi’s PM2.5, and studies cited by IITM put secondary organic aerosols at around 28% in peak winter.
Much of this pollution also originates beyond the city. Scientists describe the area over which emissions mix and travel as an airshed, and Delhi’s extends well past its borders. The city’s own sources produce roughly a third of its PM2.5 in the post-monsoon and winter months, with neighbouring NCR districts and more distant regions supplying the rest, according to estimates from IITM’s forecasting model. “A majority of the PM2.5 in Delhi comes from its immediate neighbors,” IITM’s scientists wrote in 2024. A 2022-23 study for the Delhi government estimated that 83% to 84% of the city’s secondary inorganic share came from outside its borders.
Two ways of counting
Source apportionment, as the exercise is known, works in one of two directions. Receptor studies work backwards from the air: they collect particles at a site, analyse their chemistry and match the result to the chemical fingerprints of known sources, such as potassium, a marker of burning biomass.
Models work forwards, and they rest on two parts. The first is an emissions inventory, a ledger of how many tonnes of particles and gases each source releases, from vehicles and power plants to households, construction and burning, compiled from surveys of activity such as fuel use, traffic and industry. The second is a simulation, a computer recreation of the region’s weather and air chemistry that releases those emissions, carries them on the wind, turns gases into particles and keeps track of which source each particle started from.
Each method has a weakness. A study that samples the air picks up every particle in it, including those formed from gases. But the chemistry of a sulphate or nitrate particle does not reveal whether the gas it formed from came from a power plant, a truck or a farm, so such studies can measure the secondary share without being able to divide it among the sources responsible. That is the issue with the AIRWISE realtime sampling data.
A model has the opposite problem. Because it follows each source’s emissions from the start, it can say which source produced every particle it simulates. But its answers can only be as accurate as its inventory: if the inventory overstates road dust or understates traffic, the model’s source split will be wrong in the same way. And it can only simulate the chemical reactions it has been built to handle, so any particle that forms through a reaction it did not originally account for never appears in the results. And this is what underpins the second problem with Delhi’s pollution forecasting — the emissions industry is far outdated.
Also read: Delhi env minister meets with NCR police officers over pollution-control measures
The 2016 foundation
The modern record begins with a receptor study. IIT Kanpur, working for the Delhi government, trapped particles on filters at six sites in 2013-14, analysed their chemistry and matched it against the fingerprints of known sources, submitting its findings in January 2016. They shifted sharply with the seasons. In winter, secondary particles were the largest component, at 30% of PM2.5, followed by biomass burning at 26% and vehicles at 25%. In summer, soil and road dust (28%) and coal and fly ash (26%) led. Mean PM2.5 at the sites was 377 micrograms per cubic metre in winter, more than six times the national 24-hour standard of 60, and 294 in summer. The study also compiled an emissions inventory for the city, and it ranked the sources differently from the air samples, since pollutants disperse, settle and transform before they reach a monitor: of 59 tonnes of PM2.5 emitted in Delhi each day, road dust accounted for 38% and vehicles 20%, while industry, largely power plants, produced 52% of nitrogen oxides and vehicles 36%.
The surveys that matter most for today’s estimates came the same year, when The Energy and Resources Institute (TERI) gathered the raw material for a much larger inventory, covering Delhi and 19 NCR districts, sector by sector: transport, households, industry, waste burning, construction, road dust, energy and a residual category covering crematoria, airports, restaurants and diesel generators. A decade on, those surveys remain the basis of the inventory IITM’s forecasting model uses. TERI and the Automotive Research Association of India followed in 2018 with a source study that put Delhi’s own contribution at about 36% of its winter PM2.5.
Also read: No consistent improvement of Delhi’s winter air quality in past four years: CREA analysis
What has changed since the surveys?
The forecasting model officials consult each winter still starts from an inventory built on the 2016 surveys, so the numbers it begins with predate a decade of changes to what Delhi’s vehicles, power plants and industries emit.
Delhi moved to BS-VI fuel in April 2018, two years before the rest of India, cutting sulphur in petrol and diesel from 50 parts per million to 10. Less sulphur in the tank means less sulphur dioxide from the tailpipe, the gas that forms sulphate particles. Since April 2020, only BS-VI vehicles can be sold in India, with limits on particulate matter and nitrogen oxides 50% to 89% tighter than under BS-IV, depending on the vehicle. The limits apply to new vehicles; older ones stay on the road until they are scrapped. The 705-megawatt Badarpur coal plant, which a court-appointed authority had called one of the largest contributors to Delhi’s pollution, shut in October 2018.
Age limits that bar diesel vehicles older than 10 years and petrol vehicles older than 15 are now in force. The Eastern and Western Peripheral Expressways, opened in 2018, route through-trucks around the city, though HT found then that heavy vehicles were still avoiding the eastern road. Delhi’s bus fleet, moved to CNG under a 1998 Supreme Court order, is going electric. Electric vehicles made up 12.7% of Delhi’s new registrations in 2025-26. Even so, Delhi’s PM2.5 emissions from transport rose 37% between 2010 and 2020, one study estimates.
Each of these factors essentially conflict with the numbers (or calculations) what was recorded in the inventory: how much sulphur goes into a fuel tank, how many trucks cross the city, whether a coal plant is running. IITM’s scientists have acknowledged “significant changes” in emissions between 2016 and 2021, but without an updated inventory there is no measure of how far each source’s share has moved. One with 2026 as its baseline is due by June 2027.
Also read: No tandoor at eateries, open burning of waste: List of curbs under Grap-1 in Delhi
The forecast model
The age of those surveys matters because they feed the model officials consult each winter. IITM introduced its Decision Support System in 2021 to estimate, each day, how much of Delhi’s PM2.5 will come from eight sectors within the city, 19 NCR districts, stubble burning and everything beyond those 20 districts. It is the forwards method in practice, a weather and chemistry simulation on a 10-km grid, so each value averages conditions across a 10-km square, that follows each source’s particles through more than 400 tagged tracers. Stages of the Graded Response Action Plan (Grap), the curbs that tighten as forecast air quality worsens, are imposed or lifted on the basis of IITM’s air quality forecasts and the model’s source estimates, according to a 2024 paper by IITM scientists.
For 2021-22, the model found transport to be the largest single source within Delhi, at 17% of PM2.5 after the monsoon and 14% in winter, and attributed about a quarter of the city’s pollution after the monsoon, and about a third in winter, to the districts on its border. During peak episodes, the dominant source, farm fires in early November or local emissions in winter, reached 65% to 69%. A 20% cut in emissions from human activity across the NCR would reduce Delhi’s PM2.5 by 12%, the model estimated.
IITM’s own scientists have spelled out the limits of those figures. The model’s emissions come from an inventory “compiled using surveys done in 2016”, and IITM’s website cautions that “the latest emission fields are required” for more accurate estimates. Its chemistry tracks only primary particles, so it does not simulate nitrate or secondary organic aerosols, the same secondary material that measurement studies have repeatedly found to be the largest share of Delhi’s winter PM2.5. IITM’s scientists have also conceded that “the severe AQI values (AQI > 400) are missed by the model,” and that on such days “the source apportionment in that situation would not be correct.” A 2024 paper promised nitrate and secondary organic aerosols “in the next version”.
The studies that were stopped
Measuring the air directly captures the secondary particles the model leaves out, but the Delhi government’s recent source studies have not lasted. In 2021, it stopped a study by Washington University before its data was made public, citing “unsatisfactory” results, and reassigned the work to IIT Kanpur.
IIT Kanpur, with IIT Delhi and TERI, then ran a real-time supersite at Rouse Avenue for the Delhi Pollution Control Committee through winter 2022-23, with online instruments that analysed the air’s chemistry as it was sampled. Over the season, secondary inorganic aerosols made up 32% of PM2.5, biomass burning 24% and vehicles 17%. The study went further than the 2016 study in locating the secondary share, attributing 83% to 84% of it to sources outside Delhi. The government halted the project in November 2023, again dissatisfied with its methods and results, and set up nothing in its place.
Also read: Delhi’s AQI remains ‘poor’; minimum temperature recorded at 19°C
The new lab
IITM opened its Advanced Atmospheric Chemistry and Aerosol Toxicity Analysis Laboratory (Aatal) in New Rajendra Nagar on October 24, with 25 instruments costing about ₹80 crore, to establish what is in Delhi’s air, where it originates and how it transforms. In its first daily estimates, particles formed in the air made up 29% to 42% of the lab’s PM2.5 between October 20 and 24, ahead of vehicles at 19% to 28%, and IITM cannot yet trace them to their source. “The original source is not known,” Ghude said. The lab’s PM2.5 total was also less than half the middle reading among Delhi’s monitoring stations on October 22, a discrepancy officials attributed to teething issues, HT reported. IITM expects to split the vehicle share by fuel within weeks.
Where we stand
No study has yet produced a reliable account of which emitters release the gases behind the largest share of Delhi’s winter particulate matter, or in what proportion, and the inventory behind IITM’s model will not reflect a decade of cleaner fuel, newer vehicles and a closed coal plant before June 2027.
