Delhi’s air-pollution crisis is not only a story of emissions entering the atmosphere. A new study shows that once pollution becomes severe, the particles themselves can alter sunlight, surface temperatures and wind movement in ways that make the atmosphere less capable of dispersing further pollution. The result is a feedback loop in which polluted air becomes harder to clear, particularly during the capital’s post-monsoon and winter episodes.
The study, titled “Improving aerosol-radiation interactions in the operational forecasting system – AIRWISE”, was published in Atmospheric Chemistry and Physics on September 21. Researchers from the Indian Institute of Tropical Meteorology, Savitribai Phule Pune University, the National Centre for Medium Range Weather Forecasting, the University of California, Los Angeles, and the National Center for Atmospheric Research examined Delhi’s pollution and weather conditions between September 2023 and January 2024.
Its central finding is significant for how Delhi’s pollution is understood and forecast. Aerosols — tiny particles suspended in the air — absorb and scatter sunlight. When less sunlight reaches the ground, the surface heats less during the day. That reduces the natural movement of air which normally helps mix pollutants through the atmosphere and disperse them. Near-surface winds also weaken, while the layer of air available for mixing becomes thinner.
This does not mean that atmospheric conditions create pollution independently of emissions. Rather, the research shows how high aerosol concentrations can modify local meteorological conditions and make the consequences of emissions more persistent. Pollution sources and weather therefore interact: emissions load the atmosphere, and the resulting aerosol concentration can reduce the atmosphere’s ability to dilute and move those pollutants away.
The effect was particularly visible during the November 1 to 10, 2023 pollution episode. Heavy pollution reduced the sunlight reaching the ground by up to about 80 watts per square metre, according to the study. The reduction slightly cooled the surface, weakened winds and lowered the air movement that helps disperse pollutants. The atmosphere’s ability to disperse pollution fell by up to 20 per cent in November.
The study found that the air layer in which pollution normally mixes became about 100 to 200 metres thinner during the afternoon. Winds weakened by as much as 0.4 metres per second. Under the study’s modelling, near-ground PM2.5 increased by up to 37 to 40 micrograms per cubic metre as a result of these changes.
Those effects occurred against already severe pollution levels. PM2.5 concentrations reached about 500 micrograms per cubic metre in November, while daytime levels were generally between 200 and 300 micrograms per cubic metre. The figures illustrate why the relationship between emissions and weather matters for Delhi: when the atmosphere is heavily loaded, even a reduction in dispersal capacity can leave considerably more pollution concentrated close to where people live and travel.
The January episode showed a somewhat different pattern. PM2.5 levels peaked at about 550 micrograms per cubic metre amid low temperatures, high humidity, weak winds, fog and low clouds. These conditions already favour the accumulation of pollution near the ground. The additional impact of reduced sunlight was smaller than in November because there was less sunlight available to be blocked in the first place.
Even so, the model showed that winds weakened by about 0.2 metres per second in January and that the air layer supporting pollutant dispersal became 40 to 50 metres thinner. Near-ground PM2.5 increased by about 25 micrograms per cubic metre in the model. The study said the atmosphere’s dispersal ability fell by 10 to 12 per cent during the January episode.
The distinction between November and January is important. The same broad pollution crisis can operate through different combinations of atmospheric conditions. In November, the reduction in sunlight and surface heating played a more prominent role. In January, low temperatures, humidity, fog and low clouds had already created conditions that limited dispersal, leaving less sunlight for the aerosol feedback to suppress further.
This makes air-quality forecasting an institutional challenge, not simply a matter of measuring emissions or recording pollution after it has occurred. The AIRWISE system, developed through collaboration between the Indian Institute of Tropical Meteorology, the India Meteorological Department and the National Center for Atmospheric Research, has been operational since 2018 and provides real-time air-quality forecasts across India.
The researchers said that despite its operational success, AIRWISE has limitations in reproducing extreme pollution-loading events over northern India, particularly during the crop-residue-burning season in the states surrounding Delhi. Such high-impact episodes are often underestimated partly because of inaccuracies in representing the interactions between aerosols, radiation and meteorology.
That limitation exposes a gap between the way pollution is often discussed publicly and the way it must be modelled operationally. Emissions inventories remain essential, but they are not sufficient for forecasting the concentration that residents will experience at ground level. Forecast systems also need to represent how pollutants change the physical conditions that govern their own movement.
The study’s findings therefore have two connected implications. First, better forecasts require more accurate representation of aerosol-radiation-meteorology interactions. If a forecasting system underestimates the reduction in sunlight, wind strength or mixing height during extreme episodes, it can also underestimate the persistence and intensity of near-ground PM2.5.
Second, improved forecasting cannot substitute for reducing emissions. Manoj Kumar, an analyst at the Centre for Research on Energy and Clean Air, said the findings reinforce the need to reduce air pollution at its sources rather than rely mainly on temporary measures after pollution has accumulated. He identified industry, power plants and other combustion sources among the major areas requiring consistent, targeted and time-bound emission reductions.
This is the larger urban governance question raised by the research. Delhi’s pollution response often becomes most visible when atmospheric conditions have already deteriorated: warnings are issued, restrictions are considered and short-term measures are activated. But the study indicates that once pollution reaches very high levels, the atmosphere may become less cooperative, reducing the effectiveness of natural dispersal and making recovery more difficult.
The evidence does not establish that weather feedback is the primary cause of Delhi’s pollution episodes. It shows that the interaction can amplify and prolong pollution after emissions have accumulated. That distinction matters because it places responsibility across both source control and forecasting institutions. A forecast can identify worsening conditions, but it cannot remove the emissions that create the aerosol load.
The research also adds a physical explanation to the public experience of winter pollution. When PM2.5 remains concentrated near the ground, people encounter polluted air not merely as a regional background condition but as a near-surface urban environment shaped by weak winds and a shallow mixing layer. The study’s modelled increases in near-ground PM2.5 show how changes in atmospheric structure can directly affect the concentration experienced at street level.
What the study confirms is that Delhi’s severe pollution episodes are produced by a tightly connected system of emissions, atmospheric chemistry, sunlight and local weather. What remains unresolved is how accurately operational forecasts can capture these interactions during the most extreme events. The next institutional task identified by the researchers is to improve that representation, while the public-health task remains more fundamental: reduce emissions before pollution loads the atmosphere enough to make its own dispersal harder.
