Delhi's winter pollution depends on emissions, wind, and the depth of air that can mix. These factors explain why fewer farm fires do not guarantee clean air.

The YouTube player loads only when you press Play. Playing connects to YouTube, which may use cookies and storage. See Privacy.

The puzzle: fires fell, smog did not

Punjab records 5,114 farm-fire incidents during the 2025 paddy monitoring season. The government's December 1 release reports a 93 percent fall from 2021.

The monitoring period ends on November 30. Yet the video's saved Delhi air-quality series reaches its highest value on December 14, about two weeks later.

That comparison does not show that farm fires have no effect. It shows why one source cannot explain an entire pollution season. Other emissions continue after the farm-fire season ends.

Fire counts also differ from measurements of smoke mass. Weather affects where that smoke travels and how much it dilutes. A lower count and a high pollution concentration can therefore occur together.

What the AQI number means

India's Air Quality Index, or AQI, converts pollutant concentrations into a common reporting scale. The scale runs from 0 to 500 and covers eight pollutants.

PM2.5 means fine particles with diameters generally no greater than 2.5 micrometres. A micrometre is one millionth of a metre. The category describes size, rather than a single chemical substance or source.

For PM2.5, the good category covers concentrations up to 30 micrograms per cubic metre. The severe category starts above 250 micrograms per cubic metre. These boundaries concern the relevant daily concentration, rather than an instantaneous particle count.

The index uses straight-line interpolation between each pair of category boundaries. This means it assigns intermediate concentrations intermediate index values. For example, 45 micrograms per cubic metre gives a PM2.5 sub-index near 75.

The highest valid pollutant sub-index determines the overall AQI. An AQI value therefore does not average all pollutant concentrations. It also cannot identify which activity produced the pollution.

A box of air

Imagine a box above the city. Emissions add pollution to the box. Wind carries air through its sides, while vertical mixing spreads pollution through its depth.

Concentration means the amount of pollution within a given volume of air. The same emissions can produce a higher concentration when they enter a smaller volume. Weak wind also slows the replacement of polluted air.

The animation changes these controls separately. More emissions increase the input. Faster wind increases transport through the box. A deeper mixing layer gives emissions more air in which to spread.

Real air also has chemical reactions, rain removal, and pollution arriving from distant regions. The box model omits that complexity to explain dilution. It is not a sealed container or a complete air-quality forecast.

The lid: why rising air cools

Air pressure decreases with height. A rising parcel of air expands as the surrounding pressure falls. Expansion cools the parcel when little heat enters it.

An unsaturated parcel cools by about 10°C for each kilometre of ascent. Unsaturated means that the parcel does not contain enough water vapour for condensation at its current temperature.

Whether the parcel keeps rising depends on its temperature relative to the air around it. A parcel that becomes colder and denser than its surroundings resists further ascent. This comparison explains atmospheric stability.

Winter inversions

On a clear, calm winter night, the ground loses heat. The ground then cools the air immediately above it. Warmer air can remain higher above the surface.

This arrangement has the name temperature inversion. Temperature increases through that layer, instead of decreasing with height. The warm layer limits vertical mixing of the colder air below.

The video's lid represents this resistance to mixing. No solid roof exists. Wind, sunlight, clouds, and changing weather alter the layer's depth and strength.

Weak daytime heating can leave winter mixing shallower than summer mixing. Continuing emissions then enter less air. The inversion makes emissions more concentrated; it does not create the emissions.

Ten years of Delhi lid heights

The video compares saved ERA5 model data for 2016–2025 near Delhi. ERA5 combines observations with a weather model to reconstruct past atmospheric conditions.

The afternoon calculation uses the available 08:00 and 09:00 UTC values. These correspond to 13:30 and 14:30 in India. The calculation omits missing boundary-layer values.

Recalculation gives a January mean mixing depth of about 858 metres. The April mean reaches about 2,761 metres, or nearly 3 kilometres. These are averages for the selected model location and times.

The video's early-morning curve often lies below 100 metres. That illustrates the strong daily cycle, rather than a fixed height above every part of Delhi. A model grid value also differs from a direct measurement at each street.

Ventilation coefficient

The ventilation coefficient multiplies mixing height by wind speed. Its unit is square metres per second. Greater height and faster wind both favour dispersion.

The video's calculation uses wind speed at 10 metres as a proxy. A full ventilation estimate uses a representative wind through the mixing layer. The surface-wind proxy cannot measure that entire layer directly.

The cited ventilation study discusses an afternoon value below 6,000 square metres per second together with weak wind. This is a screening criterion, not a universal boundary between safe and unsafe air.

The saved-data calculation gives December a mean proxy value of 2,247 square metres per second. April reaches 8,796, about 3.9 times higher. About 98.7 percent of the selected December samples lie below the comparison threshold, against 36.7 percent in April.

Those percentages describe the proxy and selected sampling times. They are not percentages of days with a measured health emergency.

Why PM2.5 does not fall quickly

A typical human hair has a width near 70 micrometres. A particle at the PM2.5 size boundary is about 30 times smaller across. Smaller particles have little weight relative to the drag that air exerts on them.

The video's settling example uses an ideal spherical particle with a diameter of 2.5 micrometres in still air. It assumes particle densities of 1,000–1,500 kilograms per cubic metre. It also assumes an air viscosity of 0.000018 pascal-seconds, which measures resistance to flow.

The calculation gives about 0.19–0.28 millimetres per second, or roughly 0.7–1 metre per hour. These values omit the small correction for slip between the air and particle surface.

This example explains slow settling, rather than the motion of every particle. PM2.5 includes many sizes, densities, and shapes. Turbulence, rain, and surface contact also affect removal.

Fine particles can remain in air for days under suitable conditions. They can also enter deep parts of the lungs. Their small size matters for both atmospheric transport and exposure.

Where the smoke comes from

Delhi lies in the Indo-Gangetic Plain, south of the Himalayas. Regional winds can carry pollution across city and state boundaries. Local emissions therefore combine with pollution from the surrounding region.

A published decision-support study estimates source contributions for the post-monsoon and winter seasons of 2021–2022. Its post-monsoon average assigns 34.4 percent to Delhi and 31 percent to other National Capital Region districts.

Biomass burning contributes 7.3 percent in that seasonal estimate. Other regions contribute 27.3 percent. Biomass burning means combustion of plant material, including crop residue.

These shares change with the period and weather. The study finds that a dominant source can reach 65–69 percent during some peak events. A small seasonal average therefore does not mean a source has little effect on every day.

The whole winter in one chart

The video's saved CPCB series covers October 1, 2025 through January 31, 2026. Recalculation of its 123 daily values gives a mean AQI of about 309.

The series contains 78 days above 300 and 11 days at or above 400. Only 10 days exceed 400 strictly. This distinction matters because one daily value equals 400 exactly.

The highest saved value is 461 on December 14. Another peak reaches 440 on January 18. These describe the archived chart, not today's Delhi air quality.

The sequence illustrates continued winter pollution after the main farm-fire monitoring period. It does not identify each day's sources or prove that weather alone causes each peak.

What actually works

The final volume example uses an area of 1,483 square kilometres and a mixing depth of 300 metres. Multiplication gives about 445 billion cubic metres of air.

This is an illustrative box, not a measured daily mixing volume. It explains the scale difference between cleaning a room and managing a city's atmosphere. Air also moves continuously across the box boundary.

Reducing emissions from traffic, industry, construction, and burning reduces the material entering that air. Regional coordination matters because pollution crosses administrative boundaries. Weather changes the benefit visible on any particular day, but does not make emission reductions pointless.

What this means

Delhi's winter air contains emissions that poor dispersion can concentrate. Pollution still leaves through transport and removal, so the video's closing image must not imply a perfectly sealed lid.

Use current official readings for current conditions. The historical charts explain a process and a selected winter, rather than a live warning.

FAQ

Do fewer farm fires guarantee clean air?

No. Other emissions, incoming pollution, and weather still affect concentrations.

Does AQI identify a pollution source?

No. It reports air quality from pollutant measurements. Source estimates require additional observations or models.

Is the winter lid always at the same height?

No. The mixing layer changes with time, place, and weather.

Does slow settling mean particles never leave?

No. Air transport, rain, and contact with surfaces also remove particles.

Sources