Why North India’s Air Collapses Every Winter — and What Electrification Actually Fixes

Between roughly the middle of October and the middle of February, air quality across the Indo-Gangetic plain deteriorates on a schedule you could set a calendar by. It is tempting to read that regularity as evidence of a single seasonal cause. It is not. What changes in winter is mostly the ventilation, not the emissions.

Three things happen together. The mixing layer — the depth of atmosphere that pollution disperses into — collapses from a couple of kilometres on a summer afternoon to a few hundred metres on a still winter morning. Temperature inversion puts a lid of warm air above cold surface air, so nothing rises through it. And wind speeds drop, so nothing moves sideways either.

The same quantity of smoke, dust and exhaust released into a tenth of the volume of air gives you roughly ten times the concentration. That is the entire mechanism. Emissions that are survivable in June are lethal in December because the air stops carrying them away.

This matters because the trend line is going the wrong way. Analyses of the October-to-February 2025–26 season found average PM2.5 concentrations across the region materially higher than the winter before, with seasonal averages around 91 µg/m³ against a national annual standard of 40, and peak daily readings above 390 µg/m³ at the onset of winter. On 17 and 18 January 2026, PM2.5 in Delhi crossed 320 µg/m³ — roughly double the same dates a year earlier.

It is also not a Delhi story. Of 238 Indian cities assessed for that winter, the large majority failed to meet the permissible annual limit for fine particulate matter, with Ghaziabad and Noida frequently worse than Delhi itself. The capital gets the headlines because it has the monitors and the press corps.

PM2.5 is not one substance. It is a size class — particles under 2.5 microns across, small enough to pass the nose and throat and reach deep into the lung. What matters for policy is where those particles come from, and they arrive by two quite different routes.

• •Primary particles are emitted directly as particles: soot from incomplete combustion in diesel engines and biomass fires, dust lifted from roads and construction sites, fly ash from industry.
• •Secondary particles form in the atmosphere from gases. Nitrogen oxides from engines and sulphur dioxide from fuel combustion react with ammonia — much of it from agriculture — to form nitrate and sulphate aerosols. In a cold, humid, stagnant winter atmosphere this chemistry runs faster, which is why secondary particles are a larger share of the winter load than the summer one.

The practical consequence is that you cannot fix winter air by controlling only what comes out of chimneys and exhaust pipes as visible smoke. A significant fraction of the particle mass was invisible gas when it left the source.

Paddy residue burning in Punjab and Haryana genuinely does contribute, and on a handful of days in late October and early November, with the wind blowing the right way, it contributes enormously — enough to dominate the daily figure.

But the fires end in November and the smog does not. Analyses of the season repeatedly find that Delhi’s air gets worse in December and January, after the burning has finished. Averaged across the full winter, crop fires are a minority contributor; on many days in the burning window itself, the modelled contribution is in the low single digits of percent.

Source apportionment is contested territory and the numbers vary by method, by year and by which boundary you draw. But across studies the direction is consistent: once you set aside long-range transport and look only at what is emitted inside the city and its immediate surroundings, the transport sector is the single biggest contributor to PM2.5.

The Decision Support System used for Delhi attributes something in the region of 45 per cent of local PM2.5 emissions to transport. Within the primary particle fraction specifically, vehicles come out at around 40 per cent, and some analyses of local emissions alone put the vehicle share above half. Treat any single figure with caution — the honest summary is “the largest single local source, by a clear margin, and rising with vehicle population.”

Two features of the transport contribution make it particularly difficult. It is emitted at street level, in the volume of air people actually breathe, rather than from a stack a hundred metres up. And it grows structurally with vehicle registrations, so a fleet that gets cleaner per vehicle can still emit more in aggregate.

It is worth being precise here, because both the enthusiasts and the sceptics overstate their case. Replacing a combustion vehicle with an electric one removes some emissions completely, reduces others, and does nothing at all about several.

Read down that table and the fair conclusion is this: electrification is the most effective available lever on the largest local source, and it is not sufficient on its own. Anyone claiming it solves the problem is wrong. So is anyone claiming it does not help.

The Graded Response Action Plan works by escalating restrictions as air quality worsens — construction halts, certain vehicle categories are barred, diesel generator use is curtailed, schools shift online. It is an emergency measure and it functions as one: it shaves the top off the worst episodes.

What it cannot do is change the annual average, because it is only active on the bad days and it stops the moment the wind picks up. Every year the same sequence runs — a curtailment in November, a debate in December, a rainstorm in January that clears the air and the conversation with it.

There is a symmetry worth noting for anyone running an electric fleet. The same weather that traps the pollution also degrades the vehicles fighting it.

Lithium chemistry slows down in the cold. Internal resistance rises, usable capacity falls, and a pack that gave 90 kilometres in October may give 75 in January — not because it has aged, but because it is cold. Charging is the more serious constraint: charging a lithium-ion cell below about 0°C plates metallic lithium on the anode, which causes permanent damage, and a competent BMS will refuse the charge until the pack warms.

For most of India this is a nuisance rather than a crisis, but in the north it is a real operational planning issue in December and January. Our guide on seasonal range variation covers what to expect and how to manage it, and the daily battery check is worth doing more carefully in the cold months.

If you are asking what a city, a fleet operator or a facility manager can do that changes the winter average rather than the winter headline, the honest list is short and unglamorous.

• •Electrify the vehicles that run the most hours. A commercial three-wheeler covering 100 kilometres a day displaces roughly ten times the emissions of a private car covering ten. Duty cycle, not vehicle count, is what matters.
• •Remove the standby diesel. Backup generators are a concentrated urban source and are usually replaceable by a battery system that also earns its keep the rest of the year by shaving demand charges.
• •Pave, sweep and manage construction dust. Deeply boring, consistently underfunded, and one of the few interventions that acts directly on the resuspension fraction.
• •Attack the biomass fraction with heating and cooking access, not enforcement. People burn refuse at 4 a.m. in January because they are cold.
• •Fund the monitoring. You cannot apportion what you do not measure, and most of the argument about who is responsible is downstream of thin data.

None of this is fast. But the reason the same three months look identical every year is that the sources are structural, and structural problems only respond to structural answers — which, unlike an emergency curtailment, keep working after the wind changes.