Where Delhi’s PM2.5 Actually Comes From: The 2024–25 Source-Apportionment Map
Public debate about Delhi’s air assigns the blame in cycles. October: stubble burning. November: Diwali firecrackers. December: vehicles. The science is messier and more useful. Source apportionment — the discipline of measuring exactly which sources contribute what fraction of ambient PM2.5 — has produced five major Delhi studies in the last fifteen years (CPCB 2010, IIT-Kanpur 2015–16, SAFAR-India ongoing, TERI 2018, CEEW 2023). The numbers vary by season and by methodology, but a consistent picture emerges. This is the consolidated source-apportionment map for Delhi NCR, with the major uncertainties flagged.
Key numbers — annual average
These ranges combine the five major Delhi source-apportionment studies (2010, 2015–16, 2018a, 2018b, 2023):
- Vehicles (road transport): 18–40% of PM2.5
- Industry (factories, brick kilns, power): 8–25%
- Road and construction dust: 8–30%
- Biomass and waste burning (year-round, not just stubble): 6–15%
- Residential and commercial cooking: 7–15%
- Stubble burning (annual average): 4–8%
- Diwali firecrackers: <1% (annual average; up to 35% on Diwali night)
- Long-range transport from other regions: 15–30%
The wide ranges reflect both seasonal variation and genuine methodological differences. No single study is “right”; the consolidated picture is more reliable than any individual paper.
How source apportionment works
Two main methods:
1. Emission inventory + dispersion modelling. Researchers measure or estimate emissions from each known source category (vehicles by type, industries by SIC code, brick kilns by location, etc.) and run atmospheric models to predict the resulting ambient concentrations. The IIT-Kanpur 2015 study is a major example.
2. Receptor modelling. Researchers sample PM2.5 at ambient monitoring sites, chemically analyse the samples (markers like potassium for biomass, vanadium for heavy oil combustion, elemental carbon for diesel exhaust), and statistically back-calculate which sources contributed. The TERI 2018 and CEEW 2023 studies use this approach.
Both methods have known limitations. Emission inventories miss informal/illegal sources; receptor modelling can mis-attribute sources that share chemical signatures. The best practice is to compare both and treat the overlap zone as the credible answer.
Vehicles: 18–40% of PM2.5
Vehicles are the single largest controllable PM2.5 source in Delhi. Within vehicles:
- Heavy-duty diesel (trucks, buses) — disproportionately high emission per vehicle. The 30,000+ trucks transiting Delhi nightly (it sits on the major north–south freight corridor) emit large mass loads in a short window.
- Light-duty diesel (taxis, autos pre-2020) — significant; BS6 transition has reduced per-vehicle emission by ~70% but real-world emissions remain higher than type-approval limits (ICCT remote-sensing studies).
- Petrol two-wheelers — large fleet, modest per-vehicle emission, significant aggregate contribution.
- Petrol cars — relatively clean per-vehicle; smaller share of total.
The BS6 transition (since April 2020) has reduced new-vehicle emissions substantially. Vehicle PM2.5 contribution will fall further as the old fleet retires, but the timeline is decades, not years.
Industry: 8–25%
Includes:
- Brick kilns — 3,000+ legal kilns in Delhi-NCR contribute 6–7% of regional PM2.5. The shift from Fixed Chimney Bull’s Trench Kiln (FCBTK) to zigzag technology is reducing emissions by 50–70% in converted kilns, but only ~20% of surveyed kilns have completed quality conversion.
- Power plants — Delhi region’s coal plants have been progressively shut or made compliant; current contribution is smaller than a decade ago.
- Small industries — informal factories in the NCR belt (Bahadurgarh, Faridabad, Ghaziabad, Noida) contribute substantial PM2.5 that drifts into Delhi.
Road and construction dust: 8–30%
This is the most contested category in the data. The IIT-Kanpur 2015 study estimated road dust at the very high end (~30%); other studies place it lower. The variation reflects:
- Differences in how “dust” is defined (resuspended road dust vs. construction dust vs. windblown soil)
- Seasonal variation (dry season vs. monsoon)
- Whether suburban brick kilns and construction sites are counted here or in industry
Construction activity in Delhi-NCR — particularly the continued boom in Gurugram, Noida, Greater Noida, Ghaziabad and Faridabad — generates substantial PM10 and PM2.5. The 13 identified Delhi pollution hotspots (Narela, Bawana, Mundka, Wazirpur, Rohini, RK Puram, Okhla, Jahangirpuri, Anand Vihar, Punjabi Bagh, Mayapuri, Dwarka, plus floating one) are characterised partly by construction concentration.
Biomass and waste burning (year-round): 6–15%
This is separate from stubble burning. Year-round biomass burning includes:
- Residential heating in informal settlements during winter (wood, plastic, rubber)
- Garbage burning by sanitation workers and informal scrap collectors
- Construction-site fires for warmth or scrap disposal
- Rural cooking in NCR villages still using biomass
This category is hard to enforce against because the sources are diffuse and informal. It contributes year-round, unlike stubble fires.
Cooking: 7–15%
Residential cooking — primarily LPG and PNG, with some remaining biomass in urban poor areas — contributes a steady fraction. Commercial cooking (restaurants, dhabas, street food) adds on top. The Indian-cooking-specific factor: high oil-aerosol cooking styles produce higher emission rates than e.g. European cooking.
Stubble burning: 4–8% annual, 25–40% peak
The most-discussed source produces a small annual fraction but dominates peak winter days. See stubble burning explained for full coverage. The 6-week season in mid-October to late November can push the daily contribution to 25–40% on the worst days, occasionally higher with specific wind conditions.
Diwali firecrackers: <1% annual, briefly very high
Annual contribution is negligible because the burning happens for one to two evenings. Diwali night and the day after can see firecracker contributions of 20–35% briefly, then decay. The Supreme Court 2024–25 cycle has alternately banned and permitted firecrackers; green crackers (CSIR-NEERI certified) reduce emissions by ~30% but the absolute peak is still high.
Long-range transport: 15–30%
Pollution from outside Delhi-NCR reaches the city via wind. Sources include:
- Northwest: stubble fires (Oct–Nov), Punjab/Haryana industry
- Northeast: Uttar Pradesh industries, Indo-Gangetic Plain biomass burning
- West: Rajasthan dust storms (especially May–June)
Long-range transport explains why “fixing Delhi” alone won’t deliver clean air. Regional cooperation is necessary.
Why the numbers disagree across studies
Five reasons:
1. Year of study. Vehicle fleet, brick-kiln technology, and stubble burning have all changed. A 2015 inventory is not the same as a 2024 one.
2. Season measured. Winter source mix differs from summer. Studies that focus on winter peak days emphasise stubble and inversion-trapped local sources; annual averages look different.
3. Methodology (modelling vs. receptor). These produce systematically different answers for some source categories, especially construction and biomass.
4. Geographic scope. “Delhi” vs. “Delhi-NCR” vs. “Indo-Gangetic Plain” gives different answers. Local-only studies miss long-range transport; regional studies dilute the local-source signal.
5. What counts as a “source.” Re-suspended road dust — is that vehicles (they kick it up) or dust? Cooking smoke in the home — is that residential combustion or biomass? Methodology choices shift the percentages.
What this means for action
Three takeaways:
1. No single source is the whole problem. A complete fix requires action on vehicles, brick kilns, construction, biomass burning, cooking, and stubble fires — all together.
2. The “controllable share” is large. Roughly 70% of Delhi’s PM2.5 comes from sources where Indian regulators have direct policy levers. The remaining ~30% (long-range transport, occasional dust storms) is harder.
3. The “personal protection” share is 100%. Whatever the source mix, your indoor PM2.5 is what you actually breathe for ~16 hours a day. A whole-home fresh-air system holds indoor PM2.5 under 10 µg/m³ regardless of what’s happening outside.
FAQ
Why are the percentages so different across sources? Different studies, different years, different methodologies. The consolidated picture (ranges, not single numbers) is more honest than any single percentage.
Is the IIT-Kanpur 2015 study still valid? Largely yes, with adjustments. The qualitative source ranking is similar in newer studies. Vehicle composition has shifted (BS6, EVs growing), brick kiln conversion is advancing, but the overall framework holds.
Why isn’t there a more recent definitive study? A new comprehensive Delhi source apportionment study was commissioned in 2024 (the 6th since 2010) and is in progress. Updated results expected 2025–26.
Does adding more roads reduce vehicle pollution? No, generally the opposite. More road capacity induces more travel demand. Reducing total vehicle-km is the lever that matters.
Why is biomass burning year-round, not just stubble season? Stubble burning is a 6-week agricultural phenomenon. Residential heating, garbage burning, construction-site fires, and rural biomass cooking happen year-round and collectively contribute more than stubble across the calendar.