Current events
Delhi's new source data
still leaves a big gap
IITM's new atmospheric chemistry lab publishes a daily split of Delhi's PM2.5 into 15 source categories, the first in nearly three years. Between 29% and 42% of it is secondary particles that nobody can yet trace to a source, and the forecast model beside it still runs on a 2021 emissions estimate.
What IITM has published
Reported by Hindustan Times on 25 October 2026, from IITM's own pages. The figures below are HT's reading of those pages, not Swechha's.
On Saturday 24 October 2026 the Indian Institute of Tropical Meteorology (IITM) opened the Advanced Atmospheric Chemistry and Aerosol Toxicity Analysis Laboratory (Aatal) at New Rajendra Nagar in Delhi. Its 25 instruments cost about ₹80 crore. A new web page from the lab shows one figure a day for each of 15 source categories of Delhi's PM2.5.
For 20 to 24 October the largest group was secondary particles, the ones that form in the air when gases react, at 29.3% to 42.2% of the daily total. Vehicles were next at 19.4% to 27.6%. The seven burning categories together made up 20.5% to 35.6%, and crop-residue burning alone 2.5% to 4.7%. These shares are Hindustan Times's sums of IITM's categories.
IITM has said the lab will first split the vehicle share between diesel, CNG and petrol vehicles. It has no method yet to trace secondary particles back to where they started.
Why the unexplained share matters
A source breakdown is only useful for policy if it says what to act on. A third or more of the total that cannot be assigned to a source is a third of the problem that no rule can yet be aimed at.
The comparison matters too. Delhi's last source studies were IIT Kanpur in 2016 and TERI in 2018, and the city government stopped two later attempts, in 2021 and November 2023, over dissatisfaction with their methods and results. This is the first daily figure in years, which is why its limits are worth stating plainly.
Where it does not yet add up
29.3–42.2 %
Secondary particles, share of the lab's daily PM2.5 estimate
IITM, as summed by Hindustan Times · observed 20 to 24 October 2026
19.4–27.6 %
Vehicles, share of the lab's daily PM2.5 estimate
IITM, as summed by Hindustan Times · observed 20 to 24 October 2026
47 %
The lab's PM2.5 total (55.3 µg/m³) as a share of the middle CPCB reading (117.1 µg/m³, 37 stations)
IITM and CPCB, compared by Hindustan Times · observed 22 October 2026
2021 inventory
Emissions estimate the IITM forecast model runs on, compiled from 2016 surveys
IITM, per Hindustan Times · observed 25 October 2026
Counted or measuredModelled
The lab's total reads low. On 22 October its PM2.5 total was 55.3 µg/m³ against a middle reading of 117.1 µg/m³ across 37 CPCB stations, 47% of it. IITM officials told Hindustan Times the site was new and facing teething issues, and that the PM2.5 data would reflect accurately shortly.
The headline leaves out the biggest group. IITM's page names vehicles the dominant contribution each day, which is true only if both secondary categories are set aside. On 23 October one of them, secondary organic aerosol, was level with vehicles at 19.5% to 19.4%.
The forecast model is older than the lab. The decision-support model whose split the page shows by default uses a 2021 emissions inventory built from 2016 surveys. IITM's own page says newer emission fields are needed, and a 2026 baseline is due by June 2027. The model also does not simulate nitrate or secondary organic particles, which alone were 15.5% to 21.2% of the lab's daily estimate.
The numbers were still moving. Hindustan Times saw the page's Friday and Saturday readings vanish and return revised within three hours on 24 October. The page calls the split real time but gives one figure a day per category, with no hourly data and no stated method.
How this site reads it
Swechha's Delhi air page already carries two older source studies side by side, and for the same reason: a source split is a model, and it depends on the method and the inventory behind it. IITM's daily figures are a third voice, and so far the least settled one.
Until the lab's total agrees with the monitors and the secondary share is traced, the safe reading is narrow. Vehicles, burning and dust are all real, and a third or more of the particles are not yet attributed to any of them. We will not quote a single pie chart as the cause of Delhi's air.
Known
- IITM opened Aatal on 24 October 2026 and publishes a daily split of Delhi's PM2.5 into 15 source categories.
- Secondary particles were 29.3% to 42.2% of the lab's estimate for 20 to 24 October, ahead of vehicles at 19.4% to 27.6%, on Hindustan Times's sums.
- The lab's PM2.5 total on 22 October was 47% of the middle CPCB reading that day.
- The forecast model beside it runs on a 2021 emissions inventory built from 2016 surveys; a 2026 replacement is expected by June 2027.
Uncertain
- Where the secondary particles come from. IITM says the original source is not known and is working on it.
- Whether the lab's low total is a calibration problem that will be corrected, as officials told Hindustan Times, or something that also affects the shares.
- How the lab tells dung cake, wood, crop residue and other fuels apart within its burning categories. IITM has not said.
- Pollution carried in from outside Delhi. None of the 15 categories covers it, although IITM's note on the lab names regional transport.
Sources, and what to watch next
Everything above is traceable to one of these.
- Delhi gets pollution source data, but a big chunk stays unexplained Hindustan Times, Jasjeev Gandhiok and Binayak Dasgupta, 25 October 2026 · The reporting every figure on this page comes from
- Source apportionment of PM2.5 and PM10 of Delhi NCR TERI and ARAI, 2018 · One of the two older studies this site already publishes
What to watch
- The vehicle split into diesel, CNG and petrol, which IITM expects in about a fortnight.
- Whether the lab's PM2.5 total moves into line with the CPCB monitors.
- The 2026 emissions inventory, due by June 2027.
- Any method note that explains how the 15 categories are separated.
Based on Hindustan Times, Jasjeev Gandhiok and Binayak Dasgupta, 25 October 2026. Reuse freely — CC BY 4.0. Terms, method and limitations.
Next
The explanation behind it, the live reading, and the archive.