Data analysis
Delhi is not one city,
and its air proves it
On the evening of 8 September, one Delhi monitor read 35 and another read 211. Both were CPCB stations, both were reporting normally, and both were describing the same city at the same hour.
One hour, forty-five monitors
The gap between the cleanest and the dirtiest monitor was larger than the gap between a good day and a bad one.
At 20:00 IST on 8 September 2026, forty-five of Delhi's continuous air quality monitors were reporting. The lowest index reading in the city was 35, at Alipur. The highest was 211, at Bawana, governed by PM2.5.
Twenty-eight of the forty-five — nearly two in three — were above 100, the top of CPCB's ‘Satisfactory’ band and the point at which the air is over the 24-hour standard for its governing pollutant.
The unweighted mean of all forty-five station readings was 114. Nobody in Delhi was breathing 114.
Why an average hides it
A single city figure is the number that reaches a headline, a school's decision about outdoor games, and an office's decision about whether to say anything at all. On this evening that figure sat in the middle of a range 176 index points wide.
The range is not noise. Bawana and Alipur are both in north Delhi, roughly ten kilometres apart. A reading is a description of the air at a monitor, and between two monitors there is no published value at all.
This is why every reading on this site is the worst reporting monitor, named, with the unweighted mean printed beside it as a cross-check. It answers a different question from CPCB's population-weighted city figure, and it is the question a person standing in Bawana is asking.
The readings
Every figure below carries the moment it was observed and the source that published it. None of them moves after this date.
211 AQI
Worst reporting monitor — Bawana, governed by PM2.5
CPCB, via the data.gov.in mirror · observed 8 September 2026, 20:00 IST
35 AQI
Cleanest reporting monitor — Alipur
CPCB, via the data.gov.in mirror · observed 8 September 2026, 20:00 IST
114 AQI
Unweighted mean of all reporting stations
CPCB, via the data.gov.in mirror · observed 8 September 2026, 20:00 IST
28 of 45
Stations above 100, the top of ‘Satisfactory’
CPCB, via the data.gov.in mirror · observed 8 September 2026, 20:00 IST
Counted or measuredModelled
CPCB's index is not an average of pollutants. Each station's AQI is the highest of its per-pollutant sub-indices, and the pollutant that produced it is named. At Bawana that evening it was PM2.5.
The index is also already a rolling figure: the particulate sub-indices are computed over twenty-four hours. So 211 at 20:00 describes a day-long window ending at that hour, not a spike at that minute — which makes the spread across stations harder to dismiss as a transient.
None of these figures will change. They are the readings as observed, snapshotted at that stamp; the live page moves and this page does not.
What the spread means
Separated on purpose. The second list is content, not a disclaimer.
Swechha's air archive keeps every hourly observation with the worst monitor named and the full state of the network at that hour. That is what makes a spread like this checkable after the fact rather than a claim about an evening nobody can revisit.
It is also why the site refuses to publish a single city number on its own. A city average is a real and useful quantity; it is not a description of anybody's exposure, and the distance between those two things is 176 index points on an ordinary September evening.
Known
- Forty-five monitors were reporting at that hour; the highest read 211 and the lowest 35.
- Twenty-eight of them were over 100, which is the top of CPCB's ‘Satisfactory’ band.
- PM2.5 was the governing pollutant at the worst station.
- All figures are CPCB's own, read from the data.gov.in mirror of its network feed.
Uncertain
- Why the gap was that wide on this particular evening. A local source, a wind direction and an instrument problem all produce the same shape in this data.
- What the air was doing between the monitors. There is no published value for the space between two stations and interpolating one would be an invention.
- Whether this spread is typical. The archive is weeks old, not years, and a single evening cannot answer it.
- What anybody was actually breathing. Ambient monitoring says nothing about indoor air, where most of the day is spent.
Sources, and what to watch next
Everything above is traceable to one of these.
- Real-time Air Quality Index from various locations data.gov.in, mirroring the Central Pollution Control Board · The per-station feed these readings come from
- National Air Quality Index Central Pollution Control Board · The six bands and the sub-index method
- National Ambient Air Quality Standards, 2009 Central Pollution Control Board · The 24-hour standards the index is measured against
What to watch
- Whether the spread widens as the winter inversion sets in, when the whole city's readings rise together.
- Whether the same stations sit at the top of the range repeatedly, which would point at a local source rather than at meteorology.
- CPCB's own city figure against the worst monitor over the same period — the two diverge most on exactly the evenings that matter.
Reuse freely — CC BY 4.0. Terms, method and limitations.
Next
The explanation behind it, the live reading, and the archive.