Delhi at AQI 500

Delhi at AQI 500

A calculation of station-day indices from the saved hourly record. The current window ends at 16:00 IST. This is a station record, not a count of city-wide AQI days.

Delhi needs concentrations beside the index, and a clear account of the periods in which an index cannot be calculated.

The same display value can cover different measured concentrations. Of the 1,348 station-days displayed as 500, 1,339 exceed 500 before the display cap and rounding. Another 9 reach the displayed value through rounding. These cases must not be described as identical.

A station-day means one station on one reporting date. Several stations on the same date contribute several station-days. The figure does not mean that Delhi as a whole had an AQI of 500 on that many days.

A missing or unqualified index is not a clean-air reading. The tables retain the number of eligible station-days and identify the January to August 2022 period as unavailable for this multi-pollutant calculation.

The demand
Publish the concentration behind a capped index and explain each substantial gap in the station record. Neither the cap nor a missing value should conceal what the record can establish.

What a capped index means

A numerical display limit is different from the range of a concentration instrument.

CPCB describes an index scale running to 500. Its original calculator has severe-band formulas that continue arithmetically above 500. For example, a PM2.5 concentration of 600 micrograms per cubic metre gives an uncapped formula result of about 669.2. A separate display cap reduces that result to 500; it does not erase the measured concentration.

Here C is the eligible 24-hour mean concentration. The calculation keeps the uncapped arithmetic, applies the stated 500 display cap, and rounds the displayed index half up to the nearest integer. The underlying measurement remains available.

Rounding can produce a displayed 500 just below an unrounded value of 500. It is therefore wrong to infer that every displayed 500 proves a concentration above the exact breakpoint. It is also wrong to interpret a concentration ratio as the same multiple of health harm.

PM10 and PM2.5 describe particle-size fractions. A high particle reading alone does not identify construction, traffic, burning or another emissions source. Source attribution requires additional evidence.

Totals need their denominators

An eligible station-day has enough observations for the stated multi-pollutant index. A zero and an unavailable value have different meanings.

The last column changes only the reporting window to the 24 hours ending at 10:00 IST, using the same source checks and formula. These are sensitivity results. They are not interchangeable with the primary 16:00 series.

The same January to August dates

The 2022 archive contains PM2.5 and PM10 observations during these months, but not enough qualifying AQI pollutants for the required three-pollutant index. This table cannot support a claim that no station reached 500 in that period. PM observations remain evidence even when the combined index is unavailable.

Even identical calendar windows do not remove differences in station coverage, pollutants or eligible dates. These totals are not a controlled trend in pollution. The 2026 full-period column ends on 17 October; it must not be compared as a full calendar year.

Stations and months

One station on one date is one station-day. Counts below refer to the primary 16:00 IST series.

These are the ten stations with the largest displayed-500 counts in the assessed record. Different numbers of eligible days limit a direct ranking of their air quality. The complete station table is available with the report.

Month totals also reflect the available observation record and the shorter 2026 period. They do not isolate weather, policy or an emissions source.

Episodes and concentrations

A high count across several monitors can occur on a few dates. The concentration table retains the physical measurement behind the capped display.

Four large eligible 24-hour means are shown separately for each particle fraction. Concentrations are rounded down to one decimal. These are not hourly peaks, and the table is not an estimate of a multiple of health harm. The dated ceiling results, including gas sub-indices, are retained in the downloads.

The record supports scrutiny of particular dates and monitors. It does not show that the rest of the year was safe, or that a particular agency or activity caused the measured pollution.

Anand Vihar, hour by hour

PM10 concentrations in the later saved archive for 29 August to 2 September 2026. Calendar hours below differ from the 16:00 reporting window.

All hourly values are micrograms per cubic metre and are rounded down to an integer for this matrix. A dash means no numeric value in the selected archive. An asterisk identifies a value outside the stated analysis range; it remains visible here but does not enter a mean.

For the reporting window ending at 16:00 on 30 August, PM10 has 20 valid hours and a mean of 529.0 micrograms per cubic metre; the combined displayed index is 500. The 31 August window has only 12 PM10 hours, below the required 16.

This is a later saved record, not proof of which values were available to an earlier live feed. A gap in an early download may later be filled. The archive cannot by itself establish equipment shutdown, intent, the missing concentration, or the direction of any effect on a city average.

The 16-of-24 requirement applies separately to each pollutant in the reporting window, and an AQI also needs at least three qualifying pollutants. A count of calendar-day PM10 hours is therefore not by itself a certificate that a daily combined AQI can be calculated.

Method, sources and limits

The observed record is calculated under explicit rules. It is not a claim that every sensor reading has been independently validated.

The period is 1 January 2021 to 17 October 2026. Each primary station-day ends at 16:00 IST and uses hourly labels from 16:00 the previous day through 15:00 on the stated date. Data from 31 December 2020 supply the first window.

The source is CREA's saved compilation of CPCB and DPCC station observations. The calculation selects raw observed records identified as CPCB, station level, India, Delhi and the canonical micrograms-per-cubic-metre unit. Identical duplicate source, station, pollutant and hour values count once. Conflicting duplicates block acceptance.

The existing value checks retain positive values below 1,000 micrograms per cubic metre for particles and gases other than carbon monoxide. Carbon monoxide uses a ceiling of 50,000 micrograms per cubic metre and is converted to milligrams for the calculator. Excluded observations and source fingerprints are retained in the data notes.

PM2.5, PM10, nitrogen dioxide, sulphur dioxide and ammonia use 24-hour means when present. Carbon monoxide and ozone use the largest eight-clock-hour mean wholly inside that window. At least 16 distinct hourly observations are required for a pollutant sub-index. Six of eight valid clock hours is our stated completeness convention for an eight-hour block; it is not quoted as a separate CPCB requirement.

An overall index needs at least three qualifying AQI pollutants, including PM2.5 or PM10. It is the largest pollutant sub-index, using the original CPCB calculator formulas. No unavailable pollutant is filled with zero. This report uses observed concentrations; no weather adjustment or model prediction is substituted.

The network and its availability change. A calendar period with no eligible AQI cannot establish clean air. A capped index does not erase a concentration, and missing observations do not establish how an unobserved value would change an average. Emissions, individual exposure and causation are not estimated here.

Primary documents and data
CPCB: How is AQI calculated?
CPCB: original AQI calculator
CPCB: National Air Quality Index report
CREA data API and documentation
Full report, dated results and source notes

CREA has no involvement in this analysis. Responsibility for its interpretation and any error rests with the publisher.