Your phone says AQI 180 and the government board says 300. Both can be right.

Two people standing on the same Delhi footpath can look at their phones and see 180 and 300. Neither app is broken and neither is lying. The Air Quality Index is not a reading off an instrument — it is a formula applied to readings, and almost every confusing thing about it comes from the choices baked into that formula. Once you know what those choices are, the number becomes far more useful, and far less alarming in some situations and considerably more alarming in others.

The AQI is a translation, not a measurement

What instruments actually measure is concentration: micrograms of fine particulate matter per cubic metre of air, parts per billion of ozone. Those units are precise and meaningless to most people. An index exists to turn several such numbers into one figure a person can act on.

India's National Air Quality Index, launched by the Central Pollution Control Board in 2014, does that for eight pollutants: PM10, PM2.5, nitrogen dioxide, sulphur dioxide, carbon monoxide, ozone, ammonia and lead. It reports on a 0–500 scale in six named bands — Good (0–50), Satisfactory (51–100), Moderately Polluted (101–200), Poor (201–300), Very Poor (301–400) and Severe (401–500).

The arithmetic runs in two steps. First, each pollutant's concentration is converted into its own sub-index using a table of breakpoints, with linear interpolation between them. Then the AQI is announced as the highest sub-index of the lot — not the average.

Why the scale is not a straight line

The breakpoints are the part almost nobody looks at, and they are where the index stops being intuitive.

For PM2.5, India's bands run 0–30 µg/m³ for Good, 31–60 for Satisfactory, 61–90 for Moderately Polluted, 91–120 for Poor, 121–250 for Very Poor, and anything above 250 for Severe. Notice what that does. The first hundred points of AQI cover a 60 µg/m³ span of real air. The Very Poor band alone covers 130 µg/m³ in the same hundred points.

So a jump from AQI 90 to AQI 190 is a much smaller change in the air than a jump from 290 to 390, even though both look like "a hundred points worse". The index is deliberately compressed at the dirty end. Treating AQI as a linear measure of how much pollution you are breathing will mislead you every time, particularly during a severe episode where the actual concentration can double while the index barely moves.

The max rule, and what it hides

Reporting only the worst sub-index is a defensible public-health choice: if any single pollutant is at a harmful level, the summary should say so rather than diluting it in an average. It also means the headline number tells you nothing about everything else in the air.

A station reporting AQI 210 in winter is almost always reporting PM2.5, and the ozone and NO₂ sub-indices underneath might be perfectly ordinary. A station reporting 210 on a bright May afternoon may be reporting ozone, which is a different exposure with different advice — outdoor exercise in the afternoon becomes the wrong thing to do, while a mask that stops particles does nothing about it. The category tells you how bad; only the underlying pollutant tells you what to do about it. Most dashboards do publish the responsible pollutant, usually in small type, and it is the single most under-read field on the page.

There is also a data-sufficiency rule that matters more than it sounds: an AQI should not be computed unless a minimum number of pollutants are available, including at least one of PM2.5 or PM10. A station can go quiet, and a gap is not the same as clean air.

Why the number lags the evening

Most of India's sub-indices are computed on a 24-hour average, with carbon monoxide and ozone on an 8-hour window. That single design decision explains the most common complaint about the index.

If a still, cold evening collapses the mixing layer at 7 p.m. and concentrations triple by 9 p.m., a 24-hour rolling average is still carrying twenty-two hours of better air. The index climbs slowly, understating what you are breathing at that moment. The same thing happens in reverse: the morning after Diwali or a stubble-burning peak often shows the highest AQI of the season, hours after the actual air has begun to improve, because the window is only then fully loaded with the bad night.

Averaging is the right choice for a health index — the health effects being flagged are dose-related, not instantaneous — but it means AQI answers "what has the last day been like here" rather than "what is it like right now". For the second question, the raw hourly PM2.5 concentration is the more honest field to read.

The AQI is a public-health summary wearing the costume of a measurement. It is built to trigger the right advice for most people, not to describe the air in front of your face at 9 p.m.

Why two apps disagree

The most common reason is that they are running the same air through different formulas.

The US EPA scale is popular in international apps and uses different breakpoints from India's: the same PM2.5 concentration usually produces a substantially higher number on the US scale in the moderate range, because its bands are tighter. Some apps also use the EPA's NowCast, a weighted calculation that leans on the most recent hours instead of a flat 24-hour mean, so it reacts to a worsening evening far faster than a CPCB station will. A phone reading 300 while the official board shows 180 is often not a contradiction at all — it is a faster, stricter scale describing the same afternoon.

The second reason is where the numbers come from. A regulatory continuous monitor is a heavy, calibrated instrument in a fixed enclosure, and India has several hundred of them, concentrated in cities. Many apps interpolate between them or blend in low-cost optical sensors, which infer particle mass by scattering laser light. Those sensors are genuinely useful and genuinely noisy: humidity makes particles swell and scatter more, so they over-report on damp mornings and near fog unless corrected. Placement matters as much as calibration — a sensor on a balcony over a bus stop is measuring a bus stop.

None of this makes the cheap network worthless. It makes it a different instrument, answering a hyper-local question that a station eight kilometres away cannot.

What "satisfactory" actually means

The band names are the most quietly misleading part of the index, because they read as verdicts on safety and are not.

India's Satisfactory band tops out at 60 µg/m³ of PM2.5 as a 24-hour average, which is exactly the national ambient standard. The World Health Organization's 2021 guideline for the same 24-hour window is 15 µg/m³, and 5 µg/m³ as an annual mean. A city sitting comfortably in Satisfactory all year is running at several times the WHO guideline. That is not a flaw in the arithmetic — national standards are set against what is achievable as well as what is ideal — but it does mean a green label on a dashboard is a statement about Indian policy targets, not a certificate of clean air.

The practical reading: use the category for daily decisions, use the named pollutant to choose what to do, use the raw concentration when you want to know the truth, and remember the guideline the category is not measured against.

Why it matters for students and researchers

Air quality indices sit at an awkward and interesting junction — atmospheric chemistry, instrumentation, statistics and public communication — and nearly every open question about them is interdisciplinary. How should sub-indices be combined when several pollutants are elevated at once? How should low-cost sensor networks be calibrated against reference monitors in high-humidity conditions? How well does a 24-hour construct serve populations whose exposure is dominated by a two-hour commute? What does an index built on outdoor stations say about indoor exposure, where most people spend most of the day?

Those are exactly the interdisciplinary atmospheric problems that the International Journal of Atmosphere (ISSN 3049-3900), a peer-reviewed journal launched in 2024, exists to publish — research, reviews and technical notes across atmospheric science, with an explicit emphasis on work that combines fields. For students of environmental science, meteorology and public health, and for anyone building or auditing a monitoring network, the index is a good reminder that a number reaching millions of phones is only ever as good as the assumptions someone documented in a table of breakpoints.

Frequently asked questions

How is the Air Quality Index calculated in India?

Each of eight pollutants is converted into a sub-index using CPCB breakpoint tables and linear interpolation, and the AQI reported is the highest of those sub-indices. Most pollutants use a 24-hour average concentration, while carbon monoxide and ozone use an 8-hour window.

Why does my app show a different AQI from the government value?

Usually because it uses the US EPA scale rather than India's, which produces higher numbers for the same PM2.5 concentration, and often a NowCast calculation that responds to recent hours instead of a flat 24-hour mean. Different apps also draw on different monitors, including low-cost sensors.

Is AQI the same as PM2.5?

No. PM2.5 is a concentration in micrograms per cubic metre; AQI is an index derived from it and seven other pollutants. On most Indian winter days the AQI is being driven by PM2.5, but the two numbers are not interchangeable and the scale between them is not linear.

Why does the AQI stay high the morning after the air has improved?

Because the sub-index is computed over a rolling 24-hour window. The morning after a severe night, that window is fully loaded with the worst hours, so the index peaks even as current concentrations fall.

Is an AQI in the Satisfactory band safe?

It meets India's national ambient standard, but that band extends to 60 µg/m³ of PM2.5 over 24 hours, four times the WHO's 2021 guideline of 15. Satisfactory means acceptable against national policy targets, not equivalent to clean air.