Guide

The CPCB air-quality bands

India's national scale runs 0–500. The band decides the colour of the sky above and what the advice says.

Common questions

Why is my risk score different from the AQI?

The AQI describes the air. The risk score describes you in that air — it adds your age, health condition, and what you are planning to do. The same air can be routine for a healthy adult indoors and serious for a senior with COPD about to exercise outside. The gap between the two numbers is the whole reason this site exists.

How is the risk score calculated?

It has its own section below, including every number that goes into it and where that number came from. The short answer: it starts from the air, adds how hard you will be breathing, and adds how much harder this air is on your body. It is a decision aid — not a validated clinical instrument, and not a diagnosis.

Where does the air data come from?

Ground monitoring stations run under India's CPCB network. We read CPCB's own publication first, on the government's open data site data.gov.in; when it has nothing for a place we fall back to the WAQI feed, which republishes the same network. When neither answers for a place, no number is shown for it: we do not have one, and we will not invent one. If the only measurement we have for that place is an earlier one, we tell you what it was and when it was taken, and nothing else is worked out from it — no band word, no colour, no risk score and no go-out window. That holds on the Today page exactly as it does on City Pulse. On City Pulse a CACHED tag marks a reading we are still holding rather than one that just arrived, and says how old the measurement is; NO READING marks a place we hold nothing for. Nothing old is ever presented as live.

Why does the answer sometimes ignore my question?

Answers are written for the persona shown on the Today page, not for a hypothetical one. Asking "what if I were fit?" will still return advice for the persona you have set. Change the persona and ask again — the whole page, including the answer, is rewritten for whoever you select.

What happens to what I type?

Your age, health condition and plans stay in the page address and your session — those are never written to a database. The area you pick is the one part of your persona that is stored deliberately, so the System view can show which areas get requests; it is never stored alongside your health condition. Questions are checked for prompt-injection before reaching the model. Logs keep a one-way hashed session id and status fields; the security log keeps at most a 120-character snippet of a blocked prompt. Separately, each page load adds one to a count for the width band your browser reports — a count for each band, with no time, no session id and nothing else attached.

Is this medical advice?

No. It is general guidance built from public health sources (CPCB, WHO, GINA, GOLD, AHA, ACOG, EPA). If you are unwell, or your usual medication is not working as it should, contact a doctor.

Where the air number comes from, and what it is not

The readings come from India's CPCB monitoring stations. Most of them reach us as CPCB publishes them: concentrations, in the units the instrument records. Some reach us instead through the WAQI feed, which republishes the same network but on the United States EPA index rather than India's — it moved every Indian station onto the US scale in January 2016, and says itself that its figures will therefore differ from India's own National AQI portal.

The two scales are far apart. 60 µg/m³ of PM2.5 is 100 "Satisfactory" on India's scale and about 154 "Unhealthy" on the US one. So a reading that came through the WAQI feed is converted: its US index values are turned back into concentrations, and the Indian figure is worked out from those. A reading read from CPCB directly arrives as concentrations already and is never put through that conversion. Either way the number you see here is the kind of number you see everywhere else in Delhi.

What that number is not. It is worked out from particulate matter only — PM2.5 and PM10, and sometimes from just one of those two, which the reading on the Today page says beside itself. India's official method uses up to eight pollutants and needs at least three, so on a day when a gas like ozone is the worst thing in the air, the official figure would be higher than ours. Under an answer that carries a reading, the provenance panel names which of the two sources it came from, and shows WAQI's own figure when — and only when — the reading came through WAQI.

The World Health Organization comparison

The Organization's guideline for PM2.5 is 15 µg/m³ averaged over 24 hours, and 5 µg/m³ averaged over a year. The line on the Today page makes that comparison. What it compares against depends on where the reading came from, and the two are not the same thing: CPCB publishes a 24-hour average, and the WAQI feed publishes the latest hour, which is not a daily average at all.

The guideline is also stricter than it sounds in one direction and looser in another. WHO defines the 24-hour level as the 99th percentile of a year's daily averages — that is, three or four days above it a year is still within the guideline. It is not a limit for any one day. We round the comparison to a single figure and always say "about", because the underlying reading is not precise enough to deserve more.

WHO global air quality guidelines. Particulate matter (PM2.5 and PM10), ozone, nitrogen dioxide, sulfur dioxide and carbon monoxide. Geneva: World Health Organization; 2021.

How your risk score is worked out

The exertion part of this score comes from published breathing rates (US EPA, confidence rated medium). The health-condition and age parts are our own judgement, not a validated medical model.

The score adds three things together and stops at 100: a starting figure for how bad the air is, how much air you will breathe in while doing what you have planned, and how much harder this air is on your body than on an average adult's.

What the words on the score mean

The part that comes from published research

Breathing harder means breathing in more of the same air. How much more is measured, and we use those measurements rather than guessing. The figures below are litres of air per minute (shown as cubic metres per minute, the unit the source uses), for the five age groups this site offers.

U.S. EPA, Exposure Factors Handbook: 2011 Edition (EPA/600/R-09/052F), Chapter 6, Table 6-2 -- recommended short-term inhalation values, mean column, m3/min. EPA rates its own confidence in this recommendation Medium (Table 6-3).

Which of your plans counts as which effort level is our reading, not the source's: Outdoor exercise = hard · School run = moderate · Commute = light · Stay home = at rest.

The part that is our own judgement

There is no single published number for how much worse polluted air is for a person with asthma, or COPD, or a heart condition. Rather than invent one that would look authoritative, we say plainly that the figures below are our ordering of who is most affected — the order is well supported, the exact sizes are not. The same applies to the extra weight given to children and older people, which is there for reasons that are real but do not reduce to a number: developing lungs, more air breathed per kilogram of body weight, less reserve to draw on.

Unvalidated clinical heuristic. Relative ordering only, chosen by the author from general public-health guidance; not derived from, or validated against, any published risk model.

If you think the balance here is wrong, you can see every figure and change it — the whole calculation is a hundred lines of readable code, and nothing about it is hidden from you.