What JanVayu does that the other Indian air-quality sites do not

Published: 17 September 2026 | Author: Team JanVayu | Reading time: 11 min

Dr Sarath Guttikunda published a list this month of where to get air-quality data for Indian cities. It is the most useful thing of its kind, and reading it sent us to a dataset we had missed and should not have.

It also prompted an obvious question, which somebody asked us directly: if all of that exists, what is JanVayu for?

What follows is a capability table. Two caveats belong in front of it, since we are one of the parties being compared.

Most of the sites listed are not competitors, and several supply data we use. CPCB, WAQI, OpenAQ, Sensor.Community, XKDR and CREA all appear inside JanVayu.

The verification is uneven. Every JanVayu figure below sits in a data file in our public repository, and an automated check recomputes it and stops publication if it drifts. For the other columns we have used what their pages state and what we could test ourselves. Where we could test neither, the cell says unverified. Several of these sites block automated requests, which is their right and our limit.

The short version

Most Indian air-quality sites answer one question: what is the air like right now, where a monitor is. They answer it well, and several answer it better than we do.

JanVayu is built around four questions a live map cannot answer, whatever its quality:

1. What is the air where there is no monitor? India has roughly 565 continuous stations and 584,615 mapped village boundaries.
2. Is it actually getting better, or was that the weather?
3. Who promised what, and did they do it?
4. What do I do about it on Tuesday?

The table

Legend: yes = we verified it. no = we checked and could not find it. unverified = we could not test it and their pages do not say.

Updated 18 September 2026: a Hawa Ka Hisab column was added, and two rows that read "no" everywhere but our own column were wrong. See the correction under "Weather removed from the trend".

The rows that carry the argument

Most of the table is convenience. These five are the substance.

A number for every village and every ward

983,149 administrative areas: 36 states, 785 districts, 6,471 blocks and tehsils, 319,287 gram panchayats, 584,615 village boundaries (Local Government Directory layer; Census 2011 counted 649,481 villages), 3,359 city bodies and 68,596 wards. Every one carries an annual PM2.5 figure, most carry surface heat, tree cover and built-up share.

A network of monitors cannot do this. About 565 continuous stations cannot cover 584,615 village boundaries. The satellite retrieval can, and that is the entire reason we lead with a modelled layer rather than apologising for one.

The consequence is concrete: no Indian village meets the WHO annual guideline of 5 µg/m³, 371,938 of them (63.6%) are above India's own limit of 40, and the median village sits at 43.7. You cannot get that number from a dashboard of 157 Indian cities.

Weather removed from the trend

Every "air improved by X%" headline rests on not knowing whether it was the policy or the wind. A still, cold week traps emissions and the monitors read high; a windy week scatters them and they read low, with nothing changed at the source.

We run meteorological normalisation (Grange et al., Atmos. Chem. Phys. 18, 2018) over seven years of hourly readings for 44 cities. 33 are getting cleaner once weather is removed and 11 are not. Meerut −14.6 µg/m³ a year, Varanasi −14.2, Lucknow −14.0; Chandigarh +3.1, Gwalior +2.4, Mumbai +0.8.

Correction, 18 September 2026. This section first said we knew of no other public Indian source publishing weather-normalised air quality, and the table gave every other column a flat no on that row. Both were wrong. Hawa Ka Hisab, published by Ajay Maken, MP, has done it daily for Delhi since 19 July 2026 and publishes its method in full. The table above is corrected.

Further correction, 2 October 2026: two cells in the "Raw station data via API" row were also wrong. AQICN offers a token-based station API (JanVayu's own WAQI proxy uses it), and CREA's open API returns daily, outlier-filtered station values.

We should have caught this ourselves. Our own assistant already cites that site, using its decadal figures for Delhi as an independent check on our 1980-to-2022 reconstruction. The site was in our repository while this post said no such source existed. We knew it as a history and had not looked at what else it publishes.

The two are not substitutes. Hawa Ka Hisab asks each day whether Delhi's air was worse than the same fortnight last year once weather is accounted for, comparing matched stations against a fifteen-day baseline window, using a model trained on Delhi data from 2021 onward. That is the question you need if you are holding a minister to account this week. We ask whether a multi-year trend is real, across seven years and 44 cities, and report it in micrograms per cubic metre per year. For Delhi specifically, and for anything current, theirs is the better instrument.

The official bulletin as a series, and what it shows

CPCB publishes an AQI bulletin every day at 4pm as a PDF, and has since May 2015. It covered 10 cities in 2015, 264 in 2024, and between 218 and 258 on each day we hold so far in 2026. It is the official daily figure, and it has never existed as a series because it is a decade of PDFs. UrbanEmissions parsed the archive; we now read the PDFs directly as well.

297 cities, 2015–2026, as days in each official category per city per year. 2015 and 2026 are part years and are labelled as such on the page, because these are counts of days and a part year set against a full one is not a comparison.

The finding is uncomfortable. In CPCB's own 2024 bulletin, 264 cities reported a usable year and 221 of them did so on a median of fewer than three monitoring stations. For 204 cities the median was exactly one. Agartala, Ajmer, Amritsar, Aizawl and two hundred others have a daily official air-quality figure that is a reading from one place with the city's name attached.

Every dashboard in the table above, ours included, shows those cities a number. Only this tells you how thin it is.

An RTI, pre-filled

This is the row we would keep if we had to keep one. A number that does not end in a question to somebody is a number that changes nothing.

The RTI Assistant gives you the right Public Information Officer, questions phrased as requests for records rather than opinions, and the statutory anchors. ₹10, free with a BPL card, reply due in 30 days. The RTI Clinic workshop ends with a filed application, not a drafted one.

Material you can take and teach

Four workshops as plain text files at janvayu.in/workshops, seven games, a 10-question self-check, 45 blog posts, and an Open Data API. CC BY-NC-SA 4.0, so you can cut them, translate them and put your own city's numbers in.

Where the others are better

CPCB's portal is the official record and ours is not. When the two disagree, theirs is the official one, which is why we read their bulletin.

OpenAQ is a better raw-data API than ours and does not pretend otherwise. If you want station measurements to build on, start there.

CREA's NCAP analysis is better than ours and our accountability pages lean on it. Tracing the Hazy Air is the reference, and we cite it rather than reproducing it.

The sensor networks reach places we cannot. PurpleAir, AirGradient, AirVeda, Aurassure and Sensor.Community give street-level density that a regulatory network never will. Less accurate, far denser, and the right tool for "is it worse on my street than the next one".

VayuBuddy answers questions against CPCB data, much as our assistant does. Two independent answers to the same question are worth having, and we would rather say so than pretend we are the only one.

We do not attempt IQAir's global comparability. Their World Air Quality Report is what makes an international ranking possible.

Hawa Ka Hisab covers Delhi better than we do, and is far more current. It publishes a daily report (roughly twelve pages, by its own description), a short Hindi film with it, a weekly edition, and a reconstruction of Delhi's air back to 1980. We give a city one trend line across seven years. It gives Delhi a weather-adjusted verdict every morning, measured against the same fortnight a year earlier.

Its provenance is worth stating, since the site states it itself: every page says it is published by a serving opposition MP and is not a neutral third party. JanVayu is non-partisan and that site is not. The method is published in full, which is the part to read before deciding what the analysis is worth.

What this adds up to

For whether to go running this evening, almost any of these will tell you, and several will tell you faster than we will.

The questions we built for are slower ones. What does my village breathe, when there is no monitor within fifty kilometres. Is my city's improvement real or was it a wet year. How many days was my city officially Poor, and how many monitors produced that figure. What do I write, and to whom. We know of no other single site that answers all of those, though several answer one of them better than we do.

We are built for a different question, and somebody choosing a tool should know which.

Corrections wanted, particularly on the other columns. If we have understated what a source does, that is our error and we will fix it in this post with a dated note. Several of these sites block automated requests, so parts of the table rest on their own descriptions rather than our testing, and those cells are marked. contribute@janvayu.in

Sources. Guttikunda's list, September 2026. JanVayu figures from the data/ folder of our repository, each recomputed by automated checks. CPCB bulletin figures parsed from the published PDFs. De-weathering after Grange et al. (2018), Atmospheric Chemistry and Physics 18, 6223–6239. Village and ward air from SatPM2.5 V6GL03 (ACAG / Washington University, 2024).