How Bad Is Delhi's Air? — AQI 2021-2024

Big picture first

Only about 1 in 5 days had air you could call clean

Delhi's official categories run from "Good" to "Severe." Add up four years of days and only "Good" + "Satisfactory" make up roughly a fifth of them. Nearly half of all days were "Poor" or worse. Hover any bar for the exact count.

21.8%

days rated Good or Satisfactory

46.5%

days rated Poor, Very Poor, or Severe

1,461

days of real data, zero gaps

Every day from Jan 2021 to Dec 2024, grouped into Delhi's official AQI categories

Trend over time

Air quality actually got better from 2021 to 2023 — then ticked back up

Averaging AQI by month over the full four years shows the same steep winter spikes every year, but the overall level drops from 2021 through 2023 before rising again in 2024. Hover any point to see that month's average.

Monthly average AQI, January 2021 through December 2024 · each spike is a winter

A quick note: an earlier draft of this project's README described a "clear worsening trend from 2021 to 2023." Re-running the numbers directly from the underlying data shows the opposite — average AQI actually fell each year from 2021 (216) to 2023 (189), then rose slightly in 2024 (196). This page reflects what the data shows, not the earlier draft's summary.

Season by season

Monsoon air is dramatically cleaner than the rest of the year

Winter is the worst season, as expected — but the size of the gap is the real story. Average AQI in monsoon months is less than half of what it is in winter. Autumn, often assumed to be a "shoulder season," is nearly as polluted as winter.

Average AQI by season, all four years combined

Monsoon (Jun–Sep) averages 108 — in the "Moderate" band — while Winter (Dec–Feb) averages 282, in the "Poor" band. This is an observed seasonal association; this dataset alone cannot establish the cause.

What actually drives the AQI number

PM10 and PM2.5 track AQI closely — Ozone barely does, and not in the expected direction

Each bar shows how closely a pollutant's daily level moves with the overall AQI score, on a scale from -1 to 1 (0 means no relationship). Particulate matter (PM10, PM2.5) and carbon monoxide move with AQI strongly. Ozone barely relates to it at all — and what little relationship exists is slightly negative.

Correlation of each pollutant with the overall AQI score, across all 1,461 days

PM10 (0.90) and PM2.5 (0.80) are doing most of the work behind Delhi's AQI number. Ozone's near-zero, slightly negative relationship (−0.16) is a real and slightly counterintuitive finding, not an error — it's kept in rather than dropped.

Testing a popular assumption

Weekends are not meaningfully cleaner than weekdays

Less weekday traffic and fewer construction sites running should, in theory, mean cleaner weekend air. In this data, weekday AQI averages 202.6 and weekend AQI averages 201.1 — a difference of under 1%.

202.6

average AQI, weekdays

201.1

average AQI, weekends

<1%

the actual difference

Whatever is driving Delhi's air quality, it isn't mainly a Monday-to-Friday activity pattern — seasonal weather effects (see above) swamp any weekday/weekend difference almost completely.

Putting it together

When do the worst days pile up?

Each circle is one condition: it's winter, the day was rated "Poor" or worse, or it's a weekday. A day can meet more than one condition at once. Click any region to see how many days fall into it, and what the average AQI was for exactly that combination.

Based on all 1,461 days · click a region, or its label, to see the count and average AQI

What this data can't tell us

Being upfront about the limits

This is Delhi only — no comparison to other Indian cities is included here, though the source project lists that as a natural next step.
Correlation, not causation — PM10 and PM2.5 tracking AQI closely is expected, since AQI is computed from pollutant levels; it isn't evidence of an independent cause.
No weather or traffic data — the seasonal pattern is almost certainly about monsoon rain clearing particulates, but this dataset doesn't include rainfall or traffic directly to confirm that mechanism.
All figures on this page were recomputed directly from the underlying daily CSV, not copied from the project's earlier documentation.

What's next Forecasting AQI a few days ahead, bringing in weather and traffic data to test the seasonal mechanism directly, and comparing Delhi to other major Indian cities.