Complimentary life-shortening
If Delhi's air met the WHO guideline, residents would live about 8.2 years longer. Haryana: 5.3 years. UP: 5 years. Year-round dirty air does this, and the festival is its loudest act.
Pollution Fest
Every October, north India sets its fields on fire and its governments set up a committee. We use NASA satellites to count the fires so you don't have to take anyone's word for it. What the smoke does · Why the farmer isn't the villain · What Punjab could grow · Explore the fires · Glossary
From mid-October to late November, farm smoke rolls across the Indo-Gangetic plain, meets cold still air, and settles in for the season. Delhi's official AQI (CPCB 24-hour average) was 376 at 4 PM on Diwali, Sun 8 Nov 2026, and 344 at 4 PM on Mon 9 Nov, with 15 stations in the "severe" band. On Diwali night, PM2.5 readings at several monitoring stations conveniently went blank. Here's this year's programme.
If Delhi's air met the WHO guideline, residents would live about 8.2 years longer. Haryana: 5.3 years. UP: 5 years. Year-round dirty air does this, and the festival is its loudest act.
On the worst days, farm fires made up 38% of Delhi's PM2.5 in 2024 and 22% in 2025. Across November it averages less (20% in 2024, 7% in 2025), but it turns bad air into a public-health emergency overnight.
When the fires stop, Delhi's December smog often gets worse, thanks to its own vehicles, dust and industry. Stubble is the guest who gets blamed for the whole party, then leaves early while the party carries on.
Schools go online. Construction stops. This year: a 15-day construction ban across NCR from 5 Nov, and half of office staff in Noida and Gurugram sent home. Construction workers get the holiday too. The pay is another matter.
Every tonne of straw burnt sends roughly 5.5 kg of nitrogen, 25 kg of potash, 1.2 kg of sulphur and ~400 kg of carbon up in smoke. Farmers then buy fertiliser to put it back. Truly a circular economy.
Delhi gets the headlines, but the villages next to the burning fields breathe it first and worst, with no smog towers and no TV debates. They're also where the matchstick gets lit, which tells you how little choice is involved.
Detections from the same satellite (VIIRS on NOAA-20), 15 Sep to today's date, every year. Same satellite, same window, so it's a fair comparison.
Change from 2024 to 2025: fires officially counted vs land actually burnt (satellite burn scars). Source: iFOREST, via Indian Express, 12 Nov 2026.
In Haryana, the fire count halved while the burnt area grew by a quarter. That's not a pollution fix. That's a scheduling fix.
Every satellite that can see a farm fire or its smoke over these six states, Indian and foreign, and whether this page uses it. Times are IST over Punjab: measured from our own data for the satellites we use, approximate for the rest. "Could use" means the data is public; most of it needs an account and some file wrangling.
| Satellite | Over Punjab (IST) | What it sees | Status here |
|---|---|---|---|
| NOAA-20 VIIRS USA (NOAA/NASA) | 12:50–2:40 PM, and 1–3 AM | Fires in 375 m pixels. The workhorse of the official count. | Using via NASA FIRMS |
| NOAA-21 VIIRS USA (NOAA/NASA) | 12:50–2:30 PM, and 1–3 AM | Same instrument as NOAA-20, about 50 minutes apart in the same orbit. | Using via FIRMS |
| Suomi NPP VIIRS USA (NASA/NOAA) | 1–2:40 PM | Same instrument again; the oldest of the three. | Ended 2 Nov 2026, mid-season |
| Aqua MODIS USA (NASA), launched 2002 | 3:30–4:30 PM, and drifting later | Fires in 1 km pixels. Our only polar eye on the late afternoon, thanks to orbit drift. | Using. NASA plans to start shutting it down in late 2026 or early 2027. |
| Terra MODIS USA (NASA), launched 1999 | About 9–10 AM and 8:30–9 PM | Fires in 1 km pixels, at hours when hardly anything burns: fewer than 40 detections in and around Punjab all season. | Using. Same shutdown plan as Aqua. |
| INSAT-3DS, INSAT-3DR India (ISRO), geostationary | Every 30 minutes | A fire product, but in 4 km pixels, too coarse for most single fields: "existing geostationary satellites are unable to pick up smaller fires" (Shaastra, IIT Madras). | Could use with a MOSDAC account |
| Meteosat-9 SEVIRI Europe (EUMETSAT), geostationary | Every 15 minutes | Fires in 3 km pixels at best. ISRO's own 2025 study used Meteosat data to show peak burning moving to around 5 PM (Tribune). | Could use via LSA SAF (registration) |
| GK-2A AMI South Korea (KMA), geostationary | Every 10 minutes | Fires in about 4 km pixels. A NASA researcher has published hourly Punjab and Haryana fire detections from it for 2019–2025 (dataset, paper). | Could use: free, CC BY 4.0, past seasons only |
| Sentinel-3A/3B/3C SLSTR Europe (ESA/EUMETSAT) | About 10:30 AM and 10:30 PM | Fire radiative power in 1 km pixels, with a night pass too. | Could use (Copernicus, free account) |
| FY-3D MERSI-II China (CMA) | About 2:30 PM | Fires in 1 km pixels (ESSD 2022). Same lunch-hour shift as VIIRS, so no help with the evening. | Could use (registration) |
| Sentinel-2, Landsat 8/9 Europe (ESA); USA (NASA/USGS) | About 10:30–11 AM, every few days | Burn scars at 10–30 m. Scars don't care what time the fire was lit. (FIRMS only publishes Landsat fire points for the US and Canada.) | Could use: on our burnt-area roadmap |
| Himawari-9 AHI Japan (JMA) | Every 10 minutes | Its wildfire product stops at 80°E. Punjab is at about 75°E. | Can't: Punjab is off the edge |
| FireSat USA (Earth Fire Alliance, Google, Muon Space) | At least twice a day, once complete | Fires "as small as five-by-five meters". First three satellites reached orbit in July 2026. | Not yet: India coverage and data terms unknown |
| Satellite | Over Punjab (IST) | What it sees | Status here |
|---|---|---|---|
| Sentinel-5P TROPOMI Europe (ESA/Copernicus) | About 2 PM, daily | NO₂, carbon monoxide and smoke aerosol in 3.5 × 5.5 km pixels. Good for mapping the plume. | Could use (free account) |
| NOAA-20, NOAA-21 VIIRS USA (NOAA) | Same passes as the fire data | Aerosol optical depth and a smoke/dust mask in 750 m pixels, posted publicly within about half an hour. | Could use: no account needed |
| INSAT-3DS, INSAT-3DR India (ISRO), geostationary | Every 30 minutes, daytime | Aerosol optical depth at about 10 km (method). | Could use with a MOSDAC account |
| EOS-06 (Oceansat-3) OCM-3 India (ISRO) | Daytime pass | Aerosol optical depth at 1 km (ISRO). | Could use with a MOSDAC account |
| GEMS on GK-2B South Korea (NIER), geostationary | 8 times a day | NO₂ and aerosols at 7 × 8 km. Its view starts at 75°E, so Delhi is in and Punjab is only partly in. | Partly: licence bars modified use |
| MetOp-B/C IASI, GOME-2 Europe (EUMETSAT) | About 10 AM | Carbon monoxide and NO₂ in coarse pixels (WMO OSCAR). | Could use |
| Aura OMI USA (NASA) with the Netherlands and Finland | Early afternoon | NO₂ and aerosols at 13 × 24 km. Power is expected to run short by mid-2028. | Could use, but TROPOMI is sharper |
| MOPITT on Terra, CALIPSO Canada/USA; USA/France | – | Carbon monoxide; vertical smoke profiles. | Ended: MOPITT Feb 2025, CALIPSO Aug 2023 |
| TEMPO, Sentinel-4 USA; Europe | Hourly, elsewhere | Hourly pollution maps, for North America and Europe. | Can't: wrong continent |
| An Indian air-quality satellite | – | Doesn't exist yet. ISRO's aerosol nanosatellite NEMO-AM is listed as "Planned". | Under consideration |
Smoke readings measure the whole column of air, not what you breathe at street level, and they need a clear daytime sky. Winter haze and fog leave gaps.
Pick any window. Every chart on this page except the season comparison follows it, and the URL updates, so you can send the exact view to your MLA.
| # | District | State | Detections | FRP (MW) |
|---|
Ruling parties since 2015, by government. Every major party in the region has held power somewhere during the smoke years. Pick your favourite to blame; they've all had a turn.
Coalition partners are shown under the lead party. Emoji stand in for election symbols.
Who did what, and what came of it. Tags show who acted (tap one to check who was in charge): the Union government or its agencies, a State government, or a Court.
Keep this open during the next press conference. Tap a square each time you hear one. Get five in a row to win clean air (offer not valid in NCR). Every square really happened: hit ↗ to see who said or did it.
It's late October in Sangrur. The paddy is cut. Wheat has to be sown in about two weeks or the yield suffers. The field is covered in loose straw. Our farmer has these options:
| Option | Cost | Time | The catch |
|---|---|---|---|
| 🌱 Rent a seeder to sow straight into the straw | Thousands of rupees an acre. Punjab reckoned ₹2,500/acre of compensation was fair. | Whenever the machine turns up | 3.54 lakh subsidised machines exist across four states, and every farmer needs one in the same fortnight. |
| 📦 Bale it and sell it | Baling costs money | Days | To whom? The two big straw-to-ethanol plants (Panipat, running; Bathinda, still under construction) are built to eat 2 lakh tonnes a year each: a rounding error of the straw mountain. |
| 🧪 Spray bio-decomposer | Cheap | ~20 days to break down | Wheat needed sowing yesterday. |
| 🙏 Wait for the promised incentive | Free | Haryana: "payment delays". Punjab's ₹2,500/acre ask: the Centre said no. | Your landlord and your bank still want their money on time. |
| 🔥 Matchstick | ₹1 | One evening | Up to ₹30,000 fine, if caught. Most satellites have clocked out by 3 PM. |
Put any of us in those boots and we'd be lighting the match too. Burning isn't a character flaw. It's the cheapest answer to a problem built by people who will never stand in that field.
A fine fits in a tweet and makes the problem the farmer's fault. A fix takes years, money and cooperation between rivals, and the credit might go to whoever wins the next election. So, year after year, we get the tweet.
First, a quick look at the national grain strategy, as currently practised.
So: Punjab drains its aquifers to grow rice the country already has too much of, while India spends lakhs of crores importing oil and pulses that Punjab could grow on far less water. Somewhere, a committee is calling this "food security".
| Crop | Water vs paddy | Where it fits | Returns & risk today | Nation needs it? |
|---|---|---|---|---|
| 🌽 Kharif maize | Much less | Most of Punjab, except waterlogged pockets | Roughly equal to paddy at best; no assured buyer; needs drying | ✅ Feed & ethanol demand rising |
| 🫘 Summer moong between wheat and paddy | Very little | Statewide, a ~60-day catch crop | Short crop; market price is the risk | ✅✅ Record pulse imports; also fixes nitrogen |
| ☁️ Cotton | Less | South-west Malwa: Bathinda, Mansa, Fazilka | Lower and volatile; pink bollworm | ✅ Textiles |
| 🌼 Mustard & oilseeds | Less | Some rabi land now under wheat | Decent; price support patchy | ✅✅ The biggest import bill |
| 🍊 Kinnow, vegetables, poplar | Mostly less | Abohar–Fazilka (kinnow), near cities | Can pay well; needs cold chains and buyers | ➖ Niche, not a replacement |
| 🌾 Smarter paddy where it stays | PR-126: ~25% less; direct seeding: 10–20% less | Statewide | Already 33% of paddy area (2023) | ✅ Less water and a wider gap before wheat, so less burning |
PAU on PR-126: it "requires less water, has less straw load while ensuring more window period between rice harvest and wheat sowing". The shorter the paddy season, the less the farmer needs a matchstick.
The pattern isn't complicated: farmers grow whatever the government guarantees to buy. Today that's paddy. So we get paddy, sinking aquifers and smoke, and then an emergency meeting about the smoke. Fixing it means guaranteeing a market for pulses, oilseeds and maize, and that costs real money for years. A fine costs nothing.
Data: NASA FIRMS (VIIRS NOAA-20, NOAA-21, S-NPP; MODIS Aqua/Terra), standard-processing archives where available, else near-real-time. Times in IST. Updated every 3 hours.
Terms you'll meet on this page. Dotted-underlined words show the same definitions when you hover or tap them.