Rice in India

The crop that feeds the country, floods a tenth of its farmland every monsoon, and, for six weeks each autumn, sets the north's air on fire.

Simply put

Rice likes its feet wet. A paddy is a field turned into a shallow pond, the seedlings pushed into mud by hand or machine, the water kept standing for most of four months. Under that water the soil runs out of oxygen and microbes make methane instead of carbon dioxide; the plant itself is the chimney. When the grain is ripe the field is drained and cut. What is left standing, the stubble and the loose straw, has to go somewhere before the next crop, and in the north the next crop is wheat and the gap is three weeks. The quickest way to clear a field is a match.

The seasons

India grows rice in three seasons, and most states grow it in one. Kharif is the monsoon crop: transplanted in June and July, cut from October. It is nearly all of Punjab, Haryana and Uttar Pradesh's rice, and the source of the autumn fires. In the east it is called aman and runs later, Bengal cutting into December; in Tamil Nadu the main crop, samba, is planted in August and cut in January. Rabi or boro is the winter crop of the east and south, sown around the new year and cut in April and May, almost all of it irrigated. Summer or aus rice fills the gap before the monsoon in a few states. The map's season colours come from these calendars; the windows for every state are on the Data page.

Punjab, with 4.1 million ha, grows one crop of rice a year and then wheat. West Bengal, with 2.3 million ha, grows rice on much of the same land twice. A field that grows rice twice is two floods, two harvests and two lots of straw: that is why Dhaan counts hectare-seasons, not just hectares.

Water and methane

Rice is the one major crop grown in standing water, and that is where its climate cost comes from. In a flooded soil, bacteria and archaea that only live without oxygen break down straw, roots and root exudates and release methane; up to nine-tenths of it leaves through the rice plant's hollow stems. How much depends on how long the field stays flooded and how often it is drained. The IPCC's factors say a South Asian paddy kept continuously flooded emits about 0.85 kg of methane per hectare per day; drain it once mid-season and that falls by about a third (0.71×); drain it several times and it halves (0.55×). A rainfed field that dries on its own sits at about half (0.54×).

That range is the point of the practices people are now paid to adopt (below): the same field, the same crop, half the methane, and, in Punjab, a great deal less groundwater pumped. The Method page walks the arithmetic; the map shows every district's range.

The straw

A tonne of rice leaves about 1.5 tonnes of straw. Across India that is around 207 million tonnes a year. In most of the country it is fodder, thatch, packing, fuel or simply left to rot in the flooded field before the next crop. In Punjab and Haryana it is a problem, for three reasons that arrived together: combine harvesters, which cut high and leave a standing stubble no animal wants; the wheat that must be sown within weeks; and a 2009 Punjab law that pushed rice transplanting later into June to save groundwater, which pushed the harvest later into a colder, stiller November.

Why it burns

Between the middle of October and the end of November, a farmer in Sangrur or Patiala has a field of stubble and about twenty days to get wheat into it. Machines that can sow wheat through the stubble exist (the Happy Seeder, the super seeder) and are subsidised; balers and biomass plants exist and are far away; a match is free and takes an afternoon. So the fields burn, tens of thousands of them, and the smoke rides a slow north-westerly into the plain. The burning tab shows each day's fires, the wind, and where the air over them goes.

The east burns too, later and less densely: Bengal's aman stubble in December, parts of Bihar and eastern Uttar Pradesh in November. The map shows it because the rice mask covers the country, not only the two states the news is about.

Why the counts are wrong

Every November a number is quoted: fires in Punjab this season, down so many percent. It comes from the ICAR-CREAMS laboratory and the Punjab Remote Sensing Centre, counted from NASA's satellites, and it is honest as far as it goes. The catch is when the satellites look. The polar-orbiting satellites that carry the fire sensors cross north India between about 10:30 and 13:30, and again around 01:30. A fire lit at five in the evening and burnt out by nine is invisible to all of them.

In 2025 the research group iFOREST checked, using a geostationary weather satellite that photographs India every 15 minutes and Sentinel-2's pictures of the blackened fields. Over 90% of Punjab's large fires in 2024 and 2025 were lit after 3 pm, against 3% in 2021: the burning moved to after the satellite. The government reported a 90% fall in fires; the land actually burned fell by a quarter to a third. Dhaan's burning tab publishes the 13:30 count so it can be checked against the official one, and says on every page that the count is not the fire. The measures that do not miss the evening, fields burned and hourly detections, are the next things to connect.

What changes it

The alternatives are known and most are old. Alternate wetting and drying (AWD) lets the field dry to a hand's depth below the surface before re-flooding, several times a season: less water, less methane, no loss of yield when done right. Direct-seeded rice (DSR) skips the nursery and the flooded transplanting altogether; Punjab pays farmers per acre to try it. Short-duration varieties such as PR 126 ripen weeks earlier, opening the gap before wheat and shrinking the reason to burn. In-situ management sows wheat through the stubble; ex-situ bales it for power plants, pellets and board. Bio-decomposers sprayed on the stubble rot it in the field. Carbon-credit programmes now pay for AWD and DSR by the hectare and register where they run.

What Dhaan will show is adoption: the hectares each programme reports, and, from radar, the fields that were not kept flooded. The two will not agree everywhere, and the difference is itself a finding. Not this build; see what is not connected yet.

The air downwind

Smoke from a field is mostly fine particles, PM2.5, with carbon monoxide and black carbon. Where it goes is a matter of wind and of the mixing depth of the lowest air, which in a still November morning over the Gangetic plain can be a few hundred metres. The burning tab follows the air over each day's fires for 48 hours on Maplore's own ECMWF wind and marks the cities it crosses, then shows what CPCB's stations in those cities read. It does not, and will not, put a percentage on how much of Delhi's air is Punjab's smoke: that needs a chemical transport model, and every published estimate of it is contested. Reach and readings side by side, and the reader decides.

Who works on it

• ICAR-CREAMS (Consortium for Research on Agroecosystem Monitoring and Modeling from Space, IARI, New Delhi): the daily official fire bulletins.
• Punjab Remote Sensing Centre, Ludhiana, and the Punjab Pollution Control Board: the state's counts and enforcement.
• ICAR-NRRI, Cuttack, and the International Rice Research Institute: varieties, AWD, DSR, crop calendars.
• Punjab Agricultural University, Ludhiana: PR 126 and the Happy Seeder.
• iFOREST: the 2025 analysis of when fires are lit and how much land burns.
• The Commission for Air Quality Management (CAQM): the NCR's authority on the smoke.
• NASA FIRMS, ECMWF, ESA Copernicus: the satellites and the wind.