Bengaluru: Last week, the Union government said in parliament that farm fires – which are often cited as the main cause of air pollution in Delhi-NCR and other parts of north India during the winter season – decreased by around 90% in the states of Punjab and Haryana this year when compared to 2022. 

However,  a new analysis by the International Forum for Environment, Sustainability and Technology (iFOREST) suggests that this may only be the result of using wrong methods to quantify fires and stubble burning.

Their analysis of satellite imagery shows that the timing of farm fires has shifted since 2019: farmers in Punjab and Haryana now set fire to crop residue or stubble after 3 PM during the winter months. The government’s existing system (which monitors only a three-hour window from 10:30 PM to 1:30 PM daily) misses this completely.

The report also noted that another indicator of stubble burning, i.e. the area of burnt cropland in these states, has decreased – though only by about 25-37% – this year when compared to 2022.

This shows that policy implementations may be working on the ground, the report noted.

India is undercounting farm fires

Stubble burning or crop residue burning is the phenomenon wherein farmers set fire to leftover stubble after paddy and wheat harvests. These farm fires that occur across many northern Indian states, primarily Punjab and Haryana, during the winter months have repeatedly been cited as the major source of ‘episodic’ pollution in the Indo-Gangetic plains including Delhi-NCR by studies such as this one.

Last week, the Union government said in parliament that farm fires this year (from September 15 to November 30) had decreased by around 90% in the states of Punjab and Haryana, when compared to 2022.

The official data on farm fires is published by several government platforms. Portals such as the Decision Support System for Air Quality Management in Delhi – under the aegis of the Ministry of Earth Sciences and the Indian Institute of Tropical Meteorology – rely on daily imagery from the VIIRS (Visible Infrared Imaging Radiometer Suite) satellite to quantify the number of active fires in states. 

Similarly, another government portal, the Consortium for Research on Agroecosystem Monitoring and Modeling from Space (CREAMS) of the Indian Agricultural Research Institute (IARI) relies on satellite imagery from both the VIIRS satellite and the MODIS satellite (both are polar-orbiting satellites that pass over India between 10:30 AM and 1:30 PM daily). For instance, the CREAMS bulletin listed 37 burning incidents across six states on October 12 this year: four in Haryana, 10 in Punjab, 17 in Uttar Pradesh, six in Madhya Pradesh and none in Delhi and Rajasthan.

Researchers at iFOREST, an NGO that works on sustainability and energy among other things, looked deeper. They compared data from different satellites: MODIS (Terra and Aqua satellites), VIIRS (Suomi-NPP satellite), Sentinel-2 Multispectral Instrument (MSI) burnt-area mapping, and 15-minute geostationary observations from the Spinning Enhanced Visible and Infrared Imager (SEVIRI) on Meteosat 8 and 9.

Their results, listed in the Stubble Burning Status Report 2025, released on December 8, found that farm fires now occur after 3 PM, and peak at around 5 PM everyday.

Specifically, the SEVIRI satellite data (which provides satellite images every 15 minutes), showed that in Punjab, over 90% of large fires in 2024 and 2025 occurred after 3 PM, compared to just 3% in 2021. In Haryana, most large fires have occurred after 3 PM since 2019. 

This suggests that India’s official government data undercounts farm fires. It also suggests that the government’s claim that farm fires in the two states had declined by a huge 90% was “largely a result of limitations in the current monitoring system”, the iFOREST report said.

Speaking at a webinar during the launch of the report on December 8, Chandra Bhushan, CEO, iFOREST, said that this was “incontrovertible evidence that India’s current stubble-burning monitoring system is structurally misaligned with ground realities”.

“Farmers have shifted burning to the late afternoon, while our monitoring relies on satellites that capture active fires only during a narrow time window – 10:30 AM to 1:30 PM. The result is a massive underestimation of fires, emissions, and their contribution to air pollution in Delhi. We urgently need to overhaul the system,” he said.

Area of burnt cropland has decreased

The iFOREST report also shows that the area of burnt cropland (through satellite imagery from Sentinel-2) has indeed reduced over the years.

Burnt area during the kharif cropping season (crops fed by the southwest monsoon which are harvested by September-October) in Punjab declined from a peak of 31,447 km² in 2022, to about 20,000 km² in 2025. This is a 37% reduction. 

Similarly, in Haryana, burnt area has dipped from 11,633 km² in 2019 to 8,812 km² in 2025 – a 25% reduction.

“This is good news. Something is working on the ground,” Bhushan said at the report launch on December 8. He added that this indicates that in-situ and ex-situ stubble-management practices are being adopted.

“But this is not the time to become complacent,” Bhushan commented. “Even in 2025, close to 30,000 km² of paddy fields were burnt in Punjab and Haryana, making them a major source of air-quality degradation in Delhi-NCR and the wider Indo-Gangetic region.”

Moreover, burnt crop area is also now emerging as a feature in other states too, such as Madhya Pradesh and Uttar Pradesh, the report noted; polar-orbiting satellites show that there are now more numbers of active fires in these states as well. 

“Stubble burning is now becoming a disease,” Bhushan said at the report launch. “It was largely centred in Punjab and Haryana but we are now seeing it in Madhya Pradesh and Uttar Pradesh; we also have numbers from Maharashtra and Orissa. It is now a disease which is catching the entire country.”

Therefore we have to look at stubble burning across the country, Bhushan said.

Government systems need to be updated 

At the report launch, Bhushan also said that government systems should be using a combination of active fire counts and burnt area to estimate stubble burning, as this provides a more reliable picture.

India should revise its monitoring protocols and publish data on burnt area through portals such as CREAMS and not just rely on active fire counts alone, the iFOREST report recommended.

“We cannot manage what we do not measure accurately,” said Ishaan Kochhar, Programme Lead, iFOREST. “Policy decisions are currently being shaped by incomplete information. To solve the stubble-burning problem in the Indo-Gangetic Plain, the government must urgently reform the monitoring protocol to integrate burnt-area mapping and geostationary data. We also need to expand our focus beyond Punjab and Haryana to emerging hotspots in Uttar Pradesh and Madhya Pradesh.”

The report also said that the existing Decision Support System (DSS) developed by the Indian Institute of Tropical Meteorology for Delhi’s air-quality management must revise its methodology to correctly quantify the contribution of stubble burning. Currently, the DSS relies on satellite imagery from VIIRS alone to estimate the contribution of farm fires to Delhi’s pollution. 

In ‘testing mode’

Burnt area of cropland is a much better estimate of stubble burning than active fire counts and it would be better if government portals switched to this indicator, agreed Sachin Ghude, scientist at the IITM Pune who heads the DSS and Air Quality Warning and Integrated Decision Support System for Emissions (AIRWISE).

“It is very good news that there was around 30% decrease in burnt area [in Punjab and Haryana], this is good news in terms of policy implementation,” he told The Wire over a telephonic interview, commenting on the findings of the iFOREST report. 

However, geostationary satellites that provide burnt area estimates have some drawbacks and the IITM is still ironing out ways to work this into the DSS, he said. This is still in ‘testing mode’, he said.

For instance, while geostationary satellites can indeed give more frequent data (some even capture images every ten minutes) they are very far above — around 30,000 kilometers away. 

The spatial resolution of data from geostationary satellites is thus coarse (square kilometers, versus square meters from polar-orbiting satellites) — something iFOREST researchers also noted. Thermal signals from fires would be weak at that resolution, so this is a challenge. Research on how to isolate these weak signals through algorithms is still ongoing, Ghude said.

Moreover, there is also a few days’ lag in obtaining burnt area data from these satellites — while the DSS requires quick and real-time data, Ghude said, explaining why the DSS still uses only data from polar-orbiting satellites such as VIIRS.

Since last year, the IITM team has been trying to develop a method to estimate the contribution of stubble burning to air pollution using chemical markers (such as potassium levels from stubble burning through on-ground field data), he added.

“We are trying to see how we can use this data to derive biomass burning estimates,” Ghude told The Wire.

Estimates of biomass (or crops burnt), along with crop production estimates can in turn provide an estimate of the emissions that affect air quality in the Indo-Gangetic plains, including the Delhi-NCR.

The Wire is now on WhatsApp. Follow our channel for sharp analysis and opinions on the latest developments.