Stubble burning and air pollution: Two years after Supreme Court flag, no final protocol to measure burnt area of farm fires

While satellite-based farm fire counts for both the kharif and rabi seasons are routinely monitored and analysed by the Centre, there is no officially finalised assessment of the extent of farmland that has actually been burnt.

The missing metric, known to scientists as the ‘burnt area’, is critical for gauging the true scale of stubble burning and estimating its contribution to air pollution.

Yet burnt-area estimates for key agricultural states, including Punjab and Haryana, remain “unverified” and “unfinalised,” the Indian Agricultural Research Institute (IARI) told The Indian Express in an RTI response.

At present, IARI’s CREAMS (Consortium for Research on Agroecosystem Monitoring and Modeling from Space) Laboratory monitors residue-burning activity using satellite data and publishes daily farm-fire bulletins.

It compiles paddy-burning figures across states using fire-event data received from NRSC, ISRO.

The issue had come under scrutiny in the Supreme Court on November 22, 2024, when a bench of Justice A S Oka and Justice A G Masih questioned the Commission for Air Quality Management (CAQM) over discrepancies between declining farm-fire counts and rising burnt-area estimates.

After amicus curiae Aparajita Singh presented data indicating an increase in burnt area despite a decline in reported fire incidents, Justice Oka stressed the importance of measuring the total area burnt.

Story continues below this ad

“The extent of fire can be demonstrated only by a burnt area… That’s most important,” he had observed.

Nearly two years later, the methodology to officially measure burnt areas remains unfinalised.

In its RTI reply, IARI said it was “only researching” the use of satellite imagery to map paddy residue burning areas. The research involves techniques such as thresholding of spectral indices sensitive to burning signatures and machine-learning-based classification methods.

“There is yet no standardized methodology for this activity,” the institute stated.

Story continues below this ad

On the burnt-area estimates for Punjab and Haryana for 2025, as well as the preceding four years, IARI said research was continuing to evaluate different methods using multi-date, multi-spectral satellite images.

“Burnt area statistics are yet to be authenticated and finalized and is work in progress,” the institute said.

While a standardised protocol for counting active farm fires has been in place since 2021, the CAQM formed an expert committee under the NRSC in January 2024 to standardise a separate methodology for mapping the area burnt due to paddy stubble burning.

The exercise involves using Sentinel-2 satellites operated by the European Space Agency. These satellites provide optical, near-infrared and short-wave infrared imagery, which help identify burn scars on agricultural fields through changes in the way the ground reflects infrared radiation after a fire. They revisit an area every five days.

Story continues below this ad

However, the protocol is still under validation. Validation involves comparing satellite-derived maps against independent reference data to establish how accurately they identify burnt areas, including the fires they miss and areas they incorrectly classify as burnt. CAQM’s ground-truthing exercises during the 2025 kharif and 2026 rabi, as well as one underway during the current kharif season, have found differences between satellite maps and ground observations, officials said.

A senior official aware of the matter said, “During ground truthing exercises, our teams are unable to detect more than 50 per cent of satellite detected farm fires.”

Unlike active fire counts, which record visible fire events at the time a satellite passes over an area, burnt-area mapping identifies the land affected after the event. It can, therefore, capture the cumulative extent of burning, including smaller or short-duration fires that may not be detected during satellite overpasses.

In a recent research study, scientists from IARI’s Division of Agricultural Physics outlined some of the technical challenges involved in accurately mapping burnt farmland. In the intensive agricultural belts of Punjab and Haryana, standing crops, harvested straw and scorched earth often occur close to one another during the narrow harvesting window. This can make it difficult for satellite-based models to distinguish burnt fields from dry crop residue.

Story continues below this ad

Earlier research by IARI scientists has also indicated the scale of the problem. A study published in the Indian Journal of Extension Education, which examined burnt areas between 2018 and 2021, found that an average of 49.1 per cent of Punjab’s paddy area was burnt each harvesting season.

In Haryana, the corresponding figure was 17.69 per cent. The study also found that the burnt-area footprint in Haryana grew at a compound annual rate of 9.6 per cent, substantially faster than the growth in paddy cultivation.

The absence of a standardised burnt-area assessment comes amid growing evidence that active fire counts alone may not capture the full extent of stubble burning.

A November 2025 study by ISRO’s Space Applications Centre (SAC), Ahmedabad, published in the journal Current Science, documented a shift in peak farm-burning activity towards late afternoon and evening hours in Punjab and Haryana. The study used geostationary satellite imagery to examine the changing timing of agricultural activity.

Story continues below this ad

The IARI, in its RTI response, referred this reporter to the study for information on the shift in peak farm activity and said it had no further information to share on the subject.

More recently, a report by the International Forum for Environment, Sustainability and Technology (iFOREST) highlighted the mismatch between active fire counts and the area burnt in the two states.

In Haryana, while government data showed a decline in farm-fire incidents from 1,406 in 2024 to 662 in 2025, satellite-derived estimates showed that the burnt area increased from 7,117 sq km to 8,812 sq km.

In Punjab, the number of farm-fire incidents fell from 10,909 to 5,114 between 2024 and 2025. However, the estimated burnt area declined by a much smaller margin, from 23,262 sq km to 19,757 sq km.

Story continues below this ad

It found that more than 90 per cent of large fires in both states in 2024 and 2025 occurred after 3 pm, outside the overpass times of polar-orbiting satellites used for conventional fire detection. This could result in a significant number of fires going undetected.