Delhi’s worsening air quality has once again focused attention on stubble burning in Punjab and Haryana. But the more important finding this season is not simply that farm fires are increasing. It is that the timing of those fires may be changing faster than the satellite system commonly used to count them can capture. That creates a serious measurement problem for a city trying to understand how much pollution is arriving from outside its borders.
An analysis by Hindustan Times of air-quality, weather and satellite data found that most fires in Punjab are now being reported in the afternoon and evening, while the NASA satellite used for tracking passes over the state at about 1.30 pm each day. If fires are lit after that overpass, many may not appear in the daily count. The result is not necessarily that the fires are invisible, but that the principal monitoring window may be poorly aligned with the time at which burning is taking place.
The issue matters because satellite fire counts are frequently used to describe the scale and timing of crop burning. They inform public discussion, administrative responses and explanations of Delhi’s seasonal pollution. If the counting system misses a significant share of fires, the resulting picture can understate the problem and make comparisons across years more difficult.
The air-quality numbers cited in the report show why the question has immediate significance. Delhi’s air quality index was recorded at 117 on Saturday, rose to 261 on Sunday and reached 324 by 4 pm on Monday. This was the first time in the season that the city’s air entered the ‘very poor’ category, according to the supplied report.
These changes occurred alongside a rise in attention to farm fires in neighbouring states. However, the available evidence does not establish that stubble burning alone caused Delhi’s deterioration. The HT analysis also noted the role of moisture and weather conditions and did not assign a precise share of Delhi’s pollution to crop-burning smoke. That distinction is important: the direction of incoming winds can show that smoke from burning regions may have travelled towards the capital, but it cannot by itself quantify the contribution to the city’s AQI.
The central weakness in the current monitoring picture is the mismatch between fire timing and satellite timing. NASA’s satellite passes over Punjab at approximately 1.30 pm. The 2025 stubble-burning status report by the New Delhi-based research group iFOREST, cited in the analysis, found that more than 90 per cent of large fires in Punjab during 2024 and 2025 occurred after 3 pm. In 2021, the corresponding share was reported to be only 3 per cent.
That change has two implications. First, a count based on an early-afternoon satellite pass may increasingly record only a portion of the day’s fires. Second, year-on-year comparisons may become misleading if the observation conditions remain unchanged while the timing of burning shifts. A lower satellite count would not necessarily mean fewer fires; it could mean that more fires are taking place outside the period when the satellite is observing the area.
This is a familiar challenge in urban environmental governance: the indicator is treated as a direct measure of reality even though it is produced through a particular method, at a particular time, with particular technical limitations. A satellite fire count is not the same as a physical census of every field fire. It is an observation generated by the satellite’s orbit, resolution, timing, cloud conditions and ability to detect heat.
The comparison with India’s INSAT-3DS satellite illustrates why no single dataset can provide the complete picture. Unlike the NASA satellite, INSAT-3DS can keep scanning the same broad area through the day. That gives it an advantage when fires occur in the afternoon or evening. But the report also notes that its imagery has lower detail and is more likely to detect large or intense fires than smaller or less intense ones.
The INSAT-3DS data cited by HT recorded very few evening fires in Punjab and Haryana between 15 and 28 October. The numbers then rose steadily: 17 fires were recorded on 29 October and 90 on 1 November. Even those figures may be below the actual number of incidents because the satellite is more effective at identifying larger or hotter fires. The contrast demonstrates the trade-off between observation frequency and image detail. A satellite that watches more often may not see every small fire clearly, while a higher-resolution system may miss fires that occur after it has passed.
The NASA data themselves recorded 171 fires on 1 November, the highest figure of the season at that point. Yet that number should not automatically be read as the full extent of burning across Punjab and Haryana. The report’s findings suggest that the count is best understood as a monitored minimum or an observation shaped by the satellite’s schedule, rather than a complete inventory of all farm fires.
Weather provides the second part of the explanation for Delhi’s worsening conditions. HT’s analysis found that winds at higher elevations were moving from the northwest on 31 October and 1 November. Punjab and Haryana lie in that direction from Delhi and contain the areas where crop burning was being reported. According to the weather model cited in the report, the air reaching Delhi on 1 November had spent most of the previous three days over Punjab and Haryana.
This does not turn wind direction into proof of a single pollution source. Delhi’s air quality is affected by several factors, and the supplied report does not provide a precise source-apportionment figure for the period. But the trajectory is relevant because it establishes a plausible transport pathway at the same time that fire detections were rising. The evidence therefore supports a more careful conclusion: smoke from farm fires may have contributed to Delhi’s pollution under those conditions, but the available data do not determine its exact share.
The institutional issue is that different agencies and public discussions may rely on different definitions of what counts as a fire. A system that detects only large or intense events will produce a different total from one that observes a region repeatedly but at lower resolution. A satellite passing once a day will produce a different record from one capable of scanning the same area throughout the day. These are not minor technical details. They influence how governments assess enforcement, how researchers evaluate seasonal trends and how the public interprets changes in Delhi’s air.
The evidence also shows why timing must be treated as part of the data, not merely as background information. The shift from 3 per cent of large fires occurring after 3 pm in 2021 to more than 90 per cent in 2024 and 2025, as reported by iFOREST, means that a monitoring arrangement designed around an earlier overpass may be losing visibility precisely when burning is most common. The change does not by itself explain why the timing shifted, and the supplied material does not establish the reasons. It does, however, show that the surveillance method and the behaviour being monitored are no longer closely aligned.
For Delhi, this creates a governance problem beyond the annual argument over the contribution of stubble burning. The city needs reliable information not only about whether fires are occurring, but also about when they occur, where the smoke travels and how that transport overlaps with local weather and pollution conditions. Without that fuller picture, public responses risk being based on incomplete counts or on comparisons that do not account for changes in observation conditions.
The immediate evidence confirms three facts. Delhi’s AQI worsened sharply over the reported period; winds reaching the capital had travelled over Punjab and Haryana; and fire detections increased in satellite data even as the principal NASA observation window appears to have missed many later-day fires. What remains uncertain is the precise contribution of those fires to Delhi’s pollution and the full number of burning incidents. The next meaningful step in understanding the season will be to reconcile datasets with different observation times, image resolutions and detection thresholds rather than treating any single satellite count as the complete story.
