An unsigned report from IIT Kanpur, shared by Delhi environment minister Manjinder Singh Sirsa on Tuesday, contends that the cloud seeding experiment in Delhi was somewhat successful, citing rainfall data and air quality improvements at three locations.
But as HT’s report on the experiment, published on October 29, pointed out, the report cites rain data from Windy.com, a website that displays modelled estimates rather than measured rainfall. And the claim of cleaner air suffers from equally serious methodological failures.
What IIT Kanpur claimed
The report stated: “The PM 2.5 was 221, 230, and 229 reported from Mayur Vihar, Carol (Karol) Bagh, and Burari, respectively, before cloud seeding, which got reduced to 207, 206, and 203, respectively, after the first seeding. Similarly, PM 10 was 207, 206, 209, which got reduced to 177, 163, 177 at Mayur Vihar, Carol Bagh, and Burari, respectively. Given that winds were negligible, one possible explanation is that the denser moisture content created due to seeding particles has helped in settling down a portion of these particles, which translated to these reductions.”
This is not a scientifically sound way of assessing or reporting air quality improvement. Here is why.
One: No units specified
PM2.5 and PM10 refer to particles smaller than 2.5 and 10 micrometres in diameter suspended in air. The scientific convention is to report their concentrations—the mass of pollutants in a given volume of air. India’s National Ambient Air Quality Standards, for instance, specify limits in micrograms per cubic metre.
Air quality data is also commonly presented using a unitless index that translates concentrations into health impact categories. The IIT Kanpur report cites no units at all, making it impossible to determine whether it is reporting concentrations or index values.
This ambiguity matters. While India has an official conversion formula published by the Central Pollution Control Board (CPCB), the formulae differ for PM2.5 and PM10. Without knowing what the numbers represent, the claimed “improvement” cannot be properly evaluated.
Two: No time period stated
The report does not specify the averaging period for its measurements or when exactly they were taken. Are these 15-minute averages, hourly concentrations, or 24-hour rolling averages? Without this information, it is impossible to determine how long any improvement lasted—or whether the measurements are even comparable.
Three: Natural daily patterns ignored
The experiment was conducted in two sorties, with seeding taking place roughly between 2 PM and 5 PM. This timing coincides precisely with the period when Delhi’s air naturally improves each afternoon.
The air quality data cites three locations, although the report states that data was collected from 20 locations. However, IIT Kanpur director Manindra Agarwal on Wednesday told HT in Lucknow that his team set up “15 stations in different parts of Delhi” but did not give details about their type and location.
At least one of the locations they cited, Burari Crossing, is covered by an India Meteorological Department air quality station. Hourly data from this station shows that air quality consistently improves between noon and 6pm, then deteriorates in the late evening and night.
Comparing PM2.5 and PM10 concentrations on October 28 with the three previous days reveals an identical pattern: the 2pm to 5pm window is consistently the cleanest part of the day. (See charts 1A, 1B)
This makes it impossible to isolate any improvement attributable to cloud seeding from the natural daily cycle.
