Delhi’s Diwali pollution levels may have been higher than official estimates, says CREA

Delhi’s air pollution on Diwali may have been considerably higher than the official figures indicated, with missing readings from monitoring stations potentially pushing the city’s average PM2.5 concentration up by 49%, according to an analysis by the Centre for Research on Energy and Clean Air (CREA).

The analysis examined hourly PM2.5 data from Delhi’s 59 air quality monitoring stations between 4 pm on Nov 8 and 4 pm on Nov 9, the 24-hour period used by the Central Pollution Control Board (CPCB) for its post-Diwali air quality assessment.

Of the 1,416 station-hour observations expected during this period, 255, or around 18%, were missing. CREA conducted a sensitivity analysis by assigning these missing observations a concentration of 1,000 micrograms per cubic metre (µg/m³), the upper measurement limit cited by the CPCB for its particulate matter analysers.

Under this illustrative scenario, the pooled average PM2.5 concentration rose from 269 µg/m³ to 400 µg/m³, an increase of approximately 49%.

CREA clarified that the exercise does not establish that all missing readings were caused by pollution concentrations exceeding 1,000 µg/m³.

Instead, it demonstrates how the reported average could change if the missing observations were at the system’s stated upper measurement limit. The analysis also estimated that the peak hourly PM2.5 concentration could have reached 825 µg/m³ at 2 am on Nov 9, compared with the recorded peak of 584 µg/m³ at 1 am.

Why did the monitoring data go missing?

The data gaps followed the CPCB’s clarification that its continuous ambient air quality monitoring stations are calibrated to measure particulate matter concentrations between zero and 1,000 µg/m³. Readings beyond this range are treated as outside the system’s calibrated limit and are not included in the relevant hourly averages used for air quality calculations.

In a statement issued on Nov 10, the CPCB said that during occasional extreme pollution episodes, such as Diwali firecracker bursting, readings could exceed the instruments’ designed range. Such hours may consequently appear as data gaps and do not contribute to the 24-hour average for the pollutant at the affected station.

The pollution control board also said that it was holding internal discussions and consultations on processing readings beyond the system’s design range, indicating that the existing approach could be reviewed.

The explanation came after questions were raised over the number of monitoring stations that stopped reporting particulate matter readings during the hours when pollution levels were at their highest.

There were also reports of gaps in Delhi’s air-quality monitoring network on Diwali night, with several stations recording missing hourly readings during the pollution spike. The gaps were not limited to particulate matter. PM10 had 116 missing hourly readings, while PM2.5 had 55. Among the other pollutants, sulphur dioxide recorded 253 missing observations and ammonia 221.

DPCC readings reveal much higher pollution peaks

Data released by the Delhi Pollution Control Committee (DPCC) offered another indication of the scale of the pollution episode. Several stations recorded PM2.5 concentrations far above the CPCB system’s 1,000 µg/m³ measurement limit.

Mundka recorded 3,744 µg/m³ at 2 am on Nov 9, while Jawaharlal Nehru University recorded 3,023 µg/m³ at 4 am. The NSUT West Campus station registered 3,008 µg/m³ at 5 am. At least 11 DPCC monitoring stations recorded concentrations above 2,000 µg/m³.

All 125 station-hour readings missing from the CPCB records at DPCC-operated stations were above 1,000 µg/m³ in the latter agency’s data.

The figures highlight a difference between the two datasets during the period of peak pollution. While CPCB’s system did not include readings beyond its calibrated range in the relevant calculations, DPCC data captured concentrations that crossed that threshold.

What does this mean for Delhi’s official AQI?

The missing observations have raised questions about how extreme pollution episodes are represented in the National Air Quality Index (AQI).

Delhi’s official AQI stood at 344 at 4 pm on Nov 9, placing it in the ‘very poor’ category. CREA’s adjusted PM2.5 average of 400 µg/m³ was not itself an AQI figure. The AQI is calculated using sub-indices for multiple pollutants, and the highest sub-index determines the overall index.

The CPCB has also explained that the national AQI is capped at 500, which it describes as the highest level of health concern. This means that even when pollutant concentrations rise substantially beyond the levels represented by the index’s upper categories, the displayed AQI does not continue increasing in the same way.

CREA has argued that the episode points to the need for greater transparency in reporting extreme pollution levels. Its analysis also raises the question of whether the existing AQI categories adequately communicate the scale of short-duration pollution peaks to the public.

The issue extends beyond the calculation of a daily average. Missing readings during the hours of highest exposure can make it difficult for residents, researchers and public authorities to understand the intensity and duration of pollution episodes.

A wider challenge for air-quality monitoring

The Diwali episode has brought attention to the limits of monitoring equipment and the way automated systems process readings outside their calibrated range. While the CPCB’s clarification explains why certain observations were excluded, the gaps have also prompted questions about whether the system needs a way to flag extreme concentrations rather than leaving them out of the reported averages.

For researchers, a complete record of pollution levels is important for assessing exposure and comparing pollution episodes over time. For public authorities, reliable real-time information helps inform health advisories and emergency responses, particularly for children, older people and those with respiratory or cardiovascular conditions.

The CPCB’s indication that it is reviewing how readings beyond the system’s design range are processed could therefore have implications for the way India reports severe pollution episodes in the future.

The 49% estimate should not be read as proof that Delhi’s actual average pollution was exactly 49% higher. It is a scenario based on assigning the missing observations the system’s upper measurement limit.

But the combination of missing CPCB readings, much higher DPCC measurements and the regulator’s own explanation has exposed a significant question: how should official air-quality data account for pollution levels that exceed the limits of the instruments recording them?