Assessment of Changes in Air Quality in Indian Cities Since the Launch of the National Clean Air Programme

HEI has published a new report “Assessment of Changes in Air Quality in Indian Cities Since the Launch of the National Clean Air Programme.” The report evaluates trends for particulate matter (PM₁₀ and PM_2.5) across NCAP cities since the launch of the program. Using station-level regulatory data from 2017–2024, the analysis assessed data availability and completeness, particulate matter trends, and the influence of seasonal and meteorological factors. The report also offers a framework for understanding local variation in air quality and tracking air quality progress over time.

Key findings from the report include: 

• The air quality monitoring infrastructure has expanded rapidly since NCAP was launched in 2019. Between 2017 and 2024, the number of PM_2.5 monitoring stations grew by 344% and PM₁₀ stations grew by 462%.
• When compared to the 2017 baseline, more than half of the stations show declining trends for both PM₁₀ and PM_2.5 levels. 
• While there is a declining trend, it is difficult to attribute the improvements solely to NCAP, as several overlapping policy measures including Bharat Stage-VI fuel norms and expansion of access to LPG for cooking have occurred over a similar timeframe. Furthermore, funding timelines have varied across cities complicating the attribution of changes specifically to NCAP. 
• When considering data availability, annual thresholds (e.g., 75% data availability) can be insufficient, as they can miss seasonal and peak pollution episodes.
• When considering changes in air quality, on its own, the absolute change between two years does not provide a reasonable estimate for improvements in air quality. In particular, adjusting for season and weather help reveal the true trends more clearly.  
• Rolling averages can provide more reliable signals of sustained change in air quality over time.
• Station-level averages offer better representation of local variation than citywide averages. 

This work was supported by Coefficient Giving.