Data from small monitoring networks is unreliable: case of Indian cities

An ambient monitoring network in a city requires a minimum of 4-5 stations to truly represent the spatial and temporal trends of emission intensities in an urban airshed. These locations must include representation from residential, commercial, industrial, traffic, and background activities. Operating less than the minimum number of ambient air monitoring stations will misrepresent the ground realities. Larger sample size is also necessary to capture the heterogeneity in the land use activities and source mixes across an urban airshed. Comparing a city represented by only one monitoring station with a city represented by at least 5 monitoring stations will lead to biased interpretations. With more (and at least minimum number of) monitors, the confidence intervals are narrower, helping in definite attribution of air quality, air quality index value, and air quality index category for a city. With more monitors, sensitivity to the type of statistical inference reduces. In this working paper, we are demonstrating methods to evaluate uncertainty associated with operating small monitoring networks to represent heterogeneity in the emission sources and land use types in Indian urban airsheds.