Ambient air pollution is a major environmental health risk in many cities, leading to millions of premature deaths per year globally. The regulation and management of urban air pollution is currently based on two pillars. On the one hand, a limited number of fixed site monitoring stations provides highly accurate measurement data but cannot resolve the high spatial pollution heterogeneity encountered in urban areas. On the other hand, computer models provide high-resolution concentration maps, but are bound by static emission inventories and lower time resolutions. The rise and nature of citizen-based applications, which are aimed at empowering individuals to make informed choices for their health, pose expanded requirements on air quality monitoring and modelling not covered by the pre...[ Read more ]

Ambient air pollution is a major environmental health risk in many cities, leading to millions of premature deaths per year globally. The regulation and management of urban air pollution is currently based on two pillars. On the one hand, a limited number of fixed site monitoring stations provides highly accurate measurement data but cannot resolve the high spatial pollution heterogeneity encountered in urban areas. On the other hand, computer models provide high-resolution concentration maps, but are bound by static emission inventories and lower time resolutions. The rise and nature of citizen-based applications, which are aimed at empowering individuals to make informed choices for their health, pose expanded requirements on air quality monitoring and modelling not covered by the previous paradigm. For example, applications such as finding one’s least polluted way through a city require real-time data of roadside pollution at a high spatial resolution accessible to the general public.

Towards this goal, we firstly

• assessed the ability of Hong Kong’s current air quality network, made up of fixed site monitors, to represent the population-based health effects of air pollution. Combining high-resolution air quality model, spatial population distribution and health risk factors, we propose a novel population-health based network representation index. We found that the current monitoring network reflects health risks well for particulates but is less able to represent risks for NO₂ and O₃.

Subsequently, we performed two studies aimed at the integration of new data sources:

• Based on these findings, we tested whether the identified shortcomings of the current network can be improved by the addition of smart sensors. These sensors can be characterized by a low price and simple installation but also suffer from measurement errors. We conducted a model-based study, in which up to 400 pseudo smart sensors were perturbated with the aim of simulating common sensor errors and added to the existing FSM network in Hong Kong. For PM_2.5, improvements (up to 16%) to the high baseline representativeness were achievable only by the addition of high-quality sensors and favourable environmental conditions. Due to higher levels of pollution (population-weighted average 37.3 ppb) in comparison to sensor error ranges, smart sensors of a wider quality range were able to improve network representativeness (up to 42%) for NO₂. Often, a small number of added sensors of a higher quality class led to larger improvements than hundreds of lower-class sensors. The proposed methodology can help to find the required sensor quality and quantity to achieve an improvement in network representation in a given city.

• To achieve real-time communication of roadside air quality, we transferred crowd-sourced traffic data into road segment-based congestion states. It was found that congestion states were highly correlated to roadside monitoring stations for NO₂ (R²: 0.25), and less for PM_2.5 and O₃. Based on these congestion states, we developed a meta-model based traffic adjustment module. The module can update previously calculated model results in high temporal resolution (10 minutes) with the impact of chaotic traffic events. Adjustment success for NO₂ was validated by correlating the noise-fraction of monitoring data with calculated adjustments at the locations of roadside monitoring stations.