Development of urban air monitoring with high spatial resolution using mobile vehicle sensors

Kiwon Yeom
DOI: https://doi.org/10.1007/s10661-021-09139-2
2021-06-01
Abstract:Traditionally, the equipment used to measure air pollution is expensive and placed around cities or in mobile laboratories. It might only represent a certain area and not the entire city due to the locations and limited number of monitoring stations. Nowadays, a mobile sensing is becoming an alternative option to monitor air quality in urban environment due to its ease of use, high flexibility, and low price. This paper develops a vehicular-based mobile monitoring system for real time air quality sensing and visualization across large cities with high spatial resolution. The commercially available low-cost CO, NO[Formula: see text], NH[Formula: see text] O[Formula: see text], CH[Formula: see text], SO[Formula: see text], PM[Formula: see text], temperature and humidity sensors along with the microcontroller and GPS were integrated in a sensing device installed on the roof of taxi and sport utility vehicle (SUV). The developed device was calibrated through a reference monitoring station and validated through field measurement. We first split the entire city with a uniform grid discretization. We then propose a data processing methodology based on machine learning algorithms for generating 250 representative data set from 286 million data which is collected using the vehicular based mobile sensors. Next we present the representativeness of the data set by comparison of stationary data and mobile data. We also describe the analytical results and spatial distribution with high spatial resolution throughout the city. In addition, the collected mobile sensor data is also used to show that the significant differences and spatial variability in mean levels per street. Finally, we conclude that the proposed mobile monitoring system using high spatial resolution can effectively map the air quality in metropolitan environment and provide detail about the spatial variability that cannot be done with stationary monitoring systems.
What problem does this paper attempt to address?