Spatial and temporal urban air pollution patterns based on limited data of monitoring stations

Urban air pollution is one of the serious environmental problems in the development of high-density cities, especially for traffic-related pollution. Effective urban environmental monitoring technologies are helpful to predict and control the air pollution emitted by traffic sectors. However, the complex temporal and spatial characteristics of air pollution in urban areas pose a great challenge for its monitoring, prediction, and control. This work investigated on the spatiotemporal air pollution patterns and its correlation mechanism with urban elements (such as traffic) based on limited data of monitoring stations, which is significant for the accurate prediction and regulation of urban environment. Regional Cluster Monitoring (RCM) methods were proposed through integrating multi-source data, including data from U.S. Geological Survey (USGS) Landsat, OpenStreetMap (OSM) road network, and environmental monitoring stations to categorize different monitoring stations. According to different urban areas (defined as U: 'urban' or S: 'suburban') and traffic densities (T: 'traffic' or B: 'background'), monitoring stations areas were categorized into U-B, U-T, S-B, and S-T. Then the spatiotemporal characteristics of traffic-related pollutants in different region categories were analyzed. The rapid prediction of air pollution distribution in different categories of urban regions was realized by RCM method-based Long Short-Term Memory (LSTM) neural network. Results showed that the proposed methods could accurately represent the spatiotemporal distribution characteristics of urban air pollution. The mean absolute percentage errors (MAPE) between the predicted results and actual observed values were 20.31% to 28.05% (absolute errors were 3.9-14.3 μg/m³). This work can significantly impact on low-cost & effective monitoring and control strategies for urban area, and further contribute to comprehend the spatial-temporal distribution patterns of urban traffic-related air pollution from limited data of monitoring stations.