Ozone is an important secondary air pollutant and plays different significant roles in regulating the formation of secondary organic aerosols. However, the characteristics of winter vertical ozone distributions have rarely been studied. In the winter of 2017, field experiments were performed in Shanghai, China using hexacopter unmanned aerial vehicle (UAV) platforms. The vertical profiles of ozone were obtained from 0–1200 m above the ground level. Results show that the UAV observations were reliable to capture the vertical variations of ozone. Vertical ozone profiles in the winter are classified into four categories: (1) well-mixed profile, (2) altitudinal increasing profile, (3) stratification profile, and (4) spike profile. Results show that although the average surface ozone level was relatively low, strong ozone variability and high ozone concentrations occurred at the upper air. The maximum observed ozone concentration was 220 ppb. In addition, using meteorological profiles and backward trajectories, we found that the ozone elevation aloft can be attributed to the downward transport of air flow from higher altitudes. Furthermore, ozone accumulation in the winter could be influenced by the horizontal transport of air masses for the northern part of China. This study successfully used hexacopter UAV platforms to perform vertical observations within the boundary layer. This provides systematic classification of winter ozone distribution within the boundary layer.

Accurate air quality forecasts can provide data-driven supports for governmental departments to control air pollution and further protect the health of residents. However, existing air quality forecasting models mainly focus on site-specific time series forecasts at a local level, and rarely consider the spatiotemporal relationships among regional monitoring stations. As a novelty, we construct a diffusion convolutional recurrent neural network (DCRNN) model that fully considers the influence of geographic distance and dominant wind direction on the regional variations in air quality through different combinations of directed and undirected graphs. The hourly fine particulate matter (PM_2.5) and ozone data from 123 air quality monitoring stations in the Yangtze River Delta, China are used to evaluate the performance of the DCRNN model in the regional prediction of PM_2.5 and ozone concentrations. Results show that the proposed DCRNN model outperforms the baseline models in prediction accuracy. Compared with the undirected graph model, the directed graph model considering the effects of wind direction performs better in 24 h predictions of pollutant concentrations. In addition, more accurate forecasts of both PM_2.5 and ozone are found at a regional level where monitoring stations are distributed densely rather than sparsely. Therefore, the proposed model can assist environmental researchers to further improve the technologies of air quality forecasts and could also serve as tools for environmental policymakers to implement pollution control measures.

The establishment of an effective roadside air quality forecasting model provides important information for proper traffic management to mitigate severe pollution, and for alerting resident’s outdoor plans to minimize exposure. Current deterministic models rely on numerical simulation and the tuning of parameters, and empirical models present powerful learning ability but have not fully considered the temporal periodicity of air pollutants. In order to take the periodicity of pollutants into empirical air quality forecasting models, this study evaluates the temporal variations of air pollutants and develops a novel sequence to sequence model with weekly periodicity to forecast air quality. Two-year observation data from Shanghai roadside air quality monitoring stations are employed to support analyzing and modeling. The results conclude that the fine particulate matter (PM_2.5) and carbon monoxide (CO) concentrations show obvious daily and weekly variations, and the temporal patterns are nearly consistent with the periodicity of traffic flow in Shanghai. Compared with PM_2.5, the CO concentrations are more affected by traffic variation. The proposed model outperforms the baseline model in terms of accuracy, and presents a higher linear consistency in PM_2.5 prediction and lower errors in CO prediction. This study could assist environmental researchers to further improve the technologies for urban air quality forecasting, and serve as tools for supporting policymakers to implement related traffic management and emission control policies.