| Road intersection has become a hard-hit area of urban air pollution,while due to frequently travelling or overstaying in the intersection microenvironment,crowds of commuters are being exposed to the severe traffic-related air pollution with a high health risk.Fine particulate matter(PM2.5)and black carbon?BC?that are closely related to traffic have been demonstrated to pose a serious harm to human health,but to date,it is unclear how their spatiotemporal distributions are near road intersections and studies are even rarely reported in China.Focusing on a busy road intersection and its surrounding areas in the suburb of Shanghai,this paper systematically investigated the spatiotemporal distributions of PM2.5 and BC and their relationships with local traffic,weather,land use or typical roads,and proposed a nonlinear statistical model for predicting short-term variations of air pollution in proximity to the road intersection by using the fixed-point and mobile observation methods based on portable monitors.The research results will provide strong scientific basis for the exposure assessment of air pollution near road intersections,reduction of traffic emissions and roadside planning.The innovative work and research results are summarized as follows.Firstly,a systematic assessment on the accuracy and reliability of portable monitors for air pollutant measurements was launched to develop new tools of urban air pollution observation.Through 40-day contrast tests at five outdoor monitoring stations in Shanghai,the portable monitors were confirmed having high measurement accuracy and stability as well as strong correlation with the standard reference instruments.Some measurement errors of portable monitors relative to the standard reference instruments were found due to varied measurement environments and configurations between different devices.Thus,with the standard instruments as a reference,the empirical methods considering the modification of influencing factors were proposed to calibrate the portable monitors,and results showed that the methods were reasonable.Secondly,a three-point synchronous observation method at the intersection was established to study variations of PM2.5 and BC concentrations at multiple time scales,and to reveal the contribution of local traffic to roadside PM2.5 and BC variations using a bivariate polar plot method.The main findings are reached as follows.Three sampling sites represented roadside,diffusion and background of the intersection.The diurnal variations in pollutant concentrations at sampling sites resembled the diurnal cycle in local traffic volume,which was especially strong for BC at roadside.Roadside average concentration of PM2.5 was higher than background by 9.7?g/m3?8.9%?in the winter and 4.5?g/m3?8.6%?in the spring,and for BC,it was 4.1?g/m3?69.5%?in the winter and 2.9?g/m3?96.7%?in the spring.If the intersection was dominated by heavy air pollution caused by the external source,the local traffic contribution to roadside PM2.5 and BC decreased significantly.However,that contribution rarely changed for PM2.5 but slightly increased for BC if being mainly affected by the local traffic.The generalized additive models built between PM2.5?or BC?and meteorology and traffic near the intersection indicated that,the background level was always a first contributor to PM2.5 and BC variations,followed by air pressure and solar radiation.Air pressure affected atmospheric stability and then particle diffusion,while solar radiation determined photochemical reactions and then particle generation and disappearance.Air temperature,relative humidity and dew-point temperature played a role in promoting or inhibiting particle concentrations with a seasonal variation.BC was always sensitive to local traffic change from the windward of the sampling site,while PM2.5 was largely influenced by the external high pollution driven by the westerly wind.The vehicle number of gasoline or diesel generally contributed little to PM2.5 and BC variations near the intersection,among which only the vehicle number of diesel had an appreciable contribution to the roadside BC variation in the spring.Thirdly,a mobile method for measuring the traffic-related air pollution was built to explore how the spatial distributions of PM2.5 and BC concentrations response to kinds of microenvironments around the road intersection.The main results are:?1?The average concentration of PM2.5 and BC for the whole area showed a descending order in early morning,morning,afternoon and noon during a day.The high humidity and low wind speed led to the accumulation of particles in high concentrations along arterial roads in early morning,while traffic conditions and roadside environment largely decided the spatial variation of PM2.5 and BC concentration in other periods of a day.The average concentration of PM2.5 and BC for the whole area under the scenario of no prevailing wind direction but low wind speed was 5070%higher than the prevailing wind direction scenario.Wind direction hardly affected the spatial distribution of BC,which was opposite for PM2.5.PM2.5 was very sensitive to the diffusion condition especially at leeward of the road,but BC was more sensitive to local traffic intensity.Ultrafine particle(PM1.0)had a spatiotemporal pattern similar to PM2.5,but was more sensitive to traffic.?2?The median concentrations of PM2.5,PM1.0 and BC were higher in intersections or arterial roads than other microenvironments,and was even 2.8,2.5 and 16 times higher than campus.Additionally,PM2.5 and PM1.0 concentrations were extremely high in the production and livelihood areas.The highly polluted areas were always caused by heavy traffic and poor atmospheric diffusion at roadside,and the risk of commuting exposure to particles was increased because of the unreasonable space between bicycle and vehicle lanes.?3?In general,the highway exerted an impact on both sides at a maximum range of 350 m.The decay gradient of particulate matter on both sides of the highway showed that,the open side was clearer than the side with the densely built environment and heavy local traffic,and the leeward side was more prominent than the windward side of the highway.Low wind speed lifted up the overall concentration along the highway,and made the concentration gradient of particles significant and similar on both sides of the highway.Regardless of these conditions,BC concentrations declined more than 70%with increasing distance from the highway about 300 m,which was superior to PM2.5 and PM1.0 in stability.In the leeward of the typical arterial road,the adjacent street had a clearer concentration gradient of BC than that of PM2.5,and the BC gradient was less impacted by roadside structure and local traffic along the street.Finally,a hybrid model integrating wavelet neural network with genetic algorithm?GAWNN?was proposed to improve the classic back-propagation neural network?BPNN?in low learning efficiency and weak generative ability,which is intended to provide a new method for forecasting short-term variation of air pollution at urban intersection.GAWNN was compared with BPNN on a5-min series of PM2.5 and carbon monoxide?CO?concentrations sampled in a neighbourhood of a road intersection.The comparative results indicated that,GAWNN greatly enhanced the training efficiency of neural network,provided more reliable and accurate predictions of both PM2.5 and CO concentrations at varied sampling positions and traffic periods,and also outperformed BPNN in the spatial transferability prediction.This demonstrates the potential of the application of GAWNN to forecast the fine-scale trend of air pollution with high efficiency and practicability in the vicinity of road intersections. |
PDF Full Text Request