PM2.5 concentration estimation using convolutional neural network and gradient boosting machine

Surface monitoring, vertical atmospheric column observation, and simulation using chemical transportation models are three dominant approaches for perception of fine particles with diameters less than 2.5 micrometers (PM_2.5) concentration. Here we explored an image-based methodology with a deep learning approach and machine learning approach to extend the ability on PM_2.5 perception. Using 6976 images combined with daily weather conditions and hourly time data in Shanghai (2016), trained by hourly surface monitoring concentrations, an end-to-end model consisting of convolutional neural network and gradient boosting machine (GBM) was constructed. The mean absolute error, the root-mean-square error and the R-squared for PM_2.5 concentration estimation using our proposed method is 3.56, 10.02, and 0.85 respectively. The transferability analysis showed that networks trained in Shanghai, fine-tuned with only 10% of images in other locations, achieved performances similar to ones from trained on data from target locations themselves. The sensitivity of different regions in the image to PM_2.5 concentration was also quantified through the analysis of feature importance in GBM. All the required inputs in this study are commonly available, which greatly improved the accessibility of PM_2.5 concentration for placed and period with no surface observation. And this study makes an exploratory attempt on pollution monitoring using graph theory and deep learning approach.     

Variations in traffic-related polycyclic aromatic hydrocarbons in PM2.5 in Kanazawa, Japan, after the implementation of a new vehicle emission regulation

A three-year sampling campaign was conducted at a roadside air pollution monitoring station in the urban area of Kanazawa, Japan. Due to a new emission regulation, PAHs levels decreased over the sampling campaign, exhibiting values of 706 ± 413 pg/m³ in 2017, 559 ± 384 pg/m³ in 2018, and 473 ± 234 pg/m³ in 2019. In each year, similar seasonal variations in PAHs levels were observed, with higher levels observed in winter and lower levels in summer. Among the PAHs isomer ratios, we observed that the ratio of benzo[b]fluoranthene (BbF) and benzo[k]fluoranthene (BkF), [BbF]/([BbF] + [BkF]), and the ratio of indeno[1,2,3-cd]pyrene (IDP) and benzo[ghi]perylene (BgPe), [IDP]/([BgPe] + [IDP]), showed stability over the sampling campaign and were less affected by the new emission regulation, seasonal variations, and regional characteristics. When using the combined ratio ranges of 0.66 - 0.80 ([BbF]/([BbF] + [BkF]) and 0.26-0.49 ([IDP]/([BgPe] + [IDP]), traffic emissions were clearly distinguished from other PAHs emission sources. Principal component analysis (PCA) and positive matrix factorization (PMF) were also performed to further analyse the characteristics of traffic-related PAHs. Overall, this study affirmed the effectiveness of the new emission regulation in the reduction of PAHs emissions and provided a combined range for identifying PAHs traffic emission sources.     

耦合遗传算法与RBF神经网络的PM2.5浓度预测模型

基于北京市空气质量监测点获取的空气污染物浓度数据,通过遗传算法搜索径向基人工神经网络的最优隐含层神经元数目和扩展常数,构建了耦合径向基人工神经网络算法与遗传算法的预测模型,预测北京市未来一天24h平均PM_2.5质量浓度.结果表明,预测精度与泛化性能良好.该模型不需要输入气象和地理位置信息等数据,具有依赖变量少、预测精度高（R²达0.75）和运算效率高等特征,并可以通过训练样本的驱动,使自身不断优化调整.该模型预测效果可以通过扩展输入特征、增加训练样本量等方法进一步提升,可对多种时空情境下的城市空气污染进行高效率且精确的预测.     

Variation characteristics of fine particulate matter and its components in diesel vehicle emission plumes

A rapid reaction occurs near the exhaust nozzle when vehicle emissions contact the air. Twenty diesel vehicles were studied using a new multipoint sampling system that is suitable for studying the exhaust plume near the exhaust nozzle. The variation characteristics of fine particle matter (PM_2.5) and its components in diesel vehicle exhaust plumes were analyzed. The PM_2.5 emissions gradually increased with increasing distance from the nozzle in the plume. Elemental carbon emissions remained basically unchanged, organic carbon and total carbon (TC) increased with increasing distance. The concentrations of SO₄^2?, NO₃^? and NH₄⁺ (SNA) directly emitted by the vehicles were very low but increased rapidly in the exhaust plume. The selective catalytic reduction (SCR) reduced 42.7% TC, 40% NO₃^? emissions, but increased 104% SO₄^2? and 36% NH₄⁺ emissions, respectively. In summary, the SCR reduced 29% primary PM_2.5 emissions for the tested diesel vehicles. The NH₄NO₃ particle formation maybe more important in the plume, and there maybe other forms of formation of NH₄⁺ (eg. NH₄Cl). The generation of secondary organic carbon (SOC) plays a leading role in the generation of secondary PM_2.5. The SCR enhanced the formation of SOC and SNA in the plume, but comprehensive analysis shows that the SCR more enhanced the SNA formation in the plume, which is mainly new particles formation process. The inconsistency between secondary organic aerosol (SOA) and primary organic aerosol definitions is one of the important reasons for the difference between SOA simulation and observation.     

基于遗传算法与人工神经网络相结合的玉米估产研究

在遗传算法 (GeneticAlgorithm )与误差反传 (BackPropagation)网络结构模型相结合的基础上 ,设计了用遗传算法训练神经网络权重的新方法 ,并对吉林省梨树县的玉米进行了估产研究 ,同时与BP算法和灰色系统理论模型进行了比较。经检验 ,计算值与实际值接近 ,并优于灰色理论模型 ,具有良好的预测效果 ,从而为农作物估产提供了新方法     

人工神经元网络对水处理系统建模适应性的研究

针对水处理系统的特点和研究中的难点，引入人工神经元网络的理论和思想，提出了水处理系统的神经网络分析方法．通过建立基于ＢＰ人工神经元网络的臭氧生物活性炭系统的模型，具体地考察了人工神经元网络对水处理系统建模的适应性，使水处理系统的研究迈向智能化和控制化．     

Artificial neural network model for identifying taxi gross emitter from remote sensing data of vehicle emission

Vehicle emission has been the major source of air pollution in urban areas in the past two decades. This article proposes an artificial neural network model for identifying the taxi gross emitters based on the remote sensing data. After carrying out the field test in Guangzhou and analyzing various factors from the emission data, the artificial neural network modeling was proved to be an advisable method of identifying the gross emitters. On the basis of the principal component analysis and the selection of algorithm and architecture, the Back-Propagation neural network model with 8-17-1 architecture was established as the optimal approach for this purpose. It gave a percentage of hits of 93%. Our previous research result and the result from aggression analysis were compared, and they provided respectively the percentage of hits of 81.63% and 75%. This comparison demonstrates the potentiality and validity of the proposed method in the identification of taxi gross emitters.     

大气中SO2浓度的小波分析及神经网络预测

应用小波分解和重构对SO2浓度年变化趋势进行分析,在此基础上,建立了一种分段BP神经网络预测模型,并对各段有针对性地设计了神经网络预测模型.采用主成分分析进行输入变量降维.在BP网络训练过程中,往往会出现过拟合的现象,为此,在训练过程中,将样本等间距地分离为训练集和验证集来防止这个问题.为了消除网络的权值初始化对学习系统复杂性的影响,采用了5个子网络输出取算术平均的神经网络集成的方法.预测结果表明,该模型应用于SO2浓度预测具有较高的预测精度和良好的推广能力,而且明显优于一般的神经网络模型.     

Ammonia in urban atmosphere can be substantially reduced by vehicle emission control: A case study in Shanghai, China

To investigate the impact of emission controls on ammonia (NH₃) pollution in urban atmosphere, observation on NH₃ (1 hr interval) was performed in Shanghai before, during and after the 2019 China International Import Expo (CIIE) event, along with measurements on inorganic ions, organic tracers and stable nitrogen isotope compositions of ammonium in PM_2.5. NH₃ during the CIIE period was 6.5±1.0 µg/m³, which is 41% and 32% lower than that before and after the event, respectively. Such a decrease was largely ascribed to the emission controls in nonagricultural sources, of which contribution for measured NH₃ in control phase abated by ∼20% compared to that during uncontrol period. Molecular compositions of PAHs and hopanes further suggested a dominant role of the reduced vehicle emissions in the urban NH₃ abatement during the CIIE period. Our results revealed that vehicle exhaust emission control is an effective way to mitigate NH₃ pollution and improve air quality in Chinese urban areas.     

Fine particulate matter (PM2.5/PM1.0) in Beijing, China: Variations and chemical compositions as well as sources

Particulate matter (i.e., PM_1.0 and PM_2.5), considered as the key atmospheric pollutants, exerts negative effects on visibility, global climate, and human health by associated chemical compositions. However, our understanding of PM and its chemical compositions in Beijing under the current atmospheric environment is still not complete after witnessing marked alleviation during 2013–2017. Continuous measurements can be crucial for further air quality improvement by better characterizing PM pollution and chemical compositions in Beijing. Here, we conducted simultaneous measurements on PM in Beijing during 2018–2019. Results indicate that annual mean PM_1.0 and PM_2.5 concentrations were 35.49 ± 18.61 µg/m³ and 66.58 ± 60.17 µg/m³, showing a positive response to emission controls. The contribution of sulfate, nitrate, and ammonium (SNA) played an enhanced role with elevated PM loading and acted as the main contributors to pollution episodes. Discrepancies observed among chemical species between PM_1.0 and PM_2.5 in spring suggest that sand particles trend to accumulate in the range of 1–2.5 µm. Pollution episodes occurred accompanied with southerly clusters and high formation of SNA by heterogeneous reactions in summer and winter, respectively. Results from positive matrix factorization (PMF) combined with potential source contribution function (PSCF) models showed that potential areas were seasonal dependent, secondary and vehicular sources became much more important compared with previous studies in Beijing. Our study presented a continuous investigation on PM and sources origins in Beijing, which provides a better understanding for further emission control as well as a reference for other cities in developing countries.     

Influencing factors and prediction of ambient Peroxyacetyl nitrate concentration in Beijing, China

Peroxyacyl nitrates (PANs) are important secondary pollutants in ground-level atmosphere. Accurate prediction of atmospheric pollutant concentrations is crucial to guide effective precautions for before and during specific pollution events. In this study, four models based on the back-propagation (BP) artificial neural network (ANN) and multiple linear regression (MLR) methods were used to predict the hourly average PAN concentrations at Peking University, Beijing, in 2014. The model inputs were atmospheric pollutant data and meteorological parameters. Model 3 using a BP-ANN based on the original variables achieved the best prediction results among the four models, with a correlation coefficient (R) of 0.7089, mean bias error of ? 0.0043 ppb, mean absolute error of 0.4836?ppb, root mean squared error of 0.5320?ppb, and Willmott's index of agreement of 0.8214. Based on a comparison of the performance indices of the MLR and BP-ANN models, we concluded that the BP-ANN model was able to capture the highly non-linear relationships between PAN concentration and the conventional atmospheric pollutant and meteorological parameters, providing more accurate results than the traditional MLR models did, with a markedly higher goodness of R. The selected meteorological and atmospheric pollutant parameters described a sufficient amount of PAN variation, and thus provided satisfactory prediction results. More specifically, the BP-ANN model performed very well for capturing the variation pattern when PAN concentrations were low. The findings of this study address some of the existing knowledge gaps in this research field and provide a theoretical basis for future regional air pollution control.     

BP神经网络和ARIMA模型对污水处理厂出水总氮浓度的模拟预测

污水处理厂出水总氮(TN)浓度是评价水处理效果的关键指标之一。建立BP神经网络模型对污水处理厂脱氮工艺进行模拟,引入自回归整合移动平均模型(ARIMA模型)对污水处理厂未来短期出水TN浓度进行预测。结果表明:BP神经网络模型在训练集和测试集模拟结果的平均相对误差分别为15.9%和16.5%,模型预测结果的平稳性较差;ARIMA模型对未来7 d出水TN浓度的时序预测平均误差为4.41%,预测精度较高;2个模型相结合有助于实现污水处理厂快捷和高效的在线检测。     

基于模糊神经网络的水稻农田重金属污染水平高光谱预测模型

以吉林省长春一汽厂附近3块不同重金属污染状况的水稻实验样地为例,通过地面实测获取了水稻的光谱反射率、叶片叶绿素含量、叶片及土壤重金属含量等数据.同时,在分析重金属对水稻叶片叶绿素含量影响的基础上,通过多元逐步回归分析选出对水稻叶片叶绿素含量微小变化指示灵敏的光谱参数作为模型输入层,并将水稻叶片叶绿素含量值作为输出层来表征农田重金属污染胁迫水平,最终建立了用于预测水稻农田重金属污染水平的模糊神经网络模型.结果表明,该模糊神经网络模型预测的水稻重金属污染胁迫水平与实测结果吻合度较高,预测的叶绿素含量值与实测值的拟合度较好(R2=0.985).表明在受重金属污染胁迫的情况下,水稻叶片叶绿素含量微小而复杂的变化可以通过构建模糊神经网络模型很好地模拟出来,从而确定出农田的重金属污染水平.     

Ozone pollution characteristics and sensitivity analysis using an observation-based model in Nanjing, Yangtze River Delta Region of China

Ground-level ozone (O₃) has become a critical pollutant impeding air quality improvement in Yangtze River Delta region of China.In this study,we present O₃ pollution characteristics based on one-year online measurements during 2016 at an urban site in Nanjing,Jiangsu Province.Then,the sensitivity of O₃ to its precursors during 2 O₃ pollution episodes in August was analyzed using a box model based on observation (OBM).The relative incremental reactivity...     

Ethanol mediated As(III) adsorption onto Zn-loaded pinecone biochar: Experimental investigation, modeling, and optimization using hybrid artificial neural network-genetic algorithm approach

Organic matters (OMs) and their oxidization products often influence the fate and transport of heavy metals in the subsurface aqueous systems through interaction with the mineral surfaces. This study investigates the ethanol (EtOH)-mediated As(III) adsorption onto Zn-loaded pinecone (PC) biochar through batch experiments conducted under Box–Behnken design. The effect of EtOH on As(III) adsorption mechanism was quantitatively elucidated by fitting the experimental data using artificial neural network and quadratic modeling approaches. The quadratic model could describe the limiting nature of EtOH and pH on As(III) adsorption, whereas neural network revealed the stronger influence of EtOH (64.5%) followed by pH (20.75%) and As(III) concentration (14.75%) on the adsorption phenomena. Besides, the interaction among process variables indicated that EtOH enhances As(III) adsorption over a pH range of 2 to 7, possibly due to facilitation of ligand–metal(Zn) binding complexation mechanism. Eventually, hybrid response surface model–genetic algorithm (RSM–GA) approach predicted a better optimal solution than RSM, i.e., the adsorptive removal of As(III) (10.47 μg/g) is facilitated at 30.22 mg C/L of EtOH with initial As(III) concentration of 196.77 μg/L at pH 5.8. The implication of this investigation might help in understanding the application of biochar for removal of various As(III) species in the presence of OM.     

NirS-type denitrifying bacteria in aerobic water layers of two drinking water reservoirs: Insights into the abundance, community diversity and co-existence model

The nirS-type denitrifying bacterial community is the main drivers of the nitrogen loss process in drinking water reservoir ecosystems.The temporal patterns in nirS gene abundance and nirS-type denitrifying bacterial community harbored in aerobic water layers of drinking water reservoirs have not been studied well.In this study,quantitative polymerase chain reaction (qPCR) and Illumina Miseq sequencing were employed to explore the nirS gene abundance and denitrifying bacterial community structur...