Search

Accurate prediction of air pollutant concentrations can be effective in controlling and preventing air pollution, which is crucial for both the government's policy response and the public's reasonable ability to avoid air pollution. In this study, we propose a model based on the principle of big data correlation and deep learning techniques for predicting the concentration of air pollutant concentrations. The proposed model, named STARes-SaLSTM, consists of a residual neural network with spatiotemporal attention (STAResNet) and a temporal self-attention based on long short-term memory network (SaLSTM). The spatiotemporal attention module in STAResNet enhances the original input information by weighting it in both spatial and temporal dimensions, meanwhile the features of the spatial distribution of pollutant and meteorological information across many cities are deeply extracted using ResNet. A novel self-attention method known as temporal self-attention is developed to extract the temporal dependency of air pollutants. LSTM is used as a sequence encoder in this temporal attention method to calculate queries, keys, and values to acquire a more comprehensive temporal dependence than is possible with normal self-attention. The STARes-SaLSTM model successfully predicts air pollution concentrations in the target city for the future by extracting spatiotemporal correlation of feature sequence. In comparison to previous neural network models and conventional methods, the model increases the accuracy of pollutant concentration prediction. The suggested prediction model works well for the one to 3 h prediction job, with a root mean square error (RMSE) ranging from 6.716 to 11.648. Furthermore, even for the 1–24 h prediction assignment, we executed multi-scale prediction in the target city and obtained a reasonable performance, with an average RMSE value of 20.576. • This paper solves the problem of difficult extraction of spatiotemporal features. • ResNet allows for the extraction of spatially relevant features; • The spatiotemporal attention mechanism is used to spatiotemporal extract features; • The temporal self-attention mechanism obtain temporal dependence of data. [ABSTRACT FROM AUTHOR]

The current compliance networks of automatic air-quality monitoring stations in large urban environments are not sufficient to provide spatial and temporal measurement resolution for realistic assessment of personal exposure to pollutants. Small low-cost sensor platforms with greater mobility and expected lower maintenance costs, are increasingly being used as a supplement to compliance monitoring stations. However, low-cost sensor platforms usually provide data with uncertain precision. To improve the precision, these sensor platforms require in-field calibration. Our paper aims to demonstrate that data from each individual sensor system can be corrected using that sensor system's own data to achieve much improved data quality compared to a reference. However, in this procedure, there are practical difficulties such as individual sensor outputs from the multi-sensor system not being sufficiently available due to malfunctions for instance. We explore how this can be dealt with. In our opinion, this is a novel approach, of practical importance both to users and manufacturers. We present a detailed comparative analysis of Linear Regression (univariate), Multivariate Linear Regression and Artificial Neural Networks used with a specific aim of calibrating field-deployed low-cost CO and O 3 sensors. For Artificial Neural Network models, the performance of three common training algorithms was compared (Levenberg-Marquardt, Resilient back-propagation and Conjugate Gradient Powell-Beale algorithm). Data for this study were obtained from two campaigns conducted with 25 multi-sensor AQMESH v.3.5 platforms used within the activities of the CITI-SENSE project. The platforms were co-located to reference gas monitors at the Automatic Monitoring Station Stari Grad, in Belgrade, Serbia. This paper demonstrates that Multivariate Linear Regression and Artificial Neural Network calibration models can improve the output signal. This improvement can be measured by changes in the median and interquartile ranges of statistical parameters used for model evaluation. Artificial Neural Networks showed the best results compared to Linear Regression and Multivariate Linear Regression models. The best predictors for CO, in addition to CO low-cost sensor data, were PM 2.5 and NO 2 , while for O 3 , in addition to O 3 low-cost sensor data, the most suitable input predictors were NO and aH. Based on residual error analysis, we have shown that for CO and O 3 , a certain range of concentrations exists in which calibrated values differ by less than 10% from the reference method results. In addition, it was noted that for all models, CO sensors consistently showed lower variability between platforms compared to O 3 sensors. Image 1 • Precision of measured concentrations can be enhanced by using other available sensors. • Calibration predictors for CO were CO lc , PM 2.5 and NO 2 , for O 3 were O 3, lc , NO and aH. • ANN models perform better compared to linear or multilinear regression models. • Within certain ranges of CO and O 3 the error of calibrated data is less than 10%. • CO sensors show less variability between the platforms than O 3 sensors. [ABSTRACT FROM AUTHOR]

In this paper, a new method to calculate the average spatial distribution of air pollutants based on diffusive sampling measurements and artificial neural networks evaluation is presented. Most established methods like interpolation algorithms are inflexible or limited in considering important distribution parameters such as emission sources or land use. Of special interest are air quality measurements since they provide a direct view on the actual pollutant level. With diffusive samplers, the average concentration of many gaseous species over a large area can be determined simultaneously. During a project in Cyprus, NO2 diffusive samplers were exposed at 270 sites in six month-long campaigns throughout one year providing the database for the model described in this paper. A multilayer perceptron was trained with the NO2 measurement data and distribution parameters like population density and meteorological parameters using a 1×1km grid covering Cyprus. The best fit could be achieved with an emissions inventory including previously simulated concentration plumes and population density data as input nodes for the neural network, resulting in realistic maps of the annual average distribution of NO2 in Cyprus. [Copyright &y& Elsevier]

Hazardous gas leak accident has posed a potential threat to human beings. Predicting atmospheric dispersion and estimating its source become increasingly important in emergency management. Current dispersion prediction and source estimation models cannot satisfy the requirement of emergency management because they are not equipped with high efficiency and accuracy at the same time. In this paper, we develop a fast and accurate dispersion prediction and source estimation method based on artificial neural network (ANN), particle swarm optimization (PSO) and expectation maximization (EM). The novel method uses a large amount of pre-determined scenarios to train the ANN for dispersion prediction, so that the ANN can predict concentration distribution accurately and efficiently. PSO and EM are applied for estimating the source parameters, which can effectively accelerate the process of convergence. The method is verified by the Indianapolis field study with a SF 6 release source. The results demonstrate the effectiveness of the method. [ABSTRACT FROM AUTHOR]

This paper presents a novel approach to predict carbon dioxide (CO 2), nitrogen oxides (NOx), and carbon monoxide (CO) emissions of diesel vehicles using artificial neural network (ANN), which offer high degrees of accuracy and practicality. Six operating parameters (velocity, engine speed, engine torque, engine coolant temperature, fuel/air ratio, and intake air mass flow) collected through on-board diagnostic interface were used as predictors of exhaust emissions. The importance of each parameter to the emission predictions were comprehensively analyzed by comparing the coefficient of determination, root mean square error, cumulative emissions, and instantaneous emission rates. The emission prediction accuracy of ANN tends to increase as more parameters were considered as model inputs at the same time. However, the level of accuracy improvement depends on the input parameters. For CO 2 emissions, engine torque and fuel/air ratio were good predictors for achieving high prediction accuracy. The relative importance of intake air mass flow rate and fuel/air ratio was high for NOx and CO predictions, respectively. In addition, the emission prediction accuracy of ANN depends on the vehicle type (Euro 5, Euro 6b, Euro 6d-temp). The emission prediction accuracy of vehicles equipped with after-treatment devices (selective catalytic reduction and lean NOx trap) was lower than that of vehicles without after-treatment devices. • Artificial neural networks were developed to predict CO 2 , NOx, and CO emissions. • On-road testing data were used to train artificial neural networks. • The importance of vehicle parameters to emission predictions were quantified. • Fuel/air ratio and intake air mass flow are good predictors of exhaust emissions. • NOx prediction accuracy were highly dependent on vehicle types. [ABSTRACT FROM AUTHOR]

In the paper a novel hybrid model combining air mass trajectory analysis and wavelet transformation to improve the artificial neural network (ANN) forecast accuracy of daily average concentrations of PM 2.5 two days in advance is presented. The model was developed from 13 different air pollution monitoring stations in Beijing, Tianjin, and Hebei province (Jing-Jin-Ji area). The air mass trajectory was used to recognize distinct corridors for transport of “dirty” air and “clean” air to selected stations. With each corridor, a triangular station net was constructed based on air mass trajectories and the distances between neighboring sites. Wind speed and direction were also considered as parameters in calculating this trajectory based air pollution indicator value. Moreover, the original time series of PM 2.5 concentration was decomposed by wavelet transformation into a few sub-series with lower variability. The prediction strategy applied to each of them and then summed up the individual prediction results. Daily meteorological forecast variables as well as the respective pollutant predictors were used as input to a multi-layer perceptron (MLP) type of back-propagation neural network. The experimental verification of the proposed model was conducted over a period of more than one year (between September 2013 and October 2014). It is found that the trajectory based geographic model and wavelet transformation can be effective tools to improve the PM 2.5 forecasting accuracy. The root mean squared error (RMSE) of the hybrid model can be reduced, on the average, by up to 40 percent. Particularly, the high PM 2.5 days are almost anticipated by using wavelet decomposition and the detection rate (DR) for a given alert threshold of hybrid model can reach 90% on average. This approach shows the potential to be applied in other countries’ air quality forecasting systems. [ABSTRACT FROM AUTHOR]

This paper uses artificial neural networks (ANN), combined with k-means clustering, to understand the complex time series of PM10 and PM2.5 concentrations at a coastal location of New Zealand based on data from a single site. Out of available meteorological parameters from the network (wind speed, wind direction, solar radiation, temperature, relative humidity), key factors governing the pattern of the time series concentrations were identified through input sensitivity analysis performed on the trained neural network model. The transport pathways of particulate matter under these key meteorological parameters were further analysed through bivariate concentration polar plots and k-means clustering techniques. The analysis shows that the external sources such as marine aerosols and local sources such as traffic and biomass burning contribute equally to the particulate matter concentrations at the study site. These results are in agreement with the results of receptor modelling by the Auckland Council based on Positive Matrix Factorization (PMF). Our findings also show that contrasting concentration-wind speed relationships exist between marine aerosols and local traffic sources resulting in very noisy and seemingly large random PM10 concentrations. The inclusion of cluster rankings as an input parameter to the ANN model showed a statistically significant (p < 0.005) improvement in the performance of the ANN time series model and also showed better performance in picking up high concentrations. For the presented case study, the correlation coefficient between observed and predicted concentrations improved from 0.77 to 0.79 for PM2.5 and from 0.63 to 0.69 for PM10 and reduced the root mean squared error (RMSE) from 5.00 to 4.74 for PM2.5 and from 6.77 to 6.34 for PM10. The techniques presented here enable the user to obtain an understanding of potential sources and their transport characteristics prior to the implementation of costly chemical analysis techniques or advanced air dispersion models. [ABSTRACT FROM AUTHOR]

Abstract: The bulk of existing work on the statistical forecasting of air quality is based on either neural networks or linear regressions, which are both subject to important drawbacks. In particular, while neural networks are complicated and prone to in-sample overfitting, linear regressions are highly dependent on the specification of the regression function. The present paper shows how combining linear regression forecasts can be used to circumvent all of these problems. The usefulness of the proposed combination approach is verified using both Monte Carlo simulation and an extensive application to air quality in Bogota, one of the largest and most polluted cities in Latin America. [Copyright &y& Elsevier]

Abstract: Soiling results from the deposition of pollutants on materials. On glass, it leads to an alteration of its intrinsic optical properties. The nature and intensity of this phenomenon mirrors the pollution of an environment. This paper proposes a new statistical model in order to predict the evolution of haze (H) (i.e. diffuse/direct transmitted light ratio) as a function of time and major pollutant concentrations in the atmosphere (SO2, NO2, and PM10 (Particulate Matter<10μm)). The model was parameterized by using a large set of data collected in European cities (especially, Paris and its suburbs, Athens, Krakow, Prague, and Rome) during field exposure campaigns (French, European, and international programs). This statistical model, called NEUROPT-Glass, comes from an artificial neural network with two hidden layers and uses a non-linear parametric regression named Multilayer Perceptron (MLP). The results display a high determination coefficient (R 2 =0.88) between the measured and the predicted hazes and minimizes the dispersion of data compared to existing multilinear dose–response functions. Therefore, this model can be used with a great confidence in order to predict the soiling of glass as a function of time in world cities with different levels of pollution or to assess the effect of pollution reduction policies on glass soiling problems in urban environments. [Copyright &y& Elsevier]

Abstract: This paper evaluates the effects of long-range transport patterns of air masses to the regional PM profile in a megacity, Istanbul, Turkey. Five-day hourly backward trajectories were obtained by the HYSPLIT model for selected episodic events in 2008. Self Organizing Maps (SOM), a very powerful classification tool, was used to cluster these trajectories. In total, eight cluster groups were obtained. All of the clusters were evaluated with respect to inhalable particulate matter (PM10) concentrations observed in Istanbul for the arrival times of the trajectories. Istanbul is generally under the effect of trajectories in three clusters (1, 2 and 4) (52% of all pre-selected episodic events), which have higher mean concentration values than the mean value of all the samples. These clusters typically make significant PM contributions to Istanbul''s air quality. PM loadings of the trajectories in these clusters were attributed to massive anthropogenic activity over all of Europe and southwestern air flow most likely carrying PM10 atmospheric particles originating from the Saharan Desert and other global dust generation regions located in the northern part of Africa. [Copyright &y& Elsevier]

Abstract: The new method for the forecasting hourly concentrations of air pollutants is presented in the paper. The method was developed for a site in urban residential area in city of Zagreb, Croatia, for four air pollutants (NO2, O3, CO and PM10). Meteorological variables and concentrations of the respective pollutant were taken as predictors. A novel approach, based on families of univariate regression models, was employed in selecting the averaging intervals for input variables. For each variable and each averaging period between 1 and 97 h, a separate model was built. By inspecting values of the coefficient of correlation between measured and modelled concentrations, optimal averaging periods for each variable were selected. A new dataset for building the forecasting model was then calculated as temporal moving averages (running means) of former variables. A multi-layer perceptron type of neural networks is used as the forecasting model. Index of agreement, calculated for the entire dataset including the data for model building, ranged from 0.91 to 0.97 for the respective pollutants. As suggested by the analysis of the relative importance of the input variables, different agreements for different pollutants are likely due to different sources and production mechanisms of investigated pollutants. A comparison of the new method with more traditional method, which takes hourly averages of the forecast hour as input variables, showed similar or better performance. The model was developed for the purpose of public-health-oriented air quality forecasting, aiming to use a numerical weather forecast model for the prediction of the part of input data yet unknown at the forecasting time. It is to expect that longer term averages used as inputs in the proposed method will contribute to smaller input errors and the greater accuracy of the model. [Copyright &y& Elsevier]

Abstract: In order to define efficient air quality plans, Regional Authorities need suitable tools to evaluate both the impact of emission reduction strategies on pollution indexes and the costs of such emission reductions. The air quality control can be formalized as a two-objective nonlinear mathematical problem, integrating source–receptor models and the estimate of emission reduction costs. Both aspects present several complex elements. In particular the source–receptor models cannot be implemented through deterministic modelling systems, that would bring to a computationally unfeasible mathematical problem. In this paper we suggest to identify source–receptor statistical models (neural network and neuro-fuzzy) processing the simulations of a deterministic multi-phase modelling system (GAMES). The methodology has been applied to ozone and PM10 concentrations in Northern Italy. The results show that, despite a large advantage in terms of computational costs, the selected source–receptor models are able to accurately reproduce the simulation of the 3D modelling system. [Copyright &y& Elsevier]

Artificial neural networks are functional alternative techniques in modelling the intricate vehicular exhaust emission dispersion phenomenon. Pollutant predictions are notoriously complex when using either deterministic or stochastic models, which explains why this model was developed using a neural network. Neural networks have the ability to learn about non-linear relationships between the used variables. In this paper a recurrent neural network (Elman model) based forecaster for the prediction of daily maximum concentrations of SO2, O3, PM10, NO2, CO in the city of Palermo is proposed. The effectiveness of the presented forecaster was tested using a time series recorded between 1 January 2003 to 31 December 2004 in eight monitoring stations in urban area of Palermo (Italy). Experimental trials show that the developed and tuned model is appropriate, giving small values of root mean square error (RMSE) , mean absolute error (MAE) and mean square error (MSE). In addition, the related correlation coefficient ranges from 0. 72 to 0. 97 for each forecasted pollutant, underlying a small difference between the forecasted and the measured values. The above results make the proposed forecaster a powerful tool for pollution management systems. [Copyright &y& Elsevier]

Abstract: This paper deals with tropospheric ozone modelling by using Artificial Neural Networks (ANNs). In this study, ambient ozone concentrations are estimated using surface meteorological variables and vehicle emission variables as predictors. The work is especially focused on analysing the importance of the input variables used by these models. This analysis is carried out in different time windows: all the time of study (April of 1997, 1999 and 2000), one month (April 1999), and finally, an hourly analysis. All the information extracted from these analyses can determine the most important factors in tropospheric ozone formation, thus achieving a qualitative model from the quantitative model obtained by neural networks. The relative importance of both meteorological and vehicle emission variables on the surface ozone prediction is of great interest to establish the legislative measures that permit to reduce the tropospheric ozone levels. The methodology developed in this study is applied to a small town near Valencia (Spain), but it can be generalisable to other locations. [Copyright &y& Elsevier]

The role of precursors' concentrations and meteorological conditions on the emerging ozone pollution problem in China has received wide attention, especially after the releasing of the Air Pollution Prevention Control and Action Plan (APPC) since 2013. With the decreasing trend of PM 2.5 nationwide, the effect of the strict control measures on increasing ozone variation has less been studied due to the challenge of complexity of nonlinear relationship among a number of factors on ozone formation. This paper evaluated the influence of both ozone precursors and meteorology on maximum daily average 8 h (MDA8) ozone at two urban sites and one rural site in Hebei province, China, from 2013 to 2017, by using a combined application of Kolmogorov-Zurbenko (KZ) filter and artificial neural network (ANN) model. Results showed that R2 was 0.80 between the measured and the simulated MDA8 ozone concentration when using meteorological factors and precursors' concentrations as input variables for ANN model. However, ANN model has limitation in estimating O 3 concentration peaks. The values of threat score (TS), probability of detection (POD) and false alarm rate (FAR) for MDA8 ozone concentration were 39%, 44% and 21%, respectively, throughout three studied sites in Hebei province. The annual average percentage change of precursor-related ozone was 0.67% from 2014 to 2017. Temperature, atmospheric pressure and boundary layer height were shown to account for 64% of the variability in long-term ozone levels. Ozone variation in Hebei province was reproduced mainly by meteorological parameters, and the contribution from precursors' concentration was smaller during the years when the APPC was implemented. [Display omitted] • Ozone level was estimated based on precursors' level and meteorological factors. • The ambient ozone concentration estimated by ANN model was acceptable. • Meteorology dominates the ozone variation in 2013–2017 in Hebei province. [ABSTRACT FROM AUTHOR]

The emission parameterization is a crucial part of numerical pollen dispersion models. This paper shows that Artificial Neural Networks (ANNs) can substantially improve the performance of the Ambrosia pollen emission in numerical pollen dispersion models such as COSMO-ART. Based on simultaneous measurement of Ambrosia pollen concentrations and meteorological variables in the source area, ANNs were trained to predict the diurnal profile of pollen emission. Six different combinations of explanatory meteorological variables were trained with five different ANN configurations resulting in 30 candidate emission models. The best network configuration for each combination of explanatory variables were used as emission parameterization in the numerical pollen dispersion model COSMO-ART. In addition, two benchmarks were implemented: an emission parameterization based on sigmoid functions and an artificial neural network using only time as an explanatory variable. The Ambrosia pollen seasons of 2015 and 2016 were simulated using the two benchmarks and the six emission parameterizations. The modelled diurnal profile of emission fluxes at 15 different sites from Serbia, Hungary and France with strong local pollen sources were compared with observed concentrations. Artificial Neural Networks based emission parameterization substantially improved the performance of the Ambrosia pollen emission in COSMO-ART compared to the emission based on the sigmoid functions in all these three countries. However, a time-related explanatory variable must be used. This suggests that the ANN-based emission parameterizations can be used at distant locations as well. On the other hand, the use of meteorological related parameters did not increase the performance compared with the time-only benchmark. • Artificial Neural Networks allowed to substantially improve the Ambrosia pollen emission in COSMO-ART. • The emission parametrization should include a variable representing time. • Meteorological parameters did not significantly improve the time-only Ambrosia emission parameterization. [ABSTRACT FROM AUTHOR]