Neural network and multiple regression models for PM10 prediction in Athens: a comparative assessment

Particulate atmospheric pollution in urban areas is considered to have significant impact on human health. Therefore, the ability to make accurate predictions of particulate ambient concentrations is important to improve public awareness and air quality management. This study examines the possibility of using neural network methods as tools for daily average particulate matter with aerodynamic diameter <10 microm (PM10) concentration forecasting, providing an alternative to statistical models widely used up to this day. Based on a data inventory, in a fixed central site in Athens, Greece, ranging over a two-year period, and using mainly meteorological variables as inputs, neural network models and multiple linear regression models were developed and evaluated. Comparison statistics used indicate that the neural network approach has an edge over regression models, expressed both in terms of prediction error (root mean square error values lower by 8.2-9.4%) and of episodic prediction ability (false alarm rate values lower by 7-13%). The results demonstrate that artificial neural networks (ANNs), if properly trained and formed, can provide adequate solutions to particulate pollution prognostic demands.     

Neural Network and Multiple Regression Models for PM10 Prediction in Athens: A Comparative Assessment

Abstract

Particulate atmospheric pollution in urban areas is considered to have significant impact on human health. Therefore, the ability to make accurate predictions of particulate ambient concentrations is important to improve public awareness and air quality management. This study examines the possibility of using neural network methods as tools for daily average particulate matter with aerodynamic diameter <10 µm (PM₁₀) concentration forecasting, providing an alternative to statistical models widely used up to this day. Based on a data inventory, in a fixed central site in Athens, Greece, ranging over a two-year period, and using mainly meteorological variables as inputs, neural network models and multiple linear regression models were developed and evaluated. Comparison statistics used indicate that the neural network approach has an edge over regression models, expressed both in terms of prediction error (root mean square error values lower by 8.2–9.4%) and of episodic prediction ability (false alarm rate values lower by 7–13%). The results demonstrate that artificial neural networks (ANNs), if properly trained and formed, can provide adequate solutions to particulate pollution prognostic demands.     

Forecasting Daily Source Air Quality Using Multivariate Statistical Analysis and Radial Basis Function Networks

Abstract

It is vital to forecast gas and particle matter concentrations and emission rates (GPCER) from livestock production facilities to assess the impact of airborne pollutants on human health, ecological environment, and global warming. Modeling source air quality is a complex process because of abundant nonlinear interactions between GPCER and other factors. The objective of this study was to introduce statistical methods and radial basis function (RBF) neural network to predict daily source air quality in Iowa swine deep-pit finishing buildings. The results show that four variables (outdoor and indoor temperature, animal units, and ventilation rates) were identified as relative important model inputs using statistical methods. It can be further demonstrated that only two factors, the environment factor and the animal factor, were capable of explaining more than 94% of the total variability after performing principal component analysis. The introduction of fewer uncorrelated variables to the neural network would result in the reduction of the model structure complexity, minimize computation cost, and eliminate model overfitting problems. The obtained results of RBF network prediction were in good agreement with the actual measurements, with values of the correlation coefficient between 0.741 and 0.995 and very low values of systemic performance indexes for all the models. The good results indicated the RBF network could be trained to model these highly nonlinear relationships. Thus, the RBF neural network technology combined with multivariate statistical methods is a promising tool for air pollutant emissions modeling.     

Impacts of Meteorological Factors on PM10: Artificial Neural Networks (ANN) and Multiple Linear Regression (MLR) Approaches

In this study, prediction capacities of multi-linear regression (MLR) and artificial neural networks (ANN) onto coarse particulate matter (PM₁₀) concentrations were investigated. Different meteorological factors on particulate pollution were chosen for operating variables in the model analyses. Two different regions (urban and industrial) were identified in the region of Kocaeli, Turkey. All data sets were obtained from air quality monitoring network of the Ministry of Environment and Urban Planning, and 120 data sets were used in the MLR and ANN models. Regression equations explained the effects of the meteorological factors in MLR analyses. In the ANN model, backpropagation network with two hidden layers has achieved the best prediction efficiency. Determination coefficients and error values were examined for each model. ANN models displayed more accurate results compared to MLR.     

Neural network-based study for predicting ground-level ozone concentration in large urban areas, applied to the Sao Paulo metropolitan area

Events of high concentration of ground-level ozone constitute a matter of major concern in large urban areas in terms of air quality, and public health. In the Sao Paulo Metropolitan Area (SPMA), air quality data generated by a network of air quality measuring stations have been used in a number of studies correlating ozone formation with different variables. A study was carried out on the application of neural network models in the identification of typical sceneries leading to high ground-level ozone concentrations in the SPMA. The results were then applied in the selection of variables, and in the definition of neural network-based models for estimating ozone levels from meteorological variables. When combined with existing weather prediction tools, the models can be applied in the prediction of ozone levels in the SPMA     

基于指标规范值的室内空气客观评价的概率神经网络模型

为了建立简单、普适、通用的概率神经网络的室内空气评价模型,在适当设定室内空气各项指标的参照值及指标值的规范变换式基础上,使室内空气同级标准不同指标的规范值差异尽可能小,从而用规范值表示的各指标都可用同一个规范指标"等效"替代。因此,概率神经网络隐层各类模式的基函数中心矢量的各指标分量值与同级标准所有15项指标规范值的均值等同。将基于指标规范值的概率神经网络模型用于室内空气的评价实例进行检验,验证了该模型的普适性、通用性和简便性。     

A Simple Neural Network for Estimating Emission Rates of Hydrogen Sulfide and Ammonia from Single Point Sources

Abstract

Neural networks have shown tremendous promise in modeling complex problems. This work describes the development and validation of a neural network for the purpose of estimating point source emission rates of hazardous gases. This neural network approach has been developed and tested using experimental data obtained for two specific air pollutants of concern in West Texas, hydrogen sulfide and ammonia. The prediction of the network is within 20% of the measured emission rates for these two gases at distances of less than 50 m. The emission rate estimations for ground level releases were derived as a function of seven variables: downwind distance, crosswind distance, wind speed, downwind concentration, atmospheric stability, ambient temperature, and relative humidity. A backpropagation algorithm was used to develop the neural network and is also discussed here. The experimental data were collected at the Wind Engineering Research Field Site located at Texas Tech University in Lubbock, Texas. Based on the results of this study, the use of neural networks provides an attractive and highly effective tool to model atmospheric dispersion, in which a large number of variables interact in a nonlinear manner.     

Combining neural network models to predict spatial patterns of airborne pollutant accumulation in soils around an industrial point emission source

Neural networks (NNs) have the ability to model a wide range of complex nonlinearities. A major disadvantage of NNs, however, is their instability, especially under conditions of sparse, noisy, and limited data sets. In this paper, different combining network methods are used to benefit from the existence of local minima and from the instabilities of NNs. A nonlinear k-fold cross-validation method is used to test the performance of the various networks and also to develop and select a set of networks that exhibits a low correlation of errors. The various NN models are applied to estimate the spatial patterns of atmospherically transported and deposited lead (Pb) in soils around an historical industrial air emission point source. It is shown that the resulting ensemble networks consistently give superior predictions compared with the individual networks because, for the ensemble networks, R2 values were found to be higher than 0.9 while, for the contributing individual networks, values for R2 ranged between 0.35 and 0.85. It is concluded that combining networks can be adopted as an important component in the application of artificial NN techniques in applied air quality studies.     

Carbon monoxide concentration forecasting in Santiago, Chile

In the city of Santiago, Chile, air quality is defined in terms of particulate matter with an aerodynamic diameter < or = 10 microm (PM10) concentrations. An air quality forecasting model based on past concentrations of PM10 and meteorological conditions currently is used by the metropolitan agency for the environment, which allows restrictions to emissions to be imposed in advance. This model, however, fails to forecast between 40 and 50% of the days considered to be harmful for the inhabitants every year. Given that a high correlation between particulate matter and carbon monoxide (CO) concentrations is observed at monitoring stations in the city, a model for CO concentration forecasting would be a useful tool to complement information about expected air quality in the city. Here, the results of a neural network-based model aimed to forecast maximum values of the 8-hr moving average of CO concentrations for the next day are presented. Forecasts from the neural network model are compared with those produced with linear regressions. The neural network model seems to leave more room to adjust free parameters with 1-yr data to predict the following year's values. We have worked with 3 yr of data measured at the monitoring station located in the zone with the worst air quality in the city of Santiago, Chile.     

Neural modelling of the spatial distribution of air pollutants

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 × 1 km 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.     

Forecasts using neural network versus Box-Jenkins methodology for ambient air quality monitoring data

This study explores ambient air quality forecasts using the conventional time-series approach and a neural network. Sulfur dioxide and ozone monitoring data collected from two background stations and an industrial station are used. Various learning methods and varied numbers of hidden layer processing units of the neural network model are tested. Results obtained from the time-series and neural network models are discussed and compared on the basis of their performance for 1-step-ahead and 24-step-ahead forecasts. Although both models perform well for 1-step-ahead prediction, some neural network results reveal a slightly better forecast without manually adjusting model parameters, according to the results. For a 24-step-ahead forecast, most neural network results are as good as or superior to those of the time-series model. With the advantages of self-learning, self-adaptation, and parallel processing, the neural network approach is a promising technique for developing an automated short-term ambient air quality forecast system.     

Identification of redundant air quality measurements through the use of principal component analysis

This study aims to show how principal component analysis (PCA) can be used to identify redundant measurements in air quality monitoring networks. The minimum number of air quality monitoring sites in Oporto Metropolitan Area (Oporto-MA) was evaluated using PCA and then compared to the one settled by the legislation. Nine sites, monitoring NO₂, O₃ and PM₁₀, were selected and the air pollutant concentrations were analysed from January 2003 to December 2005. PCA was applied to the data corresponding to the first two years that were divided into annual quarters to verify the persistence of the PCA results. The number of principal components (PCs) was selected by applying two criteria: Kaiser (PCs with eigenvalues greater than 1) and ODV₉₀ (PCs representing at least 90% of the original data variance). Each pollutant was analysed separately. The two criteria led to different results. Using Kaiser criterion for the eight analysed periods, two PCs were selected in: (i) five periods for O₃ and PM₁₀; and (ii) six periods for NO₂. These PCs had important contributions of the same groups of monitoring sites. The percentage of the original data variance contained in the selected PCs using this criterion was always below 90%. Thus, the results obtained using ODV₉₀ were considered with more confidence. Using this criterion, only five monitoring sites for NO₂, three for O₃ and seven for PM₁₀ were needed to characterize the region. The number of monitoring sites for NO₂ and O₃ was in agreement with what was established by the legislation. However, for PM₁₀, Oporto-MA needed two more monitoring sites. To validate PCA results, statistical models were determined to estimate air pollutant concentrations at removed monitoring sites using the concentrations measured at the remaining monitoring sites. These models were applied to a year's data. The good performance obtained by the models showed that the monitoring sites selected by the procedure presented in this study were enough to infer the air pollutant concentrations in the region defined by the initial monitoring sites. Additionally, the air pollutant analysers corresponding to the redundant measurements can be installed in non-monitored regions, allowing the enlargement of the air quality monitoring network.     

Receptor modeling application framework for particle source apportionment

Receptor models infer contributions from particulate matter (PM) source types using multivariate measurements of particle chemical and physical properties. Receptor models complement source models that estimate concentrations from emissions inventories and transport meteorology. Enrichment factor, chemical mass balance, multiple linear regression, eigenvector. edge detection, neural network, aerosol evolution, and aerosol equilibrium models have all been used to solve particulate air quality problems, and more than 500 citations of their theory and application document these uses. While elements, ions, and carbons were often used to apportion TSP, PM10, and PM2.5 among many source types, many of these components have been reduced in source emissions such that more complex measurements of carbon fractions, specific organic compounds, single particle characteristics, and isotopic abundances now need to be measured in source and receptor samples. Compliance monitoring networks are not usually designed to obtain data for the observables, locations, and time periods that allow receptor models to be applied. Measurements from existing networks can be used to form conceptual models that allow the needed monitoring network to be optimized. The framework for using receptor models to solve air quality problems consists of: (1) formulating a conceptual model; (2) identifying potential sources; (3) characterizing source emissions; (4) obtaining and analyzing ambient PM samples for major components and source markers; (5) confirming source types with multivariate receptor models; (6) quantifying source contributions with the chemical mass balance; (7) estimating profile changes and the limiting precursor gases for secondary aerosols; and (8) reconciling receptor modeling results with source models, emissions inventories, and receptor data analyses.     

基于人工蜂群算法与BP神经网络的水质评价模型

针对BP网络水质评价模型的不足,引入人工蜂群(ABC)算法,将求解BP神经网络各层权值、阀值的过程转化为蜜蜂寻找最佳蜜源的过程,提出了一种新的结合人工蜂群算法的BP网络水质评价方法(ABC-BP)。并以2000—2006年渭河监测断面的10组实测数据作为测试样本对其水质进行了评价,实验结果表明该方法得到的水质评价结果准确,并具有很强的稳定性和鲁棒性。     

Combining Neural Network Models to Predict Spatial Patterns of Airborne Pollutant Accumulation in Soils around an Industrial Point Emission Source

Abstract

Neural networks (NNs) have the ability to model a wide range of complex nonlinearities. A major disadvantage of NNs, however, is their instability, especially under conditions of sparse, noisy, and limited data sets. In this paper, different combining network methods are used to benefit from the existence of local minima and from the instabilities of NNs. A nonlinear k-fold cross-validation method is used to test the performance of the various networks and also to develop and select a set of networks that exhibits a low correlation of errors. The various NN models are applied to estimate the spatial patterns of atmospherically transported and deposited lead (Pb) in soils around an historical industrial air emission point source. It is shown that the resulting ensemble networks consistently give superior predictions compared with the individual networks because, for the ensemble networks, R² values were found to be higher than 0.9 while, for the contributing individual networks, values for R² ranged between 0.35 and 0.85. It is concluded that combining networks can be adopted as an important component in the application of artificial NN techniques in applied air quality studies.     

Prediction of sulfur dioxide concentrations at a site near downtown Santiago,Chile

We have analyzed the possibility to predict hourly averages of sulfur dioxide concentrations in the atmosphere at a site not far from the downtown area in the city of Santiago, Chile. We have compared the forecasts produced assuming persistence, linear regressions and feed forward neural networks. The effect of meteorological conditions is included by using forecasted values of temperature, relative humidity and wind speed at the time of the intended prediction as inputs to the different models. The best predictions for hourly averages are obtained with a three-layer neural network that has hourly averages of sulfur dioxide concentrations every 6 h on the previous day plus the actual values of the meteorological variables as input. Training the network with 1995 data, error in 8 h in advance prediction for 1996 data is of the order of 30%.     

Potential assessment of the "support vector machine" method in forecasting ambient air pollutant trends

Monitoring and forecasting of air quality parameters are popular and important topics of atmospheric and environmental research today due to the health impact caused by exposing to air pollutants existing in urban air. The accurate models for air pollutant prediction are needed because such models would allow forecasting and diagnosing potential compliance or non-compliance in both short- and long-term aspects. Artificial neural networks (ANN) are regarded as reliable and cost-effective method to achieve such tasks and have produced some promising results to date. Although ANN has addressed more attentions to environmental researchers, its inherent drawbacks, e.g., local minima, over-fitting training, poor generalization performance, determination of the appropriate network architecture, etc., impede the practical application of ANN. Support vector machine (SVM), a novel type of learning machine based on statistical learning theory, can be used for regression and time series prediction and have been reported to perform well by some promising results. The work presented in this paper aims to examine the feasibility of applying SVM to predict air pollutant levels in advancing time series based on the monitored air pollutant database in Hong Kong downtown area. At the same time, the functional characteristics of SVM are investigated in the study. The experimental comparisons between the SVM model and the classical radial basis function (RBF) network demonstrate that the SVM is superior to the conventional RBF network in predicting air quality parameters with different time series and of better generalization performance than the RBF model.     

A Pilot Indoor-Outdoor Study of Organic Particulate Matter and Particulate Mutagenicity

Abstract

Many large metropolitan areas experience elevated concentrations of ground-level ozone pollution during the summertime “smog season”. Local environmental or health agencies often need to make daily air pollution forecasts for public advisories and for input into decisions regarding abatement measures and air quality management. Such forecasts are usually based on statistical relationships between weather conditions and ambient air pollution concentrations. Multivariate linear regression models have been widely used for this purpose, and well-specified regressions can provide reasonable results. However, pollution-weather relationships are typically complex and nonlinear—especially for ozone—properties that might be better captured by neural networks. This study investigates the potential for using neural networks to forecast ozone pollution, as compared to traditional regression models. Multiple regression models and neural networks are examined for a range of cities under different climate and ozone regimes, enabling a comparative study of the two approaches. Model comparison statistics indicate that neural network techniques are somewhat (but not dramatically) better than regression models for daily ozone prediction, and that all types of models are sensitive to different weather-ozone regimes and the role of persistence in aiding predictions.     

Forecasting air quality time series using deep learning

This paper presents one of the first applications of deep learning (DL) techniques to predict air pollution time series. Air quality management relies extensively on time series data captured at air monitoring stations as the basis of identifying population exposure to airborne pollutants and determining compliance with local ambient air standards. In this paper, 8 hr averaged surface ozone (O₃) concentrations were predicted using deep learning consisting of a recurrent neural network (RNN) with long short-term memory (LSTM). Hourly air quality and meteorological data were used to train and forecast values up to 72 hours with low error rates. The LSTM was able to forecast the duration of continuous O₃ exceedances as well. Prior to training the network, the dataset was reviewed for missing data and outliers. Missing data were imputed using a novel technique that averaged gaps less than eight time steps with incremental steps based on first-order differences of neighboring time periods. Data were then used to train decision trees to evaluate input feature importance over different time prediction horizons. The number of features used to train the LSTM model was reduced from 25 features to 5 features, resulting in improved accuracy as measured by Mean Absolute Error (MAE). Parameter sensitivity analysis identified look-back nodes associated with the RNN proved to be a significant source of error if not aligned with the prediction horizon. Overall, MAE's less than 2 were calculated for predictions out to 72 hours.

Implications: Novel deep learning techniques were used to train an 8-hour averaged ozone forecast model. Missing data and outliers within the captured data set were replaced using a new imputation method that generated calculated values closer to the expected value based on the time and season. Decision trees were used to identify input variables with the greatest importance. The methods presented in this paper allow air managers to forecast long range air pollution concentration while only monitoring key parameters and without transforming the data set in its entirety, thus allowing real time inputs and continuous prediction.     

基于改进型灰色神经网络组合模型的空气质量预测

基于空气质量数据不足及波动较大的情况,将灰色GM（1,1）模型与人工神经网络模型组合并改进,建立改进型灰色神经网络组合模型。利用天津市2001—2008年PM10、SO2和NO2年均值作为原始数据预测2009—2010年PM10、SO2和NO2的浓度以进行模型精度检验,最后利用该模型预测2011—2015年天津市空气质量状况。结果表明,与灰色GM（1,1）模型、传统灰色神经网络组合模型相比,所建立的改进型灰色神经网络组合模型相对模拟误差小,预测结果更为可靠,可以用于空气质量预测。