Search

Understanding air quality requires a comprehensive understanding of its various factors. Most of the association rule techniques focuses on high frequency terms, ignoring the potential importance of low- frequency terms and causing unnecessary storage space waste. Therefore, a dynamic genetic association rule mining algorithm is proposed in this paper, which combines the improved dynamic genetic algorithm with the association rule mining algorithm to realize the importance mining of low- frequency terms. Firstly, in the chromosome coding phase of genetic algorithm, an innovative multi-information coding strategy is proposed, which selectively stores similar values of different levels in one storage unit. It avoids storing all the values at once and facilitates efficient mining of valid rules later. Secondly, by weighting the evaluation indicators such as support, confidence and promotion in association rule mining, a new evaluation index is formed, avoiding the need to set a minimum threshold for high-interest rules. Finally, in order to improve the mining performance of the rules, the dynamic crossover rate and mutation rate are set to improve the search efficiency of the algorithm. In the experimental stage, this paper adopts the 2016 annual air quality data set of Beijing to verify the effectiveness of the unit point multi-information coding strategy in reducing the rule storage air, the effectiveness of mining the rules formed by the low frequency item set, and the effectiveness of combining the rule mining algorithm with the swarm intelligence optimization algorithm in terms of search time and convergence. In the experimental stage, this paper adopts the 2016 annual air quality data set of Beijing to verify the effectiveness of the above three aspects. The unit point multi-information coding strategy reduced the rule space storage consumption by 50%, the new evaluation index can mine more interesting rules whose interest level can be up to 90%, while mining the rules formed by the lower frequency terms, and in terms of search time, we reduced it about 20% compared with some meta-heuristic algorithms, while improving convergence. [ABSTRACT FROM AUTHOR]

Air pollution poses a grave threat to human health and everyday life. Accurate air-quality prediction plays a crucial role in effectively preventing and controlling air pollution. A multi-graph spatial-temporal attention network is proposed to predict the air quality in a given area by analyzing the interconnections between stations and the individual characteristics of each station. Firstly, this paper constructs multi-scale spatial-temporal graphs from spatial and temporal perspectives to effectively capture the interconnections between stations, thereby comprehensively understanding the spatial-temporal relationships among them. Secondly, incorporating the nonlinear temporal correlation of station data, this paper proposes a temporal multi-graph attention-fusion module to integrate information from both neighboring stations and the station itself. Finally, predictions are made using a spatial-temporal graph network. The experiments in this paper were based on air-quality data from Beijing and Tianjin, and the following experimental results were obtained. (1) The proposed model outperforms the baselines significantly in both single-step and multi-step prediction tasks. (2) The ablation study confirms that the graph constructed in this paper contributes to improving the performance of the model. (3) Comparing the performance of attention components, the proposed attention mechanism exhibits better performance than other attention mechanisms. [ABSTRACT FROM AUTHOR]

Air quality has a significant impact on human health, in order to alleviate the air pollution and improve the ability to predict the air quality. In this paper, a prediction model of air quality index composed of variational mode decomposition and temporal convolutional network was proposed. First, in order to reduce the non-stationarity and randomness of the time series, the original air quality index sequence was decomposed by variational mode decomposition, and the decomposition number was determined by the central frequency method to decompose into multiple relatively stable sub-sequences with different frequency scales. Then, the decomposed sub-stable sequence was predicted by the time convolutional network. Finally, the prediction data were integrated and reconstructed to obtain the final prediction results. Comparing the results of other forecasting models by performance evaluation metrics, the combined forecasting model proposed in this paper reduced RMSE by 20.9%, 19.2%, 5.1%, 29.9%, 23.7% on the Beijing dataset. MAPE reduced by 26.6%, 22.3%, 19.5%, 28.9%, 15.0%, respectively. MAE decreased by 19.1%, 20.6%, 9.6%, 29.5%, 23.5%. R2 increased by 4.6%, 4.0%, 0.8%, 14.9%, 5.5% respectively. This proves the accuracy and reliability of the proposed model. [ABSTRACT FROM AUTHOR]

The profound impacts of severe air pollution on human health, ecological balance, and economic stability are undeniable. Precise air quality forecasting stands as a crucial necessity, enabling governmental bodies and vulnerable communities to proactively take essential measures to reduce exposure to detrimental pollutants. Previous research has primarily focused on predicting air quality using only time-series data. However, the importance of remote-sensing image data has received limited attention. This paper proposes a new multi-modal deep-learning model, Res-GCN, which integrates high spatial resolution remote-sensing images and time-series air quality data from multiple stations to forecast future air quality. Res-GCN employs two deep-learning networks, one utilizing the residual network to extract hidden visual information from remote-sensing images, and another using a dynamic spatio-temporal graph convolution network to capture spatio-temporal information from time-series data. By extracting features from two different modalities, improved predictive performance can be achieved. To demonstrate the effectiveness of the proposed model, experiments were conducted on two real-world datasets. The results show that the Res-GCN model effectively extracts multi-modal features, significantly enhancing the accuracy of multi-step predictions. Compared to the best-performing baseline model, the multi-step prediction's mean absolute error, root mean square error, and mean absolute percentage error increased by approximately 6%, 7%, and 7%, respectively. [ABSTRACT FROM AUTHOR]

In recent years, China has implemented several measures to improve air quality. The Beijing-Tianjin-Hebei (BTH) region is one area that has suffered from the most serious air pollution in China and has undergone huge changes in air quality in the past few years. How to scientifically assess these change processes remain the key issue in further improving the air quality over this region in the future. To evaluate the changes in major air pollutant emissions over this region, this paper employs ensemble Kalman filtering (EnKF) for integrating the national ground monitoring pollutant observation data and the Nested Air Quality Prediction Modeling System (NAQPMS) simulation data to inversely estimate the emission rates of SO2, NOX, CO, and primary PM2.5 over BTH region in February from 2014 to 2019. The results show that SO2, NOX, CO, and primary PM2.5 emissions in the BTH region decreased in February from 2014 to 2019 by 83%, 37%, 41%, and 42%, while decreases in Beijing during this period were 86%, 67%, 59%, and 65%, respectively. Compared with the prior emission inventory, the inversion emission inventory reduces the uncertainty of multi-pollutant simulation in the BTH region, with simulated root mean square errors of the monthly average concentrations of SO2, NOX, PM2.5, and CO reduced by 41%, 30%, 31%, and 22%, respectively. The average uncertainties of SO2, NOX, PM2.5, and CO inversion emissions in 2014–19 are ±14.03% yr−1, ±28.91% yr−1, ±126.15% yr−1, and ±43.58% yr−1. Compared with the uncertainty of MEIC emission, the uncertainties of all species changed by +2% yr−1, −2% yr−1, −26% yr−1, and −4% yr−1, respectively. The spatial distribution results illustrate that air pollutant emissions are mainly distributed over the eastern and southern BTH regions. The spatial gap between the inversion emissions and MEIC emissions was further closed in 2019 compared to 2014. The results of this paper can provide a new reference for assessing changes in air pollution emissions over the BTH region in recent years and validating a bottom-up emission inventory. [ABSTRACT FROM AUTHOR]

Aerosol optical depth (AOD) is the key parameter for determining the aerosol radiative effects and air quality variation. It is important to quantify nighttime aerosols using satellite-based night light images to understand their diurnal variations. This study selected high-resolution low light images from the Glimmer Imager (GLI) aboard the SDGSAT-1 satellite to examine spatial–temporal changes in night light emitted from the urban surface of Beijing. The radiance observed by SDGSAT-1/GLI was used to discern the AOD changes using the radiance background method (RB) and standard deviation method (SD) based on the characterization of the radiance from artificial light sources. Cloud cleaning processes were conducted to reduce the influence of the cloud cover in the glimmer images of the derived AOD. The results showed that there are good consistencies between the derived AOD results from the remote sensing and nighttime site observations. The radiance background method is better than the standard deviation method for deriving AOD using SDGSAT-1/GLI with the RMSE of its RB (0.0984) being greater than that of the SD (0.7653). The influence of moonlight, atmospheric absorption, and positioning errors on the results is briefly discussed. This paper shows that SDGSAT-1 can obtain relatively reliable night AOD values based on our investigations using the available satellite images taken in winter and spring, and that it has the potential to provide the scientific products of nighttime AOD. [ABSTRACT FROM AUTHOR]

Air quality plays a vital role in people's health, and air quality forecasting can assist in decision making for government planning and sustainable development. In contrast, it is challenging to multi-step forecast accurately due to its complex and nonlinear caused by both temporal and spatial dimensions. Deep models, with their ability to model strong nonlinearities, have become the primary methods for air quality forecasting. However, because of the lack of mechanism-based analysis, uninterpretability forecasting makes decisions risky, especially when the government makes decisions. This paper proposes an interpretable variational Bayesian deep learning model with information self-screening for PM2.5 forecasting. Firstly, based on factors related to PM2.5 concentration, e.g., temperature, humidity, wind speed, spatial distribution, etc., an interpretable multivariate data screening structure for PM2.5 forecasting was established to catch as much helpful information as possible. Secondly, the self-screening layer was implanted in the deep learning network to optimize the selection of input variables. Further, following implantation of the screening layer, a variational Bayesian gated recurrent unit (GRU) network was constructed to overcome the complex distribution of PM2.5 and achieve accurate multi-step forecasting. The high accuracy of the proposed method is verified by PM2.5 data in Beijing, China, which provides an effective way, with multiple factors for PM2.5 forecasting determined using deep learning technology. [ABSTRACT FROM AUTHOR]

Atmospheric oxidizing capacity (AOC) is an essential driving force of troposphere chemistry and self-cleaning, but the definition of AOC and its quantitative representation remain uncertain. Driven by national demand for air pollution control in recent years, Chinese scholars have carried out studies on theories of atmospheric chemistry and have made considerable progress in AOC research. This paper will give a brief review of these developments. First, AOC indexes were established that represent apparent atmospheric oxidizing ability (AOIe) and potential atmospheric oxidizing ability (AOIp) based on aspects of macrothermodynamics and microdynamics, respectively. A closed study refined the quantitative contributions of heterogeneous chemistry to AOC in Beijing, and these AOC methods were further applied in Beijing-Tianjin-Hebei and key areas across the country. In addition, the detection of ground or vertical profiles for atmospheric OH·, HO 2 ·, NO 3 · radicals and reservoir molecules can now be obtained with domestic instruments in diverse environments. Moreover, laboratory smoke chamber simulations revealed heterogeneous processes involving reactions of O 3 and NO 2 , which are typical oxidants in the surface/interface atmosphere, and the evolutionary and budgetary implications of atmospheric oxidants reacting under multispecies, multiphase and multi-interface conditions were obtained. Finally, based on the GRAPES-CUACE adjoint model improved by Chinese scholars, simulations of key substances affecting atmospheric oxidation and secondary organic and inorganic aerosol formation have been optimized. Normalized numerical simulations of AOIe and AOIp were performed, and regional coordination of AOC was adjusted. An optimized plan for controlling O 3 and PM 2.5 was analyzed by scenario simulation. [Display omitted] [ABSTRACT FROM AUTHOR]

Accurate monitoring of air quality can no longer meet people's needs. People hope to predict air quality in advance and make timely warnings and defenses to minimize the threat to life. This paper proposed a new air quality spatiotemporal prediction model to predict future air quality and is based on a large number of environmental data and a long short-term memory (LSTM) neural network. In order to capture the spatial and temporal characteristics of the pollutant concentration data, the data of the five sites with the highest correlation of time-series concentration of PM2.5 (particles with aerodynamic diameter ≤2.5 mm) at the experimental site were first extracted, and the weather data and other pollutant data at the same time were merged in the next step, extracting advanced spatiotemporal features through long- and short-term memory neural networks. The model presented in this paper was compared with other baseline models on the hourly PM2.5 concentration data set collected at 35 air quality monitoring sites in Beijing from January 1, 2016, to December 31, 2017. The experimental results show that the performance of the proposed model is better than other baseline models. [ABSTRACT FROM AUTHOR]

To study urban air quality, this paper proposes a novel categorical time series model, which is based on a linear combination of bounded Poisson distribution and discrete distribution to describe the dynamic and systemic features of air quality, respectively. Daily air quality level data of three major cities in China, including Beijing, Shanghai and Guangzhou, are analyzed. It is concluded that the air quality in Beijing is the worst among the three cities but is gradually improving, and its dynamics is also the most pronounced. Theoretically, the design of our model increases the flexibility of the probabilistic structure while ensuring a dynamic feedback mechanism without high computational stress. We estimate the parameters through an adaptive Bayesian Markov chain Monte Carlo sampling scheme and show the satisfactory finite sample performance of the model through simulation studies. [ABSTRACT FROM AUTHOR]

Research on the enhanced control and emission-reduction measures to improve air quality during major events could provide data theory and scientific support for air-quality improvement during non-activities. Based on the air-quality data published by the China Environmental Monitoring Station and the meteorological elements and weather conditions released by the China Meteorological Administration, this paper explored the characteristics of air-quality evolution in Beijing from 5 August to 18 September 2015 and the weather situation during the Military Parade. The results showed that: (1) Emission-reduction measures implemented for air quality by Beijing and its surrounding area were induced, and we explored the contribution of these measures to pollutants or AQI in the locality. (2) During the 2015 Military Parade, Beijing was in the front or lower part of the high-pressure system. Due to the strong effect of North or Northeast winds, the weather situation was conducive to the diffusion of pollutants. When before or after the implementation, once the atmospheric diffusion was poor, the pollutants would accumulate gradually. Thus, it can be seen that the weather situation had a great impact on air quality. (3) During the implementation, PM2.5, PM10, NO2 and other pollutants decreased significantly, of which the concentration of PM10 decreased the most, from 109 μg·m−3 down to 34 μg·m−3, and the concentration of PM2.5 decreased by 72.73%. According to the changes between before and during the implementation or during and after the implementation, the concentration of PM10 and PM2.5 increased when the implementation of the emission-reduction measures had been finished, indicating that the enhanced control measures made a great contribution to the emission reduction in particles. (4) In addition, the annual average of AQI in the three years is 87.49, and the average value of a normal year was the average value of 2013 and 2014. The average value of the normal year during the military parade is 64.63, which was 70.40% lower than the average value of AQI during the military parade. The goal of reaching the secondary standard of GB-3095-2012 was achieved, and there was still a long way to go from the primary standard. In a few words, in order to achieve the goal of better air quality throughout the year, all parties still needed to coordinate control and make joint efforts. [ABSTRACT FROM AUTHOR]

In recent years, the Beijing–Tianjin–Hebei region has become one of the worst areas for haze pollution in China. Sun photometers are widely used for aerosol optical property monitoring due to the advantages of fully automatic acquisition, simple maintenance, standardization of data processing, and low uncertainty. Research sites are mostly concentrated in cities, while the long-term analysis of aerosol optical depth (AOD) for the pollution transmission channel in rural Beijing is still lacking. Here, we obtained an AOD monitoring dataset from August 2017 to March 2019 using the ground-based CE-318 sun photometer at the Gucheng meteorological observation site in southwest Beijing. These sun photometer AOD data were used for the ground-based validation of MODIS (Moderate Resolution Imaging Spectroradiometer) and AHI (Advanced Himawari Imager) AOD data. It was found that MODIS and AHI can reflect AOD variation trends by sun photometer on daily, monthly, and seasonal scales. The original AOD measurements of the sun photometer show good correlations with satellite observations by MODIS (R = 0.97), and AHI (R = 0.89), respectively, corresponding to their different optimal spatial and temporal windows for matching with collocated satellite ground pixels. However, MODIS is less stable for aerosols of different concentrations and particle sizes. Most of the linear regression intercepts between the satellite and the photometer are less than 0.1, indicating that the errors due to surface reflectance in the inversion are small, and the slope is least biased (AHI: slope = 0.91, MODIS: slope = 0.18) in the noon period (11 a.m.–2 p.m.) and most biased in summer (AHI: slope = 0.77, MODIS: slope = 1.31), probably due to errors in the aerosol model. The daily and seasonal variation trends between CE-318 AOD measurements in the Gucheng site and fine particulate observations from the national air quality site nearby were also compared and investigated. In addition, a typical haze–dust complex pollution event in North China was analyzed and the changes in AOD during the pollution event were quantified. In processing, we use sun photometer and satellite AOD data in combination with meteorological and PM data. Overall, this paper has implications for the study of AOD evolution patterns at different time scales, the association between PM2.5 concentrations and AOD changes, and pollution monitoring. [ABSTRACT FROM AUTHOR]

Based on Beijing's air quality data and socio-economic data from 2015 to 2020, Kernel density estimation, spatial autocorrelation and random forest methods were used in this paper to analyze the air quality characteristics and influencing factors in Beijing from 2015 to 2020. The results showed that the concentration of particulate matter contributed the most to the change of AQI in six pollution components of air quality; the difference in air quality in the four seasons was decreasing; air quality was obviously related to social production activities and governance rules. As a whole, the Air Quality Index(AQI) in the north of the municipal area was low, while the air quality in the south was relatively high. The increase in per capita disposable income, green coverage area and scientific research funding could effectively reduce the degree of air pollution. [ABSTRACT FROM AUTHOR]

Air pollution caused by coal burning not only increases the cost of environmental pollution but also harms human health. It is urgent for China to change the practice of coal-fired central heating. Therefore, the effectiveness and sustainability of the Coal to Gas and Electricity policy have become the focus of all sectors of society. In this paper, eight cities in the Beijing–Tianjin–Hebei region were taken as the experimental groups and the other eleven cities as the control groups. Based on the PSM-DID model and the time-varying DID model, a quasi-natural experimental analysis was conducted to evaluate the effect of the policy of coal to gas and electricity to improve air quality in the Beijing–Tianjin–Hebei region from 2015 to 2020 and to test the sustainability of the policy. Three research conclusions are shown below: First, during the implementation of the policy, especially in 2019, the AQI index decreased significantly. Although there was a rebound thereafter, it was still lower than before. This shows that the Coal to Gas and Electricity policy has indeed improved the air quality in Beijing, Tianjin, and Hebei during its implementation. Second, the policy had a great impact on SO2 and PM10 but was relatively weak on PM2.5 and CO. Therefore, there is an urgent need to formulate scientific and accurate policies to control different air pollutants. Third, the time-varying DID model was used to identify the dynamic sustainability effect of the Coal to Gas and Electricity policy. The results showed that the policy had a strong impact in the initial stage, but its effect was greatly reduced at the end of the implementation or near the end, when it was far less obvious than in the initial stage of the policy. Therefore, in formulating relevant measures to reduce air pollution, it is necessary to fully consider the sustainability of the policy. [ABSTRACT FROM AUTHOR]

Aerosol Optical Depth (AOD) is one of the most important parameters of aerosol and a key physical quantity to characterize atmospheric turbidity and air pollution. Accurate retrieval of AOD is of great significance for air quality assessment. However, the spatial resolution of the currently widely used Moderate Resolution Imaging Spectroradiometer (MODIS) AOD products is too low to meet the application research of atmospheric environment at the regional scale. In 2013, China launched the Gaofen-1 (GF-1) satellite, which provides a new idea for AOD retrieval. In this paper, we apply the synergetic use of TERRA and AQUA satellite MODIS data to calculate the high-resolution AOD over Beijing based on the Synergetic Retrieval of Aerosol Properties algorithm (SRAP) and discussed scale conversion problems between AODs with different resolutions. To obtain the 100 m MODIS data, we use GF-1 wide-field-of-view data to downscale 1 km MODIS data based on mutual information method. The retrieved AOD has a spatial resolution of 100 m and can cover many land surface types. Preliminary validation was carried out with the Aerosol Robotic Network (AERONET) ground observation data. The correlation coefficient is about 0.88, and the root-mean-square error is about 0.15. Due to the high resolution of retrieved results, more detailed features can be provided in the spatial distribution. The experimental results show that the method has high precision, and further verification work is continuing. [ABSTRACT FROM AUTHOR]

The application of deep learning in time-series prediction has developed gradually. In this paper, we propose a deep generative network model for feature extraction of multivariate time series, namely, mutual information variational autoencoders (MI-VAE). In the architecture of the proposed model, we use the latent space of VAE for feature learning, which can extract the essential features of multivariate time-series data effectively. The latent space employed directly as a feature extractor can avoid poor interpretability of model. In addition, we introduce a mutual information term into the loss function, which improves the expression capability and accuracy of model. The proposed model, combining the merits of VAE and mutual information, extracts features for multivariate time-series data from a new perspective. The Lorenz system and Beijing air quality time series are used to test performance of the proposed model and comparative models. Results show that the proposed model is superior to other similar models in terms of accuracy and expression capability of latent space. [ABSTRACT FROM AUTHOR]

Microorganisms are an important part of atmospheric particulate matter and are closely related to human health. In this paper, the variations in the characteristics of the chemical components and bacterial communities in PM10 and PM2.5 grouped according to season, pollution degree, particle size, and winter heating stage were studied. The influence of environmental factors on community structure was also analyzed. The results showed that seasonal variations were significant. NO3− contributed the most to the formation of particulate matter in spring and winter, while SO42− contributed the most in summer and autumn. The community structures in summer and autumn were similar, while the community structure in spring was significantly different. The dominant phyla were similar among seasons, but their proportions were different. The dominant genera were no-rank_c_Cyanobacteria, Acidovorax, Escherichia-Shigella and Sphingomonas in spring; Massilia, Bacillus, Acinetobacter, Rhodococcus, and Brevibacillus in summer and autumn; and Rhodococcus in winter. The atmospheric microorganisms in Beijing mainly came from soil, water, and plants. The few pathogens detected were mainly affected by the microbial source on the sampling day, regardless of pollution level. RDA (redundancy analysis) showed that the bacterial community was positively correlated with the concentration of particulate matter and that the wind speed in spring was positively correlated with NO3− levels, NH4+ levels, temperature, and relative humidity in summer and autumn, but there was no clear consistency among winter samples. This study comprehensively analyzed the variations in the characteristics of the airborne bacterial community in Beijing over one year and provided a reference for understanding the source, mechanism, and assessment of the health effects of different air qualities. [ABSTRACT FROM AUTHOR]

Compared with mechanism-based modeling methods, data-driven modeling based on big data has become a popular research field in recent years because of its applicability. However, it is not always better to have more data when building a forecasting model in practical areas. Due to the noise and conflict, redundancy, and inconsistency of big time-series data, the forecasting accuracy may reduce on the contrary. This paper proposes a deep network by selecting and understanding data to improve performance. Firstly, a data self-screening layer (DSSL) with a maximal information distance coefficient (MIDC) is designed to filter input data with high correlation and low redundancy; then, a variational Bayesian gated recurrent unit (VBGRU) is used to improve the anti-noise ability and robustness of the model. Beijing's air quality and meteorological data are conducted in a verification experiment of 24 h PM2.5 concentration forecasting, proving that the proposed model is superior to other models in accuracy. [ABSTRACT FROM AUTHOR]

This paper examines whether mega-events-initiated planning regulations improved air quality in Chinese cities and explores the driving forces of the enforcement of such regulations. Using the 2008 Beijing Olympics as an example, we find that mega-events serve as an opportunity for cities to combat air pollution. The 2008 Olympics prompted a handful of Chinese cities to cut pollution and sustain a blue sky: Cities with air quality regulated for the Olympics cut their Air Pollution Index by about 16 points during the Games, compared to non-regulated cities, and 60% of that effect remained four years after the event. These achievements are obtained through effective mobilization of city leaders by associating air quality with their political careers. This study reveals that 1) a mega-event may improve urban environmental quality beyond the host cities and the event period, and 2) successful implementation of environmental regulations hinges on incentivizing local leaders. [ABSTRACT FROM AUTHOR]

Beijing-Tianjin-Hebei metropolitan area is a significant carbon emission center. The region's early achievement of peak carbon targets is critical to the nation's achievement of peak carbon targets. In this paper, it is proposed to use different orders of grey models to classify into three scenarios. Based on three scenarios, the grey multivariate convolutional model with new information priority accumulation is adopted to predict carbon emissions in the Beijing-Tianjin-Hebei region and select the scenario suitable for local development. The results show that: (1) The Beijing region has already achieved peak carbon, the Tianjin region may not reach its peak carbon target by 2030, and the Hebei region is expected to reach its peak carbon target by 2030. (2) The high rate of carbon emission reduction scenario will greatly improve the air quality of Beijing. The low-speed growth carbon emission scenario is more in line with the future development of Tianjin city. The low-rate carbon reduction scenario is more in line with the synergistic governance of pollution reduction and carbon reduction in Hebei Province. (3) Beijing's population policy in the most recent years has been conducive to improving the local environment. Tianjin's medium-term population policy is more in line with the local area. Hebei's medium-term industrial structure reform is favorable to local development. • Grey correlation analysis is used to analyze the factors affected carbon emissions in a region. • The factors with high correlation are predicted by using grey prediction models of different orders. • The high rate of carbon emission reduction scenario will greatly improve the air quality of Beijing. [ABSTRACT FROM AUTHOR]

• An imputation model imputes specific missing values in multivariate time series. • Our method outperforms the comparison method when specific values are missing. • Our experiments provide selection principles of hyperparameters. • Our experiment determines the best training method with introduction of noise. Data missing is a ubiquitous phenomenon in multivariate time series analysis because of failure measurements, improper installation or other unfavorable factors. Most imputation methods, such as statistical methods and machine learning methods, are mainly for dealing with random or continuous missing, but do not take into consideration the non-random specific missing situation. This paper proposes an imputation model based on the variational auto-encoders (VAE) and shift correction for specific missing values, which is also extended to the β -VAE model. The shift correction is used to correct the original probability distribution deviation caused by specific values concentrated missing. Then taking the meteorological and air quality data sets in Beijing as the example, this paper mainly explores two common specific missing situations, including large values missing and small values missing. The experimental results show that the proposed model can effectively impute specific missing values, which has high imputation accuracy and robustness. [ABSTRACT FROM AUTHOR]

Long short-term memory (LSTM), especially vanilla LSTM (VLSTM), has been widely used in air quality prediction field. However, VLSTM has many more parameters, thereby making training slow and prediction performance unstable. The VLSTM network input data have not been selected for better efficiency. In this paper, we propose an air quality prediction model based on the improved VLSTM with multichannel input and multiroute output (IVLSTM-MCMR). The proposed model includes the IVLSTM and MCMR modules. The proposed IVLSTM module is developed by improving the VLSTM inner structure of VLSTM in order to reduce the number of parameters that help to accelerate the convergence. A new historical information usage approach is further proposed to obtain a stable training process. For the MCMR module, a multichannel data input model (MC) with an improved linear similarity dynamic time warping is introduced to choose the valid data as the input of IVLSTM. A multiroute output model (MR) is designed to integrate the results from MC, in which the results of different target stations with different features are output by different routes. We evaluate the proposed model with the collected data from Beijing, China, and the experimental results show that our model achieves improvements regarding the predication performance. • An air quality prediction model IVLSTM-MCMR is proposed based on deep learning. • The number of parameters in IVLSTM-MCMR is reduced to accelerate the convergence. • The improved linear similarity dynamic time warping is introduced in IVLSTM-MCMR. • The integration of multi-channel data input and multi-route output is designed. [ABSTRACT FROM AUTHOR]

Predicting the air quality index (AQI) has been regarded as a critical problem for environmental control management. Many factors over time and space may relate to the diffusion of pollutants. In other words, there exist very intricate spatio-temporal interactions among the characteristic for revealing diffusion of pollutants. Recently, some relevant works studied the topic of AQI prediction considering spatial and temporal correlations simultaneously, but most of them either ignore geospatially topological structures to learn spatio-temporal dependency or utilize sub-modules separately to encode the spatial and temporal information. Unfortunately, ignoring geospatially topological structures or correlations among spatial properties and temporal dependencies leads that the AQI prediction model cannot deal with the prediction task well. Although few existing works utilize an integrated model to encode the spatial and temporal information with geospatially topological structures, these works focus only on catching inter-station relationships among stations, which only can reflect the spatio-temporal dependency. In this paper, we propose a spatio-attention embedded recurrent neural network, called SpAttRNN , for air quality prediction by comprehensively leveraging dynamic spatio-temporal correlations among monitoring stations to tackle the intricate spatio-temporal interactions. To capture spatial properties, a self-loop-normalized adjacency matrix is utilized to enhance graph-based attention cells to learn the relationships between monitoring stations. Moreover, the proposed graph-based attention cell is embedded into a recurrent neural network to learn dynamic spatio-temporal correlations from multiple monitoring stations simultaneously. We conducted comprehensive experimental evaluations on two real-world air quality datasets collected from Beijing's real-time air quality monitoring system. The results demonstrate that our method outperforms other state-of-the-art methods significantly. In particular, SpAttRNN delivers up to 15% improvement in terms of the mean square error (MAE) on PM 2.5 prediction compared with other existing methods. [ABSTRACT FROM AUTHOR]

Air contamination is identified with individuals' wellbeing and furthermore affects the sustainable development of economy and society. This paper gathered the time series data of seven meteorological conditions variables of Beijing city from 1 November 2013 to 31 October 2017 and utilized the generalized regression neural network optimized by the particle swarm optimization algorithm (PSO-GRNN) to explore seasonal disparity in the impacts of mean atmospheric humidity, maximum wind velocity, insolation duration, mean wind velocity and rain precipitation on air quality index (AQI). The results showed that in general, the most significant impacting factor on air quality in Beijing is insolation duration, mean atmospheric humidity, and maximum wind velocity. In spring and autumn, the meteorological diffusion conditions represented by insolation duration and mean atmospheric humidity had a significant effect on air quality. In summer, temperature and wind are the most significant variables influencing air quality in Beijing; the most important reason for air contamination in Beijing in winter is the increase in air humidity and the deterioration of air diffusion condition. This study investigates the seasonal effects of meteorological conditions on air contamination and suggests a new research method for air quality research. In future studies, the impacts of different variables other than meteorological conditions on air quality should be assessed. [ABSTRACT FROM AUTHOR]