Qingchun Guo  This email address is being protected from spambots. You need JavaScript enabled to view it.1,2,3, Zhenfang He This email address is being protected from spambots. You need JavaScript enabled to view it.1,2,4, Zhaosheng Wang5

1 School of Geography and Environment, Liaocheng University, Liaocheng 252000, China
2 Institute of Huanghe Studies, Liaocheng University, Liaocheng 252000, China
3 State Key Laboratory of Loess and Quaternary Geology, Institute of Earth Environment, Chinese Academy of Sciences, Xi’an 710061, China
4 State Key Laboratory of Urban and Regional Ecology, Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China
5 National Ecosystem Science Data Center, Key Laboratory of Ecosystem Network Observation and Modeling, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, China


Received: December 11, 2022
Revised: March 12, 2023
Accepted: March 27, 2023

 Copyright The Author(s). This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are cited.


Download Citation: ||https://doi.org/10.4209/aaqr.220448  

  • Download: PDF


Cite this article:

Guo, Q., He, Z., Wang, Z. (2023). Prediction of Hourly PM2.5 and PM10 Concentrations in Chongqing City in China Based on Artificial Neural Network. Aerosol Air Qual. Res. 23, 220448. https://doi.org/10.4209/aaqr.220448


HIGHLIGHTS

  • Artificial intelligence method is used to predict hourly PM2.5.
  • Trainbr algorithm has higher prediction accuracy than trainlm algorithm.
  • The artificial neural network has good prediction ability, with high R up to 0.9773.
 

ABSTRACT


Accurate prediction of air pollution is a difficult problem to be solved in atmospheric environment research. An Artificial Neural Network (ANN) is exploited to predict hourly PM2.5 and PM10 concentrations in Chongqing City. We take PM2.5 (PM10), time and meteorological elements as the input of the ANN, and the PM2.5 (PM10) of the next hour as the output to build an ANN model. Thirteen kinds of training functions are compared to obtain the optimal function. The research results display that the ANN model exhibits good performance in predicting hourly PM2.5 and PM10 concentrations. Trainbr is the best function for predicting PM2.5 concentrations compared to other training functions with R value (0.9783), RMSE (1.2271), and MAE (0.9050). Trainlm is the second best with R value (0.9495), RMSE (1.8845), and MAE (1.3902). Similarly, trainbr is also the best in forecasting PM10 concentrations with R value (0.9773), RMSE value (1.8270), and MAE value (1.4341). Trainlm is the second best with R value (0.9522), RMSE (2.6708), and MAE (1.8554). These two training functions have good generalization ability and can meet the needs of hourly PM2.5 and PM10 prediction. The forecast results can support fine management and help improve the ability to prevent and control air pollution in advance, accurately and scientifically.


Keywords: Air pollution, Artificial neural network, Meteorological element, Predict, PM2.5, PM10, Chongqing, Training function


1 INTRODUCTION


In the past 40 years, the rapid urbanization and economic growth in China have been accompanied by dramatic changes in land use and rapid growth in pollution emissions. Air pollution has become a major environmental problem of widespread concern to the government, the public and the international community (Huang et al., 2014; Geng et al., 2021; Li et al., 2022b; Liu et al., 2022). Air pollution affects climate change, ecosystems, virus transmission, human health and sustainable socio-economic development (Crippa et al., 2022; Guo et al., 2023b). The change in air pollution time series is nonlinear, and it is affected by meteorological conditions (Guo et al., 2021, 2023a; Islam et al., 2023). Ambient PM2.5 and PM10 are the main air pollutants. In 2019, PM2.5 pollution caused about 7 million premature deaths worldwide (Fowler et al., 2020). China is one of the regions that suffer the most serious health hazards from PM2.5 pollution. PM2.5 pollution in China led to the premature deaths of about 2.1 million people in 2017 (Geng et al., 2021). Therefore, accurate prediction of air pollution is very important for protecting people's health.

Artificial intelligence (AI), such as artificial neural network (ANN) models, has been established. ANN has been widespreadly used in hydrology, confirmed cases of COVID-19, meteorology and air pollution prediction (Zaini et al., 2022; Guo et al., 2023c). In most previous studies, the Levenberg Marquardt algorithm performs better than other algorithms, and it is proved to be the fastest and powerful algorithm for solving any nonlinear least squares problems (Pakalapati et al., 2019; Perera et al., 2020). Nevertheless, Bayesian Regularization algorithm outperforms the Levenberg Marquardt in some studies (Guo et al., 2020; Nasrudin et al., 2020; Guo and He, 2021; He et al., 2022). This is because it tends to improve the generalization ability of the network by fully iterating in the training process (Pakalapati et al., 2019). In addition, the BFGS Quasi-Newton and conjugate gradient algorithms have better performance than the traditional Gradient descent algorithms. These algorithms use the second-order method, while the Gradient Descent algorithm only calculates the first-order method (Perera et al., 2020). The general algorithm can be superior to the back-propagation algorithm because it can ensure that the network is trained with the optimal weight (Awolusi et al., 2019). These training algorithms have good accuracy. Therefore, we use different training algorithms to predict hourly PM2.5 and PM10 concentrations in Chongqing City.

 
2 MATERIALS AND METHODS


 
2.1 Meteorological Data and Air Pollution Data

Chongqing, also known as Mountain City and River City, is a national central City and an economic center in the upper reaches of the Yangtze River. By the end of 2021, the total area was 82,400 square kilometers, and the permanent population is 32124300. In 2021, Chongqing's GDP is 2789.402 billion yuan. Chongqing has a subtropical monsoon humid climate, with an average annual temperature of 16–18°C, an average annual precipitation of 1000–1350 mm, and an average annual relative humidity of 70%–80%.

Meteorological data comes from China Meterological Administration. Hourly PM10 and PM2.5 concentrations come from Environmental Monitoring of China. Meteorological data, PM10 and PM2.5 concentrations are from 0:00, November 21, 2022, to 4:00, December 7, 2022. The data is divided into two parts. The training data set is from 00:00, November 21, 2022, to 13:00, December 5, 2022, and the forecasting data is from 14:00, December 5, 2022, to 04:00, December 7, 2022.

 
2.2 Artificial Neural Network

ANN consists of one or more median layers (for facilitating the convergence of the ANN), an output layer (to obtain the collected values), and an input layer (where input variable data will be collected). Artificial neural network is a branch of AI, which copies the biological training system (Goudarzi et al., 2021). The ANN for predicting hourly PM2.5 and PM10 concentrations has 3 layers (Fig. 1). The input layer includes hour (H) (n), PM2.5 (PM10) (n), air pressure (AP) (n), sea level pressure (SLP) (n), maximum air pressure (MAP) (n), minimum air pressure (MAP) (n), maximum wind speed (MWS) (n), extreme wind speed (EWS) (n), wind direction of extreme wind speed (WDEWS) (n), average wind speed (AWS) (n), wind direction of maximum wind speed (WDMWS) (n), air temperature (AT) (n), maximum air temperature (MAT) (n), relative humidity (RH) (n), minimum air temperature (MAT) (n), minimum relative humidity (MRH) (n), precipitation (P) (n), and body temperature (BT) (n). The output layer includes PM2.5 (PM10) (n + 1). Six kinds of training algorithms with thirteen various training functions were used in the research. The six algorithms are Gradient Descent (GD), Levenberg Marquardt (LM), Conjugate Gradient (CG), Quasi-Newton (QN), Bayesian Regularization (BR) and Resilient Backpropagation (Rprop). The training functions of CG are CG backpropagation with Powell-Beale restarts (traincgb), CG backpropagation with Fletcher-Reeves updates (traincgf), CG backpropagation with Polak-Ribére updates (traincgp), and Scaled CG backpropagation (trainscg). The training functions of GD algorithms are GD (traingd), GD with momentum (traingdm), GD with adaptive learning rate backpropagation (traingda) and GD with momentum and adaptive learning rate backpropagation (traingdx). The training functions of QN utilized in the study are Broyden-Fletcher-Goldfarb-Shanno (BFGS) QN backpropagation (trainbfg) and One-step Secant backpropagation (trainoss). The training functions of BR, LM and RP are trainbr, trainlm and trainrp, respectively (Nasrudin et al., 2020).

Fig. 1. Structure of ANN for predicting PM2.5 (PM10).
Fig. 1. Structure of ANN for predicting PM2.5 (PM10).

 
2.3 Model Fitting and Evaluation

In order to evaluate the accuracy and prediction ability of the proposed ANN forecasting model, this study calculated three indicators, namely, correlation coefficient (R), root mean square error (RMSE) and mean absolute error (MAE). RMSE, MAE, and R are defined in Eqs. (1), (2) and (3), respectively.

 

where F is the number of the original (raw) PM2.5 (PM10) data, Cl is the measured PM2.5 (PM10),  is the average of the original (raw) PM2.5 (PM10), Bl is the simulated PM2.5 (PM10), and  is the average of the PM2.5 (PM10) simulated by the ANN model.

 
3 RESULTS AND DISCUSSION


 
3.1 The Change Characteristics of Air Pollution in Chongqing

The changes of PM2.5 and PM10 concentrations in Chongqing show a downward trend (Fig. 2). The annual average concentrations of PM2.5 and PM10 in 2014 are respectively 62.9 and 94.7 µg m3. However, those are 31.3 and 48.3 µg m3. These results show that the annual average concentrations of PM2.5 and PM10 from 2014 to 2022 fell by 50.2% and 48.9%, respectively. The clean air policy plays a key role in improving air quality in Chongqing. Nevertheless, the annual average concentrations of PM2.5 and PM10 in Chongqing are still higher than the new guidance values of the World Health Organization (WHO).

Fig. 2. Change characteristics of annual average PM2.5 and PM10 concentrations.
Fig. 2. Change characteristics of annual average PM2.5 and PM10 concentrations.

 
3.1 Design of Network Structures and Training Algorithms

We use error analysis to determine the network structures, training functions and transfer functions. Table 1 shows the influence of different neurons in the hidden layer on the PM2.5 prediction results. The best network structure for predicting PM2.5 concentrations (n + 1) is discriminated as 18-11-1. Table 2 displays the performances of the thirteen kinds of training functions for predicting PM2.5 concentrations (n + 1) in Chongqing City, demonstrating that the trainbr training functions performs best, followed by trainlm, trainscg, traincgb, trainbfg, and trainrp in predicting PM2.5 concentrations (n + 1). These six training functions can meet the daily needs of PM2.5 concentrations prediction. Compared to other training functions, Trainbr is the best forecasting function with R value (0.9783), RMSE (1.2271), and MAE (0.9050). Trainbr has the highest R and the smallest MAE and RMSE among the training functions. Trainlm is the second best with R value (0.9495), RMSE (1.8845), and MAE (1.3902). Table 3 shows the impact of different transfer functions on PM2.5 prediction results. Transfer function (logsig -purelin) for predicting PM2.5 concentrations in Chongqing City is better than others during training and predicting epochs. Similarly, Table 4 shows also the influence of different neurons in the hidden layer on the PM10 prediction results. The best network structure for predicting PM10 concentrations (n + 1) is distinguished as 18-17-1. Table 5 exhibits also the performances of the thirteen kinds of training functions for predicting PM10 concentrations (n + 1) in Chongqing City, revealing that the trainbr training function has also the best prediction accuracy, followed by trainlm, traincgb, trainscg, trainbfg, trainoss, and trainrp, in predicting PM10 concentrations (n + 1). These seven training functions can meet the daily needs of PM10 concentration prediction. Trainbr is the best forecasting training function compared to other algorithms with R value (0.9773), RMSE value (1.8270), and MAE value (1.4341). Trainlm is the second best with R value (0.9522), RMSE (2.6708), and MAE (1.8554). Table 6 shows also the impact of different transfer functions on PM10 prediction results. Transfer function (logsig -purelin) for predicting PM10 concentrations in Chongqing City is also better than others during training, and predicting epochs. Nevertheless, the performances of trainbr are obviously superior to those of training functions. Because of the high complexity of PM10 concentrations, the prediction accuracy of the traingda algorithm is not high, but the prediction effect of the traibr algorithm is better than the other twelve prediction training functions.

Table 1. Comparison between different neurons in hidden layer for hourly PM2.5 prediction.

Table 2. Comparison between various training algorithms for hourly PM2.5 prediction.

 Table 3. Comparison between various transfer functions for hourly PM2.5 prediction.

Table 4. Comparison between different neurons in hidden layer for hourly PM10 prediction.


Table 5. Comparison between various training algorithms for hourly PM10 prediction.

Table 6. Comparison between various transfer functions for hourly PM10 prediction.

 
3.2 Comparative Analysis of Training Algorithms

We also compared and analyzed the trained and predicted PM2.5 concentrations and PM10 concentrations. Fig. 3 exhibits the measured PM2.5 concentrations (n + 1), simulated PM2.5 concentrations (n + 1), and the predicted PM2.5 concentrations (n + 1) in Chognqing City. The both training functions (trainbr and trainlm) not only simulate PM2.5 concentrations (n + 1) very well during the training epoch, but also predict PM2.5 concentrations (n + 1) very well during the predicting epoch. Further analysis demonstrates that trainbr is better than trainlm in both training and prediction epochs. Both training functions can not only recreate the average PM2.5 concentrations, but also recreate the minimum or maximum PM2.5 concentrations. The trainbr and trainlm predicted PM2.5 concentrations (n + 1) in an acceptable accuracy level in Chongqing City. The agreement between the measured PM2.5 concentrations (n + 1) and predicted PM2.5 concentrations (t + 1) is also very good utilizing trainbr and trainlm.

Fig. 3. Comparison between the measured PM2.5 concentrations (n + 1), simulated PM2.5 concentrations (n + 1), and the predicted PM2.5 concentrations (n + 1) in Chognqing City.Fig. 3. Comparison between the measured PM2.5 concentrations (n + 1), simulated PM2.5 concentrations (n + 1), and the predicted PM2.5 concentrations (n + 1) in Chognqing City.

Fig. 4 reveals the measured PM10 concentrations (n + 1), simulated PM10 concentrations (n + 1), and the predicted PM10 concentrations (n + 1) in Chognqing City. Similarly, the both training functions (trainbr and trainlm) not only simulate PM10 concentrations (n + 1) very well during the training epoch, but also predict PM10 concentrations (n + 1) very well during the predicting epoch. Further analysis indicates that trainbr is better than trainlm in both training and prediction epochs. Both training functions can not only copy the average PM10 concentrations, but also capture the minimum or maximum PM10 concentrations. The trainbr and trainlm predicted PM10 concentrations (n + 1) in an acceptable accuracy level in Chongqing City. The agreement between the measured PM10 concentrations (n + 1) and predicted PM10 concentrations (n + 1) is also very good by utilizing trainbr and trainlm. The prediction data of traibr and trainlm is relatively close to the real data.

Fig. 4. Comparison between the measured PM10 concentrations (n + 1), simulated PM10 concentrations (n + 1), and the predicted PM10 concentrations (n + 1) in Chognqing City.Fig. 4. Comparison between the measured PM10 concentrations (n + 1), simulated PM10 concentrations (n + 1), and the predicted PM10 concentrations (n + 1) in Chognqing City.

 
3.3 Comparison with Other Literatures

There are many prediction models based on time series analysis for predicting the concentrations of PM2.5 and PM10. The staging evolving spiking neural network (Staging-eSNN) model is established for predicting PM2.5 concentrations in the next 1–24 hours. RMSE of the model is about 9.1, 10.5, 10.9, 13 µg m–3 in the next 1, 3, 6, 12 hours in Shanghai (Liu et al., 2021). MLR, SVR and ANNs are used to forecast hourly PM2.5 and O3 concentrations in Manizales, Colombia. The ANN model for forecasting O3 exhibits the highest R (0.94), and the lowest RMSE (4.63). However, the developed ANN model is mediocre for PM2.5 with R (0.43), RMSE (5.45), MAE (4.19) (Cifuentes et al., 2021). A spatial pyramid pool (SPP) network is added between LSTM and CNN to build a CNN-SPP-LSTM network for hourly PM2.5 prediction in Qingdao. Compared to the LSTM-dense, the CNN-SPP-LSTM structure algorithm caused better hourly PM2.5 prediction performance with RMSE (29.244), MAE (16.146), R (0.828) (Li et al., 2022a). The extreme gradient boosting (XGBoost) model has better hourly PM2.5 prediction ability with R2 (0.762) in Changsha than the fully connected neural network (FCNN) deep learning model with R (0.6337) (Peng et al., 2022). The novel hybrid CEEMDAN–CNN–LSTM model for daily PM2.5 prediction has the smallest RMSE (12.68) and MAE (9.6) and the highest R2 (0.86) in Binzhou City (Xu et al., 2022). The ICNN with a 1-h time step is used for forecasting hourly PM10 and PM2.5 in South Korea. The ICNN model exhibits high prediction performance, hinted by the R2 values of 0.976 and 0.975 for hourly PM2.5 and PM10, respectively, and the low RMSE values of 1.64 and 2.745 (Chae et al., 2021). The ensemble EEMD–LSTM model is proposed to predict hourly PM2.5 concentration at Batu Muda in an urban area. The EEMDLSTM model yields the lowest forecasting errors (RMSE of 4.8949, MAE of 2.7724) and highest R2 (0.9673) compared to the other deep learning models (Zaini et al., 2022). The novel hybrid WPD-SE-VMD-Q-GRU method is proposed in hourly PM2.5 forecasting in Haerbin City. the WPD-SE-VMD-Q-GRU has the best effect with MAE (2.2342), MAPE (4.4552%), RMSE (2.9437) (Zheng et al., 2022). The new hybrid CEEMDAN–COOT–VMD–JAYA–LSSVM for daily PM2.5 concentration in Xi'an and Shenyang is proposed. The prediction result of CEEMDAN–COOT–VMD–JAYA–LSSVM has good prediction effect with RMSE (2.843), MAE (1.8344), MAPE (2.94%), R2 (0.99525) (Yang et al., 2022). The Linear Mixed Effects (LME) model is used to predict the monthly-average PM2.5 concentrations to study the spatial and temporal patterns in PM2.5 concentrations in two megacities in peninsular India between 2016 and 2019. The RMSE, Mean Prediction Error (MPE) and Relative Prediction Error (RPE) of the Linear Mixed Effects (LME) model for Bengaluru are about 8.4 µg m–3, 6.4 µg m–3, and 21.4%, respectively. The corresponding values in the case of Hyderabad are about 11.3 µg m–3, 8.2 µg m–3, and 25%, respectively (Lavanyaa et al., 2022). Machine learning (ML) approaches are used to predict PM2.5 in Bishkek. The MAE, Mean Squared Error (MSE), and RMSE of the ANN are about 20, 1953, and 44 µg m–3, respectively. The corresponding values for Random Forest Regressor (RFR) with Hyperparameter Tuning (HPT) are about 13, 800, and 28 µg m–3, respectively. The corresponding values for Xgboost Regressor (XgbR) with HPT are about 13, 742, and 27 µg m–3, respectively. The calculated evaluation metrics show that the best results are obtained for XgbR with HPT and RFR with HPT (Isaev et al., 2022). The Weather Research and Forecasting (WRF) model is used for simulating the meteorological prediction data in Ho Chi Minh City. The input data of the PM2.5 predictive machine learning model includes meteorological data (humidity, wind direction, temperature, and wind speed) and PM2.5 concentration data. The R2, RMSE, MAE, and Mean Absolute Percentage Error (MAPE) of the Neural Network Regression are about 0.66, 7.97, 5.56 and 32.26 µg m–3, respectively. The corresponding values for the Extra Trees Regression 0.68, 7.68, 5.38, and 32.27 µg m–3, respectively (Minh et al., 2021). The ML models are used to forecasting PM10 and PM2.5 concentrations in selected Polish agglomerations. Employing analysis and cross-validation, XGBoost performs the best, followed by random forests and neural networks, and stepwise regression performs the worst. The four regression models obtain R values of 0.97–0.99 and RMSE values of 4.0–13.5 and 2.2–9.6 for the PM10 and PM2.5, respectively (Czernecki et al., 2021). Feedforward-Backpropagation Neural Network is used to forecast the PM2.5 concentration in Malaysia. Trainlm is the most reliable prediction model with R2 of 0.9834, RMSE of 2.398, MAE of 1.784 and MAPE of 0.106 (Chinatamby and Jewaratnam, 2023). The hybrid forecasting system based on convolutional neural network integrated with spatial-temporal attention and residual learning (STA-ResCNN) is used to forecast PM2.5 and PM10 concentrations in the Yangtze River Delta urban agglomeration. The STA-ResCNN model produces the PM2.5 prediction performance with the RMSE of 6.986, MAE of 3.918, MAPE of 12.623% in Shanghai. The STA-ResCNN model again exhibits the PM10 prediction performance with the lowest RMSE of 11.971, MAE of 6.938, MAPE of 15.048% in Shanghai (Zhang et al., 2023). The LSTM+LSTM model has a good generalization ability with an R2 accuracy rate of 0.90 in predict next hour PM2.5 in the Istanbul Metropolitan City (Eren et al., 2023). The wavelet-packet transform (WPT) driven deep learning model is used to predict the hourly PM2.5 concentration in Qingdao. The method performs the best and achieves the lowest RMSE (3.2998) and the highest R2 (0.9978) and Nash-Sutcliffe efficiency coefficient (NSE) (0.9955) (Zheng et al., 2023). Compared with other models, our ANN model has good accuracy. This situation demonstrates that this architecture is suitable for solving the prediction problem of serial data.

 
4 CONCLUSIONS


In this paper, an ANN model is proposed to predict 1-h ahead PM2.5 and PM10 concentrations in Chongqing City in China. Considering the nonlinear and complex relationship between PM2.5 (n + 1), PM10 (n + 1) and meteorology (n), we use multiple meteorological elements. Then, 13 training functions are applied to map weather (n), time (n), PM2.5 (PM10) (n) and PM2.5 (PM10) (n + 1) to establish an ANN model for continuous prediction. Finally, the established ANN is used for prediction to verify its effectiveness under various air pollution levels. It is found that the performance of the ANN produces excellent performance. In addition, the model based on Bayesian Regularization (BR) method greatly improves the performance of the model.

However, our research has a limitation. This research did not consider other regional factors that affect Chongqing. For example, air pollution caused by northern regions is transmitted to Chongqing through wind, but this research did not calculate its impact on air quality in Chongqing.

In the future, the proposed ANN will be exploited to investigate multistep ahead predicting for other cities. We will extend the time scale of the predictions to at least 24 hours. How to make a more accurate long term prediction of PM2.5 and PM10 concentrations is what we will continue to research. Besides that, by considering the influence of meteorological parameters and other predictor variables, the study can be extended to predict other air pollutants such as, O3, and NO2. The deep learning model will be also considered to develop multiple hybrid prediction models for future research and improvement, such as, RNN, LSTM, CNN, GRU, Interpolated convolutional neural network (ICNN), EEMD, PSO, Inverse distance weighting (IDW), Ordinary least squares (OLS), wavelet transform (WT), EMD-LSTM, ICNN, EMD-GRU, CNN-LSTM, and CEEMDAN-CNN-LSTM. We will consider the impact of geographic characteristics, other regional transmissions and air pollutant emissions, such as VOCs, SO2 and NOx. The feature importance ranking of the input parameters will been evaluated.

 
ACKNOWLEDGMENTS


This paper was funded by the National Natural Science Fund of China (41572150), Shandong Province Higher Educational Humanities and Social Science Fund (J18RA196), and State Key Laboratory of Loess and Quaternary Geology Found (SKLLQG2211).


REFERENCES


  1. Awolusi, T.F., Oke, O.L., Akinkurolere, O.O., Sojobi, A.O., Aluko, O.G. (2019). Performance comparison of neural network training algorithms in the modeling properties of steel fiber reinforced concrete. Heliyon 5, 1–27. https://doi.org/10.1016/j.heliyon.2018.e01115

  2. Chae, S., Shin, J., Kwon, S., Lee, S., Kang, S., Lee, D. (2021). PM10 and PM2.5 real-time prediction models using an interpolated convolutional neural network. Sci. Rep. 11, 11952. https://doi.org/​10.1038/s41598-021-91253-9

  3. Chinatamby, P., Jewaratnam, J. (2023). A performance comparison study on PM2.5 prediction at industrial areas using different training algorithms of feedforward-backpropagation neural network (FBNN). Chemosphere 317, 137788. https://doi.org/10.1016/j.chemosphere.2023.​137788

  4. Cifuentes, F., Gálvez, A., González, C.M., Orozco-Alzate, M., Aristizábal, B.H. (2021). Hourly ozone and PM2.5 prediction using meteorological data – alternatives for cities with limited pollutant information. Aerosol Air Qual. Res. 21, 200471. https://doi.org/10.4209/aaqr.200471

  5. Crippa, M., Solazzo, E., Guizzardi, D., Van Dingenen, R., Leip, A. (2022). Air pollutant emissions from global food systems are responsible for environmental impacts, crop losses and mortality. Nat. Food. 3, 942–956. https://doi.org/10.1038/s43016-022-00615-7

  6. Czernecki, B., Marosz, M., Jędruszkiewicz, J. (2021). Assessment of machine learning algorithms in short-term forecasting of PM10 and PM2.5 concentrations in selected polish agglomerations. Aerosol Air Qual. Res. 21, 200586. https://doi.org/10.4209/aaqr.200586

  7. Eren, B., Aksangür, İ., Erden, C. (2023). Predicting next hour fine particulate matter (PM2.5) in the istanbul metropolitan city using deep learning algorithms with time windowing strategy. Urban Clim. 48, 101418. https://doi.org/10.1016/j.uclim.2023.101418

  8. Fowler, D., Pyle, J.A., Sutton, M.A., Williams, M.L. (2020). Global air quality, past present and future: an introduction. Phil. Trans. R. Soc. A 378, 20190323. https://doi.org/10.1098/rsta.​2019.0323

  9. Geng, G., Zheng, Y., Zhang, Q., Xue, T., Zhao, H., Tong, D., Zheng, B., Li, M., Liu, F., Hong, C., He, K., Davis, S.J. (2021). Drivers of PM2.5 air pollution deaths in china 2002–2017. Nat. Geosci. 14, 645–650. https://doi.org/10.1038/s41561-021-00792-3

  10. Goudarzi, G., Hopke, P.K., Yazdani, M. (2021). Forecasting PM2.5 concentration using artificial neural network and its health effects in Ahvaz, Iran. Chemosphere 283, 131285. https://doi.org/​10.1016/j.chemosphere.2021.131285

  11. Guo, Q., He, Z., Li, S., Li, X., Meng, J., Hou, Z., Liu, J., Chen, Y. (2020). Air pollution forecasting using artificial and wavelet neural networks with meteorological conditions. Aerosol Air Qual. Res. 20, 1429–1439. https://doi.org/10.4209/aaqr.2020.03.0097

  12. Guo, Q., He, Z. (2021). Prediction of the confirmed cases and deaths of global COVID-19 using artificial intelligence. Environ. Sci. Pollut. Res. 28, 11672–11682. https://doi.org/10.1007/​s11356-020-11930-6

  13. Guo, Q., Wang, Z., He, Z., Li, X., Meng, J., Hou, Z., Yang, J. (2021). Changes in air quality from the Covid to the post-Covid era in the Beijing-Tianjin-Tangshan region in China. Aerosol Air Qual. Res. 21, 210270. https://doi.org/10.4209/aaqr.210270

  14. Guo, Q., He, Z., Wang, Z. (2023a). Change in air quality during 2014–2021 in Jinan City in China and its influencing factors. Toxics 11, 210. https://doi.org/10.3390/toxics11030210

  15. Guo, Q., He, Z., Wang, Z. (2023b). Long-term projection of future climate change over the twenty-first century in the Sahara region in Africa under four Shared Socio-Economic Pathways scenarios. Environ. Sci. Pollut. Res. 30, 22319–22329. https://doi.org/10.1007/s11356-022-23813-z

  16. Guo, Q., He, Z., Wang, Z. (2023c). Predicting of daily PM2.5 concentration employing wavelet artificial neural networks based on meteorological elements in Shanghai, China. Toxics 11, 51. https://doi.org/10.3390/toxics11010051

  17. He, Z., Guo, Q., Wang, Z., Li, X. (2022). Prediction of monthly PM2.5 concentration in liaocheng in China employing artificial neural network. Atmosphere 13, 1221. https://doi.org/10.3390/​atmos13081221

  18. Huang, R.J., Zhang, Y., Bozzetti, C. (2014). High secondary aerosol contribution to particulate pollution during haze events in China. Nature 514, 218–222. https://doi.org/10.1038/nature​13774

  19. Isaev, E., Ajikeev, B., Shamyrkanov, U., Kalnur, K.U., Maisalbek, K., Sidle, R.C. (2022). Impact of climate change and air pollution forecasting using machine learning techniques in Bishkek. Aerosol Air Qual. Res. 22, 210336. https://doi.org/10.4209/aaqr.210336

  20. Islam, N., Toha, T.R., Islam, M.M., Ahmed, T. (2023). Spatio-temporal variation of meteorological influence on PM2.5 and PM10 over major urban cities of Bangladesh. Aerosol Air Qual. Res. 23, 220082. https://doi.org/10.4209/aaqr.220082

  21. Lavanyaa, V.P., Varshini, S., Mitra, S.S., Hungund, K.M., Majumdar, R., Srikanth, R. (2022). Geospatial modelling for estimation of PM2.5 concentrations in two megacities in Peninsular India. Aerosol Air Qual. Res. 22, 220110. https://doi.org/10.4209/aaqr.220110

  22. Li, J., Xu, G., Cheng, X. (2022a). Combining spatial pyramid pooling and long short-term memory network to predict PM2.5 concentration. Atmos. Pollut. Res. 13, 101309. https://doi.org/​10.1016/j.apr.2021.101309

  23. Li, Z., Yue, Z., Liu, J., Zhai, Z., Meng, J., Wang, L., Wei, J., Yang, D., Hu, A., Zhu, W., Ding, N., Lu, X., Lin, C. (2022b). Sources and health risks of PM2.5-bound PAHs in a small city along with the “clean heating” policy. Aerosol Air Qual. Res. 22, 220162. https://doi.org/10.4209/aaqr.220162

  24. Liu, H., Lu, G., Wang, Y., Kasabov, N. (2021). Evolving spiking neural network model for PM2.5 hourly concentration prediction based on seasonal differences: A case study on data from Beijing and Shanghai. Aerosol Air Qual. Res. 21, 200247. https://doi.org/10.4209/aaqr.2020.​05.0247

  25. Liu, Y., Dong, J., Ma, H., Jiang, Y., Zheng, W., Luo, X. (2022). An overview: PM2.5 concentration levels in urban residential buildings during the past two decades. Aerosol Air Qual. Res. 22, 220174. https://doi.org/10.4209/aaqr.220174

  26. Minh, V.T.T., Tin, T.T., Hien, T.T. (2021). PM2.5 Forecast System by Using Machine Learning and WRF Model, A Case Study: Ho Chi Minh City, Vietnam. Aerosol Air Qual. Res. 21, 210108. https://doi.org/10.4209/aaqr.210108

  27. Nasrudin, N.A., Jewaratnam, J., Hossain, M.A., Ganeson, P.B. (2020). Performance comparison of feedforward neural network training algorithms in modelling microwave pyrolysis of oil palm fibre for hydrogen and biochar production. Asia-Pac. J. Chem. Eng. 15, e2388. https://doi.org/​10.1002/apj.2388

  28. Pakalapati, H., Tariq, M.A., Arumugasamy, S.K. (2019). Optimization and modelling of enzymatic polymerization of ε-caprolactone to polycaprolactone using Candida Antartica Lipase B with response surface methodology and artificial neural network. Enzyme Microb. Technol. 122, 7–18. https://doi.org/10.1016/j.enzmictec.2018.12.001

  29. Peng, J., Han, H., Yi, Y., Huang, H., Xie, L. (2022). Machine learning and deep learning modeling and simulation for predicting PM2.5 concentrations. Chemosphere 308, 136353. https://doi.org/​10.1016/j.chemosphere.2022.136353

  30. Perera, A., Azamathulla, H., Rathnayake, U. (2020). Comparison of different artificial neural network (ANN) training algorithms to predict the atmospheric temperature in Tabuk, Saudi Arabia. Mausam 71, 233–244. https://doi.org/10.54302/mausam.v71i2.22

  31. Xu, S., Li, W., Zhu, Y., Xu, A. (2022). A novel hybrid model for six main pollutant concentrations forecasting based on improved LSTM neural networks. Sci. Rep. 12, 14434. https://doi.org/​10.1038/s41598-022-17754-3

  32. Yang, H., Liu, Z., Li, G. (2022). A new hybrid optimization prediction model for PM2.5 concentration considering other air pollutants and meteorological conditions. Chemosphere 307, 135798. https://doi.org/10.1038/s41598-022-17754-3

  33. Zaini, N.a., Ean, L.W., Ahmed, A.N., Abdul Malek, M., Chow, M.F. (2022). PM2.5 forecasting for an urban area based on deep learning and decomposition method. Sci. Rep. 12, 17565. https://doi.org/10.1038/s41598-022-21769-1

  34. Zhang, K., Yang, X., Cao, H., Thé, J., Tan, Z., Yu, H. (2023). Multi-step forecast of PM2.5 and PM10 concentrations using convolutional neural network integrated with spatial–temporal attention and residual learning. Environ. Int. 171, 107691. https://doi.org/10.1016/j.envint.2022.107691

  35. Zheng, G., Liu, H., Yu, C., Li, Y., Cao, Z. (2022). A new PM2.5 forecasting model based on data preprocessing, reinforcement learning and gated recurrent unit network. Atmos. Pollut. Res. 13, 101475. https://doi.org/10.1016/j.apr.2022.101475

  36. Zheng, Q., Tian, X., Yu, Z., Jiang, N., Elhanashi, A., Saponara, S., Yu, R. (2023). Application of wavelet-packet transform driven deep learning method in PM2.5 concentration prediction: A case study of Qingdao, China. Sustain. Cities Soc. 92, 104486. https://doi.org/10.1016/j.​scs.2023.104486


Share this article with your colleagues 

 

Subscribe to our Newsletter 

Aerosol and Air Quality Research has published over 2,000 peer-reviewed articles. Enter your email address to receive latest updates and research articles to your inbox every second week.

7.3
2022CiteScore
 
 
77st percentile
Powered by
Scopus
 
   SCImago Journal & Country Rank

2022 Impact Factor: 4.0
5-Year Impact Factor: 3.4

Aerosol and Air Quality Research partners with Publons

CLOCKSS system has permission to ingest, preserve, and serve this Archival Unit
CLOCKSS system has permission to ingest, preserve, and serve this Archival Unit

Aerosol and Air Quality Research (AAQR) is an independently-run non-profit journal that promotes submissions of high-quality research and strives to be one of the leading aerosol and air quality open-access journals in the world. We use cookies on this website to personalize content to improve your user experience and analyze our traffic. By using this site you agree to its use of cookies.