Roadside Air Quality Forecasting in Shanghai with a Novel Sequence-to-Sequence Model
Abstract
:1. Introduction
2. Data and Methods
2.1. Study Area and Data Description
2.2. Autocorrelation Analysis
2.3. Long Short-Term Memory (LSTM) Networks
2.4. Sequence to Sequence (Seq2Seq) Model
3. Results and Discussion
3.1. Diurnal Variation
3.2. Weekly Variation
3.3. Forecasting Model Results
3.4. Urban Monitoring Station Results
4. Conclusions
- (1)
- The daily trend of CO and PM2.5 is consistent with the trend of daily traffic volume in Shanghai, and the PM2.5 suggests a strong hysteresis (roughly one hour). Morning and evening traffic rush hours are also high pollution-level periods.
- (2)
- The concentrations of air pollutants are quite different between the two roadside air quality monitoring stations Xuhui and Jing’an, and the latter is lightly affected by traffic emissions. The CO concentrations at the Xuhui station are higher, while the PM2.5 concentrations at the two stations are similar. Given that the Xuhui station is closer to the traffic emission sources than the Jing’an station, the results indicate that the CO concentrations at the two stations are mainly affected by local traffic sources, and the pollution sources of PM2.5 are not only limited to local traffic emissions.
- (3)
- This model improves the forecasting accuracy of the roadside air quality, compared with the traditional Seq2Seq model and other baseline machine learning models. The model accuracy is higher at the Xuhui station where the periodic characteristics of air pollutants are more noticeable. This result suggests that the daily periodicity caused by traffic should not be overlooked in modeling and forecasting air quality in roadside areas. Therefore, the weekly periodicity should be fully considered in air quality forecasting.
- (4)
- The proposed Seq2Seq model with weekly periodicity was also suitable for the eight urban monitoring stations in Shanghai. In contrast, the weekly periodicity demonstrates a more pronounced impact on PM2.5 forecasting. For CO forecasts, the weekly periodicity-based model is not necessarily appropriate for all monitoring stations.
- (1)
- The temporal patterns of traffic and air quality are fully evaluated and further summarized based on the two-year air quality monitoring data in megacities;
- (2)
- Weekly periodicity is taken into account in the deep learning-based air quality forecasting model, which strongly improves prediction accuracy;
- (3)
- The proposed Seq2Seq model with weekly periodicity is also applicable to urban air quality prediction (not only for traffic-related roadside air quality), and thus can be used by public authorities to make timely management adjustments to protect public health based on air quality predictions.
Author Contributions
Funding
Acknowledgments
Conflicts of Interest
References
- Amato, F.; Schaap, M.; Reche, C.; Querol, X. Road Traffic: A Major Source of Particulate Matter in Europe; Springer: Berlin/Heidelberg, Germany, 2013. [Google Scholar]
- Wu, Y.; Zhang, S.; Hao, J.; Liu, H.; Wu, X.; Hu, J.; Walsh, M.P.; Wallington, T.J.; Zhang, K.M.; Stevanovic, S. On-road vehicle emissions and their control in China: A review and outlook. Sci. Total Environ. 2017, 574, 332–349. [Google Scholar] [CrossRef] [PubMed]
- Health Effects Institute. A Critical Review of the Literature on Emissions, Exposure and Health Effects; Health Effects Institute: Boston, MA, USA, 2010. [Google Scholar]
- Song, Y.; Wang, X.; Maher, B.A.; Li, F.; Xu, C.; Liu, X.; Sun, X.; Zhang, Z. The spatial-temporal characteristics and health impacts of ambient fine particulate matter in China. J. Clean. Prod. 2016, 112, 1312–1318. [Google Scholar] [CrossRef]
- Laden, F.; Schwartz, J.; Speizer, F.E.; Dockery, D.W. Reduction in fine particulate air pollution and mortality: Extended follow-up of the Harvard Six Cities Study. Am. J. Respir. Crit. Care Med. 2006, 173, 667–672. [Google Scholar] [CrossRef] [PubMed]
- Luo, C.; Zhu, X.; Yao, C.; Hou, L.; Zhang, J.; Cao, J.; Wang, A. Short-term exposure to particulate air pollution and risk of myocardial infarction: A systematic review and meta-analysis. Environ. Sci. Pollut. Res. 2015, 22, 14651–14662. [Google Scholar] [CrossRef] [PubMed]
- Cheng, Z.; Wang, S.; Jiang, J.; Fu, Q.; Chen, C.; Xu, B.; Yu, J.; Fu, X.; Hao, J. Long-term trend of haze pollution and impact of particulate matter in the Yangtze River Delta, China. Environ. Pollut. 2013, 182, 101–110. [Google Scholar] [CrossRef]
- Skamarock, W.C.; Klemp, J.; Dudhia, J.; Gill, D.O.; Barker, D.; Wang, W.; Powers, J.G. A Description of the Advanced Research WRF Version 3; University Corporation for Atmospheric Research: Boulder, CO, USA, 2008; Volume 27, pp. 3–27. [Google Scholar]
- Grell, G.A.; Peckham, S.E.; Schmitz, R.; McKeen, S.A.; Frost, G.; Skamarock, W.C.; Eder, B. Fully coupled “online” chemistry within the WRF model. Atmos. Environ. 2005, 39, 6957–6975. [Google Scholar] [CrossRef]
- Byun, D.W.; Schere, K.L. Review of the governing equations, computational algorithms, and other components of the Models-3 Community Multiscale Air Quality (CMAQ) modeling system. Appl. Mech. Rev. 2006, 59, 51–77. [Google Scholar] [CrossRef]
- Yang, W.; Deng, M.; Xu, F.; Wang, H. Prediction of hourly PM2.5 using a space-time support vector regression model. Atmos. Environ. 2018, 181, 12–19. [Google Scholar] [CrossRef]
- Hua, H.; Jiang, S.; Sheng, H.; Zhang, Y.; Liu, X.; Zhang, L.; Yuan, Z.; Chen, T. A high spatial-temporal resolution emission inventory of multi-type air pollutants for Wuxi city. J. Clean. Prod. 2019, 229, 278–288. [Google Scholar] [CrossRef]
- Ni, X.Y.; Huang, H.; Du, W.P. Relevance analysis and short-term prediction of PM2.5 concentrations in Beijing based on multi-source data. Atmos. Environ. 2017, 150, 146–161. [Google Scholar] [CrossRef]
- Deng, W.; Xu, J.; Song, Y.; Zhao, H. Differential evolution algorithm with wavelet basis function and optimal mutation strategy for complex optimization problem. Appl. Soft Comput. J. 2020, 106724. [Google Scholar] [CrossRef]
- Song, Y.; Wu, D.; Deng, W.; Gao, X.Z.; Li, T.; Zhang, B.; Li, Y. MPPCEDE: Multi-population parallel co-evolutionary differential evolution for parameter optimization. Energy Convers. Manag. 2021, 228, 113661. [Google Scholar] [CrossRef]
- Li, X.; Luo, A.; Li, J.; Li, Y. Air pollutant concentration forecast based on support vector regression and quantum-behaved particle swarm optimization. Environ. Model. Assess. 2019, 24, 205–222. [Google Scholar] [CrossRef]
- He, H.-D.; Lu, W.Z.; Xue, Y. Prediction of particulate matter at street level using artificial neural networks coupling with chaotic particle swarm optimization algorithm. Build. Environ. 2014, 78, 111–117. [Google Scholar] [CrossRef]
- Abderrahim, H.; Chellali, M.R.; Hamou, A. Forecasting PM10 in Algiers: Efficacy of multilayer perceptron networks. Environ. Sci. Pollut. Res. 2016, 23, 1634–1641. [Google Scholar] [CrossRef]
- Gradišar, D.; Grašič, B.; Božnar, M.Z.; Mlakar, P.; Kocijan, J. Improving of local ozone forecasting by integrated models. Environ. Sci. Pollut. Res. 2016, 23, 18439–18450. [Google Scholar] [CrossRef]
- Wang, H.W.; Li, X.B.; Wang, D.; Zhao, J.; He, H.-D.; Peng, Z.R. Regional prediction of ground-level ozone using a hybrid sequence-to-sequence deep learning approach. J. Clean. Prod. 2020, 253, 119841. [Google Scholar] [CrossRef]
- Li, X.; Peng, L.; Hu, Y.; Shao, J.; Chi, T. Deep learning architecture for air quality predictions. Environ. Sci. Pollut. Res. 2016, 23, 22408–22417. [Google Scholar] [CrossRef]
- Biancofiore, F.; Verdecchia, M.; Di Carlo, P.; Tomassetti, B.; Aruffo, E.; Busilacchio, M.; Bianco, S.; Di Tommaso, S.; Colangeli, C. Analysis of surface ozone using a recurrent neural network. Sci. Total Environ. 2015, 514, 379–387. [Google Scholar] [CrossRef]
- Athira, V.; Geetha, P.; Vinayakumar, R.; Soman, K.P. DeepAirNet: Applying recurrent networks for air quality prediction. Procedia Comput. Sci. 2018, 132, 1394–1403. [Google Scholar]
- Li, X.; Peng, L.; Yao, X.; Cui, S.; Hu, Y.; You, C.; Chi, T. Long short-term memory neural network for air pollutant concentration predictions: Method development and evaluation. Environ. Pollut. 2017, 231, 997–1004. [Google Scholar] [CrossRef] [PubMed]
- Zhu, J.Y.; Zhang, C.; Zhang, H.; Zhi, S.; Li, V.O.K.; Han, J.; Zheng, Y. p-Causality: Identifying spatiotemporal causal pathways for air pollutants with urban Big data. IEEE Trans. Big Data 2017, 4, 571–585. [Google Scholar] [CrossRef] [Green Version]
- Zhao, J.; Deng, F.; Cai, Y.; Chen, J. Long short-term memory—Fully connected (LSTM-FC) neural network for PM 2.5 concentration prediction. Chemosphere 2019, 220, 486–492. [Google Scholar] [CrossRef] [PubMed]
- Liu, B.; Yan, S.; Li, J.; Qu, G.; Li, Y.; Lang, J.; Gu, R. A sequence-to-sequence air quality predictor based on the n-step recurrent prediction. IEEE Access 2019, 7, 43331–43345. [Google Scholar] [CrossRef]
- Gao, J.; Wang, T.; Zhou, X.; Wu, W.; Wang, W. Measurement of aerosol number size distributions in the Yangtze River delta in China: Formation and growth of particles under polluted conditions. Atmos. Environ. 2009, 43, 829–836. [Google Scholar] [CrossRef]
- Cheng, Z.; Luo, L.; Wang, S.; Wang, Y.; Sharma, S.; Shimadera, H.; Wang, X.; Bressi, M.; de Miranda, R.M.; Jiang, J.; et al. Status and characteristics of ambient PM2.5 pollution in global megacities. Environ. Int. 2016, 89–90, 212–221. [Google Scholar] [CrossRef]
- Wang, D.; Wang, Z.; Peng, Z.R. Performance evaluation of CAL3QHC and CALINE4 for short-term simulation of fine particulate matter and carbon monoxide concentrations at road intersection. In Proceedings of the Transportation Research Board Meeting, Washington, DC, USA, 11–15 January 2015. [Google Scholar]
- Wang, Z.; Lu, F.; He, H.; Lu, Q.-C.; Wang, D.; Peng, Z.-R. Fine-scale estimation of carbon monoxide and fine particulate matter concentrations in proximity to a road intersection by using wavelet neural network with genetic algorithm. Atmos. Environ. 2015, 104, 264–272. [Google Scholar] [CrossRef]
- Wang, Y.; Li, L.; Chen, C.; Huang, C.; Huang, H.; Feng, J.; Wang, S.; Wang, H.; Zhang, G.; Zhou, M.; et al. Source apportionment of fine particulate matter during autumn haze episodes in Shanghai, China. J. Geophys. Res. Atmos. 2014, 119, 1903–1914. [Google Scholar] [CrossRef]
- Wang, Z.; Lu, Q.-C.; He, H.-D.; Wang, D.; Gao, Y.; Peng, Z.-R. Investigation of the spatiotemporal variation and influencing factors on fine particulate matter and carbon monoxide concentrations near a road intersection. Front. Earth Sci. 2016, 11, 1–13. [Google Scholar] [CrossRef]
- Fu, Q.; Zhuang, G.; Wang, J.; Xu, C.; Huang, K.; Li, J.; Hou, B.; Lu, T.; Streets, D.G. Mechanism of formation of the heaviest pollution episode ever recorded in the Yangtze River Delta, China. Atmos. Environ. 2008, 42, 2023–2036. [Google Scholar] [CrossRef]
- Box, G.E.P.; Jenkins, G.M.; Reinsel, G.C.; Ljung, G.M. Time Series Analysis: Forecasting and Control; John Wiley & Sons: Hoboken, NJ, USA, 2015. [Google Scholar]
- Feng, R.; Zheng, H.; Gao, H.; Zhang, A.; Huang, C.; Zhang, J.; Luo, K.; Fan, J. Recurrent Neural Network and random forest for analysis and accurate forecast of atmospheric pollutants: A case study in Hangzhou, China. J. Clean. Prod. 2019, 231, 1005–1015. [Google Scholar] [CrossRef]
- Hochreiter, S.; Schmidhuber, J. Long short-term memory. Neural Comput. 1997, 9, 1735–1780. [Google Scholar] [CrossRef] [PubMed]
- Sutskever, I.; Vinyals, O.; Le, Q.V. Sequence to Sequence Learning with Neural Networks. Adv. Neural Inform. Process. Syst. 2014, 27, 3104–3112. [Google Scholar]
- Cho, K.; van Merrienboer, B.; Gulcehre, C.; Bahdanau, D.; Bougares, F.; Schwenk, H.; Bengio, Y. Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation. In Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP); Association for Computational Linguistics: Stroudsburg, PA, USA, 2014; pp. 1724–1734. [Google Scholar]
- Huang, J.; Zhang, J.; Yang, H.; Geng, F.; Chen, Y.; Li, H.; Yu, Y.; Wang, Y.; Wang, Y. Diurnal and seasonal variations in particulate matter at Shanghai during the heavy haze and non-haze periods. J. Donghua Univ. 2015, 32, 677–680. [Google Scholar]
- Shanghai Urban and Rural Construction and Transportation Development Research Institute Shanghai 5th Comprehensive Transportation Survey Report. Traffic Transp. 2015, 31, 15–18.
- Keogh, D.U.; Kelly, J.; Mengersen, K.; Jayaratne, R.; Ferreira, L.; Morawska, L. Derivation of motor vehicle tailpipe particle emission factors suitable for modelling urban fleet emissions and air quality assessments. Environ. Sci. Pollut. Res. 2010, 17, 724–739. [Google Scholar] [CrossRef] [Green Version]
- Li, J. Seasonal Characteristics of Air Pollution and Weekend Effect in Shanghai; The University of Chinese Academy of Sciences: Beijing, China, 2015. [Google Scholar]
- Graves, A.; Schmidhuber, J. Framewise phoneme classification with bidirectional LSTM and other neural network architectures. Neural Networks 2005, 18, 602–610. [Google Scholar] [CrossRef]
Station | Model | PM2.5 | CO | ||||
---|---|---|---|---|---|---|---|
RMSE | NMSE | r | RMSE | NMSE | r | ||
Xuhui | Seq2Seq with weekly periodicity | 21.51 | 0.281 | 0.735 | 0.327 | 0.124 | 0.681 |
Seq2Seq | 27.13 | 0.335 | 0.692 | 0.349 | 0.132 | 0.676 | |
Bidirectional LSTM | 27.66 | 0.360 | 0.641 | 0.355 | 0.146 | 0.577 | |
LSTM | 30.49 | 0.432 | 0.540 | 0.364 | 0.147 | 0.581 | |
Jing’an | Seq2Seq with weekly periodicity | 23.14 | 0.278 | 0.716 | 0.348 | 0.160 | 0.590 |
Seq2Seq | 25.73 | 0.312 | 0.683 | 0.360 | 0.173 | 0.587 | |
Bidirectional LSTM | 24.28 | 0.345 | 0.667 | 0.362 | 0.174 | 0.565 | |
LSTM | 33.13 | 0.503 | 0.482 | 0.370 | 0.202 | 0.475 |
Station | Model Input Parameters | PM2.5 | CO | ||||
---|---|---|---|---|---|---|---|
RMSE | NMSE | r | RMSE | NMSE | r | ||
Xuhui | PM2.5 & CO | 25.92 | 0.377 | 0.695 | 0.353 | 0.134 | 0.654 |
PM2.5 & CO + POL 1 (6 pollutants) | 24.85 | 0.321 | 0.647 | 0.322 | 0.121 | 0.661 | |
PM2.5 & CO + MEO 2 | 21.93 | 0.293 | 0.720 | 0.305 | 0.108 | 0.704 | |
PM2.5 & CO + POL + MEO (all) | 21.51 | 0.281 | 0.735 | 0.327 | 0.124 | 0.681 | |
Jing’an | PM2.5 & CO | 25.45 | 0.365 | 0.661 | 0.352 | 0.165 | 0.577 |
PM2.5 & CO + POL (6 pollutants) | 24.59 | 0.322 | 0.672 | 0.355 | 0.161 | 0.606 | |
PM2.5 & CO + MEO | 23.28 | 0.334 | 0.713 | 0.353 | 0.157 | 0.616 | |
PM2.5 & CO + POL + MEO (all) | 23.14 | 0.278 | 0.716 | 0.348 | 0.160 | 0.590 |
No. | Station | District |
---|---|---|
1142A | Fifteenth factory | Huangpu |
1143A | Hongkou | Hongkou |
1144A | Xuhui Shanghai Normal University | Xuhui |
1145A | Yangpu Sipiao | Yangpu |
1147A | Jing’an monitoring station | Jing’an |
1148A | Pudong Chuansha | Pudong |
1149A | Pudong monitoring station | Pudong |
1150A | Pudong Zhangjiang | Pudong |
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |
© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
Share and Cite
Wang, D.; Wang, H.-W.; Li, C.; Lu, K.-F.; Peng, Z.-R.; Zhao, J.; Fu, Q.; Pan, J. Roadside Air Quality Forecasting in Shanghai with a Novel Sequence-to-Sequence Model. Int. J. Environ. Res. Public Health 2020, 17, 9471. https://doi.org/10.3390/ijerph17249471
Wang D, Wang H-W, Li C, Lu K-F, Peng Z-R, Zhao J, Fu Q, Pan J. Roadside Air Quality Forecasting in Shanghai with a Novel Sequence-to-Sequence Model. International Journal of Environmental Research and Public Health. 2020; 17(24):9471. https://doi.org/10.3390/ijerph17249471
Chicago/Turabian StyleWang, Dongsheng, Hong-Wei Wang, Chao Li, Kai-Fa Lu, Zhong-Ren Peng, Juanhao Zhao, Qingyan Fu, and Jun Pan. 2020. "Roadside Air Quality Forecasting in Shanghai with a Novel Sequence-to-Sequence Model" International Journal of Environmental Research and Public Health 17, no. 24: 9471. https://doi.org/10.3390/ijerph17249471