Advertisement

PM2.5 Concentration Forecast Based on Hierarchical Sparse Representation

  • Rui Zhao
  • Bingjian LuEmail author
  • Zhenyu Lu
  • Hengde Zhang
  • Tianming Zhan
Conference paper
Part of the Lecture Notes in Electrical Engineering book series (LNEE, volume 550)

Abstract

This paper proposes hierarchical sparse representation (H-SRC) to predict PM2.5 Concentration. It selects factors from observational data in Beijing-Tianjin-Hebei. Its time is from January to March in 2013–2017. Then, it constructs 4000 samples of historical databases based on fuzzy C means algorithm (FCM). Input Meteorology factors predicted by Rapid Refresh Multi-scale Analysis & Prediction System-CHEM (RMAPS-CHEM) and European Centre for Medium-Range Weather Forecasts (ECMWF), use the first-level sparse representation to classify test samples and the second-level sparse representation to regress test samples, then it predict the PM2.5 Concentration. Experiment with the data in Beijing-Tianjin-Hebei between January and March, 2018, reveals that the method in this paper can increase the accuracy and reduce mean absolute error. The accuracy by hierarchical sparse representation is 25.28%, 13.34%, 14.28%, 23.08% higher than RMAPS-CHEM in 0–35 \( \upmu{\text{g}}/{\text{m}}^{3} \), 75–115 \( \upmu{\text{g}}/{\text{m}}^{3} \), 115–150 \( \upmu{\text{g}}/{\text{m}}^{3} \), 150–250 \( \upmu{\text{g}}/{\text{m}}^{3} \), while absolute errors are all lower than RMAPS-CHEM. At the same time, this method is easy to study and is convenience for the analysis of other meteorological data.

Keywords

Historical database Sparse representation Regress PM2.5 concentration 

Notes

Acknowledgments

This work has been supported in part by the National Key Research Program of China (Grant No. 2016YFC0203301), the National Natural Science Foundation of China (Grant No. 61773220, 61502206), the Nature Science Foundation of Jiangsu Province under Grant (No. BK20150523).

References

  1. 1.
    Wang, Y.H., Liu, Y.J.: The characteristics of long-term changes of smog and haze in China in the past 50 years and its relationship with atmospheric humidity. Chin. Sci. Earth Sci. 44(1), 37–48 (2014)Google Scholar
  2. 2.
    Fang, J.X., Zhao, C.Y., Zhu, W.H.: Concentration monitoring and forecasting model of PM_(2.5) during the heating period of Lanzhou city. Environ. Sci. Technol. 37(4), 80–84 (2014)Google Scholar
  3. 3.
    Liu, H., Rao, X.Q., Zhang, H.D., et al.: Comparative test of the forecast effect of environmental meteorology numerical prediction model. J. Meteorol. Environ. 33(5), 17–24 (2017)Google Scholar
  4. 4.
    Zhao, X.J., Xu, J., Zhang, Z.Y., et al.: Beijing regional environmental meteorology prediction system and PM2.5 forecast inspection. J. Appl. Meteorol. 27(2), 160–172 (2016)Google Scholar
  5. 5.
    Fan, J.X., Li, Q., Zhu, Y.J., et al.: Research on air pollution prediction model based on RNN. Sci. Surv. Mapp. 42(7), 76–83 (2017)Google Scholar
  6. 6.
    Dai, L.J., Zhang, C.J., Ma, L.M.: PM2.5 short-term concentration dynamic prediction model based on machine learning. Comput. Appl. 37(11), 3057–3063 (2017)Google Scholar
  7. 7.
    Akhtar, N., Shafait, F., Mian, A.: Efficient classification with sparsity augmented collaborative representation. Pattern Recogn. 65, 136–145 (2017)CrossRefGoogle Scholar
  8. 8.
    Abrol, V., Sharma, P., Sao, A.K.: Greedy double sparse dictionary learning for sparse representation of speech signals. Speech Commun. 85, 71–82 (2016)CrossRefGoogle Scholar
  9. 9.
    Abdessamad, J., Eladel, A., Zaied, M.: A sparse representation-based approach for copy-move image forgery detection in smooth regions. In: International Conference on Machine Vision, vol. 10341, 1034129 (2017)Google Scholar
  10. 10.
    Chen, X.H., Diao, Z.G., Hu, J.K., et al.: Dynamic-statistical prediction of PM2.5 concentration in China based on CMAQ model and adaptive partial least square regression. J. Environ. Sci. 35(8), 2771–2782 (2016)Google Scholar

Copyright information

© Springer Nature Singapore Pte Ltd. 2019

Authors and Affiliations

  • Rui Zhao
    • 1
  • Bingjian Lu
    • 1
    • 2
    Email author
  • Zhenyu Lu
    • 2
    • 3
  • Hengde Zhang
    • 1
  • Tianming Zhan
    • 4
  1. 1.National Meteorological CenterBeijingChina
  2. 2.School of Electronic and Information EngineeringNanjing University of Information Science and TechnologyNanjingChina
  3. 3.The Collaborative Innovation Center on Atmospheric Environment and Equipment of JiangsuNanjingChina
  4. 4.Nanjing Audit UniversityNanjingChina

Personalised recommendations