A Review on the Application of Artificial Intelligence and Multi-source Data in Refined Prediction of Atmospheric Pollution
Main Article Content
Keywords
atmospheric pollution, refined prediction, multi-source data, artificial intelligence, graph neural network, data-mechanism dual-driven
Abstract
Refined prediction of atmospheric pollution captures the dynamic evolution of pollutant concentrations at high spatiotemporal resolution, serving as the core technical support for precise governance of the atmospheric environment and early warning of air quality risks. Artificial intelligence technologies have enabled new methods for exploring the complex spatiotemporal correlations and nonlinear response relationships hidden in data. This paper systematically sorts out the research achievements of the integration of artificial intelligence and multi-source data in the field of refined atmospheric pollution prediction from 2019 to 2024, defines the technical connotation and evaluation system of refined prediction, and analyzes the principles and characteristics of typical technical routes from two dimensions: data fusion level and model design paradigm. Combined with case studies in typical regions of China, such as the Beijing-Tianjin-Hebei region and the Yangtze River Delta, the actual efficiency and engineering bottlenecks of the technical implementation are dissected. The study finds that the field is still grappling with core problems, including heterogeneous data quality across multiple sources, limited interpretability of models, and insufficient generalization in complex scenarios. Based on this, future research directions are proposed, including deep fusion of multimodal data integrating atmospheric physical mechanisms, the construction of interpretable artificial intelligence models for pollution prevention and control, and the design of lightweight models driven by edge computing. This paper aims to provide a systematic reference for technological innovation and engineering applications in this field and to promote the transformation of atmospheric pollution prediction from “data-driven” to “data-mechanism dual-driven”.
References
- [1] Xu X D, Tong T, Zhang W, et al. Fine-grained prediction of PM2.5 concentration based on multisource data and deep learning[J]. Atmospheric Pollution Research, 2020, 11(10): 1752-1762.
- [2] Dee D P, Uppala S M, Simmons A J, et al. The ERA-Interim reanalysis: configuration and performance of the data assimilation system[J]. Quarterly Journal of the Royal Meteorological Society, 2011, 137(656): 553-597.
- [3] Van der A R J, Levelt P F, Veihelmann B, et al. TROPOMI on the Sentinel-5 Precursor: A GMES mission for global air quality monitoring[J]. Remote Sensing of Environment, 2018, 212: 155-167.
- [4] Liu C, Zhang Q, Chen J. Machine learning-based fine-grained PM2.5 prediction using multi-source meteorological and monitoring data[J]. Atmospheric Environment, 2021, 262: 118609.
- [5] Li J, Zhang L, Chen Y. A systematic review of data fusion techniques for atmospheric pollution monitoring and prediction[J]. Atmospheric Environment, 2023, 301: 119456.
- [6] Guo S, Lin Y, Feng X, et al. Deep learning for spatiotemporal prediction of air pollution: A review[J]. Science of the Total Environment, 2023, 861: 163624.
- [7] Breiman L. Random Forests[J]. Machine Learning, 2001, 45(1): 5-32.
- [8] Hochreiter S, Schmidhuber J. Long Short-Term Memory[J]. Neural Computation, 1997, 9(8): 1735-1780.
- [9] Yao X, Tang X, Wei W, et al. Spatiotemporal convolutional neural networks for urban air quality prediction: A case study of Beijing[J]. Environmental Pollution, 2018, 239: 647-656.
- [10] Yu H, Yao X, Ren Y. Spatiotemporal graph neural networks for regional fine-grained air pollution prediction[J]. Computers & Geosciences, 2024, 185: 105532.
- [11] Zhang Y, Wang J, Li Z. Fine-grained prediction of VOCs concentrations using machine learning and multi-source monitoring data[J]. Journal of Hazardous Materials, 2024, 467: 132987.
- [12] Lundberg S M, Lee S I. A unified approach to interpreting model predictions[J]. Advances in Neural Information Processing Systems, 2017, 30: 4765-4774.
- [13] Li M, Han J, Xu B. Hybrid XGBoost-LSTM model for early warning of heavy air pollution in North China Plain[J]. Journal of Cleaner Production, 2023, 391: 136215.
- [14] Wang Y, Zhang M, Liu F. Interpretable machine learning for O3 fine-grained prediction in the Pearl River Delta[J]. Environmental Pollution, 2024, 338: 122654.
- [15] Yang S, Li X, Zhao Y. Lightweight deep learning model for real-time fine-grained air pollution prediction[J]. Computers in Industry, 2023, 146: 103859.
- [16] Wang L, Zhao J, Sun M. LSTM-CNN based fine-grained PM2.5 prediction in urban core area using multi-source big data[J]. Sustainable Cities and Society, 2022, 88: 104256.
- [17] Chen Y, Liu Y, Zhang L. STGAT-based regional fine-grained air pollution prediction for Yangtze River Delta urban agglomeration[J]. Atmospheric Pollution Research, 2024, 15: 101897.
- [18] An J, Kim H, Choi Y. The impact of fine-grained air quality prediction on public health protection in urban areas[J]. Environmental Pollution, 2023, 331: 121989.
Article Sidebar
This work is licensed under a Creative Commons Attribution 4.0 International License.