| Autor: |
Wu, Shuaiwen, Li, Hengkai, Zhou, Yanbing, He, Yonglan |
| Předmět: |
|
| Zdroj: |
Theoretical & Applied Climatology; Apr2024, Vol. 155 Issue 4, p2787-2799, 13p |
| Abstrakt: |
As one of China's national urban agglomerations, the middle reaches of the Yangtze River urban agglomeration (MRYRA) has suffered from serious PM2.5 pollution for a considerable period. Using deep learning methods to fit the quantitative relationship between aerosol optical depth (AOD) and PM2.5 (AOD-PM2.5) has received extensive attention. In this study, based on the correlation between AOD and PM2.5 data, a new network model BiCNN is proposed by using the satellite-ground synergy method, aiming to obtain the spatial distribution of PM2.5 concentration in the MRYRA. Data from MRYRA 2019 is used for model training and testing. The results show that BiCNN exhibits the best estimation on the annual scale, with a fit coefficient R2 of 0.836 and the smallest error, RMSE of 6.746 and MAPE of 12.497 compared to the benchmark model. The model effectively improves the relationship between AOD-PM2.5. The spatial distribution of PM2.5 in the MRYRA from 2019 to 2022 by applying the BiCNN model inversion shows that the PM2.5 concentration is higher in 2020, but the overall air quality improves year by year. Seasonal evaluation results show that BiCNN has a significant advantage in PM2.5 estimation in autumn, with R2 reaching 0.883. The spatial distribution of PM2.5 shows a spatial pattern of higher in spring and winter, lower in summer and autumn, and transition in autumn. The BiCNN model integrates the advantages of Convolutional Neural Network (CNN) and Bidirectional Long Short-Term Memory Neural Network (BiLSTM), makes full use of historical and future data, and can effectively capture the temporal and spatial characteristics between cofactors and PM2.5 data. It has achieved ideal results in the PM2.5 estimation of the MRYRA and provides an effective scientific method for air pollution control in urban agglomerations. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
Full text from SpringerLink