Search

Haze and dust pollution have a significant impact on human production, life, and health. In order to understand the pollution process, the study of these two pollution characteristics is important. In this study, a one-year observation was carried out at the Beijing Southern Suburb Observatory using the MAX-DOAS instrument, and the pollution characteristics of the typical haze and dust events were analyzed. First, the distribution of aerosol extinction (AE) and H2O concentrations in the two typical pollution events were studied. The results showed that the correlation coefficient (r) between H2O and AE at different heights decreased during dust processes and the correlation slope (|k|) increased, whereas r increased and |k| decreased during haze periods. The correlation slope increased during the dust episode due to low moisture content and increased O4 absorption caused by abundant suspended dry crustal particles, but decreased during the haze episode due to a significant increase of H2O absorption. Secondly, the gas vertical column density (VCD) indicated that aerosol optical depth (AOD) increased during dust pollution events in the afternoon, while the H2O VCD decreased; in haze pollution processes, both H2O VCD and AOD increased. There were significant differences in meteorological conditions during haze (wind speed (WD) was <2 m/s, and relative humidity (RH) was >60%) and dust pollution (WD was >4 m/s, and RH was <60%). Next, the vertical distribution characteristics of gases during the pollution periods were studied. The AE profile showed that haze pollution lasted for a long time and changed slowly, whereas the opposite was true for dust pollution. The pollutants (aerosols, NO2, SO2, and HCHO) and H2O were concentrated below 1 km during both these typical pollution processes, and haze pollution was associated with a strong temperature inversion around 1.0 km. Lastly, the water vapor transport fluxes showed that the water vapor transport from the eastern air mass had an auxiliary effect on haze pollution at the observation location. Our results are of significance for exploring the pollution process of tropospheric trace gases and the transport of water vapor in Beijing, and provide a basis for satellite and model verification. [ABSTRACT FROM AUTHOR]

The concentration of PM2.5 is an important index to measure the degree of air pollution. When it exceeds the standard value, it is considered to cause pollution and lower the air quality, which is harmful to human health and can cause a variety of diseases, i.e., asthma, chronic bronchitis, etc. Therefore, the prediction of PM2.5 concentration is helpful to reduce its harm. In this paper, a hybrid model called CNN-BiLSTM-Attention is proposed to predict the PM2.5 concentration over the next two days. First, we select the PM2.5 concentration data in hours from January 2013 to February 2017 of Shunyi District, Beijing. The auxiliary data includes air quality data and meteorological data. We use the sliding window method for preprocessing and dividing the corresponding data into a training set, a validation set, and a test set. Second, CNN-BiLSTM-Attention is composed of the convolutional neural network, bidirectional long short-term memory neural network, and attention mechanism. The parameters of this network structure are determined by the minimum error in the training process, including the size of the convolution kernel, activation function, batch size, dropout rate, learning rate, etc. We determine the feature size of the input and output by evaluating the performance of the model, finding out the best output for the next 48 h. Third, in the experimental part, we use the test set to check the performance of the proposed CNN-BiLSTM-Attention on PM2.5 prediction, which is compared by other comparison models, i.e., lasso regression, ridge regression, XGBOOST, SVR, CNN-LSTM, and CNN-BiLSTM. We conduct short-term prediction (48 h) and long-term prediction (72 h, 96 h, 120 h, 144 h), respectively. The results demonstrate that even the predictions of the next 144 h with CNN-BiLSTM-Attention is better than the predictions of the next 48 h with the comparison models in terms of mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination (R2). [ABSTRACT FROM AUTHOR]