Dataset Information

Air Pollutant Concentration Forecasting Using Long Short-Term Memory Based on Wavelet Transform and Information Gain: A Case Study of Beijing

ABSTRACT: Air pollutant concentration forecasting is an effective way which protects health of the public by the warning of the harmful air contaminants. In this study, a hybrid prediction model has been established by using information gain, wavelet decomposition transform technique, and LSTM neural network, and applied to the daily concentration prediction of atmospheric pollutants (PM2.5, PM10, SO2, NO2, O3, and CO) in Beijing. First, the collected raw data are selected by feature selection by information gain, and a set of factors having a strong correlation with the prediction is obtained. Then, the historical time series of the daily air pollutant concentration is decomposed into different frequencies by using a wavelet decomposition transform and recombined into a high-dimensional training data set. Finally, the LSTM prediction model is trained with high-dimensional data sets, and the parameters are adjusted by repeated tests to obtain the optimal prediction model. The data used in this study were derived from six air pollution concentration data in Beijing from 1/1/2014 to 31/12/2016, and the atmospheric pollutant concentration data of Beijing between 1/1/2017 and 31/12/2017 were used to test the predictive ability of the data set test model. The results show that the evaluation index MAPE of the model prediction is 7.45%. Therefore, the hybrid prediction model has a higher value of application for atmospheric pollutant concentration prediction, because this model has higher prediction accuracy and stability for future air pollutant concentration prediction.

SUBMITTER: Liu B

PROVIDER: S-EPMC7545461 | biostudies-literature | 2020 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:To curb the spread of the coronavirus, China implemented lockdown policies on January 23, 2020. The resulting extreme changes in human behavior may have influenced the air pollutants concentration. However, despite these changes, hazy weather persisted in Shanghai and became a public issue. This study aims to investigate air pollutant mass concentration changes during the lockdown in Shanghai. Air pollutant mass concentration data and meteorological data during the pre-lockdown period and the level I response lockdown period were analyzed by statistical analysis and a Lagrangian particle diffusion model. The data was classified in three periods: P1 (pre-lockdown: 10 days before the Spring Festival), P2 (the first 10 days after lockdown: during the Spring Festival celebration), and P3 (the second 10 days after lockdown: after the Spring Festival). Data for the same period in 2019 were used as a reference. The results indicate that the Spring Festival holiday in 2019 resulted in a reduction in energy consumption, which led to a decrease in PM2.5 (26.4%) and NO2 (43.41%) mass concentration, but an increase in ozone mass concentration (31.39%) in P2 compared with P1. The integrated effect of the Spring Festival holiday and lockdown in 2020 resulted in a decrease in PM2.5 (36.5%) and NO2 (51.9%) mass concentrations, but an increase in ozone mass concentration (43.8%) in P2 compared with P1. After the Spring Festival, the mass concentrations of PM2.5, SO2, and NO2 increased by 74.41%, 5.52%, and 53.28%, respectively in P3 compared with P2 in 2019. However, PM2.5 and SO2 concentrations in 2020 continued to decrease, by 14.74% and 4.61%, respectively, while NO2 mass concentration increased by 7.82% in P3 compared with P2. We also found that PM2.5 mass concentration is susceptible to regional transmission from the surrounding cities. PM2.5 and other gaseous pollutants show different correlations in different periods, while NO2 and O3 always show a strong negative correlation. The principal components before the Spring Festival in 2019 were O3 and NO2, and after the Spring Festival, they were PM2.5 and CO, while the principal components before the lockdown in 2020 were PM2.5 and CO, and during lockdown they were O3 and NO2.

Project description:A direct and quantitative linkage of air pollution-related health effects to emissions from different sources is critically important for decision-making. While a number of studies have attributed the PM2.5-related health impacts to emission sources, they have seldom examined the complicated nonlinear relationships between them. Here we investigate the nonlinear relationships between PM2.5-related premature mortality in the Beijing-Tianjin-Hebei (BTH) region, one of the most polluted regions in the world, and emissions of different pollutants from multiple sectors and regions, through a combination of chemical transport model (CTM), extended response surface model (ERSM), and concentration-response functions (CRFs). The mortalities due to both long-term and short-term exposures to PM2.5 are most sensitive to the emission reductions of primary PM2.5, followed by NH3, nonmethane volatile organic compounds and intermediate volatility organic compounds (NMVOC+IVOC). The sensitivities of long-term mortality to emissions of primary organic aerosol (POA), NMVOC+IVOC and SO2 do not change much with reduction ratio, whereas the sensitivities to primary inorganic PM2.5 (defined as all chemical components of primary PM2.5 other than POA), NH3 and NOx increase significantly with the increase of reduction ratio. The emissions of primary PM2.5, especially those from the residential and commercial sectors, contribute a larger fraction of mortality in winter (57-70%) than in other seasons (28-42%). When emissions of multiple pollutants or those from both local and regional emissions are controlled simultaneously, the overall sensitivity of long-term mortality is much larger than the arithmetic sum of the sensitivities to emissions of individual pollutants or from individual regions. This implies that a multi-pollutant, multi-sector and regional joint control strategy should be implemented to maximize the marginal health benefits. For NOx emissions, we suggest a nationwide control strategy which significantly enhances the effectiveness for reducing mortality by avoiding possible side effects when only the emissions within the BTH region are reduced.

Dataset Information

Air Pollutant Concentration Forecasting Using Long Short-Term Memory Based on Wavelet Transform and Information Gain: A Case Study of Beijing

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets