Global PM2.5 Prediction and Associated Mortality to 2100 under Different Climate Change Scenarios.

Ambient fine particulate matter (PM2.5) has severe adverse health impacts, making it crucial to reduce PM2.5 exposure for public health. Meteorological and emissions factors, which considerably affect the PM2.5 concentrations in the atmosphere, vary substantially under different climate change scenarios. In this work, global PM2.5 concentrations from 2021 to 2100 were generated by combining the deep learning technique, reanalysis data, emission data, and bias-corrected CMIP6 future climate scenario data. Based on the estimated PM2.5 concentrations, the future premature mortality burden was assessed using the Global Exposure Mortality Model. Our results reveal that SSP3-7.0 scenario is associated with the highest PM2.5 exposure, with a global concentration of 34.5 µg/m3 in 2100, while SSP1-2.6 scenario has the lowest exposure, with an estimated of 15.7 µg/m3 in 2100. PM2.5-related deaths for individuals under 75 years will decrease by 16.3 and 10.5% under SSP1-2.6 and SSP5-8.5, respectively, from 2030s to 2090s. However, premature mortality for elderly individuals (>75 years) will increase, causing the contrary trends of improved air quality and increased total PM2.5-related deaths in the four SSPs. Our results emphasize the need for stronger air pollution mitigation measures to offset the future burden posed by population age.

Development of an integrated machine-learning and data assimilation framework for NOx emission inversion.

As major air pollutants, nitrogen oxides (NOx, mainly comprising NO and NO2) not only have adverse effects on human health but also contribute to the formation of secondary pollutants, such as ozone and particulate nitrate. To acquire reasonable NOx simulation results for further analysis, a reasonable emission inventory is needed for three-dimensional chemical transport models (3D-CTMs). In this study, a comprehensive emission adjustment framework for NOx emission, which integrates the simulation results of the 3D-CTM, surface NO2 measurements, the three-dimensional variational data assimilation method, and an ensemble back propagation neural network, was proposed and applied to correct NOx emissions over China for the summers of 2015 and 2020. Compared with the simulation using prior NOx emissions, the root-mean-square error, normalized mean error, and normalized mean bias decreased by approximately 40 %, 40 %, and 60 % in NO2 simulation using posterior NOx emissions corrected by the framework proposed in this work. Compared with the emissions for 2015, the NOx emission generally decreased by an average of 5 % in the simulation domain for 2020, especially in Henan and Anhui provinces, where the percentage reductions reached 24 % and 19 %, respectively. The proposed framework is sufficiently flexible to correct emissions in other periods and regions. The framework can provide reliable and up-to-date emission information and can thus contribute to both scientific research and policy development relating to NOx pollution.

Estimation and variation analysis of secondary inorganic aerosols across the Greater Bay Area in 2005 and 2015.

As the concentrations of primary components of fine particulate matter (PM2.5) have substantially decreased, the contribution of secondary inorganic aerosols to PM2.5 pollution has become more prominent. Therefore, understanding the variations in and characteristics of secondary inorganic aerosols is vital to further reducing PM2.5 concentrations in the future. In this study, an ensemble back-propagation neural network model was built by combining 3D numerical models, observation data, and machine learning methods, to estimate the concentrations of secondary inorganic aerosols (SO2-4, NO-3, and NH+4) across the Greater Bay Area (GBA) in 2005 and 2015. The ensemble model provided a better estimation than the 3D numerical air quality model, with higher correlation coefficients (approximately 0.85) and lower root mean square errors. The model revealed that the concentrations of the SO2-4, NO-3, and NH+4 decreased by 1.91, 0.20, and 0.49 µg/m3, respectively, from 2005 to 2015. To investigate the oxidation and acidy of sulfate, the sulfur oxidation ratio (SOR), degree of sulfate neutralization (DSN), and particle neutralization ratio (PNR) were calculated and analyzed for 2005 and 2015 across the GBA region. The SOR slightly increased in summer, but decreased in other seasons in 2015, indicating the overall weaker sulfate chemical formation due to sulfur emission control measures. The increasing DSN and PNR indicated that more sulfate was neutralized due to reduced sulfur emission and increased ammonia availability. Our study suggests that more effort is needed to control ammonia emission to further reduce the concentrations of SO2-4, NO-3, and NH+4 across the GBA region in the future.

Impacts of urbanization and long-term meteorological variations on global PM2.5 and its associated health burden.

PM2.5 pollution has adverse health effects on humans. Urbanization and long-term meteorological variations play important roles in influencing the PM2.5 concentration and its associated health effects. Our results indicate that the urbanization process can enhance the PM2.5 concentration globally. The PM2.5-caused mortality density (deaths/100 km2) is also positively correlated with the urbanization degree in both developed and developing countries. The results from machine learning technique revealed that the meteorology-driven variation in PM2.5-caused health burden has increased with the increase in the urbanization degree from 1980 to 2018, suggesting that residents living in urban areas are more vulnerable to experiencing unfavorable meteorological conditions (e.g. low wind speed and planetary boundary layer height). The maximum difference in PM2.5-caused mortality due to the variation in annual meteorological conditions (between 2013 and 1986) was 270 600 (196 800-317 900). Our findings indicate an urgent need to understand the driving force behind the appearance of unfavorable meteorological situations and propose suitable climate mitigation measures.

Estimations of Long-Term nss-SO42- and NO3- Wet Depositions over East Asia by Use of Ensemble Machine-Learning Method.

Wet deposition of non-sea-salt sulfate (nss-SO42-) and nitrate (NO3-), derived from anthropogenic emissions of SO2 and NOx, exerts adverse effects on ecosystems. In this work, an ensemble back-propagation neural network was proposed to estimate the long-term wet depositions of nss-SO42- (2005-2017) and NO3- (2001-2014) over East Asia in 10 km resolution. The R2 values for the 10-fold cross-validation of annual wet depositions of nss-SO42- and NO3- were 0.90 and 0.85, respectively. The hotspots of the wet deposition of these two acidic species span southwestern, central, and eastern China. The molar ratio of NO3- to nss-SO42- increased in 10 out of 12 analyzed East Asian countries from 2005 to 2014, which indicates that the acidity in rainwater shifts from the sulfur type to nitrogen type over most of the regions. The wet deposition on the four ecosystems (forest, grassland, cropland, and freshwater body) was also analyzed. Results showed that the nss-SO42- wet deposition on 25.5% of freshwater bodies in 2015 and NO3- wet deposition on 21.7% of grassland in 2014 exceeded the ecosystem empirical critical loads (25 kg/ha sulfate and 2 kg N/ha) in East Asia. Thus, more stringent and regionally collaborative sulfur and nitrogen emission-control measures are urgently needed to protect the ecosystem of East Asia.