Importance of secondary decomposition in the accurate prediction of daily-scale ozone pollution by machine learning.

Machine learning (ML) models have been proven as a reliable tool in predicting ambient pollution concentrations at various places in the world. However, their performance in predicting the maximum daily 8-h averaged ozone (MDA8 O3), the metric often used for O3 pollution assessment and management, is relatively poorer. This is largely resulted from more irregular data fluctuations of the MDA8 O3 levels governed collectively by the synoptic condition, local photochemistry, and long-range transport. In order to improve the prediction accuracy of MDA8 O3, this study developed a secondary decomposition ML model framework which coupled the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) as the primary decomposition, the variational mode decomposition (VMD) as secondary decomposition, and the gate recurrent unit (GRU) ML model. By applying this secondary decomposition model framework on MDA8 O3 prediction for the first time, we showed that the prediction accuracy of MDA8 O3 is largely improved from R2 of 0.46 and RMSE of 30.4 µg/m3 for GRU without decomposition to R2 of 0.91 and RMSE of 12.6 µg/m3 over the Pearl River Delta of China. We also found that the prediction accuracy rate of O3 pollution non-attainments, an essential indicator for initiating contingency O3 pollution control, improved greatly from 14.9 % for GRU without decomposition to 72.5 %. The performance of O3 pollution non-attainment prediction is relatively higher in southwestern PRD, which is mainly due to greater number and severity of O3 non-attainments in southwestern cities located downwind of the emission hotspot area at central PRD. This study underscored the importance of secondary decomposition in accurately predicting daily-scale O3 concentration and non-attainments over the PRD, which can be extended to other photochemically active region worldwide to improve their O3 prediction accuracy and assist in O3 contingency control.

Integrated assessment of volatile organic compounds from industrial biomass boilers in China: emission characteristics, influencing factors, and ozone formation potential.

Industrial biomass boilers (IBBs) are widely promoted in China as a type of clean energy. However, they emit large amount of volatile organic compounds (VOCs) and the emission characteristics and the underlying factors are largely unknown due to the sampling difficulties. In this study, three wood pellet-fueled and two wood residue-fueled IBBs were selected to investigate the characteristics of VOC emissions and to discover their underlying impacting factors. The emission factor of VOCs varied from 21.6 ± 2.8 mg/kg to 286.2 ± 10.8 mg/kg for the IBBs. Oxygenated VOCs (OVOCs) were the largest group, contributing to 30.3 - 73.6% of the VOC emissions. Significant differences were revealed in the VOC source profiles between wood pellet-fueled and wood residue-fueled IBBs. Operating load, excess air, furnace temperature, and fuel type were identified as the primary factors influencing VOC emissions. The excess air coefficient should be limited below 3.5, roughly corresponding to the operating load of 62% and furnace temperature of 630 °C, to effectively reduce VOC emissions. VOC emissions also showed great differences in different combustion phases, with the ignition phase having much greater VOC emissions than the stable combustion and the ember phases. The ozone formation potential (OFP) ranged from 4.3 to 31.2 mg/m3 for the IBBs, and the wood residue-fueled IBBs yielded higher OFP than the wood pellet-fueled ones. This study underscored the importance of OVOCs in IBB emissions, and reducing OVOC emissions should be prioritized in formulating control measures to mitigate their impacts on the atmospheric environment and human health.

Effects of organochlorine exposure on male reproductive disorders in an electronic waste area of South China.

Several studies suggest that organochlorine exposure can affect male reproductive functions, causing poor semen quality, endocrine disruption, or dysregulation of thyroid hormones. This study uses multiple linear regression (MLR) models to analyze the correlation between male reproductive functions and polychlorinated biphenyl (PCBs) congeners or p,p'-DDE levels in serum, semen, and indoor dust samples. Multiple comparisons were all adjusted using the false discovery rate (FDR). The results revealed that the PCB congener levels in seminal plasma were significantly associated with the quality parameters of human semen (i.e., sperm count, morphology, and motility) and thyroid hormones after adjusting for covariates, e.g., associations of the sperm concentration with levels of CB105 (ß = -0.323, 95% CI: -0.561, -0.085, p = 0.009), CB44 (ß = 0.585, 95% CI: 0.290, 0.880, p < 0.001), and CB66 (ß = -0.435, 95% CI: -0.728, -0.143, p = 0.004) in the seminal plasma were observed. Correlations between serum pollutants levels and the semen quality, reproductive hormones, or thyroid hormones were also observed. Moreover, our results demonstrate that the quantification of PCBs in seminal plasma can better describe male reproductive disorders than that in serum or dust. Organochlorine exposure measured in serum or dust, especially in seminal plasma, was associated with semen quality, as well as reproductive and thyroid hormones, thus suggesting that the impacts of persistent pollutants on male reproductive health require further investigation.

Quantifying the relative importance of major tracers for fine particles released from biofuel combustion in households in the rural North China Plain.

Biomass burning tracers have been widely used to identify biomass burning types, but such tools can sometimes cause large uncertainties in the source attribution studies of PM2.5 (particles with an aerodynamic diameter of smaller than 2.5 µm). To quantify the relative importance of the major biomass burning tracers in PM2.5 released from biofuels combusted in the North China Plain, combustion experiments under the smoldering and flaming combustion conditions were conducted using nine types of typical household biofuels including two types of agricultural wastes, five types of hardwoods, one softwood, and one mixed wood briquette. PM2.5 samples were collected from the combustion experiments and source profiles of PM2.5 were thus determined for various biofuels under the two different combustion conditions. Carbonaceous species including organic carbon (OC) and elemental carbon (EC) were the major chemical components of the PM2.5 released from combustion of all the tested biofuels, with mass fractions of 37-45% and 4-7% under the smoldering condition and 11-25% and 7-29% under the flaming condition, respectively. Higher mass fractions of water-soluble inorganic ions (WSIIs, e.g., K+ and Cl-) in PM2.5 were observed under the flaming than smoldering combustion condition, while anhydrosugars (levoglucosan (LG) and mannosan (MN)) presented in an opposite pattern. The average LG/MN ratio in PM2.5 changed significantly with biofuel type (20-55 for agricultural wastes, 10-22 for hardwoods (except elm) and 3-6 for softwood), but varied little with combustion condition. In contrast, the K+/LG ratio in PM2.5 varied significantly between smoldering (<0.2) and flaming (>0.6) combustion conditions for all the biofuel types except softwood. Results from this study suggested that the ratio LG/MN was the best tracer for identifying the biofuel types and the ratio K+/LG is suitable for identifying the combustion conditions in this region.