Search

Long-chain unsaturated fatty acids (uFAs), such as oleic acid, undergo rapid degradation via heterogeneous reactions with atmospheric oxidants upon emission. The oxidation mechanism and kinetics have been extensively studied in laboratory experiments. However, quantitative knowledge of degradation rates under real-world atmospheric conditions is scarce. We obtained the nighttime decay rates of three cooking-related uFAs using a relative rate approach applied to bihourly measured data in urban Shanghai. The estimated lifetime of oleic acid was 6 h under conditions of similar to 12 ppb ozone and 60%-100% relative humidity encountered at our urban location or an inferred similar to 2 h at a higher ozone level of similar to 40 ppb. The decay rates of elaidic and linoleic acid are determined to be 0.62 and 1.37 that of oleic acid, respectively. This work provides the first kinetic data pertaining to real-world conditions. They are valuable for constraining the modeling of heterogeneous aging of ambient organic aerosols.

Fine particulate matter (PM2.5)-bound nonpolar organic compounds (NPOCs), including polycyclic aromatic hydrocarbons (PAHs) and alkanes, are commonly used as typical molecular markers for detailed source identification. Online thermal desorption aerosol gas chromatography–mass spectrometry (TAG) system can obtain ambient data with hourly resolution, which is of great importance for investigating the diurnal characteristics and refined source identification of NPOCs. From June to October 2020, hourly ambient aerosol samples were collected and analyzed to investigate the characteristics and sources of 14 PAHs and 15 alkanes (C21-C35) in PM2.5 using TAG at a suburban site of Baoshan district in Shanghai, China. The average concentration of summed PAHs and alkanes during the sampling period was 1.27 ± 1.4 ng/m3 and 8.87 ± 3.46 ng/m3, respectively, in which Benzo[b]fluoranthene (BbF), Benzo[ghi]perylene (BghiP) and Indeno[1,2,3-cd]pyrene (IcdP) are the dominant PAHs species, with n-Heptacosane (C27), n-Nonacosane (C29) and n-Hentriacontane (C31) being the most abundant n-alkane species. Carbon preference index (CPI) and carbon maximum (Cmax) number indicated that the sources of alkanes shifted from biogenic-oriented (such as plant wax) in the summer to anthropogenic-dominated (such as fossil fuels) in the autumn. Results from trajectory cluster analysis and potential source contribution function (PSCF) modeling showed that alkanes were mainly from the middle and lower reaches of the Yangtze River Plain including Anhui, Jiangxi, and Zhejiang provinces, while PAHs were mainly from northeastern China. Positive Matrix Factorization (PMF) model results indicated that gasoline (41.48%) and diesel (21.82%) were the two major sources of PM2.5-bound PAHs in summer and fall of 2020 in Shanghai, followed by coal consumption or catering (19.96%) and biomass burning (16.74%). Diurnal variation of PAHs sources resolved by PMF showed characteristic features consistent with the correspo

Air quality in megacities is significantly impacted by emissions from vehicles and other urban-scale human activities. Amid the outbreak of Coronavirus (COVID-19) in January 2020, strict policies were in place to restrict people's movement, bringing about steep reductions in pollution activities and notably lower ambient concentrations of primary pollutants. In this study, we report hourly measurements of fine particulate matter (i.e., PM2.5) and its comprehensive chemical speciation, including elemental and molecular source tracers, at an urban site in Shanghai spanning a period before the lockdown restriction (BR) (1 to 23 Jan. 2020) and during the restriction (DR) (24 Jan. to 9 Feb. 2020). The overall PM2.5 was reduced by 27% from 56.2 +/- 40.9 (BR) to 41.1 +/- 25.3 mu g m(-3) (DR) and the organic carbon (OC) in PM2.5 was similar, averaged at 5.45 +/- 2.37 (BR) and 5.42 +/- 1.75 mu gC m(-3) (DR). Reduction in nitrate was prominent, from 18.1 (BR) to 9.2 mu g m(-3) (DR), accounting for most of the PM2.5 decrease. Source analysis of PM2.5 using positive matrix factorization modeling of comprehensive chemical composition, resolved nine primary source factors and five secondary source factors. The quantitative source analysis confirms reduced contributions from primary sources affected by COVID-19, with vehicular emissions showing the largest drop, from 4.6 (BR) to 0.61 mu g m(-3) (DR) and the percentage change (-87%) in par with vehicle traffic volume and fuel sale statistics (-60% to -90%). In the same time period, secondary sources are revealed to vary in response to precursor reductions from the lockdown, with two sources showing consistent enhancement while the other three showing reductions, highlighting the complexity in secondary organic aerosol formation and the nonlinear response to broad primary precursor pollutants. The combined contribution from the two secondary sources to PM2.5 increased from 7.3 +/- 6.6 (BR) to 14.8 +/- 9.3 mu g m(-3) (DR), partially

Epidemiological evidence linking source-specific ambient particulate matter with aerodynamic diameter <2.5 mu m (PM2.5) and cardiac arrhythmias is limited. In this study, we investigated the impact of source-specific PM2.5 on cardiac arrhythmias in a panel of forty-five healthy adults living in Beijing, China, between 2015 and 2016. Repeated measures of 24-hour electrocardiograms were conducted during clinical visits, and daily counts of four arrhythmia events including supraventricular premature beat (SVPB), atrial tachycardia (AT), premature ventricular contraction (PVC) and ventricular tachycardia (VT) were recorded. One hundred forty-seven constituents in PM2.5 were analyzed for collected particulate samples, in which fifty-six of them above laboratory defection limits were selected for source apportionment analysis using positive matrix factorization. The average contributions of identified five major sources to PM2.5 were 45.9% from secondary nitrate/sulfate, 18.0% from coal combustion, 16.9% from crustal soil, 13.8% from biomass burning, and 5.4% from cooking. Generalized estimating equation models were used to estimate relative risks (RR) of arrhythmias in association with interquartile-range (IQR) increases in PM2.5 constituents and specific sources. Total PM2.5 mass as well as several combustion related constituents were found of significant impacts on increased risks of arrhythmia events. Among the identified sources of PM2.5, coal burning has been found the major source that associated with increased risks of SVPB, PVC and VT with RR of 1.19 [95% confidence intervals (CI): 1.04,1.36] to 1.64 (95% CI: 1.35,2.00). PM2.5 from combustion related secondary nitrate/sulfate was also found of significant impact on SVPB and AT, followed by PM2.5 from biomass burning and crustal soil. Our results indicated that PM2.5 from anthropogenic activity related sources were most responsible for increased risks of arrhythmia events. Our findings enhance the understanding of

Positive matrix factorization (PMF) is a widely used tool to do the source apportionment, compared with traditional PMF, the organic tracer based PMF has the advantage of providing more chemical information and resolve more specific source type. Although multiple studies incorporating organic tracers in PMF have been performed, very few cover comprehensive data set including other chemical components such as elements and major components. Work in this thesis aims to improve the understanding of the atmospheric behavior of the organic tracers, and the application of them as source indicators in PMF. Several case studies have been performed using organic tracers-based source apportionment. Besides, the impact of time resolution is also examined using hourly measured aerosol components. The major findings are summarized below: (1) Source apportionment of PM2.5 was performed using PMF based on one-year 24-h offline filters collected at two sites in Macau in 2015. The chemical analysis included major inorganic ions, organic carbon and elemental carbon (OC and EC), elements and organic tracers including non-polar organic compounds (NPOCs) and sugar compounds. Trajectory analysis and conditional probability function (CPF) analysis were performed on the PMF resolved factor contributions to distinguish the local and regional sources. The role of organic tracers was examined by comparing with two traditional PMF runs without organics. The results showed similar contributing sources, while noticeable differences of absolute PM2.5 contributions from three PMFs were observed. Exclusion of organics overestimated source contributions from biomass burning and vehicle exhaust factor due to absence of source-specific tracers. (2) Based on a multi-year and multi-site NPOC data set, spatial and seasonal variations, correlation analysis and ratio-ratio plots were used to investigate the source information and degradation of NPOC tracers. In summer, NPOCs showed dis

Positive matrix factorization (PMF) is a widely used tool to do the source apportionment, compared with traditional PMF, the organic tracer based PMF has the advantage of providing more chemical information and resolve more specific source type. Although multiple studies incorporating organic tracers in PMF have been performed, very few cover comprehensive data set including other chemical components such as elements and major components. Work in this thesis aims to improve the understanding of the atmospheric behavior of the organic tracers, and the application of them as source indicators in PMF. Several case studies have been performed using organic tracers-based source apportionment. Besides, the impact of time resolution is also examined using hourly measured aerosol components. The major findings are summarized below: (1) Source apportionment of PM2.5 was performed using PMF based on one-year 24-h offline filters collected at two sites in Macau in 2015. The chemical analysis included major inorganic ions, organic carbon and elemental carbon (OC and EC), elements and organic tracers including non-polar organic compounds (NPOCs) and sugar compounds. Trajectory analysis and conditional probability function (CPF) analysis were performed on the PMF resolved factor contributions to distinguish the local and regional sources. The role of organic tracers was examined by comparing with two traditional PMF runs without organics. The results showed similar contributing sources, while noticeable differences of absolute PM2.5 contributions from three PMFs were observed. Exclusion of organics overestimated source contributions from biomass burning and vehicle exhaust factor due to absence of source-specific tracers. (2) Based on a multi-year and multi-site NPOC data set, spatial and seasonal variations, correlation analysis and ratio-ratio plots were used to investigate the source information and degradation of NPOC tracers. In summer, NPOCs showed dis

State-of-the-art speaker recognition systems comprise an x-vector (or i-vector) speaker embedding front-end followed by a probabilistic linear discriminant analysis (PLDA) backend. The effectiveness of these components relies on the availability of a large collection of labeled training data. In practice, it is common that the domains (e.g., language, demographic) in which the system are deployed differs from that we trained the system. To close the gap due to the domain mismatch, we propose an unsupervised PLDA adaptation algorithm to learn from a small amount of unlabeled in-domain data. The proposed method was inspired by a prior work on feature-based domain adaptation technique known as the correlation alignment (CORAL). We refer to the model-based adaptation technique proposed in this paper as CORAL+. The efficacy of the proposed technique is experimentally validated on the recent NIST 2016 and 2018 Speaker Recognition Evaluation (SRE'16, SRE'18) datasets., Comment: 5 pages

This paper presents an experimental study on deep speaker embedding with an attention mechanism that has been found to be a powerful representation learning technique in speaker recognition. In this framework, an attention model works as a frame selector that computes an attention weight for each frame-level feature vector, in accord with which an utterancelevel representation is produced at the pooling layer in a speaker embedding network. In general, an attention model is trained together with the speaker embedding network on a single objective function, and thus those two components are tightly bound to one another. In this paper, we consider the possibility that the attention model might be decoupled from its parent network and assist other speaker embedding networks and even conventional i-vector extractors. This possibility is demonstrated through a series of experiments on a NIST Speaker Recognition Evaluation (SRE) task, with 9.0% EER reduction and 3.8% min_Cprimary reduction when the attention weights are applied to i-vector extraction. Another experiment shows that DNN-based soft voice activity detection (VAD) can be effectively combined with the attention mechanism to yield further reduction of minCprimary by 6.6% and 1.6% in deep speaker embedding and i-vector systems, respectively., Comment: SLT 2018 (Workshop on Spoken Language Technology)

Knowledge of the relative abundance of primary organic aerosol (POA) and secondary organic aerosol (SOA) forms an important scientific basis for formulating particulate matter (PM) control policies. Taking advantage of a comprehensive chemical composition data set of PM2.5 including both POA and SOA tracers (most notably, SOA tracers of a few biogenic voltaic organic compound precursors), we investigate the impact of inclusion of SOA tracers on the source apportionment of organic carbon (OC) and PM2.5 in the Pearl River Delta region of China using positive matrix factorization (PMF). In PMF runs incorporating SOA tracers (PMFw), ten PMF factors were resolved including four secondary factors: (1) SOA I (alpha-pinene, beta-caryophyllene, and naphthalene-derived SOA), (2) SOA II (isoprene-derived SOA), (3) a secondary sulfate factor, and (4) a secondary nitrate factor. In PMF tests without SOA tracers (PMFwo), the SOA I and SOA II factors could not be extracted, but the remaining eight source factors were resolved. Among the eight common source factors, the industrial emission factor, identified by high loadings of Zn and Pb, showed the largest variations between PMFw and PMFwo solutions. The source contributions of SOA I and SOA II resolved in PMFw were largely shifted to the industry emission source in PMFwo. Secondary organic carbon (SOC) summed from the four secondary factors in PMFw contributed similar to 40% (4.47 mu gC/m(3)), and the SOC estimate by PMFwo (3.51 mu gC/m(3)) was 21% lower due to the inability to extract SOA I and SOA II. Secondary PM2.5 by PMFwo was 6% lower than that by PMFw (23.7 vs 25.2 mu g/m(3)). The PMFw results indicated that SOC from specific precursors may have different formation pathways than secondary sulfate and nitrate formation processes, and their source contributions could not be properly resolved wit