Fine particulate matter (PM_2.5) remains a major air pollutant of significant public health concerns in urban areas. Tracking the long-term temporal variation of PM_2.5 is essential and valuable for evaluation of existing emissions control policies. Yet, past investigations merely focused on the bulk PM_2.5 simply using annual averaged regression, leaving gaps and details of relationship among the PM_2.5 components, source emissions and the policies.

The objective of this thesis is to understand the PM_2.5 source contributions over Hong Kong and the cities in the Pearl River Delta (PRD) region through comprehensive chemical composition data including suitable organic tracers measured over the past decade (2008-2017) and across multiple sites. The long-term temporal variation was insp...[ Read more ]

Fine particulate matter (PM_2.5) remains a major air pollutant of significant public health concerns in urban areas. Tracking the long-term temporal variation of PM_2.5 is essential and valuable for evaluation of existing emissions control policies. Yet, past investigations merely focused on the bulk PM_2.5 simply using annual averaged regression, leaving gaps and details of relationship among the PM_2.5 components, source emissions and the policies.

The objective of this thesis is to understand the PM_2.5 source contributions over Hong Kong and the cities in the Pearl River Delta (PRD) region through comprehensive chemical composition data including suitable organic tracers measured over the past decade (2008-2017) and across multiple sites. The long-term temporal variation was inspected for simple annual variations as well as more sophisticated season-specific trends. The spatial differences were also examined to understand the source origins. This thesis work has revealed several noteworthy findings as summarized below:

(1) Source apportionment of one-year PM_2.5 data at six different locations in Hong Kong was conducted using positive matrix factorization (PMF) receptor modelling. The input data matrix included sugar alcohols and unresolved complex mixture (UCM) in addition to major constituents and source indictive elements. The contrast results between PMF with and without organic tracers revealed the problems of mixed biomass burning and industrial/coal combustion emissions and the vanish of biogenic source in the absent of its relevant organic markers (arabitol and mannitol). The overestimated vehicular emissions retrieved from PMF only using EC as source indicator species can be avoided only when adding hopanes as well as the UCM into the PMF input matrix. This work highlights the advantage and importance to include organic source tracers to achieve more refined and accurate source apportionment.

(2) Monthly averaged PM_2.5 chemical composition in the decade of 2008-2017 at urban Tsuen Wan (TW) in Hong Kong was analyzed by the Seasonal and Trend decomposition by LOESS method coupled with the generalized least square and autoregessive moving average model (STL-GLS-ARMA). The trend analysis confirmed the continuing declining of PM_2.5 and that revealed that the decline was in a rate of -1.5 μgm^-3 per year, in which the sulfate ion and organic matter (OM) contributed the most in the reduction. EC and hopanes which closely related to vehicle emissions control measures implemented by the Hong Kong local government was found to have dropped by at least 60%. While the changes in species from other regional sources, such as Al and Si from dust, K⁺ and levoglucosan from biomass burning, were relatively lower (<50%) than the locally controlled species. The disproportional reduction between secondary sulfate/nitrate and their respective precursor gases was noted and attributable to the nonlinear formation mechanism. In addition, the long-time series reveals that 2011, due to strong La Niña events, is an anomaly year in that most PM_2.5 components were elevated above the adjacent years.

(3) The long-term trend analysis at urban Guangzhou (GZ) and at suburban Nansha (NS), sites in the PRD region, was also conducted by the STL-GLS-ARMA method. Significant decline of PM_2.5 was also recorded at GZ and NS for -3.0 and -1.9 μgm^-3yr^-1, respectively. Sulfate and OM were the main contributors to the reduction of PM_2.5 at those sites as well. Different from the species in Hong Kong, which all showed a decreasing trend, Cu at NS notably increased by 70%. Using mass increment ratio between episodic and normal events as an indicator, we found nitrate was the species attributed to the highly polluted days at GZ and NS whereas dust materials (e.g., Al, Si) contributed most to TW, implying the different spatial variance of PM_2.5 over the sites. This result was further supported by the evidence that the concentration of species at GZ and NS were typically higher than those at TW even for those originated from regional sources.

(4) Sensitivity tests using moving PMF were conducted to identify the most suitable parameters for resolving reliable profiles more sensitive to any changes upon the series of control measures over the past decade. The STL-GLS-ARMA trend analysis on the most appropriate PMF results derived using a time window size of 2-year and interval of 3 months were then conducted for the ten-year PM_2.5 data at TW. Among all the declining source factors, regional mixed industrial sources and local vehicular emission play the dominating roles in the reduction. While shift of source origin of regional sources from southeast to surrounding areas was found in the concentration weighted trajectories analysis.

(5) Moving PMF was also applied to the ten-year data at GZ and NS. The main contributors to the reduction of bulk PM_2.5 over the decade were secondary nitrate and sulfate at GZ and NS, respectively. The next significant sources contributing to the decline of PM_2.5 were secondary sulfate and residual oil for GZ versus secondary nitrate, industrial and vehicular emissions for NS. As for the spatial and temporal variation over the three sites, the regional source generally distributed and varied in similar pattern among them. Obvious reduction of residual oil combustion at GZ verse the steady variation of the same source at NS and TW were observed. It was later confirmed from the local/non-local estimates that the residual oil combustion at GZ were mostly locally emitted. Same transition of regional source origin was also obtained for GZ and NS. More notably, biomass burning contributed to similar amounts regardless of source origins, implying wide presence of biomass burning activities over the Southern China. Lastly, the episodic events over the PRD region were found mainly associated with the regional sources including secondary inorganic aerosol, mixed industrial sources, and biomass burning.