Supplement of Measurement report: Characterization and source apportionment of coarse particulate matter in Hong Kong: insights into the constituents of uniden- tified mass and source origins in a coastal city in southern China

. Coarse particulate matter (i.e., PM with aerodynamic diameter between 2.5 and 10 micrometers or PM coarse ) has been increasingly recognized of its importance in PM 10 regulation because of its growing proportion in PM 10 and the accumulative evidence for its adverse health impact. In this work, we present comprehensive PM coarse speciation results 15 obtained through a one-year long (January 2020–February 2021) joint PM 10 and PM 2.5 chemical speciation study in Hong Kong, a coastal and highly urbanized city in southern China. The annual average concentration of PM coarse is 14.9±8.6 μg m –3 (±standard deviation), accounting for 45 % of PM 10 (32.9±18.5 μg m –3 ). The measured chemical components explain ~75 % of the PM coarse mass. The unexplained part is contributed by unmeasured geological components and residue liquid water content, supported by analyses by positive matrix factorization (PMF) and the thermodynamic equilibrium model ISORROPIA 20 II. The PM coarse mass is apportioned to four sources resolved by PMF, namely soil dust, copper-rich dust, fresh sea salt, and an aged sea salt factor containing secondary inorganic aerosols (mostly nitrate). Back-trajectory cluster analysis reveals Cheung et al. (2011) reported an up to 25 % contribution from such unidentified mass in Los Angeles area, while Putaud et al. (2010) reported 6–43 % in urban Europe. Although it has been suggested that the unidentified mass was associated with liquid water content and mineral components, their exact contributions have remained largely uncharacterized. By using positive matrix factorization (PMF), we showed that the unidentified masses can be allocated to the resolved sources, providing qualitative and quantitative information on their origins. We propose the unidentified mass in PM coarse in our study region is mainly composed of unmeasured mineral components and liquid water content. The measured PM coarse in its entirety was successfully apportioned to various contributing sources by PMF, and the potential source origins are identified using backward air mass trajectory analysis. With the robust source apportionment analysis, we found that fugitive dust associated with regional influence is the dominant contributor of high PM coarse loading in Hong Kong. The methodology and results from this study can serve to provide guidance to other locations with similar monitoring needs. Integrated Trajectory (HYSPLIT) model using study region, elucidating the roles of coarse particles in mediating secondary aerosol formation, and examining the potential health burden of PM coarse exposure through oxidative potential measurement.


S1 PMcoarse speciation data quality
The quality of the speciation data is evaluated by examining the consistency between species concentrations obtained by different analytical methods. The evaluation is conducted for each PM size group. Deming regression is applied in the examination using the Scatter Plot computer program developed by Wu available at https://doi.org/10.5281/zenodo.832417 (Wu and Yu, 2018). The results are given in Fig. S1. Sulfate measured by IC and total S by ED-XRF exhibit an excellent consistency (R 2 = 0.99) in PM10 and PM2.5 samples ( Fig. S1a and b). Apart from validating the chemical analyses, the result also validates the performance of the four samplers given the sulfate and total S in PM10 and PM2.5 were measured on four separate filters. The result for PMcoarse is scattered (Fig. S1c) because the quantification is derived from the difference between two close values as both species mostly exist in PM2.5. Comparison between K + and K shows similar results ( Fig. S1d-f). Cland total Cl, which predominantly exist in the coarse mode, exhibit high consistency in PMcoarse, with R 2 of 0.91 (Fig. S1i). The cation and anion equivalence are highly consistent in PM10 and PM2.5, displaying slope and R 2 values close to 1 (Fig. S1j-k). The cation equivalence appears to exceed anion for PMcoarse as shown in Fig. S1l, plausibly because carbonate was not measured.
The gravimetric mass is consistent with the PM data obtained from continuous monitoring by automated analyzers (oscillating microbalance/beta attenuation) at the same site, with slope and R 2 values being 1.08-1.21 and 0.95-0.98, respectively . The sum of chemical species strongly correlates with and is lower than the gravimetric mass ( Fig. S2d-f). The difference is attributed to unmeasured ions, metal oxides, and non-carbon constituents in organic compounds. PM mass is reconstructed as the sum of geological material, organics, EC, NH4 + , SO4 2-, NO3 -, Na + , Mg 2+ , Cl -, and non-crustal elements. The geological material is estimated by multiplying coefficients accounting for oxide in the crustal elements, i.  [Fe]. Coarse mode K is considered as a component in geological material given its strongly association with Si (R 2 = 0.98). The organic mass in PM2.5 is approximated to be 1.6× [OC] assuming organic composition in typical urban atmosphere while that in PMcoarse is 2.0×[OC] considering coarse mode organics are typically associated with biological particles (Turpin and Lim, 2001;Edgerton et al., 2009). The reconstructed mass is in good agreement with the gravimetric mass for PM2.5, with slope and R 2 values of 0.90 and 0.99, respectively (Fig. S2h). The reconstructed PMcoarse, however, is notably lower than the gravimetric mass, showing a slope value of 0.74 and R 2 of 0.96 (Fig. S2i). The underestimation is plausibly due to omission of constituents in geological material (e.g., carbonate) and/or water bound on PM.

PM10
PM2.5 PMcoarse Figure S2. Evaluation of speciation data quality for PM10, PM2.5, and PMcoarse shown by comparison between (a-c) PM mass concentration obtained by continuous monitoring vs. gravimetric measurement, (d-f) sum of chemical species vs. gravimetric mass, and (g-i) reconstructed mass vs. gravimetric mass.

S2 Season division
Hong Kong is situated in the sub-tropical region along the southeast coast of China. The seasonal evolution of weather in Hong Kong is closely related to the East Asian Monsoon system. Therefore, the direction of upper-level wind is a reliable indicator for seasonal change around Hong Kong. Figure S3 shows the wind direction at ~20 km above ground level from December 2019 to February 2021, measured over the Hong Kong Observatory automatic weather station. It shows that the wind direction is mainly westerly in winter (December to February) and becomes easterly in summer (June to September). The transition seasonsspring and fallare marked by a group of variable wind directions with relatively shorter duration compared to summer and winter. Four seasons can, therefore, be identified approximately.
The exact dates of the seasonal divisions are located by identifying the arrival date of the first synoptic event that is typical in the respective season; for example, cold surge in winter and arrival of warm and humid air mass in spring. These synoptic events are identified by observing the sea level pressure and dew point changes. For example, the sudden and rapid drop in sea level pressure and rise in dew point during 5-9 March 2020 indicates the arrival of warm air mass, marking the end of winter and beginning of spring. The same indicator is also used to locate the transition from spring to summer, while the opposite is used to identify the transition from summer to fall and fall to winter. The seasonal division in this study is summarized as follow: Second Winter: 29 November 2020-28 February 2021 Figure S3. Temporal variation in wind direction at ~20 km height in Hong Kong during the study period.

S3 Source number determination in PMF
At first the optimal number of source factors was deduced by examining the Q/Qexpected value for a range of PMF solutions with different factor numbers. The Q/Qexpected value is indicative of the overall fitting of all input species and is inversely related to the fitting (Norris et al., 2014). Mathematically, the optimal factor number is the number upon which further increasing the factor number would result in much less significant improvement in the fitting, or equivalently much less reduction in Q/Qexpected value. Figure S4 presents the Q/Qexpected values and their changes as a function of factor number from two to eight. From the figure the three-factor solution appears to be the optimal solution since further increasing the factor number to four led to much less reduction in Q/Qexpected value. However, we found that an additional factor is required to better reproduce the Cu concentration. Specifically, the slope and R 2 values for modeled vs. measured Cu improve from 0.54 to 0.72 and from 0.39 to 0.56, respectively, in the four-factor solution. The fourth factor is a dust-like factor enriched in Cu (Fig. 2 in main text).
Considering Cu is an important species in PM health effects associated with reactive oxygen species formation, this factor is retained for source interpretation (Bates et al., 2019).
The five-factor solution was also assessed. However, the fifth factor, which is a secondary nitrate factor, was assessed to be chemically inexplainable after examining the charge balance of the ionic composition. Specifically, the factor composition is significantly depleted in cation to counterbalance nitrate (cation-to-anion equivalence ratio = 0.3). By contrast, the nitrate mainly exists in a sea salt factor in the four-factor solution, with a much more reasonable cation-to-anion equivalence ratio (1.2). This is also consistent with a previous study showing coarse nitrate in Hong Kong is mainly associated with sea salt (Bian et al., 2014). Taken all the analyses together, the four-factor solution is considered the optimal solution for source interpretation and analysis.