Effects of emission reductions on organic aerosol in the southeastern United States

Long-term (1999 to 2013) data from the Southeastern Aerosol Research and Characterization (SEARCH) network are used to show that anthropogenic emission reductions led to important decreases in fine-particle organic aerosol (OA) concentrations in the southeastern US On average, 45 % (range 25 to 63 %) of the 1999 to 2013 mean organic carbon (OC) concentrations are attributed to combustion processes, including fossil fuel use and biomass burning, through associations of measured OC with combustion products such as elemental carbon (EC), carbon monoxide (CO), and nitrogen oxides (NOx). The 2013 mean combustionderived OC concentrations were 0.5 to 1.4 μg m at the five sites operating in that year. Mean annual combustionderived OC concentrations declined from 3.8± 0.2 μg m (68 % of total OC) to 1.4± 0.1 μg m (60 % of total OC) between 1999 and 2013 at the urban Atlanta, Georgia, site (JST) and from 2.9± 0.4 μg m (39 % of total OC) to 0.7± 0.1 μg m (30 % of total OC) between 2001 and 2013 at the urban Birmingham, Alabama (BHM), site. The urban OC declines coincide with reductions of motor vehicle emissions between 2006 and 2010, which may have decreased mean OC concentrations at the urban SEARCH sites by > 2 μg m. BHM additionally exhibits a decline in OC associated with SO2 from 0.4± 0.04 μg m −3 in 2001 to 0.2± 0.03 μg m in 2013, interpreted as the result of reduced emissions from industrial sources within the city. Analyses using non-soil potassium as a biomass burning tracer indicate that biomass burning OC occurs throughout the year at all sites. All eight SEARCH sites show an association of OC with sulfate (SO4) ranging from 0.3 to 1.0 μg m −3 on average, representing ∼ 25 % of the 1999 to 2013 mean OC concentrations. Because the mass of OC identified with SO4 averages 20 to 30 % of the SO4 concentrations, the mean SO4-associated OC declined by ∼ 0.5 to 1 μg m −3 as SO4 concentrations decreased throughout the SEARCH region. The 2013 mean SO4 concentrations of 1.7 to 2.0 μg m −3 imply that future decreases in mean SO4-associated OC concentrations would not exceed ∼ 0.3 to 0.5 μg m. Seasonal OC concentrations, largely identified with ozone (O3), vary from 0.3 to 1.4 μg m (∼ 20 % of the total OC concentrations).

the introduction and clarified the interpretation of the PCA nomenclature relative to 143 conventional use of OC and EC source terminology (please see proposed revision for 144 new material). Our seasonal (ozone) and sulfate PCA factors correspond within error to 145 previous work that has linked biogenic and anthropogenic emissions. We have prepared 146 a comparison to the manuscripts of Xu et al. (2015a;, which appear consistent with our 147 results. We link our sulfate factor to the isoprene SOA identified by Xu et al. (2015a;, 148 and we link our seasonal factor to the LO-OOA factor. The comparisons are provided in 149 the proposed revised text. 150 Our combustion OC factor is identified through its correlation with CO, EC, and NOx. 151 This factor doesn't correlate with sulfate. (We aren't sure why the referee listed sulfate 152 and SO2 as part of the combustion component, as our results do not show either SO4 or 153 SO2 as components of the combustion factor. We hope that our revision of the section on 154 PCA will be clear about these distinctions). As noted, there is a separate sulfate OC factor 155 that is consistent with the AMS isoprene OA factor. Our combustion OC factor is not 156 linked with either isoprene or nitrate SOA, nor is it equivalent to HOA identified by AMS 157 or to POA. It is an emission source-related factor, not a composition-based factor. We 158 conclude that the combustion factor includes both fresh emissions and more aged and 159 oxidized OA, all deriving from sources that co-emit CO, EC, NOx, and, possibly, VOCs. 160 The factor represents an observable association among combustion emissions 161 notwithstanding the evolution of those emissions via atmospheric processes. We hope 162 that the new comparisons and text will clarify this result. 163 Comment "B" (link calculated OM and biomass burning estimates to other SOAS study 164 results). As indicated in the revised text, we provide an additional comparison of our 165 calculated OM to the OA and OM/OC published by Xu et al (2015a;. We also compare 166 our biomass burning OC (OCbb) to AMS BBOA estimates and estimate uncertainties for 167 our OCbb. 168 Comment "C" (context and perspective). We had previously placed most of this material 169 in appendix. We have revised the introduction to summarize historical advances in 170 understanding carbonaceous aerosol and have added to the supplement a table In much of North America, organic aerosol (OA) represents approximately half of 284 average PM2.5 mass concentrations in ambient air (Kanakidou et al., 2005). OA derives 285 from primary source emissions and secondary atmospheric processes involving reactions 286 of volatile organic compounds (VOCs) of anthropogenic and natural origins (see 287 appendix). The latter is widely recognized in the southeastern U.S. with its potential 288 source of VOCs from dense vegetation (Hand et al., 2012). Initial speculation about 289 secondary organic aerosol (SOA) in the Southeast from natural terpenoid compounds 290 dates back to 1991 (e.g., Pandis et al., 1991). With re-evaluation of particle yields from 291 isoprene acidic-photochemical oxidation in smog chambers, interest in natural SOA 292 focused on this species (e.g., Kroll et al., 2006). The early 2000s investigations involving 293 isoprene and terpenoids identified chemical mechanisms hypothetically applicable in the 294 ambient atmosphere as well as tracers for reaction products (e.g. Hallquist et al., 2009). 295

Effects of Emission Reductions on Organic Aerosol in
These hypotheses included accounting for the effect of acidity and photochemical 296 linkages with the gas and condensed phases; a part of this chemistry involves the 297 interactions with inorganic acids in the atmosphere-sulfur and nitrogen oxides, SO2 and 298 NOx. 299 In parallel with advances in organic aerosol chemistry, workers explored different 300 indirect means of estimating SOA from VOC sources. In Atlanta, Lim and Turpin (2007) 301 used the carbon tracer method to calculate summertime SOA concentrations from 302 collected filter samples. In the Southeast, Zheng et al. (2002;2006) used chemical tracers 303 extracted from filters to identify primary OA, noting that an incomplete mass balance 304 could be SOA. Kleindienst et al. (2007; and Lewandowski et al. (2013) used 305 chemical tracers to estimate SOA from isoprene and terpenes. The carbon tracer method 306 was expanded for natural species using carbon isotopes (e.g., Lewis et al., 2004;Tanner 307 et al., 2004;Zheng et al., 2006;Ding et al., 2008). These empirical approaches were 308 explored further by Blanchard et al. (2008). Identification of water soluble carbon as an 309 SOA indicator also has been used (e.g. Weber et al., 2007). More recently field studies 310 have adopted measurements from aerosol mass spectrometry combined with gas 311 chromatograph and mass spectroscopy to track indicator species for SOA components, 312 including species associated with isoprene-sulfur oxide or nitrogen oxide photochemistry 313 (e.g., Gao Hansen et al. (2003) and Edgerton et al. (2005;2006). Special 366 data from ancillary experiments in the SEARCH network supplement the long-term data. 367 We also use emission data derived from the EPA National Emission Inventory (NEI),

Emission sources and the relation of ambient to emission trends 385
This section incorporates previously published analyses by reference, extends them 386 through 2013, and integrates findings. Results related to emission changes are compared 387 with those obtained using other approaches in Section 3.6 (Synthesis). 388 Southeastern emissions in 2013 are shown by source category in Table S2new  S3new). These differences occurred even when comparing model predictions to the 420 fraction of measured OC that was not associated with O3 and SO4 (inventory OC 421 emissions do not represent SOA deriving from biogenic emissions of isoprene and other 422 gases). Ambient OC trends were more pronounced than trends predicted by the model 423 from the inventory ( Figure S3new). However, the receptor model reproduces observed 424 OC trends more readily for sites where the mobile-source contribution is greatest ( Figure  425 S3new

Ambient EC and OC concentrations and trends 457
Trends and spatial variations are evident for mean annual and seasonal EC and OC 458 concentrations (Table 1 and Figure 1). Mean EC concentrations were 2.0 to 3.5 times 459 greater at JST than at CTR, thereby indicating two-to three-fold greater influence of 460 combustion sources within Atlanta compared to rural CTR because EC is a tracer of 461 combustion (appendix). Mean OC concentrations were 1.0 to 1.8 times greater at JST 462 than at CTR, indicating urban sources of OC possibly superimposed on a relatively high 463 regional baseline. The ratio of JST EC to CTR EC declined from 2.8:1 to 2.1:1 between 464 the first and third five-year periods, while the JST OC to CTR OC ratio decreased from 465 1.5:1 to 1.2:1 between the first and third five-year periods. Since the ratio of JST 466 EC/CTR EC declined by 25% and the ratio of JST OC/CTR OC declined by 20%, the 467 declines are comparable but the difference is consistent with a greater mobile-source No season consistently exhibits the highest mean EC and OC concentrations but the CTR 476 mean OC concentrations and OC/EC ratios are highest during summer, interpreted as the 477 influence of aging and SOA formation during warmer months. In contrast, JST mean OC 478 and EC concentrations tend to be higher during autumn and winter (Table 1) (Table 1). 486 Mean OC/EC ratios are higher at CTR than at JST, again consistent with regional-scale 487 aging of ambient aerosol and a relatively greater influence of SOA at CTR. and potentially much greater than unity as oxidation and SOA formation proceed 494 (Robinson et al., 2007). 495 Temporal trends in ambient EC and OC correlated within individual sites and across the 496 SEARCH domain (e.g., CTR and JST, Figure 1), indicating regional coherence of trends 497 and seasonal variations for both EC and OC. The strong correlation of EC and OC at all 498 SEARCH sites, averaging times (annual, seasonal, monthly, daily), and seasons indicates 499 that combustion processes are a major source of OC (Table S1; Figure S2). However, significant correlations of SO4 with both EC and OC during summer suggest the 501 influence of SO4 on SOA formation in summer, consistent with results from SOAS (Xu 502 et al., 2015a,b;Budisulistiorini et al., 2015). OC correlates with both EC and SO4, but for 503 different reasons. Consequently, EC and SO4 also correlate, but not as strongly and not as 504 consistently across time scales. In summary, the EC and OC measurements indicate 505 influence of multiple emission sources or atmospheric processes affecting all SEARCH 506 sites, though differently at urban and rural locations. 507 We estimate the OM/OC ratio for the urban and rural SEARCH sites using a mass 518 balance computation based on particle composition. The sum of species concentrations, 519 including estimated particle-bound water (PBW) at laboratory temperature and relative 520 humidity (RH), is: 521 (1) 522 (inorganic species concentrations are from ion measurements). PBW at laboratory RH of 523 < 38% is represented by the coefficients f1 (1.28), f2 (1.15), and f3 (1.25) (Tombach, 2004, 524 as derived from Tang et al., 1996). The coefficient f1 is an average of the coefficients for 525 NH4HSO4 (1.27) and (NH4)2SO4 (1.29), f2 is the coefficient for NH4NO3, and f3 is a 526 weighted average reflecting higher SO4 than NO3 concentrations. MMO is the sum of the 527 concentrations of six crustal elements (Al, Ca, Fe, Mg, Si, Ti) (XRF measurements), expressed as oxides (Hansen et al., 2003). This estimate of crustal mass is likely 529 conservative, since it does not include Mn and the assumed Ca mass (as CaO) would be 530 less than the mass of CaCO3 (if present). The carbon components, metals, and chloride 531 are not adjusted for retained water at laboratory temperature and humidity. This creates a 532 potential for uncertainty in the calculation, especially in the case of OC. Atmospheric OC 533 is known to be hygroscopic at elevated humidity, but experimental data suggest that 534 water retention is minimal at < 38% RH for laboratory filter analysis (e.g., Malm et al., 535 2005; Taylor et al., 2011). Measurements made during SOAS indicate that organic-536 associated water was less than ~25% of total particle water in mid-day ambient samples 537 when ambient RH was less than ~50% (Guo et al., 2015). We estimate an OC PBW 538 uncertainty in Eq. (1) by assuming that OC PBW is 10% of OC (fOC = 1.1), which would 539 increase the calculated sum of species by 3% and decrease the OM/OC (calculated 540 below) by 0.1 units on average. 541 The difference between PM2.5 mass and the sum of species concentrations is denoted as 542 "non-measured" (NM) mass: 543 NM mass = PM2.5 mass -Sum of species (2) 544 An upper bound for OM is calculated as OM* = OC + NM mass, which assumes that all 545 NM mass is associated with OA. Any mass that is missing from the computed sum of 546 species would bias NM mass high, thereby also causing OM* to be higher than the true 547 OM. Similarly, underestimation of PBW would bias NM mass and OM* high. We 548 estimate the combined effect of missing species and PBW to result in possible 549 overestimation of OM*/OC by up to 0.2 units on average. An opposing potential bias 550 arises in Equation 2, because the FRM sampler that is used by SEARCH to provide the 551 PM2.5 mass measurement is known to lose volatile species (e.g., inorganic particle NO3). 552  The ratio of TC to Kbb (TCbb/Kbb) in biomass burning is known to vary widely among 642 fire types (e.g., wildfires differ from prescribed burns) and among fire stages (e.g., 643 temperature, or flaming vs. smoldering). The variability of emissions among and within 644 fires implies that biomass-burning tracers are more useful for estimating average impacts 645 than for quantifying burn contributions during individual events. We use a single average scaling factor based on consideration of emissions information , which 647 we check using the correlation of modern C with non-soil K ( Figure S9). Inventory 648 annual average TCbb/Kbb is in the range 28:1 -36:1 . Our 649 assumed scaling factor of 32 for TCbb/Kbb is similar to carbon isotope data from CTR 650 winter samples ( Figure S9, CTR regression slope TCmodern/Kbb = 43), when 651 prescribed burns are more common and SOA formation rates are lower. Our scaling 652 factor of 32:1 introduces a possible bias toward underestimation of OCbb of ~25% 653 relative to the CTR winter ratio of TCmodern/Kbb = 43. The higher slope of 71:1 at 654 JST could reflect a different type of biomass burning (e.g., residential wood combustion), 655 while the lesser correlation of modern TC with non-soil K (assumed to represent Kbb) at 656 BHM and higher slope at PNS potentially reflect the confounding influence of industrial 657 (BHM) or marine (PNS) sources of non-soil K. A higher scaling ratio (e.g., TCbb/Kbb 658 = 70 rather than 32) would yield higher computed TCbb and therefore higher OCbb. 659 Based on OC/EC in actual biomass-burning events observed at SEARCH sites ( Figure  660 S5), we compute OCbb = 0.9*TCbb (the ratio OCbb/TCbb could be higher in some burn 661 events). Considering the range among SEARCH sites of winter TCmodern/Kbb (22:1 662 to 82:1, Figure S9), we estimate the uncertainty in scaling from Kbb to OCbb as -30% to 663 +250%. As noted, Kbb could be overestimated if there are other sources that contribute to 664 nsK. We therefore estimate the uncertainty range for OCbb as -50% to +200% subject to 665 the constraint that OCbb < OC. 666 Due to the loss of organic tracers on a time scale of about a day or less, the biomass 698 burning OA that is estimated using the AMS with levoglucosan tracers is thought to yield 699 an estimate of relatively fresh burning as compared to aged regional burning levels (Xu et  physical and chemical processes (e.g., deposition; chemical loss processes; contrasts 726 between inland versus marine air mass transport) as well as species origins. 727

Application 728
We report two main versions of PCA, with additional versions used for sensitivity tests 729 and auxiliary information. PCA1 is applied to measurements made at SEARCH sites 730 from 2008 through 2013. The 23 gas and PM2.5 measurements comprise daily-average 731 concentrations of PM2.5 EC and OC (thermal-optical reflectance, TOR), daily averages of 732 gases NH3 (measured continuously or at 24-hour resolution) and continuous NOx and 733 NOz, secondary species (daily peak 8-hour O3, plus PM2.5 SO4, NH4, and NO3), and PM2.5 crustal elements (XRF measurements of Al, Si, and Fe), species associated with salts 735 (PM2.5 Na, Cl, Mg, and Ca ions), and trace metals (PM2.5 Zn, Cu). Both daily averages 736 and daily 1-hour-maxima of gases (CO and SO2) are included to match the temporal 737 resolution of the other daily data while also potentially capturing shorter-duration plumes. 738 Water-soluble PM2.5 K (K ion) is included as a potential indicator of biomass 739 combustion. Because some species used in PCA1 were not measured throughout the 15-740 year SEARCH program, PCA2 is carried out to interpret long-term OC trends from 1999 741 through 2013. PCA2 excludes measurements that commenced in 2008 (water-soluble Ca, 742 Mg, K, Na, and Cl). XRF Ca and nsK are used instead of water-soluble Ca and K, 743 respectively. Without Na and Cl in PCA2, salt is not detectable, as will be discussed. NH3 744 is excluded from PCA2, since those measurements began in 2004. Daily-average O3 is 745 included in PCA2 to complement daily peak 8-hour O3. 746 The sensitivity of our results to the choice of statistical method is examined by comparing 747 PCA1 and PCA2 and by using additional PCA and PMF applications. As described in 748  (Table 2). For clarity, we designate the components as: 771 (1) combustion, (2) crustal, (3) seasonal, (4) SO2, (5) SO4, (6) metals, (7) salt, and (8)  772 other. These names are used as descriptors, rather than as designated emission sources. The OC apportionments indicate statistically-significant relationships between OC and 783 four to seven PCA components (Tables S12 and S13). Mean contributions of each 784 statistically-significant component to daily OC at each site using both PCA1 and PCA2 785 are summarized in Table 3; these contributions are expressed as percentages of total OC 786 in Table S14. PCA1 and PCA2 each indicate that OC is associated with multiple 787 components at all sites. Except at YRK and OLF (PCA2 only), the overall OC 788 associations are strongest for the combustion component (Tables S2 to S9). 789 The PMF source profiles varied depending on the choice of uncertainty inputs, but 790 yielded average OC apportionments that were qualitatively comparable to PCA2 (Figure  791 S26). The PMF crustal OC and SO4-associated OC concentrations were comparable to 792 PCA ( Figure S27). However, PMF source profiles combined CO and O3, whereas PCA tended to separate O3 from CO, leading to differences in the apportionment of OC to 794 combustion and seasonal components ( Figure S27). Differences between PCA and PMF 795 occur in part because the PCA seasonal component generally comprised contrasts (e.g., 796 positive O3, negative inorganic particulate NO3) whereas PMF forced positive solutions. 797 In these applications, PCA predicted high OC concentrations more accurately than PMF 798 did ( Figure S28). 799 Combustion. All sites exhibit a suite of species associated with combustion processes 800 (EC, OC, CO, Kbb or K ion, NOx or NOz). The variations in combustion associations 801 among sites suggest different source mixes, differences in air mass ages (e.g., fresh 802 emissions at urban sites, more aged emissions at rural sites), or differing transport of 803 polluted air masses. For example, NOz is more strongly associated than NOx with the 804 combustion component at the two most rural sites, CTR and OAK. OC associated with 805 the combustion factor could therefore comprise material that would be classified as either 806 POA or SOA by other analytical approaches (e.g., HOA or MO-OOA by AMS). 807 Mean combustion OC ranges from 0.7 to 1.6 g m -3 for PCA1 (2008 -2013) and from 808 1.5 to 2.6 g m -3 for PCA2 (1999 -2013), except at YRK (Table 3). Daily PCA1 and 809 PCA2 combustion OC concentrations are correlated at all sites ( Figure S12). Mean 810 absolute differences between PCA1 and PCA2 computed combustion OC range from 0.1 811 to 0.7 g m -3 (not tabled). However, the mean PCA1 and PCA2 combustion OC 812 apportionments are averaged over different time periods, so the differences in their 813 averages are partly due to declining EC, CO, and NOx concentrations (Table 1). Trends in 814 OC components are discussed in Section 3.4.4. Mean PCA2 combustion OC ranged from 815 25 to 63% of mean OC concentrations (Table 3). The mean crustal-associated OC concentrations vary from 0.1 to 0.3 g m -3 at inland sites 830 (Table 3). Coastal sites exhibit non-significant, minor, or inverse associations of OC with 831 crustal elements (-0.1 to 0.1 g m -3 , Table 3). Inverse associations indicate that OC 832 concentrations at coastal sites are lower than average when Al, Si, and Fe concentrations 833 are elevated. PCA1 and PCA2 crustal OC concentrations correlate ( Figure S14) and 834 crustal OC correlates with Si ( Figure S15). Crustal-associated OC could derive from 835 region-wide phenomena (e.g., transport of Saharan dust), but may also stem from 836 ubiquitous and widely distributed activities that suspend crustal material. Potential 837 sources include soil-derived OC (e.g., agricultural activities, construction, or road dust), 838 or biomass burning that lofts crustal material (e.g., through plowing material into debris 839 piles). Road dust is known to include OC among its constituents (e.g.   Coastal sites show an inverse association of OC with Na and Cl (sea salt) (Table S13) and 928 a negative mean OC contribution from salt (Table 3) (Table 2). 947

Intercomparisons and uncertainty 948
For PCA3 (Table S10) (Table S11). 962 The ranges of mean OC concentrations associated with each PCA component as obtained 963 from the various applications are listed for CTR, JST, and YRK in Table 4

Temporal variations 995
Temporal variations of the 1999-to-2013 PCA2 results are described here primarily for 996 CTR and JST, representing (as in Table 1) one rural and one urban location having 997 extensive SEARCH data records. At JST, day-of-week variations are evident for the 998 combustion-derived OC and for the OC associated with crustal species (Figure S20 used as an input in the multivariate regressions that generate "POC" and "SOC" 1046 (Blanchard et al., 2008, not discussed here), and "POC" is a fitting species used in the 1047 CMB receptor modeling. As shown in Table 5, the apportionments exhibit areas of 1048 agreement as well as certain differences. Both are summarized using ratios of the values 1049 listed in Table 5. We report averages and ranges across the sites.