Air Quality and Atmospheric Deposition in the Eastern US: 20 Years of Change

Abstract. Data collected in the eastern US between 1990 and 2009 at 34 paired dry and wet monitoring sites are examined. A goal is to evaluate the air quality impacts occurring between 1990 and 2009 that are associated with concurrent legislatively mandated changes in emissions. Four 5-year periods, 1990–1994 (P1), 1995–1999 (P2), 2000–2004 (P3), and 2005–2009 (P4) are considered, with a primary focus on P1-to-P4 changes. Results suggest that legislatively mandated air pollution mitigation strategies have been successful in improving air quality and reducing atmospheric deposition in the eastern US. Respective P1-to-P4 reductions of estimated sulfur dioxide (SO2) and nitrogen oxides (NOx) emissions in the eastern US are 50 and 42%. Corresponding behavior of the following metrics associated with these emissions reductions is examined: monitored atmospheric concentrations of SO2, aerosol sulfate (SO4), and oxidized sulfur (S); dry, wet, and total deposition of S; monitored atmospheric concentrations of nitric acid (HNO3), aerosol nitrate (NO3), and their sum, oxidized nitrogen (OxN); dry, wet, and total deposition of OxN; monitored atmospheric concentration of aerosol ammonium (NH4); dry, wet, and total deposition of NH4; summed monitored atmospheric concentration of oxidized and reduced nitrogen (N); dry, wet, and total deposition of N; wet deposition of hydrogen ion (H+); monitored atmospheric concentration of ozone (O3); dry deposition of O3; and the summed monitored atmospheric concentration of aerosol NO3, SO4, and NH4 (Clean Air Status and Trends Network particulate matter – CASTNET PM). Other metrics (e.g., ratios of dry to total deposition) are also considered. Selected period-to-period changes of air quality and deposition metrics at site, regional, and seasonal scales are discussed. As an example, despite P1-to-P3 reductions in estimated emissions of both SO2 and NOx, aerosol NO3 concentration increased in the east, with widespread wintertime numerical increases in both aerosol NO3 concentration and CASTNET PM. However, a reversal of this behavior is associated with continuing P3-to-P4 reductions of SO2 and NOx emissions. Thus, additional P3-to-P4 reductions of these emissions, especially NOx, appear to have made progress in altering the chemical regime of the wintertime eastern US atmosphere so that future emissions reductions and their resulting reductions in aerosol concentrations may no longer be accompanied by sub-linear changes (or actual increases) in CASTNET PM.


Introduction
In the United States (US), pollution control legislation has been established to protect the public health and welfare by reducing emissions of selected pollutants.Both sulfur dioxide (SO 2 ) and nitrogen dioxide (NO 2 ) are designated by the US Environmental Protection Agency (EPA) as criteria pollutants, and they contribute to acidic deposition as well as airborne particulate matter (PM).In addition, nitrogen ox-  (Xing et al., 2013).ides (NO x ), the sum of nitric oxide (NO) and NO 2 , act as precursors of another criteria pollutant, ozone (O 3 ).
Nationwide, in 1990, 69 % of SO 2 emissions came from electric generating units (EGUs), whereas 82 % of NO x emissions came from the sum of EGUs (25 %) and mobile sources (57 %) (Xing et al., 2013).Sources in the eastern states contributed 76 and 58 % of the US total 1990 SO 2 and NO x emissions, respectively.The Clean Air Act Amendments of 1990 (CAAA) established phased year-round controls for SO 2 and NO x emissions from EGUs that became effective in 1995 for SO 2 and 1996 for NO x (Phase I), while additional controls became effective under Phase II in 2000.Over half of the EGUs targeted by the CAAA are located in six states along the Ohio River (source region -SR; Table 1).Beginning in 1999, O 3 season (i.e., summertime) NO x controls focusing on EGUs became effective in selected eastern states under the Ozone Transport Commission (OTC) and were superceded in 2003 by the NO x State Implementation Plan (SIP) Call.The affected states chose to meet their mandatory reductions by participating in the NO x Budget Trading Program (BTP), a market-based cap and trade program for EGUs and large industrial units.The BTP was replaced in 2009 by the Clean Air Interstate Rule (CAIR).Various mobile source NO x emissions control programs also began in the mid-to late 1990s.Fully implemented in 2006, the Tier 2 Gasoline Sulfur Program targeted both SO 2 and NO x emissions from mobile sources.These mobile source programs have a cumulative effect of reducing emissions over time as the mobile fleet is replaced.
Figure 1 shows estimates of emissions from 1990 to 2010 for both the total US and the eastern states (Xing et al., 2013).Both SO 2 and NO x emissions show weak reductions between 1990 and 1994.There is a sharp reduction in SO 2 emissions in 1995, when Phase I controls became effective.Although additional reductions of both SO 2 and NO x emissions begin to appear in 1999, they are apparent in later years (e.g., after 2005 for SO 2 ).Although a substantial portion of the long-term emissions reductions since 1990 has resulted  1).from legislatively driven emissions controls, economic factors (e.g., coal-to-natural gas fuel switching and recessioninduced reduction in demand for electricity and transportation fuel) played a sizeable role between 2008 and 2009 (Lu et al., 2012).
A major objective of the current study is to examine the monitored and estimated air quality and deposition behaviors in the eastern US (E) between 1990 and 2009 associated with changes in emissions.The time period of interest is divided into four 5-year periods: 1990-1994 (period one, P1), 1995-1999 (period two, P2), 2000-2004 (period three, P3), and 2005-2009 (period four, P4).Focusing on these 5-year periods, average P1 US SO 2 and NO x emissions have declined by 19 and 6 % by P2, by 33 and 19 % by P3, and by 49 and 37 % by P4 (Xing et al., 2013).In the current study, the east is taken to represent all states east of the Mississippi River and is divided into three geographical regions: midwest (MW), northeast (NE), and south (SO) (Fig. 2).Mean period-specific emissions in these regions are illustrated for each 5-year period relative to P1 in Fig. 3.
This paper builds on the previous work (Sickles and Shadwick, 2007a and b; hereafter known as SS07a and SS07b).In SS07a seasonal and regional distributions of concentration and deposition of selected pollutants were examined for the 1990s (i.e., P1 and P2) at eastern rural US Clean Air Status and Trends Network (CASTNET) and paired National Atmospheric Deposition Program (NADP) monitoring sites.In SS07b temporal changes in air quality and deposition in the eastern states between the early 1990s and early 2000s (i.e., P1-to-P3) were described and discussed.The record is expanded in the current paper to include a fourth 5-year period.Here, emphasis is on temporal changes occurring between the early 1990s and late 2000s (i.e., P1-to-P4) as new and continuing pollution controls have become mature and their impacts apparent.Results are displayed in a consistent manner to previous work (SS07a and b) to facilitate comparison.See Table 2 for definitions of terms and pollutant metrics used in the text, tables, and figures.

Approach
Monitoring data collected in the east between 1990 and 2009 at 34 CASTNET http://epa.gov/castnet/javaweb/index.htmldry monitoring sites and paired NADP http://nadp.sws.uiuc.edu/ wet monitoring sites provide a sufficiently complete 20-year record to permit examination.Weekly concentrations of gaseous SO 2 and nitric acid (HNO 3 ), and aerosol sulfate (SO 4 ), nitrate (NO 3 ), and ammonium (NH 4 ) (using 3-stage filter packs) and hourly concentrations of O 3 (using continuous monitors) are determined at 10 m.Dry deposition of airborne species is estimated using the infer- ential model by multiplying modeled deposition velocities (Vd) by the monitored atmospheric concentration for the gas or aerosol species of interest.Deposition velocities for gases and aerosols are estimated using the multilayer model (MLM) (Meyers et al., 1998;Wesely et al., 1985).Monitoring data are composed from reported weekly (hourly for O 3 ) measurements into estimates of means for year, site, and season.These year-site-season means, computed for atmospheric concentration, Vd, precipitation rate (PR), and dry, wet, and total deposition, are averaged to period-site-season means and used to examine differences between 5-year periods for seasons, sites, and predefined groupings of sites.The data, methodologies, data quality (i.e., precision and accuracy), covariance of concentration and Vd, impacts of nonideal landscape features, adjustments for sampling artifacts, and data completeness criteria for accepting mean estimates of measurement data are described in SS07a.
Monitoring site locations, 10 in the MW and NE, and 14 in the SO, are identified in Fig. 2 and Table 1.These 34 sites provide sparse coverage, especially in the southern part of the SO and the northeastern part of the NE.Three additional environmental groupings of monitoring sites are also defined as subsets of the eastern sites: background (BKG), high elevation (HE), and SR.These sites are chosen to reveal distinctions in pollutant behavior characteristic of their environments (Table 1 and SS07a).
CASTNET monitoring locations were initially sited to avoid the influence of major cities, highways, local activities, and point and area sources of pollutants and to be regionally representative (i.e., similar to a characteristic 80 by 80 km area surrounding the site) (Clarke et al., 1997;Baumgardner et al., 2002).In the current study, most of the CASTNET sites are located within 50 km of a paired NADP-NTN (National Trends Network) wet deposition site.
Collocated precision estimates (as CVs) of weekly CAST-NET concentration, Vd, and dry deposition at eastern sites are within 5 to 10 % for the major species (Clarke et al., 1997;Sickles and Shadwick, 2002).Corresponding precision estimates are 5 % for weekly NADP PR, wet S (i.e., wet deposition of sulfate; see Table 2), and wet OxN; 10 % for wet H + ; and 15 % for wet NH 4 (Nilles et al., 1994).These estimates for weekly network results are likely to be conservative for corresponding longer-duration 5-year period results.
Total oxidized nitrogen (NO y ) includes the primary NO x species (NO and NO 2 ) plus the secondary species (HNO 3 , nitrous acid (HNO 2 ), aerosol NO 3 , peroxyacetylnitrate (PAN), as well as other oxidized organic nitrogen species).Total reduced nitrogen (NH x ) includes both gaseous ammonia (NH 3 ) and aerosol NH 4 .In the current study, only two of the NO y species, HNO 3 and aerosol NO 3 , (their sum is OxN), and only one of the NH x species, aerosol NH 4 , are monitored.As a result, reported estimates of dry deposition of OxN and NH 4 underestimate corresponding values for NO y and NO x .
The contribution of coarse (i.e., > 2.5 µm) particles to total particulate deposition likely depends on location and proximity to sources of crustal dusts, road salt, and marine aerosols.Since the contribution of coarse particles is ignored in the MLM, where all particles are modeled as fine aerosol SO 4 particles, current estimates of particulate dry deposition are biased low.However, the non-size selective CASTNET filter pack sampler likely over-samples the fine tail of coarse particle distribution.Although this may not be too important for aerosol SO 4 and NH 4 , which occur mainly as fine particles, it would underestimate the contribution of dry deposition of aerosol NO 3 , where appreciable coarse aerosol NO 3 can occur (Wolff, 1984;Lefer and Talbot, 2001).
As noted in SS07a (and references therein), use of the CASTNET filter pack sampler introduces sampling artifacts, especially to nitrogen-containing species.Findings from Sickles and Shadwick (2008), as well as wind tunnel and modeling studies (unpublished) of particle sampling characteristics of the non-size selective open-faced CASTNET filter pack inside its weather shield over a range of wind speeds, suggest appreciable collection of both fine (< 2.5 µm) and coarse particles on the first filter (Teflon) of the filter pack.Although both aerosol SO 4 and NH 4 exist mainly in the fine fraction (< 1 µm) year round, aerosol NO 3 is multi-modal, and in non-marine environments exists mainly (>75 %) in the fine fraction in cool seasons and mainly as larger (>4 µm) particles during warm seasons (Wolff, 1984;Lefer and Talbot, 2001;Zhang et al., 2008;Lee et al., 2008).Sampling artifacts in opposing directions may arise through possible loss of aerosol NH 4 NO 3 during sampling and by aspiration of the fine tail of coarse aerosol NO 3 .Aerosol NH 4 NO 3 (nominally < 2.5 µm) collected on the Teflon filter may be lost during sampling by volatilization into gaseous HNO 3 and NH 3 , driven by departures from equilibrium (exaggerated at higher temperatures) and/or by reaction with strong acids.Results from comparison studies (Sickles andShadwick, 2002, 2008;Lee et al., 2008;Zhang et al., 2009) are consistent with loss of NH 4 NO 3 from the Teflon filter followed by collection of volatilized HNO 3 on the downstream nylon filter of the CASTNET filter pack during warm season sampling and with minimal volatilization loss during cool season sampling.Coarse alkaline particles (e.g., sea/road salt or soil/mineral dust) may contain aerosol NO 3 (possibly formed by scavenging HNO 3 while airborne), and to the extent that they are collected, these larger particles may provide a high bias to CASTNET aerosol NO 3 relative to its fine (PM 2.5 ) contribution.Some of these larger particles (if unreacted and retained on the Teflon filter) may scavenge airborne HNO 3 from sampled air during week-long sampling periods, also providing a high bias to aerosol NO 3 and a low bias to HNO 3 .
Some of these concepts are illustrated by the performance of Interagency Monitoring of Protected Visual Environments (IMPROVE) samplers with 2.5 µm size selective inlet and CASTNET samplers collocated at 10 eastern monitoring sites (Sickles and Shadwick, 2008).At the eight non-marine sites good agreement was found for aerosol SO 4 (CASTNET was biased high by 4 to 7 %), and more variability was found for aerosol NO 3 (CASTNET ranged between 25 % higher and 39 % lower).Seasonally, aerosol NO 3 showed median relative biases ranging between 5 and −10 % for all seasons, except summer, where the IMPROVE aerosol NO 3 exceeded the corresponding CASTNET values by > 40 % (this was at very low concentrations where the median bias was 0.07 µg m −3 ).Finally, although CASTNET and IMPROVE aerosol NO 3 displayed mixed comparisons at eight inland sites, CASTNET sites showed consistently higher aerosol NO 3 (40-100 %) at the two marine-influenced sites, which was attributed to over-sampling the fine tail of coarse sea salt aerosols by the CASTNET sampler.
Thus, in the current study, minimal CASTNET sampling bias is expected for aerosol SO 4 and NH 4 , which occur mainly as fine particles.However, HNO 3 and aerosol NO 3 as well as other metrics derived from them are subject to sampling biases that act in opposing directions.Volatilization of collected NH 4 NO 3 during sampling could lead to a high bias of HNO 3 and a low bias of aerosol NO 3 .In contrast, the nonsize selective sampler inlet could lead to an over-sampling of the fine tail of coarse aerosol NO 3 with a high bias of aerosol NO 3 and a potentially low bias of HNO 3 .The impacts of these biases are expected to be minimal during cool season sampling at non-marine sites.As a result, in the current study HNO 3 , aerosol NO 3 , and metrics derived from them (e.g., CASTNET PM, Dry HNO 3 , Dry NO 3 , RNO 3 and other ratios of concentration and deposition) should be viewed with caution, especially during summer.
Uncertainties in Vd, rather than concentration, likely limit the accuracy of dry deposition estimates (Hicks et al., 1991;Lovett and Lindberg, 1993).Accuracy of Vd estimates has not been well established.Based on comparisons with field measurements, the MLM may provide fairly accurate estimates of Vd for SO 2 , O 3 , and HNO 3 to many surfaces (e.g., crops; Meyers et al., 1998); however, Vd for SO 2 to forested canopies may be underestimated by ≈ 35 % (Finkelstein et al., 2000).Recent comparison at the same site of weekly CASTNET MLM Vd estimates with those from a similar model used by the Canadian Air and Precipitation Monitoring Network (CAPMoN) shows CASTNET Vd to be low (relative to CAPMoN) by approximately 50 % for SO 2 and 35 % for HNO 3 ; differences of hourly estimates between the two models could be as large as a factor of 2-3 (Schwede et al., 2011).Annual average small-scale spatial variability (0.5 km) of model-estimated Vd was estimated at 10 % for SO 2 and O 3 , 30 % for aerosol SO 4 , and 40 % for HNO 3 (Brook et al., 1997).Uncertainty of the inferential model has been estimated at 25 % for O 3 , 30 % for SO 2 and ≥40 % for HNO 3 and particles (Clarke et al., 1997).These estimates may be appropriate for areas with ideal terrain and uniform land use; however, non-ideal landscape features (e.g., irregular terrain and/or forested canopies) likely contribute low biases to concentration, estimated Vd, and dry deposition estimates, and increase their resulting uncertainty (Clarke et al., 1997;Hicks, 2006).The distribution of influential landscape features across sites suggests that their impacts result in the greatest underestimates at sites in the south and northeast geographical regions, and that their impacts should be smaller at sites in the midwestern grouping.Much of the above evidence suggests that CASTNET Vd estimates, while uncertain, likely provide conservative lower bound estimates of dry deposition.
Internetwork comparisons of annual wet deposition have been conducted for the periods 1986-1993 (Sirois et al., 2000) and 1995-2004(Wetherbee et al., 2010) at collocated NADP and CAPMoN sites.NADP results for precipitation depth and wet deposition of H + , NH 4 , S, and OxN appear to be biased low relative to CAPMoN by less than 30 %.Although there are no standards for determining accuracy, current NADP results are possibly biased low and likely accurate within 25 %.
The east has large heterogeneities in its spatial distribution of land cover (i.e., crops, forests, and suburban), of terrain (i.e., flat, rolling, complex, and montane) and of pollution sources (e.g., EGU sources in the Ohio Valley and mobile sources in urban centers).As a result, it is unlikely that the sparse coverage of CASTNET (i.e., 34 sites) can provide truly accurate regional representations.This is supported by a recent modeling study of wet deposition (Sickles et al., 2009) where current study site locations were found to over represent geographical areas subject to both high emissions and wet deposition, yielding over-estimations of corresponding regional values by 5 to 25 %.Nevertheless, for the current study, values representing areas with common geographical and environmental characteristics are approximated as the means of values from sites associated with each classifica-tion.It is assumed that these estimates are useful as rough indices of their respective geographical regions and environmental groupings.
As indicated above, the airborne concentration, estimated Vd, and dry and wet deposition estimates may be subject to sizeable uncertainties.However, assuming constant bias, their precision to a first approximation is expected to control the uncertainty associated with the current examination of period-to-period (e.g., P1-to-P4) temporal changes.
Data have been composed from reported measurements into estimates of means for year, site, and season.Yearsite-season means were computed for atmospheric concentration, Vd, PR, and dry, wet, and total deposition.Consistent with previous work (SS07a and b) these values, averaged to period-site-season means, are used to examine differences between 5-year emission periods for seasons, sites, and predefined site groupings.Seasons (e.g., data for winter, W, 1990 include data from December 1989 through February 1990; spring, Sp; summer, Su; and fall, F) are used to define the years in periods.Five-year periods have been chosen to coincide with emission patterns (noted earlier), to minimize the impacts of missing data on seasonal means, to reduce the impact of meteorological variability, and to maintain consistency with earlier studies.For example, Table 3 presents averages of year-site-season means averaged over all the years in P4 and all the sites in each specified site grouping.Analogous tabulations for P1, P2, and P3 may be found in SS07a and b.
The approach to data analysis, described in detail in SS07b, uses mean squared error derived from analysis of variance of mean estimates for each pollutant metric to examine period-to-period differences in the monitoring data and metrics derived from them.Use of the term significant implies comparison where a statistical test is satisfied at the p = 0.05 significance level, and NS implies a non-significant test result (p>0.05).Comparison with no mention of significance, using the term numerically implies nominal comparison where one comparator is numerically different from another, without statistical testing.The statistical tests that were performed apply only to absolute differences.Relative differences (changes) are used as a convenient construct to enable discussion of the absolute differences.

Results
Means of each species-specific value of P4 atmospheric concentration, Vd, PR, dry, wet, and total deposition, and selected ratios are presented in Table 3 for each site grouping.Corresponding across-site variabilities are given as coefficients of variation (CV).Results are displayed for each grouping of monitoring sites over all seasons and for the east by specific season.P1-to-P4 absolute changes in both native units and relative differences (%) are displayed in Fig. 4 for each geographical region.To facilitate identification of pat-terns of P1-to-P4 changes, significant (p ≤ 0.05) decreases are colored blue, and increases are colored pink.Six sets of period-to-period differences are considered (e.g., P1-to-P2, P1-to-P3); however, they are not tabulated due to space limitations.Although not shown comprehensively, P3-to-P4 differences are discussed in a few instances to contrast with P1to-P4 differences.Graphical illustrations of period-to-period comparisons of concentration, deposition, and related metrics as well as relative changes by site and region have been created for each of the entries shown in Table 3. Example illustrations for selected airborne concentration (i.e., S, OxN, N and CASTNET PM) and deposition (i.e., total S, total OxN, total N and wet H + ) are shown for all sites in Figs. 5  and 6.

Deposition velocity and precipitation rate
Deposition velocity and its analog, PR, may be viewed as velocities at which pollutants are delivered (i.e., deposited via dry and wet deposition) to the Earth's surface.Both Vd and PR are strongly influenced by meteorology.One of the reasons for selecting 5-year averaging periods in the current study is to reduce the confounding influences of yearto-year meteorological variability in assessing the impacts of changes in pollutant emissions on resulting ambient atmospheric concentrations, dry deposition, and wet deposition.
The variabilities of Vd and PR are fairly low.In the case of Vd, this is indicated by the 2010 CASTNET decision (http://epa.gov/castnet/javaweb/docs/CASTNET_Factsheet_2013.pdf) to discontinue meteorological monitoring and use site-specific historical average Vd at all but five monitoring sites (Bowker et al., 2011).In the current study, period-site-season-specific CVs, averaged across all periods, sites, and seasons for Vd for aerosol, O 3 , SO 2 , and HNO 3 are each < 5 %, and the corresponding value for PR is < 9 %.For the east and for each of the three geographic regions, entity-specific relative differences were determined for each of the six possible period-to-period comparisons (i.e., 4 × 6 = 24 relative differences per entity), yielding median absolute relative differences ≤ 6 % for Vd of each of the four species and PR.Similar comparisons at the site-specific level (34 × 6 = 204 relative differences per entity) yield median absolute relative differences ≤ 7 % for Vd of each of the four species and PR.
In the east and in each geographic region, seasonal P1-to-P4 changes in mean PR are not significant (p ≤ 0.05; Fig. 4), and they are usually small (≤ 10 %), except in winter.These wintertime changes contribute to a small, but significant, 1 % increase in PR in the east, 9 % increase in the northeast, and 2 % decrease in the south.Reduced significance of wintertime changes in wet deposition (e.g., wet S, wet OxN, wet H + , and wet N; Fig. 4) appears to be associated with these numerical wintertime P1-to-P4 changes in PR.

Oxidized sulfur species
As is illustrated in Fig. 4, significant P1-to-P4 reductions occur in atmospheric SO 2 concentration and dry deposition, atmospheric aerosol SO 4 concentration, atmospheric S concentration and dry deposition, and total S in each region in each season and over all seasons.For atmospheric SO 2 concentration and dry deposition, the numerically largest significant seasonal absolute reductions usually occur in winter when SO 2 concentration is at its seasonal maximum (largest in northeast); the smallest occur in summer when SO 2 concentration is at its seasonal minimum (smallest in south).Both SO 2 concentration and dry deposition show similar P1to-P4 relative reductions of approximately 50 % across all seasons and regions.
As noted earlier, P1-to-P4 changes in atmospheric aerosol SO 4 concentrations show significant reductions in each region both in each season and over all seasons.Seasonally, the numerically largest absolute and relative reductions tend to occur in summer when aerosol SO 4 concentration is at its seasonal maximum, with regionally uniform summertime absolute reductions of ≈ 3.4 µg SO 4 m −3 and relative reductions ranging regionally between 38 and 42 %.The regionality of aerosol SO 4 concentration, a secondary pollutant, is also illustrated in Table 3 by contrasting its across-site variability (CV) with that of its precursor, SO 2 (e.g., 22 % vs. 51 %).
Dry deposition of aerosol SO 4 also shows significant P1to-P4 reductions regionally.Overall relative reductions in Sites are identified by region in the same order as shown in Table 1.  1.
dry SO 4 are numerically less than the corresponding values for concentration of aerosol SO 4 (i.e., for the east, 29 % vs. 33 %).This pattern is likely due in part to significant period-to-period increases in aerosol Vd, noted earlier.Seasonally, the numerically largest absolute and relative reductions in dry SO 4 tend to occur in summer when aerosol SO 4 concentration is at its seasonal maximum, with an average relative reduction of 37 % in summer in the east.Dry SO 4 shows significant reductions in most seasons, except winter, when its concentration is at a seasonal minimum.
Since atmospheric S concentration is the sum of atmospheric SO 2 and aerosol SO 4 concentrations (expressed as mass of sulfur) and ≈ 67 % of atmospheric S is present as SO 2 (RSO 2 ; Table 3), the behavior of atmospheric S concentration is dominated by that of SO 2 .Additionally, since ≈ 85 % of the dry S is usually contributed by SO 2 (RFSO 2 ; Table 3), the behavior of dry S is likewise dominated by that of dry SO 2 .As a result, P1-to-P4 changes of both atmospheric S concentration and dry deposition are similar to those for atmospheric SO 2 concentration and dry deposition, and they are similar to each other (Fig. 4).Atmospheric S concentration and dry S are reduced significantly in each region in each season and over all seasons.Seasonally, the numerically largest significant absolute reductions usually occur in winter when atmospheric SO 2 concentration is at its seasonal maximum, and the largest of these occurs in the northeast.Regionally, P1-to-P4 relative reductions are similar for atmospheric S concentration (42 to 47 %) and dry deposition (48 to 51 %).
In contrast to atmospheric SO 2 concentration (with its summertime minimum), wet S has a seasonal summer maximum.This coincides with scavenging of SO 4 by precipitation when both atmospheric aerosol SO 4 concentration and PR are generally at their seasonal maxima.The numerically largest significant absolute and relative P1-to-P4 reductions in wet S also generally occur in summer.Changes in wet S are numerically smaller and less significant in winter due to seasonal minima of wet S and PR and to wintertime P1-to-P4 increases in PR.In general, relative reductions in wet S are similar to those of aerosol SO 4 concentration (e.g., 33 and 32 % in the east).
Except during winter, wet S generally constitutes more than 50 % of the total S ((D/T) S; Table 3) and exerts a corresponding influence on total S.As noted earlier, P1-to-P4 total S is reduced significantly in each region in each season and over all seasons (Fig. 4).Since seasonal behaviors of dry S and wet S are different but complementary, the seasonal behavior of their sum, total S, shows less seasonal variability than its constituents.Nevertheless, the numerically largest significant P1-to-P4 absolute reductions for total S occur in summer when wet S is at its seasonal maximum.Relative P1to-P4 reductions are similar across regions, and average 40 % in the east.

Oxidized nitrogen species
Sampling artifacts described in Sect. 2 suggest that sampling artifacts may provide biases in opposing directions for concentrations of HNO 3 and aerosol NO 3 , but these impacts would be minimized during non-summer sampling at nonmarine sites.Regional and seasonal changes in atmospheric HNO 3 concentration have more consistent, substantial, and significant reductions for P1-to-P4 (Fig. 4) than for P1-to-P2 or P1-to-P3 (not shown).With recent summertime reductions in NO x emissions, the seasonal timing of peak HNO 3 concentration has shifted from summer in P1 and P2 to spring in P3 and P4.In addition, P1, P2, and P3 summertime peaks of dry deposition of both HNO 3 and OxN have shifted to spring in P4.The overall P1-to-P4 relative reductions in atmospheric HNO 3 concentration in the east are 34 % vs. 0 % (NS) for P1-to-P2 and 13 % for P1-to-P3.Regionally, the numerically largest significant P1-to-P4 reductions occur in the northeast (39 %), with smaller reductions in the south (32 %) and midwest (30 %).Numerically, maximum absolute and relative seasonal reductions consistently occur in summer, with the largest of these in the northeast.Summertime P1-to-P4 relative reductions in atmospheric HNO 3 concentration are 48 % in northeast, 40 % in south, 38 % in midwest, and 42 % in east.Wintertime P1-to-P4 absolute reductions in atmospheric HNO 3 concentration are smaller, 40 to 46 % of those in summer.
Dry HNO 3 exhibits P1-to-P4 reductions that are of similar magnitude and consistency with those noted above for atmospheric HNO 3 concentration.The only exception is winter in the midwest, where P4 Vd was 9 % higher than in P1.Regionally, significant reductions occur in the northeast (41 %), south (35 %), and midwest (28 %), averaging 35 % in the east.Seasonal behavior of dry HNO 3 is similar to that of atmospheric HNO 3 concentration, with numerically maximum absolute and relative seasonal reductions generally occurring in summer.Summertime P1-to-P4 relative reductions in dry HNO 3 are 50 % in northeast, 41 % in south, 34 % in midwest, and 42 % in east.Wintertime P1-to-P4 absolute reductions in dry HNO 3 are smaller, 20 to 28 % of those in summer.
In contrast to P1-to-P3 comparisons, where atmospheric aerosol NO 3 concentration and dry deposition generally display significant increases (SS07b), P1-to-P4 comparisons in Fig. 4 show more reductions.Significant P1-to-P4 absolute and relative increases in atmospheric aerosol NO 3 concentration and dry NO 3 occur only in winter in the east (e.g., 0.20 µg NO 3 m −3 and 12 %, and 0.03 kg N (ha yr) −1 and 31 %) when atmospheric aerosol NO 3 concentration is at its seasonal maximum.The increased magnitude of winter relative increases in dry NO 3 over those of airborne concentration are likely due to previously noted significant period-to-period increases in aerosols Vd.Wintertime regional absolute P1-to-P4 increases in concentration and dry deposition are numerically strongest in the midwest (e.g., 0.27 µg NO 3 m −3 and 0.05 kg N (ha yr) −1 ).The wintertime Significance level of period-to-period relative changes indicated by bar fill: black, p ≤ 0.05; and open, p>0.05 (NS).Sites are identified by region in the same order as shown in Table 1.
P1-to-P4 increases in aerosol NO 3 concentration are considerably smaller than the corresponding P1-to-P3 increases (Fig. 7).Significant P1-to-P4 reductions in absolute and relative atmospheric aerosol NO 3 concentration occur in many of the remaining seasons, especially summer.Significant P3-to-P4 reductions in relative atmospheric aerosol NO 3 concentration occur in all seasons in the east, consistent with recent aggressive reductions of NO x emissions.
Atmospheric OxN concentration is the sum of atmospheric HNO 3 and aerosol NO 3 concentrations (expressed as mass of nitrogen), and almost 60 % of the OxN is generally present as HNO 3 (RHNO 3, ; Table 3).As a result, the behavior of atmospheric OxN concentration reflects the influence of both atmospheric HNO 3 and aerosol NO 3 concentrations.Also, since well over 90 % of the dry OxN is generally contributed by HNO 3 (RFHNO 3, ; Table 3), the behavior of dry OxN is strongly influenced by that of dry HNO 3 .Seasonally in the east, the numerically largest P1-to-P4 absolute and relative reductions in atmospheric OxN concentration and dry deposition (e.g., 0.26 µg N m −3 and 38 %, and 1.23 kg N (ha yr) −1 and 43 %; Fig. 4) occur in summer when atmospheric HNO 3 is the dominant OxN species.Summertime relative reductions are numerically more pronounced in the northeast (43 and 50 %, respectively) than the south (35 and 43 %) or midwest (36 and 34 %).In contrast, in winter, when atmospheric aerosol NO 3 concentration is at its seasonal maximum and significant P1-to-P4 increases of both atmospheric aerosol NO 3 concentration and dry deposition occur, reductions in OxN concentration and dry deposition also occur, but they are smaller in magnitude and in some cases not significant.Over all seasons, regional P1-to-P4 relative reductions in atmospheric OxN concentration and dry deposition are numerically larger in the northeast (29 and 40 %) than the south (24 and 34 %) or the midwest (21 and 27 %), and average 24 and 34 %, respectively, in the east.
Wet OxN and PR in the east are at their seasonal maxima in summer (followed closely by spring).Numerically, the largest significant absolute reductions in wet OxN occur in summer (or spring).Reduced significance of some wintertime P1-to-P4 reductions in wet OxN is associated with numerical increases in wintertime PR.Significant regional P1-to-P4 reductions in wet OxN are numerically larger in the northeast (34 %) than midwest (28 %) or south (25 %), and average 29 % in the east.
Considering the dry deposition of a comprehensive suite of NO y species, field measurements suggest that 40 to 85 % of total NO y deposition at eight rural Canadian sites occurred as wet deposition (Zhang et al., 2009), whereas a modeling study suggests that for the continental US this figure is closer to 31 % for NO y and 44 % for OxN (Zhang et al., 2012).In the current study, over 60 % of the total OxN generally is contributed by wet OxN ((D/T) OxN; Table 3); however, this estimate may be biased because it neglects previously noted uncertainties in Vd for HNO 3 and contributions to dry deposition by non-monitored NO y species.Nevertheless, wet OxN appears to exert a moderate-to-strong influence on the behavior of total OxN.Both dry OxN and wet OxN display strong seasonal P1-to-P4 reductions in summer that become weaker in winter (Fig. 4).Seasonally, the numerically largest absolute and relative reductions in total OxN occur in summer and range from 42 % in the northeast, to 35 % in the south, to 33 % in the midwest, and average 37 % in the east.Regional P1-to-P4 reductions in total OxN are 36 % in the northeast, 29 % in the south, and 28 % in the midwest, and average 31 % in the east.Thus, over all seasons the numerically largest absolute and relative reductions in atmospheric HNO 3 concentration and dry deposition, atmospheric OxN concentration and dry deposition, and total OxN generally occur in the northeast, and seasonally they tend to occur there in summer.

Reduced nitrogen species
Since NH 3 is not monitored in CASTNET, aerosol NH 4 is the reduced nitrogen species considered in the current study.Numerically, the regional maximum for atmospheric aerosol NH 4 concentration occurs near emissions sources in the source region and midwest, and the seasonal maximum in the east occurs in summer.As is illustrated in Fig. 4, between P1 and P4, atmospheric aerosol NH 4 concentration shows significant regional relative reductions that are similar, ranging from 29 % in the midwest, to 27 % in the northeast, to 24 % in the south, and averaging 26 % in the east.Seasonally, the numerically largest absolute and relative reductions of atmospheric aerosol NH 4 concentration occur in summer (when atmospheric aerosol SO 4 concentration is at its seasonal maximum and is also experiencing its largest absolute and relative reductions).Summertime P1-to-P4 relative reductions in atmospheric aerosol NH 4 concentration range from 40 % in the midwest, to 32 % in the northeast, to 26 % in the south, and average 32 % in the east.In contrast, reductions are smaller in winter (e.g., 14 % in the east) when atmospheric aerosol NH 4 concentration is small and atmospheric aerosol SO 4 concentration is also at its seasonal minimum.
Although generally showing significant reductions, dry deposition of atmospheric aerosol NH 4 shows numerically weaker P1-to-P4 changes than its atmospheric concentration.The overall P1-to-P4 relative reduction for dry NH 4 in the east (22 %) is numerically smaller than the corresponding value for its atmospheric concentration.This pattern of weaker reductions is likely due to the significant P1-to-P4 increases in aerosol Vd, noted earlier.Regional P1-to-P4 relative reductions in dry NH 4 are 26 % in the northeast and 21 % in the midwest and south.Seasonally, in winter, when the atmospheric aerosol NH 4 concentration is low, changes in its dry deposition show changes which tend to be small and not significant regionally.Mirroring aerosol NH 4 concentration, the numerically largest P1-to-P4 absolute and relative reductions in dry NH 4 also occur in summer.They range from 34 % in the northeast, to 33 % in the midwest, to 24 % in the south, and average 30 % in the east.
Numerically, wet NH 4 is also high near emissions sources in the midwest and source region, and in the east in the spring and summer seasons when PR is also relatively high.At the regional and seasonal scales, wet NH 4 shows P1-to-P4 changes that are usually not significant.The only two reductions that are significant occur overall in the midwest (12 %) and east (9 %).Two wintertime increases occur in the midwest and the east, but they are not significant.
Since the magnitude of wet NH 4 is over seven times larger than dry deposition ((D/T) NH 4 ; Table 3), the behavior of total NH 4 is strongly influenced by the behavior of wet NH 4 .Similar to wet NH 4 , total NH 4 shows mostly P1-to-P4 reductions that are usually not significant, but with non-significant wintertime increases in the midwest and the east.The only significant reduction is overall in the east (11 %).

Oxidized plus reduced nitrogen species
The sum of monitored oxidized and reduced nitrogen species is represented as N.The behavior of the resulting N metrics (i.e., atmospheric N concentration, dry N, wet N, and total N) reflects the behavior of the weighted sum of the individual constituents, described previously.
Aerosol NH 4 constitutes ≈ 67 %, HNO 3 ≈19 %, and aerosol NO 3 ≈ 14 % of the monitored airborne N concentration (RNNH 4 and RNHNO 3 ; Table 3).As a result, aerosol NH 4 exerts a moderate-to-strong influence on its behavior.Seasonally, the numerically strongest absolute and relative P1-to-P4 reductions of atmospheric N concentration occur in summer (midwest, 39 %; northeast, 34 %; south, 28 %; and east, 33 %).This seasonality coincides with the summer peaks of atmospheric N concentration and P1-to-P4 reductions in the concentration of aerosol NH 4 , HNO 3 , and OxN.Regionally, atmospheric N concentration shows relative reductions of 26 % in the midwest, 27 % in the northeast, 24 % in the south, and 26 % in the east.
Due to the large difference in deposition velocities, ≈ 75 % of dry N is contributed by dry HNO 3 , but only ≈ 21 % by dry NH 4 and ≈ 4 % by dry NO 3 (RNFHNO 3 and RNFNH 4 ; Table 3).Seasonally, the numerically strongest absolute and relative P1-to-P4 reductions in dry N mirror those of aerosol NH 4 , HNO 3 , and OxN concentration and dry deposition (and N concentration); they occur in summer (northeast, 47 %; south, 39 %; midwest, 34 %; and east, 40 %).Regional non-significant wintertime P1-to-P4 reductions may result in part from corresponding significant regional increases in dry NO 3 .Regionally, the numerically largest P1-to-P4 reductions occur in the northeast (38 %), with smaller reductions in the south (32 %) and midwest (26 %), and they average 32 % in the east.
Wet OxN and wet NH 4 contribute similarly to wet N, with maxima in spring or summer (Table 3).The numerically largest significant P1-to-P4 absolute reductions in wet N also occur in spring or summer, and peak seasonal relative reductions range between 20 and 30 %. Non-significant numerical wintertime P1-to-P4 reductions in wet N occur in the northeast and south, with a non-significant increase in the midwest.This behavior is likely associated with previously noted numerical wintertime P1-to-P4 increases in PR.Regional P1-to-P4 relative reductions in wet N are 24 % in the northeast, 20 % in the midwest, 17 % in the south, and average 20 % in the east.
Since wet N contributes ≈ 75 % of total N ((D/T) N; Table 3), the behavior of total N is strongly influenced by the behavior of wet N. The numerically largest significant P1-to-P4 absolute reductions in total N generally occur in summer, with corresponding seasonal relative reductions of 31 % in northeast, 25 % in midwest, and 26 % in south.Nonsignificant P1-to-P4 reductions in total N occur only in wintertime and are associated with similar behavior of wet OxN and wet N and numerical wintertime P1-to-P4 increases in PR.Regionally, the numerically largest P1-to-P4 reductions occur in the northeast (28 %), with smaller reductions in the midwest and south (22 %).The overall P1-to-P4 relative reduction of total N in the east is 24 %.

Wet H +
Regionally, wet H + deposition is generally heaviest in the northeast, and seasonally in the east it is heaviest during summer (Table 3).Seasonally, the numerically largest absolute reductions occur in summer, with relative reductions of similar magnitude in spring and summer, while generally smaller reductions occur in winter (Fig. 4).The numerically largest seasonal relative reductions range from 65 % in the midwest (summer), to 55 % in the northeast (spring), to 50 % in the south (spring).Regionally, P1-to-P4 regional relative reductions of wet H + are numerically larger near sources in the midwest (56 %) and smaller in the south (46 %) and northeast (44 %).The overall P1-to-P4 relative reduction in wet H + in the east is 47 %.

Atmospheric O 3 concentration and dry deposition
In the current study, hourly atmospheric O 3 concentrations, averaged to weekly values, are treated as other weekly CASTNET data (i.e., averaged to period-site-season means); however, this does not represent the relevant time period for determining compliance with the US ambient air quality standard.As is illustrated in Fig. 4, mostly nonsignificant P1-to-P4 seasonal increases in atmospheric O 3 concentration occur in winter in the northeast (13 %, NS), midwest and south (7 %, NS), and east (9 %, significant).Summertime atmospheric O 3 concentrations do show substantial and mostly significant absolute and relative reductions in each region: 6.7 ppb (17 %) in the northeast, 5.3 ppb (14 %) in the south, 2.4 ppb (6 %, NS) in the midwest, and 4.9 ppb (12 %) in the east.The numerically largest reduc-tion in atmospheric O 3 concentration occurs in summer and in the northeast, coinciding with aggressive O 3 -season NO x emissions controls during P3 and P4 in this region and upwind in the east.Unpublished examination of seasonal NO x emissions in the eastern US for the period between 1990-1992 and 2006-2008 (roughly corresponding to P1-to-P4) suggests that reductions in summertime emissions are approximately 6 % greater than those for the aggregate of the remaining three seasons over the same time period.Regional reductions of atmospheric O 3 concentration over all seasons are small and only significant in the south (5 %) and east (4 %).
Moderate P1-to-P4 increases in Vd for O 3 act in opposition to corresponding reductions in atmospheric O 3 concentration to yield P1-to-P4 changes in dry deposition of O 3 that are sometimes significant.Relatively large, significant seasonal reductions in dry O 3 , ranging from 5 to 15 %, occur regionally in summer; however, corresponding increases ranging between 15 and 25 % occur in winter.Regionally, P1-to-P4 reductions in dry O 3 occur in the northeast (8 %) and south (4 %), and a non-significant increase occurs in the midwest (1 %, NS).The P1-to-P4 relative reduction in dry O 3 in the east is 4 %.

Changes in atmospheric aerosol SO 4 and SO 2 concentration
Both P1-to-P4 absolute and relative changes in atmospheric aerosol SO 4 concentration, shown in Fig. 4, are numerically substantially less than the corresponding values for atmospheric SO 2 concentration, consistent with the findings of Reid et al. (2001), Holland et al. (2004), and SS07b for earlier time periods.For regional groupings of sites and in the east, the year-round ratios of P1-to-P4 relative reductions in atmospheric aerosol SO 4 concentration to atmospheric SO 2 concentration are similar (≈ 0.7 % decrease in aerosol SO 4 for each 1 % decrease in SO 2 ).In the east these ratios range numerically between a wintertime low (≈ 0.5 % SO 4 per 1 % SO 2 ) and a high (≈ 0.8 % SO 4 per 1 % SO 2 ) in summer when aerosol SO 4 levels are at their seasonal maxima and SO 2 levels are at their seasonal minima.Year-round ratios of P1-to-P4 absolute reductions of atmospheric aerosol SO 4 concentration to those of SO 2 concentration (both expressed as mass of sulfur) are similar in regions of higher emissions density, the midwest and northeast (≈ 0.2 µg S m −3 decrease in aerosol SO 4 for each 1 µg S m −3 decrease in SO 2 ).This ratio is numerically higher in the south (≈ 0.3), and it may be due in part to concurrent reductions of SO 2 emissions in areas of higher emissions density with subsequent transport of reduced amounts of oxidized sulfur, especially aerosol SO 4 , into the south.RSO 2 (Table 3) is the ratio of atmospheric SO 2 concentration to total atmospheric S concentration, and 1-RSO 2 is the relative amount of atmospheric S concentration present as aerosol SO 4 , representing a rough index of the extent of conversion of SO 2 to other oxidized sulfur species (i.e., aerosol SO 4 ).Significant P1-to-P4 RSO 2 reductions (not shown) of 8 % in the south, 5 % in the northeast, and 7 % in the midwest and east correspond to increases in the conversion index (e.g., 17 % in the east).This suggests that as the P1-to-P4 atmospheric SO 2 concentration in the east has decreased, the relative amount of airborne S present as SO 2 has decreased, and both the relative amount present as aerosol SO 4 and the apparent extent of conversion of SO 2 to aerosol SO 4 have increased.

Changes in dry and wet deposition
As noted in Sect.3.2, wet S exceeds dry S in the east, except in winter.The P4 ratio of dry S to total S in the east is 0.37 ((D/T) S; Table 3), down from 0.44 in P1; further, qualitatively similar behavior is noted for the BKG sites (i.e., P4 ratio 0.23 vs. 0.26 in P1) and HE sites (i.e., P4 ratio 0.40 vs. 0.50 in P1).In the east, the ratio of dry S to wet S (calculated from (D/T) S) has diminished by 24 % from P1 (0.78) to P4 (0.60).This reduction is strongest (> 30 %) in regions of high emissions density (midwest and northeast) in winter, where SO 2 concentrations are at their peak.Year-round, in the east and regionally, both P1-to-P4 absolute and relative reductions in dry S are numerically larger (by ≈ 20 to 70 %; Fig. 4) than corresponding values for wet S.However, in summer when PR is high and atmospheric aerosol SO 4 concentration is at its seasonal maximum, the absolute P1to-P4 reductions are larger (by ≈ 25 to 100 %) for wet S than dry S. In other seasons, the roles are different, and in winter, when atmospheric SO 2 concentration is at its seasonal maximum and PR is at its seasonal minimum, P1-to-P4 absolute and relative reductions in dry S are substantially larger (by over a factor of ≈ 2) than the corresponding reductions for wet S. Thus, strong emissions controls of the primary pollutant SO 2 have reduced dry S to a greater extent than wet S, permitting the latter to become an increasing relative contributor to total S. This, along with the increased apparent conversion of SO 2 to aerosol SO 4 , noted above, suggests that wet processes are becoming more prominent than dry processes at removing oxidized sulfur from the current atmosphere in the east.
Whereas deposition estimates of both primary and secondary oxidized sulfur species are available for comparison of their dry and wet P1-to-P4 changes, only changes in deposition of monitored secondary oxidized nitrogen species can be compared in the current study.In the east, dry OxN appears to be numerically smaller than wet OxN (Sect.3.3), and the P4 ratio of dry OxN to total OxN is 0.35 ((D/T) OxN; Table 3), down slightly from 0.37 in P1.Qualitatively similar behavior is noted for the BKG sites (i.e., P4 ratio 0.27 vs. 0.30 in P1) and HE sites (i.e., P4 ratio 0.41 vs. 0.51 in P1).Absolute P1-to-P4 reductions in dry OxN are generally numerically smaller than corresponding values for wet OxN (Fig. 4).However, with the recent, more aggressive NO x emissions controls, the relative P1-to-P4 reductions of dry OxN now exceed those for wet OxN more frequently than in the earlier P1-to-P3 comparison (SS07b).These findings suggest that wet OxN is becoming an increasing relative contributor to total OxN.

CASTNET PM
Although PM 2.5 is not monitored in CASTNET, many of the major inorganic constituents (i.e., aerosol SO 4 , NO 3 , and NH 4 ) are monitored.In contrast to aerosol SO 4 and NH 4 , which occur mainly as fine particles, there is evidence suggesting that aerosol NO 3 can be multi-modal in the east and is subject to sampling artifacts in opposing directions with the CASTNET sampler, but they are thought to be minimal during cooler seasons at the non-marine, eastern CASTNET sites (see Sect. 2).In the current study, CASTNET PM is taken to be the sum of atmospheric aerosol SO 4 , NO 3 , and NH 4 mass concentrations.
Significant P1-to-P4 (Fig. 5) and P3-to-P4 reductions occur in CASTNET PM concentration in each region in each season and over all seasons.This is in contrast to P1-to-P3 behavior where non-significant wintertime changes occurred (SS07b), with both numerical and significant increases in the midwest (Fig. 8).Year-round absolute and relative reductions in the east are 2.3 µg m −3 and 28 % (P1-to-P4) and 1.1 µg m −3 and 16 % (P3-to-P4).Seasonally, they peak in summer when aerosol SO 4 and NH 4 concentrations are at their maxima and are also experiencing their largest reductions.Regionally, the largest absolute and relative reductions in CASTNET PM occur in the midwest in summer (4.9 µg m −3 and 41 % -P1-to-P4 and 1.9 µg m −3 and 21 % -P3-to-P4).
For the chemically complex NH 3 -HNO 3 -H 2 SO 4 water system, conditions can arise where the availability of NH 3 limits the formation of aerosol NH 4 NO 3 .Here, subsequent reductions of airborne aerosol SO 4 , by tying up less NH 3 as aerosol (NH 4 ) 2 SO 4 , may make more NH 3 available to react with HNO 3 , increasing the level of aerosol NH 4 NO 3 .Although atmospheric NH 3 concentration is not monitored in CASTNET, when the index, RNO 3 (i.e., the ratio of atmospheric aerosol NO 3 to OxN concentrations), is less than 0.7, then formation of aerosol NH 4 NO 3 may be NH 3 -limited (Blanchard et al., 2000).Using this index and focusing on non-summer season results to minimize the impacts of sampling artifacts (see Sect. 2), results in Table 3 suggest that NH 3 -limited conditions are present in the east but are weakest in winter and year-round in the midwest.
Significant P1-to-P2-to-P3-to-P4 increases in RNO 3 occur in the east.Regionally, the numerically largest absolute and relative P1-to-P4 increases occur in the northeast (40 %), with smaller increases in the south (28 %), midwest (10 %), and east (22 %).Seasonally, the numerically largest absolute increases occur in winter.The numerically largest regional P1 winter RNO 3 value (0.70) occurs in the midwest, and this value increases significantly by 12 % to 0.78 in P4.These results suggest that between P1 and P4, NH 3 availability has increased in the east and the formation of aerosol NO 3 has become less NH 3 -limited.The least NH 3 -limited situation appears to occur in winter in the midwest (with its strong sources of NH 3 ).In addition, the relative amount of monitored OxN present as HNO 3 , RHNO 3 (i.e., 1-RNO 3 ), diminished significantly in the east between P1 (0.66) and P4 (0.59) (Sect.3.3).
For a chemical system initially containing NH 3 , HNO 3 , H 2 SO 4 , and water, an index of the extent to which ambient sulfuric acid (H 2 SO 4 ) has been neutralized by atmospheric NH 3 may be defined as the neutralization index (NI).Correcting for the presence aerosol NH 4 NO 3 , it may be computed as ([NH 4 and Hidy, 2005).The value of NI can range between zero (un-neutralized H 2 SO 4 ) and unity (100 % neutralization to (NH 4 ) 2 SO 4 ).As shown in Table 3, NI averages ≈ 0.8 in the east.Calculation of NI assumes that aerosol NO 3 is present as fine aerosol NH 4 NO 3 and neglects the possible presence other forms of aerosol NO 3 (e.g., from large particle sea/road salt or soil/mineral dust).As suggested in Sect.2, sampling artifacts may provide biases in opposing directions for aerosol NO 3 .Their impacts are expected to be minimal during cooler season sampling at non-marine sites.Although small (but significant) P1-to-P4 increases in NI occur in the east, large increases occur in summer in the northeast (15 %) and south (18 %), while small non-significant decreases occur in winter.These results suggest that between P1 and P4 an acidic aerosol is approaching neutralization due to increased relative availability of NH 3 associated with reductions of aerosol SO 4 concentration.These findings are consistent with those noted in previous paragraphs and with those for P1-to-P3 (SS07b), indicating that conditions in the east are becoming less NH 3 -limited.
Modeling efforts have indicated that nonlinear changes in PM 2.5 concentrations can occur in response to reductions in atmospheric aerosol SO 4 concentrations in the east (Ansari and Pandis, 1998;West et al., 1999;Blanchard and Hidy, 2005).Under NH 3 -limited conditions Blanchard et al. (2000) suggest that reduction of aerosol SO 4 concentrations can increase aerosol NO 3 concentrations by amounts that exceed the aerosol SO 4 reductions on a mass basis, and that reductions of ambient oxidized nitrogen concentrations (that may accompany reductions in NO x emissions) may not reduce aerosol NO 3 concentrations.These modeling studies suggest that for the NH 3 -HNO 3 -H 2 SO 4 water system, constituent concentration and temperature conditions can exist in the east so that when the atmospheric aerosol SO 4 concentration is reduced, the corresponding change in PM 2.5 concentration can be influenced nonlinearly.At low temperatures, aerosol NH 4 NO 3 will likely form when NH 3 is made available from a reduction in atmospheric aerosol SO 4 concentration, and negative to nonlinear responses ranging from −0.3 to < 1.0 ( µg PM 2.5 / µg SO 4 ) can result (Ansari and Pandis, 1998;West et al., 1999).Using model estimates, West et al. (1999) have predicted that these conditions would exist most commonly in winter in the midwest and would be uncommon in summer in the east.A 2009 field study (Stanier et al., 2012) has focused on wintertime episodes of elevated aerosol NO 3 concentrations in the midwest (Wisconsin).
Ratios of P1-to-P4 absolute changes in CASTNET PM concentrations to corresponding changes in aerosol SO 4 -(µgCASTNET PM m −3 )/ (µg SO 4 m −3 ) -have been determined.Year round, these ratios are of similar magnitude, numerically falling between 1.2 and 1.6, both for the east and for each regional grouping of sites.Seasonally in the east, these ratios range between 1.3 and 1.5 in spring, summer, and fall.However, in winter they are numerically close to 1.0 in the east and each region.This represents a change from P1to-P3 behavior, where the wintertime ratio averaged close to 0.0 in the east, and in the midwest was negative and accompanied by a numerical increase in CASTNET PM (SS07b).Wintertime numerical increases in CASTNET PM occurred at 15 sites between P1 and P3, but at only one site between P1 and P4 and at only one site between P3 and P4 (Fig. 8).This wintertime behavior of CASTNET PM appears to be driven largely by that of aerosol NO 3 (Fig. 7).These findings suggest that additional P3-to-P4 reductions in emissions of SO 2 , and especially NO x , have made progress in altering the chemical regime of the wintertime eastern US atmosphere so that future emissions reductions and their resulting reductions in aerosol concentrations may no longer be accompanied by sub-linear changes (or actual increases) in CASTNET PM.

Comparison with emissions estimates
Emissions estimates for SO 2 and NO x (Xing et al., 2013) have been aggregated to regional levels and have been used to determine current mean estimates of P1, P2, P3 and P4 emissions as well as period-to-period absolute and relative changes in regional SO 2 and NO x emissions in the east (see Sect. 2.1 for identification of the states composing each geographical region).Although current P1, P2, and P3 emissions estimates have been updated and may be slightly different compared to those reported in SS07b, using the same current version permits period-to-period comparisons to be made using consistent emissions estimates.In addition, subsequent analyses, repeated using emissions estimates (US EPA, National Emissions Inventory, Emissions Factor and Inventory Group, Office of Air Quality Planning and Standards, Research Triangle Park, NC, http://www.epa.gov/ttn/chief/net/2008inventory.html,accessed 02/12/13), yielded reasonably consistent findings with no appreciable changes to conclusions.
Continuous emissions monitors used on most EGUs provide reasonably accurate estimates of SO 2 and NO x emissions from these sources.Although EGUs are the main source of SO 2 emissions; EGUs, on-road, and off-road sources contribute substantially to total NO x emissions.Uncertainty in on-road and other non-EGU NO x emissions make estimates of total NO x emissions more uncertain than those for SO 2 (Butler et al., 2005;Parrish, 2006;Blanchard et al., 2013).
Estimates of both P1 and P4 emissions density are shown in Table 4 in the same units as deposition.For example, the P1 emissions density for SO 2 is over 60 % higher in the midwest, with its EGU sources in the Ohio Valley, than in the south (i.e., 51 vs. 31 kg S (ha yr) −1 ).P1-to-P4 relative change results for regional SO 2 and NO x emissions (Fig. 3) are also shown in Table 4 with corresponding values for selected measures of atmospheric concentration and total deposition from Fig. 4.
As indicated in Table 4, the P1-to-P4 relative reductions in SO 2 emissions in the east (50 %) are in closer numerical agreement with reductions of atmospheric concentration of the monitored primary pollutant, SO 2 (48 %), than the monitored reaction product, aerosol SO 4 (33 %).Ratios of the P1-to-P4 relative change in monitored concentration to emissions estimates for SO 2 in the east are approximately 0.97 for SO 2 and 0.66 for aerosol SO 4 , with similar findings for P3-to-P4 comparisons.Relative P1-to-P4 comparisons also show significant reductions for SO 2 emissions (50 %) and dry S (49 %).These results suggest that at the current scale, the responses of both atmospheric SO 2 concentration and dry deposition to changes in SO 2 emissions are close to 1 : 1; however, the relationship appears to be less than 1 : 1 for aerosol SO 4 .In addition, as indicated here and in Sect.4.1, the relationship between changes in the concentration of the reaction product, aerosol SO 4 , and that of the primary pollutant, SO 2 , appears to be less than 1 : 1.
These findings are generally consistent with earlier findings for P1-to-P3 comparisons (SS07b) as well as those of those of others.Butler et al. (2001)  from specific upwind source regions.Blanchard et al. (2013) found southeastern US trends of SO 2 ambient concentrations and emissions to be decreasing with a qualitative 1 : 1 proportionality, while the relationship for trends of ambient aerosol SO 4 concentrations and SO 2 emissions were decreasing but at less than 1 : 1 proportionality.Hand et al. (2012) reported linear relationships between aerosol SO 4 concentrations and SO 2 emissions that depended on location in the US, with aerosol SO 4 concentration showing the weakest response to SO 2 emissions in a region in the eastern US roughly included in the currently defined source region, midwest, and northeast.
Atmospheric pollutants are transported and transformed by physical and chemical processes between their release and deposition.For example, gaseous SO 2 may be transformed into aerosol SO 4 by several pathways, including gasphase oxidation by the hydroxyl radical and, after dissolution in water droplets, via oxidation by hydrogen peroxide and O 3 (McHenry and Dennis, 1994).As SO 4 -containing aqueous particles undergo atmospheric processing (i.e., evaporation and dissolution), there is a close linkage between the secondary pollutant's atmospheric concentration and wet removal (i.e., they have similar atmospheric timescales).Although SO 4 is then subject to both dry and wet deposition, the latter dominates (e.g., east P4 wet S is 5.2 kg S (ha yr) −1 vs. 0.4 kg S (ha yr) −1 for dry SO 4 ; Table 3).This may explain the relatively good agreement between P1-to-P4 (33 %) and P3-to-P4 (15 %) reductions in the atmospheric concentration of secondary pollutant, aerosol SO 4 , and corresponding changes (32 and 13 %) in wet S.
Since atmospheric S concentration is comprised of SO 2 and aerosol SO 4 , reductions of SO 2 emissions may be expected to yield responses of atmospheric S concentration and total S that fall between corresponding values for SO 2 and aerosol SO 4 .This is confirmed in Table 4 where reductions in SO 2 emissions in the east correspond to significant reductions of both atmospheric S concentration (i.e., P1-to-P4, 50 vs.45 %) and total S (i.e., P1-to-P4, 50 vs.40 %).
For a pollutant where all primary and secondary species are monitored (e.g., S), neglecting other sources and sinks, emissions density may be compared with total deposition to construct rough budgets for oxidized sulfur, revealing insights about the relative amounts of the species deposited and exported.Model predictions for North America (Wojcik and Chang, 1997) suggest approximately 50 % of S emissions are deposited.In the current study, P1, P2, P3 and P4 S emissions densities in the east are 38, 30, 25, and 19 kg S (ha yr) −1 , while corresponding total S values are 15, 12, 11 and 8.7 kg S (ha yr) −1 .Between P1 and P4 the relative amount of deposited S emissions increased by ≈ 8 % from 39 to 47 %, and the net exported S emissions decreased by a corresponding amount from 61 % to 53 %.Thus, the strong P1to-P4 SO 2 emissions controls are accompanied by increased apparent conversion of SO 2 to aerosol SO 4 (Sect.4.1), possible increased relative contribution of wet vs. dry deposition (Sect.4.2), increased fraction of deposited S emissions, and decreased fraction of exported S emissions.

Oxidized nitrogen species
Changes in pollutant emissions generally lead to more direct changes in the atmospheric concentration of primary rather than secondary pollutant species, where emissions as well as variable meteorological and atmospheric chemical processes are influential.In contrast to oxidized sulfur species, where the behaviors of both primary and secondary pollutants are monitored, the atmospheric concentration and deposition of only a few of the secondary oxidized nitrogen species (i.e., HNO 3 , aerosol NO 3 and aqueous NO 3 ion) are monitored in the current study.This likely propagates more uncertainty into comparisons with primary NO x emissions than is present in earlier comparisons with SO 2 emissions.
The numerically largest P1-to-P4 relative reductions in estimated NO x emissions (46 %), year-round atmospheric concentrations of HNO 3 (39 %) and OxN (29 %), dry OxN (40 %), wet OxN (34 %), total OxN (36 %), and summertime total OxN (42 %) occur in the northeast, the region with the largest P1 emissions density (Table 4).As noted in Sect.3.3, numerical maximum absolute and relative seasonal reductions in atmospheric HNO 3 concentrations also occur in summer, with largest reductions in the northeast.These changes are likely reflecting contributions from year-round control of NO x emissions by Phases I and II of the CAAA and continuing mobile source controls.However, the timing (e.g., changes most apparent in summer with peak concentrations moving from summer to spring) and the location (i.e., strongest reductions in northeast) suggest that major contributors are summertime NO x emissions control activities instituted under OTC, SIP Call, BTP and CAIR that focused on EGU sources in the eastern states.
As noted above and in SS07b, the 22 % reduction in the east of P1-to-P3 NO x emissions is associated with a 13 % reduction in HNO 3 concentration, an 8 % reduction in dry OxN, a 12 % reduction in wet OxN, an 11 % reduction in total OxN, but an 11 % increase in aerosol NO 3 concentration.Here, the ratio of P1-to-P3 relative changes of monitored HNO 3 concentration to those of estimated NO x emissions is 0.6, qualitatively consistent with the value of 0.7 found by Butler et al. (2005) for a similar time period (1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001).Recent aggressive NO x emissions controls, however, appear to have altered these relationships.As indicated in Table 4, the 42 % P1-to-P4 reduction of NO x emissions in the east corresponds to 34 % reductions in HNO 3 concentration and dry OxN, a 29 % reduction in wet OxN, a 31 % reduction in total OxN, and an 8 % reduction in aerosol NO 3 concentration.Here, the ratio of P1-to-P4 relative changes of monitored HNO 3 concentration and dry OxN to those of estimated NO x emissions is 0.8, but the corresponding ratio for aerosol NO 3 concentration is 0.2.These findings are consistent with the recent observations in the southeastern US of decreasing 1999-2010 trends of NO y concentration that exceed those for aerosol NO 3 concentration and are qualitatively proportional to decreasing NO x emissions trends (Blanchard et al., 2013).
In the east, the 42 % P1-to-P4 reduction of NO x emissions also corresponds to significant reductions of deposition, i.e., dry HNO 3 (35 %), dry OxN (34 %), wet OxN (29 %) and total OxN (31 %).The relatively good agreement between the P1-to-P4 changes in the atmospheric concentration of HNO 3 , dry HNO 3 , and dry OxN can be explained by the high RFHNO 3 (Sect.3.3; Table 3).The favorable comparison with wet OxN may be related to the high relative abundance and high water solubility of HNO 3 in comparison to other NO y species in the atmosphere.It could also be related to a possible high correlation of atmospheric concentration of HNO 3 with those of other airborne NO x reaction products that are removed from the atmosphere and measured as aqueous NO 3 ion in precipitation.
Model predictions for the continental US (Zhang et al., 2012) suggest approximately 63 % of 2006-2008 NO x emissions are deposited as NO y and 45 % as OxN.In the current study, none of the primary NO x species and only two of the secondary species (HNO 3 and aerosol NO 3 ) are monitored.Nevertheless, neglecting other sources and sinks, it is possible to compare NO x emissions density with total OxN deposition to determine the fraction of NO x emissions deposited as OxN.Using the approach described in Sect.4.4.1 for oxidized sulfur, approximately 30 % of the NO x emissions appear to be deposited as OxN during the study period.Although there are hints of a numerically small P1to-P4 increase in the relative amount of NO x emissions deposited as OxN, the large relative uncertainty associated with NO x emissions estimates (Sect.4.4) precludes a conclusion in this instance.Nevertheless, the strong P1-to-P4 reduction of NO x emissions (42 %) is accompanied by reduced atmospheric concentration of HNO 3 and OxN as well as RHNO 3 (Sects. 3.3 and 4.3), and by increased relative contribution of wet OxN vs. dry OxN (Sect.4.2).

Wet H +
In the east the P1-to-P2 reductions in estimated emissions are 21 % for SO 2 and 6 % for NO x .Concurrent changes in deposition include reductions of 17 % for wet S, 3 % (NS) for both wet OxN and wet NH 4 , and 24 % for wet H + (not shown).In the absence of sizeable P1-to-P2 changes in NO x emissions, it appears that P1-to-P2 reductions in SO 2 emissions played a major role in the resulting reductions of both wet S and wet H + in the east.P1-to-P3 reductions in estimated SO 2 and NO x emissions are 35 and 22 %, with corresponding reductions of 22 % in wet S, 12 % in wet OxN, no change in wet NH 4 , and 34 % in wet H + (SS07b).For P1-to-P4, reductions in estimated SO 2 and NO x emissions are 50 and 42 %, and corresponding reductions are 32 % for wet S, 29 % for wet OxN, 9 % for wet NH 4 , and 47 % for wet H + (Fig. 4).Respective P3-to-P4 reductions in estimated SO 2 and NO x emissions are 23 and 25 %, with corresponding reductions of 13 % in wet S, 19 % in wet OxN, 9 % in wet NH 4 , and 20 % in wet H + .
Changes in wet NH 4 between P1 and P3 are generally small, variable, and non-significant.Although significant P1-to-P4 reductions in wet NH 4 are also relatively small (≈ 9 %), they occur mainly between P3 and P4.In the east between P1 and P4, with the continuing reduction in SO 2 and NO x emissions, the fraction of wet S plus wet OxN that could be neutralized on an equivalent basis by wet NH 4 increased by approximately 8 % from 25 % in P1, to 33 % in P4.
Neglecting the relatively small temporal impacts of wet NH 4 (above) and those of other wet cations, examination of the behavior of wet S plus wet OxN (on an equivalent basis) permits rough estimation of the relative contributions of reductions in SO 2 and NO x emissions to corresponding reductions in wet H + .Year-round in the east, P1-to-P3 reductions in wet S contribute most (≈ 80 %) of the relative reduction of wet S plus wet OxN associated with reduced wet H + .Between P1 and P4, this relative contribution diminishes (to ≈ 70 %), and between P3 and P4 the relative contribution of wet S (≈ 55 %) approaches that of wet OxN.This finding illustrates the impact of recent aggressive NO x emissions controls along with continuing SO 2 controls on wet H + and the increasing relative contribution of reductions in NO x emissions to the significant and substantial reduction of wet H + in the eastern US.

Summary, conclusions, and recommendations
Data collected in the eastern US between 1990 and 2009 at 34 CASTNET dry monitoring sites and paired NADP wet monitoring sites are examined.A major objective is to evaluate the monitored air quality impacts occurring between 1990 and 2009 that are associated with concurrent legislatively mandated changes in emissions.Four 5-year periods (P1, P2, P3, and P4) are considered.Period-to-period changes in selected pollutant metrics are examined, focusing on P1-to-P4 changes.Data are composed from reported weekly measurements into period-site-season means.These means, computed for atmospheric concentration, Vd, PR, and dry, wet, and total deposition, are used to examine differences between 5-year periods for seasons, sites, and predefined regional groupings of sites.The approach to data analysis uses the mean squared error derived from analysis of variance of mean estimates for each pollutant metric to examine differences in the monitoring data and metrics derived from them.
Estimated SO 2 emissions for P4 are 50 % lower than for P1.This is accompanied in each regional site grouping in each season and over all seasons by significant reductions in atmospheric SO 2 concentration and dry deposition, atmospheric aerosol SO 4 concentration, atmospheric S concentration and dry deposition, wet S (except in winter), and total S.For atmospheric SO 2 concentration and dry deposition, the numerically largest significant seasonal absolute reductions usually occur in winter when atmospheric SO 2 concentration is at its seasonal maximum.Analogous behavior occurs in summer for atmospheric aerosol SO 4 concentration, dry SO 4 , and wet S, when atmospheric aerosol SO 4 concentration is at its seasonal maximum.In the east, significant reductions of 48 %, 33 %, and 45 % for atmospheric concentration of SO 2 , aerosol SO 4 and atmospheric S, and of 49 %, 32 %, and 40 % for dry, wet, and total deposition of atmospheric S are associated with the 50 % P1-to-P4 reduction in estimated SO 2 emissions.Consistent with earlier findings (SS07b), results suggest that at the current scale, the responses of both atmospheric SO 2 concentration and dry deposition to changes in SO 2 emissions are close to 1 : 1 and the relationship between changes in the concentration of the secondary reaction product, aerosol SO 4 , and that of the primary pollutant, SO 2 , appears to be less than 1 : 1.The strong P1to-P4 reduction of SO 2 emissions (50 %) is accompanied by increased apparent conversion of SO 2 to aerosol SO 4 , possible increased relative contribution of wet vs. dry deposition as a sink for S, increased fraction of deposited S emissions, and decreased fraction of exported S emissions.
In the east, significant P1-to-P4 reductions of 34 %, 8 %, and 24 % for atmospheric concentration of HNO 3 , aerosol NO 3 and OxN, and of 34 %, 29 %, and 31 % for dry, wet, and total deposition of OxN are associated with the 42 % reduction in estimated NO x emissions.Atmospheric aerosol NO 3 concentration in the east displays a significant P1-to-P3 increase (11 %), with the numerically strongest seasonal relative increase (31 %) in winter (SS07b).Recent aggressive reductions of NO x emissions have been accompanied by significant P1-to-P4 and P3-to-P4 reductions of 8 and 17 % in atmospheric aerosol NO 3 concentration over all seasons.In addition, the smaller P1-to-P4 wintertime increase in atmospheric aerosol NO 3 concentration (i.e., 12 % vs. 31 % for P1-to-P3) and the P3-to-P4 wintertime reduction of 14 % suggest that recent continuing NO x emissions controls are having a desirable impact.Over all seasons the numerically largest absolute and relative reductions in atmospheric HNO 3 concentration and dry deposition, atmospheric OxN concentration and dry deposition, and total OxN generally occur in the northeast (the geographical region with the highest P1 NO x emissions density), and seasonally they tend to occur there in summer.In the east, coincident timing and location suggest that aggressive summertime NO x emissions reductions by EGU sources contributed substantially to this observed behavior.The strong P1-to-P4 reduction of NO x emissions (42 %) is accompanied by reduced atmospheric concentration of HNO 3 relative to OxN, and increased contribution of wet OxN relative to dry OxN.
Estimated annual 1990 to 2009 NH 3 emissions are relatively constant, ranging between 3.5 and 3.9 Tg yr −1 (Xing et al., 2013).However, this is not confirmed through monitoring in the current study, because only the atmospheric con-centration of the secondary reduced nitrogen species, aerosol NH 4 , is monitored.In the east, both atmospheric aerosol NH 4 concentration and dry deposition show significant P1-to-P4 reductions (i.e., 26 and 22 %).The numerically largest absolute and relative reductions of atmospheric aerosol NH 4 concentration and dry deposition tend to occur regionally in the midwest and northeast and seasonally in the summer (when atmospheric aerosol SO 4 concentration is at its seasonal maximum and is also experiencing its largest reductions).Although showing a significant (9 %) P1-to-P4 reduction over all in the east, wet NH 4 shows changes at the regional and seasonal scales that are usually not significant.Since the magnitude of wet NH 4 is over seven times larger than dry deposition, the resulting reduction in total NH 4 in the east (11 %) is similar to that of wet NH 4 .
The atmospheric concentration of N, the sum of monitored oxidized and reduced nitrogen species, is dominated in the east by aerosol NH 4 (≈ 67 %) with smaller contributions from HNO 3 (≈ 19 %) and aerosol NO 3 (≈ 14 %).However, a high Vd permits HNO 3 to dominate dry deposition of N (≈ 75 %).Both the atmospheric concentration and dry deposition of N show significant overall P1-to-P4 reductions (26 and 32 %) which tend to be seasonally higher in summer.Wet deposition of OxN and NH 4 are fairly comparable, and their sum (wet N) dominates N (≈ 75 %).The overall P1-to-P4 relative reductions of wet N and total N in the east are 20 and 24 %, with the strongest reductions in summer and in the northeast.
In the east, a 47 % reduction in wet deposition of H + ion is associated with P1-to-P4 reductions of 50 and 42 % in SO 2 and NO x emissions.The relative contributions of reduced NO x vs. SO 2 emissions have increased over time, until currently (P3-to-P4), they are almost equal.Thus, recent aggressive reductions of NO x emissions along with continuing reductions of SO 2 emissions appear to contribute similarly to the significant and substantial reduction of wet deposition of H + ion in the eastern US.
Although both atmospheric O 3 concentration and dry deposition in the east show significant summertime P1-to-P4 reductions of 12 % each, significant wintertime respective increases of 9 and 19 % lead to overall significant reductions of 4 % each.The numerically largest seasonal and regional reductions of both atmospheric O 3 concentration and dry deposition (17 and 15 %) occur in summer in the northeast, coinciding with aggressive O 3 -season NO x emissions controls during P3 and P4 in the region and upwind in the east.
Although conditions in the east permitting the formation of aerosol NO 3 may be limited by the availability of NH 3 , this appears to be diminishing over time.Increasing NI between P1 and P4 suggests that an acidic aerosol exists in the east, but is approaching neutralization due to increased relative availability of NH 3 associated with reductions of aerosol SO 4 concentration.Widespread wintertime numerical increases in the atmospheric concentrations of both aerosol NO 3 and CASTNET PM are present between P1 and P3, despite reductions in estimated emissions of both SO 2 and NO x (SS07b).Currently, this behavior appears to be largely reversed, and this reversal is associated with continuing reductions of SO 2 and NO x emissions between P3 and P4.Our findings suggest that additional P3-to-P4 reductions in emissions of SO 2 , and especially NO x , have made progress in altering the chemical regime of the wintertime eastern US atmosphere so that future emissions reductions and their resulting reductions in aerosol concentrations may no longer be accompanied by sub-linear changes (or actual increases) in CASTNET PM.
Monitoring data for several atmospheric species not collected by CASTNET, including NH 3 , NO, NO 2 , PAN, other oxidized organic nitrogen species, and NO y , as well as aerosol size distributions would have facilitated data analysis and interpretation.It is recommended that addition of these determinations be considered in future monitoring network upgrades.Uncertainties in estimates of many species' deposition velocities and emissions (especially NO x ) argue strongly for their continued improvement.

Figure 2 .
Figure 2. CASTNET monitoring sites and geographical regions considered in the current study (see Table1).

Figure 6 .
Figure 6.Site-and region-specific total deposition (kg (ha yr) −1 , left -upper bar, P1; and lower bar, P4) and P1-to-P4 relative changes (%, right and bottom) for (a) Total S: total dry plus wet oxidized sulfur deposition; (b) Total OxN: sum of dry plus wet deposition of monitored oxidized nitrogen; (c) Total N: sum of total dry plus wet deposition of monitored oxidized plus reduced nitrogen; and (d) Wet H + : wet deposition of H + .Significance level of period-to-period relative changes indicated by bar fill: black, p ≤ 0.05; and open, p>0.05 (NS).Sites are identified by region in the same order as shown inTable 1.

Figure 8 .
Figure 8. Site-and region-specific wintertime CASTNET PM concentration (µg m −3 , left -upper bar: early period; and lower bar: later period) and period-to-period relative changes (%, right and bottom) for (a) P1-to-P3, W; (b) P1-to-P4, W; and (c) P3-to-P4, W. Significance level of period-to-period relative changes indicated by bar fill: black, p ≤ 0.05; and open, p>0.05 (NS).Sites are identified by region in the same order as shown in Table 1.

Table 1 .
Summary of characteristics of 34 CASTNET monitoring sites providing data in current study (see Fig.2) listed by region.

Table 3 .
Mean (CV) atmospheric concentration, deposition velocity, precipitation rate, dry, wet, and total deposition, and concentration and deposition metrics at 34 eastern US CASTNET and selected subsets of sites for the 5-year period 2005-2009 (see Table2for definitions).