4.1. Spatial Patterns in Hg Deposition
Spatial patterns of precipitation-Hg deposition in the study area were examined with statistical analysis of the weekly data and with isopleth maps of annual data. Statistical analysis revealed a spatial pattern of differences in weekly precipitation-Hg deposition. During 2001–2016, Hg deposition at IN21 (median 196.2 ng/m
2/week, 13-year record) was significantly higher (
p < 0.001) than that at IL11, IN20, IN34, MI26, MI52, OH02 and OH52 (
Figure 3,
Table 8). The sites in Indiana differed according to the same spatial pattern of the study area in that the southeastern site IN21 was significantly higher (
p = 0.001) than the northeastern site IN20 and the northwestern site IN34. In contrast, Hg deposition at 7 sites located in the central and southern part of the study area was not significantly different—IN21, IN22, IN26, IN28, IL63 and KY10. The site with the highest median weekly precipitation-Hg deposition was IL63 with 198.3 ng/m
2/week, which was significantly different from 4 sites during 2014–2016—MI26, MI52, OH52 and OH02. Some of the statistical differences among sites in the study can be attributed to the temporal changes discussed in the next section and comparisons limited to the years with complete annual records for each site (
Table 3). Annual litterfall-Hg deposition (
Table 7) and estimated atmospheric-Hg deposition (sum of litterfall-Hg and precipitation-Hg deposition) for 2007–2016 were not significantly different among 7 sites in the study area or among 5 Indiana sites (
p = 0.25 and
p = 0.54, respectively).
Isopleth maps of annual precipitation-Hg deposition were prepared for 2005–2009 and 2014–2016 (
Figure 4). These 12 years had a sufficient number of sites with complete annual records to support spatial grids of Hg-concentrations to map the study area. Color coding of annual Hg deposition on the maps indicates a recurring spatial pattern similar to the one revealed by the statistical analysis. High deposition in southeastern Indiana near IN21 was present each year, frequently extending to southern Illinois. Areas with low annual Hg deposition were mapped in Michigan and Ohio for many years and frequently included part of northern Indiana.
The isopleth maps show apparent spatial detail in precipitation-Hg deposition corresponding to the detail from the PRISM precipitation depth data, rather than the Hg concentration data but this is not an anomaly from the method. Although Hg deposition is a product of Hg concentration and precipitation depth, the data from this study area indicates that weekly Hg deposition is more highly correlated with precipitation depth (
rho 0.833,
p < 0.001) than Hg concentration (
rho 0.249,
p < 0.001). Similarly, linear regressions indicate more of the variability in weekly Hg deposition is explained by precipitation depth (coefficient of determination
r2 0.58,
p < 0.001) than by Hg concentration (
r2 0.04,
p < 0.001). A supporting observation is that sites in the southern part of the study area (IN21, IN22, IN28, IL63 and KY10) had significantly higher weekly precipitation depths (
p < 0.001) than sites in the northern part of the study area (IN34, MI52 and OH52). The substantial influence of precipitation depth on precipitation-Hg deposition has been reported by other research [
10,
16,
51,
52,
53].
The distribution of Hg emissions sites (
Figure 1), particularly those releasing 180 to 361 kg/year of Hg, is not uniformly related to the spatial pattern of annual precipitation-Hg deposition, although the density of these emissions sites appears to coincide with the area of high Hg deposition mapped for southern Indiana. A simple comparison of the magnitude of annual precipitation-Hg deposition with annual Hg emissions from stationary sources in Indiana was made using data from
Table 1 and
Table S1. If the mean statewide precipitation-Hg deposition of 11.8 µg/m
2/year based on data from IN21, IN22 and IN34 in 2014 was uniform to the land area of Indiana (9.43 × 10
10 m
2), the annual precipitation-Hg load statewide would have been 1094 kg, which is 60% of the 1852 kg annual Hg emissions in the state.
4.2. Extreme Episodes of Hg Deposition
A potential reason that southern Indiana consistently showed statistically and spatially high precipitation-Hg deposition is the frequency of extreme episodes of weekly Hg deposition in the study area, 2001–2016. Extreme episodes were defined as the top 0.5% of values (
Table 9). Extreme Hg-deposition episodes occurred at 8 of the 13 sites in 12 of the 16 years with annual Hg deposition records and 13 of the 27 episodes were recorded for IN21. Extreme Hg-deposition episodes occurred in all seasons but most were between April and September. Only two pairs of extreme episodes occurred in the same weeks, 26 April to 3 May 2011 at IN21 and KY10 and 1 to 8 April 2014 at IN21 and IN22. Notably, the Hg deposition, Hg concentration and precipitation depths were similar among each pair, indicating the potential spatial extent of some extreme episodes.
Weekly Hg-deposition in the extreme episodes was 5 to 10 times higher than in most weekly samples. The episodes ranged from 1640 to 2602 ng/m
2/week, compared with a median of 145 ng/m
2/week and a 75th percentile of 308 ng/m
2/week for all sites for all years. The extreme episodes supplied 10 to 17% of the annual precipitation-Hg deposition in a single week but the episodes did not occur more than twice in one year at any site. Hg concentrations ranged from 7.1 to 45.4 ng/L with a median 14.9 ng/L. The median weekly Hg concentration for all samples from the study area was 9.5 ng/L, which is 36% lower. For these extreme Hg deposition episodes, precipitation depths ranged from 37.3 to 265 mm with a median 120 mm. The median weekly precipitation depth for all samples from the study area was 17.0 mm, which is 61% lower. Weekly precipitation-Hg deposition amounts for the extreme episodes in the study area and at IN21 alone were not correlated significantly with precipitation depth or Hg concentration and variability in Hg deposition was not statistically explained by either factor. The data indicate that high precipitation depth and relatively high Hg concentration contributed to each extreme episode (
Table 9).
Modeled 48-h back trajectories of the hourly positions of the air masses for five extreme episodes of precipitation-Hg deposition at IN21 during 2009–2015 illustrate a further explanation for the high precipitation-Hg deposition (
Table 9,
Figure 5). The back trajectories indicate that air masses predominantly came from the south and southwest. Locations and annual Hg emissions for stationary sources overlain on the back-trajectory map show that numerous sources emitting from 23 kg to more than 300 kg Hg per year were in the path of these air masses. This analysis suggests that local and regional sources, rather than exclusively continental or global Hg emissions likely contributed to the extreme episodes and at least in part, to the spatial patterns of precipitation-Hg deposition in the study area. Other research attributed substantial contributions from local and regional Hg emissions sources to precipitation-Hg deposition in the Midwestern USA [
19,
20,
54,
55]. According to Holmes et al. [
56] and Kaulfus et al. [
57], warm weather, convective thunderstorms are believed to increase Hg deposition and this may be the case for many of the extreme episodes examined in this study. Such storms can incorporate additional GOM formed in the troposphere from GEM originating at local and regional sources [
55], in addition to effective below-cloud and in-cloud scavenging of GOM, as described by Lynam et al. [
58].
4.3. Temporal Changes in Hg Deposition
Temporal changes in Hg deposition can be observed in the study area during the 16-year study period. The Mercury and Air Toxics Standards (MATS) rules were finalized by U.S. Environmental Protection Agency for Hg emissions from power plants in 2011, effective in 2015 [
59]. Rules for other Hg emissions sources such as cement kilns, utility boilers and medical waste incinerators were finalized in 2010, 2011 and 2013, respectively [
60,
61,
62]. The expected outcome was a succession of reduced Hg emissions from these sources up to and after the effective dates.
To examine whether a temporal trend was evident in precipitation-Hg in response to these rules, the weekly data were grouped by year 2001–2012 and 2014–2016. As grouped, Hg deposition and Hg concentration were shown to be significantly lower in 2014–2016 than 2001–2013 (
p < 0.001). The median weekly Hg deposition in 2001–2013 was 152 µg/m
2/week, compared with 123 µg/m
2/week in 2014–2016, a 19% difference and the median weekly Hg concentration was 9.9 ng/L compared with 8.3 ng/L, a 16% difference. Data for Indiana showed a similar response. Hg deposition and Hg concentration were significantly lower in 2014–2016 than 2001–2012 (
p = 0.038 and
p < 0.001, respectively). The median weekly Hg deposition in 2001–2012 was 161 µg/m
2/week, compared with 135 µg/m
2/week in 2014–2016, a 16% difference and the median weekly Hg concentration was 10.2 ng/L compared with 8.3 ng/L, a 19% difference. Consistent with other analysis for the study area, weekly Hg concentrations grouped 2001–2009 (median 10 ng/L) and 2010–2012 (median 9.6 ng/L) were significantly higher (
p < 0.001) than 2014–2016 (median 8.3 ng/L). Weekly precipitation depth between these two time periods was not significantly different in the study area (
p = 0.331) or in Indiana (
p = 0.469), meaning the decreases in Hg deposition were related primarily to decreases in Hg concentration rather than decreases in precipitation depth. These data document a gradual decline in Hg concentrations in advance of the period when Hg emissions reductions were required. This observation is consistent with model results by Zhang and Jaegle [
63] for 2005–2010 in the Midwestern USA.
Wetherbee et al. [
64] analyzed trends in annual precipitation depths at a separate group of NADP monitoring sites for 2001–2015. They reported increases in precipitation at 17 of 21 sites in the study area that were statistically significant and higher than 1 cm/year at 3 of the 17 sites. This information suggests the potential for some decreases in Hg concentrations to involve dilution by increased precipitation. Typically, increased precipitation depth would cause a corresponding increase in Hg deposition but that response was not identified in the statistical analysis. Precipitation-Hg concentrations and Hg deposition exhibited a decline during the study period.
The findings from this study in the Midwestern USA indicate that reductions in Hg emissions associated with implementation of the MATS and related rules coincided with temporal decreases in precipitation-Hg concentrations, resulting in decreased atmospheric Hg deposition. Meanwhile, precipitation depths did not significantly change. At the minimum, it can be postulated that reductions of atmospheric GOM and PBM in the study area, the species most readily scavenged from the air during precipitation, were associated with the reduced precipitation-Hg concentrations and deposition. Reductions of atmospheric GEM, the most abundant species and the one considered to dominate litterfall-Hg deposition, were inconclusive based on these Midwestern data. Annual litterfall-Hg deposition, litterfall-Hg concentrations and litterfall mass at 7 sites were not significantly different in 2007–2009 compared with 2012–2016. A study by Risch et al. [
18] reported litterfall-Hg concentrations in the eastern USA were significantly lower in 2014–2016 than 2007–2009, consistent with reported declines in atmospheric GEM reported by others. The inconsistency of Midwestern litterfall-Hg data with the broader region underscores the value of the spatial focus on five states afforded by this study.