Final report Atmospheric Deposition of Mercury and …...Final report Atmospheric Deposition of...

Final report

Atmospheric Deposition of Mercury and Trace Metals to the Pensacola Bay Watershed

Jane M. Caffrey

Center for Environmental Diagnostics and Bioremediation

University of West Florida

William M. Landing

Department of Oceanography

Florida State University

Sikha Bagui

Department of Computer Science


Subhash Bagui

Department of Mathematics and Statistics


August 11, 2009

2

Introduction Atmospheric concentrations of many compounds including nitrogen, mercury, copper,

zinc, sulfur, and polyaromatic hydrocarbons have increased since the industrial revolution due to combustion of fossil fuels, industrial and manufacturing processes and agricultural operations. Deposition of atmospheric constituents to land and water surfaces occurs at a wide range of scales, from less than a kilometer to hundreds of kilometers from the emission sources. Thus, airsheds can encompass huge and often amorphous (depending on climate conditions) areas, usually 10 to 100 times greater than watershed area (Valigura 2001). Atmospheric deposition represents a significant source of nutrients and other contaminants in many watersheds (Valigura 2001). Deposition of these compounds onto land and water surfaces has a negative impact on plant and animal communities, ecosystem function and human health. In addition, accumulation of toxic contaminants in animal tissues can lead to reproductive failures. Atmospheric deposition represents 50-90% of Hg load to US waters (http://nadp.sws.uiuc.edu/). Accumulation of toxic compounds such as mercury, selenium and organic chemicals in food chains increases human health risks when fish or ducks are consumed. 42 states currently have mercury consumption advisories for freshwater fish while the Southeast and Gulf coasts have coastal fish advisories (http://nadp.sws.uiuc.edu/).

Asia has the highest mercury emissions, approximately 46% of worldwide emissions, with Europe and North America contributing between 15-16% (Pirrone et al. 1996). Eastern Europe and the former republics of the USSR represent 13% of worldwide emissions, while Africa and Oceania are 5% or less (Pirrone et al. 1996). Mercury emissions to the atmosphere are primarily associated with coal burning power plants and waste incinerators (Pirrone et al. 1996, Dvonch et al. 1998). Mercury in the atmosphere occurs in 3 different forms: elemental gaseous mercury (Hg0), reactive gaseous mercury (RGM or Hg2+) which is the oxidized form, and mercury associated with particles (HgP). Reactive gaseous mercury is usually deposited near the source of the emission (Dvonch et al. 1998, Lin and Pehkonen 1999), while elemental gaseous mercury has a residence time in the atmosphere of 1-2 yr, thus deposition may be worldwide (Lin and Pehkonen 1999). Reactive gaseous mercury can be adsorbed directly onto plant and soil surfaces, as well as associated with soot, fog droplets or rain drops (Lin and Pehkonen 1999).

The growth of monitoring networks (e.g. National Atmospheric Deposition Program (NADP)) since 1978 has provided an increasingly detailed picture on atmospheric deposition at the national and regional level. The National Atmospheric Deposition Program (NADP) including the Mercury Deposition Network (MDN) is a national network of collaborating state and federal agencies to examine long-term trends in atmospheric deposition at regional scales. The regional and national trends provide critical information necessary for determining policies to improve air and water quality at the national level. However, in order to improve water quality at the watershed level, the importance of local sources in addition to regional sources needs to be determined.

Our primary objective was to measure mercury, trace metals and major ions in rainfall over annual cycles. We evaluated the temporal and spatial patterns in atmospheric deposition by comparing the results among 3 sites in the Pensacola Bay watershed and comparing them with data collected at MDN and NADP sites in the Southeast Region along the Gulf Coast. Our primary research question was what is the spatial distribution of mercury, trace metal and nutrient deposition in the Pensacola Bay watershed?

3

This research was conducted as part of the Partnership for Environmental Research and Community Health in order to understand the contribution of mercury and trace metals from atmospheric sources to the Pensacola Bay region. Air quality in Pensacola Bay area is affected by emissions from industrial sources, automobiles and other vehicles as well as other stationary area sources such as dry cleaners, gas stations, agriculture, and construction sites (FDEP 1998). Industrial sources include a coal fired electric generating plant, a paper mill, chemical and other manufacturing companies (EPA website: http://www.epa.gov/air/data/reports.html). NOx and SO2 emissions are dominated by power generation, although automobile and other vehicles are also a large source of NOx. The major source of mercury emissions in 2002 was from Plant Crist, the coal-fired electric generating plant, which was estimated to release approximately 183 lbs. Other mercury emission sources in the Pensacola Bay watershed include coal-fired industrial boilers at the International Paper Plant, medical waste incineration and landfills (http://www.epa.gov/air/data/ntiemis.html?co~12033~Escambia%20Co,%20Florida). Nearby Mobile and Escambia Counties in Alabama have significant sources of mercury emissions, approximately 780 and 880 lbs respectively in 2002 (http://www.epa.gov/air/data/repsco.html ?co~01053%2001097~Escambia%20Co%2C%20Mobile%20Co%2C%20Alabama).

In addition to the atmospheric deposition study, we conducted two targeted studies: one to examine historic levels of mercury in the watershed and the second to examine mercury levels in Woodbine Springs, a small lake in the Pensacola Bay watershed that has previously reported to have high water column mercury levels. Understanding historical inputs and how source inputs have changed with time provides valuable context for evaluating metal contamination in aquatic environments. Contaminants deposited over long periods can be resuspended, mobilized, and reintroduced into food webs. Studies in the San Francisco Bay, USA, Minamata Bay, Japan, and Gulf of Trieste, Italy, have shown peak deposition of mercury associated with industrial activities such as gold mining or smelting (Hornberger et al. 1999, Tomiyasu et al. 2000, Covelli et al. 2001). Despite the elimination of the industrial activities at these locations, mercury levels remain high in Minamata Bay and the Gulf of Trieste and contaminated surficial sediments continue to be transported throughout the region.

Methods

The sampling design and analytical program were designed to accomplish the following goals:

1. Collect daily (24-hour integrated) rain samples from three sites in the Pensacola area to quantify the wet deposition of mercury in the greater Pensacola area on time scales from individual rain events to annual deposition.

2. Choose sites that will enable some quantification of rainfall mercury concentrations upwind and downwind of known and suspected atmospheric mercury emission point sources.

3. Analyze other trace elements and major ions and use multivariate statistical methods in an effort to identify and quantify contributions to mercury deposition from known or suspected mercury emission point sources. This process works best on event rain samples where the air mass trajectory contributing to the chemistry of that rain event can be evaluated. In general, there is not more than one significant rain event in any given 24-hour period.

4. Collect a sediment core to enable us to examine historic levels of mercury deposition in the Pensacola Bay watershed.

4

5. Collect water samples for mercury analysis from Woodbine Springs and Woodbine Lake to compare with previous samples from the Lake Watch Program We identified three sampling locations (Figure 1) which were easily accessible, but

secure from vandalism. Data loggers with rain gauges were set up at each site in September 2006 to determine the timing of the rain events and how effective the samplers were at capturing low rainfall amounts.

The first atmospheric deposition collector was installed on November 17, 2004 at Ellyson Field (30.5256 °N, 87.2033 °W). This site is located on the Florida Department of Environmental Protection (FL DEP) property. FL DEP is monitoring ozone, SO2, PM2.5, wind speed and direction at this location. This site is generally downwind from the Crist coal-fired power plant, one of the largest suspected atmospheric mercury point sources in the Pensacola Bay watershed. We have access to other air quality information being collected at this location.

The Pace site was installed on Dec 9, 2004 on the Pace Water Systems property on Woodbine Road in Santa Rosa County (30.6023°N, 87.1912 °W). Part of this property is used as a treatment wetlands and the rest is a nature preserve. The collector was installed away from the treatment wetlands in an open field adjacent to a pond. The site is east northeast of the power plant.

The Molino site was installed on December 15, 2004 in an open field at a private residence (30.6903°N, 87.4103 °W). It is northwest of the power plant and the paper mill. The collector at the Ellyson site was powered by AC, while the Pace and Molino sites were powered by 12V batteries.

Rain samples were collected within 24 hours of 0.5 cm rain events using Aerochemetrics or Loda rain collection devices (Figure 2). Standard EPA protocols for data collection and analysis were used during this study. Duplicate bottles for major ion analysis were collected for 10% of the deployments. For the rest of the deployments (90%), duplicate bottles for trace metal and Hg analysis were collected. The sampling in 2005 was disrupted for 2 weeks by Hurricane Dennis; collectors were removed from the sites to keep them from being damaged or destroyed.

5

Figure 1 – Map of Pensacola Bay showing sampling locations: Ellyson, Molino and Pace for atmospheric deposition collection, Snag Lake for sediment core collection, and Woodbine Springs for water collection. Mercury emission sources: Crist Plant, International Paper, Sacred Heart Hospital medical waste incinerator

Molino

Pace

Ellyson

InternationalPaper

Crist plant

Sacred Heart Hospital

Snag Lake

Woodbine Springs

6

Figure 2 – Photograph of rain collection site showing rain collector, datalogger, tipping bucket, temperature and humidity probes (left panel) and Teflon and polypropylene bottles inside collection bucket (right panel) Mercury Analysis

Rain samples (in FEP Teflon bottles) were acidified after collection to 0.045 M HCl plus 0.048M HNO3, then placed in a low wattage UV digestion box for at least 48 hours to completely solubilize the collected Hg. The digested samples were analyzed by CVAFS using EPA method 1631 using a Tekran 2600 Mercury Analyzer at Florida State University. Trace metal analyses

Rain samples in the replicate Teflon bottles (acidified to 0.045M HCl plus 0.048M HNO3) were used for the trace element analyses. Samples were analyzed for total metal concentrations using a Finnegan Element-I high resolution magnetic sector ICP-MS at Florida State University. Major ions and pH

The pH of the samples was measured in the laboratory immediately following collection. The major ions, nitrate (NO3

-), ammonium (NH4+), phosphate (PO4

3-), sulfate (SO42-), sodium

(Na+), chloride (Cl-), magnesium (Mg2+), calcium (Ca2+), and potassium (K+) were measured in polyethylene sample bottles. Major ion analyses were conducted using a Dionex DX500 ion chromatograph using EPA method 300.0 for anions and Dionex protocols (Application Note 141, http://www1.dionex.com/en-us/webdocs/application/industry/chempetro/AN141_V15.pdf) for cations at the University of West Florida. Sediment Core Sample Collection and Analysis Sediment cores were collected with a polycarbonate push core to preserve the sediment-water interface. Cores were sectioned in the field immediately after collection at 1 cm intervals in the upper 15-20 cm of each core, increasing to 2 cm intervals below this depth. Sub samples of each core section were taken for geochemical, radiochemical, and mercury analysis using appropriate ”clean hands/dirty hands” sample handling techniques. Once sectioned, samples were stored on ice in the dark for transportation to the lab. Stored frozen in the laboratory, all samples were thawed, weighed wet, dried, weighed dry, and homogenized using a mortar and pestle. Samples for total Hg were dried in aluminum weighing boats placed in a drying oven at 60° C for 36 hours. Samples for radionuclide and ancillary analysis were freeze-dried. Sub samples were refrigerated for grain size analysis. Sample preparation for radiometric dating includes tightly packing polypropylene vials (84 mm H x 14.5 mm OD) with homogenized

7

sediment to a uniform height of 30 mm. Net sediment weight and height were recorded prior to sealing each vial at the sediment surface with clear, 2-part epoxy resin. Geochemistry Fraction dry weight was calculated as the dry mass per wet mass. Organic matter content was determined on dry sediment by loss on ignition (%LOI) at 550° C for 2 hours in a muffle furnace. Grain size analysis was determined according to the method of Folk (1974). Ashed sediments were extracted with 1N HCl (Wigand et al. 1997) for 24 h. Extracts were analyzed for inorganic phosphorus using standard methods (Parsons et al. 1994). Sample preparation for radiometric dating includes tightly packing polypropylene vials (84 mm H x 14.5 mm OD) with homogenized sediment to a uniform height of 30 mm. Net sediment weight and height were recorded prior to sealing each vial at the sediment surface with clear, 2-part epoxy resin. Samples equilibrated for 3 weeks to allow in-growth of the radon daughters, 214Pb and 214Bi. 210Pb (t1/2 = 22.3 y), 226Ra (t1/2 = 1620 y), and 137Cs (t1/2 = 30.2 y) were measured using low background well-type intrinsic germanium detectors (Schelske et al., 1994). Total 210Pb and 137Cs were determined directly from their respective photon peaks, 46.5 keV and 661.7 keV. Excess (or unsupported) 210Pb was calculated as the difference between total 210Pb and 226Ra (assumed to represent supported 210Pb) and decay corrected to the time of sampling. As part of the 238U decay series, 210Pb eventually results from the decay of 226Ra. Radium resides in the mineral matrix of many rocks where it decays directly to 222Rn by alpha decay. Some radon remains trapped inside the mineral grains where it decays to 210Pb within this microenvironment. A chemically inert gas, radon will migrate within the mineral matrix but will not interact with other elements present. If the radon gas is produced close enough to the surface of the mineral it can escape into the surrounding pore space. When this pore space is occupied by groundwater, radon dissolves and is transported with the flow of the water.

In lake sediments, a fraction of total 210Pb is in equilibrium with 226Ra, thus providing supported levels of 210Pb. Another important component of total 210Pb in lake environments is excess (unsupported) 210Pb which may have several sources: (1) atmospheric deposition; (2) overland wash; (3) diffusion of dissolved radon from lake sediments; and (4) weathering of terrestrial 226Ra. In the atmosphere, 222Rn decays to 210Pb which quickly adsorbs onto aerosols. These aerosols wash out of the atmosphere as both wet and dry deposition and represent the most important source of excess 210Pb to lake surfaces (Benninger 1975). Most other excess 210Pb contributors are minor in comparison. Sediment cores are usually dated using the constant rate of supply (CRS) model (Appleby and Oldfield 1983). Assumptions that must hold true to apply this model to lake sediments are threefold: (1) the flux of excess 210Pb to the lake sediments is constant; (2) variations from exponential decrease in excess 210Pb with depth are dependent on variations in sedimentation rate (mass flux), not the 210Pb flux; and (3) 210Pb decays at a rate governed by first order kinetics. The rate at which 210Pb arrives in the sediments is dependent on both the mass flux of water column particles and the flux of atmospheric excess 210Pb. Given a constant atmospheric flux of excess 210Pb, the CRS model accommodates variable mass flux to the sediments such as might be caused by changes in productivity or in the supply of wind-derived inorganic material. Woodbine Springs Samples

Water samples were collected from Woodbine Springs and Woodbine Lake on July 14, 2008. Replicate Teflon bottles (n=3 Woodbine Lake, n=4 Woodbine Springs) were collected using clean hands/dirty hands techniques for mercury analysis. A sample was also collected for

8

pH, conductivity and major ions. Mercury samples were shipped next day service to FSU where samples were filtered. Filtered and unfiltered samples were run using standard protocols described above. Cations and anions were analyzed as described above. Total gaseous mercury analysis

A total gaseous mercury analyzer (from EPA Region IV) was installed at Ellyson Field between February 2005 and May 2006. Readings were integrated over 5 minute intervals. The permeation chamber was calibrated on several occasions during the study. The results are not reported here due to quality control problems in the analyses due to running the instrument without a soda-lime moisture trap. Calculations and statistics

All deposition fluxes are calculated by multiplying the measured concentrations (grams per cubic meter) by the rainfall amounts (meters per event), and are expressed as mass per square meter (per event). We tested for differences between sites, seasons and years using ANOVA. We used the event based data to compare our three sites. A correlation analysis of mercury, trace metals and major ion fluxes was performed on the data. In addition, we compared our sites with Mercury Deposition Network (MDN) and National Atmospheric Deposition Program (NADP) sites in central Northern Gulf of Mexico region (Table 1, NADP 2009) using ANOVA to test for differences among sites. For this comparison, we summed our data over the same weekly intervals as the NADP and MDN sites. The level of significance equal or less than 0.05 was used for these analyses.

Table 1 – NADP and MDN sites used to compare with Pensacola Bay data Site Location Latitude Longitude LA30 (NADP) 30.7819 °N 90.2021 °W LA 12 (NADP) 29.931 °N 91.7165 °W LA 28 (MDN) 30.5031°N 90.3769 °W MS22 (MDN) 30.985 °N 88.9319 °W AL 24 (NADP, MDN) 30.4746 °N 88.1411 °W AL02 (NADP, MDN) 30.7905 °N 87.8497 °W FL14 (NADP) 30.5486 °N 84.6004 °W FL23 (NADP) 30.1106 °N 84.9902 °W

Results

Over the course of this study from November 19, 2004 through January 2, 2008, we

collected 565 rain samples from 3 sites. These samples represent about 225 separate rain events in the Pensacola Bay watershed.

We installed rain gauges with the dataloggers at the sites in order to have an independent check on the rainfall amounts. The comparison for 2007 shows that the volume in the bottles is equivalent to the rain gauge as long as the rainfall amounts were less than 130 mm (Fig. 3). At 130 mm, the one liter bottles are full. In the three years of sampling, we had 1-2 rain events per year that exceeded 130 mm.

9

Figure 3 – Comparison of rainfall amounts from rain gauge at sites and the amount of rain collected in sample bottles. The regression line excludes amounts greater than 129 mm.

FSU participated in two mercury round robin inter-laboratory comparisons organized by FDEP for low level mercury samples. The analyses performed by FSU had excellent precision and accuracy, with FSU results coming out very close to the FDEP lab and Frontier Geosciences, two laboratories which have set the standards for mercury analysis (Figure 4, Table 2). Table 1 shows the inter-laboratory comparison between our results and FDEP as well as Frontier Geosciences. As can be seen, both inter-laboratory comparisons show our reproducibility and consistency with other laboratories.

Figure 4 – Round robin results for low level Hg analysis. FSU results are circled

y = 0.99xR2 = 0.92

0

50

100

150

200

250

300

350

0 20 40 60 80 100 120 140

Rain from weight, mm

data

logg

er ra

in m

m

EllysonMolinoPace

10

Table 2: Comparison of FDEP, FSU and Frontier Geosciences blank corrected total Hg concentration results.

Field ID Matrix FDEP (ng Hg/L)

FSU (ng Hg/L)

Frontier Geosciences (ng Hg/L)

Round Robin Average

(ng Hg/L) #17 W-Rain 2.3 2.79 #19 W-Rain 2.4 2.78 #32 W-Rain 4.4 4.81 #35 W-Rain 4.3 4.88 #80 W-Rain 2.2 2.17 #188 W-Rain 2.4 2.16 #87 W-Rain 3.4 3.76 #44 W-Rain 3.4 4.03 #7 W-Rain 3.4 5.53 #21 W-Rain 4.0 4.7

Lab QC Blank

W-QC-BLK 0.1 0.58

Lab QC Blank

W-QC-BLK 0.1 0.68

ENR012 Everglades Water

0.68 0.90 1.02 0.82

G379D Everglades Water

0.94 1.17 1.29 1.1

S5A Everglades Water

2.10 2.37 2.59 2.4

Mercury and Trace Element Concentrations in Rainfall

Concentrations of mercury and trace metals such as arsenic, selenium and aluminum were generally higher in low volume samples (Fig 5). With larger storms, trace element and major ion concentrations are diluted leading to a washout effect. This effect was most obvious for mercury and selenium and less apparent for aluminum.

Mercury deposition showed no site differences between the three Pensacola Bay watershed sites (Fig. 6). Mercury fluxes from individual events ranged between 13 and 2050 ng/m2. There were significant seasonal differences. The highest mercury deposition occurred in the summer (403 +19.9 ng/m2/event) compared to other seasons (fall 286 +18 ng/m2/event, winter 245+15 ng/m2/event, spring 333+24 ng/m2/event). Interannual differences were significant with the highest average deposition in 2007 (400.7 ng/m2/event) and the lowest in 2006 (286 ng/m2/event).

Arsenic (As) and tin (Sn) had no consistent differences among the different sites (Figure 7, Table 3). Average arsenic flux was 4.0 µg/m2/event and values ranged between 0.15 and 22.1 µg/m2/event. The average tin flux was 2.4 µg/m2/event with a range of 0.1 and 247 µg/m2/event. In contrast, selenium was different among the sites, with significantly higher deposition at Pace (7.03 µg/m2/event) than at either Ellyson (3.96 µg/m2/event) or Molino (3.65 µg/m2/event) (Table 3). Most of the trace metals had similar deposition at difference sites, except for copper, chromium and zinc (Table 3). Copper and chromium deposition rates were significantly

11

higher at Ellyson than at Molino. Zinc deposition was significantly higher at Ellyson than at either Molino or Pace.

Figure 5 – Mercury (Hg ng/L), arsenic (As µg/L), selenium (Se µG/L), and aluminum (Al µg/L) concentrations versus the volume of rain (mL) collected in sample bottles at Pensacola Bay sites.

Figure 6 – Mercury (Hg) deposition (ng/m2/event) at Ellyson (brown), Molino (orange) and Pace (green) between November 2004 and December 2007 for individual rain events (left side). Box plot (right side) for each site, showing median, 25th and 75th percentiles (box boundary), 10th and 90th percentiles (whiskers), and 5th and 95th percentiles (circles).

Ellyson Molino Pace11/04 05/05 11/05 05/06 11/06 05/07 11/07

Hg

Flux

ng/m

2

10

100

1000

10000

0

10

20

30

40

50

60

0 500 1000 1500Rain volume mL

Hg

conc

entra

tion

ng/L

EllysonMolinoPace

00.20.40.60.8

11.21.41.61.8


As

conc

entra

tion

µg/L

EllysonMolinoPace

0100200300400500600700800900


Se c

once

ntra

tion

µg/L

EllysonMolinoPace

00.20.40.60.8

11.21.41.61.8

2


Al c

once

ntra

tion

µg/L

EllysonMolinoPace

12

Figure 7 – Deposition of arsenic (As), selenium (Se) and tin (Sn) (µg/m2/event) as in Fig. 6 except outliers indicated by circles

11/04 05/05 11/05 05/06 11/06 05/07 11/07

Sn

Flux

µg/

m2

0.01

0.1

1

10

100

1000

Ellyson Molino Pace

11/04 05/05 11/05 05/06 11/06 05/07 11/07

Se

Flux

µg/

m2

0.001

0.01

0.1

1

10

100

1000

Ellyson Molino Pace

10/04 04/05 10/05 04/06 10/06 04/07 10/07

As

Flux

µg/

m2

0.01

0.1

1

10

100

1000

Ellyson Molino Pace

13

Table 3 – Average deposition of trace metals in rain events at three sites in Pensacola Bay. Average and Standard error (SE). ANOVA test of site differences, p values. NS = non significant

Ellyson SE Molino SE Pace SE ANOVA Aluminum µg/m2/event

1251 114 992 104 1120 69 NS

Antimony ng/m2/event

998 76 830 77 918 65 NS

Arsenic µg/m2/event 4.38 0.46 3.60 0.38 3.86 0.22 NS Barium µg/m2/event 21.13 1.40 17.30 1.66 20.52 1.40 NS Bismuth ng/m2/event 49.94 5.06 41.86 4.57 46.85 4.07 NS Cadmium ng/m2/event

143.2 26.9 120.2 17.5 189.1 16.1 NS

Cobalt µg/m2/event 0.62 0.05 0.51 0.05 0.58 0.03 NS Copper µg/m2/event 7.24 0.52 4.56 0.42 5.93 0.47 p = 0.007 Chromium µg/m2/event

2.51 0.26 1.53 0.12 1.96 0.11 p 0.003

Cesium ng/m2/event 120.34 11.07 105.15 12.50

118.90 7.31 NS

Gallium ng/m2/event 1023 106 885 85 955 51 NS Iron µg/m2/event 603 54 485 55 550 36 NS Lithium ng/m2/event 1351 149 1073 109 1141 69 NS Magnesium µg/m2/event

2473 613 1520 252 1630 100 NS

Manganese µg/m2/event

26.05 2.23 21.66 2.31 25.19 1.66 NS

Nickel µg/m2/event 7.60 0.90 6.25 0.59 7.67 0.49 NS Phosphorus µg/m2/event

106.4 9.8 137.3 20.5 131.9 10.5 NS

Lead ng/m2/event 6249 467 5485 595 5745 402 NS Selenium µg/m2/event

3.96 0.44 3.65 0.42 7.03 0.80 p = 0.03

Silicon µg/m2/event 1833 193 1575 211 1728 122 NS Strontium µg/m2/event

20.33 3.53 13.77 1.80 15.22 0.85 NS

Tin µg/m2/event 1.35 0.14 1.48 0.15 1.42 0.10 NS Vanadium µg/m2/event

8.18 0.96 6.30 0.76 7.55 0.51 NS

Zinc µg/m2/event 51.36 7.59 29.23 3.36 31.91 2.58 p = 0.009

14

Precipitation Chemistry The amount of precipitation was not significantly different among the 3 sites, nor were

there any significant inter-annual differences. However, seasonal differences were significant with the highest rainfall in the spring and fall and the lowest during winter. H+ fluxes were similar among the 3 Pensacola Bay sites, with no consistent differences among the stations (Figure 8). Again there were no significant inter-annual differences, but there were significant seasonal differences with the highest H+ flux in summer and the lowest during winter and fall.

Figure 8 – H+ flux (mole/m2//event) as in Fig 6. Sulfate fluxes were also similar among the different sites and had no significant seasonal

or inter-annual differences (Fig 9). The average sulfate flux was 40.43 mg SO4/m2/event and values from individual rain events ranged from 0.5 to 438 mg SO4/m2/event.

Figure 9 – Sulfate (SO42-) flux (mg SO4/m2/event) as in Fig 6.

There were no significant site-to-site differences in nitrate fluxes (Fig. 10). The average

nitrate flux was 35.6 and values ranged from 0.5 to 285 mg NO3/m2/event. Nitrate fluxes in 2007


SO

42- fl

uxm

g/m

2

0.1

1

10

100

1000 EllysonMolinoPace


H+ fl

ux

mol

es/m

2

1e-7

1e-6

1e-5

1e-4

1e-3

1e-2

1e-1

EllysonMolinoPace

15

(19.5 mg NO3/m2/event) were significantly lower than in 2005 (45.5 mg NO3/m2/event) or 2006 (43.2 mg NO3/m2/event). Nitrate fluxes were also significantly lower during the winter (16.5 mg NO3/m2/event) and fall (26.0 mg NO3/m2/event) compared to spring (56.4 mg NO3/m2/event) and summer (45.1 mg NO3/m2/event).


NO

3- flux

mg/

m2

0.1

1

10

100


Figure 10 – Nitrate (NO3

-) flux (mg NO3/m2/event) as in Fig 6.

Ammonium fluxes were about ten percent of nitrate fluxes and had no site or interannual differences (Fig. 11). Average ammonium flux was 4.35 and values ranged from 0.04 to 46.3 mg NH4/m2/event. The seasonal pattern was the same as nitrate fluxes with higher fluxes in the spring (9.2 mg NH4/m2/event) and summer (5.2 mg NH4/m2/event) and lower fluxes during the fall (1.9 mg NH4/m2/event) and winter (2.5 mg NH4/m2/event).


NH

4+ flux

mg/

m2

0.001

0.01

0.1

1

10

100

EllysonMolinoPace

Figure 11 – Ammonium (NH4

+) flux (mg NH4/m2/event) as in Fig 6.

Chloride fluxes had a larger range than the other major element fluxes, with a few fluxes exceeding 1000 mg/m2/event, particularly at Ellyson (Fig 12). This is likely due to the accumulation of sea salt aerosols in rainwater. The average chloride flux was 58.76 and values ranged from 0.6 to 2,070 mg/m2/event. There were no significant differences in chloride fluxes among the different sites, years or seasons.

16


Cl- fl

uxm

g/m

2

0.1

1

10

100

1000

10000EllysonMolinoPace

Figure 12 – Chloride (Cl-) flux (mg/m2/event) as in Fig 6.

Sodium fluxes also had a large range and generally tracked the chloride concentrations (Fig 13). As observed with the chloride fluxes, sodium fluxes had no significant differences among the different sites, years or seasons. Average sodium fluxes were 24.5 and values ranged from 0.2 to 938 mg/m2/event.


Na+ fl

uxm

g/m

2

0.1

1

10

100

1000


Figure 13 – Sodium (Na+) flux (mg/m2/event) as in Fig 6.

Comparisons with MDN and NADP sites

Mercury fluxes along the Northern Gulf of Mexico were similar among MDN and Pensacola Bay sites (Table 4). There were no interannual differences in mercury fluxes, although seasonal fluxes were significantly different from one another. As we observed with the Pensacola Bay data alone, summer fluxes were significantly higher than fluxes during the other months. Summer mercury fluxes represented an average of 44% of the total annual mercury flux at these sites and could represent as much as 60% for some sites in some years.

Sites in Pensacola Bay area were often higher than NADP sites for H+, sulfate, nitrate, ammonium, chloride and sodium fluxes (Table 4). H+ fluxes at Ellyson and Pace were significantly higher than at all other NADP sites, while fluxes at Molino were significantly

17

higher than FL14 and LA30. Sulfate fluxes at Ellyson were significantly higher than all NADP sites, Pace was significantly higher than NADP sites in Florida and Louisiana, while Molino was significantly higher than the NADP sites in Florida. Nitrate fluxes at all Pensacola Bay sites were significantly greater than NADP sites. Ammonium flux at Ellyson was greater than FL14 and FL23. Ammonium flux at the NADP site LA30 was also greater than those at FL23. Chloride fluxes at Ellyson were greater than those at Molino and all NADP sites, while Pace was greater than the NADP sites in Florida, Louisiana and at AL02. Sodium fluxes were significantly higher at Ellsyon than at all NADP sites. Pace also had significantly higher sodium fluxes than at all NADP sites except for AL24.

There were significant seasonal differences in sulfate, nitrate, ammonium and chloride fluxes with sulfate, nitrate and ammonium having higher fluxes in the spring and summer than the other seasons. Chloride fluxes were highest in the fall. Inter-annual differences were significant for sulfate and nitrate fluxes with higher fluxes in 2005 and 2006 compared to 2007. H+ and ammonium flux declined over the three years of observations. Table 4 – Analysis of variance in mercury and major ion fluxes for Pensacola Bay sites, National Atmospheric Deposition Program (NADP) sites and Mercury Deposition Network (MDN) sites along the eastern Gulf of Mexico. P values reported, NS is non significant. Seasons are spring (Sp), summer (Su), fall (Fa) and winter (W).

Constituent Site Season Year Mercury NS p<0.001 (Su> Sp, Fa & W) NS

H+ p<0.001 p<0.0001 (Su> Sp, Fa & W) p=0.028 ((2005 > 2006 & 2007) Sulfate p<0.001 p<0.0001 (Sp & Su> Fa & W) p<0.0001 (2005 > 2006>2007) Nitrate <0.001 p<0.001 (Sp & Su> Fa & W) p<0.0001 (2005 & 2006>2007)

Chloride p<0.001 p=0.03 (Fa> Sp & Su & W) NS Sodium p<0.001 NS NS

Ammonium p=0.03 p<0.001 (Sp > Su> Fa & W) p<0.001 (2005 > 2006>2007) Historic mercury deposition in the Pensacola Bay watershed The Pb-210 analysis was performed on 10 cores collected from various water bodies in

the Pensacola area. The best sediment accumulation history was in Snag Lake (Fig. 14), with a sedimentation rate of 0.48 cm/year. Mercury concentrations ranged from 60 to 400 ng/gdw with a tripling of concentrations between 1950 and 1970 (Figure 14). Based on the sediment core from Snag Lake, mercury deposition in the watershed increased during a period when industrialization was occurring in the region and nation.

18

Figure 14. Pb-210 profile from Snag Lake, Pensacola in right panel. The linear sediment accumulation rate is estimated to be 0.48±0.03 cm/yr. Mercury concentration in sediment (ng/gdw) in center panel and mercury accumulation in right panel.

Snag Lake sediments were fine grain (silt/clay) with a high water content greater than

60% (Figure 15). The organic content was highest in the 7.5 to 15 cm depth layer which corresponds to the 1980s to 1990s. Phosphorus concentrations were intermittently high during this period, with peaks in sediment phosphorus in the 1970s, early 1990s and again in the late 1990s (Figure 15). This suggests that there were intermittent high phosphorus inputs to the area during this time.

Figure 15 – Water content (%W) of sediment, organic matter content (% LOI) and sediment phosphorus concentration (µmol/g ash free dry weight) in Snag Lake sediment core.

Snag Lake

0

5

10

15

20

25

30

0% 10% 20% 30%LOI

Dep

th

Snag Lake

0

5

10

15

20

25

30

0 5 10sediment P umol/g

Dep

th

Snag Lake

0

5

10

15

20

25

30

60% 80% 100%%W

Dep

th

19

Woodbine Springs water samples Most of the mercury was in the dissolved fraction at both sites (Table 5). Total mercury

was higher in Woodbine Lake (9.2 ng/L) than in the Springs (6.7 ng/L). These values are lower than samples collected in 2004 from Woodbine Lake taken by FWC and Lake Watch, but much higher than the average for Florida Lakes (Lange 2005). Mercury concentrations were also measured in two ponds at the Pace Water Systems property where our atmospheric deposition collector is located and Downey pond which is just south of the Pace Water Systems property. Those values were similar to those found in other Florida lakes (Figure 16, Lange 2005). pH values were lower in the springs (5.0) than in the Lake (6.6), while conductivity was similar at both sites. Lake values for pH, conductivity, sulfate and chloride were similar between 2004 and 2008, while nitrate concentrations were lower in 2008 than 2004. pH values from the nearby ponds were between 6.6 and 6.7 (Figure 17, Lange 2005).

Table 5 – Woodbine Springs and Lake water samples. Mercury, pH, conductivity, sulfate, nitrate, and chloride measurements. 2004 data from Florida Lake Watch Program (Lange 2005). n.d. no data Analyte Woodbine

Springs Woodbine

Lake Woodbine

Lake Florida Lakes

average Year sampled 2008 2008 2004 2004 Total mercury (unfiltered) ng/L

6.69 +0.45 9.21+0.90 32.00 2.28

Dissolved mercury (filtered) ng/L

6.17+0.69 8.13+0.08 18.00 1.40

Particulate mercury % 7.8 11.2 44 38 pH 5.01 6.56 6.64 n.d. Conductivity mS 46.1 44.8 44.0 n.d. Sulfate mg SO4/L 0.30 0.51 0.73 n.d. Nitrate mg NO3/L 6.9 1.2 4.9 n.d. Chloride mg/L 7.2 7.6 6.9 n.d.

These samples from Woodbine Lake and Springs show high mercury values compared to

a stream survey done by the USGS in 1998 at 106 sites in 21 basins across the United States (Krabbenhoft et al. 1999, Table 6). For this comparison between the Woodbine Springs and USGS data, 2 very high mercury values (>200 ng/L) from a mine drainage in the Nevada Basin and Range were excluded. The values from Woodbine Lake fall in the 90th percentile of the USGS data and in the 84th percentile for Woodbine Springs (Figure 16).

In addition to having high mercury levels, the pH levels from Woodbine Springs are very low compared to the USGS study (Figure 17), with only one sample from their study having a lower pH value. Woodbine Lake was in the 10% percentile compared to the USGS samples.

20

Figure 16 – Total mercury concentrations (ng/L) in water samples collected by the USGS NAQWA survey (Krabbenhoft et al. 1999). Two high samples (>200 ng/L) from a mine drainage in the Nevada Basin and Range are excluded from this plot. Samples from the Mobile River and its tributaries and from South Florida water bodies are in orange. 2008 Samples from Woodbine Springs and Lake are in red. 2004 Samples in dark red (Lange 2005).

Figure 17 – pH in water samples collected by the USGS NAQWA survey (Krabbenhoft et al. 1999), our study and Lange study (Lange 2005). Colors as in Figure 16.

ACAD

ALMN

COOKDEL

RGRSL

LIN

JLT

ENMIA

MMOBLNEC

B NROK NVB

R OAH

USA

CR SA

NA SA

NTSO

FLTR

INUCOLUIR

B YE

LLYE

LL

Woo

dbine

Lk 2

008

Woo

dbine

Sp 2

008

Woo

dbine

Lk 20

04Pa

ce S

2004

Pace

N 20

04

Downe

y Pon

d 200

4

pH

4

6

8

10

ACAD

ALMN

COOKDEL

RGRSL

LIN

JLT

ENMIA

MMOBLNEC

B NROK NVB

R OAH

USA

CR SA

NA SA

NTSO

FLTR

INUCOLUIR

B YE

LLYE

LL

Woo

dbine

Lk 20

08

Woo

dbine

Sp 2

008

Woo

dbine

Lk 20

04Pa

ce S

2004

Pace

N 20

04

Downe

y Pon

d 200

4

Florid

a Lak

es 20

04

Tota

l mer

cury

ng/

L

0

5

10

15

20

25

30

35

21

Table 6 – NAWQA study basins (Krabbenhoft et al. 1999) Abbrev Study Basin Name ACAD Acadian-Pontchartrain ALMN Allegheny and Monongohela Basins COOK Cook Inlet Basin DELR Delaware River Basin GRSL Great Salt Lake Basins LINJ Long Island and N.J. Coastal Drainages LTEN Lower Tennessee River Basin MIAM Great and Little Miami River Basins MOBL Mobile River and Tributaries NECB New England Coastal Basins NROK Norther Rockies Intermontane Basins NVBR Nevada Basin and Range OAHU Oahu Island SACR Sacramento Basin SANA Santa Ana Basin SANT Santee Basin and Coastal Drainages SOFL Southern Florida TRIN Trinity River Basin UCOL Upper Colorado River Basin UIRB Upper Illinois River Basin YELL Yellowstone Basin

Discussion

These results have shown that there are no significant differences in mercury deposition

among our three sampling sites or between our sites and MDN sites in the Northern Gulf of Mexico between 2005 and 2007. The mercury deposition at the Ellyson site, which is generally downwind of the Crist coal fired power plant, was not significantly higher than at the Molino site which is furthest away from local mercury emission sources (Crist power plant and International Paper). Our results contrast with studies in Michigan and Maryland which found higher mercury deposition in urban areas compared to rural areas (Lawson and Mason 2001, Landis et al. 2002). We did observe seasonal differences, with the highest deposition rate during the summer at Pensacola Bay and MDN sites. Higher summer mercury deposition is not driven by higher rainfall amounts, but by higher concentrations in the rain since the highest rainfall amounts are usually in the spring and fall. Several potential mechanisms that could lead to enhanced mercury concentrations in rain during the summer include greater emissions associated with higher electricity use, scavenging of reactive gaseous mercury from the free troposphere by tall convective thunderstorms, and the concentration of reactive gaseous mercury by the sea breeze effect, where the diurnal alternation of onshore and offshore winds can lead to a buildup of pollutants in the air mass. As with seasonal differences, inter-annual variability was not strictly driven by rainfall amount since the year with the highest rainfall, 2005, had an intermediate mercury deposition rate.

22

The deposition of major ions (sulfate, chloride, nitrate, sodium, H+, or ammonium) was similar among the Pensacola Bay sites and often significantly higher than NADP sites. The NADP sites in Florida (FL14 and FL23) often had the lowest values of all the sites examined. This may be due to differences in how the sites were located. NADP criteria for site locations is that they should be 10 km away from urban areas or pollution sources or 40 km away from upwind pollution sources or population centers greater than 75,000 (Bigelow et al. 2001). The Florida NADP sites are much further from urban centers and major pollution sources than either the Alabama NADP or our Pensacola Bay sites. Pensacola Bay sites were not chosen by those criteria because we were interested in capturing deposition from local sources. While the Molino site is about 14 km from International Paper and 20 km away from other pollution sources in Pensacola, it still is high for sulfate and nitrate, which are often associated with fossil fuel combustion.

Correlation analysis and source tracking

Sea salt aerosols have a significant impact on chemistry of rain water in the Pensacola Bay area. Elements associated with sea water were highly correlated, including sodium, chloride, magnesium and strontium. The correlation coefficient among these elements was greater than 0.8 (p<0.001). About 21% of the rainfall events had Cl:Na ratios near the seawater Cl:Na ratio of 1.79 (Figure 18). However, chloride fluxes were enriched relative to the seawater ratio in about 63% of the rain events. This suggests that there is another source of chloride in rainwater.

The distance of the sampling site to the Gulf of Mexico determines how significant an effect sea salt aerosols have on rain water composition. Sites that are further from the Gulf have lower average annual sodium fluxes than those near to it (Figure 19). There was a similar pattern for chloride fluxes (data not shown). Fifty km appears to be the distance where the line becomes asymptotic and the sea salt effect is lost.

0.1

1

10

100

1000

01-Nov-04

30-Jan-05

30-Apr-05

29-Jul-05

27-Oct-05

25-Jan-06

25-Apr-06

24-Jul-06

22-Oct-06

20-Jan-07

20-Apr-07

19-Jul-07

17-Oct-07

Cl:N

a ra

tio

Ellyson

MolinoPace

seawater ratio

Figure 18: Plot of the ratio of chloride (Cl) and sodium (Na) fluxes at Pensacola Bay sites between November 2004 and December 2007. The seawater ratio (1.79) is plotted as a dashed line.

23

Figure 19 - Plot of average sodium (Na+) deposition (mg/m2/y) versus distance from Gulf

(km) for Pensacola Bay and NADP sites in the Gulf of Mexico. Element ratios are a useful tool for examining potential sources in rain water. Our

samples were highly enriched in some constituents relative to sodium. The sulfate to sodium ratio was 4 to 100 times greater than seawater ratio (Figure 20). Only 2% of the rain events had sulfate:sodium ratios near the seawater ratio. The highest ratios generally occurred in the summer, particularly in 2005 (Figure 20).

0.01

0.1

1

10

100

1000

01-Nov-04

30-Jan-05

30-Apr-05

29-Jul-05

27-Oct-05

25-Jan-06

25-Apr-06

24-Jul-06

22-Oct-06

20-Jan-07

20-Apr-07

19-Jul-07

17-Oct-07

SO4:

Na

ratio

EllysonMolinoPaceseawater ratio

Figure 20 - Plot of the ratio of sulfate (SO4) and sodium (Na) in rainfall fluxes at Pensacola Bay sites between November 2004 and December 2007. The seawater mass ratio (0.25) is plotted as a dashed line.

R2 = 0.70

0

200

400

600

800

1000

1200

1400

0 20 40 60 80 100 120 140 160

Distance from Gulf, km

Ave

rage

Na+

flux

mg/

m2/

y

24

In addition to sea salt, dust from the earth’s crust is a significant component of the elements found rain water. Correlation analysis revealed that elements from the earth’s crust were highly correlated with one another (r > 0.8, p <0.001). These elements were aluminum, barium, cesium, cobalt, chromium, iron, lithium, manganese, and silicon. Although mercury was significantly correlated (p < 0.01) with most of these elements, it was not a very strong relationship with most correlation coefficients at 0.40 or less (Appendix A). The ratio of mercury to aluminum in rain was much higher than the ratio found in the earth’s crust (Fig. 21, Lantzy and Mackenzie 1979). Most values generally occurred between the ratios found in coal (Levin and Goodman 2006) and industrial emissions such as fossil fuel combustion and other industrial particulate emissions (Lantzy and Mackenzie 1979). However, during some rainfall events, the mercury to aluminum ratio was higher even than the industrial emissions ratio. There was a similar pattern in ratios of other elements to aluminum. Three examples: antimony, lead and zinc are shown in Fig. 21. The ratios antimony to aluminum and lead to aluminum were usually higher than the ratio found in the earth’s crust (Fig 21). In contrast, values for the ratio of zinc to aluminum fell in between the crustal ratio and the industrial emissions ratio.

Figure 21 – Element ratios of mercury (Hg), antimony (Sb), lead (Pb) and zinc (Zn) versus aluminum (Al) in rain samples. The ratio of these elements in the earth’s crust are indicated with the dotted line, ratio in coal indicated with solid line, and ratio in industrial emissions indicated with dashed line.

Mercury was more closely correlated with a variety of other elements or major ions that are indicative of pollution sources (Table 7). The strongest correlations were with selenium, antimony, arsenic and sulfate. Selenium is often associated with coal combustion, while sulfate and nitrate are associated with fossil fuel combustion. Arsenic and selenium were significantly correlated with one another (r = 0.53) as were arsenic and antimony (r = 0.63) and selenium and antimony (r= 0.61). Mercury was also correlated with lead (r = 0.58), bismuth (r = 0.51) and gallium (r = 0.51). Cadmium and zinc were also significantly correlated with each other (r= 0.70). These results are similar to those observed in other studies where mercury and trace metals

1

10

100

1000

10000

1 10 100 1000 10000 100000Al deposition µg/m2

Hg

depo

sitio

n ng

/m2

EllysonMolinoPacecrustal ratiocoal ratioIndustrial emissions

1

10

100

1000

10000

100000


Pb

depo

sitio

n ng

/m2


0.1

1

10

100

1000


Zn d

epos

ition

µg/

m2

EllysonMolinoPacecrustal ratioIndustrial emissions

1

10

100

1000

10000


Sb

depo

sitio

n ng

/m2


25

associated with coal combustion and other industrial activities were correlated (Dvonch et al. 1998, Lawson and Mason 2001, Landis et al. 2002, Keeler et al. 2006).

Table 7 – Correlation coefficients of mercury and elements or major ions

Constituent correlated with mercury

Correlation coefficient

Selenium 0.70 Arsenic 0.67 Antimony 0.67 Sulfate 0.64 Nitrate 0.59 Lead 0.58 Gallium 0.51 Bismuth 0.51

The relationship between mercury and certain trace metal such as arsenic, selenium,

antimony, and tin are shown below (Fig. 22).

Figure 22 – Element ratios of arsenic (As), selenium (Se), antimony (Sb), and tin (Sn) to mercury in rain samples. The ratio of these elements in the earth’s crust are indicated with the dotted line, ratio in coal indicated with solid line, and ratio in industrial emissions indicated with dashed line.

From our multi-element data, there are two ways to estimate the contribution from coal combustion to the rainfall Hg deposition. Both methods require an initial correction of the rainfall chemistry data to subtract out the impacts on trace element chemistry due to mineral dust aerosols. The trace element and mercury rainfall concentrations are corrected for contributions from mineral dust, using crustal abundance ratios relative to aluminum to calculate the “excess” (xs) trace element (TE) concentrations:

0.1

1

10

100

1000

1 10 100 1000 10000

Hg deposition ng/m2

As d

epos

ition

µg/

m2


0.01

0.1

1

10

100

1 10 100 1000 10000

Hg deposition ng/m2

Se d

epos

ition

µg/

m2

EllysonMolinoPacecrustal ratioindustrial emissions

0.1

1

10

100

1000

1 10 100 1000 10000

Hg deposition ng/m2

Sn d

epos

ition

µg/

m2


1

10

100

1000

10000

1 10 100 1000 10000

Hg deposition ng/m2

Sb d

epos

ition

ng/

m2

EllysonMolinoPacecrustal ratiocoal ratioIndustrial Emissions

26

xsTE = totalTE – totalAl x (TE/Al)crust Equation 1 where xsTE = trace element concentration corrected for mineral dust, totalTE = total

trace element concentration measured in each sample, and totalAl = total aluminum concentration measured in each sample. Al is used because it is almost exclusively associated with mineral dust in aerosols. (TE/Al)crust = crustal abundance ratio for trace element/Al. The crustal abundances can vary slightly depending on whose compilation of “average crustal composition” is used. We used the average crustal abundances from Taylor (1964).

For the first method, we make the assumption that the proportions of mercury and other

trace elements vaporized during coal combustion are similar to the stack emission ratios, and that those ratios are carried downwind without alteration until the substances are scavenged and deposited by rainfall. We choose for these calculations several relatively volatile trace elements whose vaporization ratios to mercury have been reported for coal combustion (As, Se, Sn, Sb). If we further assume that the excess trace element concentrations are impacted exclusively from coal combustion, we can use the following equation to estimate the fraction of the rainfall Hg resulting from coal combustion:

%Hg from coal combustion = ∑(xsTE x (Hg/TE)coal x 100) Equation 2 ∑[Hg]dep The %Hg from coal combustion is then summed for all rain events for each year. Table 8

(column 1, CFPP Hg/TE) shows that this estimate ranges from 39-54%, depending upon which trace element is used for the calculation.

For the second approach, we use the actual rainfall data to find samples with high volatile trace element concentrations. These samples are almost certainly impacted by anthropogenic atmospheric emissions. Since we believe that rainfall Hg can result from long-range transport of RGM in addition to local and regional emission sources, we searched the database for rainfall samples showing the lowest xsHg/xsTE ratios, since they should reflect the least impact from distant sources. We averaged the ten lowest xsHg/xsTE ratios (for As, Se, Sn, and Sb) to calculate the minimum (xsHg/xsTE) ratios and used this equation to estimate the fraction of the rainfall Hg resulting from coal combustion:

%Hg from coal combustion = ∑(xsTE x (xsHg/xsTE)min x 100) Equation 3 ∑[Hg]dep The %Hg from coal combustion is then summed for all rain events for each year. Table 8

(column 2, observed minimum Hg/TE) shows that the results for the second approach range from 25-38%, depending upon which trace element is used for the calculation.

In the first example using Equation 2, we recognize that the assumption that the Hg/TE vaporization ratio is conserved downwind from coal-fired power plants (until scavenging and deposition by rainfall) may not be valid. If emitted RGM is converted to gaseous elemental Hg during transport, then the Hg/TE vaporization ratio is too high. It is also possible that the volatile TEs in our rain samples (As, Se, Sn and Sb) result from multiple sources, and not only from coal combustion. In that case the xsTE concentrations we attribute to coal combustion are inappropriately high. Both of these factors would yield “%Hg from coal combustion” estimates that are too high. In the second example using equation 3, if a significant fraction of the volatile

27

trace elements in our rain samples came from sources other than coal combustion, then we should use lower xsTE concentrations, but the (xsHg/xsTE)min ratios would increase and partially cancel the correction. For these reasons, we conclude that the second approach may produce more reliable estimates of the %Hg coming from coal combustion.

Finally, our estimates of the %Hg from coal combustion represent the “average” chemistry of the rain samples, and thus reflect the “average” chemistry of the atmosphere from which the Hg and trace elements are scavenged. Therefore, we cannot attribute the %Hg estimates we report to any particular power plant or even to emissions within the specific local area. We do believe that these estimates reflect the fact that the entire southeastern US atmosphere is impacted by emissions from coal combustion. Table 8 – Percent mercury in rainfall from regional coal combustion based on mercury/trace element ratios

Trace element CFPP Hg/TE Observed minimum Hg/TE Arsenic 39 35 Selenium 51 38 Tin 40 25 Antimony 54 36

The higher deposition of copper, chromium and zinc at the Ellyson site compared to

Molino site may be a result of industrial activity at the Ellyson site and higher emissions of these elements. Mercury and trace metal analyses of the Crist plant plume by the EPA study “Mercury speciation in coal-fired utility boiler emission plume” should provide useful information in the future that would allow us to discriminate between emissions from the Crist plant and those from other sources. Meteorological Modeling: Back trajectory and plume dispersion

Back trajectory meteorological models were run on each rain event that was collected during the sampling period. NOAA’s HYbrid Single-Particle Lagrangian Integrated Trajectory Model (Hysplit) version 4.7 was used with the archived EDAS 40km grid meteorological dataset for the models (http://www.arl.noaa.gov/ready/hysplit4.html). Each trajectory was run for 24 hours prior to the start of the rain event as recorded at the Pensacola airport (PNS). The times of the rain events at PNS were used as the end points for the back trajectories after careful comparison of rain event times collected at Southern Company’s OLF site in northwest Pensacola with the rain event times at PNS. If there was a time difference of more than two hours between the rain events at both sites, then the back trajectories were modeled for both event times and compared. Several dates were chosen to run back trajectories from each of the sample sites as well as the power plant stack and PNS. Since the Hysplit model uses the EDAS 40km dataset and all five sites are within 40km of each other, no differences between them can be seen in the models. For this reason and to be consistent with plume dispersion models, we modeled each rain event back trajectory from the Crist power plant stack location.

Each event was modeled to ending heights of 500 m, 1000 m and 5000 m, as these heights are representative of the formation of rain within and above the average boundary layer along the Gulf coast of Florida, (Hsu and Blanchard, 2005). The back trajectory model results were clustered into three groups for comparison – those originating over the continental US, those originating over the Gulf of Mexico and those which remained relatively stagnant over the Pensacola area for the 24 hour period. For each of these types of trajectories, a representative

28

model with a four-hour plume dispersion model ending at the start of the rain event is shown in Figures 23 through 25.

Figure 23 a) NOAA Hysplit 24 hour back trajectory model and b) plume dispersion model ending and beginning at Crist power plant stack for a rain event occurring on 8/8/2005 at 1800UTC at PNS. The models are representative of a land derived particle trajectory.

29

Figure 24 a) NOAA Hysplit 24 hour back trajectory model and b) plume dispersion model ending and beginning at Crist power plant stack for a rain event occurring on 7/5/2005 at 0600UTC at PNS. The models are representative of a marine derived particle trajectory.

Figure 25 a) NOAA Hysplit 24 hour back trajectory model and b) plume dispersion model ending and beginning at Crist power plant stack for a rain event occurring on 8/1/2005 at 1800UTC at PNS. The models are representative of a stagnant particle trajectory. The trajectories that were the most characteristic of these three types were compared to the trace element to aluminum ratios to determine if any pattern could be discerned. We would expect to

30

see some similarities in the ratios for each of the trajectory categories but no significant pattern emerged from the data. The mercury to aluminum ratio is illustrated in Figure 26. Many of the highest trace element to aluminum ratios, a result of low aluminum concentrations, are associated with marine trajectories, identifying the lack of continental dust. This suggests that pollutant metals have become widely dispersed in the regional atmosphere, so that even “marine” air masses are affected. The trace element to mercury ratios were compared to the trajectory type as well. Again, no significant pattern emerged from the data.

Figure 26 - NOAA Hysplit 24 hour back trajectory models were clustered by whether the origination of the air particle was continental, marine or relatively stagnant. The trajectories that were the most characteristic of these three types were compared to the Hg/Al ratio

Project Outputs This project has resulted in the publication of a master’s thesis and 9 presentations at 6 conferences. Conference abstracts are included in Appendix B. Sara Cleveland Thesis is in Appendix C. ATMOSPHERIC MERCURY INPUT TO THE PENSACOLA BAY WATERSHED By SARA D. CLEVELAND A thesis submitted to the Department of Oceanography in partial fulfillment of the requirements for the degree of Master's of Science Degree Awarded: Fall semester, 2006 Conference Presentations Total gaseous mercury concentrations in Pensacola, FL. March 2006. American Chemical

Society. Atlanta, GA.

Spatial and temporal variability in precipitation chemistry in the Pensacola Bay watershed. March 2006. American Chemical Society. Atlanta, GA.

Mercury deposition in the Pensacola Bay watershed. March 2006. American Chemical Society. Atlanta, GA.

Trace element correlations in rainfall from the Pensacola Bay watershed. March 2006. American Chemical Society. Atlanta, GA.

8th International Conference on Mercury as a Global Pollutant in Madison WI, August 8, 2006

Atmospheric deposition of mercury, trace metals and major ions in the Pensacola Bay watershed NW Florida Regional Environmental Symposium on October 5, 2007

31

Atmospheric deposition of mercury, trace metals and major ions in the Pensacola Bay watershed: implications for Choctawhatchee Bay. Symposium on Choctawhatchee Bay. June 1-2, 2008.

Atmospheric Deposition of Mercury, Trace Metals and Major Ions in the Pensacola Bay Watershed. National Monitoring Conference May 18-23, 2008. Atlantic City, NJ

Mercury and trace elements in rainfall from the Pensacola airshed: local, regional, and distant sources. 9th International Conference on Mercury as a Global Pollutant (9th ICMGP) in China June 7-12, 2009.

Local presentations Mercury meeting at Weeks Bay National Estuarine Research Reserve - October 21, 2005

Pensacola Chapter of American Chemical Society meeting - Oct 26, 2006

Concepts in Chemistry Class, University of West Florida – Spring 2008 & Fall 2008

Outcomes The information about the seasonal and spatial patterns in mercury, trace metals and

major ions in this report should prove useful to the state of Florida as it develops a TMDL for mercury. Because of event driven sampling and analyses of trace metals, this report provides a more detailed evaluation of atmospheric deposition in the region than is possible with the data from MDN and NADP networks. Relatively few studies of this sort exist to inform decision-makers as efforts are made to reduce emissions of mercury and other constituents at state and national levels.

Monitoring at these sites was extended in order to complement an EPA study, “Mercury Speciation in Coal-fired Utility Boiler Emission Plume”, at Plant Crist in February 2008 by the EPA Office of Research and Development. It has further been extended to evaluate the effect of adding scrubbers to Plant Crist with continuing support from the Electric Power Research Institute.

Summary

The mercury deposition in the Pensacola Bay watershed is similar to that found in other

sites throughout the Central Gulf of Mexico region. Mercury deposition during the summer months is higher than other months due to higher concentrations in rainfall throughout the region. Because of the large number of mercury emission sources in the Southeast region and mixing of air masses from different regions, it is not surprising that we are unable to see a direct effect of local emission sources on the amount of mercury deposited to Pensacola Bay in rain. However, the correlations of mercury with other elements and major ions suggest that coal combustion is a significant source of mercury to the region, and may account for between 25 and 54% of the mercury deposited. Based on the sediment core collected from Snag Lake, mercury deposition tripled between the 1950s and 1970s, a period of industrialization in the region and nation.

32

Deposition of constituents like H+, sulfate, nitrate, ammonium, chloride and sodium, are much higher in Pensacola Bay that at NADP sites. Chloride and sodium fluxes are higher because Pensacola Bay sites are closer to the Gulf of Mexico which is a source of sea salt aerosols. Acid rain constituents, H+, sulfate, nitrate and ammonium are most likely higher at Pensacola Bay sites than NADP sites because Pensacola Bay sites are much closer to emission sources of these constituents than NADP sites, particularly the two Florida NADP sites FL14 and FL23 which are located in rural counties far from major industrial activities.

Acknowledgments This study was a component of the “Assessment of Environmental Pollution and

Community Health in Northwest Florida”, supported by a U.S. EPA Cooperative Agreement award X-9745502 to the University of West Florida (Project Director: Dr. K. Ranga Rao). We thank Fran Aftanas, Nathaniel Davila, Autumn Dunn, Elizabeth Gaige, Brad Kuykendall, Tanner Martin, Melissa Overton, and Pam Vaughan for their assistance in field collections. We thank Sara Cleveland, Kati Gosnell, Nishanth Krishnamurthy, Jan Macauley, Jeremy Bosso and the Wetland Research Laboratory for laboratory analyses. The Florida Department of Environmental Protection, Pace Water Systems, and Carol Hatcher provided access to sampling sites. We thank Jessie Brown for her assistance with the database development. We thank Danny France and Jerry Burger from EPA Region IV for their assistance in setting up and calibrating the Tekran Elemental mercury analyzer.

References

Appleby, P.G., and Oldfield, F., 1983. The assessment of 210Pb data from sites with varying sediment accumulation rates, Hydrobiologia, 103:29-35. Benninger, L.K., Lewis, D.M., and Turekian, K.K., 1975. The use of natural 210Pb as a heavy metal tracer in the river-estuarine system, Marine Chemistry and the Coastal Environment, Church, T.M. ed., American Chemical Society, Symposium Series 18, Washington D.C. Bigelow, D.S., Dossett, S.R., and Bowersox, V.C. 2001. Instruction Manual: NADP/NTN Site selection and installation. NADP Program Office. Illinois State Water Survey. 2204 Griffith Dr. Champaign, Illionois 61820-7495. http://nadp.sws.uiuc.edu/lib/manuals/siteinst.pdf Covelli, S. J Faganeli, M. Horvat, and A. Brambati. 2001. Mercury contamination of coastal sediments as a result of long-term cinnabar mining activity (Gulf of Trieste, northern Adriatic sea). Applied Geochemistry 16: 541-558. Dvonch, J.T., J.R. Graney, F.J. Marsik, G.J. Keeler, and R.K. Stevens. 1998. An investigation of source-receptor relationships for mercury in south Florida using even precipitation data. The Science of the Total Environment 213:95-108. Folk, R.L. 1974. Petrology of Sedimentary Rocks. Hempill.

33

Hornberger, M.I., S.N. Luoma, A. van Geen, C. Fuller, and R. Anima. 1999. Historical trends of metals in the sediments of San Francisco Bay, California. Marine Chemistry 64: 39-55 Hsu, S.A. and B.W. Blanchard. 2005. Visibility and atmospheric dispersion capability of the northern Gulf of Mexico: Estimations and observations of boundary layer parameters: Final Report. U.S. Dept. of the Interior, Minerals Management Service, Gulf of Mexico OCS Region, New Orleans, LA. OCS Study MMS 2005-008. 184 p. Keeler, G.J., M.S. Landis, G.A. Norris, E.M. Christianson, and J.T. Dvonch. 2006. Sources of mercury wet deposition in Eastern Ohio, USA. Environ. Sci. Technol. 40:5874-5881. Krabbenhoft, D.P., Wiener, J.G., Brumbaugh, W.G., Olson, M.L., DeWild, J.F., and Sabin, T.J., 1999, A National pilot study of mercury contamination of aquatic ecosystems along multiple gradients, in U.S. Geological Survey Toxic Substances Hydrology Program—Proceedings of the Technical Meeting, Charleston, South Carolina, March 8-12, 1999—Volume 2 of 3—Contamination of Hydrologic Systems and Related Ecosystems: U.S. Geological Survey Water-Resources Investigations Report 99-4018B, p. 147-160. Landis, M.S., A.F. Vette and G.J. Keeler. 2002. Atmospheric mercury in the Lake Michigan Basin: Influence of the Chicago/Gary Urban Area. Environ. Sci. Technol. 36: 4508-4517. Lange, T. 2005. Statewide Mercury Investigations: Woodbine Springs Lake Investigations. Progress Report to FL Department of Environmental Conservation. DEP Contract #SP377Amendment#5. 8 p. Lantzy, R.J. and F.T. Mackenzie. 1979. Atmosheric trace metals: global cycles and assessment of man’s impact. Geochimca et Cosmochimica Acta 43:511-525 Lawson, N.M. and R.P. Mason. 2001. Concentration of mercury, methylmercury, cadmium, lead, arsenic and selenium in the rain and stream water of two contrasting watersheds in western Maryland. Wat. Res. 35:4039-4052. Levin, L. and N. Goodman, 2006, Electric Power Research Institute, 3420 Hillview Ave., Palo Alto, California 94304-1395 U.S.A., personal communication. Lin, C-J. and S.O.Pehkonen 1999. The chemistry of atmospheric mercury: a review. Atmospheric Environment 33: 2067-2079. National Atmospheric Deposition Program (NRSP-3)/National Trends Network. 2009. NADP Program Office, Illinois State Water Survey, 2204 Griffith Drive, Champaign, IL 61820. Parsons, T.R., Maita, Y., and Lalli, C.M. (1984) A manual of chemical and biological methods for seawater analysis. Pergamon Press, Oxford, 173 pp Pirrone, N., G. J. Keeler, and J.O. Nriagu. 1996. Regional differences in worldwide emissions of mercury to the atmosphere. Atmospheric Environment 30: 2981-2987.

34

Schelske, C.L., Peplow, A., Brenner, M., and Spencer, C.N., 1994. Low-background gamma counting: applications for 210Pb dating of sediments, J. Paleolimnol., 10:115-128. Taylor, S.R. 1964. Abundance of chemical elements in the continental crust: a new table. Geochimica et Cosmochimica Acta 28:1273-1285. Tomiyasu, T., A. Nagano, N. Yonehara, H. Sakamoto, Rifardi, K. Oki, and H. Akagi. 2000. Mercury contamination in the Yatsushiro Sea, south-western japan: spatial variations of mercury in sediment. The Science of the Total Environment 257:121-132. Valigura, R.A. 2001. An introduction to the first assessment of nitrogen loads to US estuaries with an atmospheric perspective. pp. 1-10. In R.A. Valigura, R.B. Alexander, M.S. Castro, T.P. Meyers, H.W. Paerl, P.E. Stacey and R.E. Turner (eds.) Nitrogen loading in coastal water bodies: an atmospheric perspective. AGU. Washington, D.C. Wigand, C. J.C. Stevenson, and J. C. Cornwell. 1997. Effects of different submersed macrophytes on sediment biogeochemistry. Aquatic Botany 56:233-244.

Appendix A Pearson correlation coefficients for Hg, and selected trace metals. Values in red are significant at p <0.01 level and have a correlation coefficient greater than 0.6; values in bold are significant at p <0.01, while values in italics are significant at p < 0.05.

Hg Al As Ba Bi Cd Co Cu Cr Ce Ga Fe Li Mg Mn Ni Al 0.45 1 As 0.67 0.19 1 Ba 0.52 0.72 0.35 1 Bi 0.51 0.45 0.46 0.64 1 Cd 0.30 0.37 0.28 0.34 0.45 1 Co 0.39 0.65 0.26 0.72 0.42 0.29 1 Cu 0.47 0.49 0.34 0.59 0.43 0.36 0.52 1 Cr 0.29 0.56 0.12 0.63 0.31 0.32 0.74 0.48 1 Ce 0.43 0.80 0.22 0.63 0.42 0.42 0.56 0.40 0.44 1 Ga 0.51 0.50 0.44 0.52 0.42 0.38 0.43 0.39 0.41 0.52 1 Fe 0.47 0.95 0.18 0.78 0.49 0.32 0.68 0.54 0.58 0.81 0.48 1 Li 0.50 0.65 0.43 0.62 0.53 0.27 0.51 0.43 0.43 0.59 0.70 0.64 1Mg 0.18 0.16 0.36 0.19 0.14 0.12 0.15 0.22 0.18 0.21 0.39 0.19 0.60 1Mn 0.48 0.80 0.30 0.78 0.52 0.33 0.66 0.55 0.52 0.75 0.52 0.87 0.64 0.32 1Ni 0.22 0.26 0.17 0.48 0.21 0.30 0.66 0.34 0.78 0.26 0.34 0.25 0.27 0.11 0.26 1P 0.21 0.27 0.24 0.29 0.32 0.39 0.23 0.45 0.19 0.36 0.28 0.32 0.25 0.17 0.46 0.14Pb 0.58 0.60 0.51 0.70 0.74 0.34 0.52 0.52 0.39 0.60 0.47 0.61 0.60 0.23 0.64 0.32Sb 0.67 0.52 0.63 0.63 0.71 0.38 0.45 0.62 0.35 0.49 0.59 0.54 0.53 0.25 0.60 0.26Se 0.70 0.22 0.53 0.39 0.39 0.25 0.22 0.39 0.10 0.27 0.42 0.24 0.33 0.27 0.30 0.13Si 0.36 0.91 0.05 0.68 0.34 0.32 0.65 0.46 0.55 0.79 0.45 0.95 0.58 0.17 0.78 0.24Sn -0.01 -0.04 0.02 0.03 0.00 0.00 0.08 0.02 0.02 -0.03 0.01 -0.03 -0.01 0.01 -0.02 0.10Sr 0.29 0.35 0.40 0.37 0.28 0.20 0.31 0.34 0.29 0.41 0.48 0.39 0.69 0.97 0.51 0.16V 0.42 0.46 0.46 0.43 0.39 0.31 0.38 0.31 0.29 0.53 0.76 0.44 0.58 0.24 0.44 0.28Zn 0.26 0.28 0.25 0.34 0.54 0.70 0.26 0.46 0.32 0.29 0.29 0.31 0.29 0.24 0.38 0.16

P Pb Sb Se Si Sn Sr V Zn Pb 0.29 1 Sb 0.36 0.76 1 Se 0.22 0.43 0.61 1 Si 0.26 0.48 0.41 0.11 1Sn -0.01 0.00 0.01 0.01 -0.03 1Sr 0.25 0.38 0.39 0.36 0.37 0.00 1V 0.17 0.50 0.50 0.48 0.43 0.00 0.35 1Zn 0.45 0.33 0.42 0.22 0.27 -0.01 0.32 0.20 1

Appendix B – Conference abstracts

American Chemical Society - March 26, 2006

GEOC 84 - Total gaseous mercury concentrations in Pensacola, FL

Jane M. Caffrey1, William M. Landing2, and Sara D. Cleveland2. (1) Center for Diagnostics and Bioremediation, University of West Florida, 11000 University Parkway, Pensacola, FL 32514, (2) Department of Oceanography, Florida State University, Tallahassee, FL 32306-4320

Total gaseous mercury concentrations have been measured continuously in Pensacola, FL since February 9, 2005. Readings were integrated over 5 minute intervals. February and March were characterized by periodic afternoon spikes of mercury exceeding 10 ng/m3. The highest concentrations observed were 35 ng/m3. These spikes were less common in April and May and concentrations greater than 10 ng/m3 did not appear in June or July. Elevated SO2 concentrations, up to 30 ppb, often coincided with mercury spikes. Spikes occurred following periods of calm morning winds. We are using the HYSPLIT trajectory analysis model to back forecast the potential source of elevated concentrations. GEOC 85 - Spatial and temporal variability in precipitation chemistry in the Pensacola Bay watershed

Jane M. Caffrey1, William M. Landing2, and Sara D. Cleveland2. (1) Center for Diagnostics and Bioremediation, University of West Florida, 11000 University Parkway, Pensacola, FL 32514, (2) Department of Oceanography, Florida State University, Tallahassee, FL 32306-4320

We evaluate the temporal and spatial patterns in atmospheric wet deposition in the Pensacola Bay watershed. Three sites were established in November 2004. Peak rainfall occurred during April 2005 when we collected over 15 cm of rain in individual rain events at the different sites. Site to site variation in rainfall amounts was significant because of the patchy nature of thunderstorm activity. pH values at the three sites ranged from 3.7 to 5.6. There were no consistent differences among the different sites. Sulfate concentrations and pH values were negatively correlated as were nitrate and pH. The marine influence is evident at two sites near Pensacola Bay, which had higher chloride and sodium concentrations than an inland site. Chloride and sodium concentration were significantly positively correlated.

37

GEOC 86 - Mercury deposition to the Pensacola Bay watershed

Sara D. Cleveland1, William M. Landing1, and Jane M. Caffrey2. (1) Department of Oceanography, Florida State University, Tallahassee, FL 32306-4320, (2) Center for Diagnostics and Bioremediation, University of West Florida, 11000 University Parkway, Pensacola, FL 32514

By sampling individual rain events over a 1-2 year period at three sites situated around a known point source of atmospheric mercury, we are attempting to quantify its influence on local mercury deposition. In the first year, we collected samples for 61 individual rain events. Total mercury concentrations in the rainwater samples range from 2-40 ng/L. The volume weighted mean rainfall Hg concentrations for the first full year of sampling range from 9.2-9.5 ng/L, and there were no significant differences in the rainfall Hg flux between the three sites. Plume dispersion modeling and air-mass back trajectory analysis are being used to interpret the Hg concentrations measured for each rain event. GEOC 87 - Trace element correlations in rainfall from the Pensacola Bay watershed William M. Landing1, Sara D. Cleveland1, and Jane M. Caffrey2. (1) Department of Oceanography, Florida State University, Tallahassee, FL 32306-4320, (2) Center for Diagnostics and Bioremediation, University of West Florida, 11000 University Parkway, Pensacola, FL 32514 A full year of individual rain event samples from three sites in Pensacola, Florida have been analyzed for mercury and a suite of other trace elements. A multi-element analytical program has been set up using a Thermo-Finnigan “Element” high-resolution ICP-MS. We have identified over 50 elements that are significantly enriched in rain samples relative to the method blank, including the alkali metals and alkaline earth elements, all three rows of the transition metals, and the rare earth elements. Plume dispersion and air-mass back trajectory analysis is also being conducted for each rain event. The goal of this research is to use trace element relationships in an effort to identify, and quantify, the impacts from various emission sources in the region on rainfall chemistry. Preliminary analyses showed that rainfall mercury deposition was significantly correlated with Cd, Pb, V, and Zn (R2>0.7).

38

8th International Conference on Merucry as a Global Polutant – August 2006 Rainfall mercury deposition and trace element correlations in the Pensacola Bay Sara D. Cleveland1, William M. Landing1, and Jane M. Caffrey2. (1) Department of Oceanography, Florida State University, Tallahassee, FL 32306-4320, (2) Center for Diagnostics and Bioremediation, University of West Florida, 11000 University Parkway, Pensacola, FL 32514 This project examines rainfall input of mercury and other trace elements to the Pensacola Bay watershed. According to the EPA, 1% of the mercury emissions to the atmosphere are from US coal fired power plants. The largest potential source in the Pensacola Bay area is the Crist coal fired power plant. By sampling at 3 sites around the Crist coal fired power plant and comparing with data collected by the Mercury Deposition Network (MDN), we can see if a known point source has a direct impact in the local environment. Comparing trace elements of interest to Al is a traditional way to estimate “excess” trace element input from anthropogenic sources. Trace elements identified by Principal Component Factor Analysis (PCFA) which cluster with Hg have been studied in more detail by this method. Choctawhatchee Basin Symposium - June 2008 Atmospheric deposition of mercury, trace metals and major ions in the Pensacola Bay watershed: implications for Choctawhatchee Bay Caffrey, J.M. (University of West Florida), Cleveland, S.(Florida Department of Environmental Protection), Gosnell, K.(Florida State University), Landing, W.M. (Florida State University) Center for Environmental Diagnostics and Bioremediation, University of West Florida, 11000 University Parkway Pensacola, FL 32514 Atmospheric deposition of mercury, trace metals and major ions have been monitored in the Pensacola Bay watershed to evaluate the temporal and spatial patterns in atmospheric wet deposition. A goal of this project was to evaluate the contribution of local sources (coal fired power plant and paper mill) to atmospheric deposition. Three sites were established in November 2004 to collect rainfall from individual rain events. Total mercury concentrations in the rainwater samples range from 2-40 ng/L. There were no significant differences in the rainfall Hg flux between the three sites or between the Pensacola Bay sites and nearby Mercury Deposition Network monitoring sites in Alabama. Volume weighted mean pH values at the three sites ranged from 4.1 to 5.4. Sulfate concentrations and pH values were negatively correlated as were nitrate and pH. There were no consistent differences in pH or sulfate concentrations among the different sites from Pensacola Bay or the National Atmospheric Deposition (NADP) sites in Mobile and the panhandle of Florida. However, nitrate concentrations in Pensacola Bay were higher than nearby NADP sites. Multivariate statistical methods are used to sort these trace elements into factors that represent potential source components that contribute to the rainfall chemistry. Ga, Bi, V, As, Se, Sb, and Sn are all correlated with mercury. Using various Hg/element ratios, we can estimate that 16-46% of the rainfall mercury in the region results from coal combustion. However, we cannot definitively distinguish local vs. regional or distant sources. The similarity in mercury and major ion deposition rates between Pensacola and Mobile Bays suggest that there is a common airshed responsible for the materials being deposited. Thus, deposition in Choctawhatchee Bay may be similar to the rest of the region.

39

National Monitoring Conference - May 2008

ATMOSPHERIC DEPOSITION OF MERCURY, TRACE METALS AND MAJOR IONS IN THE PENSACOLA BAY WATERSHED Caffrey, J.M. (University of West Florida), Cleveland, S.(Florida Department of Environmental Protection), Gosnell, K.(Florida State University), Landing, W.M. (Florida State University) Center for Environmental Diagnostics and Bioremediation, University of West Florida, 11000 University Parkway Pensacola, FL 32514 Atmospheric deposition of mercury, trace metals and major ions have been monitored in the Pensacola Bay watershed to evaluate the temporal and spatial patterns in atmospheric wet deposition. A goal of this project was to evaluate the contribution of local sources (coal fired power plant and paper mill) to atmospheric deposition. Three sites were established in November 2004 to collect rainfall from individual rain events. Total mercury concentrations in the rainwater samples range from 2-40 ng/L. The annual volume weighted mean rainfall Hg concentrations were 9.5 and 11.9 ng/L in 2005 and 2006, respectively. There were no significant differences in the rainfall Hg flux between the three sites or between the Pensacola Bay sites and nearby Mercury Deposition Network monitoring sites in Alabama. Volume weighted mean pH values at the three sites ranged from 4.1 to 5.4. Sulfate concentrations and pH values were negatively correlated as were nitrate and pH. There were no consistent differences in pH or sulfate concentrations among the different sites from Pensacola Bay or the National Atmospheric Deposition (NADP) sites in Mobile and the panhandle of Florida. However, nitrate concentrations in Pensacola Bay were higher than nearby NADP sites. The marine influence is evident at two sites near Pensacola Bay, which had higher chloride and sodium concentrations than an inland site and much higher chloride and sodium deposition than nearby NADP sites. Chloride and sodium concentration were significantly positively correlated. The ratio of chloride to sodium concentrations in Pensacola Bay rain samples is identical to the ratio of seawater. We use principal components factor analysis to identify trace elements which cluster with mercury. Preliminary analyses suggest that there are four factors associated with the mercury and trace element deposition: the Al-Si (crustal abundance) factor, the sea-salt factor, the “P” factor (linking phosphorus with copper and zinc, possibly from agriculture) and the Hg factor. The Hg factor clusters Hg with five trace elements: Ga, Bi, Sb, Pb and V. Bi and Sb are volatiles that we assume come from coal combustion. Ga, V and Pb may also come from coal combustion. KEYWORDS: Atmospheric deposition, mercury, trace metals, acid rain, sea salt aerosols, principal components analysis

40

9th International Conference on Mercury as a Global Pollutant - June 2009 Mercury and trace elements in rainfall from the Pensacola airshed: local, regional, and distant sources WM Landing, KJ Gosnell (FSU/Oceanography), Jane Caffrey (UWF/CEDB)

To understand and quantify the impact of local, regional, and distant atmospheric mercury sources to rainfall mercury deposition in the Pensacola, FL watershed, a program of event-based rainfall sampling was started in late 2004. Modified Aerochem-Metrics wet/dry rainfall samplers were deployed at three sites in the region around the Crist coal-fired power plant and event-based samples were collected continuously for four years. Samples were analyzed for total mercury and a suite of trace elements including Li, Ga, Rb, Sr, Y, Zr, Nb, Cd, Sb, Cs, Ba, La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu, Hf, Ta, Pb, Bi, Th, U, Na, Mg, Al, Si, P, Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Sb, and Sn. Nutrients and major ions were also measured on each sample. Multivariate statistical methods are used to sort these tracers into factors that represent potential source components that contribute to the rainfall chemistry. As, Se, Sn, Sb, Pb, and Bi are all significantly correlated with mercury. Using various Hg/element ratios, we can estimate that 16-46% of the rainfall mercury in the region results from coal combustion. Combining these trace element “fingerprints” with meteorological air-mass back trajectory analysis, we hope to be able to distinguish local vs. regional or distant sources of atmospheric mercury.

Final report Atmospheric Deposition of Mercury and …...Final report Atmospheric Deposition of...

Documents

Transcript of Final report Atmospheric Deposition of Mercury and …...Final report Atmospheric Deposition of...