Document JZM5aymZLGOoQBe7891Z73BK

Environ. Sci. Technol. 2002, 36, 4245-4252 Seasonal and Annual Air--Water Exchange of Polychlorinated Biphenyls across Baltimore Harbor and the Northern Chesapeake Bay HOLLY A. BAMFORD, FUNG CHI KO, AND JOEL E . BAKER* Chesapeake Biological Laboratory, University ofMaryland, Center for Environmental Science, Solomons, Maryland 20688 As part of the Atmospheric Exchange Over Lakes and Oceans Study (AE0L0S), air-water exchange fluxes of polychlorinated biphenyls (PCBs) were determined along a transect in the Baltimore Harbor from the Inner Harbor of Baltimore, MD, to the mainstem of the northern Chesapeake Bay to assess the overall contributions of urban source PCBs through air-water exchange and their impacts on coastal waters. Six sites were sampled during three intensive sampling periods in June 1996, February 1997, and July 1997 and at two sites (urban and rural) every ninth day between March 1997 and March 1998 to measure spatial and temporal variability in the PCB air-water exchange fluxes. During the intensive sampling campaigns, net total PCB (tPCBs) volatilization fluxes ranged between 5 and 2120 ng rrr2 day-1, varying on both temporal and spatial scales. Volatilization fluxes were highest in February, driven by high winds and elevated dissolved PCB concentrations. Throughout the annual sampling period, f-PCB volatilization fluxes were similar between urban (130 ,ug m-2 yr-1) and rural (120 /tg m-2 yr-1) sites. Approximately 10% of the dissolved f-PCB inventory in the water column in Baltimore Harbor exchanges with the atmosphere each day. From airwater exchange alone, the estimated residence time of dissolved f-PCBs in the harbor is approximately 10 days (estimated hydraulic residence time of water in the harbor is between 7 and 14 days), indicating that air-water exchange is an efficient removal mechanism of f-PCBs from urban coastal waters. Furthermore, the annual flux of f-PCBs volatilizing from Baltimore Harbor is approximately 12% of the gas-phase f-PCB inventory over the water, and at the northern Chesapeake Bay site, the f-PCB flux is approximately 40% of the gas-phase f-PCB inventory. This suggests that air-water exchange of f-PCBs has the potential to be a significant source of PCBs to the rural atmosphere. Introduction In coastal areas, large urban-industrial centers increase inventories of semivolatile organic contaminants in water through stormwater runoff, treated sewage discharge, riverine inputs, and wet and dry deposition and in air through vehicular and industrial emissions and reversible exchange of gaseous components across the air--water and air--soil * Corresponding author phone: (410)326-7205; fax: (410)326-7341: e-mail: baker@cbl.umces.edu. %| EXHIBIT / 93 CCC: $22.00 2002 American Chemical Society pb 09/19/2002 Deponertf Dnte Rptr interfaces (1-8). The relative importance of each process depends directly upon the source of the contaminant, the physical-chemical properties of the compound, and the micro- and macroscale meteorological conditions of the region. For polychlorinated biphenyls (PCBs), air-water exchange between the atmosphere and surface waters is a dominant transport process, influencing the PCB concentra tions, fate, and residence times in the water and regional atmosphere (9--13). A PCB mass balance study in Lake Michigan reported that the average mean PCB gross gas depositional fluxes were 30 times higher than the average PCB dry deposition flux and that the magnitude and direction of the fluxes varied seasonally (14). In the Chesapeake Bay region, Nelson et al. (15) estimated PCB gas exchange fluxes across the air-water interface of the mainstem of the Bay and determined that the annual loss of PCBs by volatilization was more than 10 times larger than inputs to the bay from wet and diy deposition. PCBs generally volatilize from the Chesapeake Bay throughout the year, demonstrating the importance of air-water exchange as a possible source of PCBs to the regional atmosphere (15). However, elevated atmospheric and water concentrations of PCBs observed around urban-industrial areas, which can lead to enhanced fluxes in coastal urban areas, were not investigated during the previous study. Gas-phase PCB concentrations are significantly higher in Baltimore as compared to over bay and rural locations (16). PCB concentrations in Baltimore Harbor surficial sediments greatly exceed those measured in the mainstem of the Chesapeake Bay (17). While there is no doubt that organic contaminant concentrations are enriched in urban environ ments relative to remote locations, the spatial and temporal extent of the urban influence on air--water exchange in the northern Chesapeake Bay has not been quantified for PCBs. To specifically address the impact of urban areas onnearshore coastal water, the Atmospheric Exchange Over Takes and Oceans Study (AEOLOS) was conducted in Chicago/southern Lake Michigan and Baltimore Harbor/northern Chesapeake Bay (2, 4, 13. 16. 18-23). Polycyclic aromatic hydrocarbon concentrations and air-water exchange fluxes have been reported for the Baltimore Harbor/northern Chesapeake Bay (4, 16). The primary objectives of this study were to determine the magnitude and direction of PCB gas exchange fluxes and to quantify the spatial and seasonal trends in these fluxes in the Baltimore Harbor/northern Chesapeake Bay area. Fur thermore, our goal was to assess the overall contributions of urban source PCBs from Baltimore through air-water exchange and its impact on PCB inventories to these coastal waters and the regional atmosphere. The strategy of this study was (i) to collect simultaneous air and water samples at multiple sites along a transect from the Baltimore Harbor to the northern Chesapeake Bay during seasonal intensive campaigns and (ii) to collect air and water samples at urban and rural locations on a regular basis for 1 yr. A gradient flux model that incorporated gaseous and dissolved concentrations, wind speeds, and temperaturecorrected Henry's law constants was used to estimate airwater exchange fluxes for PCBs. Spatial and daily variations in PCB fluxes were quantified from the samples collected during three sampling intensives, while annual gas exchange fluxes were estimated from samples collected every 9 days between March 1997 and March 1998 at the urban and rural locations. VOL. 36, NO. 20, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 4245 WATER PCB-SD0000046819 FIGURE 1. Map of northern Chesapeake Bay and Baltimore Harbor illustrating air and water collection sites during the three intensive sampling campaigns and annual sampling between March 1997 and March 1998. Experimental Methods Sampling Strategy. Surface water and air samples were collected during the three sampling intensives conducted in the Baltimore Harbor in June 1996, February 1997, and July 1997 (4, 16). Two sets of air--water pairs (urban and rural) were also collected every ninth day between March 1997 and March 1998 (Figure 1). The urban set was collected within the Baltimore Harbor area, and the rural site was collected in the northern Chesapeake Bay region approximately 41 km north-northeast of the city along the eastern shore of the bay and the Susquehanna River (Figure 1). Latitudinal and longitudinal position of each air and water site are listed in Table IS in Supporting Information. During the intensives, consecutive 12-h air samples were collected at each of the sites. The 24-h air samples were collected for one in 9-day samples. Details of air sampling are presented elsewhere (4, 15, 16) and briefly described below. Modified General Metal Works high-volume air samplers were used, which housed a glass fiber filter and polyurethane foam (PUF) plug in series. Surface water samples were collected in 18-L polished stainless steel tanks. The particulate and dissolved phases were separated by pumping the raw water through a glass fiber filter (Schleicher & Schuell no. 25; 0.7 um nominal pore size) and Amberlite XAD-2 macroresin (Sigma) (4). Meteorological data including wind speed, wind direction, and air temperature were collected from the Key Bridge by the Maryland Department of Transportation, Key Bridge Authority (35 m above the water). Analytical Procedures. Details of the analytical proce dures were previously described and will be summarized here (4, 15, 16). Prior to extraction, two PCB congeners not found in industrial syntheses or in the environment (2,3,5,6tetrachlorobiphenyl, IUPAC no. 65, and 2,3,4,4',5,6-hexachlorobiphenyl, IUPAC no. 166) were added as surrogates to each 4246 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 20. 2002 PUF and XAD sample to monitor analytical recoveiy. After Soxhlet extraction, extracts were separated on a packed column of Florisil (60--100 mesh, J. T. Baker Co.) to remove polar interferences (24). The purified extracts were concen trated and analyzed using a Hewlett-Packard 5890 gas chromatography with a DB-5 column Q&W Scientific) and an electron capture detector (GC-ECD). Each sample was analyzed for 55 individual PCB congeners and 28 chromatographically unresolved congener groups. Many of the heavier PCB congeners were not above detection in both the dissolved and the gas phases. Therefore, total PCB (f-PCB) concentra tions only included those congeners that were consistently quantified in both phases, which included 24 individual PCB congeners and 21 chromatographically unresolved congener groups. Internal standards (Ultra Scientific) consisting of 2,3,6-trichlorobiphenyl (IUPAC congener 30) and 2,2'3,4,4', 5,6,6'-octachlorobiphenyl (IUPAC congener 204) were added to each sample prior to instrumental analysis to calculate relative response factors for each congener. Each PCB congener was identified based on its retention time relative to a standard mixture of PCB Aroclors 1232, 1248, and 1262 (Ultra Scientific) whose peaks were identified in Mullin (25). Quality Control and Assurance. For the dissolved-phase samples, surrogate recoveries were 72% 22% (n = 130) for IUPAC no. 65 and 79% 24% (n = 130) for IUPAC no. 166. Vapor-phase samples surrogate recoveries were 71% 16% (n = 90) for IUPAC no. 65 and 89% 23% (n = 90) for IUPAC no. 166. On the basis of fairly consistent percent recoveries among vapor and dissolved samples, PCB congener con centrations in each sample were not surrogate corrected. Laboratory and field blanks were incorporated in the analytical analysis to quantify possible contamination due to collection, transport, and extraction. The detection limits were derived from the blanks and defined as three times the mean concentration in the blank for each PCB congener. WATER PCB-SD0000046820 TABLE 1. Mean and Range of Gaseous and Dissolved PCB Concentrations by Homologue Group and Gas Exchange Fluxes Collected from the Baltimore Harbor and Northern Chesapeake Bay Every 9 Days between March 1997 and March 1998 mean gaseous concn (pg/m3)3 min max mean dissolved concn (pg/L) min max mean flux (ng m-2 day-1)1 min max dichloro trichloro tetrachloro pentachloro hexachloro t-PCBs dichloro trichloro tetrachloro pentachloro hexachloro t-PCBs 43 113 74 51 31 330 5.7 17 17 13 7.8 70 10 24 15 7.8 5.3 67 0.6 0.8 0.64 0.8 0.9 5.1 196 614 230 191 122 1400 Baltimore Harbor 28 1.5 100 8.9 154 39 81 19 81 15 500 100 Northern Chesapeake Bay 37 23 2.0 121 58 21 105 116 46 80 89 40 44 104 70 370 430 246 346 562 442 261 199 1520 196 120 237 279 244 940 +33 +77 +119 +63 +51 +350 +26 +43 +97 +75 +76 +330 -9.4 -4.4 +10 +7.4 +3.3 +19 .+0.94 +6.2 +15 +15 +17 +90 +257 +467 +393 +206 +125 +1240 +236 +107 +304 +240 +235 +1150 ` Averages represent geometric means. 1 Positive flux = net volatilization, and negative flux = net absorption. Detection limits for individual PCB congeners ranged from 0.40 to 90 pg/L in the dissolved samples and 0.003 to 0.60 pg/m3 in the air samples. Results and Discussion Gas-Phase Concentrations. For the June 1996 intensive, the spatial and daily trends of the atmospheric PCB concentra tions in the Baltimore area have been reported elsewhere (16) and are summarized below. PCB gas-phase concentra tions duringJune varied spatially, with higher concentrations in Baltimore city at the Fort McHenry site as compared to the rural site on Hart Miller Island (HMI). Average gas-phase concentrations for t-PCBs at Fort McHenry were 1.3 ng/m3 (n = 4) and ranged between 0.3 and 2.9 n/m3, and at HMI they were 0.39 ng/m3 (n = 4) and ranged between 0.14 and 0.62 ng/m3. During southeasterly winds, the f-PCBs air concentrations at Fort McHenry were the highest (June 9, t-PCBs = 2.9 ng/m3), indicating an emission source in the heavily industrialized section along the south shores of the harbor (16). Similar to June 1996, the f-PCB gas-phase concentrations in the city at the Science Center and Curtis Creek sites were consistently higher than the HMI site during the February and July 1997 sampling intensives. In February, the f-PCB concentrations at both Science Center and Curtis Creek sites were approximately 2.5 times higher than those at HMI. The f-PCB gas-phase concentrations in February at the Science Center and Curtis Creek sites ranged between 0.20 and 0.70 ng/m3, and at HMI they ranged between 0.12 and 0.19 ng/m3. In contrast to February, July f-PCB con centrations measured during the three sample days at Curtis Creek (1.1,1.0,and 1.0ng/m3) were higherthan at the Science Center (0.51, 0.76, and 0.92 ng/m3) and at HMI (0.50, 0.24, and 0.54 ng/m3). Curtis Creek is centrally located along the industrialized section ofthe south shoreline, and volatilization from industrial/terrestrial sources, driven by higher ambient temperatures, may increase f-PCB concentrations (26, 27). During all three intensives, the congener distribution was dominated by the di-, tri-, tetra-, and pentachlorobiphenyls, accounting for approximately 90% of the f-PCB gas con centrations. Overall, gas-phase concentrations in Baltimore were highest in June and lowest in February. Elevated f-PCB gas concentrations in June were partly driven by an increase in air temperature and by a change in wind direction. During the February and July 1997 sampling intensives, the winds were predominantly from the north-northeast direction, meaning air masses from the industrialized area ofBaltimore did not extend over the northern Chesapeake Bay toward HMI. The wind speeds during July were fairly constant, averaging 2 m/s. In February, the winds were more variable, with speeds ranging from 1 to 11 m/s over the water. High wind speeds coming from the north-northeast direction may dilute the f-PCB air concentrations measured around Bal timore, contributing to low gas-phase concentrations during this time of year. Air samples collected every ninth day between March 1997 and March 1998 at the urban Science Center and the rural Stillpond site also showed a spatial gradient in f-PCB gaseous concentrations with significantly higher concentrations (a = 0.01; p < 0.001) at Science Center relative to Stillpond (Table 1). The f-PCB gas-phase concentrations at both sites were log-normally distributed, and the geometric mean of f-PCBs at the Science Center was 0.33 ng/m3 (n = 35) and at Stillpond was 0.07 ng/m3 (n = 28). Significantly higher PCB concentrations at the Science Center is a strong indication of a gas-phase source ofPCBs in Baltimore relative to background and rural locations. Gas-phase f-PCB con centrations from the Science Center are similar to those measured in the northern bay by Nelson et al. (75; range 0.20-0.92 ng/m3). Seasonal variability in f-PCB concentra tions was observed at the Science Center, with higher concentrations measured during the late summer to fall months (geometric mean 0.68 ng/m3 between July and November 1997) as compared to other times of the year (geometric mean 0.20 ng/m3 between December and March). This seasonal trend is consistent with studies conducted in the Great Lakes region, where volatilization of PCBs from industrial/terrestrial sources increases during warmer months (13, 26, 27). The congener distribution measured in air at both sites was similar to the PCB gas distribution during the intensive sampling, with di- through pentachlorobiphenyls accounting for approximately 85% of f-PCB concentrations in air. Dissolved-Phase Concentrations. During all three sam pling campaigns, elevated surface water f-PCB concentrations were measured at site 4 (Middle Branch, 1.2 0.26 ng/L (n = 10)) relative to concentrations measured outside the Key Bridge away from the Inner Harbor (0.63 0.14 ng/L (n = 16)). Site 4 is located near Gwynn's Falls, an urban outfall that discharges directly into Middle Branch. This spatial variability may have resulted from meteorological conditions (high wind speeds in February 1997; rain events in June and February), which resuspended contaminated sediments (17,, 28--30) or increased urban runoff. PCB concentrations in VOL. 36, NO. 20, 2002 1 ENVIRONMENTAL SCIENCE & TECHNOLOGY 4247 WATER PCB-SD0000046821 the sediments of the harbor range from 8 to 2150 ng/g dry weight, with higher concentrations localized around indus trial outflows and urban runoffs (IT). Dissolved f-PCB concentrations did not vary seasonally among the intensives, with harbor concentrations ranging between 0.46 and 1.2 ng/L in June, between 0.57 and 1.4 ng/L in February, and between 0.20 and 1.7 ng/L in July. These concentrations are similar to average dissolved f-PCB concentrations measured in the Chesapeake Bay in 1993 (0.92 0.47 ng/L; 15). Tri-, tetra-, and pentachlorobiphenyl congeners made up approximately 23%, 38%, and 1496, respectively, of the total dissolved PCB concentrations during the June and July intensives. The tri-, tetra-, and pentachlo robiphenyl congeners also dominated the homologue dis tribution during the February sampling intensive, making up approximately 28%, 23%, and 21%, respectively, of the total dissolved PCB concentration. Dissolved-phase f-PCB concentrations collected from site 3 in Baltimore Harbor and in the Susquehanna River every ninth day between March 1997 and March 1998 ranged between 0.10-1.5 and 0.25-0.94 ng/L, respectively, with highest concentrations measured in the spring for both locations (Table 1). Annual average concentrations were statistically similar (a = 0.01; p = 0.12) between site 3 in the harbor (0.5 0.17 ng/L; n = 35) and Susquehanna River (0.42 0.15 ng/L; n = 28) and were within the range of dissolved PCB concentrations measured previously in the northern Chesapeake Bay but higher than those in the southern Chesapeake Bay (15). Compared to other aquatic systems proximate to urban areas, the dissolved PCB concentrations in this study were significantly higher than recently reported for southern Lake Michigan (range 0.08 0.48 ng/L; 13) and Grand Traverse Bay, Lake Michigan (0.12 0.05 ng/L; 31), but lower than those in Raritan Bay (range 1.4-1.8 ng/L; 32) and New York Harbor (range 3.5-4.2 ng/L; 32). The homologue distribution for site 3 in the harbor is similar to the homologue distribution observed during the intensive sampling, where the tri-, tetra-, and pentachloro biphenyl congeners made up approximately 21%, 31%, and 17%, respectively. However, the Susquehanna River is enriched in heavier weight congeners as compared to site 3 in the harbor, with tetra-, penta-, and hexachlorobiphenyl congeners accounting for 28%, 21%, and 25%, respectively, of the total distribution. Air--Water Exchange Model. The modified version of the Whitman two-film resistance model (10, 33) was used to calculate gas exchange rates for each PCB congener mea sured. Details of these calculations have previously been published (4, 15) and are briefly summarized here. The flux of each PCB congener across the air-water interface is modeled using estimates of mass transfer coefficients describing transfer through the gas and liquid phases and the concentration gradient between the paired dissolvedand gas-phase PCB congener measurements: F=Kol(Cd-Ca/KH.) (1) where Fis the flux (ng m-2 day-1); Kol is the overall mass transfer coefficient (m/day); Cd and Ca (ng/m3) are the measured dissolved- and gaseous-phase concentrations; and Kw is the dimensionless, temperature-corrected Hemy's law constant. Kol is empirically derived from field and laboratory studies, based on correlations with wind speed and molecular diffusivities (34), and determined here from the daily wind speed and surface water temperature measured during each sampling period (4,12,13,15,33). The concentration gradient is calculated from the dissolved- and gas-phase PCB con centrations and the temperature-corrected Hemy's law constant. Fluxes were not calculated when the PCB con centration gradient was within two standard deviations of 4248 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 20, 2002 zero. Individual PCB gas and dissolved concentrations, mass transfer coefficients, concentration gradients, flux, and error in the flux for all samples are listed in Tables 2S-6S in Supporting Information. The magnitude and direction ofgas exchange calculations are veiy sensitive to Fm-. Correct assessment of the temper ature dependence of Fm- is essential in order to determine the direction and magnitude of the flux. For some PCB congeners, a 10-fold increase in KH- is measured with a 20 C change (35, 36). Each individual PCB congener was corrected for ambient temperatures using temperature dependence data from Bamford et al. (36): In Kw = -- (AHh/RT) + (ASH/R) (2) where AHh (kj/mol) is the enthalpy of phase change or transfer across the air--water interface, ASh (kj mol-1 K-1) is the entropy of phase change, Fis the ideal gas constant (8.314 J mol-1 K"'),and Fis temperature (K). Individual AHh values for PCBs that were experimentally measured or derived (36) were used. The previous study of PCB fluxes by Nelson et al. (15) estimated Fh- using an empirical relationship derived by Brunner et al. (37) and corrected to ambient temperatures using an average AHh for all PCB congeners. The differences in the fluxes due to different FH- values and AHh used in each study were evaluated by recalculating the fluxes from Nelson et al. (15) using the new temperature-corrected FH- values used in this study. The recalculated fluxes of individual PCB congeners yielded higher net volatilization for most congeners because FH- for many PCB congeners varies more with temperature than previously thought. For example, the recalculated flux for PCB congener 52 from the northern Chesapeake Bay during March 1993 changed from 16 to 33 ng m-2 day-1 net volatilization. The differences among the AHh ofindividual congeners is as high as 91 %, demonstrating the considerable variability in the Fh- temperature depen dence among individual congeners (36). Therefore, it was necessary to use AHh of individual congeners to accurately predict the temperature-specific Fh- of PCB congeners. Uncertainty in Gas Exchange Fluxes. The uncertainty in each of the fluxes was evaluated using a propagation of error analysis derived from Shoemaker et al. (38), which has been used in previous studies (4, 15, 23). For each congener (or congener groups), random measurement errors in Ca and Cd and in Fol are taken into account. The summation of the various random errors in the flux are described by where A is the coefficient of error. The values of A for Cd and Ca, determined by replicate samples and the relative standard deviations of sample surrogate recoveries, were 15% and 20%, respectively. The A for Kol was estimated by the uncertainty in the air and water mass transfer coefficients. Because of the large uncertainties in the air and water mass transfer coefficients (34) and the nonlinear effect of wind speed on the water mass transfer coefficient (39), A was assumed to be 40% although it may be as high as 200% (4, 15). The Fh- term was removed from the overall error analysis because errors in Fh- from incorrect temperature corrections are assumed to be systematic and not random errors. For PCB congeners, the estimated average uncertainty in the fluxes were 46%. Most of the uncertainty associated with the fluxes was attributed to Fol (74%) and the uncertainties due to the effect variations in wind speed have on the water side mass transfer coefficient (39). The uncertainty in the flux calculations for each congener are listed in Tables 2S-- 6S in Supporting Information. WATER PCB-SD0000046822 FIGURE 2. Gaseous and dissolved f-PCB concentrations and daily gas exchange rates measured during (A) June 1996, (B) February 1997, and (C) July 1997. Error bars represent the overall propagated error associated with the flux. An asterisk (*) indicates that no data were available. Gas Exchange Fluxes During Intensive Sampling. For the three intensives, instantaneous fluxes were calculated based on PCB dissolved concentrations measured at each of the six sites, which were paired with gas-phase concentrations interpolated by distance from the shoreline air sites. During all three intensives, f-PCBs volatilized from the surface waters to the atmosphere (positive fluxes) at all sampling sites along the harbor (Figure 2) because of relatively high dissolved PCB concentrations as compared to the gas phase. Despite elevated PCB gas-phase concentrations measured in Balti more in June, the concentration gradient favored net PCB volatilization. By homologue, tri- through pentachlorobiphenyls accounted for approximately 80% of the total fluxes. The magnitude of the PCB fluxes varied widely on daily, seasonal, and spatial scales in the harbor during all three intensives. The largest net volatilization flux occurred in Baltimore's Inner Harbor (site 6) during the February intensive, with f-PCB flux reaching2120ngm_2day_1 (Figure 2B). Despite dissolved- and gas-phase concentrations re maining relatively constant over the Februaiy sampling period, the magnitude of the flux was driven by high wind speeds recorded on February 22. Increased wind speeds from an average of 1 m/s on Februaiy 20 to 11 m/s on Februaiy 22 resulted in a 40-fold increase in the mass transfer coefficient for f-PCBs and a 57-fold increase in instantaneous f-PCBs fluxes measured in the Inner Harbor at site 6 (Figure 2B). On both February 18 and 20, Kol was < 0.2, resulting in low gas exchange fluxes relative to Februaiy 22. The impact of wind speed on the magnitude of f-PCB fluxes was also observed during the June intensive. Highest net volatilization fluxes (910 ng nr2 day"` on June 9) were calculated on days when Kol increased above 0.6 m/day (Figure 2A). The majority of the PCB fluxes were controlled by the water-side mass transfer, and wind speed has a nonlinear effect on the water side mass transfer coefficient. Turbulent mixing at the airwater interface facilitates the total mass transfer, increasing the magnitude of the fluxes. The lowest net volatilization flux of 5 ng m-2 day-1 was estimated during the July sampling intensive when veiy little turbulent mixing took place at the air--water interface. A spatial gradient was observed in PCB fluxes during the June intensive, with smallest net volatilization fluxes typically measured outside the Key Bridge (sites 1--3) and increasing around sites 4-6. Volatilization fluxes also increased at site 4 near Gwynn's Falls during the July intensive (Figure 2A). This trend was driven by the changes in the measured dissolved concentrations at these sites, possibly due to urban runoff discharged directly into the harbor or resuspension of contaminated sediments (17). Overall, there were large variations in the gas exchange fluxes for PCBs on both a temporal and a spatial scale. PCB fluxes ranged between 20 and 910 ng nr 2 day-1 during the four days in June, between 10 and 2120 ng nr 2 day-1 during the three days in Februaiy, and between 5 and 480 ng m-2 day-1 during the three days in July. Annual Gas Exchange Fluxes. Annual air--water exchange fluxes were calculated from samples taken eveiy 9 days between March 1997 and March 1998. The urban Baltimore Harbor sampling pair consists of the dissolved PCB con centrations measured at site 3 in the harbor and the PCB gaseous concentrations measured at the Science Center. The rural northern Chesapeake Bay sampling pair consists of the dissolved PCB concentrations measured from the Susque hanna River and the air samples collected at the Stillpond eastern shore site. The 35 and 28 air and water pairs were used to calculate the fluxes for the Baltimore Harbor and northern Chesapeake Bay sites, respectively. Throughout the year, PCB volatilized from the surface waters at both the Baltimore Harbor and the northern Chesapeake Bay sites (Figure 3). The direction of the fluxes were strongly influenced by elevated dissolved PCB con centrations that remained relatively constant throughout the VOL. 36, NO. 20, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 4249 WATER PCB-SD0000046823 Mar 07 Apr 87 May 87 Jun 87 Jul 87 Aug 87 Sep 87 Oct 87 New 87 Dec 87 Jan 88 Feb 66 Mar 86 Mar 97 Apr 97 May 97 Jun 97 Jut 97 Aug 97 Sep 97 Od 97 Nqv 97 Dec 97 Jan 98 Feb 98 Mar 98 FIGURE 3. (-PCB concentrations in the air and water, mass transfer coefficients (MTC), surface water temperatures, and daily fluxes measured every 9 days between March 1997 and March 1998 in (A) Baltimore Harbor and (B) northern Chesapeake Bay. Error bars represent the overall propagated error associated with the flux. year as compared to the gas-phase PCB concentrations. Despite increased PCB gas-phase concentrations measured during the summer months in Baltimore, net fluxes were positive, driven by the water-side gradient due to super saturation of dissolved PCBs. The largest concentration gradient in favor of net volatilization was measured on October 29, 1997, at the northern Chesapeake Bay site and on January 27, 1998, in Baltimore Harbor resuldng in the highest net fluxes measured during the year at both sites. To quantify the significance of the concentradon gradient, the total net flux was divided into its gross volatilization and deposition fluxes and compared individually. Throughout the year, PCB gross volatilization and gross gas deposidonal fluxes in Baltimore Harbor ranged from +180 to +1300 ng m-2 day-1 and from -6 to --180 ng m-2 day-1, respectively (Figure 4). At the northern Chesapeake Bay site, the PCB gross volatilization and gross gas deposition fluxes ranged from +92 to +1200 and from -1 to --93 ng m-2 day-1, respectively. The ratio of volatilization fluxes to deposidonal fluxes ranged between 2 and 110 at Baltimore Harbor and from 5 to 400 at northern Chesapeake Bay, demonstrating the dominance ofthe water-side concentrations in the overall concentration gradient. Lower PCB volatilization fluxes occurred during the late summer months betweenJuly and October 1997 in Baltimore Harbor due to elevated PCB gas concentrations in the city during the summer as well as lower recorded wind speeds during these months. A seasonal trend was not observed at northern Chesapeake Bay where both air and water con centrations remained relatively constant throughout the year. Overall, f-PCBs were volatilizing from the surface water throughout the year at both Baltimore Harbor and northern Chesapeake Bay locations, and a seasonal trend in PCB fluxes was observed in Baltimore Harbor with lower net volatiliza tion fluxes occurring in the late summer months. Mean daily fluxes were calculated for f-PCBs by temporally extrapolating among the fluxes calculated every ninth day between March 1997 and March 1998. Assuming that each flux represented the 4 days before and 4 days after the sample was taken, the annual average daily fluxes were 350 ng m-2 day-1 volatilization for f-PCB for Baltimore Harbor and 330 ng m-2 day-1 volatilization for northern Chesapeake Bay (Table 1). Multiplying the average daily fluxes by 365 days, the annual PCB gas exchange fluxes from the surface waters to the atmosphere are 130 and 120gm-2yr-1 for Baltimore Harbor and northern Chesapeake Bay, respectively. Com parable annual average f-PCB fluxes in Baltimore Harbor and northern Chesapeake Bay locations does not imply similar concentrations of f-PCBs in the air and water. The atmosphere in Baltimore is enriched in f-PCB during the summer months, reducing net volatilization fluxes in the summer, which results in a lower annual average similar to that measured in northern Bay. Average daily f-PCB fluxes at both Baltimore Harbor and northern Chesapeake Bay are higher than the average flux recalculated from Nelson et al. (15) for the mainstem of the Chesapeake Bay (+240 ng m-2 day-1) but similar in both direction and magnitude to the average flux recalculated from Nelson et al. (15) for the northern Chesapeake Bay (+390 ng m-2 day-1). The urban coastal region around Baltimore contributes to elevated f-PCB dissolved concentrations, generating higher net volatilization fluxes in the harbor and northern bay as compared to the 4250 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 20, 2002 WATER PCB-SD0000046824 Month FIGURE 4. Absorption, volatilization, and net gas exchange flux of f-PCBs in Baltimore Harbor between March 1997 and March 1998. Net fluxes are positive, dominated by water-side concentration gradient mainstem bay average. These fluxes are also higher than the 63 ng m-2 day-1 estimated for Lake Superior in 1992 (11) and the range of -32 to +59 ng m-2 day-1 estimated for southern Lake Michigan in 1994-1995 (13). Average f-PCB fluxes from Raritan Bay and New York Harbor in 1998 were +400 and +2100 ngm-2 day-1, respectively (32), which are higher than both Baltimore Harbor and northern Chesapeake Bay fluxes. On average, the congener distribution of the annual PCB volatilization fluxes from Baltimore Harbor were similar to the congener distribution of the intensive fluxes, with tri(20%), tetra- (35%), penta- (19%), and hexachlorobiphenyl (17%) accounting for approximately 90% of the total flux. The tri-, tetra-, penta-, and hexachlorobiphenyls also ac counted for approximately 90% of the total flux for northern Chesapeake Bay. However, unlike Baltimore Harbor, the contribution of the hexachlorobiphenyls to the total was higher than that of the trichlorobiphenyls. The total flux at northern Chesapeake Bay was approximately 29% tetra-, 23% penta-, and 23% hexa-, and the trichlorobiphenyls only accounted for 14% of the total flux. Congener distribution of volatilization fluxes at both sites followed the PCB distribution in the dissolved phase (see Figure IS in Supporting Informa tion). Implications of Air-Water Exchange to Coastal PCB Cycling. To place the magnitude ofthese fluxes in perspective, the estimated daily fluxes of PCBs from Baltimore Harbor were compared to the dissolved PCB inventory in the water column. With an average water depth of 7 m and an average dissolved f-PCB concentration of0.5 ng/L, the dissolved f-PCB inventory for this site is 3500 ng/m2. Comparing this to the average daily f-PCB flux (350 ng m-2 day-1), approximately 10% of the dissolved f-PCB inventory exchanges with the atmosphere on a daily basis. If air-water exchange was the only loss process for f-PCBs from the water column, the residence time of f-PCBs in the Baltimore Harbor is 10 days. Totten et al. (32) estimated the residence time for PCBs in Raritan Bay via air-water exchange to range between 14 and 87 days, depending on the variations in PCB dissolved concentrations and wind speeds. The estimated residence time of f-PCBs in the Baltimore Harbor is similar to the hydraulic residence time of water in the harbor, which is estimated to be between 7 and 14 days (40-42). This suggests that removal of f-PCBs via air--water exchange is an important loss process of these contaminants from coastal waters. Furthermore, it also suggests that additional sources of PCBs to the Baltimore Harbor are replenishing the dissolved PCB concentrations given that the dissolved f-PCB concentration remains relatively constant throughout the year. The Bal timore Harbor is a highly dynamic shallow system with a thin layer ofwater positioned between two large geochemical reservoirs of contaminants, atmosphere, and sediment (40). During high wind conditions, resuspension of contaminated sediment (17) may support water column f-PCB inventories, and during precipitation events f-PCBs may enter the water column via wet deposition (IS) or by two major stormwater outfalls where urban runoff directly enters into the harbor. The surface waters are enriched with f-PCBs relative to the overlying atmosphere, resulting in net PCB volatilization throughout the year where the regional atmosphere is a sink for PCBs from Baltimore Harbor. The annual flux of PCBs from the Baltimore Harbor site was also compared to the gas-phase f-PCB inventory to estimate the overall contribution of air--water exchange in controlling the gas-phase PCB concentrations in the atmo sphere over the Baltimore Harbor using a simple well-mixed box model. Simple box models do not take into account complex mixing between the surface layer and higher altitudes, and the resulting predictions are in turn rough estimates, which vary depending on the environmental parameters used in the model. However, these models are quite useful as a first approach to predicting pollutant concentrations in an air mass over a region. We assumed a 1 km atmospheric mixing height, a 20 km (the average length of the harbor) by 3 km water surface over the Baltimore Harbor (the average width of the harbor), and that the only source of gas-phase f-PCBs to this area is from air-water exchange. With a wind speed of 2 m/s perpendicular to the box and an annual PCB volatilization rate of 350 ng m-2 day-1, the calculated steady-state air concentration in the box is 0.04 ng/m3 from air-water exchange. This is ap proximately 12% of the actual f-PCB gas-phase concentration (0.33 ng/m3) measured in Baltimore. In the city where contributions of gas-phase PCBs from industrial/terrestrial sources exists, the contribution of gas-phase PCBs concen tration from air--water exchange is relatively low. Using the box model for the northern Chesapeake Bay site away from local sources, the steady-state air concentration from airwater exchange is 0.03 ng/m3. Comparing this to the average VOL. 36, NO. 20, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 4251 WATER PCB-SD0000046825 PCB gas-phase concentration over the northern Chesapeake Bay (0.07 ng/m3), air--water exchange accounts for ap proximately 40% of the actual f-PCB gas-phase concentration, suggesting that air-water has the potential to be a significant source of PCBs to the rural atmosphere. The exchange of PCBs from the water column to the overlying atmosphere is a relatively rapid process, resulting in the efficient exchange of PCBs between these two reservoirs. With a short residence time in the water column, this is an extremely acdve system where air-water exchange is an important loss mechanism controlling the transport, fate, and removal of f-PCBs from these urban coastal waters. It also has the potential to be a significant source of f-PCBs to the rural atmosphere where local emissions do not exist. Acknowledgments We express our appreciation to Steve Eisenreich for guiding the AEOLOS project and to the crew of the U.S. EPA OSV Anderson for their help in sample collection. We acknowledge Rick Hebert, Dan Liebert, and Carrie Dennenfelser for their laboratory assistance and Randolph Larsen and John Offenberg for their assistance with air analysis. The AEOLOS project was funded by a grant from the U.S. Environmental Protection Agency (EPA CR 822046-01-0). This is the Uni versity of Maryland Center for Environmental Science Contribution No. 3601. Supporting Information Available Table IS lists air and water sampling site descriptions; Tables 2S-6S list the PCB temperature-corrected Henry's law constant, gas and dissolved concentrations, mass transfer coefficients, concentration gradient, flux, and uncertainty in the flux for each congener for samples collected during the three intensives and the Baltimore and northern Chesapeake Bay annual sampling; Figure IS illustrates the average percent total of gas phase, dissolved phase, and net flux for PCBs by homologue groups in Baltimore and northern Chesapeake Bay. This material is available free of charge via the Internet at http://pubs.acs.org. Literature Cited (1) Leister, D. L.; Baker, J. E. Atmos. Environ. 1994, 28, 1499--1520. (2) Simcik, M. F.; Zhang, H. X.; Eisenreich, S.J.; Franz,T. P. Environ. Sci. Technol. 1997, 31, 2141-2147. (3) Cousins, I. T.: Jones. K. C. Environ. Pollut. 1998, 102, 105-118. (4) Bamford, H. A.; Offenberg, J. H.; Larsen, R. K.; Ko, F. C.; Baker, J. E. Environ. Sci. Technol. 1999, 33, 2138--2144. (5) Foster, G. D.; Roberts, E. C.; Gruessner, B.; Velinsky, J. Appl. Geochem. 2000, 15, 901-915. (6) Liu, M.; Baugh, P. J.; Hutchinson, S. M.; Yu, L.; Xu, S. Environ. Pollut. 2000, 110, 357-365. (7) Ngabe, B.; Bidleman, T. F.: Scott, G. I. Sci. Total Environ. 2000, 255, 1-9. (8) Wong, P. K.: Wang, J. Environ. Pollut. 2001. 112, 407--415. (9) Baker, J. E.; Eisenreich, S. J. Environ. Sci. Technol. 1990, 24. 342-352. (10) Achman, D. R.; Hornbuckle. K. C.; Eisenreich, S. ]. Environ. Sci. Technol. 1993, 27, 75-86. (11) Jeremiason, J. D.; Hornbuckle, K. C.; Eisenreich, S. J. Environ. Sci. Technol. 1994, 28, 903-914. (12) Hornbuckle, K. C.: Sweet, C. W.; Pearson, R. F.; Swackhamer, D. L. Environ. Sci. Technol. 1995, 29. 869-877. (13) Zhang, H.X.; Eisenreich, S.J.; Franz,T. R.; Baker,J. E.; Offenberg, J. H. Environ. Sci. Technol. 1999, 33, 2129--2137. (14) Miller, S. M.; Green, M. L.; Depinto, J. V.; Hornbuckle, K. C. Environ. Sci. Technol. 2001, 35, 278--285. (15) Nelson. E. D.; McConnell, L. L.: Baker. J. E. Environ. Sci. Technol. 1998, 32, 912-919. (16) Offenberg, J.; Baker, J. E. J. Air Waste Manage. Assoc. 1999, 49, 959-965. (17) Ashley, J. T. F.: Baker, J. E. Environ. Toxicol. Chem. 1999. 18. 838-849, (18) Offenberg, J. H.; Baker, J. E. Environ. Sci. Technol. 1997, 31, 1534-1538. (19) Simcik, M. F.; Franz. T. P.; Zhang, H. X.; Eisenreich, S. J. Environ. Sci. Technol. 1998, 32, 251-257. (20) Paode, R. D.; Sofuoglu. S. C.; Sivadechathep,J.; Noll, K. E.; Holsen, T. M.; Keeler, G. J. Environ. Sci. Technol. 1998, 32. 1629-1635, (21) Sofuoglu, S. C.: Paode,R. D.; Sivadechathep,J.; Noll, K. E.; Holsen, T. M.; Keeler, G. J. Aerosol Sci. Technol. 1998, 29, 281--293. (22) Offenberg, J. H.; Baker, J. E. Atmos. Environ. 2000, 34. 1509-- 1517. (23) Offenberg, J.; Baker, J. E. Atmos. Environ. 2002, 36, 1205-1220. (24) Kucklick,J. R.; Haivey, H. R.; Ostrom. R. H.; Ostrom, N. E.; Baker, J. E. Environ. Toxicol. Chem. 1996, 15, 1388--1400. (25) Mullin. M. D. PCB Workshop, U.S. EPA Large Lakes Research Station. 1985. (26) Hoff, R. M.; Brice, K. A.; Halsall, C. J. Environ. Sci. Technol. 1998, 32, 1793-1798. (27) Wania, F.; Haugen, J. E.: Lei, Y. D.; Mackay, D. Environ. Sci. Technol. 1998. 32. 1013-1021. (28) Ko, F. C. Particle-Related Biogeochemical Cycles ofHydrophobic Organic Contaminants in Estuarine Waters: University of Maryland: College Park, MD, 1994. (29) Sanford, L. P. Estuaries 1994, 17, 148--165. (30) Sanford. L. P.; Maa, J. P. Y. Mar. Geol. 2001, 179. 9-23. (31) Cohen, A. R. Episodic Particle Events and PAHand PCB Cycling in Lake Michigan; University of Maryland: College Park, MD, 2001. (32) Totten, L. A.; Brunciak, P. A.; Gigliotti, C. L.; Dachs, J.; Glenn, T. R.: Nelson. E. D.; Eisenreich, S. J. Environ. Sci. Technol. 2001, 35, 3834-3840. (33) Hornbuckle, K. C.: Jeremiason, J. D.; Sweet, C. W.; Eisenreich, S. J. Environ. Sci. Technol. 1994, 28, 1491-1501. (34) Wanninkohf, R.; Ledwell, F. R.; Crusius, J. Gas Transfer Velocities on Lakes Measured with SFn: Wilhelm, S. C., Gulliver, J. S., Eds.; American Society of Civil Engineers: New York, 1991; pp 441 -- 458. (35) Bamford, H. A.; Poster, D. L.; Baker, J. E. J. Chem. Eng. Data 2000, 45, 1069-1074. (36) Bamford, H. A.; Poster, D. L.; Huie, R. E.; Baker, J. E. Environ. Sci. Technol. 2002, 36, 4395--4402. (37) Brunner, S.; Hornung, E.; Santl, S.; Wolff, E.; Pringer, 0. G.; Altschuh, J.; Bruggemann, R. Environ. Sci. Technol. 1990, 24, 1751-1754. (38) Shoemaker, D. P.; Garland. G. W.; Steinfeld, J. I. Propagation of Errors, 3rd ed.; McGraw-Hill: New York, 1974; pp 51-58. (39) Wanninkohf, R. J. Geophys. Res. 1992. 97. 7373--7382. (40) Chao, S. Y.; Boicourt. W. C.; Wang, H. V. C. Cont. ShelfRes. 1996, 16. 1379-1397. (41) Boicourt, W. C.; Olson, P. A Hydrodynamic Study of the Baltimore Harbor System. Chesapeake Bay Institute, 1982. (42) Sanford, L. P.; Boicourt, W. C.; Rives. S. R. J. Waterway Port Coastal Ocean Eng. 1992, 118. 635--654. Received for review April 17, 2002. Revised manuscript re ceived July 26, 2002. Accepted July 31. 2002. ES0206893 4252 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 20, 2002 WATER PCB-SD0000046826