Reevaluation of Air−Water Exchange Fluxes of PCBs in Green Bay

This paper examines the importance of air−water exchange in two systems: Green Bay (GB), a long shallow embayment in NW Lake Michigan, and southern ...
0 downloads 0 Views 92KB Size
Environ. Sci. Technol. 2003, 37, 1739-1743

Reevaluation of Air-Water Exchange Fluxes of PCBs in Green Bay and Southern Lake Michigan L I S A A . T O T T E N , * ,† C A R I L . G I G L I O T T I , † JOHN H. OFFENBERG,† JOEL E. BAKER,‡ AND S T E V E N J . E I S E N R E I C H †,§ Department of Environmental Sciences, Rutgers University, 14 College Farm Road, New Brunswick, New Jersey 08901, Chesapeake Biological Laboratory, University of Maryland, Solomons, Maryland 20688, and Institute for Environment and Sustainability, European Commission, Joint Research Centre, TP 290, I-21020 Ispra (VA), Italy

Air-water exchange of persistent organic pollutants (POPs) such as polychlorinated biphenyls (PCBs) is an import process for the delivery of contaminants to water bodies, as well as for their removal, and is thus a pivotal parameter in the construction of mass balances in aquatic systems. Simultaneous measurements of PCB concentrations in the gas and dissolved phases conducted in Green Bay in 1989 and in southern Lake Michigan in 19941995 were used to estimate air-water exchange fluxes. In this work, improved Henry’s law constants for PCBs and new mass-transfer rates across the air-water interface were used to update the previous calculations. The new model calculations suggest that the net volatilization flux of PCBs out of Green Bay ranges from +170 to +5300 ng m-2 day-1, which is 2-20 times larger than previous estimates. The flux of PCBs in southern Lake Michigan exhibits net volatilization of +0.5 to +230 ng m-2 day-1 throughout the study period (May and July 1994, January 1995), whereas previous estimates reported that the net flux was seasonally absorptive. Thus, water-to-air fluxes are more important for the removal of PCBs from both Green Bay and Lake Michigan than previously recognized.

Introduction Air-water exchange of persistent organic pollutants (POPs) such as polychlorinated biphenyls (PCBs) is an import process for the delivery and removal of some of these contaminants to water bodies. Because of the importance of air-water exchange relative to other inputs (such as dry particle and wet deposition) and outputs (advection, burial), quantification of this process is crucial in building ecosystem mass balances and in choosing appropriate management strategies. This paper examines the importance of air-water exchange in two systems: Green Bay (GB), a long shallow embayment in NW Lake Michigan, and southern Lake Michigan (LM) near the urban/industrial areas of Chicago, * Corresponding author phone: (732)932-9588; fax: (732)932-8644; e-mail: [email protected]. † Rutgers University. ‡ University of Maryland. § Institute for Environment and Sustainability. 10.1021/es026093x CCC: $25.00 Published on Web 03/21/2003

 2003 American Chemical Society

IL, and Gary, IN. GB has been the object of much study, including the GB Mass Balance (GBMB) project, because of the historical contamination from 13 paper mills and five major municipal wastewater treatment facilities on the banks of the Fox River, which empties into the bay (1). Air-water exchange of PCBs was studied in southern LM as part of the Atmospheric Exchange Over Lakes and Oceans (AEOLOS) study. The hypothesis of the AEOLOS project is that enhanced emissions of hazardous air pollutants into the urban atmosphere increase atmospheric deposition to adjacent water bodies. Results from AEOLOS also contributed to the LM Mass Balance (LMMB) study. Calculation of air-water exchange requires simultaneous measurements of PCB concentrations in the gaseous and dissolved phases. Such measurements were conducted in GB as part of the GBMB study in 1989 and in southern LM as part of AEOLOS in 1994-1995. These data were used to calculate air-water exchange fluxes reported by the original researchers (2, 3). Since then, improved and different Henry’s law constants for PCBs have been reported by Bamford et al. (4, 5), and new relationships for estimating mass-transfer rates across the air-water interface have been published (6). Incorporation of these new approaches into calculations of air-water exchange in LM and GB results in significant changes in both the direction and magnitude of the estimated fluxes. Because these fluxes have a large impact on the overall mass balances in these systems, recalculation of the fluxes is warranted. The goals of this study are to estimate improved air-water exchange fluxes for PCBs in GB and LM, to compare these fluxes to the original results from Achman et al. (2) in GB and from Zhang et al. (3) in southern LM, and to assess the relative importance of air-water exchange.

Methods Details of sample collection and analysis, as well as raw data (sample concentrations, meteorological conditions), can be found in refs 2, 3, and 7-10. This section will focus on the differences between the previous calculations of air-water exchange fluxes of PCBs and the present reevaluation based on the same data set. Air-water exchange was calculated for all congeners for which both gas-phase and dissolved-phase concentrations were available (Supporting Information, Table 1). This list of congeners is referred to as “ΣPCBs” in both GB and LM, but it must be recognized that the list differs slightly between the two systems. In both cases, the list of congeners utilized probably represents at least 90% of the total PCB mass, and it is the same in this work as in the previously published air-water exchange calculations, allowing meaningful comparisons to be made. A modified two-layer model used here and in many previous studies (2, 3, 11-13) assumes that the rate of gas transfer is controlled by the compound’s ability to diffuse across the water and air layers on either side of the airwater interface. The volatilization and absorption fluxes (ng m-2 day-1) are calculated as

volatilization ) KOLCd

(1)

absorption ) KOLCa/H′

(2)

where KOL (m day-1) is the overall mass-transfer coefficient; Cd (ng m-3) is the dissolved-phase concentration of the compound in water; and Ca (ng m-3) is the gas-phase concentration of the compound in air which is divided by the “dimensionless” Henry’s law constant, H′, which has units VOL. 37, NO. 9, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1739

TABLE 1. Comparison of Net Air-Water Exchange ΣPCB Fluxes from the Literature net flux (ng m-2 day-1)

location

+170 to +5300 +13 to +1300 Lake Michigan (July 1994) +0.5 to +230 -32 to +59 New York Harbor (July 1998) +2100 Raritan Bay (July 1998) +400 Lake Superior (summer) +14.5 Chesapeake Bay (annual) +96 Lake Michigan (annual) -8.6

Green Bay (1989)

ref this work Achman et al. (2) this work Zhang et al. (3) Totten et al. (22) Totten et al. (22) Hoff et al. (28) Nelson et al. (12) Zhang et al. (3)

of Lwater Lair-1. The net gas-exchange flux (F) is then calculated by subtracting the volatilization flux from the absorption flux as

(

F ) KOL Cd -

Ca H′

)

(3)

The term (Cd - Ca/H′) describes the fugacity gradient (ng m-3). A positive (+) flux indicates net volatilization out of the water column, and a negative (-) flux indicates net absorption into the water column. KOL comprises resistances to mass transfer in both the water (ka) and air (kw)

1 1 1 ) + KOL kw kaH′

(4)

The coefficients ka and kw have been empirically defined on the basis of experimental studies using tracer gases (see refs 6 and 14 for a review). Differences in diffusivity (D) or Schmidt number (Sc) between these gases and PCBs were then used to estimate ka and kw for PCB congeners. A correction for the temperature (T) dependence of the Schmidt number for CO2 was employed here as well as by Zhang et al. (3) but not by Achman et al. (2), who assumed that ScCO2 ) 600 at all T. In the present study, as well as in the previous ones, the same relationship for ka was used. In contrast, different relations for kw were used. Achman et al. (2) invoked the relationship suggested by Liss and Merlivat in 1986 (15) for air-water exchange calculations from the GB data

kw ) 0.17u10 kw ) 2.85u10 - 9.65

for u10 < 3.6 m/s for 3.6 < u10 < 13 m/s

kw ) 5.9u10 - 49.3

for u10 > 13 m/s

(5)

In this work as well as in the work of Zhang et al. (3), the Wanninkhoff equation (6) was used:

kw,CO2 ) 0.45u101.64

(6)

At wind speeds from 0 to about 7 m s-1, this equation yields higher values of kw than the Liss and Merlivat prediction, with the maximum difference (a factor of 6) occurring at u10 ) 3.6 m s-1. Achman et al. (2) used the arithmetic mean wind speed to calculate air-water exchange. Zhang et al. (3) used a Weibull distribution (16) to calculate representative wind speeds over LM and observed that use of the Weibull distribution led to an increase in kw of up to 50%. This work has utilized the Weibull distributed wind speeds of Zhang (7) for the calculations of air-water exchange in southern LM. Unfortunately, the wind-speed data necessary to derive the Weibull distribution for the GB work are not available. 1740

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 9, 2003

Henry’s Law Constants Calculation of air/water exchange fluxes requires accurate values of H for each PCB congener, as well as the temperature dependence of H (∆HH) so that H can be calculated at any T. Bamford et al. (4, 5) recently measured H and ∆HH for many congeners and used the experimental data to estimate these parameters for the remaining congeners. Because the database developed by Bamford et al. relies on experimental values, it is felt to be a significant improvement over previously published values and was used herein. Zhang et al. (3) and Achman et al. (2) utilized the H values of Brunner (17). With the exception of PCBs 6 and 18, Brunner’s H values are lower than Bamford’s by factors of up to 56. The higher H values of Bamford will, in general, cause the calculated net flux to be more positive (greater volatilization). Achman et al. (2) previously used the temperature dependence of H for PCBs established by Tateya (18), which assumes that ∆HH is constant (28.4 kJ mol-1) for all 209 PCB congeners. Zhang et al. (3) used a constant ∆HH value of 49.9 kJ mol-1, based on measurements for three PCB congeners by ten Hulscher et al. (19). Bamford et al. (5) report ∆HH values ranging from 12 to 171 kJ mol-1 for all PCB congeners, with 170 of the 209 having ∆HH values greater than the Tateya value. (The congeners having ∆HH less than 28.4 kJ mol-1 are tetra- and pentachloro PCBs.) For the temperature range experienced in this study (roughly 0-30 °C), the difference in ∆HH can result in large differences in H values (see Supporting Information for more details). Use of the Tateya relationship will typically cause H, and therefore the net flux, to be overestimated (more positive) at low temperature (winter) and underestimated (less positive) at high temperature (summer). A final difference between the previous calculations of Achman et al. (2) and Zhang et al. (7) and the present work is the use of air temperature instead of water temperature to represent the temperature of the air-water interface. Experience in the Hudson River Estuary has demonstrated that surface skin temperature measured by remote sensing in the IR band displays a better correlation with air temperature than with bulk water temperature (20), and the data from southern LM further support this conclusion (21). Thus, it was felt that the air temperatures recorded aboard the research vessels during both the LM and GB studies were the most appropriate temperatures for use in calculating airwater exchange. Air temperature tends to be more variable than bulk water temperature, suggesting that the magnitude of air-water exchange flux is a stronger function of daily meteorological conditions than previously recognized.

Results and Discussion Green Bay. ΣPCBs ranged from 0.25 to 2.29 ng m-3 in the gas phase and from 0.35 to 7.8 ng L-3 in the dissolved phase and generally decreased from the mouth of the Fox River northward. In their original evaluation of the data, Achman et al. (2) noted that a correction for PCB sorption to colloidal organic matter was not warranted, and thus, no correction was employed here. Were such a correction to be employed, it would decrease the volatilization fluxes of the di-, tri-, tetra-, and pentachloro PCBs by about 5, 10, 20, and 45%, respectively. Congeners with 6 or more chlorines typically represent