Use of the Relative Concentration To Evaluate a Multimedia Model for

With no confident emission information, a dynamic multimedia model (POPsME) was evaluated by comparing predicted and measured relative concentrations ...
0 downloads 5 Views 228KB Size
Download PDF
Environ. Sci. Technol. 2004, 38, 1079-1088

Use of the Relative Concentration To Evaluate a Multimedia Model for PAHs in the Absence of Emission Estimates YUNAH LEE, DONG SOO LEE,* SEUNG-KYU KIM, YOON KWAN KIM, AND DONG WON KIM Graduate School of Environmental Studies, Seoul National University, Kwanak ku, Shilimdong San 56-1, Seoul 151-742, Korea

With no confident emission information, a dynamic multimedia model (POPsME) was evaluated by comparing predicted and measured relative concentrations (defined as the ratio of concentration in a medium to that in soil) for 12 polycyclic aromatic hydrocarbons (PAHs) (phenanthrene, anthracene, fluoranthene, pyrene, chrysene, benz[a]anthracene, benzo[b]fluoranthene, benzo[k]fluoranthene, benzo[a]pyrene, dibenz[a,h]anthracene, benzo[g,h,i]perylene, and indeno[1,2,3-c,d]pyrene). Field monitoring for multimedia (air particulates, water (dissolved phase and suspended solids), soil, sediment, and leaves of Pinus koraiensis and Prunus serrulata) was conducted seasonally over 1 yr from August 1999 to July 2000 at seven sites in Seoul and neighboring areas (150 km × 150 km) in Korea. The model was calibrated using the monitoring data of four PAHs (phenanthrene, pyrene, benzo[a]pyrene, and benzo[g,h,i]perylene) and tested with the remaining eight PAHs. For Csuspended solids/Csoil and Csediment/Csoil (Ci, concentration (mol/m3) in medium i), the model prediction changed with octanol-water partition coefficient (Kow) from underestimation to overestimation, revealing a limitation to the use of sorption equilibrium assumption and/or Kow for the description of sorption. Nonetheless, the model prediction generally agreed with the measured within a factor of 10 for all the monitored media. The relative concentration was shown to be a useful means to evaluate and improve the model.

Introduction Multimedia models (1) have been developed and used for exposure assessment, chemical ranking and scoring system, determination of bioaccumulation in organism and food web, and feasible remedial strategies (2). Moreover, the multimedia models are considered valuable for environmental impact assessment of new and existing chemicals and identifying chemicals with a potential for persistence and long-range transport (2). However, before the models can be used for these purposes, particularly in regulatory applications, the accuracy and performance of the model must be evaluated. The model evaluation involves a number of steps with a final one typically being the comparison of the predicted concentration and the monitored one. For instance, the per* Corresponding author phone: +82-2-880-8522; fax: +82-2-8862361; e-mail: [email protected]. 10.1021/es034792j CCC: $27.50 Published on Web 01/08/2004

 2004 American Chemical Society

formance of CHEMFrance was tested against the measured data, which were retrieved from literatures for isobutylene in air (3). Devillers et al. (4) compared the model prediction (SMCM) with the monitored data for trichloroethylene, tetrachloroethylene, and 1,1,1-trichloroethane in air and water that were available from the literature. The model prediction agreed with the monitored data within a factor of 2-4. Kawamoto et al. (5) applied EUSES (6) and ChemCAN (7) to 68 chemicals in Japan. The observed data were collected from the literature for air, water, and sediment compartments. The two models showed excellent agreement with each other. However, tests against measured data were less satisfying in that both the models significantly underpredicted the measured values for most chemicals in all the media. The discrepancies ranged up to a few orders of magnitude. By setting the background concentration in inflowing air to the geometric mean of the observed range of the air concentrations in Japan, the discrepancies were significantly reduced in all media. However, in water and sediment, the improvements were less satisfactory, leaving a number of chemicals with gaps of more than 1 or 2 orders of magnitude. CHEMGL (8) was evaluated for atrazine, benzo[a]pyrene, benzene, and hexachlorobenzene in air, water, and groundwater. The model prediction fell within 1-2 orders of magnitude of data reported in the literature. There exist more studies where similar comparisons have been made (9-13). However, certain limitations exist in these studies. First and most importantly, where the predicted concentration did not agree well with the observed, the large discrepancies were often ascribed to a lack of or uncertainties in emission estimates. While critical to the evaluation of these models, a precise emission estimate is often lacking. Furthermore, the uncertainty in the emission estimates alone could not account for the large discrepancy because, even when the predicted value in one particular medium was forced to match the observed concentration, the model prediction still did not agree in other media (5, 8, 14). Therefore, it is essential to develop a means to evaluate and improve the model’s predicting capability in the absence of precise emission estimates. Second, due to the lack of adequate multimedia monitoring data, the model evaluation that has been conducted often involves only one to two media while the multimedia models usually deal with four (air, soil, water, and sediment) or more media. Ideally, multimedia models need be evaluated on their capability to describe the distribution in as many media as included in the models. In this respect, the evaluations have not been complete. Third, as multimedia monitoring data are scarce, some data might have been used that are not adequate or consistent in time and space for the purpose of the model evaluation (15). To avoid the limitations addressed above, we propose to use the relative concentration and not the individual concentration in each medium for the model evaluation. The relative concentration refers to the ratio of concentration in a medium to that in a reference medium (on a mass per volume basis in all media), for example, concentration in air/concentration in soil (Cair/Csoil) or dissolved-phase concentration in water/concentration in soil (Cwater/Csoil ) if soil is chosen as a reference medium. The use of the relative concentration has certain advantages. First, multimedia models often adopt first-order equations with respect to concentration to describe fate and transport processes. In the linear systems, predicted concentrations are approximately proportional to the emission rate. The influence of the emission rate then approximately cancels out in the concentration ratio term (i.e., the predicted relative conVOL. 38, NO. 4, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1079

FIGURE 1. Influence of emission rate on the predicted concentration and the predicted relative concentration. E1-E3 present the predicted concentrations of anthracene in sediment at the emission rates in air of 0.01 × 0.056 (average emission rate in cell 33), 1.0 × 0.056, and 100 × 0.056 mol/h, respectively. The time changes in the relative concentrations for sediment (concentration in sediment/concentration in soil) are identical for the three emission rates and fall on a single curve. centration hardly changes with the emission rate particularly if the emission mode of entry is left unchanged). As shown in Figure 1, the predicted relative concentrations are independent of emission rate and fall on a single curve. Therefore, uncertainties of emission estimates pose less problem for the model evaluation, and the models can be improved without accurate emission estimates. Second, with an emphasis on the relative concentration, a model evaluation focuses on the model’s capability to predict the relative concentrations among different media (i.e., the multimedia distribution) rather than the individual media concentration itself. Therefore, the relative concentration effectively serves for assessing chemical fate in the multimedia environment, which is a primary goal of the models. If a model is properly calibrated following this approach, an accurate prediction of the concentration in a medium would lead to accurate prediction of the ones in other media. Third, in dynamic multimedia simulations for semivolatile organic compounds, the concentrations in soil and sediment usually require much longer time to stabilize than in other media. However, as shown in Figure 1, the relative concentration quickly stabilizes while the individual ones are still transient. This condition substantially reduces the burden of choosing an adequate simulation period to compare the model outcomes with measured data. In this study, a dynamic multimedia environmental fate model, POPsME (Persistent Organic Pollutants in Multimedia Environments), was developed for assessing persistent organic pollutants (POPs) in multimedia environment in Korea and evaluated by comparing the predicted relative concentration with the measured ones in air particulates, water (dissolved phase and suspended solids (ss)), soil, sediment, and vegetation. Polycyclic aromatic hydrocarbons (PAHs) were chosen to test as they possess a wide range of fate properties and are emitted mainly into air.

Methodology Model Description. POPsME is a dynamic mass balance model describing the changes in concentration in five wellmixed compartmentssair, surface water, soil, sediment, and vegetation. Also subcompartments are included in the compartments (e.g., particulate matter in air; suspended 1080

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 4, 2004

solids in water; and soil-air, soil-water, soil-solids, etc. in soil). Vegetation is divided into four types (coniferous plant, deciduous plant, grass, and crops). As an initial step for the model evaluation, the structure and processes of POPsME were compared with those of three models known widely: CalTOX (16), SimpleBOX (17), and TRIM (18). In general, the descriptions of major fate and transport processes are not significantly different among the models. For instance, an important feature in these four models is the atmospheric transport described by the Eulerian Box Model (19). As this approach does not take into account the turbulent air diffusion, especially vertical diffusion, the chemical concentration in air might be underestimated. A certain difference lies in treating the air particulates, although the theory of Junge (20) is adopted in all the four models for the description of vapor-particulate partitioning in air. The surface area of the particulates in POPsMe is derived from the site-specific size distribution and TSP as characterized by the landuse patterns (19, 21, 22) while in CalTOX, SimpleBOX, and TRIM, it is represented by average values measured in urban or clean continental site. Consequently, only POPsME accounts for the dependency of air-soil and air-water intermedia transport on the particle size distribution. The set of mass balance equations was solved by the finite difference method with the operator splitting technique (23, 24). The operator splitting technique was used to separate the transport part (e.g., advective flow in air and water), the dynamics, and the chemistry parts (e.g., degradation reaction, wet scavenging, particle deposition, etc.) from each other. For modeling, the measured values of important fate properties were retrieved from literature (25). Additionally, EPIWIN (26) was used to estimate the degradation rate constants as necessary. Also, Henry’s law constants and subcooled liquid vapor pressures were estimated based on the method of Paasivirta et al. (27). Analytical Methods. Table 1 lists the 12 PAHs selected in this study. The chemical analysis, quality assurance, and control methods were based on the GERG Trace Organics Contaminant Analytical Techniques of the U.S. NOAA (28). Study Area and Sampling The model domain encompassed Seoul and the neighboring areas in South Korea,

TABLE 1. PAHs Selected in This Study chemicals

abbreviation

chemicals

abbreviation

phenanthrene anthracene fluoranthene pyrene benz[a]anthracene chrysene

Phe Ant Fl Pyr BaA Chr

benzo[b]fluoranthene benzo[k]fluoranthene benzo[a]pyrene indeno[1,2,3-cd]pyrene dibenz[a,h]anthracene benzo[g,h,i]perylene

BbF BkF BaP IcdP DahA BghiP

particle rotary impactor (CPRI, APMKorea, Korea). Soil samples of about 100 g were taken from the top (3 cm) layer using a hexane-cleaned stainless steel spatula. Water was taken within the depth of 10 cm from the surface using a precleaned amber glass bottle. Sediment samples of about 50 g were collected with stainless steel van veen grab. Leaves were taken using dichloromethane-cleaned scissors for vegetation sample.

Results and Discussion

FIGURE 2. Study area and sampling locations. (circles: air and soil samples; triangles: water and sediment samples). ranging from 126°08′12′′ to 127°50′05′′ E and from 36°53′51′′ to 38°13′56′′ N (TMX 124155-274155, TMY 526138-376138) with a size of 150 km × 150 km. This area was divided into 25 cells of equal size (30 km × 30 km) by taking the land-use type and the hydrogeological characteristics into account. The land-use types include a densely populated urban area (cell 33 in Figure 2), industrial center (cell 32), crop fields, and forest. Since the Taebaek Mountains are located near the eastern boundary of the model domain, the configuration of the ground is high in the east and low in the west. Han River originating from these mountain ranges (cells 25 and cell 45) flows to the West Sea through cells 32-34. The eastward wind prevails in the entire domain all year round. Consequently, a concentration gradient was expected along the west to east direction. The sampling sites in cells 31-35 were chosen to identify the expected spatial gradient in chemical concentrations. The site in cell 31 was considered as a background. The sites in cells 15 and 55 were selected to capture the seasonal change in wind direction (e.g., southeastern wind in summer and northeastern wind in winter). At all sampling sites shown in Figure 2, chemical concentrations in the particulates in air and soil were monitored. The concentrations in sediment and water (dissolved phase and ss separately) were monitored at five sites where the Han River flows. Sampling was conducted during each of the four seasons from August 1999 to July 2000. However, vegetation samples were taken monthly during the growing season from April to September 2000 in the most urbanized area (cell 33). Pinus koraiensis and Prunus serrulata were selected as a coniferous and a deciduous plant species, respectively. Only phenanthrene, anthracene, fluoranthene, and pyrene were measured in the plant leaves. A standard operating procedure was prepared for sampling each medium and was strictly adhered to while sampling. Each air particulate sample was collected for 24 h with cascade impactor (AAPSS, Anderson, USA) and coarse

Analysis of Observed. Table 2 lists the summary statistics for the measured PAHs concentrations. While samples were collected over the four seasons, the summary statistics do not necessarily represent the annual averages because the number of samples was limited in time and the sampling events were not evenly distributed over the 1-yr period. The temporal variation of the multimedia concentrations at each site ranged from less than a factor of few to 3 orders of magnitude depending on the location, medium, and PAHs. While the temporal variation was not statistically significant in air and soil, those in water (both dissolved and ss) and sediment were statistically significant (two-way ANOVA without replicate, p < 0.05). However, no consistent time trend was observed across the sites for all the PAHs. Although the spatial (site-to-site) variations in concentration were generally within 1 order of magnitude in all the media, the spatial variations were statistically significant except for air (two-way ANOVA with replicate, p < 0.05). In vegetation, the concentrations of phenanthrene and pyrene based on lipid content were significantly different for the two plant species (two-sided t, p < 0.05), and those of four PAHs showed statistically significant difference in time for both plant species (two-way ANOVA, p < 0.05). As shown in Figure 3a-d, the relative concentrations (Ci/ Csoil, where Ci denotes the concentration in medium i) at each site varied in time largely by less than 2 orders of magnitude for all the media. The spatial variations of Cair/ Csoil and Cdissolved/Csoil for individual PAHs were not statistically significant although ranging from a factor of few to 2 orders of magnitude depending on PAH and medium. For Css/Csoil and Csediment/Csoil, the spatial variations were statistically significant (two-way ANOVA without replicate, p < 0.05) although their magnitudes appeared smaller than for Cair/ Csoil and Cdissolved/Csoil. The relative concentrations for the two plant species showed no temporal variations for the four PAHs. As no statistically significant temporal variation was observed in the relative concentration for all media, the timeaveraged values (geometric means) were used to compare with the predicted ones for the purpose of the model evaluation. It is interesting to note that the relative concentration mostly varies within 1 order of magnitude over 1 yr for each of the PAHs at a given site. The geometric means of Csediment/Csoil were particularly uniform even across the PAHs. Model Calibration. POPsME was calibrated using four PAHs of three, four, five, and six rings (i.e., Phe, Pyr, BaP, and BghiP, respectively). First, from a sensitivity analysis, a number of parameters were identified by which the model outcomes are strongly influenced. It was expected that any VOL. 38, NO. 4, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1081

1082

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 4, 2004

FIGURE 3. Spatial and temporal variations in the measured relative concentrations: (a) Cair/Csoil, (b) Cdissolved/Csoil, (c) Csuspended solids/Csoil, (d) Csediment/Csoil, and (e) Cvegetation/Csoil. (Ci denotes the concentration in medium i, in mol/m3.) Point and error bar denote the temporal geometric mean and the observed ranges of relative concentrations at each sampling site, respectively. L and N denote Prunus serrulata and Pinus koraiensis, respectively.

FIGURE 4. Comparisons of the relative concentrations between the measured and the predicted before (open symbols) and after (closed symbols) calibration for (a) phenanthrene, (b) pyrene, (c) benzo[a]pyrene, and (d) benzo[g,h,i]perylene. ([/]) Cair/Csoil, (9/0) Cdissolved/Csoil, (2/4) Csuspended solids/Csoil, and (b/O) Csediment/Csoil. (Ci denotes the concentration in medium i, in mol/m3.) Error bar denotes the concentration ranges. parameter strongly affecting the concentration in soil would become influential to the relative concentrations as soil was the reference medium. The degradation rate constants in soil and the effective soil depth, for instance, were found to

be the most influencing factors to all the relative concentrations. Additionally, octanol-water partition coefficients (Kow) significantly influenced Cdissolved/Csoil, Csuspended solids/Csoil, and Csediment/Csoil while organic carbon fraction in ss, settling VOL. 38, NO. 4, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1083

1084

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 4, 2004

FIGURE 5. Effect of chemical properties on the degree of agreement of the relative concentrations between the measured and the predicted: (a) Cair/Csoil, (b) Cdissolved/Csoil, (c) Csuspended solids/Csoil, (d) Csediment/Csoil, and (e) Cvegetation/Csoil. (Ci denotes the concentration in medium i, in mol/m3.) L and N denote Prunus serrulata and Pinus koraiensis, respectively. pL, Kow, and Koa denote vapor pressure, octanol-water partition coefficient, and octanol-air partition coefficient, respectively.

FIGURE 6. Comparisons of the estimated suspended solids-water partitioning coefficients (Kp) with those measured as a function of octanol-water partition coefficient (Kow). The estimation was made using the equation Kp ) 0.48focKow (16), where Kp is in L/kg and foc is the organic carbon fraction in suspended solids. The value of 0.1 was used for foc as most of the measured values were near 0.1 in this study. velocity of ss, and the resuspension rate of sediment particles affected Csuspended solids/Csoil and Csediment/Csoil. The calibration was conducted in a sequential manner involving adjustment of a few parameters. First, the degradation rate constant in soil was selected among the reported degradation rates. Second, the discrepancies in Cdissolved/Csoil and Csuspended solid/Csoil could be minimized by adjusting the organic carbon fraction in ss within a range of the values monitored by the Ministry of Environment, Korea (29). Finally, density of suspended solids, degradation rate in sediment, effective depth of sediment layer, and settling velocity of ss were varied to minimize the total sum of squares of the differences between the predicted and the measured relative concentrations. The largest uncertainties were associated with this final step as understanding is severely limited on the settling and resuspension of suspended solids under the field conditions. Following the calibration, the agreement between the model prediction and the measured significantly improved. As shown in Figure 4, the deviations were reduced by more than 2 orders of magnitude depending on the chemical and medium. Model Evaluation. POPsME was evaluated with the remaining eight PAHs (Ant, Fl, BaA, Chr, BbF, BkF, IcdP, and

DahA). Figure 5 shows the predicted relative concentration as compared with the measured concentration. Overall, the prediction and the observation agreed within approximately a factor of 10 for all PAHs in the multiple environmental media addressed in this study. For Cair/Csoil and Cdissolved/Csoil, the degree of agreement did not show a particular trend with respect to the fate properties such as vapor pressure and/or Kow (Figure 5a,b). However, the model prediction tended to underestimate Css/Csoil and Csediment/Csoil for the PAHs of lower Kow while overestimating those of larger Kow (Figure 5c,d). This tendency is due to the increasing disparity with Kow between the partitioning equilibrium coefficient (Kp) measured in this study and the estimated one shown in Figure 6. In POPsME, the solid-water partitioning equilibrium is3 assumed and the coefficient is estimated from Kow (Koc ) 0.48Kow) (30, 31). However, this approach has limitations. First, deviations from the hydrophobic sorption equilibrium have been found, and the degree to which the sorption approaches to equilibrium is likely to vary with chemicals due to the kinetic or steric inhibition for large molecules (32-35). Second, a single parameter (Kow) could not account for all the molecular interactions that determine the equilibrium partitioning of a given compound between two phases (36, 37). For PAHs with an increasing number of rings, VOL. 38, NO. 4, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1085

TABLE 2. Summary Statistics for PAHs Concentrations in Each Mediuma cell 32 geomean

min

cell 33 max

geomean

min

cell 34 max

cell 35

geomean

min

max

geomean

min

max

1.02E-05 3.24E-07 1.16E-06 9.93E-07 1.27E-06 6.76E-07 4.16E-07 1.57E-07 2.03E-07 3.63E-07 2.01E-06 5.75E-07

8.06E-07 6.90E-08 1.98E-07 6.17E-07 1.13E-06 1.23E-07 1.24E-07 1.50E-07 4.31E-08 2.44E-08 4.70E-07 5.61E-08

2.76E-04 1.91E-06 2.22E-06 2.20E-06 1.44E-06 3.72E-06 1.86E-06 1.63E-07 2.51E-06 5.39E-06 8.63E-06 5.90E-06

4.44E-06 2.62E-07 2.57E-06 2.31E-06 nd 6.44E-08 5.10E-08 1.80E-07 9.21E-08 4.20E-08 1.49E-07 nd

1.04E-06 1.15E-07 1.32E-06 7.69E-07 nd 6.44E-08 5.10E-08 1.49E-07 6.26E-08 4.20E-08 1.49E-07 nd

9.71E-06 5.67E-07 5.02E-06 6.96E-06 nd 6.44E-08 5.10E-08 2.18E-07 1.35E-07 4.20E-08 1.49E-07 nd

1.46E-04 6.72E-05 3.07E-05 2.39E-05 1.26E-04 3.40E-05 2.50E-04 4.64E-05 3.28E-05 2.39E-05 8.18E-06 6.41E-06

8.93E-04 1.16E-04 3.13E-04 2.79E-04 1.26E-04 1.52E-04 2.50E-04 4.64E-05 5.90E-05 2.56E-05 3.59E-04 1.33E-04

1.15E-03 2.28E-04 5.02E-04 4.22E-04 1.77E-05 1.35E-04 3.03E-04 3.94E-05 9.84E-05 2.96E-05 8.70E-05 7.06E-05

6.24E-04 9.50E-05 2.27E-04 2.12E-04 3.26E-06 1.02E-04 3.03E-04 3.94E-05 5.12E-05 1.45E-05 8.13E-06 2.23E-05

2.12E-03 5.45E-04 1.11E-03 8.37E-04 9.61E-05 1.98E-04 3.03E-04 3.94E-05 1.89E-04 6.02E-05 7.35E-04 2.24E-04

min

max

geomean

min

max

2.05E-04 1.30E-05 3.54E-05 4.53E-05 1.40E-05 9.09E-06 1.95E-04 nd 1.24E-05 2.68E-04 nd 1.29E-07

2.48E-04 2.73E-05 1.23E-04 1.37E-04 6.39E-05 6.21E-05 1.95E-04 nd 9.06E-05 2.68E-04 nd 8.13E-05

2.97E-04 5.57E-05 9.78E-05 1.31E-04 7.04E-05 6.23E-05 7.51E-05 6.61E-06 4.33E-05 3.05E-04 nd 1.39E-04

2.55E-04 3.33E-05 7.24E-05 7.60E-05 3.72E-05 2.22E-05 7.80E-06 6.61E-06 2.39E-05 3.01E-04 nd 4.59E-05

3.35E-04 9.73E-05 1.63E-04 2.24E-04 2.31E-04 2.44E-04 4.97E-04 6.61E-06 1.25E-04 3.09E-04 nd 4.24E-04

4.24E-07 5.10E-07 6.37E-07 5.86E-07 2.76E-07 4.73E-07 5.47E-07 4.64E-07 3.30E-07 5.55E-07 9.69E-08 8.32E-07

4.22E-06 5.10E-07 2.18E-06 2.14E-06 2.67E-06 2.81E-06 3.81E-06 1.91E-06 4.84E-06 7.41E-06 1.83E-06 3.38E-06

1.05E-06 1.20E-07 1.33E-06 1.23E-06 4.92E-07 9.40E-07 1.07E-06 9.24E-07 5.87E-07 8.24E-07 1.72E-07 9.18E-07

5.54E-07 3.65E-08 8.01E-07 7.50E-07 2.92E-07 6.12E-07 6.48E-07 6.11E-07 4.21E-07 5.66E-07 9.85E-08 5.74E-07

3.14E-06 3.60E-07 3.11E-06 2.71E-06 1.04E-06 2.03E-06 1.89E-06 1.35E-06 1.04E-06 1.05E-06 3.62E-07 1.51E-06

Air Phe Ant Fl Py BaA Chr BbF BkF BaP IcdP DahA BghiP

6.37E-06 9.52E-07 1.23E-06 7.58E-07 2.55E-07 8.66E-07 6.13E-07 4.23E-07 2.82E-07 2.60E-07 3.38E-07 5.20E-07

1.36E-06 5.31E-07 1.20E-07 2.29E-08 4.22E-08 3.22E-07 6.13E-07 2.08E-07 7.09E-08 3.40E-08 2.47E-08 1.21E-07

3.17E-05 1.47E-06 3.48E-06 3.66E-06 2.81E-06 2.33E-06 6.13E-07 6.96E-07 1.73E-06 1.99E-06 4.84E-06 2.23E-06

1.97E-05 2.70E-07 5.23E-07 5.02E-07 7.40E-08 1.82E-07 2.03E-07 3.25E-07 2.53E-07 6.81E-08 4.36E-07 2.35E-07

2.01E-06 7.57E-09 1.06E-07 1.66E-07 6.67E-09 5.62E-08 4.52E-08 4.39E-08 1.99E-08 3.81E-09 3.28E-08 5.50E-08

Phe Ant Fl Py BaA Chr BbF BkF BaP IcdP DahA BghiP

1.48E-03 1.77E-04 5.80E-04 4.26E-04 6.37E-05 1.52E-04 2.24E-04 3.18E-05 6.67E-05 4.66E-05 6.35E-05 1.02E-04

5.95E-04 6.84E-05 2.04E-04 2.14E-04 3.79E-05 1.02E-04 1.34E-04 3.18E-05 2.22E-05 1.11E-05 3.03E-06 4.91E-05

2.94E-03 2.60E-04 9.06E-04 6.41E-04 8.56E-05 2.76E-04 3.61E-04 3.18E-05 1.42E-04 9.17E-05 3.48E-04 2.00E-04

1.28E-03 1.66E-04 7.07E-04 8.60E-04 7.08E-05 8.99E-05 3.81E-04 6.92E-05 1.33E-04 8.19E-05 3.99E-05 2.33E-04

5.97E-04 9.28E-05 2.91E-04 2.61E-04 4.91E-05 6.19E-05 3.63E-04 4.67E-05 4.36E-05 1.55E-05 4.95E-06 3.17E-05

7.54E-04 8.30E-06 1.61E-06 1.94E-06 6.25E-07 7.26E-07 2.07E-06 4.47E-06 3.28E-06 4.37E-07 3.06E-06 5.39E-07

Dissolved Phase

cell 33

1.80E-03 5.08E-04 2.06E-03 7.04E-03 1.02E-04 1.15E-04 3.99E-04 1.03E-04 2.85E-04 2.29E-04 1.14E-04 7.97E-04

4.54E-04 8.84E-05 1.42E-04 1.21E-04 1.26E-04 8.15E-05 2.50E-04 4.64E-05 4.40E-05 2.47E-05 6.86E-05 3.34E-05

max

geomean

cell 34

cell 55

geomean

min

max

geomean

min

Phe Ant Fl Py BaA Chr BbF BkF BaP IcdP DahA BghiP

3.44E-04 5.29E-05 1.51E-04 1.32E-04 8.38E-05 7.19E-05 9.83E-05 8.16E-05 6.82E-05 6.05E-04 2.59E-03 8.40E-05

2.07E-04 2.05E-05 7.83E-05 7.85E-05 4.22E-05 4.31E-05 3.78E-05 2.65E-05 2.25E-05 2.62E-04 2.59E-03 1.39E-05

7.88E-04 3.46E-04 4.23E-04 2.04E-04 3.49E-04 1.22E-04 2.20E-04 1.54E-04 5.68E-04 2.57E-03 2.59E-03 2.50E-03

2.97E-04 3.17E-05 8.00E-05 4.31E-05 4.11E-05 4.02E-05 5.32E-05 1.93E-05 2.49E-05 2.79E-04 nd 1.40E-05

2.30E-04 2.09E-05 4.73E-05 2.02E-05 3.10E-05 2.53E-05 2.93E-05 1.77E-05 1.92E-05 2.55E-04 nd 3.89E-06

3.56E-04 4.47E-05 1.15E-04 8.09E-05 4.66E-05 6.23E-05 7.64E-05 2.11E-05 3.48E-05 3.04E-04 nd 3.43E-05

Phe Ant Fl Py BaA Chr BbF BkF BaP IcdP DahA BghiP

4.09E-06 7.42E-07 5.71E-06 6.44E-06 2.29E-06 3.99E-06 4.96E-06 1.94E-06 3.48E-06 4.47E-06 8.35E-07 5.65E-06

3.15E-06 6.02E-07 5.26E-06 4.91E-06 2.01E-06 3.38E-06 3.13E-06 1.61E-06 2.70E-06 2.36E-06 4.51E-07 4.94E-06

6.42E-06 9.23E-07 6.15E-06 8.84E-06 2.63E-06 5.07E-06 7.49E-06 2.33E-06 4.34E-06 9.39E-06 1.87E-06 6.33E-06

2.07E-06 3.54E-07 2.97E-06 2.83E-06 1.18E-06 2.13E-06 2.37E-06 2.22E-06 1.50E-06 2.38E-06 4.32E-07 2.40E-06

6.70E-07 1.22E-07 1.49E-06 1.43E-06 5.89E-07 9.27E-07 9.59E-07 9.97E-07 6.05E-07 1.05E-06 1.68E-07 9.82E-07

5.79E-06 7.32E-07 4.77E-06 4.49E-06 2.60E-06 3.83E-06 4.16E-06 4.49E-06 2.90E-06 4.31E-06 7.04E-07 4.45E-06

cell 15

Suspended Solid Phase 2.25E-04 1.88E-05 6.61E-05 7.89E-05 2.99E-05 2.38E-05 1.95E-04 nd 3.36E-05 2.68E-04 nd 3.24E-06

Sediment

cell 31

1.04E-06 5.10E-07 1.41E-06 1.37E-06 6.06E-07 1.27E-06 1.63E-06 1.05E-06 1.19E-06 1.95E-06 4.47E-07 1.78E-06

cell 32

cell 33

geomean

min

max

geomean

min

max

PHE ANT FL PY BaA CHR BbF BkF BaP IcdP

6.65E-01 5.56E-02 6.16E-01 5.71E-01 2.05E-01 5.00E-01 nd 4.28E-01 3.61E-01 5.07E-01

4.24E-01 4.17E-02 4.19E-01 3.80E-01 1.39E-01 3.63E-01 nd 2.21E-01 2.13E-01 2.49E-01

1.51E+00 9.05E-02 1.34E+00 1.24E+00 3.83E-01 9.09E-01 1.39E+00 1.06E+00 8.63E-01 1.32E+00

1.86E+00 1.41E-01 2.51E+00 1.72E+00 7.70E-01 1.70E+00 1.61E+00 1.17E+00 1.05E+00 1.40E+00

1.11E+00 6.47E-02 1.57E+00 1.07E+00 7.05E-01 1.10E+00 1.26E+00 9.56E-01 7.47E-01 1.11E+00

3.15E+00 2.30E-01 4.13E+00 2.59E+00 8.35E-01 2.65E+00 2.60E+00 1.55E+00 1.49E+00 2.09E+00

1086

9

cell 34

geomean

min

max

geomean

min

max

3.19E+00 3.38E-01 5.44E+00 3.60E+00 1.97E+00 4.84E+00 4.97E+00 2.78E+00 3.06E+00 5.13E+00

2.51E+00 2.96E-01 4.47E+00 2.87E+00 1.82E+00 4.31E+00 4.72E+00 2.00E+00 2.85E+00 3.92E+00

3.67E+00 4.22E-01 6.03E+00 4.13E+00 2.11E+00 5.27E+00 5.25E+00 3.76E+00 3.36E+00 8.26E+00

5.58E-01 3.78E-02 5.83E-01 4.62E-01 2.56E-01 5.54E-01 8.59E-01 4.28E-01 4.17E-01 7.03E-01

3.24E-01 1.47E-02 4.15E-01 3.60E-01 1.59E-01 4.34E-01 5.53E-01 1.68E-01 2.65E-01 4.31E-01

7.41E-01 6.18E-02 8.01E-01 5.95E-01 3.97E-01 7.46E-01 1.15E+00 7.37E-01 6.69E-01 1.14E+00

Soil

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 4, 2004

TABLE 2 (Continued) cell 31 geomean

min

cell 32 max

geomean

min

cell 33 max

geomean

min

cell 34 max

geomean

min

max

DahA 9.34E-02 3.28E-02 2.54E-01 2.91E-01 2.45E-01 4.31E-01 1.05E+00 7.88E-01 1.59E+00 2.00E-01 1.51E-01 2.38E-01 BghiP 5.28E-01 3.49E-01 1.19E+00 1.56E+00 1.17E+00 2.32E+00 3.99E+00 3.65E+00 4.27E+00 6.94E-01 4.29E-01 1.11E+00 cell 33 deciduous

coniferous

Jan

May

June

July

Aug

2.51E-07 1.77E-08 1.71E-07 9.59E-08

1.75E-07 1.64E-08 1.13E-07 7.16E-08

1.11E-07 1.2E-08 7.85E-08 5.61E-08

8.13E-08 7.88E-09 1.37E-07 4.47E-08

1.03E-07 1.9E-08 4.65E-08 3.91E-08

Jan

May

June

July

Aug

3.33E-07 2.49E-08 9.6E-08 4.8E-08

2.87E-07 2.04E-08 8.57E-08 4.41E-08

3.5E-07 2.28E-08 1.03E-07 5.73E-08

3.59E-07 2.09E-08 6.1E-08 4.19E-08

3.06E-07 1.81E-08 5.86E-08 4.06E-08

Vegetation PHE ANT Fl Py

a Unit: mol/m3. Concentrations in soil, sediment, and vegetation were originally measured in the unit of mass per dry mass basis. For the conversion of the unit, the values of bulk densities used were 1500, 2000, and 850 kg/m3 for soil, sediment, and vegetation, respectively. nd, not detected.

the differentiation of atomic charges decreases, resulting in lower polarity and higher hydrophobicity (38, 39). Also, soil or sediment organic matter might be more polar than octanol due to the increased aromaticity in humic material (40, 41). Therefore, the use of the equilibrium assumption and Kow could simultaneously contribute to the prediction bias. Despite these limitations to the use of Kow, the prediction and the observation for Css/Csoil and Csediment/Csoil still agreed within 1 order of magnitude. In vegetation (Figure 5e), the prediction and the observation agree within approximately a factor of 10. No remarkable trend was found with respect to the octanol-air partition coefficient (Koa). The results demonstrate that multimedia models may effectively be evaluated by using the relative concentration while accurate emission estimates are not available. The use of the relative concentration therefore is proposed as a valuable means to evaluate and improve the model, particularly for multimedia distribution of substances.

Acknowledgments Financial support for this work was provided by the Ministry of Environment, Korea, under the grants of G-7 and ECO2 projects.

Supporting Information Available A short description of POPsME. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Cowan, C. E.; Mackay, D. E.; Feijtel, T. C. J.; van den Meent D.; Di Guardo, A.; Davies, J.; Mackay, N. The Multimedia Fate Model: A Vital Tool for Predicting the Fate of Chemical; Society of Environmental Toxicology and Chemistry: Pensacola, FL, 1995. (2) OECD. Report of the OECD/UNEP workshop on the use of multimedia models for estimating overall environmental persistence and long-range transport in the context of PBTs/POPs assessment. OECD Series on Testing and Assessment 36; 2002. (3) Maddalena, R. L.; McKone, T. E.; Layton, D. W.; Hsieh, D. P. H. Chemosphere 1995, 30, 869-889. (4) Devillers, J.; Bintein, S.; Karcher, W. Chemosphere 1995, 30, 457476. (5) Kawamoto, K.; MacLeod, M.; Mackay, D. Chemosphere 2001, 599-612. (6) RIVIM (Institute of Public Health and the Environment, The Netherlands). EUSES-The European Union System for the Evaluation of Substances; European Chemicals Bureau: Ispra, Italy, 1996. (7) Mackay, D.; Di Guardo, A.; Paterson, S.; Kiscsi, G.; Cowan, C. E.; Kane, D. E. Environ. Toxicol. Chem. 1996, 32, 2442-2449. (8) Zhang, Q.; Crittenden, J. C.; Shonnard, D.; Mihelcic., J. R. Chemosphere 2003, 50, 1377-1397.

(9) Mackay, D.; Paterson, S. Environ. Sci. Technol. 1991, 25, 427436. (10) Cohen, Y.; Ryan, P. A. Environ. Sci. Technol. 1985, 19, 412-417. (11) Severinsen, M.; Andersen, M. B.; Chen, F.; Nyholm, N. Chemosphere 1996, 32, 2159-2175. (12) Jager, T.; De Bruijn, J. H. M. The EU-TGD for new and existing substances: Does it predict risk?. In Forecasting the Environmental Fate and Effects of Chemicals; Rainbow, P. S., Hopkin, S. P., Crane, M., Eds.; John Wiley & Sons: New York, 2001; pp 71-82. (13) Pederson, B. M.; Thibodeaux, L. J.; Valsaraj, K. T.; Reible, D. D. Environ. Toxicol. Chem. 2001, 20, 2114-2121. (14) Wania, F.; Mackay, D. Environ. Pollut. 1999, 100, 223-240. (15) Gobas, F. A. P. C.; Pasternak, J. P.; Lien, K.; Dunkan, R. K. Environ. Sci. Technol. 1998, 32, 2442-2449. (16) McKone, T. E. CalTOX, A Multi-media Total Exposure Model for Hazardous Waste Sites. Part II: The dynamic multi-media transport and transformation model; A report prepared for the State of California, Department of Toxic Substances Control by the Lawrence Livermore National Laboratory: Livermore, CA, 1993. (17) Brandes, L. J.; den Hollander, H.; van de Meent, D. SimpleBOX 2.0: A nested multimedia fate model for evaluating the environmental fate of chemicals; National Institute of Public Health and the Environment: Bilthoven, The Netherlands, 1996. (18) U.S. Environmental Protection Agency. TRIM. FaTE Technical Support Document. Vol. II: Description of chemical Transport and Transformation Algorithms; Office of Air Quality Planning And Standards: Research Triangle Park, NC, 1999; EPA-453/ D-99-002B. (19) Seinfeld, J. H.; Pandis, S. N. Atmospheric Chemistry and Physics. From air pollution to climatic change; John Wiley & Sons: New York, 1998. (20) Junge, C. E. Basic considerations about trace constituents in the atmosphere as related to the fate of global pollutants. In Fate and Pollutants in the Air and Water Environment; Suffet, I. H., Ed.; Wiley & Sons: New York, 1997; pp 7-26. (21) Whitby, K. T. Atmos. Environ. 1978, 12, 135. (22) Seigneur, C.; Belle Hudischewskyj, A.; Seinfeld, J. H.; Whitby, K. T.; Whitby, E. R.; Brock, J. R.; Barnes, H. M. Aerosol Sci. Technol. 1986, 5, 205-222. (23) Jacobson, M. Z. Fundamentals of Atmospheric Modeling; Cambridge University Press: Cambridge, 1999. (24) Roaches, P. J. Computational Fluid Dynamics; Hermosa Publications: Albuquerque, NM, 1985. (25) Mackay, D.; Shiu, W. Y.; Ma, K. C. Illustrated handbook of physicochemical properties and environmental fate for organic chemicals. Vol. 2: Polynuclear aromatic hydrocarbons, polychlorinated dioxins and dibenzofuran; Lewis Publishers: Boca Raton, FL, 1992. (26) Meyland, W. EPIWIN v3.02-US EPA, 1999. (27) Paasivirta, J.; Sinnkonen, S.; Mikkelson, P.; Rantio, T.; Wania, F.; Chemosphere 1999, 39, 811-832. (28) NOAA. Comprehensive Description of Trace Organic Analytical Methods. 4; 1993. (29) Ministry of Environment, Korea. http://www.me.go.kr/. (30) Karickhoff, S. W. Chemosphere 1981, 10, 833-846. VOL. 38, NO. 4, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1087

(31) Karickhoff, S. W. Pollutant sorption in environmental systems. In Environmental exposure from chemicals; Neely, W. B., Blan, G. E., Eds.; CRC Press: Boca Raton, FL, 1985; Vol. 1. (32) Sabljic, A.; Gusten, H.; Verhaar, H.; Hermens, J. Chemosphere 1995, 31, 4489-4514. (33) Gawlik, B. M.; Sotiriou, N.; Feicht, E. A.; Schulte-Hostede, S.; Kettrup, A. Chemosphere 1997, 34, 2525-2551. (34) Chiou, C. T.; Kile, D. E. Environ. Sci. Technol. 1998, 32, 338343. (35) Gustafsson, O.; Haghseta, F.; Chan, C.; MacFarlane, J.; Gschwent, P. M. Environ. Sci. Technol. 1997, 31, 203. (36) Luehrs, D. C.; Hickey, J. P.; Nilsen, P. E.; Godbole, K. A.; Gogers. T. N. Environ. Sci. Technol. 1996, 30, 143-152.

1088

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 4, 2004

(37) Goss, K.-U.; Schwarzenbach, R. P. Environ. Sci. Technol. 2001, 35, 1-9. (38) Ferreira, M. M. C.; Chemosphere 2001, 44, 125-146. (39) Chiou, C. T.; McGroddy, S. E.; Kile, D. E. Environ. Sci. Technol. 1998, 32, 264-269. (40) Chiou, C. T.; Porter, P. E.; Schmedding, D. W. Environ. Sci. Technol. 1983, 17, 227-231. (41) Gauthler, T. D.; Seitz, W. R.; Grant, C. L. Environ. Sci. Technol. 1987, 21, 243-248.

Received for review July 21, 2003. Revised manuscript received October 22, 2003. Accepted December 2, 2003. ES034792J