Page 2 of 6

6.3.2. Methodology Description and Parameter Assignments

Four of the six example scenarios of Chapter 5 served as "baselines" in the sensitivity analysis exercises. The single scenario for the off-site soil source category, example scenario #3 in Chapter 5, served as the basis for testing on these algorithms:

1) transport of vapor and particulate phase airborne contaminants from a site of contamination to a nearby site of exposure, and
2) transport of soils via erosion to nearby sites of exposure and to surface water bodies to impact bottom sediments, fish, and water.

The source strength for this scenario, in summary, was a 40,000 m2 (4 ha, 10 ac) area of soil concentrations of 1 m g/kg (ppb) within a watershed of size 4,000 ha (40,000,000 m2; 10,000 ac; 15.5 mi2) with soils otherwise at 0.0 ppb.

The high end scenario for the on-site soil source category, example scenario #2 in Chapter 5, served as the baseline for testing on these algorithms:

1) suspension and dispersion of vapor and particle-phase contaminants at a contaminated site, which was also the exposure site for the on-site source category,
2) impacts of soil concentrations and other parameters to below ground vegetations, and air concentrations and other parameters to above ground vegetations, and
3) impacts of soil, grass, and feed concentrations, and other parameters, to beef concentrations.

The source strength for this scenario, in summary, were soil concentrations within a 4,000 ha small farm of 1 ng/kg (ppt). The high end example scenario for the stack emission source category, example scenario #5, served as the basis for the testing the impact of particle depositions and ambient air concentrations on soils and biota. The ambient air concentrations and deposition rates at the site of exposure 500 meters from the stack served as the baseline source strength terms. The single scenario for the effluent discharge source category, example scenario #6, was used to evaluate the impact of parameters required for that source category on fish and water concentrations.

The source strength in that case was a discharge of 0.0315 mg/hr into a surface water body with a harmonic mean flow rate of 4.7x108 L/hr. Assignment of that baseline discharge was based on data from the 104 pulp and paper mill study, and then considering reductions is discharges which have occurred in these pulp and paper mills since the 104 mill study in 1988.

The baseline chemical for all these sensitivity runs was 2,3,7,8-TCDD; i.e., all the chemical specific parameters were those assigned to this example compound. The high and low values for parameter testings were determined starting with the 2,3,7,8-TCDD assignments. Care was not taken to encompass a range of possible values for all dioxin-like compounds. However, the ranges that were tested are mostly inclusive of the dioxin-like compounds. What will be noted and discussed below is that mostly the model response to chemical-specific parameters is linear or nearly linear, so that model responses to values outside the ranges tested can be evaluated easily.

All the initial parameter values required for all four source categories, and the values selected for high and low sensitivity analysis were listed above in Table 6-1. Following are brief discussions on the selection of these high and low values. Longer discussions on all parameter values can be found in Chapter 4, which included justifications for all parameter values selected for the demonstration of the methodologies in Chapter 5. Often, ranges of possible values were discussed in Chapter 4; those ranges were the basis of high and low parameter values selected below. Discussions in Chapter 4 are not repeated here, but are referenced below. The summaries below are organized in the same order as the parameter listings in Table 6-1.

. Contaminated and exposure site characteristics:
These are the area and distance parameters, and the soil characteristic parameters of the site of contamination and the site of exposure. The "site of contamination" refers to the bounded area of high soil concentration for the off-site source category. The "site of exposure" for these sensitivity runs is the small farm which was the basis for the definition of the "high end" example scenarios demonstrated in Chapter 5. The area of the site of exposure, AES, and site of contamination, ASC, are both 40,000 m2 in the demonstration scenarios, which is equal to 4 ha or 10 ac.

Low and high values tested were 4,000 m2 (0.4 ha, 1 ac) and 400,000 m2 (40 ha, 100 ac). The soil description parameters include soil porosity, ESLP, particle bulk density, Psoil, soil bulk density, Bsoil, and the organic carbon fraction, OCsl. The assignment of high/low values to these parameters were developed from Brady (1984) and cover a reasonable range of agricultural field soils. The no-till and tillage depths, dnot and dt, refer to the depth to which eroded soil or depositing particulates mix at the site of exposure. The no-till depth was set at 5 cm and was varied between 1 and 20 cm, and the tilled depth was varied between 10 and 30 cm. The no-till concentrations were used to estimate soil concentrations for soil related exposures: soil ingestion and soil dermal contact, and also for the beef and milk bioconcentration algorithm. The tilled concentrations were used only to estimate the concentration in below ground vegetations.

. Soil and Sediment Delivery Parameters:
Contaminated soil erodes from a site of contamination, a 4 ha site in the demonstration scenarios, to a nearby site of exposure and also to a nearby stream. The distance to the site of exposure from a site of contamination, DLe, was set at 150 meters for the example scenarios, and varied between 50 and 1000 meters in this exercise. The same initial distance of 150 meters was the distance to the nearby stream, DLw, and it was also varied between 50 and 1000 meters.

The unit amount of soil eroding off the site of contamination, SLs, was initialized at 21520 kg/ha-yr, equal to 9.6 Eng. ton/ac-yr (abbreviated t/ac-yr hereafter). Assumptions inherent in this estimate include: midcontinent range of annual rainfall erosivity (which is also the middle of the range of rainfall intensities of the US), midrange agricultural soil erosivity, a gentle 2% slope, no man-made erosion protection (ditches, etc.), and bare soil conditions. A doubling of this amount to 42,000 kg/ha-yr (19 t/ac-yr) was used as a high erosion estimate off the site of contamination.

This could reflect any number of different assumptions, such as more erosive soil, more erosive rainfall, steeper slopes, and so on. A low estimate of one-tenth the default value, at 2100 kg/ha-yr (1 t/ac-yr), could reflect all the same assumptions except a dense cover of grass or weeds, which changes the bare soil assumption leading to a "C" (cropping management factor) of 1.0 to a C of 0.1. The erosion amount of 2152 kg/ha-yr was the initial amount assumed for a second unit erosion term needed in this assessment, a unit erosion typical of land area between the contaminated and the exposure site, SLec.

The critical assumption in this initialization was that all conditions for this land area were similar to the contaminated site, except that the ground was densely covered with grass or weeds. The value of SLec was reduced to 0 kg/ha-yr for the low, which is unrealistically low but might give a sense of how the algorithm would perform if mixing with soil between the contaminated and exposure site were not considered. The high value was 21,000 kg/ha-yr, which is similar to the initial assumption for the contaminated site, could reflect similar erosion conditions between the contaminated site and the exposure site. The third unit soil loss parameter required is one which reflects average erosion conditions within the watershed draining into the water body, SLw.

This was initialized at 6455 kg/ha-yr (2.88 t/ac-yr) which reflects similar erosion conditions as the contaminated site (soil erosivity, rainfall intensity, average slopes, lack of support practices) except some erosion protection due to vegetation - C equal to 0.3 instead of 1.0. It was reduced to 2100 kg/ha-yr, which might translate to C equal to 0.1, and increased to 21,000, which was equal to the initial higher erosion from the contaminated site. The range of the enrichment ratio, ER, was noted at between 1 and 5 for its application in agricultural runoff field data and model simulations (Chapter 4, Section 4.3.1), and was given an initial value of 3 in this application. High and low values tested were 5 and 1.

An average watershed concentration of contaminant was set at 0 for the off-site demonstration scenarios, where the soil concentration of 2,3,7,8-TCDD (and the other example compounds) was set at 1 ppb. This was selected so that the off-site impact to surface water bodies could be demonstrated as an incremental impact. A concentration of 2,3,7,8-TCDD of 1 ppt was, however, justified as a "background" concentration for demonstrating the on-site source category.

The value was used to evaluate the impact of a bounded site at 1 ppb when a background concentration of 1 ppt is also assumed to exist. Three parameters reflect watershed size. These include the effective drainage area, Aw, the watershed sediment delivery ratio, SDw, and the volume of the receiving water body, VOLw. These are related and should therefore be changed in tandem. The initial watershed size of 4000 ha (15.4 mi2) was reduced to 400 ha (1.5 m2) and increased to 400,000 ha (1540 mi2). Since the water body volume was estimated using a in/yr runoff times an area, it was concurrently reduced 1 order of magnitude for the small watershed test and increased two orders of magnitude for the large watershed.

The values of SDw were estimated using Figure 4-5 (Chapter 4), which shows watershed delivery ratios as a function of watershed area. The remaining three parameters further described the water body, and were the total suspended solids, TSS, and the organic carbon contents of suspended and bottom sediments, OCssed and OCsed. The initial value of TSS of 10 mg/L is typical of a moving water body (stream, river) supportive of fish and other aquatic life. It was reduced to 2 mg/L, which is typical of a stationary water body (pond, lake, reservoir) and increased to 50 mg/L, which begins to be high for a water body expected to be supportive of fish.

The organic carbon contents were initialized at 0.05 for OCssed and 0.03 for OCsed. The premise was that they were related - that sediments in suspension were lighter and likely to be higher in organic carbon content than bottom sediments. They were also changed in tandem to 0.02 (OCssed) and 0.01 (OCsed) for a low organic carbon sensitivity test and 0.10 and 0.05 for a high organic carbon test.

. Volatilization and Dust Suspension Parameters:
Distances and areas are pertinent to estimating vapor-phase and particulate-phase air concentrations, and these have been discussed above in the first two categories. One parameter included for sensitivity testing in this category is the exposure duration, ED. It is included in these exercises because the estimation of average volatilization flux over a period of time is a function of that period of time.

The derivation of the flux model assumed contamination originates at the soil surface at time zero, and over time, originates from deeper within the soil profile. Therefore, the flux decreases over time (because residues have to migrate from deeper in the profile), and the average flux over a period of time will decrease as that period of time increases. This is further discussed in Chapter 4, Section 4.3.2., and in the original citation for the volatilization flux algorithm, Hwang, et al. (1986).

The exposure duration assumed in the high end scenarios was 20 years, this was changed to 1 and 70 years in sensitivity tests. A range of average windspeeds, Um, around the U.S. was noted at 2.8 and 6.3 m/sec, and these two values were used around the selected value of 4.0 m/sec. The frequency with which wind blows from a site of contamination to a site of exposure, FREQ, was set at 0.15, which is appropriate if one assumes that wind blows in all directions roughly equally.

It was changed to 0.05 and 0.50, which might translate to an assumption of a prevailing wind direction, either away from or towards a site of exposure. The remaining parameters, fraction of vegetative cover, V, threshold wind speed, Ut, and model specific function, F(x), all refer to the wind erosion algorithm which suspends contaminated particulates into the air. Sensitivity tests were applied to this trio for the on-site and the off-site source categories. V for the off-site scenario was initialized at zero, implying bare ground cover; it was increased to 0.9 reflecting dense ground cover in the single sensitivity test here.

It was set at 0.5 for the on-site small farm demonstration scenario, reflecting some bare ground conditions (in the agricultural fields, e.g.) as well as some dense vegetation (in other grassed areas of the farm property). It was decreased to 0 and increased to 0.9. The parameters Ut and F(x) reflect intrinsic erodibility of the soil and were varied together. Values were selected to reflect a high and low wind erodibility soil, following guidance in EPA (1985), the primary reference for the wind erosion algorithm.

. Bioconcentration and Biotransfer Parameters:
The only such factor for fish concentration estimation was the fraction of fish lipid, flipid. The brief discussion on this parameter in Chapter 4 (Section 4.3.4.1) indicated a range of around 5 to over 20%. Considering that the lipid content of edible portions of fish are less than whole fish lipid contents, a value less than 5%, 3% (0.03), was chosen as the low value, and also considering edible lipid content considerations, an upper value of 20% (0.20) was selected.

Several parameters are required for the vegetation concentration algorithm, most of which were associated with the algorithm for dry plus wet deposition of particulates. One parameter not associated with fate and transport was the dry to fresh weight conversion factor, FDW. The algorithm calculates vegetative matter concentrations on a dry weight basis, which is appropriate for the role of vegetation in the beef/milk bioconcentration algorithm.

However, ingestion rates of fruits and vegetables are on a fresh weight basis, so dry weight concentrations have to be converted to a fresh weight basis. The initial value of 0.15 assumes that fruits and vegetables are 85% liquid. The high and low values tested for this parameter were 0.30 (70% liquid) and 0.05 (95% liquid). Four parameters are described as empirical correction factors for the air-to-leaf algorithm adopted for vapor phase transfers to vegetation (three of the parameters), and for the soil-water-to-root algorithm adopted for below ground vegetation.

There is one each for the four principal vegetations considered: below ground vegetables/fruits - VGbg, above ground vegetables/fruits - VGveg, grass - VGgr, and feed - VGfeed. The concept for assignment of values to these parameters was the same, and briefly is as follows. The principal biotransfer factors (air-to-leaf and soil-water-to-root) were developed in laboratory experiments where relatively thin vegetations (azalea leaves for air-to-leaf transfers and barley roots for soil-water-to-root transfers) were used. Concurrently, there is evidence that the strongly hydrophobic/lipophilic dioxin-like compounds are found only in outer portions of vegetations and not inner portions of bulky vegetation; there is very little translocation of dioxin-like compounds into and within vegetation.

Therefore, the full vegetation concentrations of thin vegetations measured in the laboratory experiments (and the laboratory experiments did use dioxin-like compounds among the several used) would most likely mirror only the outer surface concentrations found for dioxin-like compounds in bulky vegetations, and not full vegetation concentrations of bulky vegetations. As such, an empirical correction factor, based on a surface area to volume calculation, was introduced to arrive at full vegetation concentrations for bulky vegetations. These were principally the fruits and vegetables and the surface area to volume calculations led to assignments of VGbg and VGveg of 0.01.

These were reduced to 0.001 and increased to 0.10 in sensitivity testing. The VGgr was set at 1.00 since grass was thought to be analogous to the azalea leaves. Although there is insufficient justification to change VGgr, a lower value of 0.50 was chosen. The VGfeed was set at 0.5, recognizing that some cattle feed is unprotected and thin vegetation such as hay, while others are protected grains such as corn grain. That value was changed to 0.25 and 0.75 in sensitivity testing. There is one required parameter for the dry deposition algorithm, and this is the particle deposition velocity by gravity settling, Vp, in m/yr.

The initial value of 3.2x105 m/yr, from a velocity assumption of 1 cm/sec, was given by Seinfeld (1986) as the gravitational settling velocity for 10 m m particles. This is the appropriate size to consider since the wind erosion algorithm was developed only for inhalable size particulates, those less than 10 m m (EPA, 1985). This was reduced to 0.5 cm/sec and 2 cm/sec (transformed to m/yr) for sensitivity testing. Three of the vegetation bioconcentration parameters are associated with the particulate wet deposition algorithm. These are the atmospheric washout ratio, Wp, the retention of particles on vegetation, Rw, and the annual rainfall amount, R.

The definition, derivation, and ranges for these values are described in Chapter 4, Section 4.3.4.2, and are not repeated here (the ranges are given in Table 6.1). The remaining bioconcentration parameters are the yield and crop intercept values for the three above ground vegetations: vegetables/fruits (Yveg, INTveg), grass (Ygr, INTgr), and cattle feed (Yfeed, INTfeed). Again, discussions of chosen, and high and low, values for these quantities are given in Chapter 4, Section 4.3.4.2 (and displayed in Table 6.1). It is noted that these two terms are correlated - high yields are correlated with high interception amounts. In sensitivity testing, therefore, these parameters were changed in tandem.

The remaining bioconcentration/biotransfer parameters are for the beef/milk bioconcentration algorithm. One of the parameters relates the bioavailability of soil relative to the bioavailability of vegetation, where bioavailability refers to the efficiency of transfer of a contaminant attached to a vehicle. Fries and Paustenbach (1990) developed the bioconcentration factor, BCF, from studies where cattle were given contaminated feed.

The studies of McLachlan, et al. (1990), from which BCFs for dioxin congeners were derived and used for this assessment, also used standard cattle feeds. This feed is assumed to be analogous to the vegetation in cattle diet; therefore, the experimental BCFs can be directly applied to vegetation in cattle diets. However, Fries and Paustenbach also hypothesized that soil is less bioavailable than feed, based on some rat feeding studies, and therefore the BCF developed from feed cannot directly be used on a soil concentration - it should be reduced. Information in Fries and Paustenbach led to an assignment of 0.65 for the soil bioavailability factor, Bs.

This was reduced to 0.30 and increased to 0.90 in sensitivity testing. Three parameters describe the proportion of the dry matter in the diet of beef cattle that is soil, BCSDF, grass, BCGDF, and feed, BCFDF. The sum of these three terms, by definition, equals 1.00. Beef cattle are principally pastured (where incidental soil ingestion occurs), with supplemental feeds including hay, silages, and grain, particularly in cooler climates where they are housed during the winter. Values of 0.04 for BCSDF, 0.48 for BCGDF, and 0.48 for BCFDF were used in the demonstration scenarios.

The same three parameters are required for cattle raised for dairy products: DCSDF for soil, DCGDF for grass, and DCFDF for feed. The dairy cattle model was one of very little pasturing, principally being fed high-quality grain indoors while they were in lactation: DCSDF of 0.02, BCGDF of 0.08, and DCFDF of 0.90. A final set of four parameters describes the proportion of these dietary intakes that are contaminated. Two are defined as the fraction of grazing land that is contaminated - BCGRA for beef cattle and DCGRA for dairy cattle.

The initial assumption of 1.00 for both these parameters meant that all the vegetations as well as all the soil in the cattle diets was contaminated (since soil was assumed to be ingested during grazing). The last two similarly are defined as the proportion of feed that is contaminated - BCFOD for beef cattle and DCFOD for dairy cattle. They were also set at 1.00, perhaps indicating that feed was grown on-site. Rather than change these diet fraction assumptions and extent of contamination assumptions individually or in tandem (if necessary), what is done instead is to model four different scenarios relating to cattle exposures. Also, what is done here is to model only the beef cattle exposure. Generally, the trends that result from changes in the diet pattern will be analogous between the beef and dairy cattle. These four scenarios and the parameter changes made are:

Diagram V3 6-1

. Effluent Discharge Source Category:
Section 4.6, Chapter 4, discusses briefly how data from the 104-mill pulp and paper mill study (EPA, 1990b) were used to develop initial parameters required for this source category in its demonstration in Chapter 5. The use of the 104-mill data in a model evaluation exercise is expanded upon in Chapter 7, Section 7.2.3.6. The data is also used here to assign high and low values for four of the seven required parameters for this source category.

Two have to do with flow rates: Qe which is the effluent flow rate, and Qu which is the receiving water flow rate. The range of Qe is from 105 to 107 L/hr, which are the low and high surrounding the 4.1x106 rate used in the demonstration scenario in Chapter 5. The range of Qu is 107 to 109 L/hr (excluding the top ten receiving water bodies, which were in the 1010 L/hr range and for which model did not appear to perform adequately), and these were the low and high around the 4.7x109 L/hr rate used in Chapter 5.

Two parameters describe the suspended solids content of the effluent, TSSe, and the suspended solids content of the receiving water body, TSSu. TSSe ranged from 10 to 250 mg/L in the 104-mill study, so this was the range around the 70 mg/L used as the initial value. Data from STORET used to develop TSSu led to an average of 9.5 mg/L and a range of less than 1 to 50 mg/L; a range of 2 (a reasonable value for a stationary water body such as a pond or lake) to 50 mg/L was tested. One required parameter was, of course, the rate of contaminant discharge, LD, in units of mg/hr.

The assumed value was 0.0315 mg/hr, and this decreased and increased an order of magnitude for low and high testing. The remaining two parameters are the organic carbon contents of effluent solids, OCe, and upstream river suspended solids, OCu. A range based on data was not available for these parameters. OCe was assigned a value of 0.36 based on the fact that solids in effluent discharges are primarily biosolids, and this value was one cited for surface water algae; values of 0.15 and 0.50 were tested. The value of 0.05 for OCu was the value assumed for demonstration of other source categories, where the parameter was called OCssed. The same range of 0.02 to 0.10 for OCssed was used for OCu.

. Stack Emission Source Category:
The parameters in this category listed in Table 6-1 are the only ones which are unique to this source category (one parameter, the no-till mixing depth at the exposure site, dnot, is also used for the off-site soil source category, but its assigned value was 5.0 cm for that source category, and 1.0 cm for the stack emission source category; that is why it is listed for both source categories). As seen, there are only a very few unique parameters.

Most of these are associated with surface water impact, and one series of tests evaluated the impact of parameter changes to surface water concentrations and fish concentrations. These include the contaminant deposition rates, RDEPwat and RDEPsw, which are depositions onto the watershed draining into the surface water body and the surface water body itself (units are m g/m2-yr). The initial values for these were those modeled to occur 500 meters from the stack. This assignment for the stack emission demonstration scenarios, #4 and #5 in Chapter 5, assumes that the stack is located essentially next to the water body. These depositions rates are specific to 2,3,7,8-TCDD.

Rates of 2,3,7,8-TCDD deposition at 200 meters and at 5000 meters were used as high and low values, respectively. It should be noted that depositions are higher at 200 meters and lower at 5000 meters as compared to 500 meters, but air concentrations are lower at 200 meters as compared to 500 meters. This trend occurs because wet deposition is highest nearest the stack. Total depositions are driven by these high wet deposition totals; hence total depositions at 200 meters exceed those at 500 meters. However, dispersion modeling shows that ambient air concentrations of contaminants in the vapor phase (given the wind data and all other parameters and assumptions in using the COMPDEP model for the demonstration scenarios) are highest 500-1000 meters from the stack. For sensitivity testing, differences in model performance as a function of distance from the stack will be evaluated. RDEPp is the deposition of particles themselves and was supplied in order to maintain a mass balance of solid materials entering the water body.

The default value of 0.03 g/m2-yr was taken from Goeden and Smith (1989) for a study on the impacts of a resource recovery facility on a lake. They estimated a total deposition of particles to the lake from all sources was 74.4 g/m2-yr. Assuming the stack is unlikely to contribute all sources of particles to a water body, a high value was chosen as 3 g/m2-yr, and a low value was given as 0.003. The fraction of depositing particles remaining in suspension, fsd, was initialized as 1.00 (meaning that all directly depositing particles remain in suspension) based on an argument that the small particles emitted from the stack and transported directly to the surface water body would settle to surface water bottoms much more slowly than other solids entering water bodies.

A low value of 0.00 was tested (meaning that all solids directly depositing within a year settle quickly to become bottom sediments). The average watershed mixing zone depth, dwmx, was initialized at 0.10 m (10 cm) which is midway between the 1 cm assumed for non-tilled conditions and 20 cm assumed for tilled conditions. This assumption might translate to a rural watershed comprised equally of farmed and unfarmed land. It was reduced to 1 cm and increased to 20 cm in sensitivity testing. A second series of tests evaluated biota impacts at the site of exposure, vegetables/fruits and beef/milk. Parameter inputs for these tests include the ambient air concentration and depositions at the site of exposure, Cva and RDEPe, and the no-till depth of mixing, dnot.

The no-till depth of mixing was increased from 1 to 5 cm. Concentrations and depositions of 2,3,7,8-TCDD at 200 and 5000 meters were tested. The baseline quantities at 500 meters were varied to reflect different vapor/particle partitioning assumptions. Currently, the assumption is that 2,3,7,8-TCDD emissions are 55% in the vapor phase and 45% in the particle phase. Linear adjustments to the emissions in vapor and in particle form can be made to stack emissions. Concentrations and depositions at specific locations are then adjusted in the same linear manner to reflect different vapor/particle partitioning assumptions. Two assumptions tested include 10% vapor/90% particle and 90% vapor/10% particle.

. Contaminant Physical and Chemical Properties:
The initial values for testing of this category of parameters were the ones used for 2,3,7,8-TCDD. Generally, the high and low values tested are those which may represent a range for this contaminant only, not all dioxin-like compounds. However, several of the ranges also encompass values that could be pertinent to other compounds. It should be remembered that this is simply a model performance exercise and nothing else. Also, it could be argued that some of the parameters should be changed in tandem - that there may be a relationship between soil/water adsorption, as modeled by Koc, and bioconcentration. Such relationships were not explored in these exercises. Notes on the parameters are as follows:

1. Henry's Constant, H
- The value of 1.65x10-5 atm-m3/mole was used for 2,3,7,8-TCDD. Except for a heptachloro-PCB, Henry's Constants for the dioxin-like compounds ranged from 10-6 to 10-4. Because of this, the initial value was reduced and then increased an order of magnitude for this test.

2. Molecular Diffusivity in Air, Da
- This parameter is needed for the volatilization flux algorithm. Because no values were available for the dioxin-like compounds, values were estimated based on the ratios of molecular between a dioxin-like compound of interest and a compound for which a Da was available - in this case, diphenyl. The range of values tested are 0.005 cm2/s as a low and 0.10 cm2/s around the initial value of 0.047 cm2/sec.

3. Organic Carbon Partition Coefficient, Koc:
The Koc is perhaps the single most influential parameter in this assessment, impacting surface water concentrations, vapor phase air concentrations, and directly or indirectly, all biomass concentrations (fish, vegetations, beef/milk). The literature for 2,3,7,8-TCDD shows a range of Koc under 106 (from Schroy, et al., 1985) to over 2x107 L/kg (Jackson, et al., 1986). The value selected for 2,3,7,8-TCDD was 2.69x106 based on an empirical relationship between Koc and Kow developed by Karickhoff, et al. (1979) (see Section 4.3.1., Chapter 4). The values tested were one order of magnitude less (2.7x105) and one order of magnitude more (2.7x107)
than the value initially assumed for 2,3,7,8-TCDD.

4. Air-to-Leaf Vapor Phase Transfer Factor, Bvpa:
The initial value for 2,3,7,8-TCDD was estimated as a function of a contaminant's octanol water partition coefficient, Kow, and Henry's Constant, H (Equation 4-30, Chapter 4). This arrives at the volume-based transfer factor, Bvol, which is then transformed to a mass based Bvpa (Equation 4-31). The empirical formulation for Bvpa was developed in a series of experiments by Bacci, et al. (1990, 1992). A second experiment by McCrady and Maggard (1993) demonstrated that the Bacci experiments would overestimate the transfer of 2,3,7,8-TCDD to grass leaves by approximately a factor of 40. An air-to-beef food chain validation exercise described in Chapter 7, Section 7.2.3.9, describes the rationale and model results which led to the approach taken in this assessment for assignment of Bvpa for all congeners: determine a value of Bvol based on Bacci's empirical algorithm, reduce it by a factor of 40 based on the McCrady and Maggard experiments, and transform it to a Bvpa considering the plant density and liquid content of vegetation as given in McCrady and Maggard. For 2,3,7,8-TCDD, the resulting Bvpa is 1.0x105. Plus or minus an order of magnitude will be tested as a high and low value for Bvpa.

5. Particle-Phase Fraction, f :
This fraction was used in the stack emission source category for determining the portion of emitted contaminant that was and remained in the particle phase from stack to exposure site. Details on the measured and theoretical partitioning is given in Chapter 3 of this Volume. As discussed there, measured partitioning of 2,3,7,8-TCDD in ambient showed a very small amount in the particle phase, 13%. However, speculation was that the monitoring method itself could lead to an underestimate in the particle phase, and for that reason, a theoretical approach was used to partition the dioxin. This led to a f of 0.45 for 2,3,7,8-TCDD. The stack emission demonstration will be used to evaluate the impact of instead assuming 0.20 or 0.80 for 2,3,7,8-TCDD f .

6. Root Bioconcentration Factor, RCF:
The initial value for 2,3,7,8-TCDD was estimated as a function of octanol water partition coefficient, Kow. Assuming a log Kow of 6.64, RCF was solved as 3916. Different assumptions for log Kow were used to estimate high and low values of RCF for this exercise. Examining literature Kow for the dioxin-like compounds, no log Kow are less than 6.0 (the lowest at 6.2) and only one value estimated to exceed log Kow equal 8.5. A high and low RCF were estimated, therefore, using log Kow of 6 and 8.5. This led to tested values of RCF of 1260 and 106,000.

7. Beef/milk Bioconcentration Factor, BCF:
Unlike the RCF, Bvpa, and Koc (but like the BSAF and BSSAF as noted blow), there are no empirical formulas developed for BCF as a function of more common parameters such as Kow. The literature summary and interpretation of 2,3,7,8-TCDD cattle feeding studies by Fries and Paustenbach (1990) led them to assign a value of 5.0 for 2,3,7,8-TCDD. The study by McLachlin, et al. (1990) allowed for generation of BCF values for 16 of the 17 congeners of dioxin toxicitiy equivalency, and the results from that study are used for this assessment. The 2,3,7,8-TCDD BCF was 4.3, which is close to the value of 5.0 promoted by Fries and Paustenbach (1990). Their summary, duplicated as Table 4-3 in Chapter 4, showed BCF less than 1.0 for higher chlorinated dioxin-like compounds. For sensitivity testing, values of 1.0 and 10.0 were used as low and high values for BCF.

8. Biota Sediment and Biota Suspended Solids Accumulation Factors, BSAF and BSSAF:
EPA (1993) summarizes several water column based and sediment (both suspended and bottom) based empirical parameters used to estimate fish concentrations given a water or sediment concentration. Two of these are the BSAF and BSSAF, which are used in this assessment. Although no data exists to determine values of the suspended solids factor, BSSAF, EPA (1993) suggests that BSAF values could be used. The range of BSAF values for 2,3,7,8-TCDD discussed in EPA (1993) is 0.03 to 0.30, and this was the low and high values selected for both BSAF and BSSAF. The literature summary on BSAF included in Chapter 4 of this assessment does include studies which imply higher BSAF for 2,3,7,8-TCDD. One study, which focused on bottom feeders (carp, catfish, etc.), found a BSAF for 2,3,7,8-TCDD (CDEP, 1992) of 0.86, whereas the range of 0.03 to 0.30 focused on column feeders. A high value of 2.94 (Kjeller, et al., 1990) was found in a lake in Sweden speculated to be impacted by an active pulp and paper mill. This high value appears to be an outlier not found in other field data sets.

9. First-order Plant Weathering Factor, kw:
This is used to simulate the weathering of contaminated particulates which have settled on plant matter via dry and wet deposition. Several modeling efforts have used the same kw as used in this effort; that kw is 18.01 yr-1, which corresponds to a half-life of 14 days (see Section 4.3.4.2, Chapter 4). Values of 51 (half-life of 5 days) and 8.4 (half-life of 60 days) yr-1 were used to test the impact of this parameter.

10. Dissipation Rate Constant for Eroding or Depositing Contaminants, k:
Evidence for soil degradation of the dioxin-like compounds indicates that residues even millimeters below the soil surface degrade at a very slow rate, if at all (see Chapter 2, Volume 2 of this assessment). This was the basis for not considering degradation of soil sources of dioxin-like compounds in this assessment. However, when residues migrate to impact only a thin layer of soil at a distant site, the processes of volatilization or photolysis (the one degradation process which appears to transform dioxin-like compounds in the environment) are likely to impact delivered residues. A rate constant of 0.0693, which corresponds to a 10-year half-life, was used in two instances for this methodology - for erosion of off-site soils onto exposure site soils, and for deposition of stack emissions onto exposure site soils (see Section 4.4.1 for a further discussion on this parameter). This value was changed to 0.693 (half-life of 1 year) and 0.00693 (half-life of 100 years) yr-1 in sensitivity testing.