Modeling Flux Pathways to Vegetation for Volatile and
Semi-Volatile Organic Compounds in a Multimedia Environment
Ellen J. Cooler
Atmospheric Sciences Modeling Division, Air Resources Laboratory.
National Oceanic and Atmospheric Administration,
Research Triangle Park, NC27711
E-mail, cooter.ellen@epa.gov
On assignment to the National Exposure Research Laboratory,
U.S. Environmental Protection Agency
Yoram Cohen
Department of Chemical Engineering, University of California,
Los Angeles, Los Angeles, CA 90065-1590
1 Abstract
2 This study evaluates the treatment of gas-phase atmospheric deposition in a screening
3 level model of the multimedia environmental distribution of toxics (MEND-TOX).
4 Recent algorithmic additions to MEND-TOX for the estimation of gas-phase deposition
5 velocity over vegetated surfaces are evaluated via recently published dry deposition
6 flux measurements. Model results are compared to similar estimates made by the
7 NOAA multilayer dry deposition model. Results of the evaluation indicate that
8 MEND-TOX performs quite well (r2 =.74), for a screening level model, for the
9 estimation of gas-phase dry deposition velocity of nitric acid over soybeans. As long
10 as the stated model assumptions regarding chemical properties are met, the present
11 study expands previous laboratory results for organic species to include some inorganic
12 species and open field and dry leaf, conditions.
-------
1 * 1. Introduction
2 There are four general types of multimedia modeling approaches: 1} integrated spatial
3 multimedia models; 2) linked spatial single-medium models; 3) compartmental ("well-
4 mixed" media) models; and 4) integrated spatial multimedia compartmental (ISMC)
5 models. All four modeling approaches have evolved significantly over time to reflect
6 changing environmental concerns and the current state of scientific knowledge. The
7 present study evaluates the treatment of gas-phase atmospheric deposition in the
8 Multimedia Environmental Distribution of Toxics (MEND-TOX) model, a recent
9 addition to the screening level ISMC-family. We briefly describe the model, focusing
10 on recently added algorithms to estimate gas-phase deposition velocity over vegetated
11 surfaces. MEND-TOX deposition velocity estimates are then evaluated in light of
12 recently published dry deposition flux measurements and compared to similar estimates
13 produced by the NOAA multilayer dry deposition model (MLM) (Meyers et at, 1998),
14 We conclude with suggestions for farther model improvement and areas in need of
15 additional tasic research or field monitoring efforts.
16 2. The Multimedia Model
17 MEND-TOX is part of the ISMC family of models that has been used previously to
18 evaluate the partitioning of PCBs, PAHs and TCE in California (Cohen and Clay,
-------
1 1994; Tsai, et al, 1991; Cohen, 1996; Ryan and Cohen, 1985), the Great Lakes (Vohra,
2 1996) and the Southeast Ohio River Valley (Ryan and Cohen, 1986). Like these earlier
3 models, MEND-TOX tracks the dynamic distribution of chemicals in a multimedia
4 setting based on a detailed mechanistic description of intermedia transfer processes. It
5 describes environmental media in terms of eight major compartments: air, aerosol, soil,
6 water, sediment, suspended solids, biota, and vegetation. Atmospheric (aerosol and
7 gas), vegetation and aquatic (i.e., water, biota, and suspended solids) compartments are
8 treated as well-mixed (i.e., uniform) and the chemical mass balance in these
9 compartments is expressed via ordinary differential equations. The soil and sediment
10 compartments are taken to be non-uniform, and transport is described by one-
11 dimensional convective-diffusion-reaction equations.
12 Modules to predict precipitation scavenging of gases, infiltration of dissolved
13 solutes into the soil and sediment, intermedia transfer of particle-bound chemicals by
14 dry and wet deposition, wind resuspension and sediment resuspension and deposition
15 are included in MEND-TOX. The basic output from MEND-TOX is in the form of
16 concentrations as a function of time for the uniform compartments and time-dependent
17 spatial concentration profiles for the non-uniform compartments (i.e., soil and
18 sediment). Model output also includes the mass distribution of the chemical in the
19 multimedia environment as well as the time-dependent intermedia mass fluxes,
20 MEND-TOX addresses atmospheric deposition of organic chemicals to water,
21 soil and vegetation media. Exchange across any media boundary utilizes a traditional
22 two-film theory approach expressed as
-------
N =
(1)
1 where KG (cm s"1) and A* (cm s"1) are the overall mass transfer coefficients (MTC)
2 between the gas phase of the chemical in the atmosphere and some adjoining medium,
3 X(Q.g., soil, water, vegetation), and JV is the interfacial mass flux between the two
4 media. H is a dimensionless Henry's Law constant which acts as a partitioning
5 coefficient between the air and the adjacent media. Gas-phase concentrations of the
6 chemical in air and in an adjacent medium are represented by Cg and Q, respectively,
7 It is assumed that the interfacial concentrations are at equilibrium. The values of the
8 MTCs depend on the prevailing turbulence levels in the two media, media temperature,
9 the properties of the solute such as diffusivity, or molecular size and, in some cases,
10 ' media geometry (Cohen, 1986). MEND-TOX treatment of the transfer of atmospheric
1 1 particles and gases to soil and liquid surfaces is described elsewhere (Tsai et aL, 1991;
1 2 Vohra, 1 996). The present analysis focuses only on the recent addition of algorithms
1 3 for the estimation of gas-phase mass transfer from the atmosphere to a vegetated
14 surface. A brief description of these algorithms follows.
15 Paterson et al. (1991) apply a conventional two-compartment chemical kinetic
1.6 uptake-clearance approach, defining the overall air to leaf mass transfer coefficient
17 (MTQ,Jff0tt(«lh"l)«B
y
&QAL = ~7 2 (2)
-------
with,
(3)
2 where k3 is the clearance rate constant (h"1), Fis vegetation volume (m3), A is
3 vegetation area (mz), % (h) is a chemical transfer time in the organic phase, HA (h) is a
4 transfer time in air and KOA is the octanol-air partition coefficient. In the case of gas-
5 phase deposition over vegetation, KOAL is equivalent to the deposition velocity, Vd.
6 Equation 3 implies that air and organic resistance are in series and can be
7 estimated as
and,
(5)
9 where y0 is the mass fraction of octanol in the plant, Kc (cms"1) is the MTC for the
10 plant cuticle and KA (cm s ) is the MTC for the air boundary layer. Octanol is
11 frequently used as a surrogate for Hpids (fats) so that^,, represents the lipid fraction of
12 the plant mass. Equation 3 assumes,that the primary site of chemical accumulation is
13 in the cuticular waxes, which are similar to octanol in partitioning properties.
-------
1 Paterson etal. (1991) tested these relationships for 14 organic compounds in a
2 laboratory setting. Their results indicate that when log(KOA ) is less than 7, r0 is much
3 larger than KOA tA and organic resistance dominates the estimation of Vd. When
4 \Qg(KOA) is approximately equal to 7, r0 and /f^ T^ are of similar magnitude. When
5 log(KOA) is greater than 8, KQA TA becomes much larger than TO and atmospheric
6 resistance controls the Vd estimate.
7 Estimates of TO and TA are needed to evaluate equation 3. TA can be easily
8 approximated (see Section 3,2), However, r0 requires cuticular resistance information.
9 Development of a numerical model for TO is undergoing research. Until such a model
10 is identified and can be added to MEND-TOX, a fixed value of TO is employed so that
11 estimates of Vd will vary most strongly in response to changes in rA for chemicals
12 characterized by logfK^,) values of 8 or greater.
13
14 3. Model Evaluation
15 Although MEND-TOX algorithms for chemical partitioning and particle deposition of
16 PCB's, PAHs and TCE in several geographic settings have been verified (see previous
17 references), similar field studies for gas-phase deposition to vegetation have yet to be
18 reported. One reason for this absence is a lack of suitably detailed organic flux data.
19 Lacking such information, two alternatives for evaluating the MEND-TOX algorithms
20 are explored: selective use of available flux data for inorganic species and comparison
21 of MEND-TOX estimates to those produced by alternative evaluated models.
-------
1 3.1 OBSERVATIONS OF GAS-PHASE FLUX
2 Meyers et al (1998) evaluate the performance of the MLM for inferring dry deposition
3 velocity using an extensive flux monitoring data base. Pollutant iluxes and
4 concentrations were measured from a mobile laboratory at various geographic
5 locations. In one case, ozone (03), sulfur dioxide (SO3) and nitric acid (HN03) were
6 measured above soybeans near Nashville, Tennessee, during the summer and early fall
7 of 1995.
8 At present, MEND-TOX is specifically designed to address the behavior of
9 organic chemical species. Because many of the process algorithms included in MBND-
10 TOX were derived specifically for organic species, MEND-TOX lacks some
11 mechanisms needed to adequately address inorganic chemical behavior in a multi
12 media environment. In particular, MEND-TOX mechanisms do not address reactive
13 inorganics in liquid solution. However, the Meyers et al. (1998) data base allows for
14 preliminary verification studies for the intermedia transport process associated
15 specifically with gas-phase deposition assuming that (1) the gas-phase deposition is
16 controlled primarily by atmospheric resistance factors, (2) deposition is primarily to the
17 leaf cuticle, and (3) that the leaf surfaces are dry. Of the chemicals in the Meyers et al
18 (1998) data base, only HNO3 meets the first two assumptions. Surface wetness at the
19 monitoring site was measured using an RM Young surface wetness sensor, which
20 detected surface moisture associated with dew as well as precipitation. Only dry
21 surface conditions are used in this analysis.
-------
1 Meyers et al. (1998) report the soybeans (ASGROW 5560) were planted
2 within wheat stubble on June 13, 1995. Dry deposition sampling was initiated on June
3 22. The beans went through a rapid growth period from July 10 to August 5 with the
4 leaf area index (LAI, measured using a Licor 2000 plant canopy analyzer) increasing
5 from 1 to about 6. LAI gradually decreased to about 3 by the end of September. By
6 October 11, LAI decreased to about 1 when the beans were mostly stalks and pods.
7 MEND-TOX requires a number of plant- and soil-specific parameters before
8 performing a full mass balance analysis. Since we intend to examine only those model
9 elements that directly impact the estimation of gas-phase deposition velocity, parameter
10 values for a "typical" agricultural soil and soybean crop (e.g., organic carbon content,
11 canopy height, etc.) are provided to the model. MEND-TOX is then provided with
12 reported temperature and windspeed and estimates of leaf surface characteristics and TA
13 for each 2-hr sampling period and a K04L, i.e., gas-phase Vd, estimate is produced. The
14 process is repeated for each of the 38 valid observation periods. The mean and
15 • standard deviation of monitored and MEND-TOX Vd are computed and compared.
16 Model bias, defined by Meyers et al. (1998) as (observed Vd - model Vd), is computed
17 for the MEND-TOX results and compared to that reported in Meyers et al. (1998) for
18 HN03.
19 3.2 MEND-TOX INPUT PARAMETERS
20 MEND-TOX inputs critical to the estimation of gas-phase Vd over vegetation include
-------
1 plant composition, surface area of foliage, equilibrium partition coefficient and values
2 for KA and Kc.
3 The fraction of plant mass contained in the foliage can vary significantly over
4 the physiological life of the plant as a function of cumulative biomass partitioning
5 which varies throughout the growing season. A representative partitioning pattern is
6 reported by Wilkerson et al. (1983). This pattern is modified slightly such that, stem
7 and leaf partitions are combined to represent a single MEND-TOX vegetation
• 8 compartment. During the early period of rapid vegetative growth, biomass is
9 partitioned as 53% to above ground elements and 47% to roots. This stage is followed
10 by a period of decreasing leaf and root growth and a transition to reproductive growth.
11 During this transition period, daily biomass partitioning to above ground plant elements
12 increases, in an approximately linear fashion, from 53% to 90%, with the majority of
13 the above-ground accumulation going to the stem to support pod and seed growth and
14 development. After growth of pods and seeds has begun, it is assumed that no
15 additional biomass partitions to the foliage.
16 The surface area of foliage available to intercept the depositing gas is
17 estimated as the ratio of leaf area to leaf mass. Plant physiologists refer to this ratio as
18 the specific leaf area (SLA, cm2 gm"1). SLA varies throughout the growing season as a
19 function of biomass accumulation and leaf area.
20 The fugaciry approach to the estimation of the leaf/air equilibrium partition
21 coefficient is adopted. This approach requires estimates of the fraction of air (yj,
22 water (yw) and lipid (y0) contained in the plant, A small, but non-zero value of 0.01 is
-------
1 assigned toyA. Miller(1938) suggests values of ^ranging from 0.6 for apple to 0.86
2 for cabbage plants. A value of 0.77 is reported for corn and is used here for soybean as
3 well. Wilkerson, et al. (1983) suggest that a reasonable lipid fraction for soybean (not
4 including seeds) is about 0.04.
5 The final set of input parameters is leaf volume, leaf area, TO and TA. The ratio
6 of leaf volume to leaf area, V/A, can be approximated as the average leaf thickness for
7 the plant. Laboratory studies suggest that a "typical" leaf thickness for conventional
8 soybeans ranges from 180 to 200um (Dr. James Dunphy, North Carolina State
9 University Extension Soybean Specialist, personal communication).
10 Nitric acid has been selected to demonstrate a case in which precise estimation
11 of variations of t0 to environmental factors is not critical. A fixed value of 2.0 x 10"u
12 (hr), suggested by the parameterizations described in Wesely (1989), is assigned to TO
13 throughout the analysis.
14 , A value for rx is computed for each sampling period in the HNO3 data set. In
15 this case, KA'' is assumed to be the atmospheric boundary layer resistance, .Rj.
16 Atmospheric resistance is impacted by stability or turbulence conditions, which can
17 vary widely in the field and cannot be ignored if model output are to be compared to
18 monitored values. All HNO3 observations were taken during daylight hours. Hicks et
19 al, (1987) suggest that aerodynamic resistance (Ra) during daylight hours (unstable
20 atmospheric conditions) can be computed as
* ~~
10
-------
1 , where u is the windspeed (m s"1) and tjg is the standard deviation of the wind direction,
2 a surrogate for turbulence. Upon determination of Ra, an internally consistent value for
3 the friction velocity, M, is obtained from the approximation
Ra - uu,-2 (7)
4 This procedure has been shown to provide very good estimates of M, for a wide range
5 of atmospheric stabilities (Erisman and Duyzer, 1991). An estimate of Rh, conditioned
6 on atmospheric stability, can then be estimated as
*. - *- 0)
7 where k is the Von Karman constant (0.4 ), and Sc is the Schmidt number for HN03,
8 Daily plant biomass and leaf area are needed to properly evaluate MEND-
9 TOX performance. A U.S. Department of Agriculture simulation model, the Erosion
10 Productivity Index Calculator (EPIC) (Williams et al, 1984; Williams and Renard,
11 1985), previously employed to estimate LAI for input to the MLM (Cooter and
12 Schwede, 2000) is employed. EPIC plant parameters were tuned so that model output
13 matched the LAI and phonological information reported by Meyers et al. (1998).
14 4. Results and Discussion
II
-------
1 An estimate of KOAL for each valid Meyers et al. (1998) HNOj observation was
2 estimated as described in Section 3 and compared to observations (Figure 1), A linear
3 correlation value (r2) of 0.74 was determined for the data set. Summary statistics for
4 the model estimates and observations are provided in Table 1. These results suggest
5 good overall gas-phase dry deposition velocity estimation capability for HN03 over
6 soybean (i.e., bias near 0).
7 Meyers et al. (1998) describe the performance of the NOAA-MLM model
8 given the same meteorological data and plant condition (LAI) observations. Their
9 results for soybeans, reported as model to observed bias, are provided in Table 1.
10 Meyers et al. (1998) conclude that the MLM showed good agreement in the mean Vd,
11 with very small average bias. However, for specific periods the model either
12 underpredicted or overpredicted Vd, leading to relatively low estimates of linear
13 correlation (Dr, Peter Finkelstein, NOAA/ARL, personal communication). MEND-
14 TOX provides comparable estimates of mean Vd and bias, but appears to more closely
15 follow linear patterns in the observations, as indicated by fairly strong r2 results. Model
16 characteristics of MEND-TOX and the MLM most likely to account for these estimate
17 differences include modeling approach ("big-leaf vs multilayer canopy), biomass
18 parameterization (SLA vs LAI) and estimation of atmospheric resistance components
19 (Rh explicit and Ra implicit vs both/?;, and ^explicit).
20 5. Conclusions
12
-------
1 This summary is far from a full verification of the MEND-TOX dry deposition
2 algorithms. Previous laboratory results (Peterson et a/., 1991) suggest the MEND-
3 TOX estimates should be reasonable for certain groups of organic chemicals. Given
4 that the stated model assumptions regarding chemical properties are met, the present
5 study has expanded these results to include some inorganic species under open field, as
6 opposed to controlled laboratory, and dry leaf conditions
7 The greatest strengths of the MEND-TOX model are its ease of use and wide
8 applicability for organic chemicals (within specified limits). Its liabilities are,
9 primarily, its focus on organic chemicals and an inability to fully address the effects of
10 stomatal and cuticular resistance on Vd. The latter represents a significant limitation for
11 many organic and inorganic chemical species and, as our understanding of these
12 processes improve, should be addressed in future model versions.
13 Although there are many modeling and monitoring studies of O3, sulfur and
14 nitrogen deposition to soil, water and vegetation surfaces, there are few studies similar
15 in quality and scope to Meyers et al. (1998) for gas phase organic species. Monitoring
16 for such chemicals is expensive, time consuming and difficult. However, it is only
17 through such efforts that we can realistically verify the utility of scientifically sound,
18 inexpensive, widely applicable, models such as MEND-TOX for initial screening-level
19 multimedia assessments of volatile and semi-volatile chemicals.
13
-------
1 Disclaimer
2 The information in this document has been funded wholly or in part by the United
3 States Environmental Protection Agency. It has been subjected to Agency review and
4 approved for publication. Mention of trade names or commercial products does not
5 constitute endorsement or recommendation for use.
6 References
7 Cohen, Y.: 1996, 'Volatile organic compounds in the environment: A multimedia
8 perspective', in Wang, W., Schnoor, J., and Doi, J. (eds.), Volatile Organic Compounds
9 in the Environment, ASM STP 1261, American Society for Testing and Materials, 100
10 Barr Harbor Drive, Wet Conshohocken, PA, pp7-32.
11 Cohen, Y. and Clay, R.E.: 1994, 'Multimedia partitioning of particle-bound organics',
12 Journal of Hazardous Materials 37, 507-526.
13 Cooter, E.J. and Schwede, D.B.: 2000, 'Sensitivity of the National Oceanic and
14 Atmospheric Administration multilayer model to instrument error and parameterization
15 uncertainty', J. ofGeophys. Res. 105, D5, 6695-6707.
16 Erisman, J.W. and Duyzer, J.: 1991, 'A micrometeorological investigation of surface
17 exchange parameters over heathland', Boundary Layer Meteorol. 57, 115-128.
14
-------
1 Hicks, B.B., Baldocchi, D.D., Meyers, T.P. and Matt, D.R.: 1987, 'A preliminary
2 multiple resistance routine for deriving dry deposition velocities from measured
3 quantities,' Water Air Soil Pollut. 36, 311-330.
4 Meyers, T.P., Finkelstein, P., Clarke, J., Ellestad, T.G. and Sims, P.P.: 1998, 'A
5 multilayer model for inferring dry deposition using standard meteorological
6 measurements,' J, ofGeophys. Res., 103, D17, 22,645-22,661.
7 Miller, E.G.: 1938, Plant Physiology with Reference to the Green Plant, Mcgraw Hill,
8 New York,
9 Paterson, S., Mackay, D., Bacci, E. and Calamari, D,: 1991, 'Correlation of the
10 equilibrium and kinetics of leaf-air exchange of hydrophobic organic chemicals',
11 Environ. Sci. Technol, 25, 866-871.
12 Ryan, P. and Cohen Y.: 1986, 'Multimedia distribution of particle-bound pollutants:
13 Benzo(a)pyrene test case', Chetnosphere 15, 21 -47.
14 Ryan, P.A. and Cohen, Y.: 1985, 'Multimedia modeling of environmental transport:
15 Trichloroethylene test case*, Environ. Sci. Technol. 19,412-417.
15
-------
1 Tsai, W., Cohen, Y., Sakugawa, H., and Kaplan, I.: 1991, 'Dynamic partitioning of
2 semivolatile organics in gas/particle/rain phases during rain scavenging', Environ. Set.
3 TechnoL 25,2012-2023.
4 Vohra, R.: 1996, 'Dynamic Partitioning and Intermedia Transport of Hydrophobia
5 Chemicals in Aquatic Systems', Masters Thesis, Chemical Engineering Department,
6 University of California, Los Angeles, 191pp.
7 Wesely, M.W.: 1989, 'Parameterizations of surface resistances to gaseous dry
8 deposition in regional scale numerical models', Atmos. Environ. 23, 1293-1304.
9 Wilkerson, G.G., Jones, J.W., Boote, K.J., Ingram, K.T. and Mishe, J.W.: 1983,
10 'Modeling soybean growth for crop management', Trans. oftheASAE 26,67-73.
11 Williams, J.R., Jones, C.A. and Dyke, P.T,: 1984, *A modeling approach to
12 determining the relationship between erosion and soil productivity', Trans. ASAE 27,
13 129-144.
14 Williams, J.R. and Renard, K.G.: 1985, 'Assessments of soil erosion and crop
15 productivity with process models (EPIC)', in R.F. Follett and B.A. Stewart (eds.), Soil
16 Erosion and Crop Productivity, Am. Soc. Of Agron., Madison, Wis., pp 67-103.
16
-------
Figure 1. Modeled vs measured deposition velocities for nitric acid over a, soybean
field near Nashville, TN.
5-
Measured (cm s )
Table 1. Comparison of observed, MEND-TOX andNOAA-MLM HNO3 gas-phase
deposition velocity (cm s"1) over a soybean field near Nashville, TN.
N
Mean
Std.Dev.
Measured
38
2,860
0.941
MEND-TOX
38
2.860
1.060
MEND-TOX Bias
38
0,002
0.724
NOAA-MLM Bias
38
0.220
1.010
17
-------
NERL-RTP-AMD-00-151 TECHNICAL REPORT DATA
1 . REPORT NO . 2 ,
EPA/6QO/A-00/080
4. TITLE AMD SUBTITLE
Modeling Flux Pathways to Vegetation for Volatile and Semi-Volatile Organic
Compounds in a Multimedia Environment
7 . AUTHOR { S }
'Cooler, Ellen J., 2Yoram Cohen
9. PERFORMING ORGANIZATION NAME AND ADDRESS
'Same as Block 12
'University of California
Los Angeles, CA
12. SPONSORING AGENCY NAME AND ADDRESS
National Exposure Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 277 1 1
5, REPORT DATE
6. PERFORMING ORGANIZATION CODE
8, PERFORMING ORGANIZATION REPORT NO,
10.PROGRAM ELEMENT NO.
1 1. CONTRACT/GRANT NO.
13.TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/9
15. SUPPLEMENTARY NOTES
! 6. ABSTRACT
This study evaluates the treatment of gas-phase atmospheric deposition in a screening level model of the multimedia
environmental distribution of toxics (MEND-TOX), Recent algorithmic additions to MEND-TOX for the estimation of gas-
phase deposition velocity over vegetated surfaces are evaluated via recently published dry deposition flux measurements.
Model results are compared to similar estimates made by the NOAA multilayer dry deposition model. Results of the
evaluation indicate that MEND-TOX performs quite well (rj = .74), for a screening level model, for the estimation of gas-
phase dry deposition velocity of nitric acid over soybeans. As long as the stated model assumptions regarding chemical
properties are met, the present study expands previous laboratory results for organic species to include some inorganic species
and open field and dry leaf conditions.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS ^IDENTIFIERS/ OPEN ENDED TERMS
1 8. DISTRIBUTION STATEMENT 19, SECURITY CLASS (This Report)
20. SECURITY CLASS (This Page)
c.COSATI
21, NO. OF PAGES
17
22. PRICE
------- |