Relative influences of uncertainty in physical-chemical property data and variability in climate parameters in determining the fate of PCBs

(1)

Relative influences of uncertainty

in physical-chemical property data

and variability in climate

parameters in determining the fate

of PCBs

Zhe Li

Student Degree Thesis in Chemistry 45 ECTS Master’s Level

Report passed: 26th May 2011

(2)

I

ABSTRACT

Persistent organic pollutants (POPs) present a class of substances which are of high concern due to their toxicity and ecotoxicity, long-range atmospheric transport and resistance to degradation. A combination of analytical measurement and modeling research is a powerful means for understanding chemical fate and behavior. Among all available modeling approaches, multi-compartmental fugacity modeling is a common and relatively simple way to simulate the concentrations and distributions of pollutants. This approach is also applied here to study a well-studied group of POPs – polychlorinated biphenyls (PCBs).

Climate change is potentially one of the largest global environmental problems facing society today. Climate change resulting from anthropogenic activities may impact the transport, fate and exposure of POPs due to alterations in climatic and landscape properties (e.g. temperature, wind speed, precipitation etc.). Modeling studies have previously been carried out to investigate their influence on POPs fate under defined scenarios. However, in previous work the uncertainty of chemical properties has not been considered. We hypothesize that the uncertainties in physical-chemical parameters could actually cause a greater impact predicted concentrations than variability in climate parameters such as temperature, wind speed and precipitation.

To test the hypothesis, a 4-compartment Level III fugacity model – ChemCAN spreadsheet model is used. A combination of 8 climate variables, 3 physical-chemical properties and 4 degradation properties are chosen to investigate how the variability and uncertainty affect the environmental fate of 6 selected PCBs congeners. In total, 18 forecasts are created in the investigation and outputs provided for concentrations, distributions, total amounts, long-range transport potential indicators and persistence of the chemicals. We arbitrarily model the Japan region for which we had environmental parameterization in the ChemCAN model. Monte-Carlo simulations are undertaken using the Crystal Ball® computer software to perform correlation and sensitivity analyses, in order to specifically examine and compare the influence brought from variability and uncertainty. Data for climate variables are collected and calculated from Intergovernmental Panel on Climate Change – Data Distribution Centre (IPCC-DDC), while those for chemical property uncertainty are collected and calculated from literature sources/methods.

The results are interpreted focusing on the two types of analyses – correlation and sensitivity. Within the correlation analysis between climate variables and PCB fate predictions, temperature is the most important factor affecting total amount, concentrations and distributions, especially in the atmosphere compartment. It also influences the fugacity of the PCBs in each media, as well as the reactive residence times. The uncertainty of temperature increases resulting from climate change has more influence than variability in mean temperature however. Precipitation hardly has any influence on PCB fate, although the change in precipitation due to climate change could affect distributions in surface water and surface soil for light chlorinated PCBs congeners due to the extremely high coefficient of variation. Wind speed also has little impact on predictions other than for long-range transport potential descriptors, namely characteristic transport distance and transfer efficiency. Wind speed also affects the advective residence time in the air compartment for PCBs. Within the correlation analysis between properties and PCB fate predictions, all the chosen 3 physical-chemical properties, vapor pressure, water solubility and water-octanol partition coefficient, significantly affect the predictions. Vapor pressure has more influence on predictions related to air and more significantly on lighter PCBs, while water solubility and the water-octanol partition coefficient mostly affect the heavier PCBs, especially

(3)

II

the predicted outcomes in water and sediment. Despite a fairly small uncertainty in physical-chemical properties compared to many other chemicals, the influence on PCB fate is noticeable. Degradation properties, which have a rather high uncertainty, also significantly affect PCB environmental fate and degradation half-lives in air and soil have much more significant influence than those in water and sediment.

In the sensitivity analysis, variability and uncertainty are assembled into a single simulation. When excluding degradation properties, it could be concluded that uncertainty in physical-chemical properties is usually dominant for PCBs. One notable exception is that precipitation change has a greater influence on PCB-8, which is also the lightest included in the study. If degradation properties are included, degradation half-lives in air and soil are always the dominant source of uncertainty to predictions.

The main conclusion of this thesis is that the uncertainty in chemical property data is generally more important than variability in climate parameters in controlling variance in predicted environmental concentrations. This supports the hypothesis of that was raised at the beginning of this study.

Key word: ChemCAN, PCBs, climate change, uncertainty, physical-chemical property, degradation

(4)

III

ACKNOWLEDGEMENTS

First and foremost, I would like to express my utmost respect and sincere gratitude to my supervisor Ian Cousins, who not only allowed me to work on this unique project but also taught me the importance of exact academic research. I am grateful for all the patience and help he gave me during the entire project. I enjoyed working with him on a scientific project.

I greatly appreciate the guidance on statistics and mathematics provided by Matthew MacLeod, who was always open for interesting discussions during my thesis project. I would like to thank him from all my heart being so kind and helpful all the time.

I would also like to thank my course teacher Karin Wiberg, who offered me continuous support in my studies and kindly agreed to be my examiner in the thesis project. I do wish her a fabulous career in Uppsala.

As well, I want to thank Deguo, Robin, Xiaole and Hongyan, who are senior to me as PhD students. They helped me on each tiny detail whenever I got confused. I would remember all the patient talking they gave me on each and every issue.

I would like to thank ITM for providing me a peace environment during the project and UmU for funding support throughout my research.

Thank my intimate friends Yingrong, Stella, Ge Yin and Van Anh, without whose accompany I could never make it here. Those toleration and intimacy would be my precious fortune.

Lastly, I owe special gratitude to my family for unconditional support all the time: To my parents for the motivation they gave me during those tiring times, to my aunt and cousin for the utmost concern in both my life and study, to my grandparents for their encouragements whenever I had doubts about career, and to Jack Lo for his enduring patience, understanding, and selfless love.

(5)

IV

List of Figures

Figure 2.1. Illustration of the ‘global fractionation’ hypothesis and ‘grasshopper’ effect (AMAP, 2002).

... 4

Figure 2.2. Illustration of the main environmental processes during LRAT of POPs. ... 5

Figure 2.3. Illustration of partition coefficients among air, water and octanol phases. ... 7

Figure 2.4. Basic chemical structure of PCBs (n changes from 1 to 5)... 9

Figure 3.1. Illustration for structure of the storylines and scenarios in the IPCC Special Report on Emissions Scenarios (SRES). ... 12

Figure 5.1. Map of study area. ... 15

Figure 5.2. Illustration of a four-compartmental environment fugacity Level III model for assessing the environmental fate of PCBs emissions in a regional environmental system. ... 16

Figure 6.1. Correlation analysis of climate variables to predictions of PCB-52. ... 25

Figure 6.2. Correlation analysis of property uncertainties to predictions of PCB-52. ... 27

Figure 6.3. Sensitivity analysis of variances in concentration predictions of selected PCBs. ... 31

Figure 6.4. Sensitivity analysis of variance in distribution predictions of selected PCBs. ... 31

Figure 6.5. Sensitivity analysis of variance in fugacity predictions of selected PCBs. ... 31

Figure 6.6. Sensitivity analysis of variance in persistence predictions of selected PCBs. ... 32

Figure 6.7. Sensitivity analysis of variance in LRTP indicator predictions of selected PCBs. ... 32

Figure 6.8. Sensitivity analysis of variance in total amount predictions of selected PCBs. ... 32

Figure 6.9. Sensitivity analysis of variance in concentration predictions of selected PCBs ... 34

Figure 6.10. Sensitivity analysis of variance in distribution predictions of selected PCBs ... 34

Figure 6.11. Sensitivity analysis of variance in fugacity predictions of selected PCBs ... 34

Figure 6.12. Sensitivity analysis of variance in persistence predictions of selected PCBs ... 35

Figure 6.13. Sensitivity analysis of variance in LRTP indicator predictions of selected PCBs ... 35

Figure 6.14. Sensitivity analysis of variance in total amount predictions of selected PCBs ... 35

Figure A.1. Illustration for transfer efficiency in three-compartmental environment. ... 47

Figure C.1. Correlation analysis of climate variables to predictions of PCB-8. ... 50

Figure C.6. Correlation analysis of physical-chemical properties to predictions of PCB-8. ... 55

Figure C.11. Correlation analysis of degradation properties to predictions of PCB-8. ... 56

Figure C.12. Correlation analysis of degradation properties to predictions of PCB-31. ... 57

Figure C.13. Correlation analysis of degradation properties to predictions of PCB-101... 57

Figure C.14. Correlation analysis of degradation properties to predictions of PCB-153... 57

(8)

VII

List of Tables

Table 2.1. List of the initial 21 POPs under the Stockholm Convention for reduction or elimination. .... 3

Table 2.2. Types and descriptions of different POPs environmental fate models ... 6

Table 3.1. List and assumption descriptions for the defined SRES scenarios in IPCC AR4. ... 12

Table 3.2. Possible impact of climate change on POPs behavior (Lamon et al., 2009a). ... 13

Table 3.3. Previous modeling studies that have estimated the effect of climate change of POP transport and fate. ... 13

Table 4.1. Summary of uncertainty existing in multi-media modeling. ... 14

Table 5.1. Properties of the four compartments in the Japan regional model. ... 15

Table 5.2. Summary of Z value definitions in compartments and subcompartments. ... 16

Table 5.3. Summary of gain and loss processes of contaminants among the four-media environment. 17 Table 5.4. Summary of transfer D parameters within the four-compartmental environment. ... 18

Table 5.5. Summary of physical-chemical properties, emissions and half-lives for selected PCBs... 20

Table 5.6. Summary of physical-chemical properties and their uncertainties for selected PCBs (raw data are listed in Table B.3). ... 21

Table 5.7. Summary of degradation half-lives (h) in different media and their uncertainties for PCBs. 22 Table 5.8. Summary of tested input parameters and outcomes in Monte-Carlo simulation ... 23

Table A.1. Fraction parameter descriptions in the background inflow calculation. ... 46

Table B.1. Regional information for Japan and climatic variables information, including climate scenario, model, observational and projected period, and data source. ... 48

Table B.2. Statistic data for climatic variables, including temperature (°C), precipitation (m/h), eastward wind speed and northward wind speed (m/h). ... 48

Table B.3. Raw data, including literature derived value (LDV) and final adjusted value (FAV), for physical-chemical properties of the selected PCBs. ... 49

Table C.1. Sensitivity analysis of contribution to predicted concentrations for the selected PCBs (excluding degradation property). ... 59

Table C.2. Sensitivity analysis of contribution to predicted distributions for the selected PCBs (excluding degradation property). ... 59

Table C.3. Sensitivity analysis of contribution to predicted fugacity for the selected PCBs (excluding degradation property). ... 60

Table C.4. Sensitivity analysis of contribution to predicted persistence for the selected PCBs (excluding degradation property). ... 60

Table C.5. Sensitivity analysis of contribution to predicted LRTP indicators for the selected PCBs (excluding degradation property). ... 61

Table C.6. Sensitivity analysis of contribution to predicted total amount for the selected PCBs (excluding degradation property). ... 61

Table C.7. Sensitivity analysis of contribution to predicted concentrations for the selected PCBs (including degradation property). ... 62

Table C.8. Sensitivity analysis of contribution to predicted distributions for the selected PCBs (including degradation property). ... 63

Table C.9. Sensitivity analysis of contribution to predicted fugacity for the selected PCBs (including degradation property). ... 64

Table C.10. Sensitivity analysis of contribution to predicted persistence for the selected PCBs (including degradation property). ... 65

(9)

VIII

(including degradation property). ... 66

Table C.12. Sensitivity analysis of contribution to predicted total amount for the selected PCBs

(10)

1 |

1 Introduction

1.1 Background and motivations

In the rush to industrialize, the planet has become ubiquitously contaminated with chemicals. Many thousands of substances have been released into the environment with few prior checks on their potential for causing long-term harm (Beyer et al., 2000; Hansen et al., 2008; El-Shahawi et al., 2010). Wildlife throughout the world has been contaminated and many species harmed. Human beings have also been affected. Among all the harmful substances present in the environment, persistent organic pollutants (POPs) are the most problematic (Macdonald et al., 2003; Barber et al., 2004; Lohmann et al., 2007).

POPs present a class of substances which are of high concern due to their toxicity, ecotoxicity and resistance to degradation. Together with the additional evidence of long-range transport of POPs to remote regions and the consequent threats they pose to the global environment, the international community has now, on several occasions called for urgent global actions to reduce and eliminate releases of these chemicals (Paasivirta et al., 1999; Macdonald et al., 2003; Breivik, 2004; Hansen et al., 2008). It is a challenge to fully understand the dynamic processes which determine the environmental fate of POPs, but efforts have been to measure their concentrations, distributions, migrations, and transformations etc. (Kim and Masunaga, 2005; Kobayashi et al., 2010). Besides laboratory and field studies, modeling is a useful means to study environmental fate of POPs because they facilitate the identification of key transport and fate processes from the many multiple processes occurring simultaneously (Sweetman et al., 2002; Mackay et al., 2003; Wania and Dugani, 2003; Webster et al., 2004). Multimedia compartmental mass balance models (sometimes known as fugacity models) are useful when simulating and predicting POP concentrations and distributions among media (Mackay, 1991; Di Guardo et al., 1994; Mackay et al., 1996; McKone, 1996; Wania et al., 1999).

When simulating and predicting POP concentrations and distributions using a multimedia environmental modeling approach, over- or under-estimations will exist due to various uncertainties (Morgan and Henrion, 1990). To date, only a few studies have aimed at exploring the effect of parametric uncertainty on POP fate modeling (McKone, 1996; McKone et al., 1996; Bennett et al., 1998; MacLeod et al., 2002; Hauck et al., 2008; Schenker et al., 2009). These few studies have identified the importance of uncertainty in physical-chemical properties of substances in controlling the uncertainty in model predictions. Input uncertainty is caused by a lack of knowledge of the true values and chemical properties can often be poorly defined this input uncertainty combined with the high sensitivity of chemical properties on model outputs results in a high output uncertainty.

Along with chemical pollution, climate change is another global environmental problem of high concern. Modern climate change is thought to be caused by anthropogenic activities (Macdonald et al., 2003; Macdonald et al., 2005; Lamon et al., 2009a; Noyes et al., 2009). These climate changes include (but are not limited to) atmospheric temperature, oceanic salinity, precipitation patterns, wind patterns and so forth. Global temperature is increasing, which directly results in glacier ice melting and sea level rises. Other consequences also exist, such as the changing of the oceanic and atmospheric currents, and the higher frequency of extreme weather events (Lamon et al., 2009a; Noyes et al., 2009).

The influence of climate change on the transport and fate of POPs has been previously reviewed and many potential climate related effects suggested (Macdonald et al., 2003; Macdonald et al., 2005). Monitoring datasets are not sufficiently detailed and long-term to reveal any clear effects of climate change on measured environmental concentrations. More recently, researchers have also attempted to

(11)

2 |

model the effects of climate change on predicted concentrations. However, modeling of climate change and POPs is a new area of research and there is much still to be learnt about the possibilities and appropriate methodologies (McKone et al., 1996; Dalla Valle et al., 2007; Lamon et al., 2009a; Lamon et al., 2009b; Ma and Cao, 2010).

In summary, previous work has been done on comparing the significance of the influence of physical-chemical property uncertainty and landscape/climate parameters on change on the transport and fate of POPs with the conclusion that physical-chemical properties often dominate output uncertainty. This suggest that climate change may only have a small effect on the transport and fate of POPs and that this effect may be obscured by the larger uncertainty associated with the chemical parameter uncertainty.

1.2 Hypothesis and objectives

As stated above, even though climate change has been reported to have an influence on the fate of POPs, there is still a lack of comparison between the influence from climate variables and physical-chemical property uncertainties. Polychlorinated biphenyls (PCBs) are selected to test the relationship within and between landscape and chemical properties. One reason for selecting PCBs was that they have been well-studied and thus exhibit a relatively small uncertainty in their physical-chemical properties. Another is that they are a homologues class of compounds with 209 congeners exhibited a range of fate behavior. This present study hypothesizes that the influence of chemical property uncertainty is much more significant than that of climate variability on PCB fate even though physical-chemical properties of PCBs are relatively well known. We believe that if we can prove this hypothesis for PCBs we can transfer the conclusions to most of POPs.

Vapor pressure, water solubility and octanol-water partition coefficient (KOW) are selected as the

studied physical-chemical properties since all of them are key inputs in the model. As for the climate variables, temperature, precipitation (rain fall) and wind speed are chosen, since these parameters are important environmental inputs in the fate model and can be readily extractable from climate models. We arbitrarily model the Japan region for which we had environmental parameterization in the ChemCAN model.

The study can be divided into two individual phases. First was a data collection phase. Physical-chemical property data for selected PCBs congeners were collected and estimates made of their inherent uncertainty. Climatic data were also collected from reports from the Intergovernmental Panel on Climate Change (IPCC). The second project stage was undertaking the model calculations. The model was selected and modified for use with Crystal Ball® for Excel. Monte-Carlo simulations were then run to the contribution of individual input parameters to variance in selected model outputs were quantified and ranked.

(12)

3 |

2 Persistent Organic Pollutants (POPs) – A Global Problem

2.1 POPs definition and classification

In 1995, the United Nations Environmental Program (UNEP) defined and classified POPs with the aim of reducing or eliminating their production and use. Generally, POPs are defined as a group of substances that are persistent and toxic in the environment, and could undergo long-distance transport, as well as bioaccumulation through the food web (Macdonald et al., 2003; Macdonald et al., 2005; Lohmann et al., 2007; Lamon et al., 2009a; Lamon et al., 2009b; El-Shahawi et al., 2010). Due to their capability of being transported to remote regions like the Arctic where POPs have never been used and their particular resistance to degradation, environmental contamination caused by POPs had quickly been considered as a global issue of high concern (Dalla Valle et al., 2007; Lamon et al., 2009a; Lamon et al., 2009b). So far, several international agreements have been made, focusing on the elimination of certain POPs, with typical instances being the Aarhus Protocol and Stockholm Convention on POPs. The Stockholm Convention was adopted in 2001, listing the initially twelve chemicals, which are also known as ‘the dirty dozen’, while nine more substances were added into the list in 2009.

Table 2.1. List of the initial 21 POPs under the Stockholm Convention for reduction or elimination.

Category Substance

Chemicals produced intentionally

pesticides or fungicides used in agriculture to control insects or fungus

aldrin, chlordane, chlordecone, dichlorodiphenyltrichloroethane (DDT), dieldrin, endrin, heptachlor, hexachlorobenzene (HCB),

gamma-hexachlorocyclohexane (γ-HCH, lindane), mirex, toxaphene

industrial chemicals used in various application, such as flame retardants, plasticizers, and dielectric fluids etc.

tetra-, penta, hexa- and heptabromodiphenyl ethers (PBDEs), hexabromobiphenyl, perfluorooctane sulfonic acid (PFOS),

perfluorooctane sulfonyl fluoride (PFOS-F), pentachlorobenzene (PeCB), polychlorinated biphenyls (PCBs)

Chemicals formed unintentionally

chemicals generated unintentionally or as by-products as a result of incomplete combustion or chemical reactions

hexachlorobenzene (HCB), pentachlorobenzene (PeCB), polychlorinated biphenyls (PCBs) and polychlorinated dibenzo-p-dioxins and

dibenzofurans (PCDD/Fs), and by-products of lindane [alpha-hexachlorocyclohexane (α -HCH) and

beta-hexachlorocyclohexane (β -HCH)]

Most POPs are anthropogenic in origin while only a few have natural sources. They are simply divided into two categories, referring to chemicals produced intentionally or formed unintentionally (Breivik, 2004; Lohmann et al., 2007; El-Shahawi et al., 2010). To a large extent, the former group contains pesticides and most industrial products whereas the latter primarily includes the chemicals from combustion and some industrial by-products. Within the chemicals produced intentionally, two specific groups are further divided, referring to those applied in agriculture and industry. Table 2.1 summarized the list and specific classification of initial 21 POPs given in the Stockholm Convention.

So far, numerous studies have been carried out to explore further knowledge on POPs, especially their behavior and fate in the multicompartmental environment. Since the behavior of POPs depends on the interaction of many intrinsic and extrinsic factors, it could be evidently influenced by the alteration in the environment.

(13)

4 |

2.2 POPs long-range transport and environmental fate

The environmental behavior and fate of POPs is mostly determined by their physical-chemical properties and the environmental characteristics (Hansen et al., 2008; Beyer and Biziuk, 2009; Paasivirta and Sinkkonen, 2009). Classical POPs have low water solubility, high lipophilicity and volatility or semi-volatility. Retrospective POP concentration data in remote regions provide empirical evidence that most POPs would undergo long-range transport (LRT) after emission, moving with air or ocean currents. The semi-volatility of many POPs makes the atmosphere an important transport media leading to global distribution, while for some more water soluble POPs such as HCHs and perfluorinated acids (PFOS and PFOA), ocean currents are a relatively more important pathway (Lohmann et al., 2007; Meyer and Wania, 2007).

Due to the difference in vapor pressures, compounds have different atmospheric travel distances, a process known as ‘global fractionation’, which is presented in Figure 2.1 (AMAP, 2002). This term is also a hypothesis widely accepted about the long-range atmospheric transport (LRAT) and distribution of POPs. In addition, POPs in the air phase could be readily adsorbed to particles because of their high lipophilicity and exchange between gaseous and airborne particles phase depending on temperature. These adsorbed POPs undergo either dry deposition or wet deposition (normally precipitation) to be washed out from atmosphere into the surface of soil or vegetation, or to be dissolved in water. Under warm weather conditions, the compounds re-evaporate into the atmosphere and undergo further transport via air. POPs can undergo several transport-deposition-revolatilization processes, a phenomenon which is termed the ‘grasshopper effect’ (Figure 2.1) (AMAP, 2002; Gouin et al., 2004b; Hansen et al., 2008).

Figure 2.1. Illustration of the ‘global fractionation’ hypothesis and ‘grasshopper’ effect (AMAP, 2002).

Depending on physical-chemical properties and through dynamics processes, POPs can distribute and transport among all the media in the environment and this is illustrated in Figure 2.2. POPs in the air can partition between the gas and particle phases, which influences their removal by different wet and dry deposition processes. Photodegradation is another removal pathway from air and for POPs reaction with OH radicals by oxidation (Anderson and Hites, 1996) is the most important pathway. Chemicals that have deposited to the surface of water or soil could re-volatilize into air, which is sometimes described as a secondary source to the atmosphere. Contaminants in the soil can be leached from soil into water and the process is rare (possibly flooding) and not usually included in models.

(14)

5 |

Surface waters are not pure water and comprise a spectrum of suspended particles at different concentrations, and POPs are distributed between these two phase with more hydrophobic POPs tending to partition to particulates. The particles are deposited to the underlying sediment and can also be transported back into water column via re-suspension. POPs also undergo bioaccumulation in biota and biomagnification through the food-chain. Environmental properties are key in determining the distribution and degradation of POPs after release (Paasivirta and Sinkkonen, 2009). Therefore, any climate variability has a potential to affect POPs behavior.

Figure 2.2. Illustration of the main environmental processes during LRAT of POPs.

2.3 POPs persistence and long-range transport potential

The behavior of POPs is complex and POPs mostly remain in the environment for fairly long periods of time. Therefore, it is particularly necessary to focus on hazard assessment, within which the overall persistence and long-range transport potential (LRTP) are extremely useful (Wania and Su, 2004; Wegmann et al., 2009).

Persistence for a chemical in multimedia models is often estimated as the environmental residence time. Models can estimate three types of residence time, namely: overall residence time, reaction residence time and advection residence time. Overall residence time is a measure of the time scale of total removal of the chemical from the model environment, while reaction and advection residence time present explicitly the reaction part and advection part of persistence, respectively. All these three parameters are useful in estimating how the chemical will be removed in the real environment (Mackay and Paterson, 1982; Mackay et al., 1983; MacLeod and McKone, 2004; Prevedouros et al., 2004). LRTP indicators that are often used as characteristic travel distance (CTD) and transfer efficiency (TE). CTD is a transport-oriented LRTP indicator, which is introduced to quantify the distance from the release point to the point where the concentration has dropped to 1/e or about 37% of the initial value (Beyer et al., 2000; Wegmann et al., 2009). Generally, only air and water are considered to possess CTD since both of these two media are mobile and could receive the emission. Soil and sediments, on the other hand, are not mobile media. In this study, since all the pollutants are assumed to be emitted into air, CTD in air is calculated from the emission-to-air scenario. By comparison, TE is a target-oriented LRTP indicator, which is introduced to quantify the deposition of chemicals transported

(15)

6 |

to the studied region. TE is defined as the mass flux into a selected target compartment divided by the emission mass flux (Scheringer, 2009; Wegmann et al., 2009). Equations to calculate these LRTP indicators will be further explained within the coming section of thesis.

2.4 POPs modeling

Several approaches are used to study the fate of POPs in the environment, with an important aim being to understand contaminant concentrations and fluxes among different environmental media under a variety of environmental scenarios (Wania and Mackay, 1993; McKone, 1996; Eisenberg et al., 1998; Cousins and Mackay, 2001; Cousins et al., 2002; Sweetman et al., 2002; Hauck et al., 2008; Armitage et al., 2009a; Armitage et al., 2009b). Measuring the chemical concentration is the most direct way to verify the contamination level and contaminant distribution. The most frequently measured compartments are air (gas-phase and particle-phase), water (dissolved-phase and particle-phase), soil, sediment and biota. Another useful approach is to mathematically simulate the pollutant concentrations and distributions in the environment, applying multimedia models. These simulations can theoretically reflect the relationship between pollutants properties and their environmental behavior such as their partitioning between different compartments.

2.4.1 Types of environmental POPs fate models

Generally, there are two types of environmental POPs fate models, namely: i) atmospheric chemistry transport models (CTMs), ii) multi-compartmental mass balance models (also termed as box models) (Hansen et al., 2008; Lamon et al., 2009a; Paasivirta and Sinkkonen, 2009). The former are generally characterized by having a high spatial resolution of the atmosphere and adjusted to account for the air-surface exchange processes, while the later usually treat the environment as a number of well-mixed compartments with homogeneous environmental characteristics (Mackay and Paterson, 1981; 1982; Mackay et al., 1983; Mackay et al., 1985; Mackay, 1991; Mackay and Paterson, 1991; Mackay et al., 1992b; Di Guardo et al., 1994; Maddalena et al., 1995; Mackay et al., 1996; Wania and Mackay, 1999). More types of box models are showed in Table 2.2.

Table 2.2. Types and descriptions of different POPs environmental fate models

Model Type Model Description

Lagrangian / Eulerian model Use high-resolution meteorological data, focus on surface modules to describe the two-way air-surface exchange processes of POPs

biological uptake model Describe the uptake and transfer of POPs in entire food chain, including vegetation, organism, biota and human being.

individual environmental process model

Mostly describe the partitioning, transport and transformation of POPs among environmental phases.

evaluative model Describe the behavior of POPs in a purely hypothetical environment, e.g. an environment consisting only air, water, soil and sediment.

POPs fate model Describe the actual environmental fate and behavior of POPs in real environment, regionally or globally.

2.4.2 Environmental fate model utilized in this present study

A primary purpose of this study is to forecast the concentrations and distributions of pollutants in the defined environment with respect to climatic variables and physical-chemical property uncertainties. Herein, a multi-compartmental environmental fugacity modeling method was applied. Before introducing the model, it is essential to highlight a concept – fugacity (unit: Pa), defined as the fleeing or escaping tendency for a specific pollutant from one compartment. Fugacity is linearly related to

(16)

7 |

concentration by a factor Z (unit: mol/m3∙Pa), which refers to fugacity capacity of the pollutant in the compartment (Mackay, 1991).

Equilibrium and Steady-State

When thermodynamic equilibrium occurs between two compartments (i.e. Gibb’s Free Energy is minimized), the net diffusive flux of chemical between the two compartments is zero. This equilibrium assumption ensures that only one mass balance equation is required to describe the distribution of chemical between all compartments in the model environment (Mackay et al., 1996; Cahill and Mackay, 2003; Mackay and Webster, 2006). Fugacity is a criterion of equilibrium, meaning the fugacity values for all considered compartments are the same at equilibrium. Equilibrium partitioning coefficients such as octanol-water, octanol-air and air-water partition coefficients control the distribution of POPs in the environment. It is important that these key partition coefficients are internally consistent with each other (Beyer et al., 2002). A simplified illustration for the inherent relationship between the key partition coefficients is shown in Figure 2.3, where K refers to partition coefficient and subscripts W, O and A refer to water, octanol and air, respectively.

Steady-state, or constancy with time, should not be confused with equilibrium as the two terms are conceptually distinct. To be at steady-state, equilibrium is unnecessary. A net transfer (flux) could exist for one chemical between two phases, but the total flux is constant with time (all time derivatives are zero) (Mackay and Paterson, 1981; Mackay, 1991; Mackay and Paterson, 1991). More explanation is given below.

Figure 2.3. Illustration of partition coefficients among air, water and octanol phases.

Ci and Cj stand for saturated concentrations in media I and j; Kij means the partition coefficient between

phase i and j.

Fugacity Model Level (I – IV) (Mackay, 1991)

i) Level I – equilibrium, steady-state, closed system

The models show the equilibrium distribution of a fixed quantity of conserved chemical, in a closed environment at steady-state, with no degrading reactions, no advective processes, and no intermedia transport processes. One fugacity and one mass balance equation applies.

ii) Level II – equilibrium, steady-state, open system

The models describe an open system, where a chemical is continuously discharged at a constant rate and achieves a steady-state and equilibrium condition at which the input and output rates are equal. Degrading reactions and advective processes are the loss or output processes treated. Intermedia

(17)

8 |

transport processes are not quantified. One fugacity and one mass balance equation applies.

iii) Level III – non-equilibrium, steady-state, open system

The models describe a situation which is one step more complex and realistic than the Level II models. Like the Level II model, chemical is continuously discharged at a constant rate and achieves a steady state condition in which input and output rates are equal. The loss processes are degrading reactions and advection. Unlike the Level II model, equilibrium between media is not assumed and, in general, each medium is at a different fugacity. A mass balance applies not only to the system as a whole, but to each compartment. Rates of intermedia transport are calculated using D values which contain information on mass transfer coefficients, areas, deposition and resuspension rates, diffusion rates, and soil runoff rates. It is hereby essential to define inputs to each medium separately, whereas in Level II only the total input rate was requested.

iv) Level IV – non-equilibrium, unsteady-state, open system

The models are dynamic or unsteady-state in nature. They are most often used to determine how long it will take for concentrations to change as a result of changing rates of emission.

To be specific, a Level III four-compartmental fugacity spreadsheet model was applied herein. Compartments are air, water, soil and sediment, respectively. Parameterization will be interpreted in the method section.

2.4.3 Challenges of POPs fate modeling

Even though the method for multi-compartmental environmental fate modeling of POPs is a useful tool, there still are several challenges modelers need to face (Hansen et al., 2008). Firstly the available physical-chemical properties and reaction rates of POPs, which are key model inputs, are limited in quantity and quality. Vapor pressures and solubilities in water are measured in the laboratory while partition coefficients can be either estimated from vapor pressure and solubilities or measured directly. Due to numerous uncertainties, including but not limited to method diversity, analytical limitation, random or systematic error, available data vary within a broad range and some of them may not be accurate and reliable (McKone, 1996; McKone et al., 1996; MacLeod et al., 2002; Gouin et al., 2004a; Hauck et al., 2008; Schenker et al., 2009). Physical-chemical properties and degradation rates are temperature-dependent and thus need to be corrected to the environmental temperature of relevance in a model scenario. All of these issues create problems in obtaining reliable sets of chemical properties to be used as model inputs. With the aim of creating a set of reliable and consistent physical-chemical input properties, adjustment techniques have been developed, by which the measured values or literature-derived values (LDVs) are corrected to final-adjusted values (FAVs) (Paasivirta et al., 1999; Beyer et al., 2002; Li et al., 2003; Schenker et al., 2005). Another challenge is that as release rates or emissions of POPs are difficult to obtain and usually are estimated, further increasing the uncertainty of simulation results. Breivik et al. (2007) utilized a mass balance approach, estimating the global historical emissions for selected PCBs congeners, but uncertainties exist in production amount, emission processes and emission factors (Breivik et al., 2002b; a; Breivik, 2004; Breivik et al., 2007). As one of the most studied group of POPs, PCBs were selected here. Since great efforts have been contributed to measure or estimate PCBs physical-chemical properties, it is believed that the uncertainty in chemical properties should be low compared to many other, less well studied POPs.

2.5 Polychlorinated biphenyls (PCBs)

2.5.1 PCB properties and applications

(18)

9 |

they comprise two benzene rings (biphenyl) with different numbers of chlorine atoms (from 1 to 10) substituted on the two rings. The basic chemical structure of PCBs is presented in Figure 2.3. Due to the change of chlorine atom numbers and substitution positions on the biphenyl ring, 209 PCBs congeners exist in theory, belonging to 10 homologues. Each homologue refers to all the PCBs with the same number of chlorine atoms. PCBs are highly lipophilic compounds, and their physical-chemical properties and biological activities generally vary among congeners (Mackay et al., 1992a). To be specific, with increasing degree of chlorination and molecular weight, PCB density, melting point and lipophilicity increase, and vapor pressure and water solubility decrease correspondingly.

Figure 2.4. Basic chemical structure of PCBs (n varies from 1 to 5).

PCBs are usually non-flammable and stable with high boiling point and electrical insulating properties, which make them extremely useful for industrially applications but also of great concern among POPs (Wania and Dugani, 2003; Breivik, 2004; Davis, 2004; Gouin et al., 2005; Lohmann et al., 2007; Hansen et al., 2008; Beyer and Biziuk, 2009; Lamon et al., 2009a; El-Shahawi et al., 2010). For either industrial or commercial purposes, PCBs are mostly and widely applied as insulating oil, plasticizer and numerous other products. PCBs were manufactured from 1929 until 1979 when they were banned, but they are still extensively distributed in landfills, building materials and the environment They are deemed extremely harmful to both humans and wildlife (AMAP, 2002). A concern is that these harmful compounds are still being released into the environment by improper storage and disposal, and may even have ongoing uses in certain parts of the world.

2.5.2 PCB environmental fate and degradation

There are no known natural sources of PCBs, implying that all PCBs found within the environment are the result of anthropogenic activities (Beyer and Biziuk, 2009). Once released into the environment at middle and lower latitudes, PCBs may even reach the arctic and sub-arctic via LRAT, waterways, and ocean currents (Mackay et al., 1992a; Beyer and Biziuk, 2009). It is complicated to thoroughly understand the transfer mechanisms and final sources or sinks for various PCBs, but several major processes significantly affect the environmental fate of them.

Sorption and Desorption to Organic Matter

Due to their high lipophilicity, PCBs readily sorb to organic matter in environmental media. Deposition of PCB-sorbed to organic rich particles in air and water is an important removal process from the atmosphere and water column for the heavier PCBs.

Volatilization

For the lighter lower chlorinated PCBs, air-water partitioning coefficients (KAW) are relatively high

compared to those more chlorinated congeners, thus resulting in volatilization from water surface and soil surface back into the air. The importance of volatilization on atmospheric concentrations of PCBs is well established. However, contaminants from global air masses can undergo deposition again into water and soil, resulting in the cycling at interfaces (Beyer et al., 2000; Beyer and Biziuk, 2009).

Bioaccumulation

(19)

10 |

contaminated animals (especially lipid rich food such as fish, seafood, meat and dairy products) can also accumulate these substances in tissues, resulting in adverse health effects. Consequently, PCBs affect not only individual organisms but also ultimately whole ecosystems and human populations (Mackay and Fraser, 2000; MacLeod et al., 2002; Czub et al., 2008).

Degradation

Degradation by abiotic and biotic processes is one of the main removal pathways for PCBs in the environment. Large uncertainties are associated with estimates of degradation rates. The estimated overall half-lives of PCB congeners can vary from weeks to years in air and often up to decades in biota. It is estimated that PCB congeners with low molecular weight have half-lives ranging between 6-21 years (Sinkkonen and Paasivirta, 2000; Gouin et al., 2004a).

In general, since there are broad ranges for physical-chemical properties and degradation properties of PCBs congeners according to degree of chlorination, congeners with different degrees of chlorinated are selected in this study (see method section).

(20)

11 |

3 Climate Change and POPs

3.1 Climate change

Climate change became a recognized international issue of concern in 1994, when 191 nations subscribed to the United Nations Framework Convention for Climate Change (UNFCCC), in order to discuss possible ways of resolving the global warming problem at the international level. The term ‘climate change’ was defined by Intergovernmental Panel on Climate Change (IPCC), which was established in 1988 together by the World Meteorological Organization (WMO) and United Nations Environment Programme (UNEP). It refers to a change in certain climate properties, either regionally or globally (Macdonald et al., 2003; Macdonald et al., 2005; Lamon et al., 2009a; Lamon et al., 2009b; Noyes et al., 2009). To a large extent, climate change is a human-induced process of global warming, mostly resulting from the emission of greenhouse gases (GHGs) such as carbon dioxide, nitrous oxide, methane and fluorocarbons. The feedback from climate change widely includes, but is not limited to, atmospheric temperature, oceanic salinity, precipitation pattern, wind pattern, atmospheric and oceanic currents and so forth (Ma et al., 2004; Macdonald et al., 2005). Phenomena resulting from climate change could be severe, such as the increase of sea levels, glacier ice melt and extreme weather events. Most of these changes have already been observed and are probably going to become even more severe because of the increasing concentration of GHGs (Lamon et al., 2009a; Noyes et al., 2009). Also, it is a complicated issue with important economic, health and safety, food production, security, and other implications. It is recognized as the largest environmental problems facing society today.

3.2 International Panel on Climate Change (IPCC) and Special Report on Emission

Scenarios (SRES)

Since there is much less confidence in estimates of how the climate will change corresponding to GHG concentrations at a regional scale, and no method yet exists of providing these confident predictions, an alternative approach is to specify a number of plausible future climates (New and Hulme, 2000; Giorgi et al., 2001; Hingray et al., 2007; Watterson, 2008). These are termed ‘climate scenarios’, along with those ‘economic scenarios’ and ‘environmental scenarios’, considered under different ‘emission scenarios’ (IPCC AR4, 2007).

The Special Report on Emissions Scenarios (SRES) is specifically on future emission scenarios, used for driving global circulation models to develop climate change scenarios. The SRES scenarios have been applied by IPCC for the Third Assessment Report (TAR) in 2001 and the Fourth Assessment Report (AR4) in 2007. It also replaced the IS92 scenarios used for the IPCC Second Assessment Report of 1995.

There are four storylines (labeled A1, A2, B1 and B2) in SRES, describing four scenario families which are further divided into six scenario groups. Each storyline represents different demographic, social, economic, technological, and environmental developments that diverge in increasingly irreversible ways whereas each scenario is a particular quantification of one of the four storylines. Figure 3.1 presents the structure of SRES, where ‘HS’ is marked for globally harmonized scenarios while all others of the same family are marked as ‘OS’. One of the harmonized scenarios is designated as the characteristic representative marker scenario. Table 3.1 lists and describes these six scenario groups.

(21)

12 | Figure 3.1. Illustration for structure of the storylines and scenarios in the IPCC Special Report on

Emissions Scenarios (SRES).

HS is marked for globally harmonized scenarios while all others of the same family are marked as OS.

Table 3.1. List and assumption descriptions for the defined SRES scenarios in IPCC AR4.

Scenario Key Assumption

SRES A1 Emissions A future world of very rapid economic growth, low population growth and rapid introduction of new and more efficient technology. Major underlying themes are economic and cultural convergence and capacity building, with a substantial reduction in regional differences in per capita income. In this world, people pursue personal wealth rather than environmental quality.

SRES B1 Emissions A convergent world with the same global population as in the A1 storyline but with rapid changes in economic structures toward a service and information economy, with reductions in materials intensity, and the introduction of clean and resource-efficient technologies.

SRES A2 Emissions A very heterogeneous world. The underlying theme is that of strengthening regional cultural identities, with an emphasis on family values and local traditions, high population growth, and less concern for rapid economic development.

SRES B2 Emissions A world in which the emphasis is on local solutions to economic, social, and environmental sustainability. It is again a heterogeneous world with less rapid, and more diverse technological change but a strong emphasis on community initiative and social innovation to find local, rather than global solutions.

3.3 Impact of climate change on POPs

Any temporal and spacial alteration of climate has the potential of affecting the environmental behavior of POPs. Alteration in temperature, precipitation, wind patterns and so forth might be effective in

(22)

13 |

redistributing POPs among environmental compartments. Their degradation rates are also likely to be altered. So far, a few studies have been carried out to investigate the interactions between climate change and fate modeling of POPs. Mostly, assessment and estimation are focused on specific locations, especially the Arctic since it is especially sensitive to both climate change and POP contamination (Scheringer et al., 2000; Ma et al., 2004; Dalla Valle et al., 2007; Hauck et al., 2008; Lamon et al., 2009a; Wang et al., 2010).

Table 3.2. Possible impact of climate change on POPs behavior (Lamon et al., 2009a).

Climate Change Effect on POPs Behavior

Temperature Emissions, degradation rate Precipitation (Rain Fall) Atmospheric wet deposition flux

Wind Speed Atmospheric distribution

Salinity Adsorption on suspended particulate matter (SPM)

Oceanic Current Marine spatial

Snow Rate Atmospheric deposition

Dry Period Volatilization

Ice Cover Release from ice

Organic Carbon (OC) Content Adsorption / release

Sediment Deposition Deposition rate (from water to sediment) Sediment Resuspension Re-circulation

Population Dynamics Adsorption and absorption Vegetation Mass Adsorption on vegetation biomass

3.4 Implications of climate change on fate modeling of POPs

Complex general circulation models (GCMs) have been extensively applied to project the impact of regional and global climate change on environmental systems (Mackay et al., 1992b; Mackay et al., 1996; Scheringer, 2009). Most of the IPCC projected scenarios are performed through GCMs. Although GCMs have the advantage of being able to predict the fate of POPs under climate change conditions, and have high temporal and spatial resolution, they are mathematically complicated and computationally expensive (MacLeod et al., 2005; Watterson, 2008; Lamon et al., 2009a; Lamon et al., 2009b). There have also been simplified modeling studies that attempt to predict the influence of climate change on the fate of POPs (McKone et al., 1996; Macdonald et al., 2005; Dalla Valle et al., 2007; Lamon et al., 2009b; Ma and Cao, 2010). Table 3.3 lists the implication of selected POPs modeling in climate change perspective.

Table 3.3. Previous modeling studies that have estimated the effect of climate change of POP transport

and fate.

POPs Modeling Method Reference

HCB Fugacity model McKone et al., 1996

PCB Berkeley – Trent global multimedia mass balance model (BETR Blobal)

MacLeod et al., 2005

PCB; PCDD/Fs Level IV fugacity model Dalla Valle et al., 2007 PCB Berkeley – Trent global multimedia mass balance model

(BETR Blobal)

Lamon et al., 2009

(23)

14 |

4 Uncertainty Analysis in Fate Modeling

4.1 Dealing with uncertainty in multimedia modeling

When simulating and predicting concentrations and distributions of chemical using multimedia models , over- or under-estimations always exist due to various uncertainties in input parameters (Hertwich et al., 2000; Hauck et al., 2008). Uncertainty has numerous sources. In Hertwich et al. (2000), the sources are distinguished as following: random error, statistical variation, systematic error, subjective judgment, linguistic imprecision, variability, true randomness and disagreements between experts. The uncertainty is also classified into four types: decision rule uncertainty, model uncertainty, parameter uncertainty and parameter variability. A brief description is presented in Table 4.1.

Table 4.1. Summary of uncertainty existing in multi-media modeling.

Uncertainty Type Description

Decision rule uncertainty When there is ambiguity or controversy about how to quantify or compare the objectives and model formulations.

Model uncertainty Due to the unavoidable simplification of reality and mostly depends on which model and which approach is applied.

Parameter uncertainty Refers to input parameter values that are not fully acknowledged with precision due to limited observations or estimation error.

Parameter variability Refers to statistical variance that derives from random or heterogeneous factors.

4.2 Parameter uncertainty and parameter variability

As stated in Table 4.1, parameter uncertainty and parameter variability can both be brought and affected by input parameters, with the former one mainly focusing on chemical-specific data while the later focusing on landscape data. The input parameter uncertainty and/or variability are addressed in the present study. We distinguish the parameter uncertainty to physical-chemical property uncertainty and degradation property uncertainty, and define parameter variability as the climate variability. Monte-Carlo simulations are widely used in uncertainty analysis, which is a means of converting the input parameter uncertainty into probability distributions (McKone et al., 1996; Hertwich et al., 2000; MacLeod et al., 2002). To date, only a few studies have aimed at exploring the effect of parametric uncertainty or variability on model predictions in chemical fate modeling (McKone et al., 1996; MacLeod et al., 2002; Barber et al., 2004; Hauck et al., 2008) . We are aware of only one study that considered uncertainty in modeling chemical fate, exposure and risk under climate change scenarios and this study only considered one chemical and a limited set of climate variables (McKone et al., 1996).

(24)

15 |

5 Methods

5.1 Fugacity model and parameterization for PCBs

The present study uses a ChemCAN spreadsheet model, which is a regional contaminant Level III mass-balance fate model based on fugacity concept (MacLeod et al., 2002; Webster et al., 2004). The model is parameterized for several selected PCBs congeners for the Japan region. It consists of a series of input parameters including both the landscape properties for the region and physical-chemical properties for the chemical, as well as the discharge rates. Outputs are usually predicted concentrations at steady-state and specific fluxes between compartments calculated by a system of mass-balance equations.

5.1.1 Environmental properties

The study area is showed in Figure 5.1. In the fugacity model used here, four major bulk compartments are defined, which are air, water, soil and sediment. They are subscripted as air – 1, water – 2, soil – 3, and sediment – 4. Subcompartments are further defined. To be specific, bulk air consists pure air (1,1) and air particle (1,3); bulk water consists pure water (2,2), water particle or suspended solid (2,3) and water biota (2,4); bulk soil consists soil air (3,1), soil water (3,2) and soil solid (3,3); bulk sediment consists pore water (4,2) and sediment solid (4,3). The compartmental properties in the Japan ChemCAN spreadsheet model are from Kawamoto et al. (2001) and presented in Table 5.1 as below (Kawamoto et al., 2001) .

Figure 5.1. Map of study area.

Table 5.1. Properties of the four compartments in the Japan regional model.

Bulk Comparment

(i)

Subcompartment (j) vol fraction Φij Fraction

OC in solids Bulk density (kg/m3) Bulk volume (m3) Bulk area (m2) Bulk depth (m) Air (1) Water (2) Solid (3) Biota (4) Air (1) 1.0 - 2×10-11 - - 1.227 3.8×1014 3.8×1011 1000 Water (2) - 1.0 1×10-5 1×10-6 0.1 1000 9.3×109 3.1×109 3 Soil (3) 0.2 0.2 0.6 - 0.02 1640 3.8×1010 3.8×1011 0.1 Sediment (4) - 0.8 0.2 - 0.05 1280 9.3×107 3.1×109 0.03

5.1.2 Fugacity (f), Fugacity Capacity (Z) and Transfer Coefficient (D)

As stated previously, fugacity f (Pa), which describing the ‘escaping’ or ‘fleeing’ tendency of a chemical, is proportional to concentration C (mol/m3) through the fugacity capacity, Z (mol/m3∙Pa). In a compartment i this can be expressed as below,

(25)

16 |

C𝑖 = 𝑓𝑖 × Z𝑖 (1)

Fugacity can also be considered as the partial pressure of a chemical in a phase. When at equilibrium condition, f values for a chemical in all the phases are equal. However, a Level III model is used here implying non-equilibrium and steady-state conditions. Individual f values are calculated for each compartment, and the difference between two f values refers to the net diffusive flux for the chemical between two phases.

Additionally, for each subcompartment, a Z value is expressed, all of which combine with the volume fractions to give an overall Z value for the bulk compartment. Table 5.2 summarizes all the Z values and their definitions in compartments of the model.

Table 5.2. Summary of Z value definitions in compartments and subcompartments.

Compartment Subcompartment Z (mol/m3∙Pa) Definitions

Air (1) Pure air (1,1) Z11 = 1/RT R = 8.314 Pa m3/mol K T = absolute temperature (K) H = Henry’s law constant (Pa m3/mol)

CS = aqueous solubility (mol/m3)

PS = vapor pressure (Pa) xij = fraction organic carbon KOC = organic carbon partition coefficient = 0.41 KOW

ρij = density of solids (kg/L)

𝑃𝐿𝑆= liquid vapor pressure (Pa)

KOW = octanol / water partition coefficient

Φij = volume fraction

Air particle (1,3) Z13 = 6×106/(𝑃𝐿𝑆RT)

Bulk air (1) Z1 = Z11 + Φ13Z13 Water (2) Pure water (2,2) Z22 = 1/H or C

S /PS Water particle (2,3) Z23 = x23 ρ23 KOC /H Water biota (2,4) Z24 = 0.048ρ24KOW/H Bulk water (2) Z2 = Z22 + Φ23Z23 + Φ24Z24 Soil (3) Soil air (3,1) Z31 = Z11

Soil water (3,2) Z32 = Z22

Soil solid (3,3) Z33 = x33 ρ33 KOC /H

Bulk soil (3) Z3 = Φ31Z31 + Φ32Z32 + Φ33Z33 Sediment (4) Sediment water (4,2) Z42 = Z22

Sediment solid (4,3) Z43 = x43 ρ43 KOC /H Bulk sediment (4) Z4 = Φ42Z42 + Φ43Z43

Emissions Advection Degradation Intermedia Transfer Figure 5.2. Illustration of a four-compartmental environment fugacity Level III model for assessing the

(26)

17 |

The steady-state equations describe gains and losses in each of the four compartments, and are used to solve the mass balance. Table 5.3 lists all the possible input and loss processes of a contaminant by transport and transformation that are addressed in each environmental compartment within this model. In the four-compartmental environment used here, as presented in Figure 5.1, E refers to the emissions of the contaminant into air compartment and is assumed to be the only emission herein for PCBs. B is the background inflow concentration for the contaminant, which is modified with extra equations, shown in Appendix A. As for other intermedia transfer processes and removal processes, specific D parameters are listed in Table 5.4, combined with subscriptions in Figure 5.1.

Table 5.3. Summary of gain and loss processes of contaminants among the four-media environment.

Compartment Gains Losses

Air Emissions from contaminant sources, Advection inflow, physical transformation, Diffusion from surface soil,

Diffusion from surface water,

Resuspension of deposited soil particles.

Advection outflow, physical transformation, Degradation, chemical transformation, Diffusion to surface soil,

Diffusion to surface water,

Deposition to soil (rain dissolution), Deposition to soil (aerosol deposition), Deposition to water (rain dissolution), Deposition to water (aerosol deposition). Water Emissions from contaminant sources,

Diffusion from air,

Washout by rainfall (rain dissolution), Deposition of aerosols,

Soil solution runoff, Erosion (solid runoff).

Advection outflow, physical transformation, Degradation, chemical transformation, Diffusion to air,

Diffusion to sediment, Sediment deposition.

Soil Emissions from contaminant sources, Diffusion from air,

Washout by rainfall (rain dissolution), Deposition of aerosols.

Soil leaching, physical transformation, Degradation, chemical transformation, Diffusion to air,

Resuspension of soil particles, Soil solution runoff,

Erosion (solid runoff). Sediment Deposition from water,

Diffusion from water.

Sediment burial, physical transformation, Degradation, chemical transformation, Diffusion to water,

(27)

18 | Table 5.4. Summary of transfer D parameters within the four-compartmental environment.

Compartments Process and D values Total D values

Air (1) – Water (2) Diffusion: DAW1 and DWA Wet deposition: DAW2 Dry deposition: DAW3

D12 = DAW1 + DAW2 + DAW3 D21 = DWA

Air (1) – Soil (3) Diffusion: DAS1 and DSA Wet deposition: DAS2 Dry deposition: DAS3

D13 = DAS1 + DAS2 + DAS3 D31 = DSA

Soil (3) – Water (2) Solid runoff: DSW1 Water runoff: DSW2

D32 = DSW1 + DSW2 D23 = 0

Sediment (4) – Water (2) Diffusion: DSeW1 and DWSe1 Deposition: DWSe2

Resuspension: DSeW2

D42 = DSeW1 + DSeW2 D24 = DWSe1 + DWSe2

Reaction (Degradation) DA-deg, DW-deg, DS-deg, DSe-deg N/A Advection DA-adv, DW-adv, DS-adv, DSe-adv N/A

5.1.3 Model modifications

Based on the original ChemCAN Level III spreadsheet model applied in the Japan region, some modifications are made. First of all, the influence of wind velocity on both the mass transfer coefficient (MTC) and residence time of air are taken into account, while in the original model, the MTC was a fixed value. Secondly, the background inflow concentration for PCBs, which is showed as B in Figure 5.1, is recalculated by a series of equations. Furthermore, persistence, including overall residence time, advective and reactive residence time is added into the model outputs, as well as two LRTP indicators CTD and TE, both of which had been introduced previously. Calculations are all listed in the Appendix A.

5.2 Variability and uncertainty of model inputs

There are two classes of inputs to the regional four-compartmental fate model – the landscape properties and physical-chemical properties. To compare the significance of climate variation and physical-chemical property uncertainty to PCB fate, each of the studied inputs is described in terms of a mean value and corresponding statistical distribution.

5.2.1 Climatic variables and climate scenario Climatic variables

To evaluate the influence of a climate change scenario on the environmental fate of PCBs, several variables are considered here. They are temperature, precipitation (rain fall) and wind speed Temperature is one of the climatic factors that is certain to affect POP fate through alteration of partitioning, and precipitation and wind speed could both affect mass transfer coefficients. The ultimate climate variable values involve observational means and departures from the means which denote to the climate perturbation (McKone et al., 1996; Ma and Cao, 2010). The departure values are obtained through projected climate scenarios. Observational and projected data are selected from IPCC Data Distribution Centre (IPCC-DDC). For instance, as one landscape property input parameter, temperature is presented as,

T (°C) = T̅ + ∆T (5)

Within equation (5), T is the air temperature, T̅ is the observational mean air temperature which herein is considered from 1901 to 2000, while ∆T is the projected mean of air temperature change which is

Relative influences of uncertainty in physical-chemical property data and variability in climate parameters in determining the fate of PCBs