CHAPITRE 2 SIMULATION OF THE MIGRATION OF DISSOLVED CONTAMINANTS THROUGH A SUBAQUEOUS CAPPING LAYER: MODEL DEVELOPEMENT AND APPLICATION FOR ARSENIC MIGRATION

Table des matières

TRANSCAP-1D est un modèle numérique pour la simulation de la migration verticale de contaminants dissous dans une colonne de sédiments. Le modèle a été développé afin d'évaluer l'efficacité à long terme d'une couche de recouvrement sous-marine. Il considère l'advection, la diffusion et l'effet de la bio-irrigation. Le modèle représente un milieu à double porosité, composé de pores et de trous ou tubes de vers. Le modèle numérique a été calé avec les profils de concentration de l'arsenic dissous qui ont été mesurés à deux stations du fjord du Saguenay, au Québec, Canada, après la déposition d'une couche de recouvrement naturelle suite au déluge de 1996. Dans le site à l'étude, l'arsenic n'est pas d'origine anthropique et ne peut donc pas être considéré un contaminant, cependant il a été utilisé pour le calage en raison de la disponibilité des données. Afin d'illustrer l'effet de l'incertitude et de la variabilité des valeurs d'entrée sur la réponse du modèle, nous avons réalisé des simulations numériques avec des valeurs variables de profondeurs de bio-irrigation et d'épaisseurs de la couche. Les simulations indiquent que la profondeur de bio-irrigation est un facteur important pour la distribution de contaminant dissous dans la colonne de sédiments. En outre, pour les contaminants qui coprécipitent avec des monosulfures de fer en milieu anoxique, tel l'As, la réaction de dissolution engendrée par l'oxydation des sulfures, liée à la bio-irrigation, est un facteur à considérer.

TRANSCAP-1D is a numerical model that simulates the vertical migration of dissolved contaminants in a sediment column. The model was developed to evaluate the long-term effectiveness of a subaqueous capping layer. It considers advection, diffusion and the effect of bio-irrigation. The sediments are represented as a dual porosity medium composed of sediment pores and biologically formed tubes. The numerical model was calibrated with the concentration profiles of dissolved arsenic measured in the sediments at two sampling stations in the Saguenay Fjord, in Québec, Canada, where a major flood event caused the natural capping of contaminated sediments. At the studied site arsenic is not of anthropogenic origin and cannot be considered a contaminant. Nevertheless, it has been used for calibration because of the availability of data. Several numerical simulations with variable bio-irrigation depth and variable thickness of the capping layer were performed to evaluate the effect of the uncertainty and variability of the input values on the model response. These simulations indicate that the depth of bio-irrigation is a major factor controlling the distribution of dissolved contaminants in the sediment column. Moreover, for contaminants that coprecipitate with iron monosulfide under anoxic condition, like As, the release from mineral dissolution caused by the exposure to an oxic environnement during bio-irrigation, is a factor to be considered.

Subaqueous capping consists in covering contaminated sediments to isolate them from the aquatic ecosystem, stabilize the sediments and eliminate, or at least reduce, the contaminant flux towards the water-sediment interface. This method represents an alternative for the containment of contaminated sediments, as opposed to their removal, and can be used for in situ remediation as well as for the disposal of dredged contaminated sediments. Any attempt at predicting the long-term effectiveness of a capping layer requires a sound understanding of the physical, chemical and biological factors affecting the migration of contaminants in the sediments.

Boudreau (1999) reviews diagenetic models representing the fate of metals in underwater sediments. The processes that control the migration of metals in these environments include molecular or ionic diffusion, advection caused by the compaction of the sediment particles and chemical reactions. Additionally, the benthic fauna increases the migration through bio-diffusive or non-local mixing and bio-irrigation. The bio-diffusive or non-local mixing is the random displacement of sediments and water particles caused by burrowing and ingestion/excretion of the benthic organisms. Bio-irrigation, on the other hand, results from the same organisms pumping the overlying water into their tubes and burrows to provide oxygen for their respiration and prevent the accumulation of potentially noxious chemical compounds (Jorgensen and Revsbech 1985).

Once a cap is placed over contaminated sediments, two different consolidation processes can occur in the sediments. The underlying contaminated sediments can consolidate because of increased stress caused from the weight of the cap and there can also be consolidation of the clean sediments under their own weight (Mohan et al. 1999, Mohan et al. 2000). Consolidation of the underlying sediments can release contaminated water into the upper clean sediments, whereas the consolidation of the capping layer might induce advection in this same layer. Consolidation is a process of limited duration, thus associated advection may last a few weeks to several months depending on the sediment properties (Zeman 1994). Thereafter, transport in the sediments will be dominated by diffusion.

Benthic fauna affects the redistribution of the sediment particles as well as the solute transport by burrowing, tube building, ingestion/excretion of sediments and bio-irrigation. The greatest number of organisms is found in the oxygenated zone above the redox boundary, in the top 0.02-0.05 m of the sediment column (Archer and Devol 1992, Forster and Graf 1995). However, some polychaete and oligochaete worms have been observed to penetrate sediments as deep as 0.15 m (Aller and Yingst 1978, Wang and Matisoff 1997, McCaffrey et al. 1980). Tube building worms increase the area of contact between the sediments and water, enhancing diffusive solute exchange across the tube-sediment interface. They pump overlying water into their burrows, modifying the redox and pH conditions, the microbial activity and the fluxes and reactions within and adjacent to the bio-irrigated zone (Marinelli and Boudreau 1996, Aller and Aller 1998). The effect of bio-irrigation on the oxygen concentration in the sediments is limited to the vicinity of the burrow walls (Furukawa et al. 2000, Meyers et al. 1987). Nevertheless several studies have documented an increased solute flux for both organic (Reible et al. 1996) and inorganic (Riedel et al. 1987) compounds in bioturbated sediments compared to sediments without bioturbation.

Under anoxic conditions, some heavy metals are removed from the porewater by coprecipitation or adsorption on iron sulfides and by formation of discrete solid sulfides. The exposure of the anoxic sediments to an oxic environment, which may occur during resuspension, dredging or migration of the redoxcline, may oxidize the iron sulfides resulting in the dissolution and release of the trace metals associated with these phases (Petersen et al. 1997, Morse 1994). Laboratory experiments have shown that metastable Fe monosulfides (FeS) are an important reservoir of reactive trace metals (Huerta-Diaz et al. 1998, O’Day et al. 2000, Simpson et al. 2000). As bio-irrigation modifies the redox conditions near the tubes, it may also induce the oxidation of sulfides and the release of trace metals (Emerson et al. 1984, Aller and Yingst 1978). Under fully oxidizing conditions, there may be precipitation of iron oxides along the burrow walls and the released metals may be adsorbed. But, as shown by Furukawa et al. (2001), since the burrow walls are not irrigated continuously but rather periodically, the burrow wall interface is subjected to oscillating geochemical parameters, including dissolved O2 concentration and pH. This geochemical oscillation may affect the stability of mineral phases, especially redox sensitive metals.

Several numerical models that simulate early diagenetic processes occurring in recent sediments, including remineralization of organic matter and the cycling of several elements, have recently been presented (Soetaert et al. 1996, Wang and Van Cappellen 1996, Park and Jaffé 1996). These models incorporate complex biogeochemical reactions and thus require a series of biogeochemical input parameters that can be either obtained by calibration, from the literature or through stoichiometry. Although these models can simulate complex biogeochemical systems and are very valuable for diagenetic studies, they were not designed for specific application to a capping layer and they would need to be adapted to represent the migration of contaminants through a cap. In comparison, the numerical model RECOVERY, developed by the U.S. Army Corps of Engineers (Boyer et al. 1994, Ruiz et al. 2000) was designed for simulating mass transport through capping layers. The model simulates the fluxes of organic contaminants in dissolved and particulate form and represents bioturbation by using a mixing factor but does not consider the effect of bio-irrigation.

Several studies have documented the effect of dwelling organisms on contaminant fluxes from and towards sediments (Petersen et al. 1998, Reible et al. 1996, Riedel et al. 1987, Rivera-Duarte and Flegal 1994). Three different conceptual models have been proposed to describe the effect of bioturbation on solute transport: the enhanced diffusion model (Guinasso and Schink 1975), the cylindrical diffusion model (Aller 1980) and the non-local exchange model (Emerson et al. 1984). The enhanced diffusion model uses an increased effective diffusion coefficient in the surface mixed layer to represent solute exchange. In the cylindrical diffusion model, the geometry of the irrigated burrows is simplified and represented with vertical cylinders (Figure 2.1). The non-local exchange model represents the solute transfer occurring between the sediments and the water by an exchange coefficient called "non-local". Matisoff and Wang (1998) compared these three models with experimental observations of bio-irrigated sediments and showed that the best fit to the observed concentrations is obtained with the cylindrical diffusion model and the non-local exchange model. As shown by Boudreau (1984), the cylindrical model can be reduced to the non-local exchange model when a linear gradient is assumed between the concentration inside the burrow and the horizontally averaged concentration in the sediment porewater.

Figure 2.1: Schematic diagram illustrating the simplified geometry of a tube/burrow and the bio-irrigated layer, approximated by packed cylinders, after Aller (1980). (r1 = inner radius of cylinders, r2 = half distance between two cylinders, δ = distance to the point where the concentration in the sediments equals the horizontally integrated value)

The objective of the present study is to develop a numerical model, TRANSCAP-1D, that will simulate the migration of certain dissolved contaminants, through a sediment cap in a marine environment. The model considers the main physical, chemical and biological factors affecting the migration of dissolved contaminants in sediments, including advection, diffusion, bio-irrigation and dissolution. The model was primarily designed to study the migration of contaminants in the sediments of the Saguenay Fjord, where a major flood caused the natural capping of contaminated sediments. Some simplifying assumptions concerning the governing processes, discussed in the paper, have been used for the application to the Saguenay Fjord. However, the model formulation is general and allows its application to other sites.

The description of dissolved solute transport in sediments containing bio-irrigated tubes, assuming non-local solute exchange between the sediment porewater and the tubes, requires governing equations for both the sediments and the tubes. The following assumptions are made: advection and diffusion are accounted for in both the sediments and the tubes, the effect of temperature and density gradients on solute transport are assumed negligible and we also neglect transient effects of compaction on solute transport. This last assumption is discussed later for the application to the Saguenay Fjord. Because transport processes and concentration gradients in contaminated sediments are predominantly vertical, a one-dimensional representation is used here. However, the model is easily expandable to multi-dimensional systems. Finally, sedimentation and accumulation of contaminants by organisms are not included in this study, but could be incorporated later. The main transport processes are illustrated in Figure 2.2.

The equation describing solute transport in the sediment porewater is given by:

(2.1)

where CS and CT are the solute concentration [M L-3] in the sediments and the tubes, respectively, ns is the sediment porosity [-], vs is the fluid velocity in the sediments [L T-1], Rt is the solute retardation factor [-], and β is a first-order mass transfer coefficient [T-1], corresponding to the non-local coefficient discussed in the previous section. The first term on the right-hand side of equation (2.1) represents diffusion/dispersion in the sediment, the second term represents the advective flux and the last term describes the mass transfer between sediments and tubes. The dispersion coefficient of the solute in the sediments, Ds [L2 T-1], is given by:

(2.2)

where Dd is the effective diffusion coefficient [L2 T-1] and αL is the sediment dispersivity [L]. Note that when the fluid velocity is small in the sediments, which is the case for low-permeability material, the dispersion coefficient is approximately equal to the diffusion coefficient.

Figure 2.2 : Diagram representing the sediment column, the main transport processes in the dual porosity system (sediment pores and tubes), the initial concentration and the variation of tube porosity with depth.

The equation describing solute transport in the tubes is given by:

(2.3)

where nT is the porosity of the tubes [-], Dm is the dispersion coefficient for the tubes [L2 T-1], given by an expression similar to equation (2.2), and vT is the fluid or irrigation velocity in the tubes [L T-1]. The porosity of the tubes represents the volume occupied by the tubes per unit volume of sediments. In equation (2.3), S is a general source or sink term [M L-3 T-1] representing the release of contaminant from mineral dissolution associated to bio-irrigation. This term is defined in the application section.

The number of benthic organisms that bio-irrigate their tubes is maximal at the surface of the sediments and decreases exponentially downwards (Martin and Banta 1992). To represent this exponential decrease, the mass transfer coefficient is expressed as:

(2.4)

where β1 [T-1] and β2 [L-1] are empirical coefficients that are adjusted to reproduce the variation in the number of tubes and zmax [L] is the location of the top of the sediment column. The porosity of the tubes, nT , is also modified in a similar manner:

(2.5)

where nT° is the maximum sediment porosity, located at the top of the sediment column (see Figure 2.2).

Initial and boundary conditions are required to solve both equations (2.1) and (2.3). To impose the initial conditions one specifies the initial solute concentration in the two domains, sediments and tubes. Boundary conditions are required at the top and bottom of the domain and can be either a prescribed concentration or prescribed mass flux for both domains.

The two governing equations (2.1) and (2.3) are discretized with the finite volume method (Versteeg and Malalasekera 1995). The equations are coupled through the mass exchange term and a simultaneous solution for the concentration in the sediments and tube is obtained for each finite volume. The system of equations is thus solved in a fully-coupled fashion, avoiding the use of iteration that is necessary when the equations are decoupled. The block tridiagonal matrix resulting from the discretization and assembly of the terms is solved with the Thomas algorithm adapted to block matrices. Details of the discretization are given in Appendix 2.

The governing equations (1.1) and (1.3) solved by the TRANSCAP-1D model are mathematically similar to those describing solute transport in a dual-porosity medium in the context of groundwater flow (for example, Zheng and Bennett 2002). The numerical model was verified by comparing the simulation results to the results of the semi-analytical solution of Neville et al. (2000), which has been developed for one-dimensional solute transport in a dual-porosity medium with multiple non-equilibrium processes. The semi-analytical solution solves the equations (1.1) and (1.3) for a uniformly contaminated domain and simple boundary conditions. The tubes and the sediment porosity of the TRANSCAP-1D numerical model correspond to the mobile and the immobile regions of the semi-analytical solution.

The TRANSCAP-1D model was verified against the results of the semi-analytical solution for a specific set of input values and boundary condition. The input parameters used in the semi-analytical solution are presented in Table 2.1. The system is represented by a 0.60-m long column, which was uniformly divided in 60 elements of equal volume. A constant concentration of 1.0 was specified at the basis of the column and the simulation time was set to 2.5 d. In the TRANSCAP-1D model the tube and sediment porosity were both set to 0.25, which gives a total porosity of 0.5 and respects the mobile fraction of 0.5 stated in the input parameters of the semi-analytical solution. The advection velocity was set to 0.1 m d-1, which corresponds to the Darcy velocity of 2.5 cm d-1 for a porosity of 0.25. The dispersivity of the numerical model was set equal to 0.01 m, which is equivalent to a dispersion coefficient of 10.0 cm2 d-1 multiplied by the advection velocity. Results of the simulation, shown in Figure 2.3, indicate that the numerical model reproduces almost perfectly the concentrations computed with the semi-analytical solution for the input and boundary conditions considered here.

Another test of the numerical model is the calculation of the solute mass balance for the simulations. The verification tests done here produced a relative mass balance error equal to approximately 10-12 of the calculated solute fluxes, which indicates almost perfect conservation of mass for the numerical scheme.

Figure 2.3 : Verification of the model results with the semi-analytical solution of Neville et al. (2000) .

The main motivation for developing the model was its application to the case of the Saguenay Fjord, where a layer of clean sediments was deposited over contaminated sediments during a major flood in 1996. The Saguenay Fjord is 90 km long, with a width varying between 1 and 6 km, and reaches the St. Lawrence estuary in Tadoussac (Figure 2.4). The water column in the fjord is composed of two layers: a thin freshwater layer near the surface overlying a thick layer of saline, well oxygenated water, coming from the estuary. The limit between the two layers is located at a depth of 30 m. From this limit downwards, the salinity is equal to 31 ‰ and the temperature is 1 °C. Two study sites, Stations 1 and 2, are located at the upstream section of the fjord, in the Bras Nord and the Baie des Ha! Ha! (Figure 2.4). The average water depth at the stations is about 150 m.

Several studies have documented the distribution of contaminants in the water (Gobeil and Cossa 1984, Tremblay and Gobeil 1990) and in the sediments (Barbeau and al. 1981, Pelletier and Canuel 1988, Gagnon et al. 1993, Cossa 1990) of the Saguenay Fjord. The pollution originates from the development of the industrial activity in the Saguenay region at the beginning of the 1930s. The introduction of environmental regulations in the 1970s was effective in reducing the industrial discharge of pollutants. Nevertheless, following studies on sediments of the Saguenay Fjord continued to show significant concentrations of heavy metals (Hg, Pb, Zn, Cu) and PAHs (polycyclic aromatic hydrocarbons) (Cossa 1990, Gagnon et al. 1993), related to anthropogenic sources located in the region. On the opposite, the presence of arsenic in the sediments could not be linked to the industrial activity in the Saguenay area. Studies on the distribution of arsenic in the water column suggest that arsenic most likely is introduced to the fjord from the St Lawrence Estuary and accumulates in the sediments by particles settling through the marine waters (Mucci et al. 2000b).

Figure 2.4: Map of the Saguenay Fjord showing the two stations that were used to test the model.

In July 1996, two days of intense rainfall caused severe flooding in the Saguenay region and lead to the discharge by rivers of several million cubic meters of clean sediments to the Saguenay Fjord. A turbidite deposit composed mainly of silt and clay and of an average thickness of 0.2 m settled in a few days on the contaminated bottom sediments of the Bras Nord and the Baie des Ha! Ha! (Maurice 2000). The thickness of the turbidite layer varies spatially and reaches a maximum of a few meters at the upstream section of the fjord, near the incoming rivers’ mouths. The clean sediments represent a natural barrier, or capping layer, that might isolate the fjord water from the contaminants present in the underlying sediments.

To study the impact of the catastrophic flood event on the fate of the contaminants present in the sediments, a major multidisciplinary project was initiated in 1997. Various research groups collected information on the biological, geochemical and geotechnical characteristics of the sediments, which provided us with valuable input parameters for the numerical simulations (Tremblay et al. 2003, Pelletier et al. 2003). The model was adapted to the simulation of dissolved arsenic, because its chemistry has been extensively studied (Aggett and O’Brien 1985, Belzile and Tessier 1990), especially for the sediments of the Saguenay Fjord (Saulnier and Mucci 2000, Mucci et al. 2000a, Mucci et al. 2000b). Arsenic is not a contaminant in the Saguenay Fjord, but the calibration of the model required profiles of the dissolved concentration in the sediment column, and those data were only available for As. Similar profiles of concentration versus depth were documented for arsenic (As) and iron (Fe), suggesting that in the sedimentary environment the behavior of As is related to that of Fe. The two main mechanisms controlling the distribution of As in the sediment column are the coprecipitation of As with Fe sulfides in the anoxic sediments and the adsorption onto oxihydroxides in the first few millimeters below the sediment-water interface, corresponding to the oxic layer.

In the Saguenay sediments the pyrite concentration is relatively low (5-30 μmol g-1, dry weight) whereas the Fe monosulfide concentration is abnormally high and up to seven times more abundant (Mucci and Edenborn 1992). Compared to pyrite, which has slow oxidation kinetics in presence of O2, Fe monosulfide is very reactive and dissolves rapidly if exposed to an oxic environment. Thus, it is likely that As adsorbed or coprecipitated with the Fe monosulfide in the anoxic layer can be remobilized after exposure to O2 during bio-irrigation (Figure 2.5). Other trace metals coprecipitate with Fe sulfides (Ni, Cu, Co, Hg, Pb), but compared to As their probability of remobilization is smaller since they prefer the association to less reactive mineral phases (e.g. organic matter). Therefore, the dissolution after exposure to O2 of the inorganic contaminants present in the sediments of the Fjord (Hg, Pb, Zn, Cu) is expected to be smaller than the one calculated for As.

Figure 2.5: Remobilization of As after oxidation of FeS around the tube walls.

The present application of the TRANSCAP-1D model includes the remobilization of As in the bio-tubes crossing the anoxic layer, but the coprecipitation of As with oxihydroxides is not explicitly represented. In fact, we assume that the advective transport in the tubes is very fast so that the released As reaches the surface before adsorption onto oxihydroxides. Thus, the simulation rather presents the "worst case scenario", corresponding to the maximal release and mobility of arsenic. Details of the representation of the reaction included in the model will be discussed in the following section.

Several input parameters are needed to simulate the migration of contaminants in sediments containing bio-tubes with the TRANSCAP-1D model. These parameters have been either measured in the Saguenay Fjord or they are estimated from other studies. A discussion of the values of all parameters used for the simulations, as well as limitations in some cases, is given below.

As stated in the introduction, the duration of the consolidation and the fluid advection depends on the sediment properties (Zeman 1994). The hydraulic conductivity of the sediments of the Saguenay Fjord is generally so small that fluid velocities are rarely sufficient to induce solute advection and solute transport is primarily carried by diffusion, except if there is consolidation of the sediments. Consolidation of the contaminated sediments can be evaluated knowing that the total vertical stress added with the deposition of the capping layer is about 1 kPa for an average cap thickness of 0.2 m. This stress does not reach the preconsolidation pressure of the underlying sediments, which varies between 1 and 3 kPa (Perret 1995, Locat and Leroueil 1987). Thus, consolidation of the contaminated layer and advection of contaminated water in the clean sediments are assumed to be negligible. To investigate if consolidation can lead to advection of solutes in the new sediment cap, a series of permeability tests in oedometric cells were conducted. These experiments were performed on a few intact sediment samples of the Saguenay Fjord. The test consisted in compressing the sediment sample by imposing a given pressure and measuring the permeability of the sample at different consolidation stages. The test allowed estimating the consolidation that occurred in the sediments of the fjord after the emplacement of the cap in 1996. The laboratory tests showed that the average hydraulic conductivity of the sediments is 10-8 m s-1. The compression index, Cc, has an average value of 0.55 and the void ratio is 2.3. From these values, we calculate a coefficient of consolidation equal to 0.44 m2 a-1. The corresponding consolidation time for a cap thickness of 0.2 m is thus estimated at 33 d. These results suggest that, in the case of the Saguenay Fjord, the consolidation of the capping layer lasted only a few weeks and that advection related to consolidation is a negligible long-term transport process. Thus, the fluid velocity in the sediments is assumed equal to zero in the numerical simulations.

The release of contaminant associated with the bio-irrigation is represented in the model as a source term S in the transport equation (2.3) of the tube. The source term can be expanded in the following way:

(2.6)

where γ is a kinetic rate [T-1]. This term specifically represents the release of trace metals following the exposure of FeS to the oxygenated water circulating in the tubes. It is important to note that, since the source term is located in the tubes-equation, the release only occurs in the tubes. However, the two equation (2.1) and (2.3) are coupled through a mass-transfer term, thus the release will affect the solute concentration in the sediment pores.

As shown by Saulnier and Mucci (2000), the oxidation of FeS is very rapid and the dissolved Fe concentration in water attains its maximum in a few minutes. Thus we assume that the dissolution of trace metals associated to FeS is fast and a large value for γ equal to 10 000 d-1 is used. Since the model is specifically adapted to simulate the fate of As, the parameter L represents the maximum dissolved concentration of As for a definite FeS content of the sediments and is given by:

(2.7)

where k1 and k2 are regression parameters that can be viewed as dissolution coefficients and CFeS [M L-3] is the concentration of FeS. The resuspension experiments conducted by Saulnier and Mucci (2000) on the sediments of the Saguenay Fjord indicated that the amount of As released is linearly proportional to the content of solid FeS of the resuspended sediments. These observations suggest that the oxygenated water introduced in the sediments through bio-irrigation may lead to the release of As next to the tubes and burrows. Since the conditions in bio-irrigated sediments are not comparable to resuspension, we do not expect that the coefficients k1 and k2 obtained from resuspension experiments can be used for simulating dissolution during bio-irrigation. Thus, we instead derive the coefficients directly from the concentration of dissolved As and solid FeS measured by the group of geochemists of the Saguenay Project, supervised by Prof. Alfonso Mucci, on the sediment cores sampled at the Station 1 and Station 2 of the Saguenay Fjord (see Figure 2.4). The dissolved As concentrations were measured with an atomic absorption spectrophotometer on porewater samples extracted from the sediments with a modified Reeburgh-type squeezer, and successively filtered and refrigerated until analysis. The solid FeS content was determined by measuring the H2S generated during acidification of the freeze-dried and homogenized sediment sample. The detailed description of the analytical methods can be found in Mucci et al. (2000a). The values calculated are shown in Table 2.2.

The molecular diffusion coefficient Dm of dissolved As is derived from the values presented in Vanysek (2000) and Domenico and Schwartz (1998) and is assumed equal to 2.5 x 10-5 m2 d-1. The effective diffusion coefficient Dd is calculated by multiplying the molecular diffusion coefficient Dm with a mean estimated tortuosity value of 0.64, and corresponds to 1.6 x 10-5 m2 d-1. The retardation factor is fixed to a value of 15, based on the range of the partition coefficient Kd of As presented in the literature (Fuller 1978).

The bio-irrigation velocity was obtained from values presented in the literature. Riisgard (1989, 1991) carried out laboratory experiments to measure the irrigation velocities of two species of polychaetes, Chaetopterus Variopedatus and Nereis Diversicolor . They reported velocities of 130 m d-1 for Chaetopterus Variopedatus and 26 m d-1 for Nereis Diversicolor, at a temperature of 15 °C. These velocities represent the pumping activity of two polychaete species that live in a shallow water environment (between 0 and 30 m depth). Because the fauna of the Saguenay Fjord lives at a depth of 150 m under very different conditions, we must adjust the reported velocities to the fjord’s settings. Riisgard et al. (1992) observed a linear relationship between the temperature and the pumping efficiency of Nereis Diversicolor , at temperature ranges between 8 and 28 °C. Moreover, it was observed that lower temperatures correspond to longer pauses between bio-irrigation periods. Since the model cannot represent periodical bio-irrigation, we consider that the irrigation pauses reduce the average bio-irrigation velocity. Accounting for these temperature-effects, we estimate that a velocity value of 1 m d-1 is more representative for a water temperature of 1 °C, equal to that of the bottom waters of the Saguenay Fjord.

As stated previously, the cylindrical model can be reduced to the non-local exchange model, under specific assumptions. Boudreau (1984) showed that the two models are related with an equation that calculates the non-local exchange coefficient β1 at the surface of the sediments using the parameters of the cylindrical diffusion model (Figure 2.1).

(2.8)

where r1 is the inner radius of the tubes or burrows [L], r2 is the half-distance between two tubes/burrows [L] and δ is the distance [L] from the burrow axis to a point where the concentration equals the horizontally integrated value. Observations on the sediment cores sampled at the study site suggest that the inner radius of the tubes r1 is approximately equal to 0.001 m. The half-distance between two tubes r2 is related to the quantity of tubes per unit area m [L-2]:

(2.9)

The value of m was visually determined from photographs of the undisturbed bottom sediments taken during the summer 2001. A total of 1000 structures per square meter represents a good approximation for the study site and we thus assigned to r2 a value of 0.015 m. The distance δ is assumed equal to 0.0026 m (Furukawa et al. 2000). From equation (2.8), using an effective diffusion coefficient for As equal to 1.6 x 10-5 m2 d-1, the value of β1 is equal to 0.0893 d-1. This value agrees with those published elsewhere for sediments bioturbated by polychaetes and oligochaetes worms (Table 2.3).

The sediment porosity is assumed to be equal to 0.7, which is the average of all values measured at the study site, while the porosity of the tubes n°T at the surface is calculated by assuming a cylindrical shape:

(2.10)

where m represents the quantity of tubes per unit area [L-2]. Assuming a radius r1 of 0.001 m and an average of 1000 tubes per square meter, the tube porosity at the surface of the sediments n°T is equal to 0.003.

The value of β2 in equation (2.4) and (2.5) is directly related to the depth of the burrows. Various sediment cores sampled in the Baie des Ha! Ha! have been analyzed by the axial tomodensitometer of the Regional Hospital Center of Rimouski. This non-destructive method allows to quantify the bioturbation structures (tubes and burrows) (De Montety et al. 2000). The results show that the structures can be observed down to a sediment depth of 0.15 m at two of five sampling stations in the fjord, with a maximum number of structures located in the upper 0.05 m. On the other hand, field and laboratory observations show that the burrows can reach a depth of 0.2 m. Assuming that the burrows reach a maximal depth of 0.2 m, a corresponding value of β2 of 20 m-1 is used in the model. If the burrows reach only 0.1 m, the value of β2 is increased to 50 m-1.

TRANSCAP-1D is used to simulate the evolution of the dissolved As concentration at two sampling stations located in the Bras Nord (Station 1) and in the Baie des Ha! Ha! (Station 2) (Figure 2.4). The response of the model is compared to the profiles of dissolved As measured at those stations in 1998, 2 years after the flood. The input parameters discussed in the previous section are used to approximate the general conditions at the two stations (Table 2.4).

The model simulates the evolution of contamination in a sediment column with a thickness of 0.6 m. The column is equally subdivided in 30 cells, each having a height of 0.02 m. The concentration of dissolved As at the sediment-water interface has been fixed at 1125 μg m-3 in accordance to the value measured in 1994 at the water-sediment interface by Mucci et al. (2000b). We assume solutes leaving the sediments from the upper boundary do not accumulate in water but are promptly carried away by the bottom currents. On the other hand the solute mass reaching the lower control volume by diffusion leaves the system at the same rate. Since the sediments underlying the modeled system are assumed to be clean, we define the advective flux entering the sediment column from the lower boundary to be zero. Thus the concentration at the lower limit stays at a constant value of zero. The simulation time is 2 years (730 d) and the time increment is 0.02 d.

The initial concentration of FeS corresponds to the concentration profiles that have been measured at the two sampling stations in 1998, 2 years after the flood. We assume that the distribution of FeS did not significantly change during the 2 years that followed the deposition of the cap, which agrees with the conceptual model proposed by Mucci et al. (2000a) for the migration of the oxidation front through the capping layer after its deposition. According to Mucci's model, shortly after deposition the depth of penetration of oxygen extended through the flood layer, but within three weeks the oxidation front migrated up towards the new sediment-water interface and attained a new steady state condition. Therefore it can be assumed that the distribution of FeS reached equilibrium a few weeks after deposition and did not change afterwards. The simulations begin immediately after the end of the migration of the oxidation front, when FeS reaches equilibrium, a few weeks after the flood of 1996. Thus the concentration profile of FeS is defined at the beginning of the simulation and is assumed to be constant until the end. At the top cell of the discretized sediment column the concentration of Fe monosulfide was set to zero. This corresponds to the oxic layer, which in the Saguenay sediments has a thickness varying between few millimeters to 0.02 m (Mucci 2000b).

The thickness of the capping layer is 0.2 m at Station 1 and 0.28 m at Station 2. Within this layer, the initial concentration of dissolved As is zero. For the underlying sediments, the initial concentration of dissolved As at the two sampling stations had to be extrapolated by calibrating the model to reproduce the peak concentration measured in 1997 and in 1998. The peak As concentration at Station 1 is located at a depth of 0.29 m and attains 90 000 μg m-3 whereas at the Station 2, the maximum concentration is located 0.35 m under the sediment-water interface and reaches 100 000 μg m-3.

The results for the model calibration at the two sampling stations are shown in Figure 2.6, where the simulated profiles of dissolved As are compared to the concentrations measured in 1998, 2 years after the flood. The comparison between the response of the model and the measured profiles shows that the numerical model reproduces the spatial and temporal patterns of As concentrations. The dissolved As profiles at both stations clearly show the effect of bio-irrigation and remobilization, which is represented by the increased concentration of dissolved As in the first 0.2 m beneath the sediment-water interface.

Figure 2.6 (a) and (b): Response of the numerical model for a simulation time of 2 years (dashed line) compared to the concentrations measured in 1998 (symbols) at Station 1 (a) and at Station 2 (b). The initial concentration of As is represented by the solid line.

Since the flood of 1996, a large amount of data has been collected on the sediments of the Saguenay Fjord. More than 300 stations were sampled to define the geotechnical, geochemical and biological characteristics of the sediments. The collected data show a very strong spatial variability. For example, the thickness of the capping layer varies from 0 m to 2 m, the number of benthic structures (tubes and burrows) can fluctuate from less than 500 m-2 to more than 1500 m-2 and the depth of bio-irrigation varies from 0.1 m to more than 0.2 m. These characteristics not only change from station to station, but also show a degree of variability within a single station. Moreover, it is very difficult to correlate data from samples collected at a given station, but at different times, because the exact positioning of the sampling instruments with respect to the ship location could not be determined most of the time. Also, since the sampling stations were often located at depths greater than 100 m, the tool collecting the sample drifted under the influence of the strong bottom currents.

Although the calibration of the model to the dissolved As profiles at Stations 1 and 2 is satisfactory, those stations do not necessarily reflect the conditions everywhere in the fjord. A series of simulation is presented here to illustrate the variability of the model response for six different cases (Table 2.5). The first four simulations represent the possible evolution of As depending on the cap thickness and the bio-irrigation depth. The last two simulations show the response of the model for a general contaminant, without reactions. All simulations are performed over a period of 5 years and show the evolution of the contamination at a simulated time of 1 day, 2 years and 5 years. The initial concentration of contaminant in the capping layer is zero.

The evolution of contamination using the same input parameters as those used for Station 1 (Table 2.2 and 2.4) is shown in Figure 2.7a. In that case, the cap is 0.2 m thick and the bio-irrigation reaches a depth of 0.2 m, which means that the worm tubes completely penetrate the clean sediments and create a direct connection between the contaminated sediments and the well oxygenated water of the fjord. The contact between the bio-irrigated oxygenated water and the As associated to FeS in the sediments leads to the release of As next to the tubes and burrows. This process is clearly shown by the rapid increase of the concentration in the first 0.2 m below the sediment-water interface. A similar simulation, with a depth of bio-irrigation reduced to 0.1 m, is shown in Figure 2.7b. In this case, the worm tubes do not completely penetrate the clean sediment cap. The upper part of the graphic shows a different trend compared to the previous simulation with concentrations of dissolved As decreasing above the contaminated sediments, with a minimum value at about 0.14 m, and then increasing again to a second maximum. The asymmetry observed in the upper part of the sediments after a simulation time of 2 years becomes smoother and finally disappears after 5 years. The lower section of the sediments follows the normal evolution of a diffusing contaminant peak, showing a symmetric trend. The bio-irrigation causes the subdivision of the concentrations in two regions: an upper region controlled by the pumping activity of the benthic fauna and a lower region dominated by diffusive processes.

Figure 2.7 (a) and (b): Evolution of contamination after the deposition of a capping layer of 0.2 m. The bio-irrigation depth (B. D.) attains 0.2 m in Simulation 1 (a) and 0.1 m in Simulation 2 (b).

The next simulation (simulation 3) shows the evolution of the contamination for sediments with a bio-irrigation depth of 0.2 m and a capping layer of only 0.1 m in thickness (Figure 2.8a). The effect of bio-irrigation is clearly illustrated by the decrease of the concentration peak from 100 000 μg m-3 to 25 000 μg m-3 within 2 years. This decrease results from the exchange of dissolved As between the sediments and the tubes. Once the contaminant is transferred from the sediment pores to the tubes, it is pumped very rapidly towards the upper boundary. On the other hand, bio-irrigation brings oxygenated water into the sediments, inducing the dissolution of As associated to FeS, and thus maintaining the dissolved As concentration in the upper 0.2 m of the sediments at a relatively high concentration. For simulation 4, the depth of bio-irrigation is reduced to 0.1 m, which modifies the simulated concentration profile as shown in Figure 2.8b. In this case, the effect of bio-irrigation is weaker but can still be recognized in the asymmetrical contaminant distribution, which shows a slightly higher concentration in the upper few centimeters. Because the bio-irrigation depth does not reach the contaminated layer, the peak decreases more slowly.

Figure 2.8 (a) and (b): Evolution of contamination after the deposition of a capping layer of 0.1 m. The bio-irrigation depth attains 0.2 m in Sim. 3 (a) and 0.1 m in Sim. 4 (b).

The last two simulations investigate the case of a non-reactive contaminant, where "non-reactive" means a compound that does not undergo dissolution following the exposure to oxygen. In fact dissolution may not affect the fate of other inorganic compounds. Figures 2.9a and 2.9b illustrate the response of the model for a non-reactive contaminant for a capping layer of 0.14 m and two different values of bio-irrigation depth, equal to 0.1 m and 0.2 m. The concentrations in the first 0.2 m below the sediment-water interface (Figure 2.9a) are smaller than those for the reactive case (Figures 2.7a and 2.8a). This difference comes from the exclusion of the dissolution term, thus the lack of a contaminant source in the bio-irrigated layer. A similar trend exits for the case where the bio-irrigation depth is equal to 0.2 m (Figure 2.9b). The absence of dissolution produces smaller concentrations in the bio-irrigated layer compared to reactive cases of Figures 2.7b and 2.8b. The profile in Figure 2.9b is also more symmetrical than those presented in Figures 2.7b and 2.8b.

Figure 2.9 (a) and (b): Evolution of contamination for a non-reactive contaminant after the deposition of a capping layer of 0.14 m. The bio-irrigation depth attains 0.2 m in Sim. 5 (a) and 0.1 m in Sim. 6 (b).

The total mass of contaminant released from the upper boundary of the model, 5 years after capping, is presented in Table 2.5 for the six simulations. After 5 years, the total mass of contaminant that reaches the water column in simulation 1 is 5791 μg m-2 for a bio-irrigation depth of 20 and it is equal to 2421 μg m-2 for simulation 2 with a reduced bio-irrigation depth of 0.1 m. These results indicate that the depth of bio-irrigation has an important effect on the contaminant mass leaving the sediments from the upper boundary. On the other hand, simulations 3 and 4 show similar results to the ones obtained for the first two simulations. Since simulations 3 and 4 only differ from simulations 1 and 2 by the thickness of the capping layer, this thickness does not seem to influence the total contaminant flux at the upper boundary. For simulations 5 and 6, the simulated mass released is much smaller than for the simulations with dissolution. This result is not surprising, because dissolution increases the amount of dissolved contaminant in the bio-irrigated layer, enhancing the mass leaving from the upper boundary of the sediments. Thus, the lack of dissolution leads to smaller concentrations in the bio-irrigated layer and to a smaller release of contaminant. Comparing simulation 5 and 6, we observe that in simulation 6, where the bio-irrigation depth is only 0.1 m, the release of contaminant is much smaller. This result indicates that if the burrows do not reach the contaminated layer the release of contaminant will be minimal. Thus for the case without dissolution, the bio-irrigation depth as well as the thickness of the capping layer are the more significant factors in preventing the contaminant release.

The cumulative contaminant mass released by the sediments for the 5-year period, shown in Figure 2.10, reveals the effect of the depth of bio-irrigation and dissolution on mass release. For the case with dissolution (simulations 1 and 2), the total mass release follows a linear trend with a gradient depending on the depth of bio-irrigation. Conversely, for a non-dissolving contaminant (simulation 5 and 6), the release decreases with time and the cumulated mass seems to be heading towards a plateau.

Figure 2.10: Predicted contaminant release from the sediments to the water column, over period of 5 years, for a contaminant undergoing dissolution (Sim. 1 and 2) and for a non-reactive contaminant (Sim. 5 and 6).

A numerical model that simulates one-dimensional vertical migration of dissolved compounds through a sediment cap is presented. The model has been calibrated with the dissolved As concentrations in the sediments, measured at two sampling stations in the Saguenay Fjord. A good match of the simulated and measured values indicates that the model can represent the evolution of As through a sediment cap. The calibrated simulations represent a "worst case scenario" and for the inorganic contaminants present in the sediments of the Fjord (Hg, Pb, Zn, Cu) a smaller release is expected.

Detailed data collection for the fjord's sediments indicates that physical and chemical parameters are quite variable and change according to the location of the sampling station. To illustrate how the variability of the input parameters might affect the response of the model, we present the results of hypothetical simulations where the bio-irrigation depth and the thickness of the capping layer are varied. Also, in an attempt to represent the possible evolution of the contamination for other dissolved compounds, some simulations that exclude the dissolution term from the transport equation are presented. The model was also used to calculate the total mass of contaminant that left the sediments from the upper boundary at the end of the simulation. If we compare these values for the first series of simulations, we notice that the bio-irrigation depth has a more significant influence on the release of contaminant than the thickness of the capping layer. In fact, for a contaminant having properties similar to those of As, which undergoes dissolution if exposed to oxygenated water, the contaminant concentration in the bio-irrigated layer is dominated by the dissolution reaction and this process controls the release of contaminant. On the other hand, for a non-reactive contaminant (simulations 5 and 6) the release of contaminant is highly reduced and depends very much on the penetration depth of the tubes compared to the cap thickness. The release will be increased only if the tubes reach the contaminated layer. Whether or not there is dissolution, the depth of bio-irrigation is a very important factor.

A comparison of the dissolved concentrations calculated by the model for the upper boundary of the sediments with the water quality criteria recommended by the USEPA (USEPA 1999) shows that the As concentration does not exceed the Criterion Continuous Concentration for saltwater, which is equal to 36 000 μg m-3. This criterion is an estimate of the highest concentration to which an aquatic community can be exposed indefinitely without resulting in an unacceptable effect. Moreover the As released through oxidation of FeS will migrate to the new sediment-water interface where As(III) will be oxidized to As(V) and adsorbed to oxihydroxides. Thus a significant fraction of the released As will precipitate again at the surface of the sediments with the iron oxides and only a small part will be transferred to the overlying water. In addition As leaving the sediments from the upper boundary will not accumulate in the water layer over the sediments, but will be transported by the bottom currents, thus mixing and diluting with the bottom water. Therefore we do not expect the released As to become an environmental problem in the case of the Saguenay Fjord.

Even if the Saguenay Fjord does not present real concern regarding the concentration of dissolved As in the water column, it is very instructive to study the evolution of this compound through the capping layer. In fact the data collected after the flood event could be used to calibrate the model and assess the importance of bio-irrigation in the transport of As. To better understand the system and its interactions, additional simulations are needed to incorporate in more details the variability and the uncertainty of the input parameters. A detailed sensitivity analysis, based on factorial design, has been undertaken to identify the parameters with the most significant effect on the response of the model.

Aggett, J. and O’Brien, G. A. 1985. Detailed model for the mobility of arsenic in lacustrine sediments based on measurements in Lake Ohakuri. Environ. Sci. Technol. 19 : 231-238.

Aller, R. C. 1980. Quantifying solute distributions in the biotubated zone of marine sediments by defining an average microenvironment. Geochim. Cosmochim. Acta 44 : 1955-1965.

Aller, R. C. and Aller, J. Y. 1998. The effect of biogenic irrigation intensity and solute exchange on diagenetic reaction rates in marine sediments. J. Mar. Res. 56 : 905-936.

Aller, R. C. and Yingst, J. Y. 1978. Biogeochemistry of tube-dwellings: A study of the sedentary polychaete Amphitrite ornata (Leidy). J. Mar. Res. 36 (2): 201-254.

Aller, R. C. and Yingst, J. Y. 1985. Effects of the marine deposit-feeders Heteromastus filiformis (Polychaeta), Macoma balthica (Bivalvia), and Tellina texana (Bivalvia) on averaged sedimentary solute transport, reaction rates, and microbial distributions. J. Mar. Res. 43 : 615-645.

Archer, D. and Devol, A. 1992. Benthic oxygen fluxes on the Washington shelf and slope: A comparison of in situ microelectrode and chamber flux measurements. Limn. Oceanogr. 37 (3): 614-629.

Barbeau, C., Bougie, R. and Côté, J.-E. 1981. Temporal and spatial variations of mercury, lead, zinc and copper in sediments of the Saguenay fjord. Can. J. Earth Sci. 18 : 1065-1074.

Belzile, N. and Tessier, A. 1990. Interactions between arsenic and iron oxyhydroxides in lacustrine sediments. Geochim. Cosmochim. Acta 54 : 103-109.

Boudreau, B. P. 1984. On the equivalence of nonlocal and radial diffusion models for porewater irrigation. J. Mar. Res. 42 : 731-735.

Boudreau, B. P. 1999. Metals and models: Diagenetic modelling in freshwater lacustrine sediments. J. Paleolimn. 22 : 227-251.

Boyer, J. M., Chapra, S. C., Ruiz, C. E. and Dortch, M. S. 1994. RECOVERY, a mathematical model to predict the temporal response of surface water to contaminated sediments. Technical report W-94-4, U.S. Army Engineer Waterways Experiment Station, Vicksburg, MS.

Cossa, D. 1990. Chemical contaminants in the St. Lawrence Estuary and Saguenay Fjord. In : El-Sabh, M. I., Silverberg, N. (Eds.), Oceanography of a large scale estuarine system, The St. Lawrence. Coastal and Estuarine Studies 39 : 239-268, Springer Verlag.

De Montety, L., Long, B., Desrosiers, G., Crémer, J.-F. and Locat, J. 2000.Quantification des structures biogènes en fonction d’un gradient de perturbation dans la baie des Ha! Ha! à l’aide de la tomodensitométrie axiale. Proceedings of the 53th Canadian Geotechnical Conference, Montréal, 15-18 Oct. 2000, Vol. 1, pp. 131-135.

Domenico, P. A. and Schwartz, F. W. 1998. Physical and chemical hydrogeology. Wiley, New York.

Emerson, S., Jahnke, R. and Heggie, D. 1984. Sediment-water exchange in shallow water estuarine sediments. J. Mar. Res. 42 : 709-730.

Forster, S. and Graf, G. 1995. Impact of irrigation on oxygen flux into the sediment: intermittent pumping by Callianassa subterranea and "piston-pumping" by Lanice conchilega. Mar. Biol. 123 : 335-346.

Fuller, W. H. 1978. Investigation of landfill leachate pollutant attenuation by soils. US EPA, Municipal Environmental Research Laboratory, Cincinnati, OH

Furukawa, Y., Bentley, S. J., Shiller, A. M., Lavoie, D. L. and Van Cappellen, P. 2000. The role of biologically-enhanced pore water transport in early diagenesis: an example from carbonate sediments in the vicinity of North Key Harbor, Dry Tortugas National Park, Florida. J. Mar. Res. 58 : 493-522.

Furukawa, Y., Bentley, S. J. and Lavoie, D. L. (2001). Bioirrigation modeling in experimental benthic mesocosms. J. Mar. Res. 59 : 417-452.

Gagnon, C., Pelletier, É. and Maheu, S. 1993. Distribution of trace metals and some major constituents in sediments of the Saguenay Fjord, Canada. Mar. Pollut. Bull. 26 (2): 107-110.

Gobeil C. and Cossa D. 1984. Profils des teneurs en mercure dans les sédiments et les eaux interstitielles du fjord du Saguenay (Québec): données acquises au cours de la période 1978-83. Rapport Technique Canadien Hydrographie et Sciences Océaniques : 53 .

Guinasso, N. L. and Schink, D. R. 1975. Quantitative estimates of biological mixing rates in abyssal sediments. J. Geoph. Res. 80 : 3032-3043.

Huerta-Diaz, M. A., Tessier, A. and Carignan, R. 1998. Geochemistry of trace metals associated with reduced sulfur in freshwater sediments. Appl. Geochem. 13 : 213-233.

Jorgensen, B. B. and Revsbech, N. P. 1985. Diffusive boundary layers and the oxygen uptake of sediments and detritus. Limn. Oceanogr. 30 (1): 111-122.

Locat, J. and Leroueil, S. 1987. Physicochemical and geotechnical characteristics of recent Saguenay Fjord sediments. Can. Geotech. J. 25 : 382-388

Marinelli, R. L. and Boudreau, B. P. 1996. An experimental and modeling study of pH and related solutes in an irrigated anoxic coastal sediment. J. Mar. Res. 54 : 939-966.

Martin, W. R. and Banta, G. T. 1992. The measurement of sediment irrigation rates: A comparison of the BR- tracer and 222RN/226RA disequilibrium techniques. J. Mar. Res. 50 : 125-154.

Matisoff, G., Wang, X. 1998. Solute transport in sediment by freshwater infaunal bioirrigators. Limn. Oceanogr. 43 (7): 1487-1499.

Maurice, F. 2000. Caractéristiques géotechniques et évolution de la couche de sédiment déposée lors du déluge de 1996 dans la Baie des Ha! Ha! (Fjord du Saguenay, Québec). M.Sc. Thesis, Université Laval.

McCaffrey, R. J., Myers, A. C., Davey, E., Morrison, G., Bender, M., Luedtke, N., Cullen, D., Froelich, P. and Klinkhammer, G. 1980. The relation between pore water chemistry and benthic fluxes of nutrients and manganese in Narragansett Bay, Rhode Island. Limn. Oceanogr. 25 (1): 31-44.

Meyers, M. B., Fossing, H. and Powell, E. N. 1987. Microdistribution of interstitial meiofauna, oxygen and sulfide gradients, and the tubes of macro-infauna. Mar. Ecol. Progr. Ser. 35 : 223-241.

Mohan, R. K., Mageau, D. W. and Brown, M. P. 1999. Modeling the geophysical impacts of underwater in-situ cap construction. Mar. Tech. Soc. J. 33 (3): 80-87.

Mohan, R. K., Brown, M. P. and Barnes, C. R. 2000. Design criteria and theoretical basis for capping contaminated marine sediments. Appl. Ocean Res. 22 : 85-93.

Morse, J. W. 1994 . Interactions of trace metals with authigenic sulfide minerals: implications for their bioavailability. Mar. Chem. 46 : 1-6.

Mucci, A. and Edenborn, H. M. 1992. Influence of an organic-poor landslide deposit on the early diagenesis of iron and manganese in coastal marine sediment. Geochim. Cosmochim. Acta 56 : 3909-3921.

Mucci, A., Guignard, C. and Olejczuyk, P. 2000a. Mobility of metals and As in sediments following a large scale episodic sedimentation event. Proceedings of the 53th Canadian Geotechnical Conference, Montréal, 15-18 Oct. 2000, Vol. 1, pp. 169-175.

Mucci, A., Richard, L.-F., Lucotte, M. and Guignard, C. 2000b. The differential geochemical behaviour of arsenic and phosphorous in the water column and sediments of the Saguenay fjord estuary, Canada. Aquatic Geochem. 6 : 293-324.

Neville C. J., Ibaraki, M. and Sudicky, E. A. 2000. Solute transport with multiprocess nonequilibrium: a semi-analytical solution approach. J. Cont. Hydr. 44 : 141-159.

O’Day, P. A., Carroll, S. A., Randall, S., Martinelli, R. E., Anderson, S. L., Jelinski, J. and Knezovich, J. P. 2000. Metal speciation and bioavailability in contaminated estuary sediments, Alameda Naval Air Station, California. Environ. Sci. Technol. 34 : 3665-3673.

Park, S. S. and Jaffé P. R. 1996. Development of a sediment redox potential model for the assessment of postdepositional metal mobility. Ecol. Model. 91 : 169-181.

Pelletier, E. Desrosiers, G., Locat, J., Mucci, A. and Tremblay, H. 2003. The origin and behavior of a flood capping layer deposited on contaminated sediments of the Saguenay Fjord (Quebec). In: Contaminated sediments: Characterization, Evaluation, Mitigation/Restoration and Management strategy Performance, ASTM STP1442, J. Locat, R. Galvez-Cloutier, R.C. Chaney, and K.R. Demars, Eds. ASTM International, West Conshohocken, PA, pp. 3-18.

Pelletier , E. and Canuel G. 1988.Trace metals in surface sediment of the Saguenay fjord, Canada. Mar. Pollut. Bull. 19 : 336-338

Perret, D. 1995. Diagenèse mécanique précoce des sédiments fins du fjord du Saguenay. Ph.D Thesis, Université Laval

Petersen, W., Willer, E. and Willamowski, C. 1997. Remobilization of trace elements from polluted anoxic sediments after resuspension in oxic water. Wat. Air Soil Poll. 99 : 515-522.

Petersen, K., Kristensen, E. and Bjerregaard, P. 1998. Influence of bioturbating animals on flux of cadmium into estuarine sediment. Mar. Environ. Res. 45 (4/5): 403-415.

Reible, D. D., Popov, V., Valsaraj, K. T., Thibodeaux, L. J., Lin, F., Dikshit, M., Todaro, M. A. and Fleeger, J. W. 1996. Contaminant fluxes from sediment due to tubificid oligochaete bioturbation. Wat. Res. 30 (3): 704-714.

Riedel, G. F., Sanders, J. G. and Osman, R. W. 1987. The effect of biological and physical disturbances on the transport of arsenic from contaminated estuarine sediments. Estuarine, Coast. Shelf Sci. 25 : 693-706.

Riisgard, H. U. 1989. Properties and energy cost of the muscular piston pump in the suspension feeding polychaete Chaetopterus variopedatus. Mar. Ecol. Progr. Ser. 56 : 157-168.

Riisgard, H. U. 1991. Suspension feeding in the polychaete Nereis diversicolor. Mar. Ecol. Progr. Ser. 70 : 29-37.

Riisgard, H. U., Vedel, A., Boye, H. and Larsen, P. S. 1992. Filter-net structure and pumping activity in the polychaete Nereis diversicolor : effects of temperature and pump-modelling. Mar. Ecol. Progr. Ser. 83 : 79-89.

Rivera-Duarte, I. and Flegal, A. R. 1994. Benthic lead fluxes in San Francisco Bay, California, USA. Geochim. Cosmochim. Acta 58 : 3307-3313.

Ruiz, C. E., Schroeder, P. R. and Aziz, N. M. 2000. RECOVERY: A contaminant sediment-water interaction model. ERDC/EL SR-D-00-1, U.S. Army Engineer Research and Development Center, Waterways Experiment Station, Vicksburg, MS.

Saulnier, I. and Mucci, A. 2000. Trace metal remobilization following the resuspension of estuarine sediments: Saguenay Fjord, Canada. Appl. Geochem. 15 : 191-210

Simpson, S. L., Rosner, J. and Ellis, J. 2000. Competitive displacement reactions of cadmium, copper, and zinc added to a polluted, sulfidic estuarine sediment. Environ. Toxicol. Chem. 19 : 1992-1999.

Soetaert, K., Herman, P. M. J. and Middelburg J. J. 1996. A model of early diagenetic processes from the shelf to abyssal depths. Geochim. Cosmochim. Acta 60 (6): 1019-1040.

Tremblay, G.-H. and Gobeil, C. 1990. Dissolved arsenic in the St Lawrence Estuary and the Saguenay Fjord, Canada. Mar. Pollut. Bull. 21 (10): 465-469.

Tremblay, H., Desrosiers, G., Locat, J., Mucci, A. and Pelletier, É. 2003. Characterization of a catastrophic flood sediment layer: geological, geotechnical, biological, and geochemical signatures. In: Contaminated sediments: Characterization, Evaluation, Mitigation/Restoration and Management strategy Performance, ASTM STP1442, J. Locat, R. Galvez-Cloutier, R.C. Chaney, and K.R. Demars, Eds. ASTM International, West Conshohocken, PA, pp. 87-101.

USEPA 1999. National recommended water quality criteria - Correction. United States Environmental Protection Agency, Office of Water 4304. EPA 822-Z-99-001.

Vanysek P. 2000. Ionic Conductivity and diffusion at infinite dilution. In CRC handbook of chemistry and physics, 81st edition, Cleveland, Ohio.

Versteeg, H.K. and Malalasekera, W. 1995. An introduction to computational fluid dynamics : the finite volume method. Longman Scientific & Technical, Burnt Mill, Harlow, Essex.

Wang, X. and Matisoff, G. 1997. Solute transport in sediments by a large freshwater oligochaete, Brachiura sowerbyi. Environ. Sci. Technol. 31 : 1926-1933.

Wang, Y. and Van Cappellen, P. 1996. A multicomponent reactive transport model of early diagenesis: application to redox cycling in coastal marine sediments. Geochim. Cosmochim. Acta 60 (16): 2993-3014.

Zeman, A. J. 1994. Subaqueous capping of very soft contaminated sediments. Can. Geotech. J. 31 : 570-577.

Zheng, C. and Bennett, G. D. 2002. Applied Contaminant Transport Modeling. 2nd edition, John Wiley, New York.