MODELING THE TRANSPORT OF HEAVY METALS THROUGH A CAPPING-LAYER: THE CASE STUDY OF THE FLOOD SEDIMENTS DEPOSITED IN THE SAGUENAY FJORD, QUEBEC

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 Fjord. These clean sediments represent a potential barrier, or capping-layer, for the upward migration into the Fjord of heavy metals and PAHs present in the underlying sediment.

The evaluation of the long-term efficiency of the capping layer requires the knowledge of physical, chemical and biological factors affecting the migration of contaminants in sediments. These factors include advection related to consolidation, diffusion, chemical reactions, and the effect of the burrowing activity of benthic organism. Permeability tests in oedometric cells indicate that, for the clean sediments of the Saguenay Fjord, advection is a negligible long-term transport process. On the other hand, dwelling worms may play an important role because they create tubes that increase the actual surface of exchange for contaminants and they pump water in their burrows (bio-irrigation). This activity modifies the redox conditions next to the tubes and may lead to heavy metal remobilization.

To evaluate the long-term efficiency of the capping-layer, we have developed a numerical model that simulates one-dimensional vertical migration of dissolved heavy metals through a sediment cap. The model solves a system of coupled partial differential equations representing the contaminant flux in both the sediment pore water and in the irrigated burrows. The coupling is made via a non-local transfer term that represents the exchange of contaminant between the pore-water and the burrow-water. Using the data obtained through field and laboratory characterization for the Saguenay Fjord, a detailed sensitivity analysis is presented to illustrate the effect of various parameters on contaminant migration through the layer, and also to assess the impact on model predictions arising from the uncertainty in the knowledge of these parameters.