The continuous demand on new residential and economic areas of the modern society has to face up with problems posed by polluted sites related to former industrial activities, typically located in suburbs areas. These sites, known as brownfields, are often located nearby navigable rivers to facilitate transport operations of industrial manufacturing, which increase their potential environmental threat due to the possible migration of pollutants in groundwater to surface water bodies through groundwater discharge. In this context, the objective of this research, performed in the scope of the FP6-IP AquaTerra project, was to contribute to a better assessment of the risk of groundwater contaminant dispersion for a brownfield located next to the Meuse River (Belgium), in a context where strong groundwater - surface water interactions prevail. The brownfield of interest corresponds to the site of the former coke factory of Flémalle. Resulting from industrial activities, soil and groundwater located in the alluvial aquifer are heavily contaminated with various types of organic (BTEX, PAHs, mineral oils...) and inorganic (As, Zn, Cd...) pollutants. To do so, detailed characterisation campaign was performed, consisting of, on the one hand, classical field experiments such as pumping tests, injection tests and tracer experiments; on the other hand, advanced and original field experiments such as detailed monitoring of groundwater - surface water interaction and dynamics, and the development and application of an innovative tracer technique, the Finite Volume Point Dilution Method (FVPDM), used to quantify and monitor groundwater fluxes. Monitoring and field works data was subsequently used to develop and calibrate a groundwater flow model using the finite difference code MODFLOW, with an automatic parameter estimation approach based on an original combined regional scale (zonation) and local scale (pilot points) approach. A transport model was also developed using MT3DMS and calibrated using tracer experiments performed in the brownfield. This groundwater flow and transport model was used to better quantify the dynamics of groundwater - surface water interactions and to model various scenarios of contaminant dispersion through the aquifer - river system. For these scenarios, benzene was considered because it is one of the main pollutants encountered in the site, its large solubility and mobility in groundwater and its acute toxicity. These scenarios were established considering various groundwater flow conditions (steady state vs. transient) and various hydrodispersive processes possibly affecting the mobility of benzene in groundwater, namely advection, hydrodynamic dispersion, sorption - desorption and, as evidenced by the research results of the University of Neuchâtel (Switzerland), benzene degradation under sulphate reducing conditions. These simulations indicate that benzene attenuation is mainly controlled by ongoing benzene degradation processes, aquifer heterogeneity and river stage fluctuations. Based on this analysis, the risk of benzene dispersion is low, and monitored natural attenuation (MNA) is a valuable option with (1) monitoring benzene at control planes downstream from the sources; (2) further investigation on risk of sulphate depletion in the alluvial aquifer; and (3) further investigation on mobilisation/immobilisation of heavy metals related to dynamics of organic pollutant plumes.

L'étude géologique préliminaire de la retenue d'accumulation souterraine de Sidi -lfni (Anti -Atlas, Maroc), a été menée en considérant que la connaissance de l'organisation structurale régionale est fondamentale pour analyser les écoulements souterrains en milieu rocheux. Pour cela, les systèmes de décrochement qui ont façonné la région, ont été analysés; ainsi, ont été établies la genèse et la géométrie de ces décrochements. Les travaux de reconnaissances (sondages, essais Lugeon, mesures piezométriques ont permis de préciser l'importance de la notion de " maille de fracturation " en milieu rocheux fracturé. Il apparaît que les données de reconnaissance doivent toujours être interprétées en fonction de l'objet géologique considéré. La confrontation des cartes piezométriques et du schéma de répartition de la fracturation semble indiquer qu'il existe une relation claire entre la position de zones structurales de nature différente et la distribution de la perméabilité à l'échelle régionale. Quant au projet initial, les données acquises montrent qu'il était très optimiste quant à la possibilité d'accumuler ainsi, puis d'exploiter les volumes d'eau envisagés.