Modeling of the solid-solution partitioning of heavy metals and arsenic in embanked flood plain soils of the rivers Rhine and Meuse.

The aim of this study is to predict the solid-solution partitioning of heavy metals in river flood plain soils. We compared mechanistic geochemical modeling with a statistical approach. To characterize the heavy metal contamination of embanked river flood plain soils in The Netherlands, we collected 194 soil samples at 133 sites distributed in the Dutch part of the Rhine and Meuse river systems. We measured the total amounts of As, Cd, Cr, Cu, Ni, Pb, and Zn in the soil samples and the metal fraction extractable by 2.5 mM CaCl2. We found a strong correlation between heavy metal contamination and organic matter content, which was almost identical for both river systems. Speciation calculations by a fully parametrized model showed the strengths and weaknesses of the mechanistic approach. Cu and Cd concentrations were predicted within one log scale, whereas modeling of Zn and Pb needs adjustment of some model parameters. The statistical fitting approach produced better results but is limited with regard to the understanding it provides. The log RMSE for this approach varied between 0.2 and 0.32 for the different metals. The careful modeling of speciation and adsorption processes is a useful tool for the investigation and understanding of metal availability in river flood plain soils.