Quality of estimated Freundlich parameters of cadmium sorption from pedotransfer functions to predict cadmium concentration

To predict the cadmium concentrations of soil solutions C-Cd under varying properties of soil matrix and electrolyte, we established Cd sorption isotherms for 225 samples from sandy, northern German arable and forest soils, and from these derived the Freundlich parameters k and M. Standard electrolyte was 5 mM Ca(NO3)(2). As the initially (native) sorbed fraction of Cd, we used the amount extracted with 0.025 M Na-2-EDTA at 20 degreesC, 2 h shaking. The average value of parameter M was 0.815 (-). There was some correlation of M with pH: samples above pH 6 had an average M of just 0.730. The main information about the sorption properties of the soils was contained in k, which could be predicted by multiple regressions from pH, organic carbon (OC,%) and clay content (%) for one subset of Ap horizons (r(2) = 0.96). When all 225 samples were combined, no more statistical influence was found for the variable "clay" on the multiple regression models. Clay, therefore, is not included in the final model in which Freundlich k (mg(1-M) L-M kg(-1)) is given by -0.993 pH(0.537) OC0.783. The resulting values are valid for 5 mM Ca(NO3)(2), 20 degreesC and were used, together with the mean M of 0.815 and the Freundlich equation, to predict the Cd concentrations of the soil solutions (C-Cd) of the 225 investigated sites, both for the current load of Cd and assuming higher contamination. In a large number of samples, the estimated C-Cd exceeded current drinking water threshold values and other solution-based critical limits when a total load of 1 mg Cd kg(-1) soil was assumed. In a final step, we corrected the predicted C-Cd for the strength of the electrolyte, here defined by the Ca2+ concentration, and the proportion of the complexing Cl- among the accompanying anions. The approach appears to be promising, but there are still some clear deficiencies concerning the prediction of the Freundlich exponent M and the influence on k of the independent variables DOC, time, and temperature on the one hand, and the contents of clay or oxides on the other.