Evaluation of field-scale variability of heavy metal sorption in soils by scale factors — Scaling approach and statistical analysis

Abstract

The ability of soils to sorb heavy metals is quantified by sorption isotherms. The field-scale variability of heavy metal sorption isotherms across fields of apparently “homogeneous” soil is often very large and makes the upscaling of point measurements to larger scales problematic. This may be overcome by scaling of sorption isotherms, which is a method that potentially reduces the wide spread of the isotherms into a reference or average isotherm, respectively, but preserves the variation through calculated scale factors. At two study sites near Hannover, Germany (loess soil at Lathwehren, loamy to sandy soil at Vinnhorst), we investigated the field-scale variability of the sorption isotherms of cadmium (Cd), copper (Cu), lead (Pb) and zinc (Zn). For each site, 50 samples were taken along a 250m transect at two depths. Further, for each site and depth a composite sample was mixed from aliquots of the 50 transect samples. Sorption isotherms of single heavy metals were measured, along with a range of soil properties, including pH, CEC, OC, and texture. The isotherms were successfully parameterized by the Freundlich equation and were spatially very variable. Calculation of scale factors for the sorption isotherms was successful, as scaling reduction of variance was high (from 64% to 99%). We then tested if correlations between scale factors of different heavy metal sorption isotherms, and also to soil properties existed. Such correlation were expected, because heavy metals (e.g. Cd and Zn) are competitively adsorbed, and the respective soil properties directly relate to ion sorption in soil. Significant correlations between scale factors of heavy metal sorption isotherms were only found at one site and depth (in the loamy to sandy subsoil). Thus, these relationships were site- and depth specific and are not generalizable. In addition, significant correlations between scale factors of heavy metal sorption isotherms and soil properties occurred only sporadically, and were not transferrable. One possible reason for this might be that the data were transformed (difference transformation) prior to correlation analysis to obtain normality. A further objective was to prove, whether the average isotherms can be represented by a single measured isotherms of the composite samples from the area. This was found to hold at the loess site, but not at the loamy to sandy site. This indicates that scaling is favorable to sites not too much differing in soil texture. Although correlation analysis revealed only sporadic and not transferrable correlations, a multiple linear regression equation for the Lathwehren subsoil was found to predict Cu sorption scale factors from organic carbon content and cation exchange capacity. From our investigation we conclude that scaling is a useful tool to quantify and express field-scale variability of heavy metal sorption isotherms in soils. However, a prediction of scale factors from simple soil properties was only partly successful and needs further research efforts.