Critical evaluation of soil contamination assessment methods for trace metals.

Correctly distinguishing between natural and anthropogenic trace metal contents in soils is crucial for assessing soil contamination. A series of assessment methods is critically outlined. All methods rely on assumptions of reference values for natural content. According to the adopted reference values, which are based on various statistical and geochemical procedures, there is a considerable range and discrepancy in the assessed soil contamination results as shown by the five methods applied to three weakly contaminated sites. This is a serious indication of their high methodological specificity and bias. No method with off-site reference values could identify any soil contamination in the investigated trace metals (Pb, Cu, Zn, Cd, Ni), while the specific and sensitive on-site reference methods did so for some sites. Soil profile balances are considered to produce the most plausible site-specific results, provided the numerous assumptions are realistic and the required data reliable. This highlights the dilemma between model and data uncertainty. Data uncertainty, however, is a neglected issue in soil contamination assessment so far. And the model uncertainty depends much on the site-specific realistic assumptions of pristine natural trace metal contents. Hence, the appropriate assessment of soil contamination is a subtle optimization exercise of model versus data uncertainty and specification versus generalization. There is no general and accurate reference method and soil contamination assessment is still rather fuzzy, with negative implications for the reliability of subsequent risk assessments.