Analysis of procedures for sampling contaminated soil using Gy's Sampling Theory and Practice

Soil sampling is a critical step in environmental site assessment studies. The representativeness of soil samples has a direct influence on financial, liability, environmental and public health issues associated with the outcome of remediation activities. Representativeness must be quantified for assessing and designing soil sampling procedures. Gy's Sampling Theory and Practice (STP) was used to analyze the reproducibility of two soil sampling procedures, namely a procedure based on grab sampling (GSP) and an alternative procedure (ASP) developed from STP principles. Sampling reproducibility, a component of sampling representativeness, was determined by theoretical calculations and experimental measurement of relative variances in trace metals concentrations at each stage of both sampling procedures. The ASP significantly increased the reproducibility of soil sampling compared to the GSP. Larger relative variances occurred during field sampling for the ASP and during laboratory sampling for the GSP. They were due to subsample mass reduction without control over particle size. Relative theoretical and experimental variances were in agreement. However, large discrepancies were observed for all sampling stages of both procedures between absolute theoretical and experimental relative variances. In the case of Pb, theoretical calculations were closer to experimental measurements when using a calculated value of the liberation factor (l) based on mineralogical data rather than l = 1. It was shown that the b-exponent had a large influence on theoretical variances. Increasing the estimate of b from 0.5 to 1 largely improved the agreement between theory and experiment. Finally, 99% of experimental relative variance was explained by sampling errors compared to analytical errors.

► Gy's sampling theory is used for the analysis of representativeness in contaminated soil sampling.
► A theory-based sampling procedure is compared to grab sampling.
► Representativeness is significantly less in grab sampling.
► 99% of representativeness loss is due to sampling errors.
► Experimental and theoretical agreement is limited by uncertainties in theoretical parameters.