Multi-Incremental Sampling: greater representativeness, less variability

Two aliquots collected just a few meters apart, on the same day, can show completely different contaminant concentrations, not because of analytical error, but because soil is not homogeneous. This phenomenon is one of the main sources of uncertainty in the investigation of contaminated sites and drove the development of alternative sampling protocols, such as the one tested in this study.

The problem: discrete sampling captures points, it does not portray the area

The most widely used method in the investigation of contaminated sites is Amostragem Discreta (AD, or Discrete Sampling): an aliquot is collected at a specific point, and the result is taken as representative of the local condition. According to the ITRC, this method offers low representativeness, because a single collection point often fails to portray the actual heterogeneity of the substrate being evaluated.

This limitation pushes investigation programs in two problematic directions: increasing the density of points, which raises analytical cost, or accepting a less dense grid, with greater uncertainty. To resolve this impasse, the ITRC developed Amostragem Multi Incremental (AMI, or Multi-Incremental Sampling): dozens of aliquots are systematically collected within a Unidade de Decisão (UD, or Decision Unit) and combined into a single composite sample, estimating the average concentration of the UD with greater reliability and lower cost per square meter investigated.

Study design

The study was conducted in an area contaminated by historical disposal of Municipal Solid Waste, in Porto Alegre, RS. Two Decision Units were delimited (UD1 and UD2), each with four investigation pits, spaced 5 m apart and with an average depth of 4.50 m. In AD, one aliquot was collected per pit, four samples per UD, analyzed individually; in AMI, nine aliquots per pit (36 per UD) were homogenized, dried, ground, sieved, and resampled in the laboratory, forming a single composite sample. All samples were subjected to pH, Electrical Conductivity, and X-Ray Fluorescence (FRX/XRF) testing, the latter qualitative and semi-quantitative.

Finding 1: AMI reduced the variability between nearby points

The pH of the samples ranged between 7.19 and 9.06 across the entire area. The pattern was not uniform between UDs: between AD1 and AMI1 there was no relevant variation, but between AD2 and AMI2 the difference was significant, showing that even nearby points within the same UD can diverge markedly under discrete sampling.

Electrical Conductivity resulted in elevated concentrations in all samples, regardless of the method, a consistent indication of impact in the area.

Finding 2: not all elements respond equally to the choice of method

The FRX (XRF) identified silicon as the majority element (above 50%) in all samples, an expected result for soils in the region. Iron and aluminum appeared in intermediate amounts, and copper, zinc, and manganese in trace amounts, a set typical of soils impacted by MSW. The most relevant point lies in the element-by-element agreement between AD and AMI: calcium, zinc, and copper showed very similar results in the two methods; nickel and chromium diverged the most, evidence that the sensitivity of each element to the choice of method is not uniform.

The counterpoint: AMI is not a universal solution

The results favor AMI for estimating the average concentration of a UD with more consistency than AD, but this advantage carries technical trade-offs that must be understood before adopting the method as a default.

• AMI does not locate hotspots precisely. Because the composite sample mixes dozens of aliquots, a point of very high concentration tends to be diluted into the average of the UD; by design, AMI is not the appropriate tool for delimiting a specific point of contamination.
• The reproducibility of AMI was not statistically quantified in this study. The literature attributes greater reproducibility to the technique by reducing the weight of small-scale variability, but this study did not include field duplicates to calculate the coefficient of variation between replicates, as recommended by the ITRC; this is a step for future work, not a result already demonstrated.
• The FRX (XRF) analysis was qualitative and semi-quantitative, not quantitative at laboratory precision. This is adequate for preliminary diagnosis, but it limits the numerical precision of comparisons between methods.
• The study covers a single site and two Decision Units. It is consistent with the international literature on AMI, but it does not replace validation in other types of contamination or geological conditions before generalizing the conclusions.

What this means for contaminated-site investigations

For companies investigating areas with a history of waste disposal or diffuse metal contamination, AMI is advantageous in the preliminary diagnostic stages: a single composite sample estimates the average concentration of an entire UD at a proportionally lower cost than sampling multiple discrete points with the same coverage. The method is especially useful in human health risk assessment, which considers average exposure rather than a single point, or in broad initial screening. For precisely delimiting a contamination source, however, Discrete Sampling or high-resolution investigation remain more suitable.

Practical recommendations

1. Use AMI to estimate the average concentration of a UD, not to locate hotspots. Combine it with discrete or high-resolution investigation when the goal is to delimit a specific point.
2. Delimit the Decision Unit with technical criteria before sampling. A poorly delimited UD can dilute real contamination or aggregate areas with distinct conditions into a single result.
3. Include field duplicates in the sampling program. This makes it possible to quantify the reproducibility of the method through the coefficient of variation between replicates, following the ITRC, strengthening the technical defensibility of the report.
4. Document the number of increments per UD and the sample processing steps. Homogenization, drying, grinding, and sieving affect the representativeness of the final sample and must be recorded in the technical report.
5. Choose the method according to the objective of the investigation, not by default. Preliminary screening and risk assessment benefit from AMI; delimitation of point sources tends to require discrete sampling or high-resolution investigation.

Methodological note

This study was conducted as a single comparative case, at a site contaminated by MSW, with two Decision Units evaluated. The X-Ray Fluorescence (FRX/XRF) analysis was qualitative and semi-quantitative, adequate for preliminary diagnosis, but it does not replace reference quantitative methods for regulatory compliance. The study did not include field duplicates, so the reproducibility of AMI was not statistically quantified in this case. The conclusions are specific to soils impacted by MSW, and their generalization to other contexts should be undertaken with technical caution.

Study source: Kempfer, T.A.S.; Zapata, R.E.S.; Kieling, A.G. (2021). "Influência do Método de Amostragem na Análise de Metais em Área Contaminada por Resíduos Sólidos Urbanos (RSU)" (Influence of the Sampling Method on Metal Analysis in an Area Contaminated by Municipal Solid Waste). Presented at CIGRAC 2020+1, International Conference on the Management and Rehabilitation of Contaminated Sites, Lisbon, Portugal, May 11 to 14, 2021. Book of Abstracts, p. 46-47. ISBN 978-989-33-1802-7. (link pending validation) Study conducted by LZ Ambiental in collaboration with a partner research institution.