5.3 Field Collection

5.3.1 Surface ISM Samples

ISM samples are composed of increments collected from specific points throughout the DU. The positioning of the collection points can be set using one of three approaches, as described in Section simple random sampling (SRS), random sampling within a grid, and systematic random sampling. SRS involves determining random locations across the entire DU (see Figure 4-9). Note that “random” in this context does not mean wherever the sampling team feels like taking a sample and that a formal approach to determining the random increment locations must be used. With random sampling within a grid, the DU is overlain with a sampling grid and soil increments are collected from random locations determined in each grid cell (see Figure 4‑8). Systematic random sampling is similar except that only the initial grid cell sampling location is randomly determined and the same relative location is sampled in each of the other grid cells (see Figure 4‑7).

As predicted by statistical sampling theory and demonstrated by the ISM simulations discussed in Section and Appendix A.1, SRS yields the most representative (least biased) estimate of the mean. However, it is also the least practical to implement since field staff have to navigate to predetermined locations nonuniformly positioned within the DU. SRS also may result in a sampling pattern that leaves large portions of a DU unsampled, which may not be acceptable to regulators, risk managers, members of the public, or other stakeholders. In practice, systematic random sampling is most often chosen for ease of implementation and to avoid the appearance of over- or underrepresentation of subareas within a DU, as may occur with SRS. Refer to Superfund Representative Sampling Guidance, Vol. 1 (USEPA 1995b) for additional information.

Incremental soil samples are prepared by collecting multiple increments of soil (typically 30 or more) from a specified DU and physically combining these increments into a single sample.

Incremental soil samples are prepared by collecting multiple increments of soil (typically 30 or more) from a specified DU and physically combining these increments into a single sample, referred to as the “incremental sample.” When the individual increment mass is adequate, this number of increments (n) generally results in a soil sample with a contaminant concentration representative of the estimate of the mean contaminant level within a DU (i.e., a representative sample). That is, even when the distribution of individual data points (i.e., discrete sample results) is nonnormal, the distribution of sets of means from the population will approach a normal or Gaussian shape as the number of increments (n) increases (Jenkins et al. 2005). See Section 4 of this document on the statistical basis of ISM for a more detailed discussion of increment number(s), adequate increment mass, and representativeness.

As sampling theory indicates, the number of increments collected depends on the amount of distributional heterogeneity present within the DU for the constituent of interest. A variety of factors may influence the amount of distributional heterogeneity within a DU. These include, but are not limited to, the following:

  • contaminant type and physical characteristics
  • soil type and physical characteristics
  • contaminant release mechanism (e.g., spill, area-wide application, munitions range)
  • others

As the DU gets significantly larger, the amount of distributional heterogeneity may increase. In these cases, depending on site specifics, CSM, and DQOs, it may be necessary to increase the number of increments per DU to 50 or more. Collection of a greater number of increments in each DU typically reduces the GSE (i.e., minimizes the variation among replicate samples). Alternatively, splitting larger DUs into two or more smaller DUs should be considered. It is not normally necessary to increase the number of increments unless there is reason to believe the DU has more distributional heterogeneity than can be controlled with 30–50 increments. See Section of this document for the statistical information and evaluation of the number of increments for ISM sampling.

In general, a minimum of 30–50 increments is sufficient for most DUs. However, in published reports for solid/particulate-type chemicals of concern (COCs) (e.g., energetics/explosives, particulate metals, etc.) 50–100 increments per DU have been collected. USEPA SW-846 Method 8330B recommends collecting 30 or more evenly spaced increments to build a sample with a total mass of >1 kg. It is anticipated that as ISM matures, additional information on the optimal number of increments for other types of contaminants may become more readily available. The number of increments to be collected from each DU of a site investigation should be evaluated during systematic planning as part of the DQO process and documented in the sampling and analysis plan (SAP).

Generally, a minimum of 30 increments should be collected for each DU, with each increment weighing 20–60 g.

In general, individual soil increments typically weigh 20–60 g. Final ISM field samples typically weigh 500–2500 g. To minimize FE to an acceptable level, it may be necessary to calculate the target bulk ISM sample mass for collection prior to field implementation and ISM collection (see Section 2 Hyperlink 14 and Hyperlink 18) (Pitard 1993, Ingamells and Pitard 1986). It may be necessary to collect bulk ISM samples >2500 g to reduce FE to an acceptable level. Additionally, note that sieving of soil samples to the <2 mm particle size reduces the amount of soil mass available for preparation and analysis, so this fact needs to be taken into consideration during systematic planning if minimizing FE is a DQO. Additionally, sieving is not applicable for the collection of VOC samples (see Section 5.4.2). Based on the required final mass of the ISM sample, as dictated by FE considerations and the number of increments determined by distributional heterogeneity, the minimum mass of the individual increments can be calculated. The number of increments to be collected per DU, the sampling depth, and the targeted mass should all be specified in the sampling and analysis plan. The number of increments to be collected per DU, the sampling depth, and the targeted mass should all be specified in the SAP. The mass of any single increment depends on the depth of interest, soil density, moisture content, and the diameter or size of the sample collection tool. Typically, the mass of the final ISM sample is sufficient for the planned analyses, any additional QC requirements, or repeat analyses due to unanticipated field, laboratory, and/or QC failures. The number of increments to be collected per DU, the sampling depth, and the targeted mass of each sample should all be specified in the sampling plan as described in the following formula for estimating sampling equipment requirements based on a predetermined ISM mass and number of increments:



targeted mass of sample (g)

Ds increment length (cm)
n number of increments

soil or sediment density (g/cm3)

Ɵ diameter of sample core (cm)

These parameters, along with the density of the soil or sediment matrix, assist in the selection of the sampling tool to collect the appropriate individual increment mass for the total ISM sample (Walsh 2009).

Table 5-1 (Walsh 2009) and Figure 5-3 are provided as examples for estimating increment mass that can be collected for a given sampling depth and soil density, once the DU size, number of increments and total ISM sample mass have been established. Generally, a minimum of 30 increments should be collected for each DU, with each increment weighing 20–60 g. Individual increment mass should be similar provided the soil density and DU thickness are fairly uniform. Typically, however, individual increments are not weighed in the field during collection. Similar mass per increment is assumed with similar volume collected. Due to practical limitations, increments of similar volume rather than of similar mass are collected, provided that the thickness of the DU is fairly uniform. For DUs of nonuniform thickness, the available thickness at each increment location is collected to ensure spatial coverage and the increment is not required to have similar volume or mass.

Table 5-1. Estimated sample mass for set increment length and substrate density

Corer diameter
Number of increments to obtain desired ISM sample mass
500 g 500 g 1000 g 1500 g 2000 g
Soil density 1.6 g/cm3, increment length 2.5 cm
2.0 40 60 80 119 159
3.0 18 27 35 53 71
4.0 10 15 20 30 40
Soil density 1.8 g/cm3, increment length 2.5 cm
2.0 35 53 71 106 141
3.0 16 24 31 47 63
4.0 9 13 18 28 35

Substrate density may vary from 0.75 g/cm3 (for Loess) to 1.81 (for Gravely Sand) with substrate densities typically ranging 1.6–1.8 g/cm3 (Walton 1988, Domenico and Schwartz 1990).

Estimated sample mass based on number of increments for set increment and substrate density.

Figure 5-3. Estimated sample mass based on number of increments for set increment and substrate density.

Click here to download the Excel File for the Incremental Soil Mass Calculator
(Save to hard drive for full functionality)

If replicate ISM results indicate data variability is too high (i.e., interferes with decision making for the DU), additional data evaluation, sample analysis, and/or resampling may be required to achieve project-specific DQOs. Note, however, that high variability between ISM replicates may also be a result of laboratory processing and subsampling procedures, which can be evaluated by examining the results of laboratory replicates (if analyzed). High data variability determined to be a result of DU heterogeneity and/or field sampling error may require revision(s) to the ISM design and implementation, including DU modification, additional increments, and/or increased increment mass (see Section 5.3.5).

Soil density across the DU should be similar. Contaminant distribution within different soil types should also be considered when determining DUs.

Soil density across the DU should be reasonably uniform (e.g., the same general soil classification can be expected throughout the DU). When the surface of the DU contains both vegetated and nonvegetated areas, it is very likely that less soil (less increment mass) will be obtained from the vegetated regions within the DU. If a site has obvious areas with different soil lithologies and/or densities (e.g., areas of sand with areas of fat clay, areas of peat, etc.), those different soil type areas should be factored into DU determinations (i.e., location, shape, size of DUs). Assumed differences in contaminant concentrations in the different soil types should also be considered. In these cases, it may be necessary to redefine the DU to account for the possible heterogeneity of contaminant concentration.

For surface/exposed soil, common sampling depths are 2.5, 5, 10, or 15 cm; however, depths can be greater depending on the DQOs and CSM, including expected vertical distribution of the chemical of potential concern (COPCs) (due to infiltration, buried utilities/facilities, stockpiles, etc.), the exposure scenario, and/or regulatory requirements. Additional depths and/or DUs may be required for vertical delineation. Contaminant dilution should also be considered when determining increment depth. For surface-deposited energetics at active U.S. Department of Defense (DOD) training ranges, soil profile samples have shown 1–2 orders of magnitude decreasing concentrations within the top 10 cm (USEPA 2006c). For these types of sites, the desired sampling depth is approximately 2 cm, based on research conducted at Cold Regions Research and Engineering Laboratory (CRREL); greater increment depths result in dilution of the contaminant concentration. In general, the location, lateral extent, and depth of the DU should be selected to represent an area of known or expected similarity. For greater depths, use of a smaller-diameter sampling tool may be desirable but often is impractical due to presence of pebbles, rocks, and vegetation. In general, however, the smallest diameter sampling tool applicable to particle size requirements is recommended to minimize delimitation and extraction error and to attain the necessary soil mass (see Section 5.2). Alternative technologies as appropriate for site-specific conditions should be considered, as appropriate.

A square, rectangular, circular, or other naturally or structurally defined DU (e.g., 5 m perimeter around the exterior of a building) is first subdivided or gridded-off into uniform cells or subareas based on the desired number of increments to be obtained. That is, the number of cells is equivalent to the number of increments. Using the systematic random design, a random position is established for a given cell, and then the same position is repeated in all of the remaining cells in the DU. For the random sampling within grids design, a random position is designated and sampled in each cell. A random starting point or random position for each cell can be obtained with dice or a random number generator. The process is repeated for replicate samples: i.e., a new random position is established for the single collection point to be repeated in all of the cells, or for each cell, depending on the sampling design. A Global Positioning System (GPS) device should be used to delineate the DU. It may or may not be necessary to determine the exact location of each increment depending on the DQOs specified during the systematic planning process.

Figure 5-4. Example DUs from industrial (A), residential (B), and agricultural (C) sites

Depending on the size of the DU and terrain features, placement of markers (e.g., pin flags and posts) at the corners and or edges can assist with a visual delineation of the cells or subareas where increments are to be collected. That is, the markers can define lanes, grids, or collection points. When DUs are square or rectangular, the conversions for the spacing (steps) between increment collection points (cells) are fairly straightforward to calculate. For example, a square-shaped DU could be divided into five rows, with six increments collected from each row in a systematic random fashion, with an initial random starting point. For more rectangular-shaped DUs, fewer rows might be used with more increments per row collected (Figure 5‑4). Row lengths and increments per row may be modified as needed for odd-shaped DUs. However, with other shapes, it is recommended that the perimeter be marked and flags be prepositioned across the DU in one or more perpendicular lines. Then a trial run with no sample collection is performed to quickly establish the distance between increment collection points to achieve the desired number of increments, while using the flags as guides that were positioned within or around the DU.

Although ISM sample collection may be performed by a single individual, a two-person team is often the most efficient method: ideally one person collects the increments, and the other holds the sample container (e.g., clean polyethylene bag) and keeps track of the number of increments. However, site conditions may dictate that three or more individuals are required for the collection of a single ISM sample. The User’s Manual for the CRREL Multi-Increment Sampling Tool (Walsh 2009) lists common sampling supplies and vendors that would be appropriate for SVOCs and metals. Sampling tools are set for the appropriate depth. Flags may be used to mark DU boundaries and to aid in visualizing the travel paths and/or to mark the actual increment locations. The ISM sampler starts in one corner or end of the DU and collects an increment at the predetermined positions. For the systematic random sampling design, the location of the first increment is determined randomly, and subsequent increments are collected in the same relative location within each grid, resulting in a serpentine collection pattern ending at the opposite corner or end of the DU from where sampling was started (see Figure 5-5). Note that, for simplicity, Figure 5-5B depicts collection of duplicate ISM samples rather than the recommended triplicates. Additional guidance on ISM can be found in the following documents:

  • Method 8330B, Appendix A (USEPA 2006c)
  • Protocols for Collection of Surface Soil Samples at Military Training and Testing Ranges for the Characterization of Munitions Constituents (Hewitt et al. 2007)
  • User’s Manual for the CRREL Multi-Increment Sampling Tool (Walsh 2009)
  • Technical Guidance Manual (HDOH 2008b)
  • Implementation of Incremental Sampling (IS) of Soil for the Military Munitions Response Program, Interim Guidance Document (IGD) 09-02 (USACE 2009)

Illustrations of systematic random incremental sampling pattern used for collecting samples in square (A) and circular areas (B).

Figure 5-5. Illustrations of systematic random incremental sampling pattern used for collecting samples in square (A) and circular areas (B).