Global Approaches to Environmental Management on Military Training Ranges - Tracey Temple

Скачать книгу

steps of the process are controlled to the greatest extent feasible (figure 2.2). The exception is if no effort is put into reducing sample preparation error, which is unlikely in an ideal world as it is much easier to reduce this error than sampling error. The problem with this assumption is if the sample preparation is done separately from the sample collection, as often happens as different contractors typically handle these two processes. The effort put into minimizing sampling error should not be perceived as excessive, but sample preparation is also critical to achieving reproducible data.

Case Error Source GEE Error fromsampling
Sampling Sample prep Analytical
1 1000% 300% 20% 1044% 92%
2 1000% 300% 3% 1044% 92%
3 1000% 5% 20% 1000% >99%
4 1000% 5% 3% 1000% >99%
5 25% 300% 20% 302% 1%
6 25% 5% 20% 32% 61%
7 25% 5% 3% 26% 95%
image

      Figure 2.1. Graphical representation of results for sample error reduction efforts.

image

      Figure 2.2. Total error distribution when error sources are controlled to currently achievable levels. Global estimation error (GEE) is 26% (Case 7, table 2.2).

      There are other sources of error associated with sampling and characterization. A few of these will be discussed here.

      Materialization error, a component of sampling error, relates to how a sample is collected. It is associated with the tools used for the collection of the sample. If the tool is not large enough to collect all particle sizes (delimitation error) or difficulty is encountered in obtaining or extracting the material from the sampling device (extraction error), the consistency of the sample increments cannot be maintained and the quality of the sample will be degraded. Consistency of the shape of the increment throughout its depth is also very important, as is the uniformity of the increments that make up a sample (figure 2.3). Munitions constituents such as energetics and metals tend to initially accumulate on the surface of ranges. A coring tool that samples at a fixed diameter and can be set for a pre-determined depth will provide reasonably consistent increments. Sampling tools such as spoons and trowels will skew the sample data by overemphasizing the very top of the sample and not consistently reaching the target depth. Thus, they should only be used in circumstances where a coring tool will not work, such as in very loose sandy soils or soils with stones larger than the coring bit diameter. The increment geometry must allow the concentration of the contaminant to be represented consistently throughout the depth of the increment.

image

      Figure 2.3. Increment shapes from different tools: (a) corer, (b) trowel and (c) spoon.

      Sample handling can also affect the sample quality. Temperature, holding times, cross contamination and sample preservation are all factors that need to be controlled when handling samples. Sample processing errors associated with activities such as field splitting of samples, sample drying, comminution and subsampling need to be controlled as well as quantified through lab QA procedures. Sample and subsample extraction can be classified as a sample preparation procedure or an analytical lab procedure. In either case, QA also needs to be performed to ensure no loss of the analyte in the subsample. Finally, data analysis errors may also occur. Any of these errors may be fatal to the quality of the sample and the characterization effort, reinforcing the need to be vigilant during every step of the process. Figure 2.4 illustrates these errors.

image

      Figure 2.4. Error sources associated with a land-based DU characterization.

      The global estimation error can be minimized (but not eliminated) using practices derived from sampling theory. GSE can be addressed by taking many increments from the area being sampled (DU) and combining them into a single sample. Each increment should represent a similar amount of area within the DU to best characterize the complete area being sampled. The increments should also be uniform in mass. FSE can be addressed by sampling a sufficient mass to overcome the differences in composition in the particles collected in the samples. It is important that each increment mass, depth and surface area sampled be as close and uniform as possible to ensure that each increment can be considered equal.

      Research at CRREL and Envirostat over the last 30 years has demonstrated what is required to minimize error associated with characterizing surface soils for energetics’ residues. We have developed the Multi-Increment® sampling (MIS) method for the environmental characterization of surface soils for heterogeneously-distributed munitions constituents. Sample increments are collected in a systematic fashion starting from a random location in one corner or location of the DU (figure 2.5). Reproducibility between samples has increased dramatically since MIS was implemented. Whereas prior sampling efforts resulted in proximate discrete samples varying by three orders of magnitude or more in a DU, using MIS has reduced variability to less than one order of magnitude, sometimes less than a factor of two between replicates. Whole population sample collection of high-order detonation residues on ice was compared to MIS of post-detonation residues on snow for the same operation and munition with no significant difference between the data sets [25].

image

      Figure 2.5. A replicate random-systematic sampling pattern in a square decision unit (left). Map of non-square detonation DUs from field research on munitions (right).

      Using specially designed tools, DU surface soils can be characterized quickly, efficiently,

Скачать книгу