LITMIR.BIZ

      Figure 2.7. Field labeling of samples: bags with tags (left) and close-up of tag (right).

      Many soil samples will contain very low concentrations of munitions constituents. Cross contamination between samples from potentially high concentration DUs with samples known to be from areas of low potential contamination is very possible. A sample must be double-bagged in the field after documentation and stored separately during transportation and when in the lab. This precaution significantly improves the sample quality by reducing the variance between sample replicates.

      Using the correct sample preparation techniques is also critical in maintaining sample quality [26, 27]. Subsampling of the DU sample must be done carefully. A large source of error occurs during comminution, the process used to reduce the variance in particle sizes in the sample, which allows more uniform distribution of the analytes within the total mass of the sample. Diminution typically is conducted with a grinder. Puck grinders, used in the mining industry, have been found to achieve the best results based on lab, field and project studies. It is important to follow the correct grinding procedures, as the analyte can be lost during grinding, as was found when using a ball mill. Subsampling should be done in the lab following grinding, never in the field using sample splitting. The process must ensure that the subsamples represent the sample, much as the sample must represent the DU from which it was taken. If not, the subsample will be no better than a grab sample. In section 2.7 we will go into detail on field splitting, sample preparation in the processing lab, grinding to reduce sample particles to a more consistent size (comminution) and subsampling.

2.5 Sampling

      Sampling is the process of obtaining a portion of a total population that represents that population as a whole. That portion is called the sample. Samples must be representative of the area as a whole from which they are collected or they will be meaningless, representing only themselves. How is a representative sample collected? This section will guide the reader through the process required to ensure that the samples collected will fulfill the objectives that the sample is required to meet. Subsampling, the process of obtaining a portion of a collected sample that represents the sample as a whole and thus the total population, will be discussed in section 2.7.

2.5.1 Objectives

      Before going to the field to obtain samples, a sampling plan must be created. A key element of the sampling plan is the establishment of objectives that the samples are to meet. In the US, the Environmental Protection Agency has established a method based on what it calls the ‘data quality objectives (DQO) process’ [28]. The DQO process is iterative and will evolve as the project progresses. The SQC process, described in section 2.3.1, is a more direct approach that pares the development of objectives down to its essentials. Other countries have processes similar to these. The key elements for any sampling plan are that the sample must represent the mean concentration of the analytes in the DU, the samples must be reproducible (replicate samples from the field, not subsamples in the lab), and the data resulting from the analyses of the samples must be defensible. Without a robust sampling plan, the samples collected will not fulfill the objectives of the project.

      At this point, it is worth saying something about ‘hot spots’. The concept of hot spots has been around since samples were first collected. However, the definition of a hot spot has not [29, 30]. There are no mass, distribution, or availability criteria associated with a hot spot. Thus, hot spots are meaningless. If there is any analyte present in a DU, the highest concentration, the hottest hot spot, will be a million parts per million: a particle of pure substance (if it is present in that form). How many hot spots by this definition will occur within a DU? Many! Is there enough mass associated with these ‘hot spots’ to pose a risk to the receptor within the DU? The only way to determine this is to estimate a mean value of the analyte within the DU. Thus, the most important estimate that can be made within a DU is the mean concentration of the analyte. Hot spots are irrelevant.