2.5.5 Device and Preparation Errors
Source: Gerlach and Nocerino 2003.
Delimitation error (DE) is a result of using an incorrect shape for the sampling device that removes each increment from the population or the incorrect use of a correct sampling device. For example, an incorrectly shaped sampling tool biases the grain sizes included in that sample. A sampling tool should be of a shape and size so that every fragment of the population of interest has an equal probability of being included in the sample. This error is a common source of bias in environmental samples, both in the field and in the laboratory. Figure 2-12 illustrates that, depending on the sample device, some particles have a greater chance of being included in the sample/subsample than others. The sampling interval depicted in Figure 2-12 has a higher proportion of larger particles at the bottom of the interval. This might be the case, for example, in an in situ soil scenario. On the other hand, this particle distribution pattern might be reversed, for example, in the case of soil jars in the laboratory. Subsampling with the rounded scoop preferentially gathers particles from the top, which tend to be the larger particles when stratification occurs as the sample is arranged in a “slabcake” shape in preparation for subsampling. With its narrower bottom width, a rounded scoop discriminates against the particles at the bottom of the sampling interval, which tends to be the smaller sizes in many if not all subsampling scenarios. By design, the rectangular scoop tool is a more inclusive tool and gathers particles of various grain sizes consistently throughout the sampling interval. In Gy theory, a sampling tool that promotes DE is termed "incorrect”; one that reduces DE is called “correct.”
Extraction error (EE) also results from the use of incorrect sampling devices. Unlike DE, which is only a function of the shape of the sampling device, EE is a function of the sizes of both the tool and the soil particles and the correct use of the sampling device. This error occurs because an inappropriate sampling device can bias the fragments that are included or excluded from being captured by the device. This scenario often plays out when the sampling device is too small and the cutting edge of the tool pushes all or certain particles (e.g., larger sized particles) aside rather than including them in the sample. EE is also a common source of bias in both the field and in the laboratory. An EE that commonly occurs in the field is when full recovery of the core is not attained when using a split-spoon or direct push-sampler. Figure 2-13 shows a sampling device that gives all particles an equal chance of being included in the sample, depending on where the center of gravity lies with respect to the cutting edge of the device. As the sampling device is used, particles are included or excluded with equal probability, thus reducing EE. To reduce EE, a correct sampling device should have a “mouth” size at least three times the size of the largest particle (Gerlach and Nocerino 2003).
Figure 2-14 illustrates both DE and EE delimitation and extraction errors A volume of soil is depicted in a two-dimensional (2-D) plane with larger particles concentrated at the bottom. Coring Device A minimizes the DE and EE because it samples the full thickness of the material and does not discriminate against the larger particle sizes. Coring Device B demonstrates EE and is an incorrect device for this matrix. Its mouth is too small to include larger particles. Coring Device C (the shovel) demonstrates DE because the sample profile it delimits cannot sample the full thickness of the DU. Coring Device D also illustrates DE because the delimited sample profile does not encompass the full shape of the DU in the vertical plane.
Figure 2-13. Vertical view of a sampling device that minimizes EE.
Figure 2-14. Increment DE and EE from sampling device selection.
Increment DE and EE are controlled by the proper use of correct sampling tools.
Preparation error (PE) is the sum of errors introduced by analyte loss, cross-contamination, or chemical or physical alteration of the sample that biases sample results relative to the true mean. Some of these errors are controlled by traditional QA/QC procedures such as sample preservation, holding times, and blanks.