Index > 3 Characterisation of radioactively contaminated sites >

3.10.2 Statistical tests

The statistical tests are only applied to measurements made at discrete locations. Specific details for conducting the statistical tests are given in Section 3.10. When the data clearly show that a survey unit meets or exceeds the release criterion, the result is often obvious without performing the formal statistical analysis. Table 3.52 describes examples of circumstances leading to specific conclusions based on a simple examination of the data.

Radionuclide not in background and radionuclide-specific measurements made:
Survey result Conclusion
All measurements less than DCGLW Survey unit meets release criterion
Average greater than DCGLW Survey unit does not meet release criterion
Any measurement greater than DCGLW and the average less than DCGLW Conduct Sign test and elevated measurement comparison

Radionuclide in background or radionuclide non-specific (gross) measurements made:
Survey result Conclusion
Difference between largest survey unit measurement and smallest reference area measurement is less than DCGLW Survey unit meets release criterion
Difference of survey unit average and reference area average is greater than DCGLW Survey unit does not meet release criterion
Difference between any survey unit measurement and any reference area measurement greater than DCGLW and the difference of survey unit average and reference area average is less than DCGLW Conduct WRS test and elevated measurement comparison

Table 3.52 Summary of statistical tests

Both the measurements at discrete locations and the scans are subject to the elevated measurement comparison (EMC). The result of the EMC is not conclusive as to whether the survey unit meets or exceeds the release criterion, but is a flag or trigger for further investigation. The investigation may involve taking further measurements to determine that the area and level of the elevated residual radioactivity are such that the resulting dose or risk meets the release criterion1 . The investigation should also provide adequate assurance, using the DQO process, that there are no other undiscovered areas of elevated residual radioactivity in the survey unit that might otherwise result in a dose or risk exceeding the release criterion. In some cases, this may lead to re-classifying all or part of a survey unit – unless the results of the investigation indicate that reclassification is not necessary.

The objective of compliance demonstration is to provide some level of confidence that the release criterion is not exceeded. As previously stated, 100% confidence in a decision cannot be proven because the data always contain some uncertainty. The use of statistical methods is necessary to provide a quantitative estimate of the probability that the release criterion is not exceeded at a particular site. Statistical methods provide for specifying (controlling) the probability of making decision errors and for extrapolating from a set of measurements to the entire site in a scientifically valid fashion [USNRC-2002], [EPA-1994], [Harnett], [Kume].

Clearly stating the null hypothesis is necessary before a statistical test can be performed. The null hypothesis recommended for use in EURSSEM is: ‘The residual radioactivity in the survey unit does not exceed the release criterion’. This statement directly addresses the issue of compliance demonstration for the regulator (and other stakeholders) and places the burden of proof for demonstrating compliance on the site owner or responsible party.

The statistical tests can be applied for all sites, but should be applied at sites that were subjected to a historical site assessment (HSA). At this point, the results of the historical site assessment have been reviewed and the site is determined to be impacted based on existing data and professional judgment as described in Section 2.4. An impacted site, by definition, is expected to contain areas of contamination, so this statement of the null hypothesis is reasonable for these sites.

The information needed to perform a statistical test is determined by the assumptions used to develop the test. EURSSEM recommends like MARSSIM [USNRC-2002] the use of non-parametric statistical tests because these tests use fewer assumptions, and consequently require less information to verify these assumptions. The tests described in EURSSEM (see Section 3.10) are relatively easy to understand and to implement compared to other statistical tests.

Site conditions can also affect the selection of statistical tests. The distribution of contamination is of particular concern at sites with residual radioactivity. Is the contamination distributed uniformly, or is it located in small areas of elevated activity? Is the residual radioactivity present as surface, volumetric, or subsurface contamination?

To demonstrate the use of the Radiation Site Survey Investigation Process at radiation sites, EURSSEM addresses soil and groundwater. However, EURSSEM concentrates mainly on surface soil for the final status survey to demonstrate compliance. This represents a situation that is expected to commonly occur at sites with radioactive contamination, and allows the survey design to take into account the ability to directly measure surface radioactivity using scanning techniques. Other contaminated media may be identified during the historical site assessment (HAS) or preliminary surveys (i.e., scoping, characterization, remedial action support). If other contaminated media (e.g., subsurface contamination, volumetric contamination of building materials) are identified, methodologies for demonstrating compliance other than those described in this manual may need to be evaluated or developed.
The next sections are dealing with examples how statistical data processing can be performed in the following situations:

  • Contaminant not present in background;
  • Contaminant present in background;
  • Sites with a relatively uniform distribution of contamination;
  • Small areas of elevated activity.

The examples are from final surveys as a final survey is in general the most complex and important one.

1 Rather than, or in addition to, taking further measurements the investigation may involve assessing the adequacy of the exposure pathway model used to obtain the DCGLs and area factors, and the consistency of the results obtained with the historical site assessment and the scoping, characterization and remedial action support surveys.