Return to Index

The ESER Quality Assurance Program consists of five ongoing tasks which measure:

1. method uncertainty;
2. data completeness;
3. data accuracy, using spike and laboratory control samples;
4. data precision, using split samples, duplicate samples, and recounts; and
5. the presence of contamination in samples, using blanks.

The following discussion summarizes the results of the quality assurance program for the period from October 1 to December 31, 2003.

METHOD UNCERTAINTY

The Quality Assurance Project Plan (QAPP) establishes data quality and method quality objectives for the ESER surveillance program (Stoller 2002). Since the primary concern is with detection, the lower bound for the method uncertainty is set at zero. The upper bound is established as the average maximum concentration from the past seven years of applicable data. Each individual result is checked for acceptance on the basis of the result, whether it is below the lower limit (i.e., a negative value), greater than the upper limit, or between the lower and upper limit (the most common occurrence). The calculated method uncertainty is then compared to the 1s measured uncertainty. A sample is deemed acceptable when the measured 1s uncertainty is less than the calculated uncertainty. Those results that did not meet this requirement are shown in Table 6.

Table 6.                    Analytical results determined to be unacceptable.

DATA COMPLETENESS

The QAPP specifies a 98 percent completeness goal for all regularly scheduled sample types. Data completeness was 100 percent during the fourth quarter for all sample types with the exception of low volume air which was 92 percent. The October 15 sample for the Mud Lake duplicate (Q/A-2) and the December 3 sample from Dubois were invalid due to insufficient sample volume collection. Dubois was the result of an electrical malfunction while the Mud Lake sample was due to a mechanical failure.

One of 33 ⁹⁰Sr analysis of waterfowl tissue was rejected due to a low yield.

DATA PRECISION

Data precision is a measure of the variability associated with a measurement system. Precision is measured using duplicate samples, split samples, and recounts. The QAPP specifies that sample results should agree within ±20 percent or 3s, whichever is greater. For environmental samples at levels that are within the normal range found by the ESER, the 3s criterion is the one that applies in nearly all cases. Mathematically, the 3s criterion is expressed as:

│X-Y│ < 3 × (sqrt(s_x² + s_y²)),

where:

X is the result of the regular sample
Y is the result of the duplicate sample
sx is the uncertainty of the regular sample
sy is the uncertainty of the duplicate sample

Another measure of duplicate sample results is the relative percent difference. This value is the difference in the two results divided by the mean of the two results.

Revisions to the QAPP will establish Warning and Control limits for duplicate/recount analysis. This method will evaluate the absolute difference between the duplicate/recount when the original result is below the upper bound and the standard relative percent difference when the original result if greater than or equal to the upper bound.
 

Field duplicate samples

Duplicate samples collected during the fourth quarter included milk from Blackfoot, potatoes from Taber, drinking water from Mud Lake, and surface water from Twin Falls.

Duplicate milk samples collected during the fourth quarter and analyzed for gamma-emitting radionuclides were found to be within the 3s criterion for ¹³¹I and ¹³⁷Cs.

Duplicate air samplers are operated at two locations adjacent to regular air samplers. In the fourth quarter of 2003 these samplers, designated as Q/A-1 and Q/A-2, were in operation at the Blackfoot CMS and Mud Lake, respectively. Particulate filters were analyzed for gross alpha and gross beta activity. All valid results from the duplicate samplers met the 3s criterion for gross alpha during the fourth quarter. Eleven of 12 valid samples met the 3s criterion for gross beta activity during the same period. One sample from each location did not meet this criterion and had relative percent differences of 16 percent and 17 percent, respectively.

Composite air samples from the two QA samplers were submitted for analysis at the end of the fourth quarter for gamma spectrometry at the EAL and for ⁹⁰Sr at Severn-Trent. Only ²⁴¹Am was not within the 3s criterion.

A comparison of duplicate results can also show bias in the sampling system. For example, if one set of results is consistently lower or higher than the other one might suspect that this bias was due to a leak in the system or variations in the calibration of the flow meter. Figure 14 and Figure 15 show the ratio of results (QA duplicate sampler/main sampler) over time. A ratio of one means that the results of both samplers are exactly the same. The figures show that the bias is small (<2) and not consistent, indicating that there is no obvious bias in the duplicate sampling systems. The average bias ratios during the fourth quarter are 1.0 and 1.0 for Blackfoot gross alpha and gross beta, respectively, and 0.9 and 1.0 for Mud Lake gross alpha and gross beta, respectively.

Lab split samples

The EAL splits and analyzes a number of milk, precipitation, and atmospheric moisture samples each quarter. The laboratory tests each result using both the ±20 percent criterion and the 3s criterion, although it considers the former test meaningless for analyses producing fewer than 15 total counts and questionable even where counts are on the order of 100. The latter criterion is applied in nearly all cases at the levels seen in environmental samples analyzed for the ESER program. Results of the EAL split sample analyses met the criteria for acceptance during the fourth quarter 2003.

The EAL also recounts a number of samples of each media type. The lab tests each recount using both 20 percent criterion and the 3s criterion, subject to the limitations described in the previous paragraph. All fourth quarter 2003 results were within the criteria for acceptance.

DATA ACCURACY

Accuracy is a measure of the degree to which a measured value agrees with the "true" value for a given parameter; accuracy includes elements of both bias and precision. During the fourth quarter of 2003, spikes of the following types were submitted for analyses by ESER:

• Strontium-90 in milk by Severn-Trent.

The current QAPP specifies a required accuracy of ±25 percent for ⁹⁰Sr in milk. The laboratory was within the accuracy criteria for this radionuclide. Revisions to the QAPP will require spikes to be within a certain calculated range based on the concentration of the spike added. The ⁹⁰Sr blank for the fourth quarter met also this new criterion.

Severn-Trent prepares an internal laboratory control sample (LCS) for analysis with each batch of samples submitted by the ESER. During the fourth quarter these consisted only of ⁹⁰Sr and actinides in air and waterfowl, ⁹⁰Sr in potatoes and milk, and ¹³⁷Cs in waterfowl. The QAPP specifies accuracies of ± 10 percent for ⁹⁰Sr and actinides in air, ± 20 percent for ⁹⁰Sr, ¹³⁷Cs, and actinides in waterfowl, and ± 25 percent for ⁹⁰Sr in potatoes and milk. Only the LCS for ¹³⁷Cs in waterfowl failed to meet the applicable criteria. All other waterfowl LCS results were within parameters: ⁹⁰Sr was +3.3 percent, ^239/240Pu was +11.5 percent, and ²⁴¹Am was -2.0 percent. Since the ¹³⁷Cs LCS was higher than the acceptance criteria this would indicate a positive, or high, bias to the associated samples. Since ¹³⁷Cs was not detected above the 3s value in any sample the failure of this LCS had no impact on the sample results.

The ISU EAL also prepares internal laboratory spikes. During the fourth quarter of 2003, twelve analyses were conducted on NIST-traceable standards for gamma-emitting radionuclides. Geometries tested included low-volume air filter composites, a 10-charcoal cartridge batch screen, a single charcoal cartridge screen, and a 500 mL and 1000 mL water sample. A total of 48 analytical results were generated. All of the results within the ±20 percent range.

Nineteen tritium spiked analyses were also run during the quarter. All results met the ±20 percent criterion, with the exception of one result. However, this result was within the three sigma criterion (see Data Precision section). A tritium milk spike also was tested and met criteria.

BLANKS

The ESER Program submits field blanks along with the regular samples to test for the introduction of contamination during the process of field collection, laboratory preparation, and laboratory analysis. The current low-volume air sampling program includes the use of two field blanks, designated as Blank A and Blank B, that each accompanies one of the air filter collection routes. These blank filters are also submitted as quarterly composites. After gamma spectrometry analysis, one of the blanks is analyzed for ⁹⁰Sr and the other for transuranics (²⁴¹Am, ²³⁸Pu and ^239/240Pu).

The QAPP also specifies that one milk sample blank will be submitted per year (although this is now being done monthly) and one precipitation blank for each month. The precipitation blanks are also used for atmospheric moisture samples collected during the month. Blanks for milk and gamma-emitting radionuclides were in control.

The QAPP does not specify requirements for blank performance, but ideally the result should be within ±3s of zero on most analyses. Four gross alpha and one gross beta result were greater than the ±3s criterion. For those weeks where the blank samples exceeded the ±3s criterion calls into question the validity of the results from the associated field samples. An out-of-control blank could suggest significant filter contamination or laboratory contamination. Revisions to the QAPP will detail blank acceptance criteria, again based on the upper limit and method uncertainty. Using this method three gross alpha blank measurements were deemed to be at the warning level and two were considered out-of-control. These samples correlated with those that missed the ±3s criterion. In addition to the gross alpha results six ¹³¹I results received a warning and three were deemed out-of-control.

The EAL also analyzes reagent blanks to help determine if the analysis will yield a zero result when no activity is present. Two such blanks were analyzed for tritium in water and one for tritium in milk during the fourth quarter. The results were less than the calculated MDCs or less than the 3s criterion. Severn-Trent also analyzes a laboratory blank with each sample set. The blanks were in control for ²³⁸Pu, ^239/240Pu and ²⁴¹Am in air. All ⁹⁰Sr analyses (air, milk, and waterfowl) were outside the ±3s control limit. An out-of-control laboratory blank suggests laboratory contamination or other analytical problems.

In summary the quality assurance and data quality objectives for analyses were met in the fourth quarter of 2004 with the following exceptions:

• ¹³⁷Cs and ¹³¹I in air filters/charcoal cartridges;

• ⁹⁰Sr in quarterly composites;

• Gross alpha and tritium in drinking and surface water;

• ¹³⁷Cs and ⁹⁰Sr in potatoes; and

• ⁹⁰Sr in wheat.
   

Return to Index