LOW-VOLUME AIR SAMPLING
Radioactivity associated with airborne particulates was monitored
continuously by 18 low-volume air samplers (two of which are used as
replicate samplers) at 16 locations during the third quarter of 2002
(Figure 1 below). Three of these samplers are located on the INEEL, nine
are situated off the INEEL near the boundary, and six have been placed at
locations distant to the INEEL. Samplers are divided into INEEL, Boundary,
and Distant groups to determine if there is a gradient of radionuclide
concentrations, increasing towards the INEEL. Each replicate sampler is
relocated every year to a new location. One replicate sampler was placed
at Arco (Boundary location) and one at Howe (Boundary location) during
2002. An average of 14,347 ft3 (406 m3) of air was
sampled at each location, each week, at an average flow rate of 1.3 ft3/min
The weekly particulate filters collected during the quarter for each location were composited and analyzed for gamma-emitting radionuclides. Composites were also analyzed by location for 90Sr, or 238Pu, 239/240Pu, and 241Am as determined by a rotating quarterly schedule.
Charcoal cartridges were analyzed for gamma-emitting radionuclides, specifically for iodine-131 (131I). Iodine-131 is of particular interest because it is produced in relatively large quantities by nuclear fission, is readily accumulated in human and animal thyroids, and has a half-life of eight days. This means that any elevated level of 131I in the environment could be from a recent release of fission products.
Gross alpha results are reported in Table C-1. Median gross alpha concentrations in air for INEEL, Boundary, and Distant locations for the fourth quarter of 2002 are shown in Figure 2. The data were tested for normality prior to statistical analyses. The data produced a distribution that was strongly right skewed. Taking the natural log of skewed data is often done to evaluate if the data may be log-normally distributed. Using this technique the data were shown to be not log-normally distributed either. Box and whisker plots are commonly used when there is no assumed distribution. Each data group in Figure 2 is presented as a box and whisker plot, with a median, a box enclosing values between the 25th and 75th percentiles, and whiskers representing the non-outlier range. Note that outliers and extreme values are identified separately from the box and whiskers. Outliers and extreme values are atypical, infrequent, data points that are far from the middle of the data distribution. For this report, outliers are defined as values that are greater than 1.5 times the height of the box, above or below the box. Extreme values are greater than 2 times the height of the box, above or below the box. Outliers and extreme values may reflect inherent variability, may be due to errors associated with transcription or measurement, or may be related to other anomalies. A careful review of the data collected during the fourth quarter indicates that the outliers and extreme values were associated with samples from the week of December 10, 2002. Additional investigation concluded that the values were not due to mistakes in collection, analysis, or reporting procedures, but rather reflect natural variability in the measurements; in this instance the results were influenced by a period of inversions over the collection week. The outliers and extreme values lie within the range of measurements made within the past five years. Thus, rather than dismissing the outliers, they were included in the subsequent statistical analyses. Further discussion of box plots may be found in Determining Statistical Differences under Helpful Information.
Figure 2 graphically shows that the gross alpha measurements made at INEEL, Boundary, and Distant locations are similar for the fourth quarter. If the INEEL were a significant source of offsite contamination, concentrations of contaminants should be statistically greater at Boundary locations than at Distant locations. Because there is no discernable distribution of the data, the nonparametric Kruskal-Wallis test of multiple independent groups was used to test for statistical differences between INEEL, Boundary, and Distant locations. The use of nonparametric tests, such as Kruskal-Wallis, gives less weight to outliers and extreme values thus allowing a more appropriate comparison of data groups. A statistically significant difference exists between data groups if the (p) value is less than 0.05. Values greater than 0.05 translate into a 95 percent confidence that the medians are statistically the same. The p-value for each comparison is shown in Table D-1. There were no statistical differences in gross alpha concentrations between groups for the fourth quarter. The outliers and extreme values for each location group are associated with a single week (December 10, 2002) that experienced multiple atmospheric inversions during the collection week.
Comparisons of gross alpha concentrations were made for each month of the quarter (Figures 3 – 5). Again the Kruskal-Wallis test of multiple independent groups was used to determine if statistical differences exist between INEEL, Boundary, and Distant data groups. There were no statistical differences in gross alpha between groups for any month (Table D-1).
As a further check, comparisons between gross alpha concentrations measured at Boundary and Distant locations were made on a weekly basis. The Mann-Whitney U test was used to compare the Boundary and Distant data because it is the most powerful nonparametric alternative to the t-test for independent samples. INEEL sample results were not included in this analysis because the onsite data, collected at only three locations, are not representative of the entire INEEL and would not aid in determining offsite impacts. No statistical differences were noted between Boundary and Distant locations for the any week of the quarter (Table D-2). More detail on the statistical tests used can be found in Determining Statistical Differences under Helpful Information.
Gross beta results are presented in Table C-1. Gross beta concentrations in air for INEEL, Boundary, and Distant locations for the fourth quarter of 2002 are shown in Figure 6. The data were tested and also found to be neither normally nor log-normally distributed. Box and whiskers plots were used for presentation of the data. Outliers and extreme values were retained in subsequent statistical analyses because they are within the range of measurements made in the past five years, and because these values could not be attributed to mistakes in collection, analysis, or reporting procedures. As in the case of alpha activity, the quarterly data for each group appear to be similar and were determined using the Kruskal-Wallace test to be statistically the same (Table D-1). As with the gross alpha, the gross beta results showed outlier and extreme values are associated with the week of December 10, 2002.
Monthly median gross beta concentrations in air for each sampling group are shown in Figures 7 – 9. Statistical data are presented in Table D-1. There were no statistical differences in gross beta between groups for any month during the quarter (Table D-1).
Comparison of weekly Boundary and Distant data sets, using the Mann Whitney U test, indicates a difference between the two location groups for four weeks (November 13th and 20th, and December 3rd and 17th) (Table D-2). The Boundary group was statistically greater than the Distant group in all cases. Significant statistical differences were found between stations of both the Boundary or Distant location groups for the weeks evaluated. The largest differences were the result of high gross beta readings of the northern Boundary locations Howe, Howe Q/A, Monteview, and Mud Lake. These stations, especially Mud Lake, often tend to have higher results than the other stations. This is primarily related to wind born particulates derived from the surrounding agricultural fields that have been recently harvested. In addition to this potential source, the week prior to collection on December 3 was strongly influenced by inversions in this portion of the Snake River Plain.
Iodine-131 was not detected in any of the analyzed batches of charcoal cartridges above the respective 2s value. Weekly 131I results for each location, including individual recount data, are listed in Table C-2 of Appendix C.
Weekly filters for the fourth quarter of 2002 were composited by location and analyzed for gamma-emitting radionuclides, including 137Cs. Composites were also analyzed for 90Sr, 238Pu, 239/240Pu, and 241Am. Plutonium-238 was not detected in any sample. Samples collected from seven air monitoring stations (EFS, FAA Tower, Howe, Howe Q/A-2, Idaho Falls, Monteview, and Van Buren Avenue) showed 239/240Pu results greater than their related 2s values. The Howe, Idaho Falls, and Van Buren Gate samples also had concentrations of 241Am above their 2s values. No 137Cs was measured above the 2s value for any sample in the fourth quarter 2002.
Strontium-90 was also measured slightly above the 2s value in the Mud Lake sample.
Figure 10 below shows concentrations of 241Am and 239/240Pu detected above the 2s concentrations as patterned. Occasional detection of human-made radionuclides is not unusual and represents natural variations in concentrations of radionuclides introduced by historical nuclear weapons testing. The concentrations measured during this quarter are consistent with those recorded in the past. All results were far less than their respective DCGs. All results for composite filter samples are shown in Table C-3, Appendix C.
Figure 10. Americium-241 and Plutonium-239/240 detected in quarterly composite air filters (by locations).