Scientific Investigations Report 2006–5232

**U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2006–5232**

A quantitative evaluation was initiated to determine if water-quality
data from samples collected after purging 1 and 3 wellbore volumes of water
were statistically equivalent at a 95-percent confidence level. Water samples
were collected from 17 INL wells with long purge times (figs.
1, 2). One quality assurance replicate sample
also was collected for evaluation. Equation 1 was used to evaluate the statistical
equivalence of 337 constituent pairs using (1) standard deviations reported
by the analytical laboratory for radiochemical measurements, and (2) estimated
standard deviations for non-radioactive constituents. Regression equations used
to calculate the *F-pseudosigma* estimates of standard deviation are
available at http://bqs.usgs.gov/bsp/regress.htm.
Equation 1 also was used to re-evaluate 63 constituent pairs generated from
water samples collected at 11 wells (figs. 1,
2) by Bartholomay (1993). Bartholomay collected samples from 10 wells during
spring and autumn and at the TRA DISP well (fig. 1)
during autumn after purging 1, 2, and 3 wellbore volumes from the wells. Analytical
results for the samples, standard deviations, and calculated Z-values are provided
in tables 1-4. If the Z-value
was less than 1.96, the sample pair was considered equivalent at the 95-percent
confidence level.

Samples collected for this study after purging 1 and 3 wellbore volumes of water from a well were considered a sample pair. Sample pairs from the 17 wells along with the quality assurance replicate sample from USGS 22 (fig. 1, tables 1-3) generated 337 constituent pairs for evaluation using equation 1. Z-values for 331 pairs (98.2 percent) were less than 1.96 and were considered equivalent (tables 1-3).

Samples collected by Bartholomay (1993) after purging 1, 2, and 3 wellbore volumes were analyzed for tritium and strontium-90 and the results were compared using a statistical technique derived from a test method recommended by the American Society for Testing and Materials (1988). To consider the Bartholomay (1993) results in this study, equation 1 was used to re-evaluate his tritium and strontium-90 data. Samples collected by Bartholomay (1993) after purging 1 and 3, 1 and 2, and 2 and 3 wellbore volumes of water from a well were considered sample pairs. The 63 sample pairs from the 11 wells generated 126 constituent pairs for evaluation. Z-values for 120 pairs (95.2 percent) were less than 1.96 and were considered equivalent (table 4). A further breakdown of data from Bartholomay (1993) indicates that statistical equivalence is slightly stronger in autumn (97 percent) than in spring (93.3 percent). Additionally, agreement is slightly better between 1 and 2 wellbore volumes (97.6 percent) than between 1 and 3 wellbore volumes (95.2 percent) or between 2 and 3 wellbore volumes (92.8 percent).

When all 1 and 3 wellbore volume constituent pairs were considered together, 371 of 379 pairs (97.9 percent) were statistically equivalent. When the 1 and 2 or the 2 and 3 wellbore volume constituent pairs were included, 451 of 463 (97.4 percent) were statistically equivalent.

Because of the good quantitative agreement between constituent pairs, there is no indication that changing from purging 3 wellbore volumes of water to purging 1 wellbore volume of water effected comparability of water-quality data. On the basis of a smaller data set from Bartholomay (1993), it appears that purging 2 wellbore volumes also is an acceptable practice for data comparability.