# C.3.2.1 DU 1

This DU was a 105 × 105 foot square that comprised one-quarter of an acre. It was investigated with three different sampling approaches:

- A grid was placed on the site with each grid cell being 17.5 × 21 feet such that there were 30 cells covering the site. A systematic random sampling approach was used to collect ISM samples composed of 30 increments. Three such ISM samples were collected.
- A grid was placed on the site with each grid cell being 10.5 × 10.5 feet such that there were 100 cells covering the site. A systematic random sampling approach was used to collect ISM samples composed of 100 increments. Three such ISM samples were collected.
- Ten discrete samples were collected using simple random sampling (i.e., the locations of the 10 samples were randomly allocated across the site).

The discrete samples were collected identically to the increments; thus, the volume of the ISM samples was roughly 30 or 100 times the volume of each discrete sample. Each sample or increment was expected to be representative of the soils in the top 6 inches bgs. Data from each sampling approach were analyzed, and a 95% UCL was calculated for each. For the discrete samples, a 95% UCL can be collected directly from the set of *n* observations; it is not necessary to repeat the discrete sampling protocol multiple times to calculate a 95% UCL. For the ISM approach, a 95% UCL can be calculated because three replicate ISM samples (each based on 30 or 100 increments) were collected. As explained in Section 4 and Appendix A, while both discrete and ISM sampling may be expected to yield unbiased estimates of the mean for most sampling protocols, they represent different distributions with different standard deviations (SDs). Therefore, the methods can be expected to yield similar estimates of the mean but different confidence limits for the estimate of the mean. The 95% UCLs were compared to the Florida Department of Environmental Protection (FDEP) cleanup level of 2.1 mg/kg arsenic in soil to determine whether the site presents an unacceptable human health risk.

For the ISM approach with 30 increments, concentrations among the three replicates ranged 1.8–1.9 mg/kg with an arithmetic mean and SD of 1.8 and 0.08 mg/kg, respectively. For the ISM approach with 100 increments, concentrations among the three replicates were all roughly 1.7 mg/kg with an arithmetic mean and SD of 1.7 and 0.03 mg/kg, respectively. The 95% UCLs calculated using either Student’s-*t* or Chebyshev yielded approximately the same result (rounded to two significant figures). The 95% UCLs were 2.0 and 1.8 mg/kg for the 30- and 100-increment samples, respectively. Since the upper-bound estimates of the mean are both below the action level of 2.1 mg/kg, either ISM sampling design would have provided evidence that arsenic at this site does not pose an unacceptable risk and that the site could be left in its current condition for the impending residential development.

For the n = 10 discrete samples collected from DU 1, arsenic concentrations ranged 0.7–5.4 mg/kg with an arithmetic mean of 2.0 mg/kg, SD of 1.4 mg/kg, and coefficient of variation of 0.7, which indicates that the data exhibit low skew. The data are not normally distributed, so a bootstrap technique was used to calculate the UCL. The 95% UCL using a bias-corrected accelerated bootstrap is 3.0 mg/kg. That level is above the threshold of interest and is considered an indication that the arsenic in soil at this site might cause an unacceptable risk for residents.

It is interesting that the data collected via discrete samples and the data collected via ISM lead to different results for this DU. In one case, the data show no unacceptable human health risk due to arsenic at this DU. In the other case, the data show that there is, indeed, an unacceptable risk due to arsenic at this site. In addition, for the ISM approach, the decision to collect three replicates allowed for an evaluation of the confidence in the estimate of the mean. Since all of the individual ISM results were within approximately 10%–20% of the action limit, any single result may have introduced uncertainty about the level of protectiveness of the risk assessment. Demonstrating that three individual ISM results and the corresponding 95% UCL are all below the action level provides stronger evidence that arsenic does not pose an unacceptable risk for DU 1. The ISM samples that are based on 90 (3 × 30 increments) or 300 (3 × 100 increments) sample locations achieve better spatial coverage of the site than the 10 discrete samples, but both types of sampling approaches yield an unbiased estimate of the mean. It is important to recognize that any of these sampling approaches might be considered reasonable for this site, yet they lead to different conclusions and may even lead to different decisions regarding the need for remediation.