Investigation-Scale Evaluation of Multi-Incremental Sampling Methodology

A series of discrete and multi-incremental (MI) samples were collected from a former scrap metal dump (SMD) located on Kure Atoll, Northwest Hawaiian Islands. The 170-foot by 80-foot dump area was known from previous investigations to be contaminated with polychlorinated biphenyls (PCBs) from historic disposal of trans- formers and other electrical equipment. A 10-foot spaced sampling grid was established within the SMD. A total of 396 discrete soil samples were analyzed from three decision units (DUs) established at three depths (0-4 inches, 28-36 inches, and 48-60 inches below ground surface (bgs)). The mean PCB concentration measured in the 10-foot grid samples collected from the shallow, intermediate and deep DUs were 0.47, 0.84 and 31.9 milligrams per kilogram (mg/kg) respectively. Much of the contamination in the deeper portions of the SMD is present in a localized spill area that contains isolated "nuggets" of soil with very high levels of PCBs. Splits of the 121 to151 discrete samples collected from within each designated DU were combined as individual increments to produce a single MI sample for each DU. Because splits of each sample point used to create the MI samples were analyzed sepa- rately as a discrete sample in the field, the resulting analytical data set allows a good comparison of discrete versus MI sample data and the advantages and limitations of each. Analysis of the data showed that the MI samples accurately reflected the mean PCB con- centration within the three DUs of the SMD despite the heterogeneous, log-normal distribution of PCBs documented by the discrete sample data. A Monte-Carlo analysis was conducted on the field PCB data collected to simulate the range of PCB concentrations that would have been determined for the SMD using a traditional Remedial Investigation (RI) approach involving the collection of eight, dis- crete soil samples. This analysis found that in the deep soils, the standard RI approach would have yielded a false negative result (i.e. underestimated the representative concen- tration of PCBs for this DU) with respect to action levels published by the Hawai'i Department of Health (HDOH) about 44% of the time, despite the "true" mean PCB concentration (31.9 mg/kg) being significantly higher than the 1.1 mg/kg PCB action level. This analysis also found that 7% to 15% of the time the RI sampling approach would not have detected PCBs at levels above the field analysis reporting limit (0.25 mg/kg) in the DU soils. These false negative results obtained from the RI approach may have eliminated the rationale to conduct additional characterization sampling of the contaminated SMD site.