Evaluation of volatile organic compound (VOC) blank data and application of study reporting levels to groundwater data collected for the California GAMA Priority Basin Project, May 2004 through September 2010

Volatile organic compounds (VOCs) were analyzed in quality-control samples collected for the California Groundwater Ambient Monitoring and Assessment (GAMA) Program Priority Basin Project. From May 2004 through September 2010, a total of 2,026 groundwater samples, 211 field blanks, and 109 source-solution blanks were collected and analyzed for concentrations of 85 VOCs. Results from analyses of these field and source-solution blanks and of 2,411 laboratory instrument blanks during the same time period were used to assess the quality of data for the 2,026 groundwater samples. Eighteen VOCs were detected in field blanks or source-solution blanks: acetone, benzene, bromodichloromethane, 2-butanone, carbon disulfide, chloroform, 1,1-dichloroethene, dichloromethane, ethylbenzene, tetrachloroethene, styrene, tetrahydrofuran, toluene, trichloroethene, trichlorofluoromethane, 1,2,4-trimethylbenzene, m- and p-xylenes, and o-xylene.

The objective of the evaluation of the VOC-blank data was to determine if study reporting levels (SRLs) were needed for any of the VOCs detected in blanks to ensure the quality of the data from groundwater samples. An SRL is equivalent to a raised reporting level that is used in place of the reporting level used by the analyzing laboratory [long‑term method detection level (LT-MDL) or laboratory reporting level (LRL)] to reduce the probability of reporting false-positive detections. Evaluation of VOC-blank data was done in three stages: (1) identification of a set of representative quality‑control field blanks (QCFBs) to be used for calculation of SRLs and identification of VOCs amenable to the SRL approach, (2) evaluation of potential sources of contamination to blanks and groundwater samples by VOCs detected in field blanks, and (3) selection of appropriate SRLs from among four potential SRLs for VOCs detected in field blanks and application of those SRLs to the groundwater data. An important conclusion from this study is that to ensure the quality of the data from groundwater samples, it was necessary to apply different methods of determining SRLs from field blank data to different VOCs, rather than use the same method for all VOCs.

Four potential SRL values were defined by using three approaches: two values were defined by using a binomial probability method based on one-sided, nonparametric upper confidence limits, one was defined as equal to the maximum concentration detected in the field blanks, and one was defined as equal to the maximum laboratory method detection level used during the period when samples were collected for the project. The differences in detection frequencies and concentrations among different types of blanks (laboratory instrument blanks, source-solution blanks, and field blanks collected with three different sampling equipment configurations) and groundwater samples were used to infer the sources and mechanisms of contamination for each VOC detection in field blanks. Other chemical data for the groundwater samples (oxidation-reduction state, co-occurrence of VOCs, groundwater age) and ancillary information about the well sites (land use, presence of known sources of contamination) were used to evaluate whether the patterns of detections of VOCs in groundwater samples before and after application of potential SRLs were plausible. On this basis, the appropriate SRL was selected for each VOC that was determined to require an SRL.

The SRLs for ethylbenzene [0.06 microgram per liter (μg/L)], m- and p-xylenes (0.33 μg/L), o-xylene (0.12 μg/L), toluene (0.69 μg/L), and 1,2,4-trimethylbenzene (0.56 μg/L) corresponded to the highest concentrations detected in the QCFBs and were selected because they resulted in the most censoring of groundwater data. Comparisons of hydrocarbon ratios in groundwater samples and blanks and comparisons between detection frequencies of the five hydrocarbons in groundwater samples and different types of blanks suggested three dominant sources of contamination that affected groundwater samples and blanks: (1) ethylbenzene, m- and p-xylenes, o-xylene, and toluene from fuel or exhaust components sorbed onto sampling lines, (2) toluene from vials and the source blank water, and (3) 1,2,4-trimethylbenzene from materials used for collection of samples for radon-222 analysis.