Interpretive Reports
Determination of Study Reporting Limits for Pesticide Constituent Data for the California Groundwater Ambient Monitoring and Assessment Program Priority Basin Project, 2004–2018—Part 1: National Water Quality Laboratory Schedules 2003, 2032, or 2033, and 2060
Fram, M.S., and Stork, S.V., 2019, U.S. Geological Survey Scientific Investigations Report 2019-5107, 129 p.
ABSTRACT
The California Groundwater Ambient Monitoring and Assessment Program Priority Basin Project (GAMA-PBP) is a long-term cooperative project designed to assess the quality of groundwater resources used for public and domestic drinking water supplies in the State of California, to monitor and evaluate changes to that quality, to investigate the human and natural factors controlling water quality, and to improve the availability of comprehensive groundwater quality data and information. Between May 18, 2004, and May 3, 2018, the GAMA-PBP collected 3001 groundwater samples for analysis of pesticide constituents by the U.S. Geological Survey (USGS) National Water Quality Laboratory (NWQL)(note that pesticide constituents includes parent compounds and degradates). Of these samples, 2994 were analyzed for pesticide constituents on schedules 2003, 2032, or 2033 (65 to 84 constituents), and 840 were analyzed for pesticide constituents on schedule 2060 (58 constituents). The original dataset reported by the NWQL to the USGS National Water Information System (NWIS) database contained a total of 2,688 detections of 78 pesticide constituents and 253,825 non-detections. In this original dataset, 33 percent of the 3,001 samples analyzed had reported detections of one or more pesticide constituents. This report describes the GAMA-PBP data-quality objectives for pesticide data, the procedures used to establish study reporting limits, and use of those reporting limits to censor the data from the NWQL so that the final data published by the GAMA-PBP meet these data-quality objectives. The final GAMA-PBP dataset for samples collected from May 2004 to May 2018, after censoring, had a total of 1,632 detections of 37 pesticide constituents. In the final GAMA-PBP dataset, 25 percent of the 3,001 samples analyzed had detections of one or more pesticide constituents.
This report describes the GAMA-PBP data-quality objectives for pesticide data, the procedures used to establish study reporting limits, and use of those reporting limits to censor the data from the NWQL so that the final data published by the GAMA-PBP meet these data-quality objectives. The final GAMA-PBP dataset for samples collected from May 2004 to May 2018, after censoring, had a total of 1,632 detections of 37 pesticide constituents. In the final GAMA-PBP dataset, 25 percent of the 3,001 samples analyzed had detections of one or more pesticide constituents.
The presence of pesticides in groundwater is commonly evaluated by calculating detection frequencies. Detection frequencies for pesticides are sensitive to detection limits and method performance for concentrations near those limits; therefore, the two primary data quality issues addressed in the GAMA-PBP data-quality objectives for pesticides are (1) establishing criteria for classifying data from the laboratory as detections or non-detections for the purpose of data reporting by the project and (2) accounting for changes in analytical methods or method performance over time. The GAMA-PBP addresses these issues by developing study reporting limits that are used as the boundary between detections and non-detections for the reporting of GAMA-PBP results. These reporting limits are defined from method detection limits (MDLs) provided by the NWQL, unless examination of results from laboratory set blanks (LSBs) and GAMA-PBP field blanks indicates that a higher concentration censoring limit is warranted. The GAMA-PBP selected the MDL as the primary choice for defining study reporting limits for consistency with U.S. Environmental Protection Agency (EPA) guidelines for reporting detections of pesticides and other organic constituents.
A five-step procedure is used to develop study reporting limits and censor the GAMA-PBP dataset accordingly. The effect of the censoring at each step is described to provide information about the relative effect of each step on the overall censoring of the dataset. Steps 1 and 2 can be implemented at the time the data are received, whereas steps 3−5 require information accumulated over an extended period.
- Step 1: Reject results that were most likely the result of specific contamination instances attributable to unusual field or laboratory conditions during sample collection or processing. Two such instances were identified, leading to rejection of 25 detections, which were assigned a data-quality indicator code of “Q†for “reviewed and rejected†in the NWIS database.
- Step 2: Use the NWQL MDLs in effect at the time each sample was analyzed as the reporting limit. A total of 506 detections were censored on this basis.
- Step 3: Use the maximum MDL established by the NWQL during July 2004–August 2018 (MDLmax) as the reporting limit. The rationale for using the MDLmax as the reporting limit is based primarily on the observation that the concentrations of MDLs generally increased over time. A total of 438 detections were censored on this basis.
- Step 4: Use the LSBs to identify periods of greater potential laboratory contamination bias and define raised reporting limits to be used during those periods. These periods were defined by using a moving average detection frequency approach. For consistency with the NWQL procedures for defining raised reporting limits on the basis of detections in LSBs, the raised reporting limits were defined as equal to three times the highest concentration measured in an LSB during the period. A total of 25 detections in groundwater samples analyzed during periods of increased laboratory contamination bias were censored.
- Step 5: Use the LSBs and field blanks to identify potential contamination bias from field or laboratory processes outside of the time periods identified in step 4. The NWQL protocols were used to define the MDLs from blanks analyzed outside of the periods identified in step 4. If an MDL defined from blanks was greater than the MDLmax, the MDL defined from blanks was used to censor the data. One constituent had a study reporting limit defined on this basis, and a total of 62 detections in groundwater samples were censored.
As of 2019, the USGS NWIS database does not have the capability to store both the original value reported by the NWQL and the final value published by the GAMA-PBP that reflects application of the quality-control censoring described in this report. In the interim, while this capability is developed, the 1,031 results censored in steps 2−5 are blocked from public release in NWIS, and the GAMA-PBP has published the original and final values in a USGS data release accompanying this report. The entire GAMA-PBP final dataset for pesticide constituents on schedules 2003, 2032, or 2033, or on schedule 2060 is publicly available in that USGS data release, through the USGS GAMA-PBP public web portal, and through the California State Water Resources Control Board GAMA public groundwater information system.