USGS - science for a changing world

Data Series 285

U.S. GEOLOGICAL SURVEY
Data Series 285
(ver 1.1, August 2018)

Back to Table of Contents

Quality-Control Data Analysis

QC samples were collected with approximately 10 percent of the samples in Southern Sacramento Valley GAMA study unit. QC samples were collected to assess the bias and variability of ground-water data resulting from sample collection, processing, storage, transportation, and laboratory analysis. Three types of QC samples were collected: blanks, sequential replicates, and laboratory matrix spikes. Additionally, surrogate compounds were added to selected samples to assess the general performance of some analytical methods. Assessment of the quality-control data resulted in censoring less than 0.03 percent of the analyses.

Surrogates

Surrogate compounds were added to all ground-water and QC samples that were analyzed for VOCs, gasoline-oxygenates, NDMA, 1,2,3-trichloropropane, and pesticide compounds. Surrogates are compounds not normally found in the environment, but have similar physical and chemical properties to the target analytes. Prior to laboratory analysis, each sample was spiked with one or more surrogates. Surrogate recovery data were used to evaluate the capability of the analytical methods to detect the target analytes in each sample and to assess bias and variability that were due to matrix effects and gross laboratory processing errors. Surrogate data in blanks and samples were used to identify general problems that may arise during sample analysis; surrogate data in ground-water samples were used to evaluate matrix interferences. A 70- to 130-percent recovery of surrogates was considered acceptable in this report. In samples with low surrogate recoveries, target analytes may not have been detected if present in low concentrations; samples with high surrogate recoveries indicate that the target analytes will be detected if present, but the concentrations may be exaggerated. If all surrogates for a sample were outside of the acceptable range, then there may have been a problem with the analytical method. If one or more, but not all, surrogates for a sample were outside of the acceptable range, then the sample chemistry may have interfered with the capability of the methods to detect and quantify the target analytes (matrix interference). ">Surrogate compounds were added to all ground-water and QC samples that were analyzed for VOCs, gasoline-oxygenates, NDMA, 1,2,3-trichloropropane, and pesticide compounds. Surrogates are compounds not normally found in the environment, but have similar physical and chemical properties to the target analytes. Prior to laboratory analysis, each sample was spiked with one or more surrogates. Surrogate recovery data were used to evaluate the capability of the analytical methods to detect the target analytes in each sample and to assess bias and variability that were due to matrix effects and gross laboratory processing errors. Surrogate data in blanks and samples were used to identify general problems that may arise during sample analysis; surrogate data in ground-water samples were used to evaluate matrix interferences. A 70- to 130-percent recovery of surrogates was considered acceptable in this report. In samples with low surrogate recoveries, target analytes may not have been detected if present in low concentrations; samples with high surrogate recoveries indicate that the target analytes will be detected if present, but the concentrations may be exaggerated. If all surrogates for a sample were outside of the acceptable range, then there may have been a problem with the analytical method. If one or more, but not all, surrogates for a sample were outside of the acceptable range, then the sample chemistry may have interfered with the capability of the methods to detect and quantify the target analytes (matrix interference).

Table 5 summarizes surrogate recoveries for ground-water and QC samples. Median surrogate recoveries for all analyte groups were acceptable. All gasoline oxygenate and 1,2,3-trichloropropane samples had surrogate recoveries within acceptable limits. Some individual samples had low recoveries of VOC, pesticide, or NDMA surrogates; some analytes may not have been detected in these samples if present at low concentrations. Some individual samples had high recoveries of VOC and pesticide surrogates; the concentrations of detected VOCs or pesticide compounds measured in these samples may have been greater than the actual concentration.

Blanks

Two types of blanks were collected for the Southern Sacramento Valley GAMA study unit: field and source-solution blanks. All blanks were collected using nitrogen-purged blank water that was certified to be free of VOCs, pesticide compounds, nutrients, dissolved organic carbon, major ions, and trace elements above their reporting levels. Associated blanks and ground-water samples in this study were defined as samples that were collected at the same site.

Source-solution blanks were collected to verify that the blank water used for the associated field blank had no detectable concentrations of VOCs, gasoline oxygenates, NDMA, 1,2,3-trichloropropane, or perchlorate. Source-solution blanks were collected at the sampling site by pouring blank water directly into sample containers that were then stored, shipped, and analyzed in the same manner as the other blank and ground-water samples.

Field blanks were collected to evaluate potential bias introduced by sampling equipment, processing, shipping, or analysis. Field blanks were collected at selected sampling sites. Depending on the list of analytes (depth-dependent, slow, intermediate, or fast) collected at a particular sampling site, blank water was either pumped or poured through the same sampling equipment (fittings and tubing, and at monitoring wells, sampling pump) used to collect ground water, then processed and transported using the same methods for the ground-water samples.

A constituent was of potential QC concern with reference to blank data when all of the following criteria were met: (1) the constituent was detected in one or more field blanks and in ground-water samples, (2) the concentration detected in the field blank was greater than the concentration in the associated source-solution blank, and (3) the minimum concentration detected in ground-water samples was less than the maximum concentration detected in field blanks plus one half of that constituent’s reporting level. If the results for a constituent met the above criteria, the pattern of detections in blanks and ground-water samples was evaluated. If a constituent was detected in at least one associated blank and ground-water sample at similar concentrations, then all detections in ground-water samples were censored. If a constituent was detected in the field blanks, but not in the associated ground-water sample, then the ground-water data were not censored.

Detections in ground-water samples that were determined to be of QC concern were censored and reported as nondetections and flagged with a “V” remark code. The threshold for censoring data was determined by summing the blank concentration and the long-term method detection level (LT-MDL), or half of the method detection level (MDL), for the constituent in question. For example, if the highest concentration of toluene detected in a field blank was 0.02 μg/L, and the LT-MDL for toluene was 0.02 μg/L, then the concentration of toluene in the associated ground-water sample would have to be greater than or equal to 0.04 μg/L to be considered a detection in ground water. This method of censoring was based on the assumption that the concentrations in the field blank and the associated ground-water sample were comparable.

One constituent, oxamyl, was detected in 7 of the 9 monitoring wells sampled in this study at levels below or near the reporting level with values ranging from 0.01 to 0.08 μg/L. Oxamyl was also detected in an associated field blank. Because oxamyl was detected in a field blank, was detected at levels below or near the reporting level, was detected in 7 of the 9 monitoring wells sampled with the same sampling equipment, and is very rarely detected in ground-water, oxymyl has been determined to be of QC concern. All oxamyl values were censored and reported as nondetections and flagged with a "V" remark code.

The following constituents were censored with reference to blank data: antimony, benfluralin, cadmium, chromium (total, National Research Program [NRP] method), chromium(VI), DCPA, dissolved organic carbon, fluoride, hexafluoropropene, lead, toluene, trifluralin. Table 6 lists the constituents that did not pass QC blank analysis and were censored in ground-water data, along with the censoring level used, the minimum ground-water concentration, and the maximum field blank concentration. Some constituents were censored only in one or more groups of samples (by equipment and sampling method used) when the analysis of blank data from only that group showed contamination. The lead concentration in the field blank was assumed to come from the fittings used on the sampling point at that well; as other field and ground-water samples had no detections of lead, it was interpreted to affect only the ground-water data at wells sampled on the intermediate and depth-dependent lists of analytes.

Replicates

Replicate samples were collected to assess the variability attributed to processing and analysis of inorganic and organic constituents. All replicates were sequential; the replicate sample was collected after the ground-water sample and analyzed using the same method. The variability in these sequential replicates was assumed to be largely from sample collection and analysis; the variability that was due to changes in ground-water composition with time were assumed to be minimal in the time frame of sample collection (in minutes). The relative standard deviation (RSD) was used to calculate the variability between replicate pairs. The RSD is defined as 100 times the standard deviation divided by the mean concentration for each replicate pair of samples. If one value in a sample pair was reported as a nondetect, and the other value was reported below the reporting level, the RSD was set to zero because the values were analytically identical. If one value in a sample pair was reported as a nondetect, and the other value was greater than the LRL or MRL, then the nondetection value was set equal to one-quarter of the reporting level, and the RSD was calculated. Values of RSD less than 20 percent were considered acceptable in this study. ">Replicate samples were collected to assess the variability attributed to processing and analysis of inorganic and organic constituents. All replicates were sequential; the replicate sample was collected after the ground-water sample and analyzed using the same method. The variability in these sequential replicates was assumed to be largely from sample collection and analysis; the variability that was due to changes in ground-water composition with time were assumed to be minimal in the time frame of sample collection (in minutes). The relative standard deviation (RSD) was used to calculate the variability between replicate pairs. The RSD is defined as 100 times the standard deviation divided by the mean concentration for each replicate pair of samples. If one value in a sample pair was reported as a nondetect, and the other value was reported below the reporting level, the RSD was set to zero because the values were analytically identical. If one value in a sample pair was reported as a nondetect, and the other value was greater than the LRL or MRL, then the nondetection value was set equal to one-quarter of the reporting level, and the RSD was calculated. Values of RSD less than 20 percent were 
considered acceptable in this study.

Table 7 lists the constituents that have RSDs greater than 20 percent. Of 185 replicate pairs that had detections, only 18 pairs had an RSD of greater than 20 percent. The RSDs of most pairs are less than 5 percent. Because most (90 percent) of the replicate analyses had acceptably low variabilities, the ground-water sample data collected for this study are presumed to have had relatively little variability introduced by the sampling and analytical procedures.

Laboratory Matrix Spikes

Laboratory matrix spikes were QC samples used to evaluate the bias and variability of analytical results that were due to interferences from the chemistry of the ground water sampled (matrix interferences). Spike samples were collected and analyzed for VOCs, pesticide compounds, perchlorate, 1,2,3-trichloropropane, NDMA, radium-226, radium-228, gross alpha and beta radiation, and coliphage. Laboratory matrix spikes were prepared by adding solutions containing known concentrations of target analytes to replicate ground-water samples before sample preparation and analysis at the laboratory. The constituents added in matrix spikes were the same as those being analyzed. Constituents with low recoveries were of potential concern because they may not have been detected if present in low concentrations; low recoveries also were a concern if environmental concentrations were close to a threshold level; a concentration less than a threshold could be falsely indicated. Constituents with high recoveries were of potential concern if the environmental concentrations were greater than a threshold level because a high recovery could falsely indicate a concentration greater than a particular threshold. Recoveries between 70 and 130 percent for matrix spikes were considered acceptable in this study.

Constituents that had low recoveries in at least one spiked sample are listed in table 8A. These constituents may not have been detected in some ground-water samples when present in low concentrations. Table 8A also indicates which low-recovery constituents were detected in ground-water samples, showing that, in some samples, these constituents were not missed because of low recoveries. Constituents that had high recoveries in at least one spiked sample are listed in table 8B. Concentrations of these constituents detected in ground-water samples (indicated in table 8B) may have been over-measured; none of these high-recovery constituents were detected at concentrations near an MCL or other high threshold. All VOCs and pesticide compounds not listed in tables 8A and 8B, along with perchlorate, 1,2,3-trichloropropane, NDMA, radium-226, radium-228, gross alpha and beta radiation, and coliphage, had acceptable spike recoveries.

Back to Table of Contents

AccessibilityFOIAPrivacyPolicies and Notices

Take Pride in America logoUSA.gov logoU.S. Department of the Interior | U.S. Geological Survey
URL: https://pubs.usgs.gov/ds/285
Page Contact Information: Publications Team
Page Last Modified: Monday, 06-Aug-2018 17:57:20 EDT