8277 - Laserfiche WebLink

Home Browse Search

criteria during analysis, and only the concentration is estimated, not the presence (National Water Quality Laboratory, 1996). Throughout this report, estimated concentrations above the MRL were treated as normal concentrations, whereas estimated concentrations below the MRL were treated as "less than" concentra- tions. Acknowledgments We thank Bob Boulger, Jeff Foster, Dave Hartle, Dennis Smits, and Rick Szmajter, USGS, for their efforts in collecting the water samples for the data analyzed here. Dave Butler, USGS, provided valuable insight into site selection for the synoptic study of agricultural drains in the Grand and Uncompahgre River Valleys. Thanks also are extended to Joe Doma- galski and Bob Kimbrough for technical review of the manuscript and to Mary Kidd for editorial review, John Evans for design, and Alene Brogan for manu- script and layout. QUALITY-CONTROL METHODS AND RESULTS Quality-control (QC) samples were used to assess the bias and variability of water-quality data that may be introduced by sample collection, processing, shipping, and analysis. Without QC samples, the quality of the data collected cannot be determined. Field-blank, replicate, and field-matrix- spike samples were collected and processed as QC samples using the same equipment and procedures as the stream-water or environmental samples in order to evaluate the bias and variability of the pesticide data. Pesticide surrogate solutions were added to each envi- ronmental and quality-assurance sample in the labora- tory as an additional QC mechanism. The same QC procedures were used in the fixed-station and synoptic sampling. Quality-control data for both sampling programs were analyzed together. Further information on QC in the NAWQA Program and QC samples is described in Shelton (1994). Field-blank samples were used to test for bias, the systematic error inherent in sampling and analyt- ical methods (Spahr and Boulger, 1997). Blank samples were organic-free water that did not contain detectable concentrations of the pesticides of interest. The blank water, Baker Analyzed HPLC (J.T. Baker Company, Phillipsburg, New York), was processed through the sampling equipment and handled as if it were a typical stream sample. Because pesticides in a water sample can be detected at low concentrations in the laboratory, field blanks are important in deter- mining possible contamination of a sample from collection, processing, cleaning, shipping, and anal- ysis procedures. Replicate samples were two or four samples collected simultaneously or in sequence in order to yield samples of nearly identical composition. Repli- cates provide information on the precision of concen- tration values and the consistency in identifying targeted pesticides. As such, replicate samples test for variability, the degree of random error in independent measurements of the same quantity (Spahr and Boulger, 1997). Replicate analysis is based on the comparison of the analytical results of the environ- mental sample with the analytical results of the QC sample or samples. Replicates were analyzed by the relative percent difference between replicate samples, calculated by dividing the difference in concentration between two replicate samples by the mean concentra- tion of the two replicate samples and multiplying the result by 100. Where more than two samples were collected at a site, a range of relative percent differ- ences was calculated. The field-matrix-spike sample was an environ- mental sample that had been spiked with a solution containing targeted pesticides and then analyzed as a typical water-quality sample. Specifically, one liter of each environmental sample to be analyzed by GC/MS was spiked with the targeted pesticides at a concentra- tion of 0.1 µg/L, while another liter, for HPLC anal- ysis, was spiked with targeted pesticides at a concentration of 1 µg/L. The spiked samples were used to identify the recoverability of pesticides in the environmental samples. Pesticide recovery can be affected by pesticide degradation, interference from other organic substances or matter in the sample, and laboratory analysis. Spike solutions were provided by the USGS NWQL. Pesticide-spike-recovery data were calculated in three steps. First, the expected or theoretical concen- tration of the spiked sample was determined by multi- plying the concentration of the spike solution with the amount of spike added to the environmental sample and then dividing the result by the sample volume. QUALITY-CONTROL METHODS AND RESULTS 13