Quality of major ion and total dissolved solids data from groundwater sampled by the National Water-Quality Assessment Program, 1992–2010

Abstract

Proper interpretation of water quality requires consideration of the effects that contamination bias and sampling variability might have on measured analyte concentrations. The effect of contamination bias and sampling variability on major ion and total dissolved solids data in water samples collected in 48 of the 52 National Water-Quality Assessment Program study units from 1992 to 2010 is discussed in this report. Contamination bias and sampling variability can occur as a result of sample collection, processing, shipping, and analysis. Contamination bias can adversely affect interpretation of measured concentrations in comparison to standards or criteria. Sampling variability can help determine the reproducibility of an individual measurement or whether two measurements are different. Field blank samples help determine the frequency and magnitude of contamination bias, and replicate samples help determine the sampling variability (error) of measured analyte concentrations. Quality control data were evaluated for calcium, magnesium, sodium, potassium, chloride, sulfate, fluoride, silica, and total dissolved solids. A 99-percent upper confidence limit is calculated from field blanks to assess the potential for contamination bias. For magnesium, potassium, chloride, sulfate, and fluoride, potential contamination in more than 95 percent of environmental samples is less than or equal to the common maximum reporting level. Contamination bias has little effect on measured concentrations greater than 4.74 mg/L (milligrams per liter) for calcium, 14.98 mg/L for silica, 4.9 mg/L for sodium, and 120 mg/L for total dissolved solids. Estimates of sampling variability are calculated for high and low ranges of concentration for major ions and total dissolved solids. Examples showing the calculation of confidence intervals and how to determine whether measured differences between two water samples are significant are presented. Introduction The U.S. Geological Survey’s (USGS) National-Water Quality Assessment (NAWQA) Program was implemented in 1991 in order to describe current water-quality conditions and how they are changing and to improve scientific and public understanding of natural and human factors impacting those conditions. These objectives are being achieved through extensive monitoring within 52 study units, which consist of large river basin and aquifer systems throughout the United States. In Cycle I (1991–2001) and Cycle II (2002–12), much of the work involved gathering comparable information on water quality in both surface water and groundwater. Estimates of contamination bias and sampling variability resulting from sample collection, processing, shipment, and laboratory analysis are needed to quantify how much variability in water-quality measurements can be explained by field and laboratory methods, as compared to environmental factors (Mueller and Titus, 2005). Quality-control (QC) samples, such as field blank or replicate samples, are collected at the same time as the environmental samples in order to evaluate contamination bias and sampling variability. Contamination bias is the systematic error that can occur during sample collection, processing, shipping, or laboratory analysis. Contaminants can be introduced into water samples by exposure to airborne gases and particulates or from inadequately cleaned sample collection or analytical equipment (Mueller and Titus, 2005). Variability is the degree of random error in independent measurements of the same quantity, and “sampling variability” (termed by Mueller, 1998, p. vii) is the variability introduced by sample collection, field processing, shipping, and laboratory analysis. Contamination bias and sampling variability are evaluated by collecting and analyzing QC samples. The frequency and magnitude of contamination bias are determined from field blank samples, and the sampling variability of measured analyte concentrations is determined from 2 Quality of Data from Groundwater Sampled by the National Water-Quality Assessment Program, 1992–2010 replicate samples. The distribution of concentrations in field blank samples is used to estimate the potential distribution of contamination in the environmental samples. Similarly, the distribution of variability in the replicate sets is used to estimate potential sampling variability in the results from environmental samples. Estimates from a particular set of field blanks or replicates can be applied to a particular set of environmental samples to describe similar sample collection and analytical methods, sample collection site characteristics, and sample collection during a specific time period.