Page 19: Guidelines for Canadian Recreational Water Quality – Third Edition
The Guidelines recommend that when assessing water quality for adherence with the guideline values for the recommended indicators of faecal contamination, a minimum of five samples should be collected, at a minimum frequency of one sample per week.
Composite sampling (collection of multiple samples from across a stretch of beach, combining the samples into one large composite, analysing a subsample of the resulting mixture) presents a possible means for increasing the area covered under a beach monitoring program, while potentially minimizing the costs associated with analysis.
Authorities and beach operators and/or service providers may wish to investigate composite sampling as a tool to potentially improve the utility of their faecal indicator monitoring programs.
With some monitoring programs, single samples may be used to characterize long stretches of swimming water, across several days of swimmer activity. It is known that the fluctuation of water quality can be significant, even over short distances and time periods. Collecting multiple samples more frequently is recommended, because increasing the number of samples increases the reliability of the data (Whitman and Nevers, 2004). The costs associated with increased monitoring, however, can be prohibitive. One solution that has been proposed to address this problem is the use of composite sampling (U.S. EPA, 2005a).
The process of composite sampling requires collecting multiple samples. Equal volumes from each sample are then mixed together to form a composite, which is then analysed as a single sample. This technique can broaden the coverage of a sampling strategy where the analysis of large numbers of samples would otherwise be required. Subsequently, it can be used to increase sampling reliability without significantly affecting the costs associated with monitoring (Patil, 2002). Composite sampling has numerous applications in biomonitoring and environmental sampling and has been used for assessing contamination in a variety of media, including soils, air, water and biological tissue. Recently, investigations have been conducted to explore whether composite sampling can be applied to the assessment of the quality of recreational waters (Kinzelman et al., 2006). Preliminary evidence has indicated that when properly conducted, composite sampling can be used in making water quality decisions with a degree of accuracy comparable to that of traditional sampling regimens.
There are challenges associated with composite sampling that need to be taken into consideration before introducing this technique into a recreational water monitoring program. These are described briefly below.
Potential sources of bias
Composite sampling adds another layer of uncertainty to the water quality results, since a subsample is being used to estimate the average indicator density over all samples, and this estimate in turn is used to characterize the water quality for the whole beach. It is suggested that including more samples in the composite compensates for the effects of this bias. There is the potential for the presence of an individual sample with a high concentration to be masked when combined with samples with lower concentrations, owing to the effects of dilution (Kinzelman et al., 2006). Samples from hot spots (areas where poor water quality is likely to be persistent) should not be composited with other samples. The identification of hot spots may be determined by conducting an Environmental Health and Safety Survey (EHSS) or through an initial period of intensive sampling. In programs where hot spots have been characterized and determined as unlikely to occur, the occurrence of a single sample with high concentrations may be considered to be the result of natural, random variability (Kinzelman et al., 2006).
Comparing results with the guideline values
The composite sampling result approximates the arithmetic mean of the indicator counts of the individual samples. However, when analysing bacteriological water quality data, the geometric mean is recommended as the best estimate of central tendency of microbial populations. The guideline values for the recommended indicators of faecal contamination are based on geometric mean values. In determining whether the results are in accordance with the guidelines, operators/service providers or responsible authorities would need to convert the composite result to an approximation of the geometric mean.
Wymer et al. (U.S. EPA, 2005a) pointed out that the difference between the composite value and the geometric mean can be compensated if the variance (v) of the log10 indicator densities is known. An estimate of the variance can be calculated from historical log10 data. Once the variance has been determined, multiplying the count per 100 mL obtained from the composite sample by the factor 10−1.15v produces a value that is approximately equivalent to the geometric mean of the individual samples (U.S. EPA, 2005a).
Kinzelman et al. (2006) produced swimming beach water quality data comparing the accuracy of composite sampling with traditional monitoring practices at two Lake Michigan beaches in Racine, Wisconsin. Water samples were collected over 68 days in 2003 from two public swimming beaches and analysed for E. coli using single-sample analysis with arithmetic and geometric mean averaging and composite sample analysis.
The resulting data indicated that composite sampling appeared to be an effective alternative to traditional monitoring procedures. In general, the value of the composite sample fell within the range of the single-sample values, and the data indicated an approximate 1:1 ratio between the composite sample and the arithmetic mean of the individual samples. In comparing what would have been the ultimate management decision (i.e., issue a water quality advisory or allow the beach to remain open) resulting from the use of composite sampling versus the results from individual analyses (singly or with averaging), the outcome for both methods remained constant at one beach and differed in only two instances at the other. Compositing resulted in additional advisories in both instances. Therefore, compositing appeared to introduce neither bias nor additional variability into the monitoring results (Kinzelman et al., 2006). Verification studies, performed on a smaller scale in subsequent years, have yielded similar results, and Racine has successfully used composite sample analysis for compliance monitoring since 2004.
Under the appropriate circumstances, compositing of samples may present a viable alternative to current monitoring schemes that employ a single sample to characterize water quality over long stretches of swimming beaches. Composite sampling may encourage more sampling to take place, thus expanding the coverage of the monitoring program and increasing sampling reliability, while at the same time maintaining the costs associated with monitoring or even lowering them. Monitoring programs that require a large number of samples to be analysed could benefit from adoption of this approach.
Page details
- Date modified: