8.7.4 Interpretation and reporting
Interpretation of results and reporting do not normally involve the analyst. Nevertheless, strict adherence to analytical control procedures make it possible for queries about unusual or anomalous results to be referred back to the analyst and sampler, through an audit trail.
Although the absolute values may differ with different approaches described, there is a high level of agreement on the water microbiological quality qualification of the beach in relation to compliance. Notice from Table 8.8 how, in bathing area A, only the geometric mean allows compliance in relation to a 1,000 cfu per 100 ml standard. However, in the regulations that govern the use of the geometric mean there is only one standard and the mean is calculated from five consecutive results taken over a period of one month (i.e. the running geometric mean). In this approach there is a strong influence from the most frequent values (8 of the 12 samples rank from 16 to 740 cfu per 100 ml). In the second example, bathing area B, the 90th percentile is the most restrictive, because this approach is highly influenced by the single high value obtained.
One of the features of microbiological studies of water quality at recreational areas is the wide variations in results, temporally and spatially (Fleisher and McFadden, 1980; Gameson, 1982; Tillett, 1993; PHLS, 1994) that are much greater than those caused by laboratory procedures. With effective quality control in the laboratory, variability caused by procedures, such as sub-sampling from the same bottle or by the enumeration procedures themselves, are little greater than expected from the assumption of random distribution of bacteria in the sample and of the Poisson theory, particularly when the mean number is low.
A single limit standard leads to water of borderline quality and with low variability, consistently passing, whereas water which is usually of high quality, but is occasionally affected by intermittent pollution, would fail, even though the former arguably poses a greater risk to health. More detailed study of the results in the latter case might identify the causes of failure and enable advice to be given to the public not to use the water when poor conditions are expected, or enable remedial action to be taken. Bathing area A shows consistently bad quality, with a higher geometric mean (597) and a lower log standard deviation (0.69) than bathing area B (15 and 0.98 respectively). The failure of bathing area B is caused by a single, unusually high count (1,600 cfu per 100 ml) that may be due to a rain effect on the second sampling date.
8.7.5 Control charts
One way of identifying systematic changes in water quality, or of pinpointing sudden deterioration of water quality at a recreational area, is to create control charts (see Chapter 4). Water quality data points are plotted sequentially, as they are obtained, against time on a chart. Any existing historical data from specific bathing areas can be used to create control charts and helps to identify patterns of behaviour. The occurrence of high values is used to initiate investigation. Such values may be set to coincide to a guideline standard. More conventionally, two upper limit values are set on a control chart, at the mean plus twice and three-times the sample standard deviation, i.e. at m + 2s and m + 3s respectively. These represent values that would be expected to be exceeded