In order to maintain a high degree of accuracy and confidence in the water quality data all personnel associated with this project were trained according to standard operating procedures for the methods listed in Table 2 (page §). Single-blind quality control tests were conducted as part of the IWS laboratory certification process. The 2001/2002 results are presented in Table 36 (page §). All results from the single-blind tests were within acceptance limits.
Laboratory duplicates were analyzed for at least 10% of all water quality parameters except the Hydrolab data. Laboratory duplicates were used to create control charts27 that track analytical performance over time. Upper and lower acceptance limits (± 2 std. dev. from mean pair difference) and upper and lower warning limits (± 3 from mean pair difference) were developed using 2000–2001 data (upper examples in Figures 176–183, pages §–§), and used to evaluate laboratory duplicates from 2002 (lower examples in Figures 176–183). The control charts indicate that the laboratory duplicates have been consistent over time.
Separate field duplicates were collected and analyzed for at least 10% of all of the water quality parameters except the Hydrolab data. To check the Hydrolab measurements, duplicate samples were analyzed for at least 10% of the Hydrolab measurements using water samples collected from the same depth as the Hydrolab measurement. The field duplicates results were in close agreement, given that they came from different water samples (Figures 184–188, pages §–§). Field duplicates are rarely as close as laboratory duplicates. As in previous years, systematic bias was observed in the conductivity results because the Hydrolab field meter is more sensitive than our laboratory meter. This appears as a flattening of the laboratory conductivity response at ~60 μS (Figure 184). In addition, the conductivity probe in the current Hydrolab unit is more sensitive than the Surveyor II Hydrolab used in the early 1990s, which creates the appearance of a decrease in the lake’s conductivity over time (Figures 67–71, pages §–§). These conductivity differences were generally ≤ 5 μS. There was a small systematic bias in the pH data, with the Hydrolab results showing a more extreme range than the laboratory pH results. This is most likely due to slight changes in the amount of dissolved CO2 and associated inorganic carbon ions (bicarbonate and carbonate) that occurred after the samples were collected. This type of pH shift is common in low alkalinity water samples.
The median difference between Hydrolab and Winkler dissolved oxygen values was 0.47 mg/L, and most of the samples differed by less than 1.0 mg/L. As in previous years, the only extreme differences occurred in samples collected in late summer from near the thermocline in basins 1 or 2. These differences are to be expected, and are caused by collecting water samples from slightly different depths at the thermocline. During the summer of 2002, however, we experienced a significant drop in Hydrolab sensitivity when measuring extremely low oxygen concentrations. Despite frequent repairs and equipment replacement by Hydrolab, we were not able to measure oxygen concentrations less that 2.0 mg/L with any accuracy. Because of this, we censured all oxygen values less than 2.0 mg/L from the 2002 data records. In addition, the Hydrolab results were consistently higher than the Winkler results in the 12–15 mg/L range. Although the Hydrolab unit we used in 2001/2002 is relatively new, and has been sent to Hydrolab repeatedly for repair and service, it does not appear to be as accurate as previous units. We are working with the City to test alternate field meters.
As part of our regular field quality control protocols, we measure initial and ending surface Hydrolab readings at each site. This is done to verify that we allowed a sufficient equilibration time during the first few samples for the Hydrolab to record accurate values. The results, shown in Figure 189 (page §), indicate that there was no consistent bias between surface and bottom conductivity, pH, or temperature readings. There was a slight tendency for the ending dissolved oxygen values to be higher than the initial values. This may be consistent with the loss of Hydrolab meter sensitivity that caused censuring of the low oxygen values.