USGS Home
Contact USGS
Search USGS

Evaluation of Street Sweeping as a Stormwater-Quality-Management Tool in Three Residential Basins in Madison, Wisconsin

By William R. Selbig and Roger T. Bannerman

Appendixes

Appendix 2. Quality assurance and quality control

Appendix 2-1. Results of blank-sample analyses.

Appendix 2-2. Results of replicate-sample analyses.

Appendix 2. Quality assurance and quality control

Field and sample-processing equipment blanks were collected at the control and test-basin monitoring stations to evaluate the integrity of the stormwater-quality sampling process, identify if sample contamination existed and if so, to identify possible sources. Blank samples were obtained by drawing deionized water through the suction line and sampler into a collection bottle. The Teflon sample line and automatic sampler were not cleaned before obtaining blank samples. Blank water collected in the 10-liter glass sample bottle was then split using a Teflon-lined churn splitter into plastic laboratory-prepared sample bottles. Samples were placed on ice and delivered to the Madison Department of Public Health (MDPH) for analysis. Deionized blank water was also used to isolate individual elements of the sampling process from source to delivery. These samples were not delivered to the MDPH unless erroneous concentrations were found in the original blank sample. Blank-sample results are detailed in table 2-1. Replicate samples were also collected to evaluate the inherent variability in the sampling analyses and methods. The bias and variability identified by analysis of blanks and replicates were within acceptable limits except for dissolved zinc, dissolved chloride, and total cadmium.

Total zinc concentrations often were above detectable limits but were typically an order of magnitude lower than those measured in stormwater-quality samples. Dissolved zinc concentrations, however, were detected in blanks at concentrations greater than in field samples at each of the stormwater-quality monitoring stations. The median dissolved zinc concentration, for all blank samples with dissolved zinc concentrations above the limit of detection, was 1.0 micrograms per liter. The minimum dissolved zinc concentration from a stormwater-quality sample taken at the control, air-sweeper, and high-frequency basins was 0.8, 1.0, and 1.3, respectively. Analyses of deionized water used for blank samples, with no exposure to sample or processing equipment, indicated a presence of total and dissolved zinc concentrations that was greater than the detectable range in field samples. It is possible that zinc concentrations detected in many blank samples may have come from the deionized water. Contamination of dissolved zinc could also have come from a source other than the deionized water. Additional blank samples taken to isolate individual elements of the sampling process had dissolved zinc concentrations that were slightly higher than the deionized water. Deionized water was used only as a rinse during sample bottle cleaning, therefore any dissolved zinc residue remaining on field sample collection bottles after cleaning would be insignificant when compared to the mass of water in the field samples.

Dissolved chloride concentrations in blanks were also greater than the detectable limit and concentration in field samples at each of the monitoring stations on several occasions. Follow-up tests of the deionized water used in the blank sample process resulted in concentrations of dissolved chloride less than the detectable limit. One explanation for elevated dissolved chloride levels in the control basin may be due to the periodic drainage of a private swimming pool into basin storm drains. Similarly, fire hydrants in each basin were flushed twice per year onto adjacent streets. Water feeding the hydrants originated from a chlorinated city water supply. For these reasons, and since no obvious source of chloride was detected in the stormwater-quality sampling equipment, dissolved chloride was disregarded from statistical evaluation.

Finally, total cadmium was detected in the control basin blank sample on one occasion. This sample had been contaminated in the analytical laboratory and does not reflect the efficacy of the stormwater-quality sampling equipment.

Replicate samples were submitted to verify reproducibility in the sample acquisition and splitting process as well as analytical methods conducted in the laboratory. Replicate samples were checked for precision on the basis of an absolute relative percent difference (RPD). Replicate-sample results are detailed in table 2-2. On several occasions within each basin, the RPD values exceeded 50 percent. The majority of these samples were acquired prior to identifying the potential bias in concentration data due to the presence of sand-size particles in a whole-water sample (Selbig and others, 2007). Changes to the sample-splitting and laboratory analytical techniques were implemented in May 2004. The RPD values for replicate samples collected after May 2004 were mostly within an acceptable range of error. Of the RPD values exceeding 50 percent after the new methods were used, the majority occurred from a single replicate (table 2-2). Due to laboratory error, the new sample-splitting techniques were not used to process this sample which was likely the cause of the differences. This sample was therefore not used as part of the replicate analysis.

Appendix table 2-1. Results of blank-sample analyses.

[mm/dd/yyyy, month/day/year; µg/L, microgram per liter; mg/L, milligram per liter; <, less than]

Appendix table 2-1. Results of blank-sample analyses—Continued.

[mm/dd/yyyy, month/day/year; µg/L, microgram per liter; mg/L, milligram per liter; <, less than]

Appendix table 2-2. Results of replicate-sample analyses.

[mm/dd/yyyy, month/day/year; µg/L, microgram per liter; mg/L, milligram per liter; RPD, relative percent difference; %, percent; --, not applicable; values in bold indicate exceedance of allowable error]

Appendix table 2-2. Results of replicate-sample analyses—Continued.

[mm/dd/yyyy, month/day/year; µg/L, microgram per liter; mg/L, milligram per liter; RPD, relative percent difference; %, percent; --, not applicable; values in bold indicate exceedance of allowable error]

Appendix table 2-2. Results of replicate-sample analyses—Continued.

[mm/dd/yyyy, month/day/year; µg/L, microgram per liter; mg/L, milligram per liter; RPD, relative percent difference; %, percent; --, not applicable; values in bold indicate exceedance of allowable error]