Improving the Accuracy of Sediment-Associated Constituent ...

0 downloads 0 Views 198KB Size Report
Jan 9, 2007 - William R. Selbig,* Roger Bannerman, and George Bowman. ABSTRACT. Sand-sized particles (.63 mm) in whole storm water samples col-.
Published online January 9, 2007

Improving the Accuracy of Sediment-Associated Constituent Concentrations in Whole Storm Water Samples by Wet-Sieving William R. Selbig,* Roger Bannerman, and George Bowman ditional experiments were performed in the analytical laboratory to improve the accuracy of copper and zinc concentrations by analyzing the entire contents of a sample container rather than an aliquot. Instead of subsampling, wet-sieving the whole storm water sample could eliminate the bias from and improve the precision of solid-phase storm water data. To ensure the success of this modified approach, the proper sieve size must be selected to minimize the bias and precision error. Two studies have identified two different particlesize divisions critical to reducing the errors associated with processing the whole storm water sample with a churn splitter. Horowitz et al. (1997) found unacceptable bias and precision when using churn splitters to process whole storm water samples with SSC greater than 1000 mg L21 or particles larger than 250 mm. The other study concluded that churn splitters are unable to representatively split whole storm water samples with particles larger than 63 mm since the stirring action is unable to overcome the tendency of sand grains to settle (Meade, 1985, cited in Capel and Larson, 1996). Other samplesplitting methods such as cone splitters may produce less bias but may also demonstrate poor precision when the original sample contains sand-sized particles (Horowitz et al., 1997). Based on these studies, initial sieving of a sample with a 63-mm sieve should improve the quality of the splitting process, but wet-sieving with a 63-mm sieve would add considerable time to the processing of most whole storm water samples. Hence, the first objective of this study was to further evaluate the sieve size needed to reduce the bias and improve precision when splitting whole storm water samples for SSC and total recoverable metal analyses. Subsampling to obtain the sample volume required by an analytical method is another potential source of error when determining sediment and sediment-associated constituent concentrations in samples with sand-sized particles. Gray et al. (2000) determined that one of the more commonly used methods to quantify concentrations of solid-phase material in natural waters, total suspended solids (TSS), was “fundamentally unreliable,” particularly when more than 25% of the material by weight is in the sand-sized fraction (U.S. Geological Survey, 2001). This method measures the dry weight of sediment from a known volume of a subsample of the original. A better representation of solid-phase material in natural waters, regardless of the absolute or relative amount of sandsized material in the sample, measures the dry weight of all the sediment without subsampling (Gray et al., 2000).

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

ABSTRACT Sand-sized particles (.63 mm) in whole storm water samples collected from urban runoff have the potential to produce data with substantial bias and/or poor precision both during sample splitting and laboratory analysis. New techniques were evaluated in an effort to overcome some of the limitations associated with sample splitting and analyzing whole storm water samples containing sand-sized particles. Wet-sieving separates sand-sized particles from a whole storm water sample. Once separated, both the sieved solids and the remaining aqueous (water suspension of particles less than 63 mm) samples were analyzed for total recoverable metals using a modification of USEPA Method 200.7. The modified version digests the entire sample, rather than an aliquot, of the sample. Using a total recoverable acid digestion on the entire contents of the sieved solid and aqueous samples improved the accuracy of the derived sediment-associated constituent concentrations. Concentration values of sieved solid and aqueous samples can later be summed to determine an event mean concentration.

C

of solids and pollutants associated with the solids in storm water runoff frequently requires subsampling or obtaining aliquots from a larger sample volume for laboratory analysis of multiple parameters. There is a growing body of evidence that demonstrates that using both churn splitting and aliquots can introduce significant bias and/or poor precision into the sediment and sediment-associated constituent concentration results (Capel and Larson, 1996; Horowitz et al., 1997; Gray et al., 2000). Much of the variability can be attributed to the presence of entrained sediment in urban runoff whose size can have a large mass fraction in the sand-sized range (Horwatich et al., 2004). Whole storm water samples collected from storm sewers frequently contain suspended sediment concentrations (SSC) exceeding 1000 mg L21 with a large percentage of sediment particles larger than 250 mm (Furumai et al., 2002; Waschbusch, 2003; Horwatich et al., 2004). The purpose of this paper is to present modifications to samplesplitting procedures and laboratory-analysis methods to improve the accuracy of storm water-contaminant concentration data. Laboratory created sediment suspensions were wet-sieved as part of the sample-splitting procedures to reduce errors associated with the presence of sand-sized particles in a storm water sample. AdHARACTERIZATION

W.R. Selbig, U.S. Geological Survey, 8505 Research Way, Middleton, WI 53562. R. Bannerman, Wisconsin Dep. of Natural Resources, 101 S. Webster Street–FH/4, Madison, WI 53703. G. Bowman, Wisconsin State Lab. of Hygiene, 2601 Agriculture Drive, Madison, WI 53718. Received 12 Apr. 2006. *Corresponding author ([email protected]). Published in J. Environ. Qual. 36:226–232 (2007). Short Communications doi:10.2134/jeq2006.0147 ª ASA, CSSA, SSSA 677 S. Segoe Rd., Madison, WI 53711 USA

Abbreviations: EMC, event mean concentration; SSC, suspended sediment concentration; TSS, total suspended solids; WSLH, Wisconsin State Laboratory of Hygiene.

226

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

SELBIG ET AL.: IMPROVING ACCURACY OF SEDIMENT-ASSOCIATED CONSTITUENT CONCENTRATIONS

One of the more commonly used methods to determine total recoverable metals concentration in aqueous samples also recommends analysis a subsample of the original. A summary of water quality concentrations in urban streams and storm sewers in Wisconsin showed most of the total recoverable metals are associated with the particulate matter (Bannerman et al., 1996). Therefore, use of a subsampling procedure might under- or overestimate the total recoverable metals concentration in the sample. As with the sediment analysis, the potential errors in the total recoverable metals analysis might be reduced by processing the entire sample. Therefore, a second objective of this study was to compare the results of a total recoverable analysis performed on a subsample with a similar analysis on the whole storm water sample. The magnitude of the subsampling error was investigated using two metals, zinc and copper, since they are both of concern in urban runoff (Bannerman et al., 1983, 1996). MATERIALS AND METHODS Experimental Design The approach to selecting the best sieve size for splitting whole storm water samples and the best laboratory method for total recoverable metals analysis is based on comparing test results from water samples with known SSC and total recoverable metals concentrations when various modifications to these procedures are used. Results of the tests that best match the known values would provide guidance on the optimum sample splitting and laboratory analysis procedures.

Preparation of the Water Samples for Evaluation of Sieve Sizes Reference samples with known sediment concentrations and particle sizes were created in the laboratory using sedi-

227

ment captured in situ by a storm sewer bedload sampler during rainfall runoff. After the sediment was collected, it was dried and sieved into the following U.S. Geological Survey particlesize classifications: .2000, 500–1000, 250–500, 125–250, 63– 125, and ,63 mm (Guy, 1969). Size-separated dry bedload sediment was added to deionized water in a tared, 14-L, Teflon-lined churn splitter to create simulated urban-runoff samples with known SSC (Fig. 1). Two SSCs were made for each of three ranges of particle sizes (Table 1). These samples were then well mixed for a period of approximately 1 min using protocols developed by the USGS when processing whole storm water samples in a churn splitter (USGS, 1999). Five aliquots of approximately 250 mL were withdrawn from the churn splitter and transferred to laboratory bottles, placed on ice, and delivered to the Wisconsin State Laboratory of Hygiene (WSLH) for SSC analysis. The WSLH was chosen to conduct all analytical testing of water quality samples because of its participation in the USGS Standard Reference Sample Program (USGS, 2003). Analysis results from these samples were compared to the known initial concentrations. It was hypothesized that the particle sizes showing the least amount of variance from a known initial concentration after being processed in the churn splitter should indicate the most appropriate sieve size to use on whole storm water samples with sand-sized particles.

Evaluation of Total Recoverable Metal Analysis Methods Two analytical methods are used to determine total recoverable metals concentrations in aqueous samples. The wholebottle analysis method uses an acid digestion on the entire contents of a 250-mL sample bottle submitted to the laboratory. The standard USEPA Method 200.7, on the other hand, performs the acid digestion on a 25-mL aliquot of the sample. When taking an aliquot from a sample containing sand-sized particles, the particles that fall out of suspension quickly may be omitted from the sample analyzed. Tests comparing the

Fig. 1. Diagram of processes to prepare dry bedload, sieved sediment, and water samples for analyses at the Wisconsin State Laboratory of Hygiene (WSLH).

228

J. ENVIRON. QUAL., VOL. 36, JANUARY–FEBRUARY 2007

Table 1. Prepared suspended sediment concentrations for selected particle size ranges. Particle size range

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

mm 63 to 125 250 to 500 500 to 1000

Prepared suspended sediment concentrations 21

1000 – –

mg L – 2000 2000

5000 5000 5000

results of these two methods were performed on a wide range of particle sizes and SSCs (Table 2). Prepared sediment suspensions were split with a churn splitter and aliquots submitted to the laboratory. The bedload sediment used to prepare the reference samples for these tests was processed in two different ways. Concentrations of zinc and copper were determined for both the dry bedload sediment and the sediments sieved from the sediment-water mixture in the churn splitter. The dry bedload sediment was used as a target because it precludes any bias or precision problems potentially associated with the churn splitter. However, since there is no volume associated with the dry bedload sediments, the concentration units from analysis on the dry sediment (mg kg21) are not directly comparable to the concentration units used for the water samples (mg L21). Since the sieved sediment should result in dry concentrations (mg kg21) of zinc and copper similar to the ones observed for the dry bedload sediment, the volume of the prepared sediment suspension was used to convert the dry weight concentrations to the concentrations in water. A total of four aliquots were removed from the churn splitter for each particle-size range and corresponding SSC (Fig. 1): three aliquots of approximately 250 mL for the whole-bottle analysis and one 250-mL aliquot for the standard USEPA analysis. The aliquots were then preserved with nitric acid, placed on ice, and delivered to the WSLH within 24 h for analysis. The whole-bottle and standard USEPA Method 200.7 samples were then analyzed for total recoverable zinc and copper (Fig. 2).

Wet-Sieving Sediment from a Churn Splitter The remaining sediment-water mixture was weighed using a large-capacity balance, then passed through a single 500-, 250-, or 63-mm nylon sieve to capture the remaining solid material for a specific particle-size fraction (Fig. 1). All material from a single pass captured on the sieve was transferred into a 250-mL polypropylene container using deionized rinse water. The container was then placed on ice and delivered to the WSLH within 24 h. A total recoverable zinc and copper analysis was performed on the dried sieved sediment. To determine if any contaminants leached from the sediment particles in the churn splitter into the deionized water during the mixing process, a sample of the water passing through each sieve was collected in a sample container, preserved with nitric acid, placed in ice, and submitted to the WSLH for total recoverable copper and zinc.

Table 2. Suspended sediment concentrations prepared for total recoverable metals tests. Particle size range mm 63 to 125 250 to 500 500 to 1000

Prepared suspended sediment concentrations 21

2000 – –

mg L 5000 5000 5000

– 10 000 10 000

Dry Bedload Sediment Analysis In addition to the aliquots from the splitter and sieved solid samples, a small amount of dried bedload sediment was sent to the WSLH for zinc and copper total recoverable metals analysis (Fig. 1).

Analytical Methods for Total Recoverable Metals Figure 2 depicts the procedures and methods used by the Wisconsin State Laboratory of Hygiene when analyzing the dry bedload sediment, sieved sediment, and aqueous samples from the splitter.

Split Water Samples Historically, the USEPA has recommended the “total recoverable method” of sample preparation as an indicator of the bioavailable pool of trace elements in sediments (USEPA, 1986). We followed this approach as specified in USEPA Method 200.7 for liquids (USEPA, 2001) in the first of two approaches evaluated. The second method also used USEPA Method 200.7, but instead of subsampling the original sample, the entire contents of the 250-mL sample bottle (whole-bottle method) were digested using a total recoverable acid digestion. Analyses for each method were performed in triplicate (Fig. 2). The digestion protocol performed on all sediment suspensions was as follows: First, the samples were weighed to determine the volume. Nitric and hydrochloric acid were added to the contents of each vial to bring the concentration to 2.5 and 5%, respectively. Each vial was then heated in an Environmental Express Hotblock at 958C using USEPA Method SW846 3005A (USEPA, 1993), and subsequently analyzed for zinc and copper by inductively coupled plasma emission spectroscopy (ICP) per USEPA Method 200.7 (USEPA, 2001). Resulting concentrations were reported in mg L21.

Dry Bedload and Sieved Sediment The sieved solid fraction was quantitatively transferred to an acid-washed, tarred porcelain evaporating dish with the aid of deionized rinse water. The dish was dried at 103 to 1058C overnight to a constant weight. The dried bedload sediment did not go through this process since it was delivered to the laboratory dry. The dried solids were then transferred to a clean, dry 250-mL polypropylene bottle. A 0.25- to 2.0-g portion of the dried solid was weighed and introduced directly into a 125-mL digestion vial and 10 mL of deionized water was added to rehydrate the sample. Although a 0.5-g sample is typically used for this type of solids analysis, we used a 2-g subsample where sample mass permitted. Using a 2-g sample should improve reproducibility, as errors associated with heterogeneities are reduced when subsampling sediments with sand-sized particles. The digestate protocols used for the aqueous samples were duplicated for the sieved solid samples. Resulting concentrations were reported in mg kg21.

RESULTS Selection of Sieve Size for Processing Whole Storm Water Samples At particle sizes larger than 250 mm, a positive bias in the SSC (Table 3) is observed in the churn split samples. For particles larger than 250 mm, the average percentage difference between the target SSC and the split water samples ranges from 26 to 85% Within each particle size

229

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

SELBIG ET AL.: IMPROVING ACCURACY OF SEDIMENT-ASSOCIATED CONSTITUENT CONCENTRATIONS

Fig. 2. Diagram of analytical techniques used to determine total recoverable zinc and copper concentrations on dry bedload, sieved sediment, and water samples from the churn splitter.

class, there is a positive relationship between percentage difference and SSC. A dramatic decrease in the degree of positive bias occurs for the samples in the particle size range of 63 to 125 mm. In this lower range, average percentage differences were 1 and 6% for suspendedsediment concentrations of 1000 and 5000 mg L21, respectively. This indicates that sieving a whole storm water sample with a 125-mm sieve before using the churn splitter will greatly reduce the positive bias in fine-fraction SSCs for samples with sand-sized particles. Subsampling precision was also degraded in water samples with larger particles. Coefficient of variation (COV) values increased from 0.01 for the 63- to 125-mm particles to 0.25 for the 500- to 1000-mm particles. Removing particles greater than 125 mm from a whole storm water sample before it is split will improve precision in the fine-fraction SSCs.

Both the precision and bias errors observed for the larger particle sizes indicate the churn splitter is not able to keep all the large particles in suspension. Since the churn paddle is unable to distribute sand grains in the splitter evenly, a gradient in concentration occurs inside the churn splitter where sand is more concentrated near the bottom (where the spigot is) than the top (Meade, 1985, cited in Capel and Larson, 1996). This accumulation of larger particles near the bottom of the splitter would also be highly variable depending on the particle size distribution, SSC, density of the particles, and the rate of mixing in the splitter. Any subsampling errors caused by the use of the churn splitter may also result in biased measurements of particle-associated contaminant levels, both on a volumetric and mass basis. Wet-sieving a whole storm water sample before splitting will require more effort to process a sample and

Table 3. Influence of churn splitter on suspended sediment concentrations (SSC) at differing initial concentrations and particle size ranges (Std. Dev., standard deviation; COV, coefficient of variation). Particle size mm 63–125

Mean Std. dev. COV 250–500

Mean Std. dev. COV 500–1000

Mean Std. dev. COV

Replicate number

SSC Target 5 1000 mg L Result 21

1 2 3 4 5

mg L 1000 1020 1020 1010 1010 1012 8 0.01 – – – – –

1 2 3 4 5

– – – – –

1 2 3 4 5

21

Percentage difference % 0 2 2 1 1 1

SSC Target 5 2000 mg L Result

21

SSC Target 5 5000 mg L

Percentage difference

Result

21

%

3340 3020 2800 2540 2150 2770 454 0.16 3250 2700 1850 2450 2350 2520 512 0.20

67 51 40 27 8 39

mg L 5350 5240 5300 5330 5190 5282 66 0.01 6470 8140 8600 7500 7760 7694 800 0.10 6600 12 300 11 000 8540 7740 9236 2354 0.25

mg L – – – – –

63 35 28 23 18 26

21

21

Percentage difference % 7 5 6 7 4 6 29 63 72 50 55 54 32 146 120 71 55 85

230

J. ENVIRON. QUAL., VOL. 36, JANUARY–FEBRUARY 2007

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

more careful tracking of the volumes in the splitter. The mass of the sediment on the sieve must be added to the mass of sediment determined by the SSC analysis on the split samples to compute the true SSC. A final SSC will require knowledge of the volume of water in the splitter.

Selection of Method for Total Recoverable Metals Analysis Differences between the dry bedload sediment and sieved solid copper and zinc concentrations tended to increase with increasing particle size (Table 4). One explanation could be greater heterogeneities of constituent concentration found within a subsample of larger particles. Since chemical analysis of dry solids requires only a 0.25- to 2.0-g subsample, there would be many more particles present in a 63- to 125-mm subsample than one composed of 500- to 1000-mm particles, thus creating a more homogenous mixture. In general, concentrations of copper and zinc determined in the sieved sediment were reasonably close to those measured in the dry bedload sediment for each particle-size fraction (Table 4). Therefore, we can assume the sieved solid concentration, when converted to mg L21, are reasonable reference concentrations for each method performed on the aqueous samples. Assuming 1 kg of water is equal to 1 L of water, the following equation can be used to convert the concentration of copper and zinc as a solid, in mg kg21, to a concentration as a liquid, in mg L21: Sm

1 1000 2 3 C 5

s

C1

[1]

V

where Cl 5 concentration of sieved solid represented in mg L21; Sm 5 mass of sieved solids after drying, in grams; Cs 5 concentration of sieved solid, in mg kg21; and V 5 volume of water sieved, in liters. Water density corrections are required should the SSC concentration exceed 8000 mg L21 (ASTM, 1999). Concentrations of copper and zinc determined in the two methods evaluated are compared to those for the sieved sediment in Table 5. The whole-bottle method of USEPA 200.7 came close to the reference values established from the sieved sediment concentration with percentage differences ranging from 26 to 28% for copper and 216 to 20% for zinc. However, the standard USEPA Method 200.7 consistently produced lower copper and zinc concentrations than the sieved solid for each particle-

size fraction. Though we cannot definitively determine the reason for the bias, it is likely that sedimentation or fractionation of particles within the bottle is the primary cause. Therefore, when taking an aliquot from a bottle, the potential for nonrepresentative sampling is great. Depending on how and where the aliquot is acquired from a sample container, the resulting concentrations could either be positively or negatively biased. An aliquot collected near the bottom of a container may contain a greater percentage of larger particles than an aliquot collected near the top. For this reason, the whole-bottle modification of USEPA Method 200.7 is recommended as the most appropriate technique for measuring metals concentration in an unfiltered water sample. The observed bias for the zinc and copper concentrations was less than the large positive bias observed for the SSCs. This may partly be explained by the effect of increasing metal concentrations with decreasing particle sizes (Table 5) (Bannerman et al., 1983). Smaller particles in each of the particle-size ranges tend to be transferred (aliquoted) with less error than the larger particles. Increasing the concentration of larger particles in the water sample has a significant effect on the SSCs, but a similar effect on metal levels is not observed because these large particles tend to have lower copper and zinc concentrations. Table 5 also demonstrates how a single value from the triplicate analysis can bias the average concentration with respect to the sieved solid reference. The italicized value for zinc in the 500- to 1000-mm particle-size fraction can likely be attributed to heterogeneities in larger sand-sized particles. Due to laboratory error, only two concentrations were recorded for copper in the 63- to 125-mm particle-size fraction. Addition of a third reliable value would have provided useful information to estimate the true value. While concentrations of copper and zinc appear somewhat more variable for the wholebottle analysis compared to the standard USEPA analysis, the improved accuracy of the whole-bottle analysis more than compensates for the difference in the precision.

Verification of Methods The main goal of many composite water quality samplers is to achieve a discharge-weighted concentration such that constituent fluxes can be accurately quantified. Many water quality studies use autosamplers to collect discrete water quality samples during a storm hydrograph that are later combined into a single composite sample. The presence and magnitude of a particular con-

Table 4. Comparison of dry bedload and sieved solid copper and zinc concentrations (RPD, relative percentage difference). Copper Particle size range mm 63–125 250–500 500–1000

SSC†

Raw bedload

21

mg L 2000 5000 5000 10000 5000 10000

† SSC, suspended sediment concentration.

Sieved solid 21

mg kg 41.2 4.3 8.1

37.0 41.7 3.7 3.6 6.3 6.0

Zinc Absolute RPD % 11 1 14 17 25 30

Raw bedload mg kg 151.0 17.4 20.2

Sieved solid 21

138.3 156.7 19.7 17.9 30.2 29.9

Absolute RPD % 9 4 13 3 40 36

SELBIG ET AL.: IMPROVING ACCURACY OF SEDIMENT-ASSOCIATED CONSTITUENT CONCENTRATIONS

231

Table 5. Triplicate concentration values for two sediment-associated metals using the whole-bottle and standard USEPA methods for aqueous samples (RPD, relative percentage difference). Copper

Zinc

USEPA 200.7 Method

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

Particle size fraction mm 63–125

SSC†

Aliquot 53 49 53 52 237% 111 114 72 99 2102% 12 11 11 11 256% 11 10 11 11 2216% 12 11 13 12 2142% 16 16 15 16 2245%

Mean RPD 5000 Mean RPD 10 000

500–1000

USEPA 200.7 Method Sieved solid mg L

2000 Mean RPD 5000

250–500

Whole bottle

Mean RPD 5000 Mean RPD 10 000 Mean RPD

70 69 60 67 26% 360 194

68 74 71 71 190 209 202 200

277 28% 23 24 26 24 27% 38 62 40 47 28% 34

19 16 18 18 35 34 32 34 29 29 29 29

31 33 11% 56 71 77 68 21%

59 53 50 54

Aliquot

Whole bottle

Sieved solid

218 204 225 216 222% 606 553 508 556 235% 45 33 35 38 2147% 52 57 52 54 2212% 47 47 57 50 2180% 55 57 58 57 2369%

272 291 291 285 7% 698 725

271 252 269 264

21

712 26% 123 88 133 115 19% 201 233 191 208 20% 135 126 131 28% 141 246 299 229 216%

764 740 754 753 111 90 78 93 159 177 166 167 148 175 100 141 271 247 280 266

† SSC, suspended sediment concentration.

taminant in a composite sample is represented by an event mean concentration (EMC) for the runoff period. Wet-sieving separates a whole storm water composite sample into aqueous and solid phases containing one or more particle-size fractions. Summing individual concentrations of each phase and fraction, after unit conversion, results in a single EMC. Using the methods described in this report, a reference sample of unfractionated composite bedload sediment was prepared in the laboratory to test the accuracy of summing individual concentrations to determine an EMC. The prepared sample was wet-sieved into three particle size fractions of greater than 250, 125 to 250, and 63 to 125 mm. The remaining aqueous solution that passed through the sieves contained only particles less

than 63 mm. Each sieved fraction and aqueous sample was submitted to the WSLH for analysis of total copper and zinc using the whole-bottle method of USEPA 200.7. A similar matrix of dried bedload sediment containing the same proportions of particle size fractions as the whole storm water sample was also submitted to the WSLH to represent the target EMC. All analyses were performed in triplicate. The results of the test are detailed in Table 6. After converting the concentrations of sieved solids from mg kg21 to mg L21 using Eq. [1], each fraction was added to the aqueous concentration to compute a measured EMC. This value was then compared to the reference EMC measured in the composite bedload sample. Both total copper and zinc EMCs were slightly greater,

Table 6. Triplicate concentration values for two sediment-associated metals using the whole-bottle USEPA method for aqueous samples. Sieved solid by particle size Replicate number

.250

125–250

63–125

Aqueous

EMC Measured (sieved solid 1 aqueous)†

EMC Reference (composite bedload)

mg L21 Total Copper 1 2 3 MEAN Total Zinc 1 2 3 MEAN

Percentage difference %

0.015 0.014 0.012 0.013

0.044 0.039 0.041 0.041

0.023 0.021 0.021 0.022

0.021 0.019 0.026 0.022

0.102 0.093 0.100 0.098

0.106 0.094 0.086 0.095

23 21 14 3

0.058 0.048 0.050 0.052

0.175 0.171 0.171 0.172

0.067 0.082 0.078 0.076

0.088 0.091 0.094 0.091

0.387 0.392 0.393 0.391

0.354 0.336 0.348 0.346

9 14 12 11

† EMC, event mean concentration.

232

J. ENVIRON. QUAL., VOL. 36, JANUARY–FEBRUARY 2007

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

on average, than those measured in the composite bedload by 3 and 11%, respectively. The agreement between the measured and reference event mean concentrations supports the validity of the process of wet-sieving whole storm water samples for improved accuracy of sedimentassociated constituent concentrations.

CONCLUSION Limitations associated with churn and cone splitters can be overcome by wet-sieving sand-sized particles from the aqueous portion of a whole storm water sample. This physical separation increases the efficiency of churn splitters and improves the overall accuracy of the constituent-concentration data. Once sand-sized particles are separated from the sediment-water mixture, they can be analyzed independently using the same analytical methods as the parent aqueous sample. The process of transferring sieved solids from the nylon sieve to the sample container had negligible effects on the measured copper and zinc concentrations. Analysis of the deionized water used after transferring solids indicated no leaching of metals from the solid particles into the residual water. Digesting the entire contents of a sample container reduced bias normally associated with taking an aliquot from a sample containing sand-sized particles. Of the two analytical methods used to determine metals concentrations in an aqueous sample, those using an acid digestion on the entire contents of the sample container (the whole-bottle version of USEPA Method 200.7) produced concentrations that were closer to the sieved sediment concentration. The presence of outliers in the concentration data using the whole-bottle version of USEPA Method 200.7 suggests difficulties in maintaining a high level of precision. The standard USEPA Method 200.7 produced results with high precision but lacked accuracy. The difference in precision is a function of subsampling an original sample. By taking an aliquot near the top of the sample container, particles that fall out of suspension quickly may be omitted from the sample analyzed. Similarly, an aliquot taken from the bottom of a sample container might acquire a sediment-enriched subsample. Therefore, in obtaining an aliquot from a sample container for laboratory analysis, the likelihood of nonrepresentative sampling is great. When compared to the corresponding sieved solid reference concentrations, concentrations of copper and zinc using the whole-bottle version of USEPA Method 200.7 displayed greater accuracy over the standard USEPA Method 200.7. Individual concentrations of sediment-associated constituents from a whole storm water sample separated into multiple particle size fractions can be summed to determine an event mean concentration. Increasing the accuracy of constituent concentrations in sieved sedi-

ment would improve event mean concentrations for each storm event. This could result in improved concentration data that is required for environmental managers in their decision-making process. REFERENCES ASTM. 1999. D 3977-97, Standard test method for determining sediment concentration in water samples. Annual book of standards. Water and Environmental Technology. Vol. 11.02. ASTM, West Conshohocken, PA. Bannerman, R.T., K. Baun, M. Bohn, P.E. Hughes, and D.A. Graczyk. 1983. Evaluation of urban nonpoint source pollution management in Milwaukee County, WI. Vol. 1. Water Planning Division, PB84114164. USEPA, Washington, DC. Bannerman, R.B., A.D. Legg, and S.R. Greb. 1996. Quality of Wisconsin storm water, 1989–94. U.S. Geol. Surv. Open-File Rep. 96458 USGS, Washington, DC. Capel, P.D., and S.J. Larson. 1996. Evaluation of selected information on splitting devices for water samples: Water Resources Investigation Rep. 95-4141. USGS, Washington, DC. Furumai, H., H. Balmer, and M. Boller. 2002. Dynamic behavior of suspended pollutants and particle size distribution in highway runoff. Water Sci. Technol. 46(11–12):413–418. Gray, J.R., G.D. Glysson, L.M. Torcios, and G. Schwartz. 2000. Comparability of suspended-sediment concentration and total suspended solids data. Water Resources Investigation Rep. 00-4191. USGS, Washington, DC. Guy, H.P. 1969. Laboratory theory and methods for sediment analysis. Techniques of water resources investigations. Book 5. USGS, Washington, DC. Horowitz, A.J., T.S. Hayes, J.R. Gray, and P.D. Capel. 1997. Selected laboratory tests of the whole storm water sample splitting capabilities of the 14-liter churn and the Teflon cone splitters, U.S. Geological Survey Office of Water Quality Tech. Memorandum No. 97.06. USGS, Washington, DC. Horwatich, J.A., S.R. Corsi, and R.T. Bannerman. 2004. Effectiveness of a pressurized stormwater filtration system in Green Bay, Wisconsin: A study for the environmental technology verification program of the U.S. Environmental Protection Agency. U.S. Geological Survey Scientific Investigations Rep. 2004-5222. USGS, Washington, DC. Meade, R.H. 1985. Suspended sediment in the Amazon River and its tributaries in Brazil during 1982–84. U.S. Geol. Surv. Open-File Rep. 85-492 USGS, Washington, DC. USEPA. 1986. Quality criteria for water 1986. EPA 440/5–86–001. USEPA, Washington, DC. USEPA. 1993. Test methods for evaluating solid wastes. Physical/ chemical methods. SW-846 Method 3005A. 3rd ed. Office of Solid Waste and Emergency Response, Washington, DC. USEPA. 2001. Method 200.7: Trace elements in water, solids, and biosolids by inductively coupled plasma-atomic emission spectrometry. Rev. 5. EPA-821-R-01-010. USEPA, Washington, DC. U.S. Geological Survey. 1999. National field manual for collection of water-quality data: Processing of water samples. Techniques of Water-Resource Investigation, 09-A5. USGS, Washington, DC. U.S. Geological Survey. 2001. Collection and use of total suspended solids data. Office of Water Quality Technical Memorandum No. 2001.03. USGS, Washington, DC. U.S. Geological Survey. 2003. Results of the U.S. Geological Survey’s analytical evaluation program for standard reference samples. U.S. Geological Survey Open-File Report 03-261. USGS, Washington, DC. Waschbusch, R.J. 2003. Data and methods of a 1999–2003 street sweeping study on an urban freeway in Milwaukee County, Wisconsin. U.S. Geological Survey Open File Report 03-93. USGS, Washington, DC.