Environ. Sci. Technol. 2008, 42, 511–516
Measuring Solids Concentration in Stormwater Runoff: Comparison of Analytical Methods SHIRLEY E. CLARK* AND CHRISTINA Y. S. SIU School of Science, Engineering, and Technology, Penn State Harrisburg, 777 West Harrisburg Pike, Middletown, Pennsylvania 17057
Received February 13, 2007. Revised manuscript received September 04, 2007. Accepted September 13, 2007.
Stormwater suspended solids typically are quantified using one of two methods: aliquot/subsample analysis (total suspended solids [TSS]) or whole-sample analysis (suspended solids concentration [SSC]). Interproject comparisons are difficult because of inconsistencies in the methods and in their application. To address this concern, the suspended solids content has beenmeasuredusingbothmethodologiesinmanycurrentprojects, but the question remains about how to compare these values with historical water-quality data where the analytical methodology is unknown. This research was undertaken to determine the effect of analytical methodology on the relationship between these two methods of determination of the suspended solids concentration, including the effect of aliquot selection/ collection method and of particle size distribution (PSD). The results showed that SSC was best able to represent the known sample concentration and that the results were independent of the sample’s PSD. Correlations between the results and the known sample concentration could be established for TSS samples, but they were highly dependent on the sample’s PSD and on the aliquot collection technique. These results emphasize the need to report not only the analytical method but also the particle size information on the solids in stormwater runoff.
Introduction Suspended solids (SS) are one of the most common parameters measured in stormwater monitoring. Stream impairment studies identified siltation (sediment) as one of the primary causes of water-quality beneficial use impairment with urban runoff as one source. Phase I National Permit Discharge Elimination System permit holders, (large municipal separate stormwater system authorities or MS4s), typically are required to monitor SS as part of their permitting process. A review of the National Stormwater Quality Database, available at http://unix.eng.ua.edu/∼rpitt/Research/ms4/ mainms4.shtml, indicates that SS (reported as total suspended solids [TSS]) were measured at 90% of the Phase I MS4 sites surveyed (3390 observations/3770 sites). Measurement of stormwater solids also is important because many pollutants in runoff (metals, inorganic nutrients, microorganisms, and trace organic compounds) often partition to stormwater runoff solids. Knowledge of pollutant partitioning allows for the prediction of fate and transport * Corresponding author phone: (717) 948-6127; fax: (717) 9486580; e-mail:
[email protected]. 10.1021/es070371g CCC: $40.75
Published on Web 12/06/2007
2008 American Chemical Society
of both the solids and pollutants, as well as the prediction of sedimentation treatment efficiency. Typically, pollutants (e.g., heavy metals and polycyclic aromatic hydrocarbons [PAHs]) associate with the finer particles. At higher pollutant concentrations, however, the association favors the larger particles in the runoff (1). Furumai et al. (2), found a highway drainage system containing high SS concentrations also showed greater levels of heavy metals and PAHs. Much of the past research on stormwater particulates has used the TSS methodology (from the International Organization of Standards [ISO], the U.S. Environmental Protection Agency [U.S. EPA], or the Standard Methods [SM] for the Examination of Water and Wastewater) to quantify these solids. These TSS methods are based on the analysis of an aliquot or subsample of the original sample, and the methods outline the required technique. In contrast, the U.S. Geological Survey (USGS), in much of their monitoring efforts, has used and promoted a different methodology to quantify the amount of particles in a water sample: the suspended sediment concentration (SSC) method, a whole-sample analysis method. They argue that the traditional TSS methods miss many of the larger solids often found in stream waterquality samples. However, others have argued that the larger particles quantified in the SSC method are not those transported large distances in storm sewers and streams. This disagreement over the appropriate methodology has led to several research teams analyzing for solids using both methods, especially for manufactured stormwater treatment device certification in the U.S. This dual analysis of the same solids ensures that comparisons can be made to the collected data, regardless of the methodology used to collect the comparison data set. Investigations of the correlations between the SS analytical methods (3) have shown that these two methods often give varying results (4, 5). All groups, including this research group, showed TSS measurements were generally lower than the SSC measurements with a TSS concentration bias ranging from 20 to 40% lower than the corresponding SSC concentration for identical samples analyzed with different methodologies. This led some researchers to argue that the simplest solution to this dichotomous information involving SS in stormwater runoff is to develop correlations between TSS and SSC that can be applied to the appropriate data set, as needed, to ensure that the comparisons and evaluations are valid, regardless of the analytical method used. Because of concerns that a single correlation between TSS and SSC could not be established because of variability of the analytical techniques and of particle size distributions (PSDs) found in runoff samples from different locations, this research team investigated the three most-common methods for SS analysis using triplicate samples of known concentrations of two silica-based standards (d50 ) 100 and 500 µm). The purpose was to evaluate the analytical results of these methods, given two well-characterized solids mixtures and a consistent analytical filter type and pore size.
Methods Two common TSS methods are U.S. EPA Method 160.2 (6) [identical in technique to ISO 11923 (7)] and SM Method 2540D (8). These solid-characterization methods differ in two important ways: (1) the specification, or lack thereof, of the nominal pore size of the filter and (2) the method used to collect the sample aliquot. Both TSS methods are documented to have problems in the capture of the larger particles found in simulated stormwater. For example, SM VOL. 42, NO. 2, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
511
TABLE 1. Comparison of Three TSS/SSC Analytical Methods method requirements
U.S. EPA TSS (160.2) and ISO (11923)
SM TSS (2540D)
filter nominal pore size
not specified
0.85 and >0.94 for TSSlower,SiO2 for SM but >0.99 for both EPA sets. The regression equations explained the majority of the sample-set variability. The plume of solids traveling along the sample jar bottom made obtaining representative sample aliquots with a pipet difficult.
FIGURE 4. TSS Analysis (SM 2540D): Effect of PSD and aliquot collection location. See SI-1 for regression equations. As a result, standard deviations were high, and R2 values were lower than those for the “shake-and-pour” method. At mid-depth sampling locations, the R2 values for the SiO2-only standard were similar to those seen for the “shakeand-pour” aliquot technique, unlike those for the sand–SiO2 mixture. This indicated that the lack of large particles in the mixture reduced the variability in obtaining a representative aliquot. When the results for each mixture were compared, the impact of the standard’s PSD was statistically significant, regardless of sampling location. All equations were statistically differentiable from the TSS ) [known] equation, which is a true representation of the concentration. For SiO2, recoveries were approximately 75–80%, while for the sand– SiO2 mixture, the recoveries were approximately 40–45%. Similar to the results seen using the “shake-and-pour” methods, the PSD of the sample had an effect on the sample recoveries. As described above, the samples with the larger d50 values (sand–SiO2) had higher recoveries (greater than 1.0 in most samples) when the samples were taken near the bottom of the bottle but a smaller recovery when the sample was taken higher in the water column. This effect also was seen for the SiO2-only samples, but it was not quite as large. Effect of Subsampling Location. Results at the mid-depth sampling location showed that the sampling location is important when using SM 2540D to analyze for SS containing a d50 of ∼100 or 500 µm. Two particular results were noted during this series of tests. First, sampling near the bottom of the bottle recovered solids up to 150% of the known concentration in the aliquot. The recoveries were less than 100% when sampling at the mid-depth location. Recoveries were similar for both standards when the sample was taken near the bottle bottom; however, they were statistically significantly different when the sampling location was the midpoint. Second, the confidence intervals were much larger (reflecting larger standard deviations) when the samples were collected near the bottom of the bottle. This confirmed the visual observation that, when the samples were mixed on the stir plate, the larger particles in the mixture did not mix evenly in the water column either vertically or horizontally (across the bottle from vortex to wall). Therefore, solids captured in the widebore pipet at any given subsample collection time were reflective of where the solids “plume” was located compared to the pipet. The solids at the mid-depth sampling location were the smaller particles and were more likely to be well-
mixed in the bottle. Analytical results were more consistent, then, at the mid-depth sampling location. However, this location impacted the method recoveries, especially in samples where a greater fraction of the solids was the larger particles that were not well-mixed in the water column (predicted settling times for 500 µm particles in these bottles are ∼1 minute, and those for 100 µm silicas are 26 min). This confirmed the results seen in these experiments that accurately capturing and measuring the larger sands in an aliquot are difficult. The maximum median particle size for this experiment was 500 µm; this size range and greater are classified as coarse sand or gross solids. Researchers have reported various median sizes for the particles in urban runoff: ∼30 µm in stormwater (19), e10 µm in erosional runoff (20), and >1000 µm in locations similar to highways (21). Stormwater PSD is affected by “. . .source area type, rain conditions, upstream controls,. . .rain characteristics, soil type, and on-site erosion controls” (19). The samples used in this series of tests, although on the high end, were within the ranges reported in the literature to compare the effectiveness of the analytical technique to represent the entire range of solids seen in stormwater. The idea of sampling for both TSS and SSC is still open to debate, given that the solids analyzed by either method are not comparable to each other across a wide range of conditions. However, having both sets of data will allow other investigators to compare their results both to current samples analyzed by only a single method and to historical data analyzed by TSS. While the SSC method produces the most “accurate” and “precise” results, this was not the method commonly employed in the historical literature. To compare current results to historical data, samples need to be analyzed by the TSS method used in the historical analyses. It is incumbent upon the analyst, however, to realize and understand the increase of error from the start of sampling to the end of analysis no matter which analysis is performed. It also is incumbent upon the researcher who is comparing results to historical data to ensure that the analytical methodology is replicated to the maximum extent possible. This research project, one of many steps in solids’ concentration determination, specifically outlined the inconsistencies and potential for misinterpretation between these solids analysis methods (such as the filter pore size, which, if specified, differs between methods and between researchers, and the method of obtaining subsamples for filtering). This research also confirmed that, whenever possible and especially when evaluating stormwater treatment device performance, the mass balance method for calculating solids concentration (when the mass of solids added is known) is the most accurate. In addition, when the influent solids concentration is not known, the whole-bottle analysis method is able to reflect that concentration well, with high recoveries, although not 100%, and results that are statistically indistinguishable from the known solids addition. This research also showed that correlations can be established between the methods and between known additions and the analytical results, but they are dependent on the PSD of the solids in the sample. PSD analysis typically was not performed on historical samples. Therefore, accurate correlations for historical comparison may not be possible. For the future, it is incumbent upon stormwater researchers to specify precisely their methodology. Not only will the data be more comparable, but it also will give a better indication of the type of errors involved in the entire analysis. Future research and monitoring either should measure both TSS and SSC (for historical comparison and determination of the “actual” concentration) or should report PSD information so that the appropriate correlations can be applied between TSS and SSC based on the analytical methodology. VOL. 42, NO. 2, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
515
Acknowledgments The authors express their sincere appreciation to all the current and former graduate and undergraduate students who contributed to this project through their work: James Elligson, Brad Mikula, Christopher Roenning, and Julia Hafera. Their work was instrumental in developing the questions upon which this paper is based.
(9) (10)
Supporting Information Available Tables of t-test values and regression equations used in slope coefficient comparisons. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Morquecho, R. Tracking stormwater heavy metal speciation changes from source areas to receiving waters, Ph.D. Dissertation, University of Alabama, Tuscaloosa, AL, 2005. (2) Furumai, H.; Balmer, H.; Boller, M. Dynamic behavior of suspended pollutants and particle size distribution in highway runoff. Water Sci. Technol. 2002, 46, 413–418. (3) Clark, S. E.; Pitt, R. Comparison of stormwater solids analytical methods for performance evaluation of manufactured treatment devices. J. Environ. Eng. In press. (4) Gray, J. R.; Glysson, G. D.; Turcios, L. M.; Schwarz, G. E. Comparability of Suspended-Sediment Concentration and Total Suspended Solids Data; U.S. Geological Survey, Water-Resources Investigation Report 00-4191; U.S. Geological Survey: Denver, CO, 2000. (5) Guo, Q. Development of adjustment and scaling factors for measured suspended solids removal performance of stormwater hydrodynamic treatment devices. World Environmental & Water Resources Congress 2005: Impacts of Global Climate Change; Anchorage, AK, May 15–19, 2005; American Society of Civil Engineers: Reston, VA, 2005. (6) Method 160.2: Total Suspended Solids (TSS) (Gravimetric, Dried at 103–105°C); U.S. Environmental Protection Agency: Washington, D.C, 1999 (Revised November 1999). (7) Water Quality––Determination of Suspended Solids by Filtration through Glass-Fibre Filters, 1st ed. ; ISO 11923:1997; International Organization for Standardization: Geneva, Switzerland, 1997. (8) Standard Methods for the Examination of Water and Wastewater, 20th ed.; Clesceri, L. S., Greenberg, A. E., Eaton, A. D., Eds.;
516
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 2, 2008
(11) (12) (13) (14) (15) (16) (17)
(18) (19) (20) (21)
American Public Health Association, American Water Works Association, Water Environment Federation: New York, 1999. Method D 3977-97(B): Standard Test Methods for Determining Sediment Concentration in Water Samples. American Society for Testing and Materials: West Conshocken, PA, 1997. The Technology Assistance Reciprocity Partnership (TARP) Protocol for Stormwater Best Management Practice Demonstrations; New Jersey Department of Environmental Protection: Trenton, NJ, 2003; http://www.state.nj.us/dep/dsr/bscit/ Documents.htm (accessed Jan 2007); adopted by CA, MA, MD, NJ, PA, and VA; available with the NJ Amendment to the TARP Tier II Protocol. U.S. Silica Company. SIL-CO-SIL Ground Silica. http://www. u-s-silica.com/scs.htm (accessed Feb 2007). SigmaPlot, version 9.0; Systat Software Inc.: San Jose, CA, 2004. Gy, P. Sampling for Analytical Purposes; John Wiley & Sons LTD: West Sussex, England, 1998; English edition. Gy, P. M. In Sampling of Particulate Materials: Theory and Practice; Development in Geomathematics 4; Elsevier Scientific: Amsterdam, 1979. Pitard, F. F. Pierre Gy’s Sampling Theory and Sampling Practice: Heterogeneity, Sampling Correctness, and Statistical Process Control, 2nd ed.; CRC Press: Boca Raton, FL, 1993. Pitt, R.; Clark, S. E. Sample processing protocols to ensure subsample representativeness. J. Environ. Eng. Submitted for publication. Khambhamettu, U.; Clark, S. E.; Pitt, R. Protocols for quantifying solids removal performance during controlled testing of manufactured treatment devices. World Environmental & Water Resources Congress 2007: Restoring Our Natural Habitat; Tampa, FL, May 15–19, 2007; American Soceity of Civil Engineers: Reston, VA, 2007. Wuensch, K. L. Comparing correlation coefficients, slopes, and intercepts, http://core.ecu.edu/psyc/wuenschk/docs30/ CompareCorrCoeff.doc (accessed Jan 2007), 2006. Burton, G. A.; Pitt, R. E. Stormwater Effects Handbook: A Toolbox for Watershed Managers, Scientists, and Engineers; CRC Press: Boca Raton, FL, 2002; p 452. Hittman Associates. Methods to Control Fine-Grained Sediments Resulting from Construction Activity; Pb-279092; U.S. EPA: Washington, DC, 1976. Cristina, C.; Tramonte, J.; Sansalone, J. J. A granulometry-based selection methodology for separation of traffic-generated particles in urban highway snowmelt runoff. Water, Air, Soil Pollut. 2002, 136, 33–53.
ES070371G
