Quantifying Nutrient Removal by Enhanced Street Sweeping

In the early 2000s, two major federal programs offered promise of improved water quality within and downstream of cities. First, EPA’s stormwater program moved stormwater drainages into the National Pollutant Discharge Elimination System (NPDES), subjecting them to regulation. Most cities with populations greater than 10,000 now have active municipal separate storm sewer system (MS4) programs. Second, EPA’s total daily maximum load (TMDL) program compelled states to develop plans to restore impaired surface waters.

So far, there is little evidence that these programs have resulted in widespread improvement of urban water quality, at least with respect to nutrients. As an example, about 140 lakes in the Minneapolis-St. Paul metropolitan region have been designated as “nutrient-impaired,” but only one has been delisted as the result of deliberate management. Some of the lack of lake response to changes in management may be due to legacy effects, such as recycling of phosphorus (P) from lake sediments, but there is growing concern that we are not achieving nutrient reduction goals. Moreover, cities have learned that the costs of their stormwater programs have been high, compelling them to think about improving the economic efficiency of stormwater control measures. Hence, a decade into the TMDL and stormwater programs, some cities have taken a new look at a very old street management practice: street sweeping.

Early analysis of the effect of street sweeping conducted during EPA’s National Urban Research Program (NURP) concluded that street sweeping was not effective at reducing event mean concentrations of P in stormwater (USEPA 1983). Even today, street sweeping has been relegated to the diminutive category of “housekeeping practices” in MS4 programs, not to be taken too seriously. This attitude is changing. The influence of leaf litter and organic matter on nutrient loads in street sediments has often been noted in recent years (Waschbusch et al. 1999, Seattle Public Utilities 2009, Law et al. 2008, Sansalone and Rooney 2007, Minton and Sutherland 2010). In a way, this is obvious. Surely the leaves collecting above the storm grate (Figure 1) are a source of nutrients!

The Prior Lake Street Sweeping Experiment
In 2009, the city of Prior Lake, MN, a leafy southwestern suburb of the Twin Cities, started an ambitious project to quantify nutrient removal by street sweeping, partnering with the University of Minnesota through an EPA 319 grant. The study had several defining characteristics that made it unique. First, it was conducted as a factorial design, with two treatments: frequency of sweeping (once, twice, and four times per month) versus tree canopy cover (low, medium, and high, although we later evaluated findings based on a continuous gradient of canopy covers). Second, rather than attempt to measure the effects of sweeping on stormwater loadings (as was done in the NURP study) we measured solids, nitrogen (N), and P in swept material, allowing us to estimate sweeping load recoveries directly.

Figure 2. Street sweeping research team. Left to right: Ross Bintner, Chris Buyarksi, Sarah Hobbie, and Paula Kalinosky

Furthermore, whereas many studies of street sweeping discarded coarse material (e.g., leaves and seeds) that don’t pass through a 2-millimeter sieve (e.g., Townsend et al. 2002, Rochfort et al. 2009), we collected coarse organic material trapped above the sieve and analyzed it separately. Third, unlike many studies that were of short duration or ceased before autumn leaf fall (e.g., Selbig and Bannerman 2007, Vaze and Chiew 2004), sweeping was conducted from just after snowmelt up until the first snowfall. Finally, we kept detailed records of costs–for labor, equipment maintenance, and fuel–and included capital depreciation for the sweepers.

Sweeping started in August 2010 and continued through July 2012 using a Tymco Model 600 regenerative air sweeper. Three hundred ninety-two sweeping samples were collected and stored frozen until analysis by the research team (Figure 2). Samples were sieved to isolate fines (passing through a 2-millimeter sieve) and coarse material (not passing). The coarse material was then floated in 3 liters of water. Material that floated to the surface was collected and termed coarse organic matter (COM). Heavier material that had been attached to the COM settled to the bottom; this material was dried and re-sieved. The fraction that passed through the sieve was added to the original fines fraction, and coarser material (mostly sand and pebbles) was weighed but not analyzed for nutrients. Wet and dry weights of the original samples and components were recorded. The fines and the COM were analyzed for carbon (C), N, and P. The water used for the flotation step was analyzed for nutrients, and the mass of C, N, and P, but nutrients dissolved in the flotation step were only 1 to 2% of the total nutrient mass of swept material. High-resolution land cover for the city of Prior Lake was developed by the University of Vermont Spatial Analysis Laboratory ( www.uvm.edu/rsenr/sal ) using object-based image analysis that combines satellite imagery and LiDAR data (Kilberg et al. 2011). This enabled us to estimate percentage of canopy cover over streets for each sweeping route. Additional methodological details are reported in Kalinosky et al. (in process a).

Figure 3. Trend in percent dry solids, N, and P in the COM fraction of sweepings for the two-year sweeping period

What Did We Find?

The composition of fines and COM was different. As one might expect, the chemical composition of COM was quite different from that of the fine fraction (“street dirt”) (Table 1). On average, the percentage of organic matter and the N content was about 10 times higher for COM than for fines; the P content of COM was about 2.4 times that of fines.

Figure 4. Cost efficiency for P load removal by sweeping for the two-year sweeping study for Routes L4 and H2

Reducing P loads by Street Sweeping Can Be Very Cost Efficient. The cost of sweeping, including labor, fuel, maintenance, and amortization of capital, was $23 per curb-mile swept. The cost efficiency of P removal varied depending on season and route (Figure 4). For the worst case (Route L4 with 5% street canopy swept four times per month), cost efficiency rose to $600 per pounds of P in mid-summer. For the most cost-efficient route, Route H2 (15% street canopy, swept twice per month), the cost of P removal dropped to less than $100 per pounds of P during spring and fall.

Tools for Planning and Implementing Enhanced Street Sweeping          

Planning Level Tool. Findings from the Prior Lake study were used to develop a spreadsheet planning tool to aid public works and streets departments that want to plan enhanced street sweeping. The spreadsheet tool is based on a multiple regression equation that includes percentage of tree canopy, frequency of sweeping, and time of year (Kalinosky et al. in process b). The spreadsheet tool allows users to enter two types of information: fixed baseline data (average street percent canopy along each route being evaluated, the curb length of each route, and the cost of sweeping per curb mile) and sweeping frequency per route, which can be varied in scenarios. The spreadsheet calculates expected pounds of N, P, and solids removed per route and across all routes, and calculates the cost efficiency for P removal for each month and for the entire year. The user can then vary sweeping frequency along each route to informally optimize sweeping, asking one of two questions: 1) How much P can be removed with a given sweeping budget? or 2) How much would it cost to reduce the P load by a fixed amount? The latter question might be asked in the context of TMDL load reduction goals.

Note: The actual measured value for the P load from the study was 505 pounds.

Because the spreadsheet planning tool is based upon findings from Prior Lake, predictions can be made only for cities with similar types of trees (north-temperate deciduous trees that drop leaves in autumn), with over-street percent canopy up to about 20%. We plan to continue sweeping studies in other cities with much higher canopies; as we do so, we will expand the spreadsheet calculator.

Implementation Level Tool. During the implementation of enhanced street sweeping, most cities would want to calculate P load recoveries in their sweeping loads. Our research analytical protocol is more rigorous (and expensive) than most cities would want to adopt, at least for a prolonged period. Hence, we asked the question, What is the least amount of information a city would need to accurately predict sweeping P load recoveries? To answer this question, we used a statistical approach (fivefold cross validation) to compare estimated recoverable P loads with the measured P recoveries (the “true” value) for each route with P loads estimated with varying specificity of information. From simplest to more complex, we examined 1) curb miles swept (determined by odometer or truck-mounted GPS),

2) the fresh (wet) weight of the sweeping load (determined by a truck scale), and 3) the dry weight of the sweeping load. Remarkably, predictions of P load recovery using these varying techniques were all within 10% of the measured P load (Table 2). This means that once a city developed several simple regression relationships for its streets, P load recovery could be monitored with very simple, inexpensive measurements, such as fresh weight of the sweeping load.

Conclusion and Knowledge Gaps
This research shows that spatially and temporally targeted enhanced street sweeping can be a cost-efficient way to prevent nutrients from entering storm drains. No city is likely to want to sweep four times per month year round, as we did in the “high-frequency” treatment. More likely, cities will find that it is more effective (in terms of P load removal, pounds P per curb-mile) and cost efficient (cost per pound P removed) to sweep more often along routes with high canopy cover and less along routes with sparser canopy cover. Even for high canopy streets, multiple sweepings per month will probably be justified only during the spring and autumn.

There are several knowledge gaps that need to be filled as cities move toward enhanced sweeping. First, the highest percentage of canopy cover in our study was only 20%, far lower than canopy covers in many older residential neighborhoods. Other studies are needed to extend this range. Second, most of the trees in Prior Lake were deciduous, so results cannot be extended to areas with markedly milder climates where tree leaf fall is greater due to higher tree growth. As we move toward greater reliance on street sweeping and other forms of nutrient source reduction, we also need to understand how reduced transport from vegetated landscapes to streets translates to stormwater quality at the end of the pipe and to water quality in the lakes and streams of our cities.

References
Kalinosky, P., L. Baker, S. Hobbie, R. Bintner. In process (a). “The Influence of Tree Canopy on the Character and Quantity of Solids Recovered Through Street Sweeping.”

Kalinosky, et al. In process (b). “Modeling Recovery of Solids and Nutrients Through Street Sweeping.”

Kalinosky, P., L. Baker, S. Hobbie. 2013. Quantification of Nutrient Removal by Street Sweeping: The Prior Lake Street Sweeping Project. International Low Impact Development (LID) Symposium, St. Paul, MN, August 8–13, 2013.

Kilberg, D., M. Martin, and M. Bauer. 2011. Digital Classification and Mapping of Urban Tree Cover: City of St. Paul. Department of Forest Resources, University of Minnesota.

Law, N. L., K. DiBlasi, and U. Ghosh. 2008. Deriving Reliable Pollutant Removal Rates for Municipal Street Sweeping and Storm Drain Cleanout Programs in the Chesapeake Bay Basin. Center for Watershed Protection, EPA project CB-9732222-01.

Minton, G., and R. C. Sutherland. 2010. “Street Dirt: A Better Way of Measuring BMP Effectiveness.” Stormwater, March/April 2010.

Rochfort, Q., K. Exall, J. P’ng, V. Shi, V. Stevanovic-Briatico, S. Kok, and J. Marsalek. 2009. “Street Sweeping as a Method of Source Control for Urban Stormwater Pollution.”Water Quality Research Journal of Canada 44:48–58.

Sansalone, J. J., and R. Rooney. 2007. Assessing the Environmental Benefits of Selected Source Control and Maintenance Practices for MS4 Permits. Florida Stormwater Association.

Seattle Public Utilities. 2009. Seattle Street Sweeping Pilot Study, Monitoring Report. Seattle, WA: Seattle Public Utilities in association with Herrera Environmental Consultants.

Selbig, W. R., and R. T. Bannerman. 2007. Evaluation of Street Sweeping as a Stormwater-Quality-Management Tool in Three Residential Basins in Madison, WI. US Geological Survey, Report #2007-5156. 103 pp.

Sutherland, R. C. “Street Sweeping 101.” 2011. Using Street Sweepers to Improve Water and Air Quality. Stormwater, January/February 2011.

Townsend, T., Y. Jang, P. Thurdekoos, M. Booth, P. Jain, and T. Tolaymat. 2002. Characterization of Street Sweepings, Stormwater Sediments, and Catch Basin Sediments in Florida for Disposal and Reuse. University of Florida: Florida Center for Solid and Hazardous Waste Management.

United States Environmental Protection Agency. 1983. Results of the National Urban Runoff Program, Vol. 1, Final Report. Washington, DC: US Environmental Protection Agency, Water Planning Division, Report #PB84-185552.

Vaze, J., and F. H. S. Chiew. 2004. “Nutrient Loads Associated with Different Sediment Sizes in Urban

Stormwater and Surface Pollutants.” Journal of Environmental Engineering 130: 391–96.

Waschbusch, R. J., W. R. Selbig, and R. T. Bannerman. 1999. Sources of Phosphorus in Stormwater and Street Dirt From Two Urban Residential Basins in Madison, Wisconsin, 1994–95. Madison, WI: US Department of the Interior, US Geological Survey, Water Resources Investigation Report 99-4021.

About the Author

Lawrence Baker, Paula Kalinosky, Sarah Hobbie, Ross Binter, and Chris Buyarksi

Lawrence Baker is a research professor in the Ecological Engineering Group in the Department of Bioproducts and Biosystems Engineering at the University of Minnesota. Paula Kalinosky is a water-quality consultant at Emmons and Olivier Resources Inc. Sarah Hobbie is a professor in the Department of Ecology, Evolution, and Behavior at the University of Minnesota and a resident fellow in the Institute on the Environment at the University of Minnesota. Ross Bintner is a civil engineer with the City of Edina, MN. Chris Buyarksi is a junior scientist/laboratory manager in the Departments of Ecology, Evolution, and Behavior, and Forest Resources at the University of Minnesota.