S.M. Stai and Westwood Professional Services
Vegetated buffers, also known as wetland buffers, riparian buffers, buffer strips, or vegetated filter strips, are areas of vegetation between developed land and surface water. Buffers are established or protected for many different reasons, including hydrologic event modification, aquatic and wildlife habitat protection, aesthetic value, and open space preservation. The potential for buffers to aid in water quality protection has gained particular attention in Minnesota in recent years.
The primary purpose of this study was to determine the buffer width that represents the point of “diminishing returns.” Specifically, the objective was to assess how a vegetated buffer’s sediment and phosphorus retention capacity changes as a function of downslope distance from the point of entry by residential stormwater runoff. A secondary purpose was to determine the effect that buffer slope has on pollutant retention. The focus on residential land use reflected the interest of the project sponsors: the Metropolitan Council, the Builders Association of the Twin Cities, and the National Association of Home Builders.
Study site and plot set-up
The study site was located behind a commercial office building in Eden Prairie, Minnesota. The site sloped away from the building for approximately 300 feet (91.4 m) toward Purgatory Creek (Figure 1). A review of historical aerial photography indicated that land use at the site was primarily agricultural until construction of the current building took place in 1997. The site was selected for four main reasons. First, because the sloped area had been seeded with a native prairie mix nine years prior to the study, the vegetation was well established and was representative of mature buffers that would typically be found at residential subdivisions in the Twin Cities Metropolitan Area (TCMA). Second, the site’s native soil was the Lester-Malardi complex, a well-drained loam and a state soil common to the TCMA. Third, the site contained slopes at varying degrees up to 50 percent, allowing for examination of buffer effectiveness as a function of slope as well as width. Finally, stormwater runoff generated by the impervious surfaces of the building roof and parking lot was captured and conveyed through pipes to a holding pond, thereby preventing overland flow (and hence interference) of naturally generated commercial runoff into the study area.
Four plots were established on the site, with each plot representing a different slope between 2:1 and 5:1 (Figure 1). Each plot consisted of three transects, each of which was 8 feet (2.4 m) wide and separated from adjacent transects by at least 2 feet (0.6 m). Transect locations were selected to meet the slope requirements for the plot, to be at least 40 feet (12.2 m) in length, and to exclude uneven ground and woody vegetation to the maximum extent possible. Each transect was outfitted with a runoff collector at four intervals (5, 10, 20, and 40 feet [1.5, 3, 6, and 12 m]) downslope from the head of the transect (Figure 2). The runoff collector consisted of a 6-inch (15.2-cm) PVC pipe cut to a 2-foot (0.6-m) length, capped at the bottom, and partially submerged in the ground (Figure 3). Each collector received runoff from a 2-foot (0.6-m) wide flow path, and the runoff was directed to the collector by means of a V-shaped piece of lawn edging called the diverter (visible in Figure 3, though mostly obscured by grass). A hole cut in the apex of the diverter was equipped with a mesh screen to prevent passage of debris and with a PVC coupler, which was connected to the runoff collector with Tygon tubing.
Synthetic runoff tests
Synthetic runoff testing (level 3 assessment) was selected for this study because of its advantages over monitoring. The synthetic runoff could be prepared with known concentrations of phosphorus and sediment and applied in known volumes under controlled, repeatable conditions without reliance on unpredictable natural rainfall.
The synthetic runoff system had three main components (Figures 4-5): a water tank for mixing and holding the synthetic runoff, an eductor manifold installed inside the tank to keep phosphorus dissolved and sediment in suspension during a trial, and a nozzle manifold placed at the head of the transect to deliver runoff in the form of laminar flow. The water tank was a 625-gallon round plastic livestock tank that measured 8 feet (2.4 m) by 2 feet (0.6 m). The eductor manifold consisted of a 2-inch (5-cm) PVC pipe, capped at both ends, and equipped with four polypropylene eductors and a T-joint at the center (Figure 4). A circulating pump was used to cycle water through the eductor manifold, which kept water circulating in the tank. A transfer pump delivered water from the tank to the transect through the nozzle manifold, which consisted of a 2-inch (5-cm) PVC pipe capped at one end and six brass flood-jet nozzles attached via T-joints to the pipe (Figure 5). The fan of water generated by each nozzle was approximately 2 feet (0.6 m) wide, corresponding to the 2-foot (0.6-m) wide flow path of each collector. The four central nozzles were lined up with the center of each flow path during a trial. The two outer nozzles were included to ensure that equal volumes were delivered to each flow path on account of slight overlap by the nozzle fans.
The experimental design consisted of three treatments: Control, Two-Year, and 100-Year. The Control and Two-Year treatments represented the runoff volume estimated from a standard residential driveway in a 2.8-inch (7.1-cm) storm event, and the runoff volume of 100-Year treatments was estimated from a 5.9-inch (15-cm) storm event. Controls and Two-Year treatments involved 450 gallons (1700 L) of synthetic runoff applied to a transect during a given trial. The tank was always filled with 600 gallons (2270 L) in preparation for a trial, because an extra 150-gallon (570-L) volume was needed in the tank at all times in order to keep the eductor manifold submerged and operational. The 100-Year trials involved 900 gallons (3400 L) of synthetic runoff applied to a transect in two applications of 450 gallons (1700 L) each, separated by the time it took to refill the tank (approximately 45 minutes).
Synthetic runoff for Control trials consisted of tap water alone. The synthetic runoff for Two- and 100-Year treatments consisted of 0.6 oz (1.8 g) phosphorus (i.e., 0.34 oz [9.5 g] Na3PO4) and 4.5 lbs (2,043 g) sediment per 600 gallons (2270 L) of tap water. Sediment was obtained by sifting topsoil from the study site to a size of = 3.9 x 10-5 inches (= 850 microns).
Trials were conducted from August through October 2006. Each of the three treatments was applied to each of the three transects in each of the four plots and replicated three times for a total of 108 trials.
Data collection and analysis
Collectors were observed from the beginning of a trial (i.e., the point at which runoff entered the transect) to the end (i.e., when the tank level had reached 150 gallons [570 L], approximately 30 minutes after starting). The volume of runoff reaching each downslope distance was calculated by adding the known volume inside the collector to the estimated volume of collector overflow. The volume of collector overflow was estimated by measuring the duration and rate of overflow through an extra PVC coupler installed on the downslope side of collectors. Whether or not the diverter overflowed upstream of the collector was also noted, in order to indicate cases where the runoff volume calculated for a collector was an underestimate of the actual volume reaching a given distance downslope. Volume reduction was calculated per collector as a percentage based on the runoff volume applied per flow path in a trial (e.g., 1700 L per transect / 6 nozzles = 283 L per flow path for Control and Two-Year trials).
Water samples were analyzed at the Metropolitan Council’s water quality lab located at the Metropolitan Wastewater Treatment Plant in St. Paul. Synthetic runoff was sampled from the central four nozzles of the manifold during two different trials in order to determine the actual concentration of total solids and total phosphorus applied to transects during Two- and 100-Year trials (Table 1). A sample of tap water was analyzed in order to estimate the background concentrations of total solids and total phosphorus in Control trials (Table 1). The tap water was also analyzed for total dissolved solids. The background concentration of 170 ppm total solids was made up of approximately 44 ppm total suspended solids and 126 ppm total dissolved solids. The additional ~179 ppm total solids applied to transects was assumed to represent primarily total suspended solids added during synthetic runoff preparation.
Samples of runoff were collected from each collector receiving runoff and analyzed for total solids and total phosphorus. The samples reported here were collected from the initial volume of runoff to enter a milk jug placed inside the collector. Samples were also collected from the final volume to overflow the collector; differences between “beginning” and “end” samples were not substantial. The level of total solids and total phosphorus in runoff samples at each collector was converted to mass by multiplying the concentration of total solids or total phosphorus by the volume of runoff received at the collector in a given trial. Removal efficiency was calculated per collector as a percentage based on the mass of total solids or total phosphorus applied per flow path in a given trial.
Collectors typically received about twice the average runoff volume in 100-Year trials as they did in Control or Two-year trials (Table 2).
As expected, average runoff volume was generally related to the steepness of the slope (Table 2). Figure 6 illustrates that there was a slight upward trend in runoff volume as slope increased for Control and Two-Year trials, and a more pronounced upward trend for 100-Year trials.
The average reduction in runoff volume was generally related to buffer width (Figure 6). The 4:1 and 3:1 plots showed a relatively consistent decrease in volume between 5’ and 40’ in all three types of trials. For the 5:1 and 2:1 plots, the 5’ and 10’ collectors tended to receive similar volumes, followed by a sharp decrease in runoff at 20’.
The 4:1 plot was the only plot to have runoff that reached the 40’ collector. This happened once during a Two-Year trial and once during a 100-Year trial. Both cases occurred on transect II during the third round of trials toward the end of the experiments. These cases could not be attributed to changes in plot characteristics, errors in trial procedures, or initial soil moisture conditions. Rather, the two occurrences of runoff at 40’ on transect II most likely resulted from gradual trampling of aisle vegetation between transects I and II and repeated flow of runoff through the aisle during trials on transect I. As the experiments progressed, it appeared that excess runoff from transect I began to flow toward the foot of transect II. This most likely increased soil moisture in the aisle and in the downslope portions of transect II. This process seems to have inhibited the infiltration of runoff upslope of the 40’collector on transect II.
As expected, the concentration of total solids in runoff reaching collectors was consistently higher in Two- and 100-Year trials than in Control trials (Figure 7). In some cases the Control total solids was higher than that of the tap water, which suggests that plots may have been contributing solids to the runoff (particularly the 3:1 plot).
In the Two- and 100-Year trials, much of the total solids removal (~60-95%) appeared to occur within the first 5’ and did not increase substantially between 5’ and 20’ (Figure 8). The 5:1 and 2:1 plots appeared to be slightly more effective at removing total solids compared to the 4:1 and 3:1 plots. Removal efficiency was generally higher in the Two- and 100-Year trials compared to Control trials. Because the higher level of total solids in treatment trials was presumed to be primarily attributable to total suspended solids, the results suggest that most of the total solids removal occurring in treatment trials was serving to remove suspended solids from the runoff.
In the only cases where runoff reached 40’ (i.e., on the 4:1 plot), the average total solids concentrations shown in Figure 7 are misleading because a runoff sample was obtained in only one of nine trials for each of the two treatment types. In the 100-Year trial, however, the total solids concentration was an outlier at 1510 ppm. This large value was likely due to runoff from the adjacent transect, as described above.
The concentration of total phosphorus in runoff reaching collectors was consistently higher in Two- and 100-Year trials than in Control trials (Figure 9). The presence of total phosphorus in excess of the background level during Control trials (primarily in the 3:1 plot), in conjunction with the observation made above for total solids in Control trials, suggests that the plots themselves were contributing some sediment-bound phosphorus to the runoff.
The overall effect of increasing buffer width on average total phosphorus removal efficiency appeared to be greater than the effect of width on total solids removal (Figure 10). In Control trials, the 5:1 plot was not informative because no runoff volume reached 20’, but the other plots experienced more pronounced increases in total phosphorus removal efficiency between 5’ and 20’ than they had for total solids (Figure 8). The 4:1 plot saw the largest increase between 5’ and 10’, while the 3:1 and 2:1 plots saw the largest increase between10’ and 20’. This result suggests that, as with width, slope may have had more of an effect on total phosphorus removal than on total solids removal.
Most total phosphorus removal appeared to occur within the first 5’. Average total phosphorus removal did not increase substantially between 5’ and 20’ for the 4:1 and 3:1 plots, but did increase somewhat between 10’ and 20’ for the 5:1 and 2:1 plots (Figure 10). As with total solids, the 5:1 and 2:1 plots appeared to be slightly more effective at removing total phosphorus compared to 4:1 and 3:1.
The observation that average total solids was more likely than average total phosphorus to approach background levels by 20’ may provide some clues to the mechanisms responsible for phosphorus removal in the buffer. Some phosphorus was likely being removed from the runoff as a function of volume infiltration and through the deposition of phosphorus-bearing sediment. The potential for phosphorus removal by plant adsorption may not have been fully realized due to the vegetative characteristics of the buffer and/or the width of buffer through which the runoff passed.
Conclusions and Recommendations
According to the results of this study, the relationship of buffer width to volume and to total solids and total phosphorus removal is asymptotic. Most reductions in volume, total solids, and total phosphorus occurred within the first 5’, and subsequent reductions were relatively gradual. All runoff volume was infiltrated or retained within 20’ in most cases, even on 50% slopes, and sometimes within 10’. Because runoff generally did not flow beyond 20’, examination of the effect of width on total solids and total phosphorus removal was limited to this same interval. There is evidence that buffer width is a more important determinant of total phosphorus removal than total solids removal.
The effect of width varied by slope. The 4:1 and 3:1 shared some similarities though the 3:1 appeared less effective overall; both plots showed a stepwise decrease in volume between 5’ and 20’, and a pattern of declining total solids and total phosphorus that reached a plateau at approximately 10’. The 5:1 and 2:1 plots behaved the most similarly; both showed a more pronounced decrease in volume between 10’ and 20’, and a more delayed decline in total solids and total phosphorus that approached a point of diminishing returns between 10’ and 20’.
The effect of slope was less clear than width and varied by both treatment and parameter. In general the effect of slope was more pronounced in 100-Year trials. Slope appeared to have a greater effect on volume reduction than on total solids or total phosphorus removal, and there was some indication that slope had more of an effect on total phosphorus removal than on total solids removal. Overall shallower slopes (5:1 and 4:1) did appear to be more effective at pollutant removal than steeper slopes (3:1 and 2:1). The 3:1 plot seemed more prone to erosion and this may have limited the buffer’s ability to reduce pollutant levels at rates comparable to the other plots.
The high percentage of volume reduction and the high removal efficiency of total solids and total phosphorus within 20’ were almost certainly a function of both the soil and vegetative characteristics of the site. The well-drained loam promoted high infiltration rates, while the dense, well-established vegetation further facilitated infiltration and retention of runoff. Consistency in behavior between the 5:1 and 2:1 plots, in spite of their extremely different slopes, was likely due to high similarity in composition and percent cover of their vegetation. The 4:1 and 3:1 plot may also have behaved similarly in part because of vegetation characteristics; these plots shared two of the same dominant plant species. The 4:1 plot was unique in some respects; it was characterized by a primarily grassy composition while the other plots were dominated by herbaceous plant species.
The results of this study suggest that buffer widths of 10-20’ may be effective at reducing the runoff volume and the levels of total solids (especially suspended solids) and total phosphorus that characterize residential stormwater runoff in the TCMA. The effectiveness is largely related to certain circumstances, namely sheet flow occurring in unsaturated well-drained soil and well-established, primarily herbaceous vegetative cover. Relatively steep slopes offer some benefit though they appear not to be as effective as more shallowly sloped buffers, especially under extreme rainfall conditions.
Return to Case Studies.