Professional Documents
Culture Documents
Prepared for
City of Burnsville
June 2006
Table of Contents
1.0 Introduction............................................................................................................................................1
2.0 Study Methodology................................................................................................................................3
2.1 Paired Watershed Approach ....................................................................................................... 3
2.2 Stormwater and Rainfall Monitoring .......................................................................................... 4
3.0 Results and Discussion ..........................................................................................................................5
3.1 Stormwater Monitoring Results .................................................................................................. 5
3.2 Conclusions and Recommendations for Further Study............................................................. 12
References...................................................................................................................................................13
List of Tables
List of Figures
List of Appendices
In an ongoing effort to protect Crystal Lake from excess phosphorus and large volumes of
stormwater runoff from surrounding hard surfaces, the Metropolitan Council, along with the City of
Burnsville, Minnesota and the Dakota County Soil and Water Conservation District funded a
prototypic rainwater garden system to infiltrate street runoff. While the City had been interested for
some time in constructing rainwater gardens, questions about their effectiveness remained. To better
document the effect of rainwater garden implementation, this project involved the completion of a
“paired watershed” study, in which two very similar residential areas are monitored —one was the
study’s control site and the other treatment watershed employs 17 new rainwater gardens within a 25-
lot, 5.3-acre neighborhood with traditional curb and gutter (see Figure 1). The project, a retrofit of a
1980s neighborhood, involved individual designs for each resident-participant’s property and close
attention to homeowner education and easy maintenance. The gardens were primarily designed to
capture street runoff through the installation of curb cuts at each garden. The depressions feature
gradual side slopes, limestone retaining walls, and colorful plantings. They were carefully sized to,
at a minimum, accept the first 0.9 inches of rainfall runoff from the impervious surfaces in the
subwatershed for each storm event.
Existing soils and utilities were surveyed in 2002 to identify potential garden sites. Seventeen sites
were identified in the treatment watershed; thirteen along Rushmore Drive and four in a backyard
swale that drains to Rushmore Drive (see Figure 2). Each garden along the street was designed to
have a curb cut to capture street runoff. Individual homeowners were involved in creating final
planting designs for each basin. Gardens were constructed in September 2003. Curb cuts were
installed in May 2004. The contractor cut the sod, excavated below grade, backfilled with
topsoil/compost mix, and installed edging and retaining walls. Homeowners planted plants in
September 2003. After planting, the contractor placed shredded wood mulch and sod to finish
gardens.
Both the control and treatment watersheds were monitored before and after rainwater garden
construction to facilitate the statistical evaluation of the paired watershed data.
Watershed
Period
Control Treatment
This “paired watershed” study was conducted by selecting two similar and adjacent subwatersheds in
the Crystal Lake watershed—one to serve as the study control and the other to be the site of 17
rainwater gardens (Figure 1). Stormwater runoff was monitored both prior to and after installation of
the gardens. As per Clausen and Spooner (1993), a linear regression and analysis of variance was
conducted on the paired data from the calibration period to evaluate the significance of the
relationship. At the end of the treatment period the significance of the effect of the rainwater gardens
was determined by completing an analysis of variance on the treatment regression equation and
comparing the difference between the slopes, and confidence levels of the calibration and treatment
regressions.
The monitoring data from the flow meters and rain gauge were downloaded on a regular basis and
used to determine the flow and rainfall volumes associated with each of the runoff events at each
monitoring location. The paired flow volume data that was available from both watershed
monitoring locations was compiled in a spreadsheet, along with the rainfall data associated with each
runoff event. In each case the runoff volume was also expressed as runoff, in inches, and as a runoff
coefficient by dividing the flow volumes by the watershed area, and then by the rainfall amount from
each runoff event.
Appendix A provides a summary of each of the rainfall/runoff events during the calibration and
treatment periods. Figure 3 shows the linear regression that was conducted on the paired runoff
volume data from the calibration period (shown in Appendix A) to evaluate the significance of the
relationship. The graph shows that there was very good agreement between the runoff volumes
collected at each site during the calibration period, since the regression coefficient of determination
is high (R2 = 0.89) and the y-intercept is low (0.05). Figure 3 also shows the 95 percent confidence
levels around the slope of the linear regression. Since the confidence level was within 15 percent of
the slope of the regression line, it indicated that there was enough data available from the calibration
period to justify the construction of the rainwater gardens and transition to the treatment period.
The rainwater gardens were constructed in the fall of 2003 and brought on-line with curb cuts later in
the spring of 2004. The treatment period monitoring began at the end of May, 2004 (see Appendix
A). Figures 4 and 5 shows how the runoff hydrographs and volumes varied at each monitoring site
during the calibration and treatment periods for moderate and larger rainfall events. Both figures
show that, for similar rainfall events, the runoff rate and volume from the treatment watershed (with
the rainwater gardens) was greatly reduced relative to the data from the control watershed. The
variability of the runoff volumes from the control watershed for similar rainfall events, as shown in
Figures 4 and 5, also underscore the importance of using a paired watershed approach to accurately
evaluate the changes due to BMP implementation.
Figure 6 Linear Regressions of Runoff Volume Data for Calibration and Treatment Periods
Pre-construction (2002-2004)
28 rainfall events = 23.77 inches total
Control Watershed (7.5 acres) = 153,313 cu. ft. runoff (5.69”)
Treatment Watershed (5.5 acres) = 111,120 cu. ft. runoff (5.78”)
Post-construction (2004-2005)
48 rainfall events = 18.97 inches total
Control Watershed (7.5 acres) = 151,897 cu. ft. runoff (5.58”)
Treatment Watershed (5.5 acres) = 7,861 cu. ft. runoff (0.41”)
Figure 7 Runoff Volume Reduction Associated with Rainwater Gardens
While the results of the monitoring from this study provided statistically significant conclusions, it is
recommended that the City repeat the exact same monitoring program again every three to five years
to document the long-term functionality of rainwater gardens in a residential setting.
Clausen, J.C., and J. Spooner. 1993. Paired Watershed Study Design. Prepared for U.S.
Environmental Protection Agency. Office of Water. Washington, D.C. 841-F-93-009.