e c o l o g i c a l e n g i n e e r i n g 3 1 ( 2 0 0 7 ) 225–231

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/ecoleng

Quantifying the effect of slope on extensive green roof stormwater retention Kristin L. Getter a,∗ , D. Bradley Rowe a , Jeffrey A. Andresen b a b

Department of Horticulture, Michigan State University, East Lansing, MI 48824, United States Department of Geography, Michigan State University, East Lansing, MI 48824, United States

a r t i c l e

i n f o

a b s t r a c t

Article history:

Impervious surfaces, such as rooftops, parking lots, and roads, increase runoff and the

Received 12 February 2007

potential for flooding. Green roof technologies, which entail growing plants on rooftops,

Received in revised form

are increasingly being used to alleviate stormwater runoff problems. To quantify the effect

31 May 2007

that roof slope has on green roof stormwater retention, runoff was analyzed from 12 exten-

Accepted 19 June 2007

sive green roof platforms constructed at four slopes (2%, 7%, 15%, and 25%). Rain events were categorized as light (<2.0 mm) (0.08 in.), medium (2.0–10.0 mm) (0.08–0.39 in.), or heavy (>10.0 mm) (>0.39 in.). Data demonstrated an average retention value of 80.8%. Mean reten-

Keywords:

tion was least at the 25% slope (76.4%) and greatest at the 2% slope (85.6%). In addition,

Vegetated roof

runoff that did occur was delayed and distributed over a long period of time for all slopes.

Runoff

Curve numbers, a common method used by engineers to estimate stormwater runoff for an

Eco-roof

area, ranged from 84 to 90, and are all lower than a conventional roof curve number of 98, indicating that these greened slopes reduced runoff compared to traditional roofs. © 2007 Elsevier B.V. All rights reserved.

1.

Introduction

Impervious surfaces continue to expand as we construct buildings, roads, and parking lots. In the United States, it is estimated that 10% of residential developments and 71–95% of industrial areas and shopping centers are covered with impervious surfaces (Ferguson, 1998). Two-thirds of all impervious area is in the form of parking lots, driveways, roads, and highways (Water Resources Group, 1998). Covering natural surfaces causes many problems. Greater runoff (Scholz-Barth, 2001) increases the potential for flooding, reduces infiltration into the groundwater system (Barnes et al., 2001), and can potentially contaminate surface waters due to particulate matter in the runoff (USEPA, 1994; Ferguson, 1998). Other problems with impervious surfaces include higher ambient air temperatures (USEPA, 2003), increased noise, poorer air quality (Liesecke and Borgwardt, 1997; Yok Tan and Sia, 2005), and a loss of biodiversity (Bastin et al., 1999).

∗

Corresponding author. Tel.: +1 517 355 5191x1341. E-mail address: [email protected] (K.L. Getter). 0925-8574/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.ecoleng.2007.06.004

Green roofs are one potential remedy for these problems. Establishing plant material on rooftops provides numerous ecological and economic benefits, including stormwater management, energy conservation, mitigation of the urban heat island effect, increased longevity of roofing membranes, and mitigation of noise and air pollution, as well as a more aesthetically pleasing environment in which to work and live (Getter and Rowe, 2006; Liesecke, 1998; Liu and Minor, 2005 GRHC; Meng and Hu, 2005; Simmons and Gardiner, 2007; VanWoert et al., 2005; Villarreal and Bengtsson, 2005). Many consider the reduction of stormwater runoff to be the greatest environmental service that green roofs provide. In a green roof system, much of the precipitation is captured in the media or vegetation and eventually evaporates from the soil surface or is released back into the atmosphere by transpiration. While the chosen type of green roof system (design, substrate depth, and plant species) will affect retention, research has shown reductions of 60–100% in runoff

226

e c o l o g i c a l e n g i n e e r i n g 3 1 ( 2 0 0 7 ) 225–231

(Liesecke, 1998; Moran et al., 2004; DeNardo et al., 2005; VanWoert et al., 2005). Since green roofs retain stormwater, they can mitigate the effects of impervious surface runoff. Peck (2005) estimated that if 6% of all buildings in Toronto had green roofs, it would result in the same stormwater retention impact as building a $60 million (CDN) storage tunnel. Likewise, in Washington, DC, if 20% of all buildings that could support a green roof had one, they would add over 71 million liters (19 million gallons) to the city’s stormwater storage capacity and store approximately 958 million liters (253 million gallons) of rainwater in an average year (Deutsch et al., 2005). In Germany, two researchers found no significant difference in retention amounts across differently sloped roofs (Liesecke, 1999; Schade, 2000), while other scientists are establishing differences (VanWoert et al., 2005; Villarreal and Bengtsson, 2005). The contradicting results may be due to rainfall patterns at different locales. Rainfall intensity, duration, and initial substrate moisture content all influence retention. Dry substrate conditions prior to rainfall result in greater stormwater retention compared to initially wet conditions (Villarreal and Bengtsson, 2005; Connelly and Liu, 2005). Environmental differences may also influence the choice for substrate depth and plant material, which may in turn influence stormwater retention. A common and widespread method for estimating stormwater runoff for a region or area is the curve number (CN) method developed by the USDA Soil Conservation Services (USDA SCS), now the USDA Natural Resources Conservation Service (USDA NRCS). This method states the relationship between rainfall and runoff with the equation F/S = Q/P, where F is the actual retention (P − Q), S the potential retention, Q the actual runoff, and P the potential runoff or total rainfall (NRCS, 2004). The potential retention (S) can then be converted to a curve number with the formula CN = 25,400/(254 + S) where S is in mm (Hawkins, 1993). Curve numbers are dimensionless and range from 0 (no runoff) to 100 (all precipitation results in runoff). All impervious surfaces such as paved roads and conventional roofs are assigned a CN of 98 (NRCS, 2004). Since green roofs are more frequently being used as a tool for managing storm runoff, the objective of this study was to quantify the effect of slope on stormwater retention and develop curve numbers for green roofs at four different slopes.

2.

Materials and methods

2.1.

Green roof testing platforms

Twelve roof platforms with dimensions of 2.44 m × 2.44 m (8.0 ft × 8.0 ft) were constructed by ChristenDETROIT Roofing Contractors (Detroit, MI) at the Michigan State University Horticulture Teaching and Research Center (East Lansing, MI). Each platform replicated a commercial extensive green roof, including insulation, protective, and waterproofing membrane layers. Construction details are outlined in VanWoert et al. (2005). Aluminum sheet metal troughs were attached on the low end of the platforms to direct stormwater runoff through the measuring devices used to quantify runoff. The wood-framed

platforms included sides that extended 20.3 cm (8 in.) above the platform deck, also covered with waterproofing membrane. All platforms were placed with the low end of the slope facing south to maximize sun exposure.

2.2.

Drainage system and vegetation carrier

Each platform was covered with a Xero Flor XF108 drainage layer (Wolfgang Behrens Systementwicklung, GmbH, Groß Ippener, Germany) installed over the waterproofing system, which allowed excess water to flow off the roof. For additional water holding capacity, a 0.75 cm (0.26 in.) thick moisture retention fabric (Xero Flor XF159) capable of retaining up to 5.92 kg m−2 of water was placed over the drainage layer. Above the retention fabric was the vegetation carrier (Xero Flor XF301).

2.3.

Plant establishment

Growing substrate (Table 1) was placed on top of the vegetation carrier at a depth of 6.0 cm (2.4 in.). The water retention fabric and substrate together have the potential to hold up to 12.0 mm (0.5 in.) of rainfall. Seeds were sown and established on the growing substrate per VanWoert et al. (2005). Species included Saxifraga granulata L. (meadow saxifrage), Sedum acre L. (biting stonecrop), Sedum album L. (white stonecrop), Sedum kamtschaticum ellacombianum Fisch. (kamtschatka stonecrop),

Table 1 – Initial physical and chemical properties of substrate Component Total sand (%) Very coarse sand (1–2 mm) (%) Coarse sand (0.5–1 mm) (%) Medium sand (0.25–0.5 mm) (%) Fine sand (0.10–0.25 mm) (%) Very fine sand (0.05–0.10 mm) (%) Silt (%) Clay (%) Bulk density (g/cm3 ) Pore space (%) Air filled porosity (%) Water holding capacity at 0.01 MPa (%) pH Conductivity (EC) (m mho/cm) Nitrate (ppm) Phosphorus (ppm) Potassium (ppm) Calcium (ppm) Magnesium (ppm) Sodium (ppm) Sulfur (ppm) Boron (ppm) Iron (ppm) Manganese (ppm) Zinc (ppm) Copper (ppm)

Unit 91.18 21.96 40.80 24.66 3.36 0.40 5.60 3.22 1.16 41.41 21.43 17.07 7.9 3.29 203 65.8 622 214 60 164 184 0.5 9.0 15.7 5.7 0.6

Analysis per A&L Great Lakes Laboratories, Inc., Ft. Wayne, Indiana.

e c o l o g i c a l e n g i n e e r i n g 3 1 ( 2 0 0 7 ) 225–231

Sedum pulchellum Michx. (bird’s claw sedum), Sedum reflexum L. (crooked stonecrop), Sedum spurium Bieb. ‘Coccineum’ (creeping sedum), and Sedum spurium Bieb. ‘Summer Glory’ (creeping sedum). Full coverage was achieved (no substrate exposed) and maintained since July 2002.

2.4.

Treatments

Platforms were set at one of four slopes (2%, 7%, 15%, and 25%) in a completely randomized design (CRD) with each slope replicated three times. Platforms were adjusted to the appropriate slope in April 2005. Because a sloped platform reduces the horizontal area upon which rain falls, the effective area of each platform was calculated based on slope and original platform area. Thus, the effective platform areas were 5.49 m2 (59.07 ft2 ), 5.48 m2 (58.94 ft2 ), 5.43 m2 (58.41 ft2 ), and 5.32 m2 (57.26 ft2 ) for the 2%, 7%, 15%, and 25% treatments, respectively. Because this study utilized roof platforms that were three years old at the beginning of the study, substrate samples were taken at the conclusion of this study in order to quantify substrate changes over time. Soil cores (13.0 cm (5.1 in.)) were taken at three random places amongst the twelve platforms and were analyzed for organic matter (loss on ignition at 550 ◦ C), pore space, free airspace, and water holding capacity (A&L Great Lakes Laboratories, Fort Wayne, IN). These results were compared with previous analysis of fresh substrate.

2.5.

Data collection

Rainfall and runoff were recorded on a CR10X datalogger (Campbell Scientific, Inc., Logan, UT) that was placed alongside three AM16T multiplexers. Twelve TE525WS tipping bucket rain gages (Campbell Scientific, Inc., Logan, UT) were each situated underneath a platform to collect runoff from the aluminum troughs and a 13th tipping bucket measured rainfall. The twelve runoff tipping buckets were covered with a plastic 10 in round plant saucer that accommodated a hole to allow water from the aluminum trough to enter the rain gage while also excluding rainfall. Accuracy of the rain gages was reported by the manufacturer to be ±1%, +0 and −2.5%, and +0 and −3.5% for rainfalls of <25.4 mm h−1 , 25.4–50.8 mm h−1 , and 50.8–76.2 mm h−1 , respectively. Data were recorded continuously from 26 April 2005 until 1 September 2006. The datalogger was programmed to collect values every minute and totals were put out every 5 min, 24 h a day throughout the period. Data were downloaded off the datalogger and onto a laptop computer every week.

2.6.

Data analysis

Retention data were analyzed from all rain events that occurred during temperatures above 0 ◦ C (32 ◦ F) as a percentage of total rainfall for each rain event. In order to exclude melting precipitation in runoff data, analysis was limited to dates between 26 April 2005 and 22 November 2005 and between 12 April 2006 and 1 September 2006. Retention is defined here as precipitation that did not run off the platforms. Independent rain events were defined as precipitation events separated by 6 or more hours. In the event runoff was

227

still occurring 6 h after the first event, the two events were combined. Rain events were arbitrarily categorized by relative intensity as light (<2.0 mm) (0.08 in.), medium (2.0–10.0 mm) (0.08–0.39 in.), or heavy (>10.0 mm) (>0.39 in.). The range of each category was chosen to obtain rain event sample sizes that were similar across all three categories. There was a total of 62 rain events. Data were analyzed two ways. In the first, mean percent retention per rain event was analyzed using an ANOVA model with roof slope and rainfall category as fixed effects. Although original means are presented, all retention values were transformed prior to analysis using an arcsine square root transformation to stabilize the variance and normalize the data set (Underwood, 1998). Significant differences between treatments were determined using multiple comparisons by LSD (PROC MIXED, SAS version 8.02, SAS Institute, Cary, NC). The second analysis was to determine curve numbers for each green roof slope by regressing for S in the formula F/S = Q/P (PROC REG, SAS version 8.02, SAS Institute, Cary, NC) and then converting S to a curve number with the equation CN = 25,400/(254 + S) where S is in mm (Carter and Rasmussen, 2006).

3.

Results and discussion

During the study there were 94 days with quantifiable precipitation, resulting in a total of 62 rain events that were used in analysis (Fig. 1). The maximum precipitation for 1 day was 38.1 mm during the study, while a maximum single rain event exceeded 40 mm (Fig. 2). Rainfall was distributed as 16 light (<2 mm), 24 medium (2–10 mm), and 22 heavy (>10 mm) rain events (Fig. 2). Daily minimum ambient air temperatures during the data collection period ranged from −6.7 ◦ C to 25.3 ◦ C (19.9–77.5 ◦ F) and daily maximum ambient air temperatures ranged from −2.3 ◦ C to 34.8 ◦ C (27.9–94.6 ◦ F) (Fig. 1). The ANOVA model showed rain category and slope, as well as the interaction of both, to be significant (Table 2). Representative hydrographs (Fig. 3) and cumulative hydrographs (Fig. 4) illustrate the effect of roof slope on quantity of runoff and overall delay for light, medium, and heavy rain events. Initial runoff delay for these rain events is minimal for all slopes. This contradicts DeNardo et al. (2005), VanWoert et al. (2005), and Carter and Rasmussen (2006) who reported 4 h, 40 min, and 34 min initial delays, respectively. Perhaps the intensity of the rainfall or moisture condition of the substrate prior to these storm events explains the difference. Another explanation for these contradicting results is that with the exception of the 2% slope, all slopes evaluated in this study are steeper than DeNardo et al. (2005), VanWoert et al. (2005), and Carter and Rasmussen (2006) and this study also has a shallower media depth than DeNardo et al. (2005) and Carter and Rasmussen (2006). In addition, this roof ecosystem had been established for 3 years prior to commencing this study, which is older than all of the previously mentioned studies. This greater maturity may effect the hydraulic conductivity of the substrate. Mentens et al. (2006) indicated that roof age was not correlated to the quantity of retention, but roof age may affect the time pattern of retention. Overall pore space and changes in pore

228

e c o l o g i c a l e n g i n e e r i n g 3 1 ( 2 0 0 7 ) 225–231

Fig. 2 – Frequency of rain events included in the study from 26 April 2005 to 22 November 2005 and 12 April 2006 to 1 September 2006. Rainfall measurements were taken from tipping bucket rain gauges mounted at the research site.

Runoff was spread out over time across all treatments with the final runoff lasting 4 h 20 min, 10 h 45 min, and 13 h 45 min for light, medium, and heavy rain events, respectively, after rainfall stopped. These results are similar to VanWoert et al. (2005), Liu (2003), and Moran et al. (2004). Liu and Moran also found that this delayed runoff was at a lower flow rate. By slowing down the rate of runoff and releasing it out over a longer period of time, green roofs can help mitigate the erosional power of runoff that does enter streams, either through direct runoff or stormsewers. It can also prevent combined stormwater sewer systems from overflowing, by allowing it to process runoff for a longer time at a lower flow rate. These results may influence stormwater management practices or design of municipal stormwater and sewage systems. The green roofs retained an average of 80.2% of precipitation averaged across all slopes and rain categories (Table 4). Mean retention was least at the 25% slope (75.3%) and greatest at the 2% slope (85.2%). In addition, retention values decreased as slope increased. Retention values were highest for light rain events (94.2%) and lowest for heavy rain events (63.3%).

Fig. 1 – Daily maximum and minimum temperatures (◦ C) and precipitation (mm) throughout the study (1 April 2005 to 1 September 2006). Data are from the Michigan Automated Weather Network’s East Lansing weather station located adjacent to the research site.

size may occur over time as a result of settling or as a result of changes in organic matter content. In this study, mature substrate exhibited greater values for porosity, free airspace (macropores), organic matter, and water holding capacity at the conclusion of this study relative to the initial substrate (Table 3). Increased free air space, resulting from channels formed by decaying roots or burrowing insects, may increase preferential macropore flow through the substrate, thereby resulting in quicker initial runoff.

Table 2 – ANOVA table for rainfall retention over the 2-year period (26 April 2005 to 22 November 2005 and 12 April 2006 to 1 September 2006) from four roof platform treatments replicated three times Source of variation Model Rain categorya Slopeb Category × slope Error Corrected total

Degrees of freedom

Sum of squares

Mean squares

11 2 3 6

80.5 74.6 1.5 2.6

7.3 37.3 0.5 0.4

598 609

63.3 143.8

0.1

F-Statistic 69.1 352.2 4.7 4.1

P-value <.0001 <.0001 .0029 .0005

Retention is the dependent variable. Roof slope and rain category are independent variables. a b

Vegetated roof platforms set at 2%, 7%, 15%, and 25% slope with 6.0 cm (2.4 in.) of substrate. Rain event categories were light (<2.0 mm) (0.08 in.) (n = 16), medium (2.0–10.0 mm) (0.08–0.39 in.) (n = 24), heavy (>10.0 mm) (>0.39 in.) (n = 22), and overall (n = 62).

229

e c o l o g i c a l e n g i n e e r i n g 3 1 ( 2 0 0 7 ) 225–231

Fig. 3 – Runoff hydrographs of selected representative rain events: (A) heavy (23.37 mm) (0.92 in.), (B) medium (5.08 mm) (0.2 in.), and (C) light (1.02 mm) (0.04 in.). Lines represent runoff (mm) from a 2%, 7%, 15%, or 25% roof slope with 6.0 cm (2.4 in.) of media. Values are averages of three replications measured using tipping bucket rain gauges mounted at the research site.

This demonstrates that the substrate has a limited storage capacity; once it is saturated the precipitation runs off. Retention data agree with VanWoert et al. (2005) and Carter and Rasmussen (2006). However, our retention values are much higher than DeNardo et al. (2005), Liesecke (1998), and Mentens et al. (2006) who all reported an average of 45%, 40–50%, and 45% retention, respectively. This may be due to

Fig. 4 – Cumulative hydrographs of selected representative rain events: (A) heavy (23.37 mm) (0.92 in.), (B) medium (5.08 mm) (0.2 in.), and (C) light (1.02 mm) (0.04 in.). Lines represent runoff (mm) from a 2%, 7%, 15%, or 25% roof slope with 6.0 cm (2.4 in.) of media. Values are averages of three replications measured using tipping bucket rain gauges mounted at the research site.

differences in substrate depth, antecedent substrate moisture status, slope, or precipitation patterns. But it is most likely due to the fact that all of the latter researchers used large storms in their stormwater testing. In addition, Liesecke (1998) and Mentens et al. (2006) followed FLL (http://www.f-ll.de/english.html) guidelines which employ nearly saturated antecedent moisture conditions followed by a simulated rain that constitutes a 100-year storm. This is very different from

Table 3 – Organic matter content and physical properties of initial substrate prior to planting (2002) and after 5 years on a green roof (2006) Sample Initial substrate Mature substrate

Organic matter (%) 2.33 4.25

Pore space (%) 41.41 81.84

Analysis per A&L Great Lakes Laboratories, Inc., Ft. Wayne, Indiana.

Free airspace (%) 21.43 14.40

Water holding capacity (%) 17.07 67.44

230

e c o l o g i c a l e n g i n e e r i n g 3 1 ( 2 0 0 7 ) 225–231

Table 4 – Mean percentage ± the standard deviation of total rainfall retention over the 2-year period (26 April 2005–22 November 2005 and 12 April 2006–1 September 2006) from four roof platform treatments replicated three times Treatmenta

Lightb (%) ± ± ± ±

3.4 b Ac 3.1 c A 3.2 c A 2.9 c A

2% 7% 15% 25%

93.3 94.0 94.0 95.5

Overall

94.2 ± 3.3 c

a b

c

Medium (%) 92.2 89.5 88.6 87.8

± ± ± ±

9.5 b A 12.7 b A 13.3 b A 16.3 b A

89.5 ± 12.8 b

Heavy (%) 71.4 66.4 58.4 57.1

± ± ± ±

18.1 a C 18.5 a B 17.4 a A 16.1 a A

63.3 ± 18.4 a

Overall (%) 85.2 82.2 78.0 75.3

± ± ± ±

15.9 B 18.3 AB 21.0 A 22.3 A

80.2 ± 19.6

Retention from vegetated roof platforms set at 2%, 7%, 15%, and 25% slope with 6.0 cm (2.4 in.) of substrate. Rain event categories were light (<2.0 mm) (0.08 in.) (n = 16), medium (2.0–10.0 mm) (0.08–0.39 in.) (n = 24), heavy (>10.0 mm) (>0.39 in.) (n = 22), and overall (n = 62). Mean separation in rows and columns by LSD (P ≤ 0.05). Lowercase letters denote comparisons across rain categories within individual slopes (n = 12). Uppercase letters in columns denote differences among slopes (n = 12).

our natural conditions, which would have varying antecedent moisture conditions and varying storm volumes. For example, on 18 June 2006, 42.2 mm (1.7 in.) of precipitation occurred following an earlier rain event of 7.62 mm (0.3 in.). These relatively dry antecedent conditions retained 68%, 64%, 57%, 58% for 2%, 7%, 15%, and 25% slopes, respectively. In contrast, on 28 August 2006, 28.7 mm (1.1 in.) of rain fell after a rain just 1.5 days before of 8.1 mm (0.32 in.). The relatively wet antecedent conditions retained 45%, 30%, 27%, 29% for 2%, 7%, 15%, and 25% slopes, respectively. In addition, organic matter composition and age of the substrate may affect retention volumes as well. While reports from Germany (Mentens et al., 2006) indicate that roof age does not affect the quantity of retention, our 5-year-old substrate had nearly twice the water holding capacity as new substrate (Table 3). Increases in organic matter and micropores may increase water-holding capacity, which increases total retention, but increased macropores (channels) also reduced the initial delay. Schade (2000) and Liesecke (1998) concluded that green roof slope did not affect retention amounts for slopes ranging from 2% to 58%. Our results are contradictory in that the effect of

roof slope was significant when comparing 2% and 15% slopes, as well as 2% and 25% slopes. This difference again is probably due to these researchers using wet antecedent moisture conditions and 100-year stormwater volumes in simulated conditions, which is different from our study design. Maybe for a 100-year storm event, slope does not influence retention, but for normal rain events it does. Green roofs will function the majority of the time under normal weather conditions, not in 100-year storms where the substrate is inundated with water. Curve numbers were calculated to be 84, 87, 89, and 90 for 2%, 7%, 15%, and 25% slopes, respectively. All of these numbers are lower than a conventional roof curve number of 98, indicating that all of these greened slopes had less runoff than traditional black roofs (Fig. 5). This agrees with VanWoert et al. (2005) who compared conventional gravel ballasted roofs with green roofs and found that traditional roofs retained the least rainfall. Curve numbers also increase in value as slope increased, indicating more runoff as slopes became steeper (Fig. 5). Using these curve numbers in the equation CN = 25,400/(254 + S) and solving for potential retention (S) we find that S ranges from 28.2 mm to 48.4 mm (1.1–1.9 in.). These findings are similar to Carter and Rasmussen (2006) who found a curve number of 86 (S = 40.5 mm) (S = 1.6 in.) for a green roof with <2% slope and 7.62 cm (3.0 in.) of substrate. Other land and surface cover types which have the similar curve numbers to these range from clay soil pastures in fair condition to gravel roads atop clay soil (NRCS, 2004). These curve numbers will assist engineers and stormwater managers in estimating stormwater runoff peak rates and runoff quantities of larger watersheds that implement green roofs.

4.

Fig. 5 – Curve numbers (CN) from vegetated roof platforms set at 2%, 7%, 15%, and 25% slopes with 6.0 cm (2.4 in.) of substrate over the 2-year period (26 April 2005–22 November 2005 and 12 April 2006–1 September 2006) from four roof platform treatments replicated three times.

Conclusion

This study demonstrated that green roof slope does have an effect on runoff retention quantities. Retention values decreased as slope increased and was significant for slopes between 2% and 15% as well as between 2% and 25%. In addition, green roof curve numbers were shown to be much lower than traditional roofing materials, which are typically assigned a curve number of 98. In this study, curve numbers ranged from 84 to 90, resulting in a potential retention (S) ranging from 28.2 mm to 48.4 mm (1.1–1.9 in.). These conclusions are applicable to the midwestern United States and other geographical areas with similar climates.

e c o l o g i c a l e n g i n e e r i n g 3 1 ( 2 0 0 7 ) 225–231

The Michigan State University campus covers 21.0 km2 (5200 acres) and has 1.1 km2 (12 million ft2 ) of flat roof surface. If all of these roofs were greened similar to the roof platforms in this study, then based on a mean retention of 80.2%, these roofs would have retained 377,041 m3 (99,603,827 gallons or 13,315,095 ft3 ) during 2005. Of course, retention on any roof depends on rainfall distribution throughout the year, the intensity of each event, ambient air temperatures, plant selection, and the influence of local environmental conditions on evapotranspiration.

references

Barnes, K., Morgan, J., Roberge, M., 2001. Impervious surfaces and the quality of natural built environments. Report prepared for Towson University’s NASA/Raytheon/Synergy project. Bastin, L., Thomas, C.D., 1999. The distribution of plant species in urban vegetation fragments. Landscape Ecol. 14, 493–507. Carter, T.L., Rasmussen, T.C., 2006. Hydrologic behavior of vegetated roofs. J. Am. Water Res. Assoc. 42 (5), 1261–1274. Connelly, M., Liu, K., 2005. Green roof research in British Columbia—an overview. In: Proceedings of the 3rd North American: Green Roof Conference on Greening Rooftops for Sustainable Communities, Washington, DC, May 4–6. The Cardinal Group, Toronto, pp. 416–432. DeNardo, J.C., Jarrett, A.R., Manbeck, H.B., Beattie, D.J., Berghage, R.D., 2005. Stormwater mitigation and surface temperature reduction by green roofs. Trans. ASAE 48 (4), 1491–1496. Deutsch, B., Whitlow, H., Sullivan, M., Savineau, A., 2005. Re-greening Washington, DC. A green roof vision based on environmental benefits for air quality and storm water management. In: Proceedings of the 3rd North American: Green Roof Conference on Greening Rooftops for Sustainable Communities, Washington, DC, May 4–6. The Cardinal Group, Toronto, pp. 379–384. Ferguson, B.K., 1998. Introduction to Stormwater: Concept, Purpose, Design. John Wiley and Sons, Inc., New York. Getter, K.L., Rowe, D.B., 2006. The role of extensive green roofs in sustainable development. HortScience 41 (5), 1276–1285. Hawkins, R.H., 1993. Asymptotic determination of runoff curve numbers from data. J. Irrig. Drain. Eng. 119 (2), 334–345. Liesecke, H.J., Borgwardt, H., 1997. Abbau von luftschadstoffen ¨ durch extensive dachbegrunungen (degradation of air ¨ 46 (5), pollutants by extensive green roofs). Stadt und Grun 245–251. ¨ Liesecke, H.J., 1998. Das retentionsvermogen von ¨ dachbegrunungen (water retention capacity of vegetated ¨ 47 (1), 46–53. roofs). Stadt und Grun ¨ Liesecke, H.J., 1999. Extensive begrunung bei 5◦ dachneigung ¨ 48 (5), (extensive roof greenings on a 5◦ slope). Stadt und Grun 337–346. Liu, K., 2003. Engineering performance of rooftop gardens through field evaluation. In: Proceedings of the 18th

231

International Convention of the Roof Consultants Institute, pp. 93–103. Liu, K., Minor, J., 2005. Performance evaluation of an extensive green roof. In: Proceedings of 3rd North American Green Roof Conference: Greening Rooftops for Sustainable Communities, Washington, DC, May 4–6. The Cardinal Group, Toronto, pp. 385–398. Meng, Q., Hu, W., 2005. Roof cooling effect with humid porous medium. Energy Buildings 37, 1–9. Mentens, J., Raes, D., Hermy, M., 2006. Green roofs as a tool for solving the rainwater runoff problem in the urbanized 21st century? Landscape Urban Plan. 77, 217–226. Moran, A., Hunt, B., Jennings, G., 2004. A North Carolina field study to evaluate green roof runoff quantity, runoff quality, and plant growth. In: Proceedings of the 2nd North American Green Roof Conference: Greening Rooftops for Sustainable Communities, Portland, OR, June 2–4. The Cardinal Group, Toronto, pp. 446–460. NRCS, 2004. National Engineering Handbook: Hydrology. National Soil Conservation Service (Formerly Soil Conservation Service), USDA, Washington, DC (Chapter 4). Peck, S.W., 2005. Toronto: A Model for North American Infrastructure Development. In: EarthPledge. Green Roofs: Ecological Design and Construction. Schiffer Books, Atglen, PA, pp. 127–129. ¨ Schade, C., 2000. Wasserruckhaltung und Abflu␤beiwerte bei ¨ ¨ ¨ 49 (2), dunnschichtigen extensivbegrunungen. Stadt und Grun 95–100. Scholz-Barth, K., 2001. Green roofs: stormwater management from the top down. Environ. Des. Constr. 4 (1), 63–70. Simmons, M., Gardiner, B., 2007. The effects of green roofs in a sub-tropical system. GRHC. Underwood, A.J., 1998. Experiments in Ecology: Their Logical Design and Interpretation using Analysis of Variance. University Press, Cambridge. USEPA (United States Environmental Protection Agency), 1994. The Quality of Our Nation’s Water: 1992. USEPA Office of Water, Washington, DC, EPA-841-S-94-002. USEPA, 2003. Cooling Summertime Temperatures: Strategies to Reduce Urban Heat Islands. USEPA, Washington, DC, EPA-430-F-03-014. VanWoert, N.D., Rowe, D.B., Andresen, J.A., Rugh, C.L., Fernandez, R.T., Xiao, L., 2005. Green roof stormwater retention: effects of roof surface, slope, and media depth. J. Environ. Qual. 34 (3), 1036–1044. Villarreal, E.L., Bengtsson, L., 2005. Response of a Sedum green-roof to individual rain events. Ecol. Eng. 25 (1), 1–7. Water Resources Group, 1998. Water laws: understanding the problems facing urban watersheds, 20 October 2005. http://www.waterlaws.com/guest/guest1.html. Yok Tan, P., Sia, A., 2005. A pilot green roof research project in Singapore. In: Proceedings of 3rd North American Green Roof Conference: Greening Rooftops for Sustainable Communities, Washington, DC, May 4–6. The Cardinal Group, Toronto, pp. 399–415.