EFFECTS OF GREEN ROOF PLANT SELECTION ON ENHANCING STORMWATER
QUALITY UNDER STEEP SLOPED ROOF CONDITIONS
By
Kevin S. Krogulecki

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Environmental Design – Master of Arts
2014

ABSTRACT
EFFECTS OF GREEN ROOF PLANT SELECTION ON ENHANCING STORMWATER
QUALITY UNDER STEEP SLOPED ROOF CONDITIONS
By
Kevin S. Krogulecki
Green roofs utilize surfaces on a building to grow vegetation in some type of soil
substrate. This practice is not new, but most recently, it has been used to mitigate some of the
negative effects of urbanization (e.g., stormwater runoff -- quality and quantity). Green roofs
have been proven to reduce the total amount of stormwater runoff from roof surfaces while
increasing the duration of time that runoff occurs, thus reducing peak flow for rivers in a
watershed. Also, green roofs have proven to affect stormwater runoff quality, both positively
and negatively. While some studies have shown that green roofs can act as a sink with respect to
some pollutants in stormwater runoff, others suggest that green roofs can act as a source for other
types of pollutants.
In this study, experimental sloped green roof plots were constructed to monitor green roof
runoff water quality versus the runoff from conventional shingle and steel roofing surfaces. As a
part of the study, two green roof vegetative types (Sedum & native grass) were compared to bare
media only and evaluated for storm water runoff quality. From this study, it was determined that
the green roofs with vegetation will uptake more nutrients than non-vegetated roofs but few
differences could be determined between vegetation types with respect to water quality runoff.
Runoff from green roofs proved to be higher for both nutrients and turbidity than roofs with
conventional roofing materials; it also was found that green roofs increase pH when compared to
roofs with conventional roofing materials.

Copyright by
Kevin S. Krogulecki
2014

ACKNOWLEDGEMENTS

The development and completion of this project has not been done alone. I have been
fortunate enough to have numerous people that have helped me throughout the field, lab and
technical writing aspects of this research project and thesis. First and foremost I would like to
thank Dr. Joanne Westphal for the guidance, insight, and flexibility during this process. She has
been a great advisor, educator and mentor in all aspects of the scholarly process, and she has
allowed me to maximize my graduate experience to the fullest, even beyond what was
anticipated or expected by myself. Second, I would like to thank Dr. Brad Rowe for being such
a great wealth of knowledge in the green roof research and industry arena. His presence on my
committee and throughout this process has proven invaluable to my success and to the
completion of this graduate thesis. Additionally, Dr. Milind Khire and Dr. Jeffrey Andresen
have been great mentors and advisors on the more technical aspects of the environment with
respect to natural processes and pollutants.
Through this process, I have made life-long friends with fellow MSU green roof graduate
student researchers Erik Cronk, Jeremiah Johnson, and Jeremey Monsma, whom I would like to
thank for their support and insight throughout this process. I also would like to thank my
graduate research assistant Samantha Padgett for managing and undertaking the task of sample
collection which allowed me to continue my professional landscape architecture career
simultaneously.
Last but not least, I need to thank my family. The continual support of my father, Stan,
mother, Barb and sister, Katie not only fueled my drive to complete this process, but also
provided me with continual support during my graduate career.

iv

TABLE OF CONTENTS

LIST OF TABLES ......................................................................................................................... vi
LIST OF FIGURES ....................................................................................................................... ix
INTRODUCTION ...........................................................................................................................1
LITERATURE REVIEW ................................................................................................................2
Effects of Urbanization on the Environment .......................................................................2
Green Roof History ..............................................................................................................7
Benefits of Green Roofs.......................................................................................................9
Green Roofs and Stormwater Runoff Quality ...................................................................13
HYPOTHESIS ...............................................................................................................................25
Null Hypothesis 1 ..............................................................................................................25
Hypothesis 1a .....................................................................................................................25
Hypothesis 1b.....................................................................................................................25
Hypothesis 1c .....................................................................................................................26
Null Hypothesis 2 ..............................................................................................................26
Hypothesis 2a .....................................................................................................................26
Null Hypothesis 3 ..............................................................................................................26
Hypothesis 3a .....................................................................................................................27
Null Hypothesis 4 ..............................................................................................................27
Hypothesis 4a .....................................................................................................................27
MATERIALS AND METHODS ...................................................................................................28
Green Roof Testing Platforms ...........................................................................................28
Water Collection ................................................................................................................29
Data Analysis .....................................................................................................................31
RESULTS ......................................................................................................................................32
Nutrients .............................................................................................................................32
pH.......................................................................................................................................35
Turbidity ............................................................................................................................36
DISCUSSION ................................................................................................................................37
CONCLUSION ..............................................................................................................................43
APPENDIX ....................................................................................................................................45
BIBLIOGRAPHY ..........................................................................................................................89

v

LIST OF TABLES

Table 1:

One-way ANOVA table for respective water quality characteristics in stormwater
runoff from five roof platform treatments replicated three times. Study was
completed over 3 growing seasons (22 July, 2010 – 24 September, 2012). Levels
of pollutants in stormwater runoff are the dependent variables. Roof treatments are
the independent variables .........................................................................................46

Table 2:

Tukey’s HSD test for multiple comparisons of Nitrate-Nitrogen (NO3-N) for all
roof treatments during the study’s 3 growing seasons (22 July, 2010 – 24
September, 2012) .....................................................................................................47

Table 3:

Tukey’s HSD test for multiple comparisons of Nitrite-Nitrogen (NO2-N) for all
roof treatments during the study’s 3 growing seasons (22 July, 2010 – 24
September, 2012) .....................................................................................................48

Table 4:

Tukey’s HSD test for multiple comparisons of Ammonia-Nitrogen (NH3-N) for all
roof treatments during the study’s 3 growing seasons (22 July, 2010 – 24
September, 2012) .....................................................................................................49

Table 5:

Tukey’s HSD test for multiple comparisons of Phosphate-Phosphorous (PO4-P) for
all roof treatments during the study’s 3 growing seasons (22 July, 2010 – 24
September, 2012) .....................................................................................................50

Table 6:

Tukey’s HSD test for multiple comparisons of Turbidity (NTU) for all roof
treatments during the study’s 3 growing seasons (22 July, 2010 – 24 September,
2012) ........................................................................................................................51

Table 7:

Tukey’s HSD test for multiple comparisons of pH for all roof treatments during the
study’s 3 growing seasons (22 July, 2010 – 24 September, 2012) ..........................52

Table 8:

Homogenous subset of recorded NO3-N data for all roof treatments during the
study’s 3 growing seasons (22 July, 2010 – 24 September, 2012) ..........................53

Table 9:

Homogenous subset of recorded NO2-N data for all roof treatments during the
study’s 3 growing seasons (22 July, 2010 – 24 September, 2012) ..........................53

Table 10:

Homogenous subset of recorded NH3-N data for all roof treatments during the
study’s 3 growing seasons (22 July, 2010 – 24 September, 2012) ..........................53

Table 11:

Homogenous subset of recorded PO4-P data for all roof treatments during the
study’s 3 growing seasons (22 July, 2010 – 24 September, 2012) ..........................54

vi

Table 12:

Homogenous subset of recorded NTU data for all roof treatments during the
study’s 3 growing seasons (22 July, 2010 – 24 September, 2012) ..........................54

Table 13:

Homogenous subset of recorded pH data for all roof treatments during the study’s
3 growing seasons (22 July, 2010 – 24 September, 2012).......................................54

Table 14:

Data from rain event 22 July, 2010. All concentrations except pH and turbidity are
in milligrams per litre (mg/L) and/or parts per million (PPM) ................................55

Table 15:

Data from rain event 14 August, 2010. All concentrations except pH and turbidity
are in milligrams per litre (mg/L) and/or parts per million (PPM) ..........................56

Table 16:

Data from rain event 19 August, 2010. All concentrations except pH and turbidity
are in milligrams per litre (mg/L) and/or parts per million (PPM) ..........................57

Table 17:

Data from rain event 3 August, 2011. All concentrations except pH and turbidity
are in milligrams per litre (mg/L) and/or parts per million (PPM) ..........................58

Table 18:

Data from rain event 19 September, 2011. All concentrations except pH and
turbidity are in milligrams per litre (mg/L) and/or parts per million (PPM) ...........59

Table 19:

Data from rain event 25 September, 2011. All concentrations except pH and
turbidity are in milligrams per litre (mg/L) and/or parts per million (PPM) ...........60

Table 20:

Data from rain event 27 September, 2011. All concentrations except pH and
turbidity are in milligrams per litre (mg/L) and/or parts per million (PPM) ...........61

Table 21:

Data from rain event 14 October, 2011. All concentrations except pH and turbidity
are in milligrams per litre (mg/L) and/or parts per million (PPM) ..........................62

Table 22:

Data from rain event 20 October, 2011. All concentrations except pH and turbidity
are in milligrams per litre (mg/L) and/or parts per million (PPM) ..........................63

Table 23:

Data from rain event 28 May, 2012. All concentrations except pH and turbidity are
in milligrams per litre (mg/L) and/or parts per million (PPM) ................................64

Table 24:

Data from rain event 2 June, 2012. All concentrations except pH and turbidity are
in milligrams per litre (mg/L) and/or parts per million (PPM) ................................65

Table 25:

Data from rain event 19 June, 2012. All concentrations except pH and turbidity are
in milligrams per litre (mg/L) and/or parts per million (PPM) ................................66

Table 26:

Data from rain event 2 August, 2012. All concentrations except pH and turbidity
are in milligrams per litre (mg/L) and/or parts per million (PPM) ..........................67

vii

Table 27:

Data from rain event 20 August, 2012. All concentrations except pH and turbidity
are in milligrams per litre (mg/L) and/or parts per million (PPM) ..........................68

Table 28:

Data from rain event 8 September, 2012. All concentrations except pH and
turbidity are in milligrams per litre (mg/L) and/or parts per million (PPM) ...........69

Table 29:

Data from rain event 19 September, 2012. All concentrations except pH and
turbidity are in milligrams per litre (mg/L) and/or parts per million (PPM) ...........70

Table 30:

Data from rain event 24 September, 2012. All concentrations except pH and
turbidity are in milligrams per litre (mg/L) and/or parts per million (PPM) ...........71

viii

LIST OF FIGURES

Figure 1:

Graphic representations of land cover change and its subsequent effect on
stormwater runoff. (Source: http://www.fairfaxcounty.gov/nvswcd/images/
drainageproblem/runoff.jpg) ....................................................................................72

Figure 2:

Graphic representations of urban stormwater runoff catchment area and transport
through stormwater utility systems. (Source: http://www2.cincinnati.com/blogs/
gardening/ 2011/11/28/rain/) ...................................................................................73

Figure 3:

Image showing the impacts of excessive nutrient runoff to receiving water bodies.
(Source: http://www.lakeforest.edu/academics/programs/environmental/courses/
seniorseminar/springbreak/students/newcomer.php) ...............................................74

Figure 4:

Graphic representation of land type and use and its effect on air temperature.
(Source: http://indymedia.org.au/2013/06/02/cities-to-get-much-hotter-asheatwaves-amplify-urban-heat-island-effect) ..........................................................75

Figure 5:

Artist’s illustration of the Hanging Gardens of Babylon. (Source: http://batkya.
deviantart.com/art/Hanging-Gardens-of-Babylon-203309919) ..............................76

Figure 6:

Project location map. Study was performed at the Michigan State University
Landscape Architecture Alumni Research Site, Old Mission, MI (Krogulecki,
2014) ........................................................................................................................77

Figure 7:

Picture taken during construction illustrating the use of .045 mm EPMD rubber
waterproof membrane (Krogulecki, 2009) ..............................................................78

Figure 8:

Plan view detail of plot dimensions and randomized treatment layout (Krogulecki,
2014) ........................................................................................................................79

Figure 9:

Picture showing the three green roof vegetation treatments after planting
(Krogulecki, 2009) ...................................................................................................80

Figure 10:

Daily precipitation totals (inches) during the study (22 July 2010 – 24 September
2012). Data collected from the Michigan State University Old Mission weather
station. (Source: http://www.agweather.geo.msu.edu/mawn/ station.asp?id=
old&rt=24) ...............................................................................................................81

Figure 11:

Daily maximum and minimum temperatures (°F) during the study (22 July 2010 –
24 September 2012). Data collected from the Michigan State University Old
Mission weather station. (Source: http://www.agweather.geo.msu.edu/mawn/
station.asp?id=old&rt=24) .......................................................................................82

ix

Figure 12:

Mean concentration of Nitrate-Nitrogen (NO3-N) in collected runoff samples for
each individual roofing treatment during the study period (22 July 2010 – 24
September 2012) ......................................................................................................83

Figure 13:

Mean concentration of Nitrite-Nitrogen (NO2-N) in collected runoff samples for
each individual roofing treatment during the study period (22 July 2010 – 24
September 2012) ......................................................................................................84

Figure 14:

Mean concentration of Ammonia-Nitrogen (NH3-N) in collected runoff samples for
each individual roofing treatment during the study period (22 July 2010 – 24
September 2012) ......................................................................................................85

Figure 15:

Mean concentration of Phosphate-Phosphorous (PO4-P) in collected runoff samples
for each individual roofing treatment during the study period (22 July 2010 – 24
September 2012) ......................................................................................................86

Figure 16:

Mean Turbidity (NTU) levels in collected runoff samples for each individual
roofing treatment during the study period (22 July 2010 – 24 September 2012) ....87

Figure 17:

Mean pH levels in collected runoff samples for each individual roofing treatment
during the study period (22 July 2010 – 24 September 2012) .................................88

x

INTRODUCTION
This thesis contains a study performed to evaluate the effect of vegetation selection on
green roof design, it is aimed at enhancing the performance capabilities of these roofing systems
to mitigate the negative stormwater runoff quality characteristics of green roof systems. Some of
these undesirable stormwater quality characteristics include increases in nutrient levels and
excess sediment (turbidity) in runoff. These two types of contaminants can affect both the
chemical and physical characteristics of receiving water bodies to the point where severe damage
can be done to aquatic environments.
Water runoff samples were collected from simulated sloped green roof installations and
analyzed for specific water quality properties including various nutrients (Nitrate-Nitrogen (NO3N), Nitrite-Nitrogen (NO2-N), Ammonia-Nitrogen (NH3-N), and Phosphate-Phosphorous (PO4P)), turbidity, and pH levels. Two different types of vegetation coverage including
conventionally used Sedum species (Sedum acre, S. album, S. ellacombianum, S. kamsatchatkum,
and S. reflexum) and more unconventional native grass species (Buchloe dactyloides and
Bouteloua gracilis) were monitored and compared to three types of non-vegetated surfaces (bare
green roof media, asphalt shingles, and steel roofing). Roof platforms designed to represent
actual green roof installations were built in 2009 using these five roofing types. Each platform
was constructed with a 30° slope to mimic conventional 4:12 housing pitch found on the
majority of Midwestern homes. Green roof stormwater runoff produced from natural rain events
was collected during the growing seasons of 2010, 2011 and 2012.

1

LITERATURE REVIEW
Effects of Urbanization on the Environment
Human influence on the natural environment is an inevitable part of man’s existence on
earth, but the extent to which mankind alters our environment is under constant debate. As
human populations have grown, so has our footprint. Today, 30-50% of the earth’s surface is
now estimated to have been transformed (Vitousek et al., 1997). Transformations of the earth
include deforestation, urbanization, and agricultural practices that directly and indirectly
influence the natural processes of the previously undisturbed land. As a result of these changes
in the earth’s land cover, other natural resources like air, water, soil, vegetation, and wildlife
have been altered. Vital natural processes such as the hydrologic cycle and regular climate
events have been impacted to a degree that humans have been forced to address these issues in
certain ways, some more unconventional than others. We now try to manage certain components
of these natural cycles in attempt to reverse negative effects that have been caused by poor land
use choices.
Green infrastructure has the ability to reduce the negative hydrological and climatic
effects of urbanization, facilitate more ecosystem services within the urban environment by
supporting biodiversity, and provide new business opportunities in green technology through
green roof design, installation, maintenance, plant propagation, and material manufacturing.
The most influential component of urbanized areas having the greatest effects on the
hydrologic cycle is the total amount of impervious area (TIA). Impervious or “hardened”
surfaces such as roadways, parking lots, sidewalks and rooftops do not allow precipitation to
infiltrate the ground to naturally replenish stream, river and groundwater reservoirs and/or
encourage evapotranspiration (Figure 1). Infiltration of water into the soil profile and

2

evapotranspiration from photosynthesizing plants are both key forces in the hydrologic cycle. In
the United States, Ferguson (1998) estimates that 11-39% of residential areas and 71-95% of
industrial, commercial, and shopping districts are comprised of impervious surfaces.
Because of urbanization, replacement of vegetation with impervious or hardened surfaces
has caused significant changes in water runoff, evapotranspiration, and water infiltration
components of the hydrologic cycle. As the percent of impervious surfaces increases, the area of
land available for infiltration and evapotranspiration of precipitation is reduced. The amount of
stormwater available for direct surface runoff also becomes greater, leading to increased
frequency and severity of flooding and sediment erosion (National Research Council, 2009). As
little as 10% impervious surface in watersheds can affect downstream water bodies to the point
where aquatic habitats can be altered (Shaver et al., 1995). These habitat alterations can include
reduced available oxygen in water, increased water temperature and increased vegetation growth;
such factors, for example, can lead to a reduction of invertebrate and fish species population and
diversity. Excess runoff can also inhibit water filtration facilities from functioning properly, and
in some cases increases, the occurrence of combined sewer overflows (CSOs). In combined
sewer systems, wastewater and stormwater are not separated from one another. High volumes of
stormwater runoff overwhelm the systems and cause CSOs, allowing wastewater and stormwater
to be directly released into receiving streams, lakes or oceans (Figure 2). As of 2004, 746
communities in the United States were still operating with combined sewer systems with nearly
half of the communities being located in the New England and Great Lakes regions (USEPA,
2008a).
Large amounts of time and money have been spent on infrastructure upgrades to address
the negative effects of CSOs in the United States. In 1987, the Clean Water State Revolving

3

Fund (CWSRF), a national program, was established under the federal Clean Water Act as a
means to partially fund water quality projects. As of 2002, state and local expenditures reported
under the CWSRF reached $0.44 billion per year. This does not account for communities with
projects that are only partially funded or choose not to participate in the program at all (USEPA,
2004).
Since the mid-1990s, a growing number of municipalities have been implementing
stormwater utility fee programs that affect individual residential, commercial, and industrial
property owners. The utility fee is determined by the percentage impervious surface area of each
individual property. In 80% of all stormwater utility fee assessment programs, the equivalent
residential unit method (ERU) is used to determine the fee owed for each parcel. Regardless of
the parcel’s total area, the ERU method bills property owners for an amount proportional to the
impervious area on a parcel using a typical single-family home’s impervious footprint as the
basis of measurement. The revenue from these programs is mainly used to provide support for
annual stormwater facility operational costs by funding construction projects like facility
upgrades and CSO programs. In the case of South Burlington, Vermont, a stormwater utility
established in 2006, generates more than $1 million annually (USEPA, 2009a).
Increased stormwater runoff not only affects the quantity of water entering aquatic
systems, but also alters the water quality of these aquatic systems in several ways. During
precipitation events, stormwater runoff from impervious surfaces can pick up pollutants such as
oils, metals, soil particles, salts, animal wastes and fertilizers. The pollution load from urban
area runoff is significantly higher than from secondary domestic sewage effluent and is
considered to be second in significance only to runoff from agricultural lands (Deletic and
Maskimovic, 1998). Increased pollutants in water systems can increase the biological activity of

4

aquatic vegetation such as algae; this process is known as eutrophication, which increases
biological oxygen demand and reduces dissolved oxygen levels (Figure 3). This process proves
harmful to aquatic ecosystems and can reduce or eliminate certain aquatic species. Water
temperature also is an important factor of stormwater runoff that can influence the water quality
of receiving water bodies. Urban runoff is typically higher in temperature than natural runoff or
groundwater supplies. This can have a significant effect on aquatic biota because as water
temperatures increase, dissolved oxygen in the water decreases (Schueler, 1987).
Much research had been done with respect to water quality beginning in the mid-to-late
1900s due in part to the formation of the Federal Water Pollution Control Act, or Clean Water
Act in 1972. This law was originally established in 1948. In 1972 it underwent complete
revisions, most notably revisions that called for ambitious programs aimed at enhancing surface
water quality. This act, which governs many aspects of human interaction and influence on the
land, has since been expanded and continues to be a major influence in environmental protection.
The majority of the Clean Water Act is comprised of two parts. One part frames the
authorization of government financial assistance to fund municipal infrastructure projects such as
sewage treatment plant construction or combined sewer system separation (i.e. the Clean Water
State Revolving Fund). The other part sets regulatory requirements to be followed by municipal
and industrial dischargers. These requirements originally focused on point source pollution
which is described as pollution being discharged from a single identifiable source such as pipes
and outfalls. Amendments in 1987 specifically identified and authorized measures to address the
issue of non-point source pollutants such as stormwater runoff from farm lands, forests,
construction sites, and urban areas. These non-point source pollutants are now estimated to
represent more than 50% of the nation’s remaining water pollution sources (Copeland, 2010).

5

Removal of natural vegetation to permit urbanization can greatly impact the immediate
climate in urban areas (Figure 4). The differences in the thermal characteristics of surface
materials that overlie urban areas versus those that overlie natural areas have profound
implications not only for microclimates, but also for stream and watershed health (Barnes et al.,
2001). Vegetated areas naturally shade surface materials and cool the air through
evapotranspiration. Large amounts of unshaded surfaces, like those found in urban areas, can
absorb heat from the sun and serve as a heat “sink”, increasing their own temperature and the
temperature of the surrounding areas. Also, since impervious surfaces make up the majority of
the urban landscape, water is not allowed to infiltrate into the soil to be evaporated or transpired
by plants. Evaporation of moisture serves to lower surface temperatures since the energy used in
evaporation is not available for sensible heating (heat exchanged by a body that creates a change
of temperature) of the surrounding area (Barnes et al., 2001).
The urban heat island effect, along with increased albedo (a coefficient that describes the
reflectiveness of a surface) of the urban area, cause greater internal building temperatures, and in
turn increases demand for summertime cooling, and increased energy expenses. For every 0.06°
C (1° F) increase in summertime temperature, peak utility loads in medium and large cities
increase by an estimated 1.5 – 2.0 percent (USEPA, 2003). In the United States, about one sixth
($40 billion/year) of all electricity generated, is used to air condition buildings. Of the $40
billion/year, about half is used in cities classified as “heat islands”, where the air conditioning
demand has risen 10% within the last 40 years (Rosenfeld et al., 1998).
Increased temperatures due to the urban heat island effect also increase the formation of
ground level ozone through the combination of volatile organic compounds and nitrogen oxides.
Ground level ozone is a public health hazard that can cause respiratory and cardiovascular illness

6

(Lo and Quattrochi, 2003). Also, the increased temperature in urban areas contributes to about
1,800 deaths per year in select metropolitan cities with excessive heat events; this phenomenon
has the greatest impact on the Northeast and Midwest regions (USEPA, 2006). Other than
mortality, increased daytime surface temperatures, reduced nighttime cooling, and higher air
pollution levels associated with urban heat islands can affect human health by contributing to
general discomfort, respiratory difficulties, heat cramps, exhaustion, and non-fatal heat stroke
(USEPA, 2008b).
Green Roof History
Green roofs have not always looked or served the same functions as we know them to do
today. Ornamental roof gardens seem to have originated in the 7th and 8th century B.C in the
Tigris and Euphrates River valleys, purported to produce one of the most well-known rooftop
gardens, the hanging gardens of Babylon (Dunnett and Kingsbury, 2008) (Figure 5). Rooftop
gardens also were seen in 16th -17th century Mexico and 17th century Russia. Canada claims
several Viking and French examples of sod roofs, exported to Newfoundland and Nova Scotia by
early Norse or French explorers, respectively (Peck et al., 1999). Sod roofs have been
traditionally used in the colder climates of Scandinavia, Canada, and Iceland where seven or
more layers of birch bark serve as a waterproof membrane and local soil is used to weigh the
birch bark down. This practice ensures that the birch bark stays in place. The soil protects the
birch bark and acts as insulation for the structure below. Local soil seed banks would germinate
and spread to cover the roof in native vegetation.
Rooftops in Germany around 1900 consisted of wood construction overlaid with tarboard. To prevent fires, the tar-board was covered with a layer of sand and gravel which
attracted a layer of unintentional vegetation (Werthmann, 2007). This vegetation process

7

intrigued several people and led to modern green roof systems that originated at the beginning of
the 20th century in Germany. Green roofs also were studied and used at that time as a way to
combat the damaging influence of solar radiation on building rooftops. An example of one such
demonstration project is the Moos Lake Water Filtration plant in Zurich, Switzerland, whose
green roof construction in 1914 led to its registration in the new millennium as a world
wildflower preserve due to the indigenous orchid seeds that came with the soil media.
Due primarily to urbanization, growing environmental concerns that arose in the 1970’s
created opportunities for the introduction of more progressive problem solving solutions in green
roof design. Green roofs were accepted because of the multiple environmental benefits
documented in the first volume of green roof technical guidelines published in 1982 by the
German Landscape, Research, Development and Construction Society, an interdisciplinary
research team (Oberndorfer et al. 2007; FLL, 1995). Since then, green roofs have been installed
across Europe and have gained a foothold in America, steadily increasing in numbers over the
past decade. Research continues in both Europe and North America as universities work to
document the impact of green roofs on the environment, economy, and energy resources (Wark
and Wark, 2003)
Today, modern green roofs can be categorized as intensive (having deeper substrate, a
wide variety of plant species, and typically accommodating public foot travel) or extensive
(having minimal maintenance requirements, shallower media depth, possible sloped applications,
and typically not tolerant of public foot travel) (Getter and Rowe, 2006). Green roofs have been
proven to mitigate the negative hydrologic and climatic effects of urbanization. Through their
structural substrate, vegetative, and water holding components, green roof systems are shown in
studies to absorb 75% of precipitation that falls on them, which translates to an immediate

8

discharge reduction of 25% of normal levels (Peck et al., 1999). Green roof albedo (ranging
from 0.7 – 0.85) has also been found to be equivalent to other high-albedo, energy efficient
roofs, such as white roofs, which have an albedo of approximately 0.8 (Gaffin et al., 2005).
Lower indoor air temperatures were also measured in buildings with green roofs, and estimated
heating and cooling loads were lower compared to buildings with traditional roofing systems
(Gaffin et al., 2005).
Benefits of Green Roofs
Green roofs are best known for their impact on the volume of stormwater released as roof
runoff. Green roof systems have the capacity to retain large amounts of water because of their
substrate (soil media), vegetation and water retention components (i.e. fleece fabric mats and a
plastic reservoir layer). Water can be held in the porous growing media of green roofs and
evaporated into the atmosphere or used for evapotranspiration by the vegetation. Due to these
and other components of the overall system, green roofs have been shown to retain anywhere
from 39-85% of received rainwater (VanWoert et al., 2005a; Teemusk and Mander, 2007;
Monterusso et al., 2004; Cronk, 2012). These retention rates differ significantly compared to
conventional flat roofs which have a typical slope of about 2% and can consist of stone ballast,
tar, steel or shingle roofing material. These conventional roofing systems are shown to retain
approximately only 4% of fallen rainwater (Mantens et al., 2006), which is probably due to
evaporation. A study done by Carpenter (2011) compared three types of roofs; all roofs were
full-scale and included green (vegetated), stone ballast roof, and asphalt roofing materials. A
total of 21 storm events were observed. It was found that the green roof had an overall retention
rate of 68.25% while the stone ballast roof had a retention rate of 57.75%. The study showed

9

that conventional roofing materials had much lower stormwater retention capabilities, with the
asphalt roof retaining only 10.45%.
Green roofs first evolved using vegetation from native soil seed banks to weigh down and
protect layers of birch bark for waterproofing. More recently, green roof design has evolved to
allow the facilitation of a larger plant palette ranging from the widely used Sedum spp. to larger
plants or grasses and even trees. When looking at some of the more commonly used plants used
in these systems, (Dunnett and Kingsburry, 2008) found that vegetation does in fact influence the
amount of runoff from green roof systems. When comparing three different vegetation types
(grass species, forb species, and Sedum species), it was found that grass species were the most
effective for reducing runoff, followed by forbs, and then Sedums. Plant attributes such as height,
density, leaf shape and orientation, and evapotranspiration rates affect the amount of runoff
released. This information is important to further green roof design and maximize the potential
for green roofs to mitigate stormwater issues in urban areas.
Roof slope also has an effect on stormwater retention capabilities of green roofs which
can be attributed to the influence of gravity on the movement of water through the green roof
system. Getter (2007a) used experimental green roof platforms at four different slopes to
investigate the water retention rate differences under these conditions. All of the green roof
platforms were designed in the same fashion, allowing the change in slope to be the only
variable. The platforms were set at slopes of 2%, 7%, 15%, and 25% were found to have overall
rainwater retention of values of 85.2%, 82.2%, 78.0%, and 75.3% respectively. This shows a
direct correlation between roof slope and rainwater retention capabilities with flat roofs retaining
more stormwater in the system.

10

Several investigations have found that green roofs can be an effective stormwater
management tool due to their retention and evapotranspiration properties. Traditional roofing
installations that may use shingle, tar, or metal materials, for example, act as impervious surfaces
and will create higher stormwater runoff. This fact has been noticed by policy makers and
government entities to the point that green roofs have been advocated and encouraged to help
manage stormwater problems in urban areas. In situations where governments impose a
stormwater runoff utility fee, green roofed areas are considered a porous surface and will not be
included in a property’s overall impervious footprint. This can be a great financial benefit to
landowners as pointed out by Clark (2008) who compared traditional roofing practices versus
green roof installations with respect to assessed stormwater utility fees in Ann Arbor, MI. The
city of Ann Arbor assesses individual properties based on the overall square foot for residential
use and square acre for industrial use of their impervious surface area, which includes rooftops
with traditional roofing materials. Clark calculated that a conventional roof of 2000 m² (21,527
ft²) would be assessed $520 per year, whereas a green roof of the same size would be assessed $0
per year. This is an example of how green roofs provide a financial incentive for installation in
addition to their documented environmental benefits.
The urban heat island effect is a phenomenon that describes the increased ambient air
temperature in urban areas due to the increase of mechanical operations, reduced albedo of road
and building surfaces, and reduction of vegetation that creates shade and cools the air through the
process of evapotranspiration. This occurrence has been considered a contributor to many issues
including increased building energy consumption for cooling and increased adverse health
effects due to extreme summer temperatures in urban areas. Much of the urban landscape
consists of densely located buildings where the rooftop represents the footprint of each building.

11

Therefore, green roofs have a potentially positive impact on the surrounding temperature due to
several factors.
Traditionally constructed rooftops, especially on office or industrial buildings, use
materials that have a low albedo (low reflective, high absorptive heat properties). These
materials include, but are not limited to, tar roofs and darker colored asphalt shingles both of
which will absorb more solar energy. This increases air temperature immediately above the roof
when compared to vegetated surfaces.
Atmospheric air quality has been an issue of concern due to the increased burning of
fossil fuels and the subsequent release of airborne pollutants into the atmosphere. This issue is
of greatest concern in and around urban areas where there is a higher concentration of industrial
activity and increased transportation operations such as the use of fossil-fuel burning
automobiles. Green roofs have been identified as a way to reduce problems caused by airborne
pollutants, namely the reduction of nitrogen oxides, carbon dioxide, polycyclic aromatic
hydrocarbons (PAHs), and heavy metals in the environment. Rowe (2011) has summarized
several studies that have addressed this issue as well as discussed the importance of green roofs
as a tool for remediating environmental pollution on many levels.
Clark (2008), analyzing the many benefits of green roofs, estimated that a 2000 m2 green
roof could take up approximately 530 kgno2/year (variance: 700 kg²no2/year²) . This translates
into $890 (variance: 2.0E6 $²) -$3390 (variance: 2.8E7 $²) of public health benefits per person.
This data was derived from studies done by the United States Environmental Protection Agency
(USEPA) that placed economic values on NOx removal with respect to its influence on
premature death and cases of chronic bronchitis. In addition to the potential NOx removal by
green roofs, their ability to sequester carbon also has been investigated. Getter (2009a) looked at

12

green roofs planted with primarily Sedum species and analyzed the amount carbon content over a
2 year period. She found that after two years, the twenty experimental plots examined (13.06 m2
in total area), sequestered 375 g C·M-2 by means of above and below ground biomass and
substrate organic matter. This shows that green roofs can be a significant air pollutant removal
tool, especially when applied on a larger, city-wide scale.
Green Roofs and Stormwater Runoff Quality
As previously mentioned, green roofs are best known for their positive impact on
stormwater runoff issues. This is particularly true in urbanized areas with large amounts of
impervious surfaces. In addition to the research conducted on stormwater quantity, there also
have been investigations of green roofs with regards to the quality of the stormwater runoff.
In most instances, natural rainwater falls and lands on the ground or on rooftops
constructed out of traditional roofing materials. This rainwater then runs off the surface quickly,
and makes its way to a roof gutter, storm sewer system, or natural body of water. This is the
same with runoff released from green roofs, but unlike roofs with traditional roofing materials,
this rainwater must first pass through the several layers contained in a green roof system. This
introduces new variables in the equation. It has been found that stormwater released from green
roof systems is significantly different in its physical and chemical properties than natural
rainwater or rainwater runoff from roofs with traditional roofing materials. Even though water
quality has not been as thoroughly investigated as water quantity, there are studies that have
attempted to analyze this issue. The main investigation in researching green roof runoff water
quality is whether or not green roofs can improve stormwater runoff quality (compared to runoff
from roofs using conventional roofing materials or uninhibited natural rainwater) or if they

13

introduce or increase additional pollutants otherwise non-existent in natural rainwater or
traditional runoff.
Rowe (2011) summarized the majority of research studies pertaining to water quality of
green roof runoff, and it was found that there exists wide variation in the findings (Alsup et al.,
2010; Bliss et al., 2009; Carpenter and Kaluvakolanu, 2011; Bernstsson, 2010; Berndtsson at al.,
2009; Berndtsson at al., 2006; Emilsson et al., 2007; Gregorie and Clausen, 2011; Hathaway et
al., 2008; Monterusso et al., 2004; Retzlaff et al., 2008; Teemusk and Mander, 2007; Toland et
al., 2012; USEPA, 2009; Vijayaraghavan et al., 2012). Among water quality characteristics that
have been investigated are various nutrients (K, Na, NO3-N, NO2-N, NH4-N, PO4-P, Total N,
Total P and SO4), metals (Cd, Cr, Cu, Fe, Mg, Mn, Ni, Pb, Zn), polycyclic aromatic
hydrocarbons (PAHs), chemical oxygen demand (COD), biological oxygen demand (BOD),
electric conductivity (EC), turbidity, color, and pH. All of these variables affecting water quality
can come from different sources, whether from the natural environment or anthropogenically
introduced, and all play significant roles in the natural balance of our earth’s many ecosystems.
This study focuses on the concentration levels of different forms of nitrogen and phosphorous in
stormwater emanating from two types of green roofs, a brown roof (no vegetation), and two
types of conventional roofing (stell and fiberglass shingle). It also includes investigations of
turbidly and pH levels from these roof types.
When examining the roll of nutrients in the natural environment, it is known that nitrogen
and phosphorous are important in plant growth and establishment. Their use as fertilizers in crop
production and their detrimental effects when introduced into the environment in above average
levels is well established in the literature. Several forms of nitrogen are readily available in the
environment including the forms of nitrate (NO3-), nitrite (NO2-) and ammonia (NH3). These

14

three forms are part of the nitrogen cycle and are products of nitrogen fixation and nitrification
processes. NH3 occurs when dinitrogen gas (N2), the most abundant and unusable form of N in
the biosphere, is fixated by bacteria in the soil to create ammonia (NH3). NH3 can appear as a
gas, moving freely throughout the soil, or dissolved in water. When dissolved in water, the NH3
can be transported much more easily throughout the water column and into the groundwater
supply. This natural production and movement of NH3 within the hydrologic cycle tends to
produce very minimal negative effects on aquatic species when transported to rivers, streams and
water bodies of varying sizes. Higher levels of NH3 and subsequent pollution of natural waters
tends to come from other man made practices such as fertilization and the use of organic
compost in farming practices. These larger amounts of NH3 can be detrimental to aquatic
organisms such as fish species, especially in waters of higher temperature and lower pH levels.
It has been shown that high concentrations of NH3 in water reduces hatching success and growth
rate in fish, including morphological development. It also leads to pathological changes in the
cell tissues of gills, liver, and kidneys. Exposure to higher concentrations of NH3 (ranging from
0.304 mg/L – 1.2 mg/L of NH3 in this study, uncorrected for pH) also causes loss of equilibrium,
hyperexcitibility, and increases in breathing, cardiac output and oxygen uptake. Fish have
displayed convulsions, coma and death (USEPA, 1985) under ammonia poisoning. In humans,
this phenomenon is very rare and can only occur when introduced in extremely high levels.
The formation of nitrite (NO2-) is a product of the nitrification process within the larger
nitrogen cycle. Bacteria play a role in this process. Ammonium (NH4+, an ionized form of NH3)
is converted into NO2-. From this, NO2- is transformed into nitrate (NO3-) by denitrifying
bacteria (Robertson and Groffman, 2007). Due to the ease of this transformation process, NO3-,
the end product of nitrification, is generally in greater abundance than NO2-. Since nitrite and

15

nitrate act similarly within the environment and produce many of the same effects, the two are
often combined and categorized together under the term nitrates. These nitrates, much like
ammonia, occur naturally in the environment and are vital to many ecological and biological
processes. Naturally occurring, these highly soluble nitrates generally produce no harmful side
effects when they are transported through the soil column and into the groundwater supply or to
natural water bodies. The problem arises when increased levels of nitrates are introduced
through the use of agricultural or home use fertilizers or as byproducts of untreated animal waste.
This makes areas around home and agricultural lands of great importance due to the increased
incidence of nitrates; these forms of nitrates are able to reach drinking water sources through soil
infiltration to the water table or by overland flow to streams, rivers and lakes.
Human exposure to above average levels of nitrates can be, at times, very detrimental to
one’s health. When overexposed, infants and children can develop what is called “blue-baby
syndrome” which is seen most often when infants are exposed to nitrates from drinking water
used to make formula. Adults diagnosed with chronic abstractive pulmonary disease also are
susceptible to impaired oxygen delivery to vital organs due to nitrate affinity for hemoglobin
(Westphal, 2014). This syndrome is named for the blue coloration of the skin caused by
impaired oxygen delivery to tissues due to the nitrate’s interruption of regular hemoglobin
function. Increased nitrate exposure also can increase the incidence of brain tumors, leukemia,
nose and throat tumors in children and cancer in adults (USEPA, 2007).
In addition to the negative human health effects of nitrates, NO2-, NO3- and NH3 also alter
natural ecosystems and can cause significant, unwanted outcomes. These same environmental
effects also are apparent when there is an increase of the nutrient phosphorous (P) into the
environment. Phosphorous is not found often as a free element, P, in the environment because of

16

its high reactivity with other elements like hydrogen and oxygen. When phosphorous goes
through its natural oxidation process, the final product is orthophosphate (PO4); this ion is taken
up by virtually every plant and used in almost every metabolic process including respiration and
photosynthesis. Though it is an important nutrient involved in many plant functions, it is often
the least available nutrient in a great number of aquatic and terrestrial ecosystems. Therefore, it
is often the limiting component to complete various environmental processes, both good and bad.
This means that a small amount of phosphorous allows many processes to be completed (e.g.
algae blooms) that eventually devoid water systems of oxygen. A large source of phosphorous
introduced into the environment comes from the use of fertilizers for crop production that
supplement soils that are naturally deficient of orthophosphates, the main form of phosphorous
taken up by plants. Introduction of phosphorous into the environment is evident by the fact that
90% of mineral orthophosphate used worldwide is derived from fertilizers (McDonald et al.,
2001).
Phosphate and nitrogen are essential to many plant functions but also are drivers of
several unwanted environmental outcomes when introduced at higher than average levels to the
environment. As discussed previously, anthropogenic introduction of these nutrients into the
environment can be directly linked to agriculture practices such as the application of fertilizers or
livestock grazing that produces untreated livestock waste. Though a large contributor,
agriculture practices are not the only source of nitrogen and phosphorous in the environment.
Residential, commercial and municipal landscape application of fertilizers, leaky septic tanks,
combined sewer overflows, atmospheric deposition, and as by-products of mining practices also
contribute these nutrients to groundwater supplies by overland runoff or by infiltrating through

17

the soil column into the water table. This transport of nutrients to various water bodies can lead
to eutrophication, the ecosystem’s response to the addition of artificial and natural substances.
Eutrophication of aquatic systems increases the growth of aquatic plant and animal
species, such as blue-green algae, creating elevated biomass levels. These plants and organisms
will eventually die and decompose. The process of decomposition uses oxygen, and therefore,
the more plant and organism biomass that decomposes, the less oxygen is available for all other
processes within that aquatic environment. These lower oxygen levels (low dissolved oxygen
content) can create numerous negative environmental effects including shifts in aquatic plant and
organism species abundance and composition, reduced water clarity, decreased aesthetic value,
changes in water taste and odor, water supply filtration problems, elevated pH, shifts in fish
species composition towards less desirable species, and probability of fish kills (Smith et al.,
1999).
The issues of increased nutrient introduction into the environment are well documented
and are of obvious concern when examining aquatic systems and groundwater supply near and
around agricultural land. Although these areas are of great concern, there also are concerns that
arise near and around urbanized areas. In urban areas, stormwater that runs off of a landscape
comprised of impervious surfaces is channeled directly into stormwater sewer systems in many
cases. Some of these systems are connected to water treatment plants that try to filter as many
pollutants picked up by the stormwater as possible. In many instances though, the stormwater is
drained directly to receiving rivers, streams, lakes or other water systems. This means that all
pollutants, including nutrients such as nitrates and phosphates, are picked up during the
stormwater’s travels and are directly introduced in greater than normal levels into these aquatic

18

systems. In high enough concentrations and under certain circumstances, these pollutants can
cause detrimental effects to these aquatic systems and their connected environments.
Green roofs can provide many environmental benefits including increased avian and
insect habitat (Eaken 2012; Monsma, 2012), increased energy savings due to added building
insulation, and cooler surrounding temperatures due to their increased albedo and
evapotranspiration properties. Although these are well documented motives to install green
roofs on new and existing structures, the ability to manage stormwater is probably the greatest
benefit when considering a green roof in areas with high impervious surfaces, such as urban
areas. The ability to retain large amounts of rainfall and prolong the release of any runoff makes
green roofs a useful tool to reduce the impact on urban stormwater systems. While stormwater
quantity has been heavily researched, fewer studies have examined the quality of stormwater that
runs off of these green roof systems.
Green roof systems can vary in several ways; for example, they can be a built-up system
(soil placed directly on roofing components and planted) or a modular system (pre-grown green
roof cells or trays). These different systems have different structural materials (e.g., plastic trays
or drainage layers, rubber for roofing membranes, fabric water retention layers, etc.) and can use
growing media comprised of different mixes. Each variation in structure supports an array of
plant species adapted to survive the extreme climate found on rooftops. As stormwater falls on
these green roof systems, it travels through and makes contact with these different materials; it
picks up pollutants and particles along the way. The unnatural additions found in the subsequent
stormwater runoff can alter the physical and chemical properties of the released water and in
many cases, are characterized as pollutants to the surrounding environment. This water,
especially in urban environments, can make its way directly to receiving aquatic systems with

19

minimal, or no filtration at all. This makes examining the quality of stormwater runoff from
green roofs an increased environmental concern as it can have direct implications on the health
of receiving rivers, streams, lakes and other water bodies.
Pollutants such as metals have been examined in green roof runoff as well. Alsup et al.
(2010) concluded that different growing substrates (soils) and different vegetation forms can
influence the level and type of metal pollutants contained in green roof runoff. She found that
none of the green roofs treated with different substrates were significant sources of Chromium
(Cr), Copper (Cu), Iron (Fe), Nickel (Ni), or Zinc (Zn). There was evidence, however, that one
substrate, Lassenite, was the only substrate tested that could be considered a source of
Magnesium (Mn). This occurrence did show Mn in greater concentrations than what is
recommended by USEPA water quality criteria, but in non-toxic amounts that only affect the
taste of drinking water and the possibility of staining of cloth when washed with water
containing Lassenite. The work of Alsup, et. al (2010) performed on substrates in planting plots
with and without vegetation, found that leachate from 45.4% of the examined plots exceeded
USEPA standard with regard to lead (Pb) concentrations.
Berndtssson et al. (2006) examined stormwater runoff from full scale green roofs of
different ages with respect to concentrations of cadmium (Cd), chromium (Cr), copper (Cu), Iron
(Fe), potassium (K), magnesium (Mn), lead (Pb), zinc (Zn), nitrate-nitrogen (NO3-N), ammonianitrogen (NH3-N), total nitrogen (Tot-N), phosphate-phosphorous (PO4-P) and total phosphorous
(PO4-P). It was found that the examined green roofs acted as a source of these nutrients in green
roof runoff when compared to natural rainwater with an exception to NO3-N. The authors found
that nitrate-nitrogen was retained by the green roofs, either by their vegetation, growing media,
or both. In addition, Phosphate-phosphorous was found to be released in greater amounts from

20

green roofs that were newer in establishment. The oldest green roof performed better with
respect to phosphate-phosphorous release which leads one to believe that phosphate-phosphorous
levels in green roof runoff can decrease over time as a green roof system matures.
In an experiment conducted by Bliss et al. (2009), the authors compared an extensive
green roof to a conventional ballast roof to examine water retention capabilities and quality of
stormwater runoff with respect to sulfate (SO4), phosphorous (P), chemical oxygen demand
(COD), Tot-N, Pb, Zn, and Cd. While the green roof reduced runoff volumes by up to 70%, it
was found that stormwater runoff from the observed green roof had higher concentrations of
phosphorous and elevated levels of COD when compared to the conventional ballast roof. Like
Bliss et al. (2009), Teemusk and Mander (2007) found that green roofs increase pH (less acidity)
of stormwater runoff from green roofs. They also found that the green roofs performed better
than the observed runoff from the studied bituminous roof with respect to pollutant
concentrations in stormwater runoff during light and moderate rains. During heavy rains though,
green roofs demonstrated higher pollutant concentrations in stormwater runoff due to the
accumulation of the pollutants in the green roof system that were eventually flushed out during
heavy rain events.
Berghage and his co-authors, in a study reported by the USEPA (2009b), examined
smaller green roof installations and their stormwater runoff properties with respect to pH, electric
conductivity (EC), turbidity, color, nitrate, P, K, Ca, Fe, Mg, Mn, Na, Zn, and Sulfur (S). The
green roofs proved to release stormwater with higher concentrations of nutrients (K and P) as
well as demonstrating greater hardness with higher levels of Ca and Mg. Though there showed
greater release of some pollutants, green roofs appeared beneficial in removing atmospheric
nitrate in this study. They also examined the concept of pollutant loading in green roof; this

21

concept compares minimal runoff with higher pollutant concentrations from green roofs to
higher runoff volumes with lower pollutant concentrations from conventional roofs. They found
that when factoring in the total volumes of stormwater with respect to pollutant concentration
levels, that green roofs acted similarly to conventional roofs with respect to pollutant loads in
stormwater runoff.
At a study performed at the Michigan State University campus, Monterusso et al. (2004)
compared four commercially available green roof systems. Each system was planted with Sedum
spp. and native grass species. The research monitored the stormwater runoff quantity and quality
with respect to various systems and vegetation coverage types. The authors examined
stormwater runoff quality with respects to NO3-N and Tot-P. There appeared to be no
correlation of phosphorous concentrations in stormwater runoff and green roof system type.
However, it was found though that green roofs planted with Sedum spp. and with shallower
substrate released greater amounts of nitrate in stormwater runoff. This may indicate that green
roofs with native grass vegetation may produce different results than green roofs planted with
Sedum spp. when looking at pollutant concentrations in stormwater runoff.
Furthering research on the subject, Hathaway et al. (2008) collected rainfall runoff from
an extensive green roof in North Carolina planted with predominately Sedum species. The
rainfall runoff was examined for levels of TKN (Total Kjeldahl nitrogen), NO3 + NO2, NH3,
Total Nitrogen (TN) and Total Phosphorous (TP). During the course of this study, it was
observed that this green roof consistently released greater amounts of TN and TP in its
stormwater runoff when compared to rainfall and runoff from the adjacent gravel ballast roof.
Emillson et al. (2007) also looked at different forms of nitrogen and phosphorous in
relation to fertilization treatments of green roof and subsequent levels in stormwater runoff.

22

Traditional and controlled release fertilizers (CRFs) were applied to experimental green roof
plots in attempts to determine overall nutrient levels, nutrient level decline over time, and
influence of vegetation composition on nutrients in stormwater runoff. In the controlled
greenhouse experiment, experimental plots with three coverage types, including shoot-grown
vegetation (S. album, S. acre, and Phedimus spurius), established pre-grown vegetated mats (2
years old at time of study) and plots with no vegetation (growing media only) were satisfied. It
was found that traditional fertilizer applications released greater amounts of nutrients in runoff
when compared to applications using CRFs. With respect to vegetation coverage type, it was
found that plots with vegetated coverage generally released runoff with lower levels of nitrogen
compounds. This also was true with respect to phosphorous; the un-vegetated plots released
greater amounts of phosphorous in runoff. This demonstrates that vegetation coverage does
have an influence on nutrient concentrations in stormwater runoff.
A study performed in Southeast Michigan by Carpenter and Kaluvakolanu (2011)
compared green roof, gravel ballast and asphalt roofing applications with respect to stormwater
retention and stormwater runoff quality. Unlike the majority of other studies that examined
nutrient runoff from green and traditional roofs, this experiment showed that total phosphorous
runoff from green roofs was less than that of the gravel ballast and asphalt roofs. Nitrate
concentrations were highest from the stone roof with the green roof and asphalt roof releasing
similar levels in stormwater runoff. Although the phosphorous levels in runoff from green roofs
were uncharacteristically lower than the traditional roofing materials, the green roof showed
statistically greater amounts of total solids in runoff, which is congruent with the majority of
research performed on green roof systems.

23

In examining the past research performed on green roofs with respect to water quality of
stormwater runoff, it is evident that there are many factors that can influence the concentration of
metals and nutrients in released stormwater. Substrate composition, system materials,
maintenance practices, roof age, vegetation composition, and location are just some of the factors
that can influence green roof runoff quality. With respect to nutrient, metal and sediment loads
in stormwater runoff, it has been shown that green roofs can perform both better or worse than
systems using traditional roofing materials. This study aims to reduce as many influencing
factors as possible and focus on green roof vegetation composition and its influence on the
quality of green roof runoff.

24

HYPOTHESIS
This study compares green roof systems of the same construction, materials, and slope
with vegetation type being the only differentiating factor (ie., Sedum, native grass, and media
only). The goal is to compare the quality of stormwater runoff from these green roof systems in
attempt to determine the best green roof vegetation selection to improve stormwater runoff
quality.
Null Hypothesis 1
There is no relationship between green roof vegetation treatment and concentration levels
of Nitrate-Nitrogen (NO3-N), Nitrite-Nitrogen (NO2-N), Ammonia-Nitrogen (NH3-N) and
Phosphate-Phosphorous (PO4-P) in stormwater runoff from these systems.
Hypothesis 1a:
There is a significant (p≥0.05) relationship between green roof vegetation treatments and
concentration of the nutrients Nitrate-Nitrogen (NO3-N), Nitrite-Nitrogen (NO2-N), AmmoniaNitrogen (NH3-N), and/or Phosphate-Phosphorous (PO4-P) in stormwater runoff with the native
grass roofing treatment releasing runoff with lower concentrations of these nutrients than the
Sedum roofing treatment.
Hypothesis 1b
There is a significant (p≥0.05) relationship between green roof vegetation treatments and
concentration of the nutrients Nitrate-Nitrogen (NO3-N), Nitrite-Nitrogen (NO2-N), AmmoniaNitrogen (NH3-N), and/or Phosphate-Phosphorous (PO4-P) in stormwater runoff with the native
grass roofing treatment releasing runoff with lower concentrations of these nutrients than the
media-only roofing treatment.

25

Hypothesis 1c
There is a significant (p≥0.05) relationship between green roof vegetation treatments and
concentration of the nutrients Nitrate-Nitrogen (NO3-N), Nitrite-Nitrogen (NO2-N), AmmoniaNitrogen (NH3-N), and/or Phosphate-Phosphorous (PO4-P) in stormwater runoff with the Sedum
roofing treatment releasing runoff with lower concentrations of these nutrients than the mediaonly roofing treatment.
Null Hypothesis 2
There is no significant (p≥0.05) relationship between green roof and traditional
(steel and shingle) roofing treatments with respect to concentration of the nutrients NitrateNitrogen (NO3-N), Nitrite-Nitrogen (NO2-N), Ammonia-Nitrogen (NH3-N), and PhosphatePhosphorous (PO4-P) in stormwater runoff from these systems.
Hypothesis 2a
There is a significant (p≥0.05) relationship between green roof and traditional (steel and
shingle) roofing treatments with respect to concentration of the nutrients Nitrate-Nitrogen (NO3N), Nitrite-Nitrogen (NO2-N), Ammonia-Nitrogen (NH3-N), and/or Phosphate-Phosphorous
(PO4-P) in stormwater runoff with all green roof treatments (sedum, grasses or media only)
releasing stormwater runoff with higher concentrations of these nutrients than both traditional
steel and shingle roofing treatments.
Null Hypothesis 3
There is no significant (p≥0.05) relationship between green roof and traditional roofing
treatments with respect to pH levels in stormwater runoff from these systems.

26

Hypothesis 3a
There is a significant (p≥0.05) relationship between green roof and traditional roofing
treatments with respect to pH levels in stormwater runoff with all green roof treatments releasing
stormwater runoff with higher pH levels (less acidic) than both traditional steel and shingle
roofing treatments.
Null Hypothesis 4
There is no significant (p≥0.05) relationship between green roof and traditional roofing
treatments with respect to turbidity (NTU) levels in stormwater runoff from these systems.
Hypothesis 4a
There is a significant (p≥0.05) relationship between green roof and traditional roofing
treatments with respect to turbidity (NTU) levels in stormwater runoff with all green roof
treatments releasing stormwater runoff with higher turbidity levels than both traditional steel and
shingle roofing treatments.

27

MATERIALS AND METHODS
Green Roof Testing Platforms

In the summer of 2009 five roof testing platforms were built at the Michigan State
University Landscape Architecture Alumni Research Site on the Old Mission Peninsula in
Traverse City, MI (Lat. 44.8822°N, Lon. 85.5059°W, Elev. 210 m (689 ft.)) (Figure 6). Each
platform was constructed using a 1.22 m x 2.44 m (4.0 ft. x 8.0 ft.) sheet of plywood that acted as
roof sheeting. Trusses were built to hold the platforms at a 4:12 pitch (33.33% slope); this pitch
represents the range of typical slopes found on single family structures in Michigan. Each of the
platforms were raised from the ground an average of 1.22 m (4.0 ft) to facilitate stormwater
runoff collection; the platforms were located in an open field away from any obstructions to
rainfall. Platforms were partitioned into three sections, each section having the dimensions of
0.81 m (2.66 ft, width) x 1.22 m (4.0 ft, run) creating a roof area of 0.93 m² (10 ft2). Each section
was lined with 0.045 mm EPDM rubber membrane that acted as the roof waterproofing
protection layer (Figure 7). Partition front, back and side boards were 15.2 cm (6 in.) high to
hold the green roof components and a 2.54 cm (1 in.) drain was cut at the bottom end to allow
each partition to drain separately. A 0.75 cm (.26 in) thick water retention fabric (Xeroflor
XF158, Xeroflor America, LLC, Durham, NC) was placed over the EPDM membrane in each
section and covered the drainage outlet. No drainage layer was incorporated into the partitions
containing the green roofs.
Five different roofing treatments were included in this experiment, and each treatment
was replicated three times (5 platforms with 3 partitions each = 15 partitions) (Figure 8). The
roofing treatments were green roof with Sedum species, green roof with native grass species,
green roof with no vegetation (media only), conventional roof with fiberglass shingle roofing
material, and conventional roof with steel roofing material. All green roof sections were filled to
28

a depth of 15.2 cm (6 in.) with engineered growing media composed of 80% heat-expanded slate
(Permatill, Stalite, Salisbury, NC), 15% USDA grade course sand, and 5%compost from
municipal yard waste. Proportions were mixed on a volume basis. Sedum species selected for
planting were S. acre, S. album, S. ellacombianum, S. kamtschaticum, and S. reflexum and sowed
into growing media at a rate of 0.9 kg (2 lbs.) per square meter. Native grasses included Buchloe
dactyloides and Bouteloua gracilis and were planted as plugs [7.62 cm (3 in.)] at 10.16 cm (4 in.)
on center (Figure 9). The vegetated green roof partitions were irrigated twice daily during the
first three weeks for establishment; no fertilizers were used in this study. The third green roof
treatment with no vegetation contained only growing media of the same consistency and depth as
the planted partitions. The two conventional roofing treatments used were three-dimensional (30
year) fiberglass shingles and galvanized steel roofing; these two treatments represent typical
residential roofing materials found in the region.
Water Collection
Stormwater runoff from each individual partition was collected via a plastic gutter system
that drained into a TB-4 tipping bucket (Campbell Scientific Inc.) for quantity measurements.
An additional tipping bucket was located on site to record rainfall (2011 and 2012 seasons only).
Outflow from the tipping buckets was directed into 18.93 L (5 gal.) plastic carboys by
polypropylene tubing. Growing media moisture content was recorded throughout the study
using CS616 water content reflectometers (Campbell Scientific Inc., Logan, UT). Each green
roof partition was monitored for moisture using two reflectometers: one located 0.30 m below
the top of the partition and one located .030 m above the drainage outlet at the bottom and
inserted 10.16 cm (4 in.) deep into the green roof media. All electronic data was recorded using

29

a CR-1000 datalogger (Campbell Scientific, Inc.) programmed to collect data values every
minute and totaling them every 5 minutes, 24 hours a day during the study.
Water quality samples were collected for individual rain events that occurred during the
growing seasons of 2010 (3 rain events), 2011 (6 rain events) and 2012 (8 rain events). No data
was collected during the winter after the onset of freezing temperatures. Rainfall event
categorizations were guided by a previous study done by Getter et al. (2007a) and were
determined as precipitation events that were preceded, at minimum, by 6 hours of no rainfall.
After rainfall had commenced, a 6 hour waiting period occurred before samples were taken to
allow for adequate stormwater runoff. If precipitation began again during the 6 hour waiting
period, the rain events were combined. Representative samples from the collection carboys were
taken in 500 ml polypropylene sample containers within 24 hours of the rain event ending.
These samples were frozen on-site until transportation to East Lansing could occur for analysis.
Samples were analyzed using various methods to determine levels of the monitored water
quality characteristics. Year 2010 samples tests were conducted at Michigan State University’s
Environmental Engineering Laboratories, East Lansing, MI. All nutrient concentrations (mg/L)
were determined using Vacu-vial ampoules supplied by CHEMetrics, Inc., USA (Samples tested
for: Ammonia-Nitrogen, Salicylate Method; Nitrite-Nitrogen, Azo Dye Formation Method;
Nitrate-Nitrogen, Cadmium Reduction Method; Phosphate-Phosphorous, Stannous Chloride
Method). Turbidity was determined using a Hach 2100N Turbidimeter (NTU) (Hach Company,
Loveland, CO), and pH was determined using an Orion 720A pH Meter (Thermo Fisher
Scientific, Waltham, MA). For year 2011 and 2012, nutrient levels were determined using a
DR/890 Colorimeter supplied by Hach Co., USA (Sample tested for: Ammonia-Nitrogen,
Salicylate Method; Nitrite-Nitrogen, Diazotization Method; Nitrate-Nitrogen, Cadmium

30

Reduction Method; Phosphate-Phosphorous, Ascorbic Acid Method). Sample pH was
determined using a remote reading pH meter supplied by McMaster-Carr, USA.
Data Analysis
Data were analyzed as mean nutrient, pH, and turbidity concentration levels in
stormwater runoff using a One-Way ANOVA model with roof treatments as the independent
variable. For each of the water quality characteristics, significant differences between roof
treatments were determined using post-hoc analysis for multiple comparisons with Tukey HSD
adjustments. All statistics were run using IBM SPSS 20.0 statistical software (IBM Corp.
Released 2011. IBM SPSS Statistics for Windows, Version 20.0. Armonk, NY: IBM Corp).

31

RESULTS
During the study (22 July, 2010 – 24 September, 2012) there were a total of 17 sampled
rain events. During this data collection period, daily precipitation totals ranged from 0.25 mm to
58.32 mm for the individual rain events (Figure 10). Daily low temperatures ranged from 18.2°C to 23.3°C and daily high temperatures ranged from -12.2°C to 35.3°C (Figure 11).
Through analysis of the collected data, there are several statistical similarities and differences
between roof treatment types with respect to each individually tested water quality parameter.
Results of the One-Way ANOVA model are shown in Table 1. The Tukey’s HSD Model
of Multiple Comparisons showed significant similarities between different roof treatments, both
green and conventional, when looking at the individually tested water quality parameters (Tables
2-7). Continuing to divide the data, the Homogenous Subsets further illustrate the similarities of
the different roof treatments with respect to the individually monitored water quality
characteristics (Tables 8-13).
Nutrients
With respect to the observed average Nitrate-Nitrogen (NO3-N) levels, concentrations in
stormwater runoff ranged from 0.07-.93 mg/L for green roof with Sedum, 0.15-1.20 mg/L for
green roof with grasses, 0.43-5.10 mg/L for green roof with media only, 0.10-.97 mg/L for
conventional roof with metal, 0 .02-.53 mg/L for conventional roof with shingle, and 0.20-.80
mg/L for rainwater (Figure 12). Media only roof (no vegetation cover) produced the highest
concentration of NO3-N on 2 August, 2012 at 5.10 mg/L while the conventional shingle roof
produced the lowest concentration of NO3-N on 14 August, 2010 at .02 mg/L.
Observed average Nitrite-Nitrogen (NO2-N) levels for concentrations in stormwater
runoff ranged from 0.003-0.148 mg/L for green roof with Sedum, 0.002-0.148 mg/L for green

32

roof with grasses, 0.003-0.149 mg/L for green roof with media only, 0.003-0.037 mg/L for
conventional roof with metal, 0.003-0.036 mg/L for conventional roof with shingle, 0.001-0.017
mg/L and 0.003-0.036 mg/L for rainwater (Figure 13). Media only roofs produced the highest
concentration of NO2-N on 2 August, 2012 at 0.149 mg/L while green roofs with grasses
produced the lowest concentration of NO2-N on 24 September, 2012 at 0.002 mg/L.
With respect to the observed average Ammonia-Nitrogen (NH3-N) levels, concentrations
in stormwater runoff ranged from .01-2.00 mg/L for green roof with Sedum, .01-1.67 mg/L for
green roof with grasses, .00-.70 mg/L for green roof with media only, .02-1.16 mg/L for
conventional roof with metal, .00-1.00 mg/L for conventional roof with shingle, and .03-1.00
mg/L for rainwater (Figure 14). Green roofs with Sedum produced the highest concentration of
NH3-N on 3 August, 2011 at 2.00 mg/L while green roofs with media only and conventional
roofs with shingle both produced the lowest concentration of NH3-N on 24 September, 2012 and
20 October, 2011, respectively, at .00 mg/L.
Looking at the observed average Phosphate-Phosphorous (PO4-P) levels, concentrations
in stormwater runoff ranged from 0.11-5.39 mg/L for green roof with Sedum, 0.12-3.31 mg/L for
green roof with grasses, 0.12-3.73 mg/L for green roof with media only, 0.01-1.71 mg/L for
conventional roof with metal, 0.01-1.66 mg/L for conventional roof with shingle, and 0.09-2.10
mg/L for rainwater (Figure 15). Green roofs with Sedum produced the highest concentration of
PO4-P on 3 August, 2011 at 5.39 mg/L while conventional roof with metal and conventional roof
with shingle both produced the lowest concentration of PO4-P together on 2 July, 2010, August
14, 2010, and 19 August, 2010 at 0.01 mg/L.
When comparing the different green roof treatments, it was found that there is a
significant (p≥0.05) relationship between green roof vegetative treatments and concentration of

33

the nutrients Nitrate-Nitrogen (NO3-N), Nitrite-Nitrogen (NO2-N), Ammonia-Nitrogen (NH3-N),
and Phosphate-Phosphorous (PO4-P) in stormwater runoff with the native grass roofing treatment
releasing runoff with lower concentrations of these nutrients than the Sedum roofing treatment
(Hypothesis 1a: Accepted). This relationship was found due to the fact that Sedum planted green
roofs produced runoff that contained significantly higher levels of Phosphate-Phosphorous (PO4P) than green roofs planted with native grass. There were no significant differences between
Sedum and grass covered green roofs with respect to NO3-N, NO2-N and NH3-N.
In addition, it was found that here is a significant (p≥0.05) relationship between green
roof vegetative treatments and concentration of the nutrients Nitrate-Nitrogen (NO3-N), NitriteNitrogen (NO2-N), Ammonia-Nitrogen (NH3-N), and Phosphate-Phosphorous (PO4-P) in
stormwater runoff with the native grass roofing treatment releasing runoff with lower
concentrations of these nutrients than the media-only roofing treatment (Hypothesis 1b:
Accepted). This relationship was found due to the fact that the media only green roofs produced
runoff that contained significantly higher levels of Nitrate-Nitrogen (NO3-N) than green roofs
planted with native grass. There were no significant differences between media only and grass
covered green roofs with respect to NO2-N, NH3-N and PO4-P.
Furthermore, it was found that there is a significant (p≥0.05) relationship between green
roof vegetative treatments and concentration of the nutrients Nitrate-Nitrogen (NO3-N), NitriteNitrogen (NO2-N), Ammonia-Nitrogen (NH3-N), and Phosphate-Phosphorous (PO4-P) in
stormwater runoff with the Sedum roofing treatment releasing runoff with lower concentrations
of these nutrients than the media-only roofing treatment (Hypothesis 1c: Accepted). This
relationship was found due to the fact that the media only green roofs produced runoff that
contained significantly higher levels of Nitrate-Nitrogen (NO3-N) than Sedum planted green

34

roofs. In addition to the previously stated, it was shown that Sedum planted green roofs
produced higher levels of Phosphate-Phosphorous (PO4-P) than the green roofs with media only.
There were no significant differences between Sedum and media only green roofs with respect to
NO2-N and NH3-N.
When comparing green roof treatments to traditional roofing treatments, it was found that
there is a significant (p≥0.05) relationship between green roof and traditional (steel and shingle)
roofing treatments with respect to concentration of the nutrients Nitrate-Nitrogen (NO3-N),
Nitrite-Nitrogen (NO2-N), Ammonia-Nitrogen (NH3-N), and Phosphate-Phosphorous (PO4-P) in
stormwater runoff with all green roof treatments releasing stormwater runoff with higher
concentrations of these nutrients than both traditional steel and shingle roofing treatments
(Hypothesis 2a: Accepted).
pH.
Average pH levels in stormwater runoff ranged from 6.53-8.94 pH for green roof with
Sedum, 6.66-9.71 pH for green roof with grasses, 6.57-9.38 pH for green roof with media only,
5.55-7.05 pH conventional roof with metal, 6.06-6.75 pH for conventional roof with shingle, and
6.16-6.93 pH for rainwater (Figure 17). Green roofs with grasses produced the highest pH levels
on 14 August, 2010 at 9.7 pH while conventional roof with metal produced the lowest pH levels
on 19 August, 2010 at 5.55 pH.
When comparing green roof treatments to traditional roofing treatments, it was found that
there is a significant (p≥0.05) relationship between green roof and traditional roofing treatments
with respect to pH levels in stormwater runoff with all green roof treatments releasing
stormwater runoff with higher pH levels than both traditional steel and shingle roofing
treatments (Hypothesis 3a: Accepted).

35

Turbidity
With respect to the observed average Turbidity, levels in stormwater runoff ranged from
0.507-4.085 NTU for green roof with Sedum, 0.626-7.484 NTU for green roof with grasses,
0.603-5.093 NTU for green roof with media only, 0.112-3.063 NTU for conventional roof with
metal, 0.096-4.227 NTU for conventional roof with shingle, and 0.070-2.730 NTU for rainwater
(Figure 16). Green roofs with grasses produced the highest levels of turbidity on 3 August, 2011
at 7.484 NTU while conventional roof with shingle produced the lowest levels of turbidity on 25
September, 2011 at 0.096 NTU.
When comparing green roof treatments to traditional roofing treatments, it was found that
there is a significant (p≥0.05) relationship between green roof and traditional roofing treatments
with respect to turbidity (NTU) levels in stormwater runoff with all green roof treatments
releasing stormwater runoff with higher turbidity levels than both traditional steel and shingle
roofing treatments (Hypothesis 4a: Accepted)

36

DISCUSSION
Significant relationships between green roof treatments themselves, as well as
differences between green roof and conventional roofing treatments allow for comparison of
different roofing treatments and the subsequent outcomes of stormwater runoff quality released
from these systems. It was originally assumed that the green roofs with native grass species
would absorb the most, and release the least amount of nutrients in observed stormwater runoff.
This has proven to not always be the case. It was also assumed that green roofs, of all vegetative
types, would produce stormwater runoff with significantly different levels of turbidity and pH
when compared to traditional roofing materials. This was proven to be true with green roofs, or
all roofing treatments producing stormwater runoff with higher levels of pH and turbidity than
the tested traditional roofing treatments.
There was no significant difference in pH concentrations between green roof
treatments, but the study did show that green roof planted with media only did produce
significantly higher levels of turbidity then green roof treatment of Sedum and Grasses.
The strongest and most consistent outcome when comparing green roof planting
treatments with respect to nutrient levels in stormwater runoff was the fact that green roofs with
media only (no vegetation) released more NO3-N than grass and Sedum planted roofs for this
specified green roof system. This leads to believe that green roofs with vegetation, whether
grass or Sedum planted, are absorbing more NO3-N due to the vegetation uptake of this major
nitrogen component in observed green roof systems. Berghage in USEPA (2009b) suggested
similar relationships from observed green roof installments and speculated that new roofs
without full plant cover may contribute to nitrate runoff as opposed to fully established green
roofs, which may not. NO3-N is probably the most important nitrogen compound in natural

37

systems as it makes up the majority of TKN (total Kjeldahl nitrogen). TKN defines the total
nitrogen in a system by combining the components of nitrate-nitrogen, nitrite-nitrogen,
ammonia-nitrogen, and organic nitrogen. When looking at the vegetated green roofs, Sedum
planted roofs released higher amounts of PO4-P than green roofs planted with native grasses.
This would lead one to believe that native grass green roofs uptake more phosphorous-phosphate
than Sedum planted green roofs. This could guide designers to use native grasses as a means of
controlling PO4-P runoff from green roofs. Since phosphorous is usually the limiting nutrient in
natural systems, this finding is more important than relations regarding to nitrogen
concentrations when focusing on nutrient availability and aquatic plant growth.
It was initially thought that between the green roof treatments (i.e. Sedum versus
Grasses), roofs planted with grasses would uptake more, and release fewer nutrients in
stormwater runoff. The reasoning for this was that the grasses have more above ground biomass
that would require more nutrients to support growth. Also, grasses have a more extensive and
dense rooting system that would allow for the uptake of larger amounts of available nutrients.
This fact was not distinguishable between the grass and Sedum planted roofs as there was not
enough significance to say that the grasses performed better at absorbing nutrients than the
Sedum species with respect to all observed nutrients. The vegetated green roofs, though,
performed better than media only green roofs most consistently with respect to nitrate-nitrogen.
This leads one to believe that the plants on green roofs are taking up more available nitratenitrogen in the system. No conclusion could be made between vegetated and non-vegetated
roofs with respect to NO2-N, NH3-N and PO4-P. If this is correct, a case can be made to
maximize the health and density of vegetation on a green roof as it would uptake more and
release fewer nutrients than roofs with large areas of bare soil. This fact could influence

38

maintenance and watering regimes in attempts to achieve the greatest green roof plant coverage
as possible.
Though this study did not address the issue, decomposition of above ground plant
material and soil organics can (and may have) contribute to overall available nutrients, and pH in
a system and could potentially increase the nutrient levels, pH, and turbidity in stormwater runoff
of green roofs with vegetation. Also, seeing as the plants were still early in their establishment,
total vegetation coverage and root mass also could have affected the amount of nutrients uptake
of these different green roof systems as observed in this study. Ideally, studies should continue,
or be re-administered as the experimental roofs develop over a 10-20 year period.
Until now, no green roof studies have been performed to quantify and categorize the
results of green roof stormwater runoff quality with these exact system parameters. Previous
research dealing with green roof runoff water quality has been difficult to compare to this study
because of the comparative inconsistencies in the materials and methods of the tested green roof
systems (i.e. individually observed water quality characteristics, media depth and composition,
plant coverage, root mass ratio to above ground biomass, roof age, geographic location,
maintenance and fertilization practices, etc.). A significant influence on stormwater runoff
quality can be brought about by these characteristics and individual studies cannot be compared
unless all independent variables are consistent when comparing green roofs with different
vegetation types.
Testable hypothesis H2a was accepted due to the fact that all green roof treatments
(grass, Sedum, and media only) were found to release more NO3-N, NO2-N, NH3-N, and PO4-P
than both the steel and shingle roof treatments. This corresponds to the findings of Berghage in
USEPA (2009b) and is generally contributed to the stormwater interaction with the vegetative

39

materials and growing media organics. Berndtsson et al. (2006), when researching existing
green roofs of different sizes, component design, vegetation treatments and age, found multiple
studied green roofs to be a “sink” of NO3-N in which stormwater runoff from these roofs
contained less Nitrate-Nitrogen than collected rainwater. Alternatively, he did find green roofs
to be a source of PO4-P from all observed green roofs. Gregorie and Clausen (2011) found
similar results from an existing 248 m² green roof consisting of extensive modules planted with
Sedum spp. when compared to a conventional bituminous tar roof with the green roofs proving to
be a sink for NO3-N and a source of PO4-P. Opposite of these findings, Toland et al. (2012) was
not able to distinguish a significant difference between conventional and green roofs with respect
to NO3-N, NO2-N and NH4-N in stormwater runoff.
The highest Phosphate-Phosphorous (PO4-P ) concentration of all observed green roof
treatments came on August 3, 2011 from Sedum planted green roofs at a level of 5.39 mg/L in
runoff. This is well above the Michigan Department of Environmental Quality standard for point
source discharges of 1.0 mg/L (MDEQ, 2012), but can be attributed to the fact that this study
collected mainly the “first flush” of green roof runoff. This first flush is the highest in nutrient
concentration but is then combined with the remainder of cleaner, less polluted green roof runoff,
which would dilute the water and reduce the overall concentrations of pollutants at that scale
(Berndtsson et al., 2008). More research can be done to determine the effect of mass loading of
pollutants on a larger, more applicable scale.
The highest Nitrate-Nitrogen (NO3-N) concentration of all observed green roof
treatments came on August 2, 2012 from media only green roofs at a level of 5.10 mg/L. This is
well below the guidelines for drinking water standards set by the U.S. EPA at 10.0 mg/L
(USEPA, 2012a), but is above other Great Lake Region state standards like Minnesota who

40

limits NO3-N concentrations to 0.012 mg/L for lakes with trout and 0.09 mg/L for shallow lakes
and reservoirs in Minnesota. (USEPA, 2012b). Again, this could be contributed to the fact that
this study collected mostly first flush runoff of these smaller-scaled, representative green roof
installations. These reductions of Nitrate-Nitrogen and Phosphate-Phosphorous over time were
also observed in Berndtsson (2010), Berndtsson et al. (2006), Hathaway et al. (2008) and Van
Seters et al. (2009).
Testable hypothesis H3a was accepted due to the fact that all green roof treatments were
found to release stormwater runoff with a higher pH level than both conventional steel and
shingle roofs. This corresponds to findings of Berndtsson at al. (2009), Retzlaff et al. (2008),
and Berghage in USEPA (2009b) that also found green roofs releasing stormwater with higher
pH levels. Higher pH levels indicate less acidic, and more alkaline, water properties. This is
generally good for the environment as it is a combatant to the dwindling occurrences of “acid”
rains found downwind of industrialized as well as urbanized areas.
Testable hypothesis H4a was accepted due to the fact that all green roof treatments
produced stormwater runoff with higher turbidity levels than both conventional steel and shingle
roofs. This contradicts findings from Bliss (2009) in which green roofs were found to produce
less turbid runoff than runoff from a conventional ballasted membrane roof. However, green
roofs still produced higher turbidity levels than what was found in collected rainwater. Due to
the presence of granular growing media and the media’s proximity to the roof’s drainage outlet,
solids are allowed to release from the system at different amounts depending on the system’s
filtration system.
Though the influences of roof age on runoff water quality of green roofs was not a focus
of this study, nor were statistical comparisons made to compare these two variables, it is clear

41

that the age of the green roof affected both turbidity as well as pH levels in green roof runoff.
Throughout the 3 year sampling period, it was observed that pH and turbidity levels of green
roof treatments leveled off and became more consistent with runoff from the conventional
roofing treatments. Berghage in USEPA (2009b) found similar results and stated that over time,
a green roofs pH buffering capacity (mainly a product of the growing media make-up) would
decline over time, much like older European roofs, and acidification of stormwater runoff would
occur. In this study, the decline in observed turbidity in green roof runoff was attributed to the
establishments, construction methods, maintenance and filtering components of these roofs.
Essentially, loose media that was present from initial construction was believed to be washed out
over the first season and turbidity levels in stormwater runoff continued to decline over the
course of this study. A more complex and thorough runoff filtration system could be used to
reduce the initial levels of turbidity in stormwater runoff of the observed green roofs and lower
levels to what was typically seen from conventional roof runoff.
Though a roof slope of 33.33% was a consistent factor across all green roof and
traditional roofing treatments, it is believed that roof slope has minimal effect on the
concentrations of observed runoff quality characteristics in stormwater runoff. Additional
studies with roof slope being the only variable could be performed to determine if it truly has an
effect on water quality of stormwater runoff from green and conventional roofing treatments.

42

CONCLUSION
In conclusion, this study focused on steep sloped green, media and conventional roofing
types under steep slope (33.33%) conditions, though roof slope was thought to have no effect on
the outcomes of this study. It focused on runoff water quality characteristics among these five
roofing materials. In regards, the data corroborated earlier studies on the quality of runoff water
emanating from different roof treatments. For example, some green roof vegetation treatments
performed better than others with respect to individual water quality characteristics. It also
solidifies findings of past research that green roofs can be a less than 100% successful treatment
with respect to the quality of any water released from these systems. Given the study design,
more research could be done with fully established large scale green roofs to provide quantifiable
information that could be applied on city or region wide models to predict the impact of green
roofs on stormwater runoff quality in an area.
Green roof component design, including media composition, construction materials and
methods could also be tailored to increase the water quality of green roof runoff. Green roof
growing media amendments, such as biochar with a high exchange capacity, can increase the
retention of nutrients in green roof systems, releasing less in stormwater runoff and making the
nutrients more available for plant uptake (Beck et al., 2012)
When looking at building and site systems as a whole, green roofs can be a legitimate
mean of stormwater quantity management as shown by Cronk (2012) but may require further onsite treatment of runoff to address increased nutrient, pH and turbidity levels. Additional on-site
measures can be taken to address these water quality issues in the form of recycling green roof
stormwater runoff to a rooftop before costly in ground bioretention and infiltration systems are
used. Addressing this issue on-site, before any site runoff can occur, will combat one of the few

43

negative aspects of green roofs and continue to hold green roofs as a viable tool in stormwater
management.

44

APPENDIX

45

Table 1: One-way ANOVA table for respective water quality characteristics in stormwater
runoff from five roof platform treatments replicated three times. Study was completed
over 3 growing seasons (22 July, 2010 – 24 September, 2012). Levels of pollutants in
stormwater runoff are the dependent variables. Roof treatments are the independent
variables.
One-Way ANOVA for Study Sample

Mean
NO3-N
(mg/L)

Mean
NO2-N
(mg/L)

Mean
NH3-N
(mg/L)

Mean
PO4-P
(mg/L)

Mean
NTU

Mean pH

Type of
Variance
Between Groups
(All Treatments)
Within Groups
(Individual
Treatments)
Total
Between Groups
(All Treatments)
Within Groups
(Individual
Treatments)
Total
Between Groups
(All Treatments)
Within Groups
(Individual
Treatments)
Total
Between Groups
(All Treatments)
Within Groups
(Individual
Treatments)
Total
Between Groups
(All Treatments)
Within Groups
(Individual
Treatments)
Total
Between Groups
(All Treatments)
Within Groups
(Individual
Treatments)
Total

Sum of
Squares

Degrees of
Freedom

Mean
Square

FStatistic

PStatistic

5.877

4

1.469

38.529

0.000

0.381

10

0.038

6.258

14

0.002

4

0.000

10.238

0.001

0.000

10

0.000

0.002

14

0.074

4

0.019

8.689

0.003

0.021

10

0.002

0.095

14

6.921

4

1.730

342.491

0.000

0.051

10

0.005

6.972

14

7.606

4

1.902

71.519

0.000

0.266

10

0.027

7.872

14

5.606

4

1.402

133.514

0.000

0.105

10

0.010

5.711

14

46

Table 2: Tukey’s HSD test for multiple comparisons of Nitrate-Nitrogen (NO3-N) for all roof
treatments during the study’s 3 growing seasons (22 July, 2010 – 24 September,
2012).
Multiple Comparisons - Tukey's HSD
Dependent
Variable
Mean
NO3-N

Mean
Difference
(I-J)

95% Upper
Bound

Treatment
(J)

G

S

0.18941

0.15944

0.758

-0.3353

0.7141

M

-1.3449*

0.15944

0

-1.8696

-0.8202

0.28

0.15944

0.446

-0.2447

0.8047

CS

0.31431

0.15944

0.344

-0.2104

0.839

G

-0.18941

0.15944

0.758

-0.7141

0.3353

M

-1.53431*

0.15944

0

-2.059

-1.0096

0.09059

0.15944

0.977

-0.4341

0.6153

CS

0.1249

0.15944

0.93

-0.3998

0.6496

G

1.3449*

0.15944

0

0.8202

1.8696

S

1.53431*

0.15944

0

1.0096

2.059

CM

1.6249*

0.15944

0

1.1002

2.1496

CS

1.65922*

0.15944

0

1.1345

2.1839

G

-0.28

0.15944

0.446

-0.8047

0.2447

S

-0.09059

0.15944

0.977

-0.6153

0.4341

M

-1.6249*

0.15944

0

-2.1496

-1.1002

CS

0.03431

0.15944

0.999

0.4904

0.559

G

-0.31431

0.15944

0.344

-0.839

0.2104

CM
S

CM
M

CM

CS

S
M
CM

Standard
Error

95% Lower
Bound

Treatment
(I)

Significance

-0.1249

0.15944

0.93

-0.6496

0.3998

-1.65922*

0.15944

0

-2.1839

-1.1345

-0.03431

0.15944

0.999

-0.559

0.4904

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum Species), M (Media Only Roof
Treatment), CM (Conventional Metal Roofing) & CS (Conventional Shingle Roofing).
All roof treatments were replicated three times.
Mean differences denoted with “*” are significant at a level ≥ 0.05

47

Table 3: Tukey’s HSD test for multiple comparisons of Nitrite-Nitrogen (NO2-N) for all roof
treatments during the study’s 3 growing seasons (22 July, 2010 – 24 September,
2012).
Multiple Comparisons - Tukey's HSD
Dependent
Variable
Mean
NO2-N

Mean
Difference
(I-J)

95% Upper
Bound

Treatment
(J)

G

S

0.00688

0.00566

-0.0118

-0.0118

0.0255

M

-0.00729

0.00566

-0.0259

-0.0259

0.0113

CM

0.01808

0.00566

-0.0006

-0.0006

0.0367

CS

0.02402*

0.00566

0.0054

0.0054

0.0427

G

-0.00688

0.00566

0.0255

-0.0255

0.0118

M

-0.01418

0.00566

-0.0328

-0.0328

0.0045

CM

0.0112

0.00566

-0.0074

-0.0074

0.0298

CS

0.01714

0.00566

-0.0015

-0.0015

0.0358

G

0.00729

0.00566

-0.0113

-0.0113

0.0259

S

0.01418

0.00566

-0.0045

-0.0045

0.0328

CM

0.02537*

0.00566

0.0067

0.0067

0.044

CS

0.03131*

0.00566

0.0127

0.0127

0.05

G

-0.01808

0.00566

-0.0367

-0.0367

0.0006

S

-0.0112

0.00566

-0.0298

-0.0298

0.0074

-0.02537*

0.00566

-0.044

-0.044

-0.0067

0.00594

0.00566

-0.0127

-0.0127

0.0246

G

-0.02402*

0.00566

0.012

-0.0427

-0.0054

S

-0.01714

0.00566

0.075

-0.0358

0.0015

M

-0.03131*

0.00566

0.002

-0.05

-0.0127

S

M

CM

M
CS
CS

Standard
Error

95% Lower
Bound

Treatment
(I)

Significance

CM
-0.00594
0.00566
0.827
-0.0246
0.0127
Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum Species), M (Media Only Roof
Treatment), CM (Conventional Metal Roofing) & CS (Conventional Shingle Roofing).
All roof treatments were replicated three times.
Mean differences denoted with “*” are significant at a level ≥ 0.05

48

Table 4: Tukey’s HSD test for multiple comparisons of Ammonia-Nitrogen (NH3-N) for all
roof treatments during the study’s 3 growing seasons (22 July, 2010 – 24 September,
2012).
Multiple Comparisons - Tukey's HSD
Dependent
Variable
Mean
NH3-N

Mean
Difference
(I-J)

95% Upper
Bound

Treatment
(J)

G

S

0.00863

0.03768

0.999

-0.1154

0.1327

M

0.09216

0.03768

0.18

-0.0319

0.2162

CM

0.10251

0.03768

0.12

-0.0215

0.2265

CS

0.19196*

0.03768

0.003

0.0679

0.316

G

-0.00863

0.03768

0.999

-0.1327

0.1154

M

0.08353

0.03768

0.249

-0.0405

0.0276

CM

0.09388

0.03768

0.169

-0.0301

0.2179

CS

0.18333*

0.03768

0.005

0.0593

0.3074

G

-0.09216

0.03768

0.18

-0.2162

0.0319

S

-0.08353

0.03768

0.249

-0.2076

0.0405

CM

0.01035

0.03768

0.999

-0.1137

0.1344

CS

0.0998

0.03768

0.134

-0.0242

0.2238

G

-0.10251

0.03768

0.12

-0.2265

0.0215

S

-0.09388

0.03768

0.169

-0.2179

0.0301

M

-0.01035

0.03768

0.999

-0.1344

0.1137

CS

0.08945

0.03768

0.2

-0.0346

0.2135

G

-0.19196*

0.03768

0.003

-0.316

-0.0679

S

-0.18333*

0.03768

0.005

-0.3074

-0.0593

-0.0998

0.03768

0.134

-0.2238

0.0242

S

M

CM

CS

M

Standard
Error

95% Lower
Bound

Treatment
(I)

Significance

CM
-0.08945
0.03768
0.2
-0.2135
0.0346
Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum Species), M (Media Only Roof
Treatment), CM (Conventional Metal Roofing) & CS (Conventional Shingle Roofing).
All roof treatments were replicated three times.
Mean differences denoted with “*” are significant at a level ≥ 0.05

49

Table 5: Tukey’s HSD test for multiple comparisons of Phosphate-Phosphorous (PO4-P) for all
roof treatments during the study’s 3 growing seasons (22 July, 2010 – 24 September,
2012).
Multiple Comparisons - Tukey's HSD
Dependent
Variable
Mean
PO4-P

Treatment
(I)

Treatment
(J)

G

S

M

CM

Significance

95% Lower
Bound

95% Upper
Bound

0.05804

0

-0.642

-0.26

-0.0902

0.05804

0.554

-0.2851

0.1008

CM

1.16196*

0.05804

0

0.971

1.353

CS

1.18098*

0.05804

0

0.99

1.372

G

0.45098*

0.05804

0

0.26

0.642

M

0.36078*

0.05804

0.001

0.1698

0.5518

CM

1.61294*

0.05804

0

1.4219

1.8039

CS

1.63196*

0.05804

0

1.441

1.823

G

0.0902

0.05804

0.554

-0.1008

0.2812

S

-0.36078*

0.05804

0.001

-0.5518

-0.1698

CM

1.25216*

0.05804

0

1.0612

1.4432

CS

1.27118*

0.05804

0

1.0802

1.4622

G

-1.16196*

0.05804

0

-1.353

-9710

S

-1.61294*

0.05804

0

-1.8039

-1.4219

M

-1.25216*

0.05804

0

-1.4432

-1.0612

0.01902

0.05804

0.997

-0.172

0.21

G

-1.18098*

0.05804

0

-1.372

-0.99

S

-1.63196*

0.05804

0

-1.823

-1.441

M

-1.27118*

0.05804

0

-1.4622

-1.0802

CS
CS

Standard
Error

-0.45098*

M

S

Mean
Difference (IJ)

CM
-0.01902
0.05804
0.997
-0.21
0.172
Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum Species), M (Media Only Roof
Treatment), CM (Conventional Metal Roofing) & CS (Conventional Shingle Roofing).
All roof treatments were replicated three times.
Mean differences denoted with “*” are significant at a level ≥ 0.05

50

Table 6: Tukey’s HSD test for multiple comparisons of Turbidity (NTU) for all roof treatments
during the study’s 3 growing seasons (22 July, 2010 – 24 September, 2012).
Multiple Comparisons - Tukey's HSD
Dependent
Variable
Mean
NTU

Mean
Difference
(I-J)

95% Upper
Bound

Treatment
(J)

G

S

0.09747

0.13314

0.944

-0.3407

0.5356

M

0.63599*

0.13314

0.005

0.1978

1.0742

CM

1.79093*

0.13314

0

1.3528

2.2291

CS

1.42728*

0.13314

0

0.9891

1.8654

G

-0.09747

0.13314

0.944

-0.5356

0.3407

M

0.53852*

0.13314

0.016

0.1003

0.9767

CM

1.69346*

0.13314

0

1.2553

2.1316

CS

1.32981*

0.13314

0

0.8916

1.768

G

-0.63599*

0.13314

0.005

-1.0742

-0.1978

S

-0.53852*

0.13314

0.016

-0.9767

-0.1003

CM

1.15495*

0.13314

0

0.7168

1.5931

CS

0.7913*

0.13314

0.001

0.3531

1.2295

G

-1.79093*

0.13314

0

-2.2291

-1.3528

S

-1.69346*

0.13314

0

-2.1316

-1.2553

M

-1.15495*

0.13314

0

-1.5931

-0.7168

CS

-0.36365

0.13314

0.118

-0.8018

0.0745

G

-1.42728*

0.13314

0

-1.8654

-0.9891

S

-1.32981*

0.13314

0

-1.768

-0.8916

M

-0.7913*

0.13314

0.001

-1.2295

-0.3531

S

M

CM

CS

Standard
Error

95% Lower
Bound

Treatment
(I)

Significance

CM
0.36365
0.13314
0.118
-0.0745
0.8081
Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum Species), M (Media Only Roof
Treatment), CM (Conventional Metal Roofing) & CS (Conventional Shingle Roofing).
All roof treatments were replicated three times.
Mean differences denoted with “*” are significant at a level ≥ 0.05

51

Table 7: Tukey’s HSD test for multiple comparisons of pH for all roof treatments during the
study’s 3 growing seasons (22 July, 2010 – 24 September, 2012).
Multiple Comparisons - Tukey's HSD
Dependent
Variable
Mean
pH

Mean
Difference
(I-J)

95% Upper
Bound

Treatment
(J)

G

S

0.18667

0.08366

0.244

-0.0886

0.462

M

0.11353

0.08366

0.665

-0.1618

0.3888

CM

1.30118*

0.08366

0

1.0259

1.5765

CS

1.38078*

0.08366

0

1.1055

1.6561

G

-0.18667

0.08366

0.244

-0.462

0.0886

M

-0.07314

0.08366

0.9

-0.3485

0.2022

CM

1.11451*

0.08366

0

0.8392

1.3898

CS

1.19412*

0.08366

0

0.9188

1.4694

G

-0.11353

0.08366

0.665

-0.3888

0.1618

S

0.07314

0.08366

0.9

-0.2022

0.3485

CM

1.18765*

0.08366

0

0.9123

1.463

CS

1.26725*

0.08366

0

0.9919

1.5426

G

-1.30118*

0.08366

0

-1.5765

-1.0259

S

-1.11451*

0.08366

0

-1.3898

-0.8392

M

-1.18765*

0.08366

0

-1.463

-0.9123

S

M

CM

CS
CS

Standard
Error

95% Lower
Bound

Treatment
(I)

Significance

0.07961

0.08366

0.87

-0.1957

0.3549

G

-1.38078*

0.08366

0

-1.6561

-1.1055

S

-1.19412*

0.08366

0

-1.4694

-0.9188

M

-1.26725*

0.08366

0

-1.5426

-0.9919

CM
-0.07961
0.08366
0.87
-0.3549
0.1957
Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum Species), M (Media Only Roof
Treatment), CM (Conventional Metal Roofing) & CS (Conventional Shingle Roofing).
All roof treatments were replicated three times.
Mean differences denoted with “*” are significant at a level ≥ 0.05

52

Table 8: Homogenous subset of recorded NO3-N data for all roof treatments during the study’s
3 growing seasons (22 July, 2010 – 24 September, 2012).
Mean NO3-N

Treatment

N
3
3
3
3
3

CS
CM
S
G
M
Significance

Subsets for
Alpla = 0.05
1
2
0.2682
0.3025
0.3931
0.5825
1.9275
0.344
1.000

Table 9: Homogenous subset of recorded NO2-N data for all roof treatments during the study’s
3 growing seasons (22 July, 2010 – 24 September, 2012).
Mean NO2-N
Treatment
CS
CM
S
G
M
Significance

N
3
3
3
3
3

Subsets for Alpla = 0.05
1
2
3
0.0108
0.0167
0.0167
0.0279
0.0279
0.0279
0.0348
0.0348
0.0421
0.075
0.058
0.166

Table 10: Homogenous subset of recorded NH3-N data for all roof treatments during the study’s
3 growing seasons (22 July, 2010 – 24 September, 2012).
Mean NH3-N

Treatment
CS
CM
M
S
G
Significance

N
3
3
3
3
3

53

Subsets for
Alpla = 0.05
1
2
0.2294
0.3189
0.3189
0.3292
0.3292
0.4127
0.4214
0.134
0.120

Table 11: Homogenous subset of recorded PO4-P data for all roof treatments during the study’s 3
growing seasons (22 July, 2010 – 24 September, 2012)..
Mean PO4-P
Treatment
CS
CM
G
M
S
Significance

N
3
3
3
3
3

Subsets for Alpla = 0.05
1
2
3
0.3043
0.3233
1.4853
1.5755
1.9363
0.997
0.554
1.000

Table 12: Homogenous subset of recorded NTU data for all roof treatments during the study’s 3
growing seasons (22 July, 2010 – 24 September, 2012).
Mean NTU
Treatment
CM
CS
M
S
G
Significance

N
3
3
3
3
3

Subsets for Alpla = 0.05
1
2
3
0.9586
1.3222
2.1135
2.6520
2.7495
0.118
1.000
0.940

Table 13: Homogenous subset of recorded pH data for all roof treatments during the study’s 3
growing seasons (22 July, 2010 – 24 September, 2012).
Mean pH

Treatment
CS
CM
S
M
G
Significance

N
3
3
3
3
3

54

Subsets for
Alpla = 0.05
1
2
6.4196
6.4992
7.6137
7.6869
7.8004
0.870
0.244

Table 14: Data from rain event 22 July, 2010. All concentrations except pH and turbidity are in
milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: July 22, 2010
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

pH

G1

0.32

0.064

0.21

0.20

4.43

9.01

G2

0.23

0.077

0.31

0.23

3.38

8.66

G3

0.40

0.064

0.45

0.21

2.65

8.04

G Average

0.32

0.069

0.33

0.21

3.49

8.57

M1

2.06

0.058

0.69

0.20

3.49

8.83

M2

1.97

0.053

0.31

0.20

2.69

7.79

M3

2.11

0.078

0.67

0.24

3.28

7.33

M Average

2.05

0.063

0.56

0.22

3.15

7.98

S1

0.17

0.048

0.45

0.15

3.48

7.99

S2

0.10

0.082

1.17

0.18

5.19

8.69

S3

0.15

0.096

1.53

0.36

9.86

8.78

S Average

0.14

0.075

1.05

0.23

6.18

8.49

CM1

0.09

0.038

0.07

0.01

4.56

6.98

CM2

0.15

0.032

0.06

0.01

0.94

6.49

CM3

0.11

0.041

0.07

0.01

0.85

6.33

CM Average

0.11

0.037

0.07

0.01

2.12

6.60

CS1

0.21

0.039

0.12

0.01

2.32

6.14

CS2

0.18

0.037

0.12

0.01

2.05

6.14

CS3

0.09

0.037

0.05

0.01

1.49

6.36

CS Average

0.16

0.037

0.10

0.01

1.95

6.21

Treatment

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

55

Table 15: Data from rain event 14 August, 2010. All concentrations except pH and turbidity are
in milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: August 14, 2010
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

G1

0.20

0.046

0.11

0.09

5.41

10.28

G2

0.68

0.080

0.48

0.16

3.14

9.32

G3

0.64

0.062

0.12

0.13

3.83

9.53

G Average

0.50

0.063

0.24

0.13

4.13

9.71

M1

1.99

0.128

0.14

0.13

3.97

9.38

M2

1.98

0.097

0.09

0.09

4.72

9.73

M3

2.00

0.159

0.19

0.18

3.22

9.02

M Average

1.99

0.128

0.14

0.13

3.97

9.38

S1

0.18

0.057

0.20

0.17

4.09

9.15

S2

0.20

0.054

0.08

0.14

5.12

9.20

S3

0.15

0.061

0.33

0.19

3.05

9.09

S Average

0.18

0.057

0.20

0.17

4.09

9.15

CM1

0.02

0.023

0.03

0.01

1.07

7.30

CM2

0.27

0.011

0.07

0.01

1.73

6.44

CM3

0.10

0.012

0.09

0.01

1.85

6.75

CM Average

0.13

0.016

0.06

0.01

1.55

6.83

CS1

0.01

0.013

0.02

0.01

1.29

6.75

CS2

0.03

0.018

0.05

0.01

0.98

5.61

CS3

0.01

0.021

0.01

0.01

0.55

5.83

CS Average

0.02

0.018

0.03

0.01

0.94

6.06

Treatment

pH

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum Species),
M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

56

Table 16: Data from rain event 19 August, 2010. All concentrations except pH and turbidity are
in milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: August 19, 2010
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

pH

G1

0.06

0.037

0.14

0.12

5.32

9.30

G2

0.15

0.061

0.22

0.11

5.51

9.97

G3

0.32

0.037

0.20

0.13

4.56

9.52

G Average

0.18

0.045

0.19

0.12

5.13

9.60

M1

0.86

0.035

0.21

0.08

6.56

9.56

M2

0.57

0.034

0.20

0.11

5.10

8.80

M3

1.96

0.035

0.26

0.17

3.62

8.44

M Average

1.13

0.034

0.22

0.12

5.09

8.93

S1

0.06

0.025

0.26

0.08

5.91

8.21

S2

0.08

0.027

0.15

0.13

5.44

9.55

S3

0.06

0.030

0.29

0.13

5.29

9.05

S Average

0.07

0.027

0.23

0.11

5.55

8.94

CM1

0.09

0.016

0.21

0.01

3.10

5.72

CM2

0.15

0.010

0.35

0.01

3.07

5.32

CM3

0.15

0.010

0.39

0.01

1.94

5.61

CM Average

0.13

0.012

0.32

0.01

2.70

5.55

CS1

0.12

0.020

0.32

0.01

3.73

6.26

CS2

0.25

0.019

0.45

0.01

2.24

6.02

CS3

0.08

0.014

0.24

0.01

2.85

6.10

CS Average

0.15

0.018

0.34

0.01

2.94

6.13

Treatment

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

57

Table 17: Data from rain event 3 August, 2011. All concentrations except pH and turbidity are in
milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: August 3, 2011
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

pH

G1

0.9

0.023

2.00

2.66

5.87

8.92

G2

1.5

0.010

1.00

2.62

6.94

8.21

G3

1.2

0.037

1.00

3.20

9.64

8.34

G Average

1.2

0.023

1.33

2.83

7.48

8.49

M1

4.6

0.020

1.00

2.75

3.22

7.92

M2

4.2

0.044

1.00

2.57

3.74

7.83

M3

5.5

0.018

0.00

2.46

4.77

7.69

M Average

4.8

0.027

0.67

2.59

3.91

7.81

S1

0.7

0.011

2.00

5.17

5.81

8.01

S2

0.4

0.014

2.00

5.50

6.18

7.81

S3

0.1

0.003

2.00

5.50

5.20

7.87

S Average

0.4

0.009

2.00

5.39

5.73

7.90

CM1

0.6

0.005

0.62

0.19

0.22

6.65

CM2

0.2

0.005

0.41

0.20

0.37

6.69

CM3

0.4

0.010

0.55

0.51

0.71

6.40

CM Average

0.4

0.007

0.53

0.30

0.43

6.58

CS1

0.2

0.010

0.28

0.22

1.79

6.31

CS2

0.5

0.006

0.11

0.10

1.72

6.35

CS3

0.6

0.006

0.16

0.16

1.38

6.41

CS Average

0.4

0.007

0.18

0.16

1.63

6.36

Rainwater

0.3

0.005

0.16

0.16

0.63

6.61

Treatment

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

58

Table 18: Data from rain event 19 September, 2011. All concentrations except pH and turbidity
are in milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: September 19, 2011
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

G1

1.3

0.032

0.59

2.66

5.87

7.41

G2

1.0

0.025

0.46

1.72

5.55

7.61

G3

0.5

0.024

0.33

1.92

5.57

7.97

G Average

0.9

0.027

0.46

2.10

5.66

7.66

M1

1.7

0.006

0.52

2.75

3.46

7.53

M2

0.5

0.012

0.67

2.15

4.89

7.55

M3

2.3

0.017

0.27

2.46

4.19

7.93

M Average

1.5

0.012

0.49

2.45

4.18

7.67

S1

0.5

0.002

0.50

2.40

4.34

7.90

S2

0.3

0.012

0.45

2.27

5.89

7.60

S3

0.9

0.019

0.55

2.09

5.19

7.61

S Average

0.6

0.011

0.50

2.25

5.14

7.70

CM1

0.2

0.001

0.18

0.12

0.94

6.31

CM2

0.2

0.003

0.30

0.22

0.85

6.85

CM3

0.2

0.004

0.51

0.30

1.32

6.63

CM Average

0.2

0.003

0.33

0.21

1.03

6.60

CS1

0.2

0.005

0.26

0.22

2.44

6.48

CS2

0.3

0.004

0.28

0.17

2.26

6.55

CS3

0.3

0.002

0.16

0.06

1.32

6.57

0.27

0.004

0.23

0.15

2.01

6.53

0.2

0.004

0.17

0.15

1.88

6.52

Treatment

CS Average
Rainwater

pH

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

59

Table 19: Data from rain event 25 September, 2011. All concentrations except pH and turbidity
are in milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: September 25, 2011
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

pH

G1

0.4

0.022

0.42

1.12

3.63

7.55

G2

0.4

0.016

0.63

1.39

3.86

7.99

G3

0.3

0.025

0.36

1.40

3.99

7.64

G Average

0.4

0.021

0.47

1.30

3.82

7.73

M1

0.6

0.012

0.35

1.70

1.88

7.63

M2

0.5

0.017

0.59

1.05

1.58

7.47

M3

0.2

0.017

0.47

1.71

1.92

7.57

M Average

0.4

0.015

0.47

1.49

1.79

7.56

S1

0.2

0.014

0.31

1.67

2.07

7.62

S2

0.3

0.004

0.41

1.53

2.84

7.58

S3

0.4

0.010

0.49

1.73

2.56

7.64

S Average

0.3

0.009

0.40

1.64

2.49

7.61

CM1

0.0

0.006

0.22

0.12

0.08

6.97

CM2

0.1

0.003

0.10

0.09

0.12

6.74

CM3

0.2

0.001

0.15

0.13

0.14

6.61

CM Average

0.1

0.003

0.16

0.11

0.11

6.77

CS1

0.1

0.004

0.19

0.28

0.08

6.46

CS2

0.2

0.001

0.24

0.23

0.06

6.59

CS3

0.3

0.005

0.11

0.19

0.15

6.60

CS Average

0.2

0.003

0.18

0.23

0.10

6.55

Rainwater

0.2

0.006

0.23

0.23

0.07

6.50

Treatment

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

60

Table 20: Data from rain event 27 September, 2011. All concentrations except pH and turbidity
are in milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: September 27, 2011
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

pH

G1

0.3

0.015

0.27

1.31

3.48

7.69

G2

0.6

0.007

0.16

1.20

4.27

7.79

G3

0.2

0.021

0.20

1.23

3.97

7.83

G Average

0.4

0.014

0.21

1.25

3.91

7.77

M1

0.6

0.013

0.29

1.17

2.37

8.00

M2

0.8

0.011

0.21

1.29

2.14

7.37

M3

0.1

0.027

0.17

1.66

2.84

7.81

M Average

0.5

0.017

0.22

1.37

2.45

7.73

S1

0.2

0.020

0.25

1.68

2.51

7.79

S2

0.3

0.010

0.18

1.67

3.09

7.60

S3

0.7

0.014

0.27

1.72

2.59

7.54

S Average

0.4

0.015

0.23

1.69

2.73

7.64

CM1

0.2

0.003

0.07

0.14

0.08

6.29

CM2

0.1

0.004

0.15

0.06

0.21

6.39

CM3

0.2

0.009

0.20

0.13

0.13

6.53

CM Average

0.2

0.005

0.14

0.11

0.14

6.40

CS1

0.2

0.005

0.23

0.49

0.10

6.55

CS2

0.2

0.003

0.15

0.12

0.19

6.60

CS3

0.2

0.003

0.10

0.06

0.03

6.64

CS Average

0.2

0.004

0.16

0.22

0.11

6.60

Rainwater

0.2

0.001

0.17

0.17

0.13

6.59

Treatment

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

61

Table 21: Data from rain event 14 October, 2011. All concentrations except pH and turbidity are
in milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: October 14, 2011
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

pH

G1

0.6

0.009

0.10

1.48

3.43

7.96

G2

0.7

0.010

0.17

1.44

4.21

7.79

G3

0.5

0.015

0.05

1.22

4.01

7.83

G Average

0.6

0.011

0.11

1.38

3.88

7.86

M1

0.9

0.013

0.09

1.29

2.36

8.00

M2

0.3

0.008

0.17

1.00

2.14

9.37

M3

1.0

0.019

0.14

2.06

3.36

7.81

M Average

0.7

0.013

0.13

1.45

2.62

8.39

S1

0.6

0.020

0.09

1.45

2.78

7.79

S2

0.6

0.014

0.05

1.67

3.05

7.60

S3

0.9

0.010

0.15

1.72

2.35

7.54

S Average

0.7

0.015

0.10

1.61

2.73

7.64

CM1

0.4

0.006

0.09

0.14

0.09

6.29

CM2

0.4

0.006

0.07

0.06

0.07

6.39

CM3

0.4

0.007

0.06

0.13

0.23

6.53

CM Average

0.4

0.006

0.07

0.11

0.13

6.40

CS1

0.3

0.004

0.05

0.49

0.39

6.55

CS2

0.1

0.008

0.06

0.12

0.18

6.60

CS3

0.4

0.007

0.04

0.06

0.24

6.64

CS Average

0.3

0.006

0.05

0.22

0.27

6.60

Rainwater

0.4

0.007

0.05

0.09

0.11

6.67

Treatment

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

62

Table 22: Data from rain event 20 October, 2011. All concentrations except pH and turbidity are
in milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: October 20, 2011
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

pH

G1

0.6

0.005

0.06

1.38

2.47

7.49

G2

0.5

0.004

0.05

1.53

1.82

7.51

G3

0.4

0.010

0.04

1.43

2.42

7.95

G Average

0.5

0.006

0.05

1.45

2.24

7.65

M1

0.4

0.010

0.04

1.31

0.82

7.49

M2

0.3

0.011

0.04

1.07

1.37

7.50

M3

1.0

0.007

0.02

1.41

1.06

7.53

M Average

0.6

0.009

0.03

1.26

1.08

7.51

S1

0.2

0.009

0.03

2.49

1.33

7.53

S2

0.2

0.004

0.03

1.87

1.57

7.33

S3

0.2

0.005

0.04

1.79

1.84

7.10

S Average

0.2

0.006

0.03

2.05

1.58

7.32

CM1

0.2

0.002

0.01

0.22

0.77

7.83

CM2

0.3

0.003

0.01

0.13

0.81

7.29

CM3

0.1

0.004

0.00

0.28

0.74

7.05

CM Average

0.2

0.003

0.01

0.21

0.77

7.39

CS1

0.2

0.002

0.02

0.16

0.65

6.75

CS2

0.2

0.003

0.01

0.07

0.79

6.71

CS3

0.1

0.004

0.00

0.10

0.60

6.69

CS Average

0.2

0.003

0.01

0.11

0.68

6.72

Rainwater

0.2

0.004

0.03

0.24

0.37

6.39

Treatment

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

63

Table 23: Data from rain event 28 May, 2012. All concentrations except pH and turbidity are in
milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: May 28, 2012
NitrateNitrogen
(NO3-N)
0.200

NitriteNitrogen
(NO2-N)
0.006

AmmoniaNitrogen
(NH3-N)
0.170

PhosphatePhosphorous
(PO4-P)
3.380

Turbidity
(NTUs)
2.240

pH
6.55

S2

0.200

0.003

0.140

3.050

1.810

6.75

S3

0.000

0.003

0.080

4.100

3.570

6.55

S Average

0.133

0.004

0.130

3.510

2.540

6.62

G1

1.300

0.019

0.230

2.420

0.891

6.99

G2

0.700

0.063

0.120

3.850

1.200

7.04

G3

0.300

0.000

0.170

3.670

0.981

6.77

G Average

0.767

0.027

0.173

3.313

1.024

6.93

M1

2.000

0.011

0.200

4.120

0.945

7.06

M2

1.700

0.071

0.160

3.220

0.431

7.09

M3

3.800

0.012

0.160

3.850

0.434

7.00

M Average

2.500

0.031

0.173

3.730

0.603

7.05

CM1

0.400

0.004

0.600

1.440

3.110

6.26

CM2

0.500

0.004

0.730

2.260

2.690

6.32

CM3

0.500

0.010

0.710

1.420

3.390

6.44

CM Average

0.467

0.006

0.680

1.707

3.063

6.34

CS1

0.500

0.003

0.030

2.000

2.470

6.57

CS2

0.500

0.010

0.050

1.800

4.050

6.78

CS3

0.600

0.007

0.030

1.170

6.160

6.89

CS Average

0.533

0.007

0.037

1.657

4.227

6.75

Rain

0.800

0.006

0.030

2.100

2.730

6.86

Treatment
S1

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

64

Table 24: Data from rain event 2 June, 2012. All concentrations except pH and turbidity are in
milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: June 2, 2012
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

pH

S1

0.400

0.002

0.020

1.170

1.240

7.44

S2

0.600

0.002

0.020

1.640

0.707

8.76

S3

0.300

0.000

0.000

1.190

0.974

7.96

S Average

0.267

0.035

0.013

1.557

1.553

6.53

G1

0.400

0.002

0.020

1.170

1.240

7.44

G2

0.600

0.002

0.020

1.640

0.707

8.76

G3

0.300

0.000

0.000

1.190

0.974

7.96

G Average

0.433

0.001

0.013

1.333

0.974

8.05

M1

1.900

0.004

0.000

1.620

1.320

7.80

M2

1.500

0.002

0.000

2.300

0.448

8.22

M3

3.300

0.004

0.070

1.820

0.593

7.60

M Average

2.233

0.003

0.023

1.913

0.787

7.87

CM1

0.200

0.004

0.490

0.410

0.748

6.09

CM2

0.200

0.003

0.130

0.250

2.030

6.14

CM3

0.300

0.004

0.010

0.060

0.555

6.33

CM Average

0.233

0.004

0.210

0.240

1.111

6.19

CS1

0.400

0.003

0.020

0.220

1.160

6.34

CS2

0.200

0.003

0.440

0.670

1.460

6.23

CS3

0.400

0.005

0.290

0.400

2.440

6.30

CS Average

0.333

0.004

0.250

0.430

1.687

6.29

Rain

0.300

0.003

0.070

0.210

0.421

6.31

Treatment

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

65

Table 25: Data from rain event 19 June, 2012. All concentrations except pH and turbidity are in
milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: June 19, 2012
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

pH

S1

0.300

0.008

0.220

2.000

0.680

6.54

S2

0.300

0.008

0.070

2.020

1.210

6.81

S3

0.100

0.003

0.100

1.650

0.987

6.93

S Average

0.233

0.006

0.130

1.890

0.959

6.76

G1

0.400

0.019

0.240

1.810

0.734

6.94

G2

0.600

0.016

0.070

1.520

0.816

7.14

G3

0.300

0.006

0.260

1.320

0.447

7.04

G Average

0.433

0.014

0.190

1.550

0.666

7.04

M1

2.800

0.030

0.240

2.040

1.330

7.16

M2

1.300

0.009

0.040

1.530

0.595

7.17

M3

4.300

0.015

0.070

2.160

2.810

7.39

M Average

2.800

0.018

0.117

1.910

1.578

7.24

CM1

0.500

0.016

1.000

0.220

0.348

6.58

CM2

0.200

0.007

1.000

0.510

0.562

6.48

CM3

0.400

0.015

1.000

0.240

0.538

6.53

CM Average

0.367

0.013

1.000

0.323

0.483

6.53

CS1

0.400

0.015

0.000

0.370

0.318

6.36

CS2

0.400

0.012

1.000

0.210

0.467

6.43

CS3

0.300

0.019

1.000

0.120

6.160

6.46

CS Average

0.367

0.015

0.667

0.233

2.315

6.42

Rain

0.500

0.010

1.000

0.370

2.730

6.37

Treatment

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

66

Table 26: Data from rain event 2 August, 2012. All concentrations except pH and turbidity are in
milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: August 2, 2012
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

pH

S1

0.200

0.001

0.180

4.910

0.625

8.67

S2

1.000

0.020

0.320

3.670

1.240

7.21

S3

0.100

0.011

0.230

4.020

1.920

6.98

S Average

0.433

0.011

0.243

4.200

1.262

7.62

G1

0.400

0.016

0.340

2.750

0.939

7.49

G2

0.900

0.022

0.190

2.470

0.765

7.83

G3

1.100

0.026

0.390

3.430

1.310

8.56

G Average

0.800

0.021

0.307

2.883

1.005

7.96

M1

5.200

0.076

0.470

3.730

1.560

8.66

M2

5.100

0.322

0.690

2.490

1.470

7.71

M3

5.000

0.049

0.360

2.170

0.557

7.12

M Average

5.100

0.149

0.507

2.797

1.196

7.83

CM1

0.600

0.049

0.100

0.100

0.419

6.52

CM2

0.100

0.066

0.700

1.980

0.760

6.26

CM3

0.500

0.060

0.610

0.460

0.829

5.56

CM Average

0.400

0.058

0.470

0.847

0.669

6.11

CS1

0.300

0.003

0.590

0.360

0.769

6.45

CS2

0.300

0.004

0.160

0.580

0.713

6.53

CS3

0.200

0.002

0.060

0.320

1.020

6.55

CS Average

0.267

0.003

0.270

0.420

0.834

6.51

Rain

0.300

0.002

0.730

0.490

0.614

6.16

Treatment

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

67

Table 27: Data from rain event 20 August, 2012. All concentrations except pH and turbidity are
in milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: August 20, 2012
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

pH

S1

0.100

0.060

1.000

3.790

0.580

7.21

S2

1.000

0.375

0.010

4.120

1.000

6.93

S3

0.300

0.008

0.470

3.110

1.040

7.11

S Average

0.467

0.148

0.493

3.673

0.873

7.08

G1

0.800

0.077

1.000

1.390

0.951

6.25

G2

1.400

0.227

2.000

2.470

1.640

6.95

G3

0.700

0.100

2.000

1.950

0.623

6.78

G Average

0.967

0.135

1.667

1.937

1.071

6.66

M1

1.200

0.086

1.000

2.170

0.582

7.13

M2

4.900

0.073

0.280

3.470

1.010

7.26

M3

5.200

0.046

0.580

2.050

0.748

6.45

M Average

3.767

0.068

0.620

2.563

0.780

6.95

CM1

0.800

0.103

0.490

0.230

0.671

7.12

CM2

0.800

0.042

1.000

0.350

0.572

6.54

CM3

1.300

0.076

2.000

1.180

0.648

6.21

CM Average

0.967

0.074

1.163

0.587

0.630

6.62

CS1

0.300

0.002

1.000

1.250

1.210

6.32

CS2

0.300

0.010

1.000

0.500

0.429

6.47

CS3

0.400

0.010

1.000

0.230

0.570

6.11

CS Average

0.333

0.007

1.000

0.660

0.736

6.30

Rain

0.600

0.010

1.000

1.140

0.949

6.38

Treatment

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

68

Table 28: Data from rain event 8 September, 2012. All concentrations except pH and turbidity
are in milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: September 8, 2012
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

pH

S1

0.700

0.040

0.670

0.970

0.981

7.04

S2

1.300

0.032

1.000

1.360

1.310

6.63

S3

0.500

0.057

0.700

1.210

1.070

6.87

S Average

0.833

0.043

0.790

1.180

1.120

6.85

G1

0.600

0.082

0.530

2.210

0.821

6.93

G2

0.600

0.041

1.000

1.410

0.679

7.10

G3

0.900

0.067

0.590

1.010

1.183

6.91

G Average

0.700

0.063

0.707

1.543

0.894

6.98

M1

1.000

0.061

0.660

0.430

0.873

6.68

M2

0.800

0.066

0.410

0.710

1.228

6.87

M3

1.100

0.071

0.490

0.330

0.764

7.19

M Average

0.967

0.066

0.520

0.490

0.955

6.57

CM1

0.300

0.032

0.310

0.120

0.839

6.43

CM2

0.400

0.011

0.200

0.270

0.437

6.21

CM3

0.200

0.008

0.220

0.240

0.521

6.54

CM Average

0.300

0.017

0.243

0.210

0.599

6.39

CS1

0.600

0.016

0.100

0.430

0.771

6.38

CS2

0.400

0.031

0.080

0.190

0.682

6.44

CS3

0.300

0.014

0.350

0.290

0.811

6.21

CS Average

0.433

0.020

0.177

0.303

0.755

6.34

Rain

0.300

0.017

0.110

0.200

0.693

6.32

Treatment

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

69

Table 29: Data from rain event 19 September, 2012. All concentrations except pH and turbidity
are in milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: September 19, 2012
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

pH

S1

0.900

0.023

0.310

1.140

0.643

6.97

S2

1.200

0.037

0.470

0.970

0.702

7.34

S3

0.700

0.044

0.590

1.030

0.579

7.21

S Average

0.933

0.035

0.457

1.047

0.641

7.17

G1

0.600

0.036

0.520

1.090

0.489

6.61

G2

0.600

0.041

1.000

1.680

0.721

6.88

G3

0.500

0.067

0.630

0.820

0.667

7.04

G Average

0.567

0.048

0.717

1.197

0.626

6.84

M1

1.300

0.081

0.430

0.930

0.812

6.68

M2

0.600

0.033

0.670

2.110

0.935

6.83

M3

0.800

0.046

1.000

1.200

0.741

6.75

M Average

0.900

0.053

0.700

1.413

0.829

6.57

CM1

0.400

0.015

0.170

0.330

0.692

6.94

CM2

0.200

0.014

0.220

0.210

0.437

6.41

CM3

0.100

0.023

0.040

0.430

0.497

6.37

CM Average

0.233

0.017

0.143

0.323

0.542

6.57

CS1

0.100

0.019

0.150

0.370

0.881

6.59

CS2

0.300

0.044

0.160

0.220

0.643

6.21

CS3

0.200

0.009

0.270

0.100

0.410

6.18

CS Average

0.200

0.024

0.193

0.230

0.645

6.33

Rain

0.200

0.013

0.190

0.190

0.693

6.37

Treatment

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

70

Table 30: Data from rain event 24 September, 2012. All concentrations except pH and turbidity
are in milligrams per litre (mg/L) and/or parts per million (PPM).
Rain Event: September 24, 2012
NitrateNitrogen
(NO3-N)

NitriteNitrogen
(NO2-N)

AmmoniaNitrogen
(NH3-N)

PhosphatePhosphorous
(PO4-P)

Turbidity
(NTUs)

pH

S1

0.300

0.002

0.010

1.230

0.621

6.73

S2

0.300

0.002

0.000

0.930

0.357

6.81

S3

0.200

0.001

0.010

0.640

0.543

7.13

S Average

0.267

0.002

0.007

0.933

0.507

6.89

G1

0.300

0.003

0.020

0.430

0.473

7.32

G2

0.200

0.002

0.000

0.840

0.921

7.01

G3

0.300

0.002

0.020

0.920

0.833

6.97

G Average

0.267

0.002

0.013

0.730

0.742

7.10

M1

0.900

0.004

0.010

1.040

0.721

7.32

M2

0.400

0.004

0.000

0.720

1.370

6.89

M3

1.200

0.011

0.000

0.880

0.739

7.12

M Average

0.833

0.006

0.003

0.880

0.943

6.57

CM1

0.300

0.004

0.000

0.210

0.461

7.12

CM2

0.300

0.003

0.010

0.110

0.323

6.32

CM3

0.400

0.004

0.040

0.200

0.679

6.36

CM Average

0.333

0.004

0.017

0.173

0.488

6.60

CS1

0.200

0.001

0.080

0.100

0.831

6.61

CS2

0.300

0.004

0.010

0.120

0.611

6.52

CS3

0.200

0.003

0.000

0.110

0.528

6.21

CS Average

0.233

0.003

0.030

0.110

0.657

6.45

Rain

0.400

0.004

0.050

0.190

0.799

6.35

Treatment

Treatment Index: G (Green Roof w/ Grass Species), S (Green Roof w/ Sedum
Species), M (Media Only Roof Treatment), CM (Conventional Metal Roofing) & CS
(Conventional Shingle Roofing)

71

Figure 1: Graphic representations of land cover change and its subsequent effect on stormwater
runoff. (Source: http://www.fairfaxcounty.gov/nvswcd/images/drainageproblem/
runoff.jpg)

72

Figure 2: Graphic representations of urban stormwater runoff catchment area and transport
through stormwater utility systems. (Source: http://www2.cincinnati.com/blogs/
gardening/ 2011/11/28/ rain/)

73

Figure 3: Image showing the impacts of excessive nutrient runoff to receiving water bodies.
(Source: http://www.lakeforest.edu/academics/programs/environmental/courses/
seniorseminar/springbreak/students/newcomer.php)

74

Figure 4: Graphic representation of land type and use and its effect on air temperature. (Source:
http://indymedia.org.au/2013/06/02/cities-to-get-much-hotter-as-heatwaves-amplifyurban-heat-island-effect)

75

Figure 5: Artist’s illustration of the Hanging Gardens of Babylon. (Source: http://batkya.
deviantart.com/art/Hanging-Gardens-of-Babylon-203309919)

76

Figure 6: Project location map. Study was performed at the Michigan State University
Landscape Architecture Alumni Research Site, Old Mission, MI (Krogulecki, 2014).

77

Figure 7: Picture taken during construction illustrating the use of .045 mm EPMD rubber
waterproof membrane (Krogulecki, 2009).

78

Figure 8: Plan view detail of plot dimensions and randomized treatment layout (Krogulecki,
2014).

79

Figure 9: Picture showing the three green roof vegetation treatments after planting (Krogulecki,
2009).

80

Figure 10: Daily precipitation totals (inches) during the study (22 July 2010 – 24 September
2012). Data collected from the Michigan State University Old Mission weather
station. (Source: http://www.agweather.geo.msu.edu/mawn/ station.asp?id=old&rt=
24)

81

Figure 11: Daily maximum and minimum temperatures (°F) during the study (22 July 2010 – 24
September 2012). Data collected from the Michigan State University Old Mission
weather station. (Source: http://www.agweather.geo.msu.edu/mawn/station.asp?id=
old&rt=24)

82

Figure 12: Mean concentration of Nitrate-Nitrogen (NO3-N) in collected runoff samples for each
individual roofing treatment during the study period (22 July 2010 – 24 September
2012).

83

Figure 13: Mean concentration of Nitrite-Nitrogen (NO2-N) in collected runoff samples for each
individual roofing treatment during the study period (22 July 2010 – 24 September
2012).

84

Figure 14: Mean concentration of Ammonia-Nitrogen (NH3-N) in collected runoff samples for
each individual roofing treatment during the study period (22 July 2010 – 24
September 2012).

85

Figure 15: Mean concentration of Phosphate-Phosphorous (PO4-P) in collected runoff samples
for each individual roofing treatment during the study period (22 July 2010 – 24
September 2012).

86

Figure 16: Mean Turbidity (NTU) levels in collected runoff samples for each individual roofing
treatment during the study period (22 July 2010 – 24 September 2012).

87

Figure 17: Mean pH levels in collected runoff samples for each individual roofing treatment
during the study period (22 July 2010 – 24 September 2012).

88

BIBLIOGRAPHY

89

BIBLIOGRAPHY

Alsup, S., Ebbs, S., Retzlaff, W. 2010. “The exchangeability and leachability of metals from
select green roof growth substrates.” Urban Ecosystems 13: 91-111.
Barnes, K., J. Morgan, and M. Roberge. 2001. “Impervious Surfaces and the Quality of Natural
Built Environments.” Report prepared for Towson University's NASA/Raytheon/Synergy
project.
Beck, D.A., Johnson, G.R., Spolek, G.A. 2011. “Amending greenroof soil with biochar to affect
runoff water quantity and quality. Environmental Pollution 159: 2111-2118.
Berndtsson, J.C. 2010. “Green roof performance towards management of runoff water quantity
and quality: A review.” Ecological Engineering 36: 351-360.
Berndtsson, J.C., Bengtsson, L., and Jinno, K. 2009. “Runoff water quality from intensive and
extensive vegetated roofs.” Ecological Engineering 35: 369-380.
Berndtsson, J.C., Bengtsson, L., and Jinno, K. 2008. “First flush effect from vegetated roofs
during simulated rain events.” Hydrology Research 39: 171-179.
Berndtsson, J.C., Emilsson, T., and Bengtsson, L. 2006. “The influence of extensive vegetated
roofs on runoff quality.” Science of the Total Environment 355: 48-63.
Bliss, D.J., Neufeld, R.D., and Ries, R.J. 2009. “Storm water runoff mitigation using a green
roof.” Environmental Engineering Science 26: 407-418.
Carpenter, D. and Kaluvakolanu, P. 2011. “Effect of roof surface type on storm-water runoff
from full-Scale roofs in a temperate climate.” SPECIAL ISSUE: Urban Storm-Water
Management in the 21st Century, Journal of Irrigation Drainage and Engineering 137: 161–169.
Clark, C., Adriaens, P. and Talbot, B. 2008. “Green roof valuation: A probabilistic economic
analysis of environmental benefits.” Environmental Science & Technology 42: 2155-2161.
Copeland, C. 2010. “Clean Water Act: A Summary of the Law - Congressional Research Service
Report for Congress.” Congressional Research Service. 7-5700, RL30030.
Cronk, E. 2012. “Stormwater Retention Capabilities of Steep Sloped, Extensive Green Roof
Systems” MA thesis, Michigan State University, East Lansing, MI.
Deletic, A. B., Maskimovic C.T. 1998. “Evaluation of water quality factors of stormwater runoff
from paved areas.” Journal of Environmental Engineering 124: 869–879.

90

Dunnett, N. and N. Kingsbury. 2008. Planting Green Roofs and Living Walls. Oregon: Timber
Press, Inc.
Durhman, A.K., D.B. Rowe, and C.L. Rugh. 2007. “Effect of substrate depth on initial growth,
coverage, and survival of 25 succulent green roof plant taxa.” HortScience 42: 588-595
Durhman, A.K., D.B. Rowe, and C.L. Rugh. 2006. “Effect of watering regimen on chlorophyll
fluorescence and growth of selected green roof plant taxa.” HortScience 41: 1623-1628.
Eaken, C. 2012. “Assessing Wildlife Habitat Contributions of Green Roofs in Urban Landscapes
in Michigan and Illinois, U.S.A.: Measuring Avian Community Response to Green Roof
Factors” MA thesis, Michigan State University, East Lansing, MI.
Emilsson, T.U., Berndtsson, J.C., Mattson, J.E., and Rolf, K. 2007. “Effect of using conventional
and controlled release fertilizer on nutrient runoff from various vegetated roof systems.”
Ecological Engineering 29: 260-271.
Ferguson, B.K. 1998. Introduction to Stormwater: Concept, Purpose, Design.
New York, NY: John Wiley and Sons, Inc.
FLL (Forschungsgesellschaft Landschaftsentwicklung Landschaftsbau)
(Landscape, Research, Development and Construction Society). 1995. Guideline for the
Planning, Execution, and Upkeep of Green-roof Sites. Bonn, Germany
Gaffin, S., C. Rosenzweig, L. Parshall, D. Beattie, R. Berghage, G. O’Keefe, and D. Braman.
2005. “Energy balance modeling applied to a comparison of white and green roof cooling
efficiency.” P. 583-597. In Proc. of 3rd North American Green Roof Conference: Greening
Rooftops for Sustainable Communities, Washington, DC. 4-6 May 2005. The Cardinal Group,
Toronto.
Getter, K.L., D.B.Rowe, G.P. Robertson, B.M. Cregg, and J.A.Andresen. 2009a. “Carbon
sequestration potential of extensive green roofs.” Environmental Science and Technology 43:
7564-7570.
Getter, K.L., D.B.Rowe, and B.M. Cregg. 2009b. “Solar radiation intensity influences extensive
green roof plant communities.” Urban Forestry and Urban Greening 8: 269-281.
Getter, K.L. and D.B.Rowe. 2009c. “Substrate depth influences sedum plant community on a
green roof.” HortScience 44: 401-407.
Getter, K.G., D.B. Rowe, J.A. Andresen. 2007a. “Quantifying effect of slope on extensive green
roof stormwater retention.” Ecological Engineering 31: 225-231
Getter, K.L. and D.B. Rowe. 2007b. “Effect of substrate depth and planting season on Sedum
plug establishment for green roofs.” Journal of Environmental Horticulture 25: 95-99.

91

Getter, K.L. and D.B. Rowe. 2006. “The role of extensive green roofs in sustainable
development.” HortScience 41: 1276-1285.
Gregorie, B.C., and Clausen, J.C. 2011. “Effect of a modular extensive green roof on stormwater
runoff and water quality.” Ecological Engineering 37: 963-969.
Hathaway, A.M., W.F. Hunt, and G.D. Jennings. 2008. “A field study of green roof hydrologic
and water quality performance.” Transactions of the American Society of Agricultural and
Biological Engineers 51: 37-44.
C.P. Lo, and D. A. Quattrochi. 2003. “Land-use and land-cover change, urban heat island
phenomenon, and health implications: a remote sensing approach.” Photogrammetric
Engineering and Remote Sensing 69: 1053-1063.
Mantens, J., D. Raes, and M. Hermy. 2006. “Green roofs as a tool for solving the rainwater
runoff problem of the 21st century?” Landscape and Urban Planning 77: 217-226.
McDonald, A.E., B.R. Grant and W.C. Plaxton. 2001. “Phosphite (Phosphorous Acid): Its
Relevance in the Environment and Agriculture and Influence on Plant Phosphate Starvation
Response.” Journal of Plant Nutrition 24: 1505-1519.
Michigan Department of Environmental Quality (MDEQ). 2014. Accessed 24 September 2012.
http://www.michigan.gov/deq/0,4561,7-135-3313_3682_3713-10416--,00.html
Monsma, J. 2012. “A comparison of Insect and Spider Species Presence and Composition Across
Multiple Vegetated Roof Systems” MA thesis, Michigan State University, East Lansing, MI.
Monterusso, M.A., D.B. Rowe, and C.L. Rugh. 2005. “Establishment and persistence of Sedum
spp. and native taxa for green roof applications.” HortScience 40: 391-396.
Monterusso, M.A., Rowe, D.B., Rugh, C.L., and Russell, D.K. 2004. “Runoff water quantity and
quality from green roof systems.” Acta Hort. 639: 369-376.
National Research Council, Water Science and Technology Board. 2009. Urban Stormwater
Management in the United States. Washington: National Academies Press.
Oberndorfer, E., Lundholm, J., Bass, B., Coffman, R., Doshi, H., Dunnett, N., Gaffin, S., Kohler,
M., Liu, K., Bowe, B. 2007. “Green Roofs as Urban Ecosystems: Ecological Structures,
Functions and Services.” Bioscience 57: 823-833.
Peck, S.W., C. Callaghan, M.E. Kuhn, and B. Bass. 1999. “Greenbacks from
Green Roofs: Forging a new industry in Canada.” Report prepared for the Canada Mortgage and
Housing Corporation.

92

Retzlaff, W., Ebbs, S., Alsup, S., Morgan, S., Woods, E., Jost, V., and Luckett, K. 2008. “What
is That Running Off of My Green Roof?” Proceedings of the Greening Rooftops for Sustainable
Communities, Baltimore, Maryland, 30 April – 2 May, 2008.
Robertson, G.P. and P.M. Groffman. Nitrogen transformation. Pages 341-364 in E.A. Paul,
[ed.] 2007. Soil Microbiology, Biochemistry, and Ecology. New York, New York: Springer.
Rosenfeld, A.H., A. Hashem, J.J. Romm and M. Pomerantz. 1998. “Cool communities: strategies
for heat island mitigation and smog reduction.” Energy and Buildings 28: 51-62
Rowe, D.B. 2011. “Green Roofs as a Means of Pollution Abatement.” Environmental Pollution
159: 2100-2110.
Rowe, D.B., M.A. Monterusso, and C.L. Rugh. 2006. “Assessment of heat-expanded slate and
fertility requirements in green roof substrates.” HortTechnology 16: 471-477.
Schueler, T.R. 1987. “Controlling Urban Runoff: A Practical Manual for Planning and Designing
Urban BMPs.” Washington Metropolitan Water Resources Planning Board, Washington, DC.
Shaver, E., Maxted, J., Curtis, G., and Carter, D. 1995. “Watershed protection using an
integrated approach.” Proceedings of the Engineering Foundation Conference, Mount Crested
Butte, Colorado, August 7-12, 1994.
Smith, V.H., G.D. Tilman and J.C. Nekola. 1999. “Eutrophication: impacts of excess nutrient
inputs on freshwater, marine, and terrestrial ecosystems.” Environmental Pollution 100: 179-196.
Teemusk, A., and Mander, U. 2007. “Rainwater Runoff Quantity and Quality Performance from
a Greenroof: The Effects of Short-Term Events.” Ecological Engineering 30: 271-277.
Toland, D.C., Haggard, B.E., and Boyer, M.E. 2012. “Evaluation of Nutrient Concentrations in
Runoff Water from Green Roofs, Conventional Roofs, and Urban Streams.” Transactions of the
Agricultural and Biological Engineers 55: 99-106.
U. S. Environmental Protection Agency. 2012a. Basic information about nitrate in drinking
water. Accessed 24, 2014. http://water.epa.gov/drink/contaminants/basicinformation/nitrate.cfm
U. S. Environmental Protection Agency. 2012b. Water: State, Tribal & Territorial Standards –
Repository of Documents: Minnesota. Accessed 24, 2014.
http://water.epa.gov/scitech/swguidance/standards/wqslibrary/mn_index.cfm
U. S. Environmental Protection Agency. 2009a. Funding Stormwater Programs. EPA 901-F-09004. Washington, D.C.: USEPA.
U. S. Environmental Protection Agency. 2009b. Green Roofs for Stormwater Runoff Control.
EPA/600/R-09/026. Washington, D.C.: USEPA.

93

U. S. Environmental Protection Agency. 2008a. A Screening Assessment of the Potential Impacts
of Climate Change on Combined Sewer Overflow (CSO) Mitigation in the Great Lakes and New
England Regions. EPA 600/R-07/033F. Washington, D.C.: USEPA.
U. S. Environmental Protection Agency. 2008b. Reducing Urban Heat Islands: Compendium of
Strategies. EPA 430-D-08-001. Washington, D.C.: USEPA.
U. S. Environmental Protection Agency. 2007. Nitrates and Nitrites TEACH Chemical
Summary. Toxicity and Exposure Assessment for Children’s Health. Document obtained from
U.S. EPA website, http://www.epa.gov/teach/teachsummaries.html, (last accessed April 20,
2010).
U. S. Environmental Protection Agency. 2006. Excessive Heat Events Guidebook., EPA 430-B06-005. Washington, D.C.: USEPA.
U. S. Environmental Protection Agency. 2004. Report to Congress: impacts and control of CSOs
and SSOs. EPA/833/R-04/001. Office of Wastewater Management, Washington, DC;
Washington, D.C.: USEPA.
U. S. Environmental Protection Agency. 2003. Cooling summertime temperatures: Strategies to
Reduce Urban Heat Islands. EPA 430-F-03-014. Washington, D.C.: USEPA.
U. S. Environmental Protection Agency. 1985. Ambient Water Quality Criteria for Ammonia –
1984. EPA 440/5-85-001. Washington, D.C.: USEPA.
Van Seters, T., Rocha, L., Smith, D., MacMillan, G. 2009. “Evaluation of green roof for runoff
retention, runoff quality, and leachability.” Water Research 44: 33-47.
VanWoert, N., D.B. Rowe, J.A. Andresen, C.L. Rugh, R.T. Fernandez and L. Xiao. 2005a.
“Green Roof Stormwater Retention: Effects of Roof Surface, Slope, and Media Depth.” Journal
of Environmental Quality 34: 1036-1044.
VanWoert, N.D, D.B. Rowe, J.A. Andresen, C.L. Rugh, and L. Xiao. 2005b. “Watering regime
and green roof substrate design affect Sedum plant growth.” HortScience 40: 659-664.
Vijayaraghavan, K., U.M. Joshi and R. Balasubramanian. 2012. “A Field Study to Evaluate
Runoff Quality from Green Roofs.” Water Research 46: 1337-1345.
Vitousek, P. M., H. A. Mooney, J. Lubchenco, J. M. Melillo. 1997. “Human domination of the
Earth’s ecosystems.” Science 227: 494-499.
Wark, C.G., Wark, W.W. 2003. “Green Roof Specifications and Standards:
Establishing an emerging technology.” The Construction Specifier 56: No. 8.
Werthmann, Christian. Green Roof – A Case Study. New York: Princeton Architectural Press.
2007.
94