Energy and Buildings 43 (2011) 3548–3557

Contents lists available at SciVerse ScienceDirect

Energy and Buildings

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

Seasonal heat flux properties of an extensive green roof in a Midwestern U.S.

climate
Kristin L. Getter a , D. Bradley Rowe a,∗ , Jeff A. Andresen b , Indrek S. Wichman c
a
Department of Horticulture, Michigan State University, East Lansing, MI 48824, USA
b
Department of Geography, Michigan State University, East Lansing, MI 48824, USA
c
Department of Mechanical Engineering, Michigan State University, East Lansing, MI 48824, USA

a r t i c l e i n f o a b s t r a c t

Article history: Green roofs, or vegetated roofs, can reduce heat flux magnitude through a building envelope as a result of
Received 10 March 2011 insulation provided by the growing medium, shading from the plant canopy, and transpirational cooling
Received in revised form 8 September 2011 provided by the plants. This study quantifies the thermal properties of an inverted 325 m2 retro-fitted
Accepted 18 September 2011
extensive green roof versus a traditional gravel ballasted inverted roof in a Midwestern U.S. climate
characterized by hot, humid summers and cold, snowy winters. In autumn, green roof temperatures
Keywords:
were consistently 5 ◦ C lower than corresponding gravel roof temperatures. Even during chilly and moist
Building energy
conditions, the heat flux leaving the building was lower for the green roof than the gravel roof. Tem-
Building insulation
Eco-roof
peratures at the top of the insulation layer were more variable for both green roof and gravel roof on
Vegetated roof winter days with no snow cover than on days with snow cover. Variation in temperatures between roof
types in spring was similar to those in autumn. Peak temperature differences between gravel and green
roof were larger in summer than other seasons (sometimes by as much as 20 ◦ C). Over the course of a
year (September 2005–August 2006), maximum and minimum average monthly temperatures and heat
fluxes were consistently more extreme for the gravel roof than the green roof.
© 2011 Elsevier B.V. All rights reserved.

1. Introduction
Green roofs, or vegetated roofs, can reduce heat flux through
a building envelope because the “soil” or growing media acts as
an insulation layer, the plant canopy shades the roof, and plants
are believed to provide transpirational cooling [1–5]. In previous
work, Theodosiou [4] found that during a Mediterranean sum-
mer, thicker growing media increased the thermal inertia, but also
allowed taller plants and greater leaf area index (LAI) which pro-
vided more shading and transpirational cooling than thinner media
profiles. These properties of a green roof resulted in 20 ◦ C lower
temperatures beneath a green roof compared to a conventional
roof on a hot day and 10 ◦ C warmer on a cold day in temperate
regions [6]. In tropical regions the difference may be greater. Sim-
mons et al. [7] compared six different extensive green roof designs
with conventional black roofs and reflective white roofs in Texas.
When ambient air temperature reached 33 ◦ C, membrane tempera-
tures on conventional black and white reflective roofs reached 68 ◦ C
and 42 ◦ C, respectively, but only ranged between 31 and 38 ◦ C on
the green roofs. A study performed in the Pacific Northwestern U.S.
∗ Corresponding author. Tel.: +1 517 355 5191x1334.
E-mail address: rowed@msu.edu (D.B. Rowe).
found a 13% reduction in winter heat flux through a green roof and
72% in summer heat flux relative to a conventional gravel ballast
roof [8].
Because green roofs reduce heat transport (both into and out of
the building), they have the potential to reduce energy consump-
tion for heating and cooling. Sailor [9] integrated the elements of
a green roof energy balance into Energy Plus, a building energy
simulation model supported by the U.S. Department of Energy.
His simulations suggested 2% reductions in electricity consump-
tion and 9–11% reductions in natural gas consumption. Based on
his model of a generic building with a 2000 m2 green roof, savings
range from 27.2 to 30.7 GJ/year of electricity and 9.5 to 38.6 GJ/year
of natural gas, depending on climate and green roof design. An addi-
tional 25% reduction in electricity use was proposed from indirect
heat island reduction achieved from large-scale green roof imple-
mentation throughout an urban area [10].
Since buildings consume 39% of total energy used and 71% of all
electricity consumption in the U.S. [11], green roof implementation
on a wide scale could significantly impact energy consumption with
subsequent monetary savings and carbon emission reductions.
Assuming national commercial average prices of $0.1026/kWh [12]
and $10.06/Mcf [13] for electricity and natural gas, respectively,
Sailor’s [9] 2000 m2 generic green roof energy reductions trans-
late to a monetary savings of $965–1144 for electricity and natural
gas combined annually ($0.48–.57/m2 of green roof), depending

0378-7788/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.enbuild.2011.09.018
 K.L. Getter et al. / Energy and Buildings 43 (2011) 3548–3557
on climate and roof design. Other research indicated that if the
campus of Michigan State University implemented green roofs
on all low-sloped roof surfaces, CO2 emission reductions up to
3.64 × 106 kg/year would be possible due to decreased electricity
and natural gas consumption [14]. This is the equivalent of taking
650 vehicles off the road each year [15]. In addition to CO2 avoid-
ance, the green roof itself can also sequester carbon in the plant
material and growing medium [16].
Potential energy consumption reductions depend on many
factors such as green roof growing media composition, depth,
and moisture content, plant species, supplementary irrigation
and local climate. In addition, building type and construction
details may influence the thermal properties of a green roof build-
ing. Researchers in Athens, Greece found that energy savings by
installing a green roof ranged from 2% for well-insulated build-
ings versus 37–48% for non-insulated buildings [17]. In the U.S.
and other temperate countries, inverted roofs are constructed on
buildings by placing the thermal insulation above the waterproof-
ing layer. This protects the waterproofing from extreme changes
in temperature as well as UV radiation and mechanical damage. A
green roof over an inverted roof may not yield the range of thermal
benefits as a non-inverted roof, although these differences have not
been extensively studied.
Due to building weight restrictions and implantation costs, shal-
low growing media extensive green roofs are more common than
deeper intensive roofs [18–21]. Older buildings will likely benefit
the most from green roofs because newer buildings often contain
deeper layers of insulation [22]. Snow cover may also provide a very
effective natural layer of insulation [23], although very few studies
have been performed on energy analysis in climates where snow
cover is common during winter. The objective of this study was
to quantify thermal properties of an extensive green roof installed
over an inverted roof against a traditional gravel ballasted inverted
roof in a Midwestern U.S. climate characterized by hot humid sum-
mers and cold snowy winters.
2. Methodology
2.1. Roof construction
A 325.2 m2 (3500 ft2 ) extensive green roof was installed on
a portion of the Plant and Soil Sciences Building roof on the
Michigan State University campus in East Lansing, MI on 21 May
2004. Structural reinforcement of the building was not neces-
sary because the vegetated section had a saturated weight of
less than eight pounds per square foot, which is roughly equiv-
alent to the removed gravel ballast. This building is a steel-deck
inverted roof assembly insulated with extruded polystyrene. The
study area was covered with a root barrier (XF112; Xero Flor Amer-
ica LLC, Durham, NC) installed over the existing roof insulation.
A drainage mat (XF108H) was placed over the root barrier, fol-
lowed by a water retention fleece (XF159). The growing media
and plants were installed by laying out pre-grown vegetated mats
(XF301) which consisted of a carrier containing at least 5 cm (2.0
in) of proprietary media (XeroTerr® ) consisting of a combination of
heat expanded slate (PermaTill, Salisbury, NC), sand, and organic
matter. Physical properties of the growing medium were sam-
pled and analyzed (Great Lakes Laboratories, Inc., Fort Wayne,
IN) and consisted of 91.18% total sand (21.96% very coarse sand
(1.0–2.0 mm), 40.8% coarse sand (0.5–1.0 mm), 24.06% medium
sand (0.25–0.5 mm), 3.36% fine sand (0.10–0.25 mm), 0.4% very
fine sand (0.05–0.1 mm)), 7.16% silt, and 1.66% clay. The bulk den-
sity was 1.17 g/cm3 , capillary pore space was 19.96%, non-capillary
pore space was 21.43%, and the water holding capacity at 0.01 MPa
was 17.07%.
3549
Fig. 1. Location of temperature, heat flux, and moisture sensors on extensive green
roof and gravel ballasted roof.
2.2. Plant material and culture
Plant species on the pre-grown mats were Hylotelephium
spectabile, Hylotelephium verticillatum, Saxifraga granulate, Sedum
acre, Sedum album, Sedum ellacombianum, Sedum floriferum, Sedum
kamtschaticum, Sedum pulchellum, Sedum reflexum, Sedum sexangu-
lare, and Sedum spurium ‘Coccineum’. Slow release fertilizer (LESCO
14-14-14 Professional Landscape and Ornamental All-Purpose Fer-
tilizer; Cleveland, OH) was applied on 15 April 2005 at a rate of
14 g/m2 and again yearly each spring. Irrigation, which occurred
on an as-needed basis by hand watering during 2004, ceased
thereafter.
2.3. Environmental monitoring
Temperature, heat flux, soil moisture, and ambient weather
conditions were recorded using a CR10X Datalogger (Campbell
Scientific, Inc., Logan, UT), three AM25T 25-Channel Solid-State
Multiplexors (Campbell Scientific, Inc., Logan, UT), and a TDR-100
Time-Domain Reflectometer (Campbell Scientific, Inc., Logan, UT).
The gravel portion of the roof and the vegetative side of the roof
each had three measurement stations, spaced 3.0 m (10 ft) apart.
Each station had five thermocouples (Campbell Scientific Type T
105T-L) installed in the following profile locations: 1 m (3.3 ft)
above the roof inside a non-aspirated solar radiation shield; on
top of the growing media (or gravel); on top of the insulation;
on top of the roofing membrane; and inside the building directly
against the ceiling (Fig. 1). Thermocouple accuracy ranged from
±0.365 ◦ C, ±0.1 ◦ C, and ±0.25 ◦ C for temperatures in the range of
−73 ◦ C, 20 ◦ C, and 50 ◦ C, respectively. A universal heat flow sen-
sor (Concept Engineering, Old Saybrook, CT) was installed on top of
the roof insulation at each measurement station. The sensor, which
was deployed so that negative and positive readings would mea-
sure heat entering and leaving the building, respectively, had an
accuracy of ±5%.
A soil moisture probe (Moisture Sensor CS605-L 3-rod TDR
probe-RG58, Campbell Scientific) was installed horizontally at each
site either directly on top of the water retention fabric (for the
green roof) or directly on top of the insulation (for the gravel roof)
so that all prongs were covered either below the media or below
3550 K.L. Getter et al. / Energy and Buildings 43 (2011) 3548–3557

Fig. 3. Monthly total precipitation (mm) from a nearby Enviro-weather (Michigan

Fig. 2. Monthly average maximum and minimum air temperatures (◦ C) on the stud- Automated Weather Network’s) East Lansing, MI weather station.
ied green roof and gravel roof compared to a nearby Enviro-weather (Michigan
Automated Weather Network’s) East Lansing, MI weather station.

Fig. 4. Thermocouple and heat flux data over 24-h periods for various environmental conditions during the autumn season. Time shown in military time. Data symbols only
shown on every other data point for clarity of graphing. Negative and positive heat flux readings represent heat entering and leaving the building, respectively.
 K.L. Getter et al. / Energy and Buildings 43 (2011) 3548–3557
Fig. 5. Volumetric moisture content (%) over 24-h periods for various environmental
conditions during three seasons. Time shown in military time. Data symbols only
shown on every other data point for clarity of graphing.
the gravel. Probe rods measured 30 cm in length and 0.475 cm in
diameter and had a radius of influence of 50 mm. Probe offsets and
calibrations were performed using PCTDR software which is based
on Castiglione and Shouse’s work [24] for time-domain reflectom-
etry electrical conductivity errors. The TDR probes placed under
the gravel produced semi-quantitative measurements (especially
during and after rainfall) due to the radius of influence measuring
much air. This was not the case with the green roof TDR probes
since probes were laid at the bottom of the media and were thus
covered on the top side by at least 50 mm of media and below by
water retention fabric. Even though the gravel TDR probes were
under the influence of air, their use in this study still serves as a
useful comparison for the presence of water.
Irradiance levels were measured using a Pyranometer (L1200X-
L, LI-COR Biosciences, Lincoln, NE) which had an maximum absolute
error of ±5% in natural daylight. Precipitation was measured using a
tipping bucket rain gage (TE525WS, Campbell Scientific, Inc., Logan,
UT) which had an accuracy of ±1%, +0 and −2.5%, and +0 and −3.5%
for rainfalls <25.4 mm/h, 25.4–50.8 mm/h, and 50.8–76.2 mm/h,
respectively. The datalogger was programmed using PC208W Dat-
alogger Support Software (Campbell Scientific, Inc., Logan, UT) to
sample temperature, heat flux, and solar irradiance every minute
(0.0167 Hz) and the volumetric moisture content was sampled
every 5 min (0.00333 Hz). These data were averaged and reported
3551
Fig. 6. Volumetric moisture content and heat flux on a 24 h rainy autumn day
(November 26, 2005) for (A) gravel roof and (B) green roof. Time shown in military
time. Negative and positive heat flux readings represent heat entering and leaving
the building, respectively.
at hourly intervals. Precipitation data were polled every minute,
then totaled and reported at hourly intervals. Precipitation data
were excluded in the winter months, due to the inability of
measuring solid precipitation. Data were collected and analyzed
for the 1 year period between 1 September 2005 and 31 August
2006.
3. Results and discussion
During the study period, average monthly minimum and max-
imum ambient air temperatures 1.5 m above ground level at a
nearby East Lansing, MI Enviro-weather station (Michigan Enviro-
weather Network) ranged from −7.7 ◦ C in February to 28.5 ◦ C in
July (Fig. 2). Averaged across each month, the ambient air temper-
ature 1 m above the gravel roof was always warmer than that of
the green roof, which was warmer than the ground-level nearby
Enviro-weather station. Precipitation data, including snow water
equivalent and snow depth, were obtained from a NOAA Coopera-
tive Climatological Station 3 km south of the study site and totaled
955 mm for the period (Fig. 3). The green roof was covered with
snow for a total of 78 days (i.e., approximately 21% of the year)
whereas the gravel portion of the roof had only 34 days for same
time period or approximately 9% of the year, less than half the green
roof amount (data not shown).
3.1. Autumn
During autumn (1 September 2005–30 November 2005), ambi-
ent air temperatures 1 m above the gravel roof and green roof were
very similar for different environmental conditions (Fig. 4), with
the vegetated air temperatures slightly lower. Temperatures above
the insulation layer followed a diurnal pattern in the absence of
precipitation (Fig. 4A and B), starting the day cooler than the roof
membrane temperatures and peaking above the roof membrane
3552 K.L. Getter et al. / Energy and Buildings 43 (2011) 3548–3557

Fig. 7. Thermocouple and heat flux data over 24-h periods for various environmental conditions during the winter season. Time shown in military time. Data symbols only
shown on every other data point for clarity of graphing. Negative and positive heat flux readings represent heat entering and leaving the building, respectively.

temperatures mid-day as solar radiation increased. The above-
insulation green roof peak temperatures were consistently 5 ◦ C
lower than the corresponding gravel temperatures. However, when
outside air temperatures were lower than 10 ◦ C and in the presence
of precipitation (Fig. 4C), the roof membrane surface temperatures
sharply decreased, with the gravel temperatures remaining much
higher than the green roof temperatures (Fig. 4C). In these chilly,
moist conditions, heat flux was from the building. The heat loss
(taken as positive) is noticeably lower for the green roof.
Volumetric moisture of the green roof growing medium on a
rainy day tended to rise steadily throughout the day, whereas
gravel moisture content rose slightly during rainfall but then
declined quickly afterwards (Fig. 5A). This changing volumetric
moisture content appeared to be correlated with heat flux (Fig. 6).
As volumetric moisture increased (due to precipitation), heat flux
decreased for both gravel and green roof (i.e., lower heat losses from
the building). Once precipitation stopped, at 1200 h (Fig. 5A), the
gravel portion of the roof drained much more quickly than the green
roof and heat flux leaving the roof quickly rebounded (increased) as
volumetric moisture returned to near zero values (Figs. 5A and 6A).
In contrast, growing media moisture for the green roof remained
high after precipitation ceased and heat flux values remained much
lower than before the precipitation event (Fig. 5B). Fig. 6 shows that
(after ∼1400 h), when the heat flux was outward (loss) for the gravel
roof, it was inward (gain) for the green roof. When plotted over a
season, this relationship between heat flux and volumetric mois-
ture content was not observed and could not be quantified, likely
because of the varying air temperatures over the large timeframe
(data not shown).
3.2. Winter
Winter (1 December 2005–28 February 2006) temperatures in
the top of the insulation layer with no snow cover were much more
 K.L. Getter et al. / Energy and Buildings 43 (2011) 3548–3557 3553

Fig. 8. Thermocouple and heat flux data over 24-h periods for various environmental conditions during the spring season. Time shown in military time. Data symbols only
shown on every other data point for clarity of graphing. Negative and positive heat flux readings represent heat entering and leaving the building, respectively.

variable than on days with snow cover (Fig. 7), especially under
sunny conditions (Fig. 7A). When both the green roof and gravel
roof were both covered with snow, their top of insulation and top
of roof membrane temperatures were nearly identical (Fig. 7C), as
reported for other cold snowy climates [25]. In general, a lower
quantity of heat leaves the building over the green roof than the
gravel roof (Fig. 7).
3.3. Spring
Trends during the spring season (1 March 2006–31 May 2006)
were similar to those in autumn (Fig. 8). Temperatures above the
insulation layer followed a diurnal pattern in the absence of precip-
itation (Fig. 8A and B), starting cooler than the roof membrane tem-
peratures early in the day and peaking above the roof membrane
temperatures mid-day as the influence of solar radiation accumu-
lated. At their peak, the above-insulation green roof temperatures
were consistently 3–5 ◦ C lower than the corresponding gravel
temperature. With precipitation, that pattern was disrupted
(Fig. 8C). In fact, temperatures on top of the insulation on the
green and gravel roofs never exceeded the top of membrane tem-
peratures. While both the gravel and green roof top of the roof
membrane temperatures were fairly constant throughout the day
in the absence of precipitation (both staying near 20 ◦ C), the pres-
ence of moisture changes this pattern. Nevertheless, springtime
gravel and green roof temperatures remain numerically close.
In the absence of precipitation, heat fluxes for the gravel and
green roof were similar and followed a diurnal pattern symmetri-
cally opposite that of solar radiation (Fig. 8A and B). At the daily
solar radiation peak, the heat flux entering the building (nega-
tive) was greatest through the gravel roof. With precipitation, heat
flux values were very different between the gravel and green roof
(Fig. 8C). In this case, gravel ballasted areas remained neutral in
terms of heat loss/gain, whereas the green roof exhibited a steady
negative heat flux entering the building. This is likely caused by the
volumetric moisture differences between the gravel and green roof
3554 K.L. Getter et al. / Energy and Buildings 43 (2011) 3548–3557

Fig. 9. Thermocouple and heat flux data over 24-h periods for various environmental conditions during the summer season. Time shown in military time. Data symbols only
shown on every other data point for clarity of graphing. Negative and positive heat flux readings represent heat entering and leaving the building, respectively.

growing media. The green roof growing medium remained above
20% volumetric moisture content all day while the gravel never
peaked above 15% and quickly returned to near 0% after rain cessa-
tion (Fig. 5B). This is consistent with research from China showing
rapid changes in soil heat flux corresponding to large soil moisture
changes [26].
3.4. Summer
Ambient air temperatures 1 m above the gravel and green roof
in the summer season (1 June 2006–31 August 2006) were similar
to trends during spring (Fig. 9). Temperatures above the insulation
also show similar trends to those in spring, but the peak differences
between gravel and green roof were larger in the summer (some-
times as much as 20 ◦ C) than other seasons. Temperatures on top
of the roof membrane were similar for gravel and green roof. The
gravel heat flux was consistently lower than the green roof dur-
ing daytime (more heat entering the building during the day) and
nearly equal at night, although the green roof began losing build-
ing heat earlier than the gravel roof. Differences between green roof
and gravel flux during the summer in the presence of rain were not
as large as those during spring, despite large differences in volu-
metric moisture content (compare Figs. 5C and 9C). The volumetric
moisture content of the gravel roof briefly spiked to 10%, but quickly
fell to near zero within hours, whereas the green roof media stayed
between 20% and 30% (Fig. 5C).
Not surprisingly, heat gain in the summer was lower on cloudy
and rainy days than on sunny days, indicating a strong dependence
on solar radiation, which is in agreement with previous research
[26]. However, on rainy days, this reduction could also be caused by
evaporative cooling or evapotranspiration. Meng and Hu [2] found
evaporative cooling could account for up to 25 ◦ C cooling at the
 K.L. Getter et al. / Energy and Buildings 43 (2011) 3548–3557 3555

Fig. 11. Daily volumetric moisture content values averaged for each month over the
course of a year for gravel and green roof.

green roof ranged from 12% to 25% higher than the gravel during
the growing season (April through September; Fig. 11).
Fig. 12 plots two heat transfer quantities of interest at different
times of the year. The first quantity is the temperature difference
across the roof from the ceiling inside the building to the top of the
media (not the air) outside the building, see Fig. 1. Fig. 12A and C

Fig. 10. Daily maximum and minimum temperatures values averaged for each
month over the course of a year for (A) top of roof insulation and (B) top of roof
membrane for gravel roof and green roof.

surface and 5 ◦ C cooling beneath a moist 30 mm growing medium.

Others have found 30 ◦ C reductions of a concrete slab temperature
on hot summer days where moisture was present in the planting
layer [27].

3.5. Annual pattern

Over the course of a year (September 2005–August 2006), aver-

age minimum and maximum roof temperatures and heat fluxes
demonstrate the role that roof vegetation plays (Fig. 10). At the top
of the insulation, the maximum and minimum average monthly
temperatures on the gravel roof were consistently more extreme
compared to the green roof (Fig. 10A). The same was not true at
the top of the roofing membrane (Fig. 10B). Here the gravel maxi-
mum temperatures were higher than the green roof temperatures,
Fig. 12. Proportionality ratio between temperature and heat flux (see Eq. (1)) for
but the gravel minimum temperatures were warmer, especially in
three 3-day periods (A) 2 September 2005–4 September 2005; (B) 3 February 2006–5
the winter months, likely caused by the saturated conditions of February 2006; (C) 1 May 2006–3 May 2006. Data symbols only shown on every
the green roof media (Fig. 11). Volumetric moisture content of the other data point for clarity of graphing.
3556 K.L. Getter et al. / Energy and Buildings 43 (2011) 3548–3557
Table 1
Monthly and seasonal cumulative heat flux (W m−2 ) of a green roof and conventional gravel ballasted roof.
Season Month Cumulative heat flux (W m−2 )
Green roof
Autumn September 1224
October 2244
November 3349
Winter December 3061
January 909
February 3899
Spring Mar 163
April 440
May 1170
Summer Jun 347
July 99
August 215
Note: The heat flux sensor was placed such that a negative and positive reading signifies heat entering and leaving the building respectively.
shows variations in September and May, respectively. Fig. 12B for
February shows that these temperature differences are essentially
identical. The gravel roof temperature difference was always larger
at midday (solar flux) than the green roof difference. The second
quantity is the ratio of the temperature difference to the heat flux,
Tinside ceilling − Toutside roofsurace
K=
Q the building (thus providing cooling), whereas for the gravel roof
where Q is the measured heat flux at the roof membrane (Fig. 1):
Q is negative when heat is transferred out of the building, positive
when heat is transferred into the building. Quantity K has units
[W/m2 K]−1 . Quantity K is a quantitative measure of heat transfer
across the roof. When the heat transfer is exactly steady, quan-
tity K is referred to as the “R-value.” When the heat transfer is
not exactly steady the standard definition of the “R-value” [28] is
not possible, but the ratio K may instead be employed to indicate
which heat transfer intervals are steady, unsteady or quasi-steady.
Fig. 12A–C shows that a quasi-steady condition is achieved when
the influences of solar irradiance are negligible. K is negative since a
positive temperature difference (inside building warmer than out-
side) produces a negative (outward) heat flux. In the quasi-steady
state K ∼ −45 m2 K/W. When the heat flux changes direction and
undergoes transition from ‘into’ to ‘out of’ the building (or vice
versa) K can become very large and spike upward or downward.
The concept of ratio K is explicitly useful only under quasi-steady
conditions away from the large mid day solar heat flux variation.
The quasi-steady value, K ∼ −45 m2 K/W, obtained only by aver-
aging data points outside the high solar flux variation periods, is
essentially constant throughout the year. This indicates that in the
quasi-steady state K is a characteristic quantity for this roof and not
a climate-dependent variable. The quasi-steady gravel and green
roof K values are nearly identical, but the gravel roof fluctuations
are almost always larger. Data for summer are not included because
quasi-steady conditions are rarely realized: the solar flux is impor-
tant during the larger part of the day.
As noted by Jim and He [26], and as seen in Eq. (1), heat flux on
these roofs is strongly influenced by temperature (i.e., heat flux Q is
lower in the summer and higher in the winter; Table 1). Averaged
over each season, gravel and green roof heat flux values were most
similar during the fall and spring (Table 1). This is likely caused
by the transitional nature of these seasons. However, as indicated
by the large variations in heat transfer during spring and autumn
(Fig. 12A and C), it is also likely that a good portion of the green roof
may be barren due to plant die-back, affecting the performance of
the green roof [29].
Monthly cumulative heat flux averaged Seasonal average
over a season (W m−2 ) reduction (%)
Gravel roof Green roof Gravel roof
427 2272 2164 −5
2357
3708
3430 2623 3017 13
1134
4487
218 591 610 3
620
993
−74 220 −327 167
−540
−367
The green roof reduced the heat flux (out of the building) by an
average of 13% in winter. This value is identical to the reduction
reported by Spolek [8]. The largest difference in heat flux occurred
during the summer, which is in agreement with previous studies
[6,8,25,29]. Here, however, the average reduction in heat flux was
167% (Table 1), larger than other findings. Average summer heat
fluxes demonstrated that the green roof heat flux is often out of
(1)
it is opposite – heat flux is always entering the building. Had a
deeper growing medium been used, it may have allowed for larger
plants with potentially larger LAI which could reduce canopy air
temperatures [5] and further reduce the heat flux into the building
[30–32].
4. Conclusions
Results demonstrate how roof temperatures and heat flux are
influenced by an extensive green roof in Michigan during different
seasons of the year. The greatest impact was during the summer
as the green roof reduced heat flux through the building envelope
by an average of 13% in winter and 167% during summer. Summer
cumulative monthly heat flux values show a net heat gain into the
building for the gravel roof while the green roof showed a cooling
effect on the building. In terms of temperature, maximum and min-
imum average monthly temperatures over the course of the year
were consistently more extreme for the gravel ballasted roof than
the green roof and the gravel roof was up to 20 ◦ C warmer during
the summer. The transition seasons (autumn and spring) showed
similar responses between the green and gravel roof, as well as
the winter when there was snow cover on both roofs. This study
agrees with other reports that heat transfer and thermal differ-
ences between green roofs and gravel roofs appear to be primarily
influenced by solar radiation, ambient outside temperature, and
volumetric moisture content of the growing medium. The presence
of snow was also a controlling factor. For all variables measured, the
gravel roof generally exhibited larger fluctuations than the green
roof.
It was shown the under quasi-steady state conditions Eq. (1)
correlated the temperature difference and heat flux data through
the experimentally measured quantity K, whose quasi-steady state
numerical value (K ∼ −45 m2 K/W) was identical across all four sea-
sons. This indicates that in the quasi-steady state the heat transfer
across the roof (whether green or gravel) can be readily calculated.
Transitional states (for example, around noon when the solar irra-
diance is important) are not correlated well by Eq. (1). In fact, these
 K.L. Getter et al. / Energy and Buildings 43 (2011) 3548–3557
transitional time periods are characterized by large positive and
negative swings in the K-value.
The results from this study suggest the potential for energy use
reductions for green roofed buildings with associated monetary
savings but also environmental benefits as carbon emissions are
reduced by lower energy consumption. The magnitude of these sav-
ings will be influenced by many factors including climate, type of
roof, the amount of insulation provided during building construc-
tion, growing media depth and composition, plant selection, and
whether the roof is irrigated. This particular study was conducted
on an inverted roof which would minimize the benefits because the
insulation is located above the roofing membrane. It also took place
in a location that experiences hot, humid summers along with cold,
snowy winters on a sedum dominated shallow depth roof without
irrigation. Increasing media depth or supplying irrigation would
allow the use of plants with greater biomass and leaf area, which
in turn would lead to higher evapotranspiration rates. In an area
with a completely different climate such as the tropics, plant selec-
tion would be completely different making these results difficult to
extrapolate.
Acknowledgements
Funding for this study was provided by Ford Motor Co., Deaborn,
MI; XeroFlor America, Durham, NC, and the MSU Office of Vice
President of Finance and Operations. The authors acknowledge
technical assistance from Clayton Rugh, XeroFlor America, Durham,
NC; Bernd Leinauer, New Mexico State University, Las Cruces, NM;
and Abagail Varner and Jiawei Qin, Michigan State University, East
Lansing, MI.
References
[1] E.P. Del Barrio, Analysis of the green roofs cooling potential in buildings, Energy
and Buildings 27 (1998) 179–193.
[2] Q. Meng, W. Hu, Roof cooling effect with humid porous medium, Energy and
Buildings 37 (2005) 1–9.
[3] T. Takakura, S. Kitade, E. Goto, Cooling effect of greenery cover over a building,
Energy and Buildings 31 (2000) 1–6.
[4] T.G. Theodosiou, Summer period analysis of the performance of a planted roof
as a passive cooling technique, Energy and Buildings 35 (2003) 909–917.
[5] N.H. Wong, Y. Chen, C.L. Ong, A. Sia, Investigation of thermal benefits of rooftop
garden in the tropical environment, Building and Environment 38 (2003)
261–270.
[6] A. Teemusk, U. Mander, Temperature regime of planted roofs compared with
conventional roofing systems, Ecological Engineering 36 (2010) 91–95.
[7] M.T. Simmons, B. Gardiner, S. Windhager, J. Tinsley, Green roofs are not created
equal: the hydrologic and thermal performance of six different extensive green
roofs and reflective and non-reflective roofs in a sub-tropical climate, Urban
Ecosystems 11 (2008) 339–348.
[8] G. Spolek, Performance monitoring of three ecoroofs in Portland, Oregon, Urban
Ecosystems 11 (2008) 349–359.
3557
[9] D.J. Sailor, A green roof model for building energy simulation programs, Energy
and Buildings 40 (2008) 1466–1478.
[10] H. Akbari, S. Konopacki, Calculating energy-saving potentials of heat island
reduction strategies, Energy Policy 33 (2005) 721–756.
[11] U.S. Green Building Council Research Committee, A National Green Building
Research Agenda. http://www.usgbc.org/ShowFile.aspx?DocumentID=3402,
2007.
[12] U.S. Energy Information Administration, Annual Electric Power Industry Report,
Form EIA-861, 2009.
[13] U.S. Energy Information Administration, Monthly Report of Natural Gas Pur-
chases and Deliveries to Consumers, Form EIA-857, 2011.
[14] D.B. Rowe, Green roofs as a means of pollution abatement, Environmental
Pollution 159 (2011) 2100–2110.
[15] U.S. Environmental Protection Agency, Office of Transportation and Air Quality,
Emission Facts: Greenhouse Gas Emissions from a Typical Passenger Vehicle,
EPA420-F-05-004, Washington, DC, 2005.
[16] K.L. Getter, D.B. Rowe, G.P. Robertson, B.M. Cregg, J.A. Andresen, Carbon
sequestration potential of extensive green roofs, Environmental Science and
Technology 43 (2009) 7564–7570.
[17] A. Niachou, K. Papakonstantinou, M. Santamouris, A. Tsandgrassoulis, G. Miha-
lakakou, Analysis of the green roof thermal properties and investigation of its
energy performance, Energy and Buildings 33 (2001) 719–729.
[18] N. Dunnett, N. Kingsbury, Planting Green Roofs and Living Walls, Timber Press
Inc., Portland, OR, 2004.
[19] K.L. Getter, D.B. Rowe, The role of extensive green roofs in sustainable devel-
opment, HortScience 41 (2006) 1276–1285.
[20] E. Oberndorfer, J. Lundholm, B. Bass, R.R. Coffman, H. Doshi, N. Dunnett, S. Gaffin,
M. Kohler, K.K.Y. Liu, D.B. Rowe, Green roofs as urban ecosystems: ecological
structures, functions, and services, BioScience 57 (2007) 823–833.
[21] E. Snodgrass, L. Snodgrass, Green Roof Plants: A Resource and Planting Guide,
Timber Press Inc., Portland, OR, 2006.
[22] H.F. Castleton, V. Stovin, S.B.M. Beck, J.B. Davison, Green roofs: building
energy savings and the potential for retrofit, Energy and Buildings 42 (2010)
1582–1591.
[23] T. Zhang, Influence of the seasonal snow cover on the ground thermal regime:
an overview, Review of Geophysics 43 (2005), doi:10.1029/2004RG000157,
RG4002.
[24] P. Castiglione, P.J. Shouse, The effect of ohmic cable losses on time-domain
reflectometry measurements of electrical conductivity, Soil Science Society of
America Journal 67 (2003) 414–424.
[25] K.K.Y. Liu, B.A. Baskaran, Thermal Performance of Green Roofs through
Field Evaluation, National Research Council, Canada, NRCC-46412, 2003,
http://www.nrc-cnrc.gc.ca/obj/irc/doc/pubs/nrcc46412/nrcc46412.pdf.
[26] C.Y. Jim, H. He, Coupling heat flux dynamics with meteorological con-
ditions in the green roof ecosystem, Ecological Engineering 36 (2010)
1052–1063.
[27] S. Onmura, M. Matsumoto, S. Hokoi, Study on evaporative cooling effect of roof
lawn gardens, Energy and Buildings 33 (2001) 653–666.
[28] J.P. Holman, Heat Transfer, 8th ed., McGraw-Hill, NY, 1997, pp. 26–75 (Chapter
2).
[29] K.K.Y. Liu, B.A. Baskaran, Thermal Performance of Extensive Green Roofs
in Cold Climates, National Research Council, Canada, NRCC-48202, 2005,
http://www.nrc-cnrc.gc.ca/obj/irc/doc/pubs/nrcc48202/nrcc48202.pdf.
[30] R. Kumar, S.C. Kaushik, Performance evaluation of green roof and shading
for thermal protection of buildings, Building and Environment 40 (2005)
1505–1511.
[31] N.H. Wong, D.K.W. Cheong, H. Yan, J. Soh, C.L. Ong, A. Sia, The effects of rooftop
garden on energy consumption of a commercial building in Singapore, Energy
and Buildings 35 (2003) 353–364.
[32] C.Y. Jim, S.W. Tsang, Biophysical properties and thermal performance
of an intensive green roof, Building and Environment 46 (2011) 1263–
1274.