Science of the Total Environment 512–513 (2015) 82–93

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Establishment and performance of an experimental green roof under extreme climatic conditions Petra M. Klein a,⁎, Reid Coffman b a b

School of Meteorology, University of Oklahoma, Norman, OK, USA College of Architecture and Environmental Design, Kent State University, Kent, OH, USA

H I G H L I G H T S • • • •

Native plant species established more readily than exotic succulents under extreme conditions. In mixed community establishment, Bouteloua gracilis proved to be the most dominant species. Green roof temperatures and buoyancy fluxes at 1.5 m tended to be lower than over a concrete roof. Higher net radiation over the green roof was compensated by higher evapotranspiration rates.

a r t i c l e

i n f o

Article history: Received 16 September 2014 Received in revised form 9 January 2015 Accepted 11 January 2015 Available online 20 January 2015 Editor: Simon Pollard Keywords: Green roof Native species Drought Albedo Urban heat island mitigation Surface energy balance

a b s t r a c t Green roofs alter the surface energy balance and can help in mitigating urban heat islands. However, the cooling of green roofs due to evapotranspiration strongly depends on the climatic conditions, and vegetation type and density. In the Southern Central Plains of the United States, extreme weather events, such as high winds, heat waves and drought conditions pose challenges for successful implementation of green roofs, and likely alter their standard performance. The National Weather Center Experimental Green Roof, an interdisciplinary research site established in 2010 in Norman, OK, aimed to investigate the ecological performance and surface energy balance of green roof systems. Starting in May 2010, 26 months of vegetation studies were conducted and the radiation balance, air temperature, relative humidity, and buoyancy fluxes were monitored at two meteorological stations during April–October 2011. The establishment of a vegetative community trended towards prairie plant dominance. High mortality of succulents and low germination of grasses and herbaceous plants contributed to low vegetative coverage. In this condition succulent diversity declined. Bouteloua gracilis and Delosperma cooperi showed typological dominance in harsh climatic conditions, while Sedum species experienced high mortality. The plant community diversified through volunteers such as Euphorbia maculate and Portulaca maculate. Net radiation measured at a green-roof meteorological station was higher than at a control station over the original, light-colored roofing material. These findings indicate that the albedo of the green roof was lower than the albedo of the original roofing material. The low vegetative coverage during the heat and drought conditions in 2011, which resulted in the dark substrate used in the green roof containers being exposed, likely contributed to the low albedo values. Nevertheless, air temperatures and buoyancy fluxes were often lower over the green roof indicating that higher evapotranspiration rates compensated for the higher net radiation at the green roof. © 2015 Elsevier B.V. All rights reserved.

1. Introduction City air temperatures are typically higher than in surrounding rural areas, a phenomenon described as the urban heat island (Arnfield, 2003; Grimmond et al., 2010; Oke, 1982). Heat stress for urban populations is thus often exacerbated during heat waves, which are predicted ⁎ Corresponding author at: School of Meteorology, University of Oklahoma, 120 David L. Boren Blvd., Norman, OK 73072, USA. E-mail addresses: [email protected] (P.M. Klein), [email protected] (R. Coffman).

http://dx.doi.org/10.1016/j.scitotenv.2015.01.020 0048-9697/© 2015 Elsevier B.V. All rights reserved.

to become more frequent and intense due to the increase of greenhouse gas emissions (Patz et al., 2005). The elevated urban air temperatures can be explained by changes in the surface energy and radiation balance in built-up areas. Compared to natural, vegetated environments, cities have different aerodynamic and thermal properties, a lower albedo (i.e. they absorb more solar radiation), and are typically impermeable surfaces that retain less water. Several cities have launched programs to increase urban vegetation as measures for improving urban climate and air quality (e.g. Pincetl, 2010). Bowler et al. (2010) conducted a systematic review of empirical evidence about cooling effects of urban

P.M. Klein, R. Coffman / Science of the Total Environment 512–513 (2015) 82–93

vegetation. They concluded that on average urban parks are about 1 °C cooler than nearby, built-up areas. Shading effects of trees and other taller vegetation along with increased evapotranspiration (ET) appear to contribute to these cooling effects. While there is evidence that urban vegetation can play an important role in sustainable urban development and in mitigating urban heat islands, further research is needed to successfully implement urban greening initiatives (Bowler et al., 2010; Gago et al., 2013). The scale of the cooling effects beyond the green spaces is still an open question. This is particularly true for vegetated roof systems. Berardi and GhaffarianHoseini (2014) present a comprehensive review of the environmental benefits of green roofs. They were shown to reduce heating and cooling loads of buildings, mitigate storm water run-off, and improve air quality in cities (Baik et al., 2012; Berndtsson, 2010; Jaffal et al., 2012; Rowe, 2011; Oberndorfer et al., 2007), but the scale of their microclimate benefits is still not well defined. A number of recent studies have discussed the performance of vegetated roofs under different climate conditions (Dvorak and Volder, 2013b; Farrell et al., 2012; Fioretti et al., 2010; Lin et al., 2013; Olivieri et al., 2013; Ouldboukhitine et al., 2012; Coutts et al., 2013) and for different types of roof insulations (D'Orazio et al., 2012). Temperatures at the surface and below the substrate of vegetated green roof systems were found to be significantly lower than surface temperatures of traditional roofs (Dvorak and Volder, 2013b; Gaffin et al., 2009; Susca et al., 2011). While some studies reported that the surface temperatures decreased by as much as 60 °C (Saadatian et al., 2013), much more moderate average reductions (18 °C) and even an increase in surface temperatures under drought conditions were observed for green roof systems using dark substrates and low-lying groundcover-type plants with limited surface cover (Hien et al., 2007). The overall thermal performance of green roofs appears to be strongly dependent on climatic conditions, particularly the rainfall and irrigation amounts, and it also varies diurnally (Coutts et al., 2013; Lin et al., 2013; Zinzi and Agnoli, 2012). Uncertainties also remain concerning the benefits of green roofs in reducing air temperatures above roofs and improving the microclimate in nearby streets. Experimental studies have shown that air temperatures above green roofs are reduced compared to traditional roofs but the influence is limited to just a couple of meters above the roof surface (Wong et al., 2003). Ouldboukhitine et al. (2014a) measured air temperatures in scaled-down street canyons with traditional and green roofs and found that the vegetation reduced the air temperature of the street canyon by 0.8 °C. However, green roof systems with dark substrates and limited vegetation cover can also lead to an increase in air temperatures (Hien et al., 2007). Some modeling studies also suggest that green roofs can reduce air temperatures in nearby streets: Peng and Jim (2013) found that the observed temperature reductions were quite small (≈10 °C or less) while Alexandri and Jones (2007) reported that greening of roofs can reduce average street-canyon temperatures by as much as ≈ 10 °C in hot, dry climates. A number of studies have also focused on the potential of using different roof materials, which can include both cool (reflective materials with high-albedo) and green roofs, as city-wide heat-island mitigation strategies (Berardi and GhaffarianHoseini, 2014; Santamouris, 2014). Cool roofs primarily alter the radiation balance of the surface with the higher albedo leading to lower net radiation, which reduces the amount of energy available for partitioning into sensible and latent heat (Li et al., 2014). Green roofs, on the other hand, are known to increase latent heat fluxes due to higher ET rates (Santamouris, 2014). Scherba et al. (2011) found that replacing black roofs by white or green roofs drastically reduces the sensible heat flux to the environment. Li et al. (2014) investigated city-scale impacts of green and cool roofs using numerical simulations with the Weather Research and Forecasting (WRF) model and an urban canopy model. Green roofs with abundant surface moisture were found to lead to comparable reductions in urban surface and air temperatures as cool roofs,

83

whereby the reduction of surface temperatures was much more pronounced than the reduction in 2-m air temperatures. Oleson et al. (2010) studied the influence of cool roofs on urban temperatures using global-scale models and found that urban maximum temperatures decreased by 0.6 °C. In a recent review of green roof and cool roof studies, Santamouris (2014) concluded that an increase in the albedo of roofing materials can lead to decreasing ambient urban air temperatures with an average reduction rate of ≈ 2 °C per 0.1 increase in roof albedo. The same paper points out that the effectiveness of green roofs in reducing air temperatures is more variable and also depends on the height of the buildings. One variable influencing green roof benefits is its vegetation where the conventional approach is to install a relatively uniform palette of plants possessing low height, spreading growth habit, and shallow rooting depths, such as species from the Sedum genus (Dunnett and Kingsbury, 2004; Snodgrass and Snodgrass, 2006). Sedums, the most common green roof plant, show strong adaptability to roof conditions where they tolerate the harsh environmental conditions of cold, heat and drought. In temperate regions Sedums have been shown to outperform North American natives in shallow roof soils (Monterusso et al., 2005). However, these species have not been proven adaptable in all geographic locations. Sedums common to the nursery trade, may be unfit for hot dryland locations. Although some members of the genus can adapt to water and temperature changes by changing their photosynthetic pathways, many members of Sedum genus possess limitations in high temperatures (Williams et al., 2010). Additionally, Dvorak and Volder (2013a) argue that using exotic sedums exclusively ignores ecoregional context and could put new roofs at risk of failure or operation at underperforming levels. Consequently, an alternative approach gaining interest is the use of the prairie vegetation as analog for vegetative roof plant community. A review of nearly two dozen green roofs deduced that the prairie biome is highly adaptive, offers greater diversity and taps context for green roofs in new locations (Sutton et al., 2012). One of the challenges is achieving vegetative community stability on new roofs. Vegetation is often installed as plugs, but Sedums are often propagated by sowing leaf cuttings or installed as pregrown mats. Mats are optimal for coverage but lack diversity. Meanwhile, native plants are most commonly plugged, but can also be seeded (Brenneisen, 2005; Sutton, 2013). Because Sedums tend to behave as stress tolerators with slower growth, it is possible that they could complement the rapid establishment of pioneering prairie species. Sedums have nursed native North American wildflowers during establishment in roof settings (Butler and Orians, 2009). In addition, it has been suggested that planting multiple species that possess varying growth forms could optimize water loss and roof surface cooling (Wolf and Lundholm, 2008). Thus, increasing plant diversity at installation could positively impact plant coverage and community stability, while secondarily improving temperature and hydrologic benefits. As several studies suggest, questions remain concerning the performance of green roof systems, particularly for green roofing systems with low-lying plants and under variable climatic conditions. In the Central Plains of the United States, climatic conditions vary widely during the different seasons including periods of extreme temperatures and drought in the summertime. In the spring and fall thunderstorms can lead to intensive rainfall events. Ambient wind speeds typically remain quite high and pose further challenges for green roof maintenance and performance. Obtaining further knowledge about the feasibility and environmental impacts of green roof systems under such conditions motivated this research study. The location, at the University of Oklahoma in Norman, Oklahoma is particularly interesting as Oklahoma has strong north–south oriented temperature gradients while the moisture gradients are in east–west direction. Thus, the study in Norman, which is located in the central part of the state, allows assessing the cooling efficiency of green roofs for a wide range of thermodynamic environments.

84

P.M. Klein, R. Coffman / Science of the Total Environment 512–513 (2015) 82–93

2. Materials, methods and timing In Spring 2010, an experimental green roof was installed on the roof observation deck of the National Weather Center (NWC) in Norman, Oklahoma (Figs. 1 and 2). The green roof consisted of 160, 61 × 122 cm (2 ft × 4 ft), individual trays covering 116.1 m2 (1250 ft2) of deck area. Each tray was delivered with identical growing media, stored for less than 24 h and moved to the roof via freight elevator. Once the entire paddock was completed with the help of students (Fig. 1a), the trays were both secured to each other and to the metal edger at the perimeter of the paddock (Fig. 1b). For vegetation studies, the trays were divided equally between 10.1 cm (4 in.) and 20.2 cm (8 in.) depths and partitioned into experimental treatments of installation type: 32 trays each of plugged, seeded, and plugged and seeded. Plugged treatments were installed with eight 21/4 in. plugs of succulents two months prior to installation. Seeding was applied one month after installation to the designated treatments at the same time period and at the recommended commercial rates for perennial grasses and wildflowers. Individual plugs and trays were randomized. All treatments were typically irrigated 3 times a week by hand in addition to natural rainfall. The amount of water applied during each irrigation equates to ≈0.2–0.3 in. of rainfall (Fig. 3). Species were recorded in a census study with drop aerial digital photography taken from 1 m above the surface and accessed for area of coverage by species using AutoCAD. By overseeding succulent plugs with a greater diversity of grasses and forbs a maximal heterogeneity (N 30) could be achieved in the plant community. It was anticipated that seeding a greater diversity of grasses and forbs would lead to higher species richness in that category. The large loss of plugged succulent species and the low rate of seed germination within the study period compromised the comparison of experimental treatments. Continuous data was collected for time series analysis in order to understand the relationships between species (Y) and time (t) for the overall roof. All trays were recorded for species presence and cover was classified by type of vegetative growth: succulent, grass,

herbaceous, and volunteers, those species not installed or seeded but observed. Species richness was recorded in census method of all combined trays to describe the collective plant community and grouped by vegetative growth classifications: succulents, grasses and herbaceous, and volunteers. Species richness by vegetative group was then examined for trend patterning. For atmospheric studies, two, 10-ft tall, meteorological stations monitored the effects of the NWC green roof on the surface energy balance. One of the stations was located above the green-roof area (Fig. 2a) while a second control site was installed ≈10 m north of the green roof area over the original concrete roof tiles (Fig. 2b). A list of the sensors installed and the variables measured on both stations is given in Table 1. Data collection began in October 2010 but remained irregular until Spring 2011. Data were collected and stored as 5-min averages on a total of 4 data loggers from where they were manually downloaded on a regular basis. The raw data sets were processed to identify possible outliers and to calculate hourly averages and daily statistics, which are further discussed in Section 3.2. The atmospheric study period chosen for this paper ran 159 days (April, 12 until October, 11 2011 months 12–18 of the vegetation studies), whereby data are missing for May, 24–June, 14, and June, 25–26. The data sets thus include measurements for a total of 159 days. The microclimate observed at the two roof stations is discussed in Section 3. To document the regional-scale weather conditions during the study period, data from the Oklahoma Mesonet site in Norman Oklahoma were analyzed. The Norman Mesonet site is located in a grassland area about 8 km north of the NWC (more detailed information about the characteristics of the Norman Mesonet site including panoramic photos, aerial photos, and topographic maps can be found at http://www. mesonet.org/index.php/sites/site_description/nrmn). The analysis of the Norman Mesonet data confirmed that drought conditions with very low rainfall rates and extreme heat prevailed in central Oklahoma during the study period (Table 2). On 55 days the daily maximum temperature exceeded 100 °F (37.78 °C) and 132 days were without

Fig. 1. Photos of (a) installation, (b) initial setup in May 2010, and (c) plant coverage in Spring 2012.

P.M. Klein, R. Coffman / Science of the Total Environment 512–513 (2015) 82–93

85

Fig. 2. Google Earth satellite image of the NWC roof observation deck and photos of the (a) green-roof and (b) control meteorological stations. Their respective locations on the roof are marked on the satellite image. The area covered by the green roof is clearly visible in the satellite image as dark shaded rectangular.

measureable rainfall (Fig. 3). Winds prevailed from the south with average daily wind speeds ranging between 1.8–11.0 m/s. These extreme conditions posed several challenges for the plant establishment and performance of the Experimental Green Roof. In order to convey the conditions the vegetation study period chosen for the paper ran 791 days (April 30, 2010 until June 30, 2012) and is discussed in more detail in Section 3. 3. Results 3.1. Vegetation studies The intent of the vegetative study was to maximize surface coverage through complementary plant types, while examining the establishment of both sedums (the workhorse of temperate roofs) and prairie natives. Of the thirty two plant species originally installed, ten species were present after six months (Table 3). Those numbers increased to only eleven species by month twenty six with the majority of species being grasses and herbs. The low total number of established species was due to both limited germination of the

short grass prairie seed and the concurrent death of several sedums species (Fig. 4). In the succulents 4 of 8 species were present (50%) and grasses and herbs 9 of 24 (38%). Due in part to higher installed species richness, native prairie exceeded exotic succulent richness by month eighteen and continued their trends to the end of the study. At about that same time volunteer “weed” species emerged, as well, colonizing the bare substrate areas left by high sedum mortality. The most abundant species in the community at month eighteen were Bouteloua gracilis (57%), Delosperma cooperi (15%), Euphorbia maculate (6%), Portulaca maculate (8%), Sedum spurium ‘Voodoo’ (8%), Oxytropis lambertii (3%), and Dalea purpurea (2%). Only small changes in this structure occurred by month twenty six, mainly that B. gracilis was becoming slightly more dominant. The effect of low germination and high succulent mortality created an extremely sparse vegetative coverage. For the first fifteen months over 82% of roof surface area remained exposed bare substrate (Fig. 5). The plastic edging around the trays accounted for 6% of the area. During this time succulents; Sedum kamtschaticum; Sedum alba ‘Coral Carpet’, Sedum rupestre Angelina and the xeric plant Santolina chamaecyparissus

86

P.M. Klein, R. Coffman / Science of the Total Environment 512–513 (2015) 82–93

Fig. 3. Rainfall amounts observed at the Norman Mesonet site (blue diamonds) and equivalent rainfall amounts due to the irrigation of the green roof (red circles) during the 2011 study period. The green line shows the 2-d average total rainfall amount.

died. D. cooperi, Sedum tetractinum ‘Vera Jameson’, Sedum mexicanum and S. spurium ‘Voodoo’ survived and expanded coverage, but only slightly. Succulent advancement was accounted for by the growth of one plant, D. cooperi, which expanded coverage incrementally. Volunteers colonized within a year and covered almost 4.5% of the roof by month 26. Wildflowers were seasonal and more frequent in the early summer. Observed wildflowers included; Thelesperma filifolium, Erysimum capitatum, Coreopsis tinctoria, Liatris punctata, Argemone polyanthemos and Ratibida columnifera. 3.2. Atmospheric studies The atmospheric studies focused on identifying impacts of the green roof on the radiation balance and thermal environment above the two different roof surfaces. Even though the area covered by the green roof containers was limited and close to the control station over the concrete roof, differences in air temperatures and relative humidity were observed. The monthly averages of daily maximum and minimum temperatures at 1.5 m were both lower over the green roof than over the Table 1 Sensors installed at two meteorological observation stations on the roof of the National Weather Center in Norman, OK. Level

Sensor

Variables measured

Top level (≈3.0 m)

03002 R.M. Young Wind Sentry Set (cup anemometer and wind vane) CSI CSAT Sonic anemometer

Horizontal wind speed and wind direction All three components of the wind vector plus sonic temperature Air temperature and relative humidity Incoming solar radiation All three components of the wind vector plus sonic temperature Air temperature and relative humidity Net radiation Incoming shortwave radiation Outgoing shortwave radiation Incoming longwave radiation Outgoing longwave radiation

Vaisala HMP35-C Temp/RH probe

Lower level (≈1.5 m)

LICOR Pyranometer CSI CSAT Sonic anemometer

Vaisala HMP35-C Temp/RH probe NR-LITE Net-radiometer CNR1 radiometer (only on the control station over the concrete roof)

concrete roof (Fig. 6a,b), while relative humidity values at the same sampling height showed the opposite trends (Fig. 6c,d). The differences of the monthly trends were however small and within the range of the measurement uncertainties for the sensors used. Histograms of the differences between the green and concrete roof in daily maximum and minimum temperature (Fig. 7a) and relative humidity (Fig. 7b) values show that the reduction of maximum air temperatures can exceed 1 °C (on ≈ 14% of the days) and the increase in minimum relative humidity can be higher than 2% (on ≈25% of the days), while increases/decreases in daily maximum/minimum temperature/relative humidity over the green roof were only rarely observed and of much smaller magnitude. The radiation balance over a surface can be expressed as


R ¼ SD þ SU þ LD þ LU;

ð1Þ

whereby SD is the incoming shortwave, SU is the outgoing shortwave, LD is the incoming longwave radiation, and LU is the outgoing longwave radiation. Positive/negative values are assigned to radiation terms that are directed towards/away from the surface. Using the definition of the albedo α of a surface Eq. (1) can be recast into


R ¼ ð1−α ÞSD þ LD þ LU:

ð2Þ

Table 2 Summary of climatic conditions during the study period covering 159 days during April– October 2011. Results are based on observations at the Oklahoma Mesonet site in Norman, OK. Characteristic values

Value

Highest recorded daily maximum air temperature Tmax Highest recorded daily minimum air temperature Tmin Number of days with daily maximum air temperature Tmax N 37.78 °C (100 °F) Number of days with daily minimum air temperature Tmin N 25.00 °C (77 °F) Total rainfall rate Rtot (sum for 159 days) Number of days without measurable rainfall Number of days with daily rainfall rate Rdaily N 1 in.

43.30 °C 29.77 °C 55 (34.59%) 46 (28.93%) 17.95 in. 132 (83.54%) 6 (3.77%)

P.M. Klein, R. Coffman / Science of the Total Environment 512–513 (2015) 82–93 Table 3 List of installed vegetation by plant type and species with percentage of seed mix and potted quantities (n = 32). Short grass prairie wildflowers (n = 18) Scientific name

Common name

Abronia fragrans Argemone polyanthemos Castilleja integra Chrysopsis villosa Coreopsis tinctoria Dalea candida Dalea purpurea Erysimum capitatum Liatris punctata Linum lewisii Oenothera caespitosa Oxytropis lambertii Penstemon angustifolius Penstemon grandiflorus Penstemon secundiflorus Ratibida columnifera Thelesperma filifolium Thermopsis rhombifolia

Snowball Sand Verbena Prickly Poppy Indian Paintbrush Silky Golden Aster Plains Coreopsis White Prairie Clover Purple Prairie Clover Wallflower Dotted Gayfeather Blue Flax Stemless Even. Primrose Showy Locoweed Pagoda Penstemon Shell Leaf Penstemon Orchid Beardtongue Prairie Coneflower Green Threadleaf Arroyo Golden Banner

Percentage 0.00% 1.99% 0.16% 0.10% 9.92% 7.97% 7.85% 0.97% 2.81% 9.79% 0.24% 2.97% 9.24% 9.83% 0.00% 9.75% 7.69% 9.99%

Short grass prairie grasses (n = 6) Scientific name

Common name

Percentage

Bouteloua gracilis Buchloe dactyloides Hilaria jamesii Pascopyrum smithii Sporobolus airoides Sporobolus cryptandrus

Blue Grama Buffalo Grass Galleta Western Wheatgrass Alkali Sacaton Sand Dropseed

21.22% 39.34% 2.29% 28.79% 1.47% 1.48%

temperatures Ts and surface emissivities ε. The albedo effect is only relevant during the day. Differences in Ts and ε play a role during day and night, but are most prominent at night when the shortwave radiation terms in Eq. (1) are both zero and the radiation balance reduces to


4

R ¼ LD þ LU ¼ LD−ε  σ  T s ;

Common name

Quantity

Delosperma cooperi Santolina chamaecyparissus Sedum kamtschaticum Sedum mexicanum Sedum rupestre ‘Angelina’ Sedum tetractinum ‘Vera Jameson’ Sedum alba ‘Coral Carpet’ Sedum spurium ‘Voodoo’

Hardy Ice Plant Gray Santolina Russian Stonecrop Mexican Sedum Angelina Stonecrop Vera Jameson Sedum Coral Carpet Sedum Voodoo Sedum

96 96 96 96 96 96 96 96

Over the concrete roof all four components of the radiation balance were directly measured with a high-quality, expensive CNR1 radiometer and net radiation was also monitored with a cheaper NRlite net radiometer. Due to budget constraints, only net radiation R⁎ was measured with two NRlite net radiometers over the green roof. Changes in the radiation balance due to the green roof installation must thus be deduced from any observed differences in net radiation R⁎. A comparison of monthly averaged values of daily R⁎ maxima (Fig. 8) shows that the values were between ≈50 and 93 W m−2 higher over the green roof than over the concrete roof. Scatter plots between the daily maximum (Fig. 9a) R⁎ values from the four different sensors installed highlight that the values from the CNR1 and NRLite sensors at the control station agree very well, while the maxima values at the green roof site (average of the 2 NRLite sensors) tend to be 17% higher. An opposite trend can be noted when comparing the minimum R⁎ values (Fig. 9b) with nearly 25% lower absolute values measured by the NRLite sensors at the green roof station than with the CNR1 sensor at the control station. For the minimum R⁎ values, which were observed at night, the comparison of the NRLite and CNR1 values at the control station shows however also a systematic bias with ≈10% lower absolute values measured by the NRLite sensor. Such bias could potentially also contribute to the lower absolute values observed at the green roof station but it likely does not explain the 25% reduction. Differences in net radiation R⁎ at the two stations can be due to differences in the albedo which affect the net shortwave radiation and/or differences in net longwave radiation, caused by variations in surface

ð3Þ

where σ is the Stefan–Boltzmann constant. Incoming longwave radiation LD depends primarily on the temperature of the atmosphere and can be assumed to have the same value over both measurement stations. The values of the surface emissivities ε for both the green and concrete roof are not really known and using tabulated values for ε published in the literature could lead to high uncertainties in the calculated surface temperatures. We thus decided not to provide values for Ts. However, to account for possible differences in the net longwave radiation when computing the albedo α, the ratio a of outgoing longwave radiation LUgr at the green roof station to its counterpart LUco at the control station was estimated using the net radiation values at night: a¼

LU gr R
gr −LD ¼ : LU co LU co

ð4Þ

⁎ shown in Fig. 8b, reUsing the minimum net radiation values Rgr,min sults in an average value of ā = 0.96, while ā ≅ 1.00 is computed if all ⁎ are used in Eq. (4). The differences in outnighttime hourly values of Rgr going longwave radiation at the two measurement sites are thus quite small, and the average albedo of the green roof can be estimated as α gr ¼ 1−

Succulents (n = 8) Scientific name

87

R
gr −LD−aLU co ≈0:18−0:21; SD

ð5Þ

for the range of estimated ā-values. These values are lower than the average, measured albedo for the concrete roof α co ¼

  SU co ≈0:28: SD

ð6Þ

The surface energy balance of the roof surface is described by


R ¼ HS þ HL þ HG ;

ð7Þ

where HS is the sensible heat flux, HL, the latent heat flux, and HG the ground heat flux. The first two terms (HS, HL) describe the energy fluxes to/from the atmosphere, the last term (HG) to/from the roof material. Latent and ground heat fluxes were not measured during the study period, but sensible heat fluxes can be assessed using the sonic anemometer measurements. From these measurements, buoyancy fluxes H (¼ ρcp w0 T v 0 ), can be computed as covariances between the vertical velocity components w and the sonic temperatures Ts that are nearly identical to virtual temperatures Tv. To obtain true sensible heat fluxes   H S ¼ ρcp w0 T 0 , the buoyancy fluxes would need to be corrected but such corrections require accurate, high-resolution moisture measurements, which were not available (Liu et al., 2001). We thus used the measured buoyancy fluxes to investigate how the green roof alters the surface energy balance. The complex geometry of the roof, which has several elevated rooftop structures in close vicinity of the meteorological stations (partially seen in Figs. 1 and 2), further complicates the analysis of the surface energy balance. The elevated roof structures cause very complex flow patterns, which in turn result in high spatial variability of the turbulent fluxes, independent of the surface parameters. The data were thus carefully screened before analyzing the heat flux observations. Only hourly records for which the wind direction at the 3-m level of the concrete site was between 200°–250° were included in the analysis. This wind direction sector was identified based on the geometry of the roof (there are no elevated structures upstream and the concrete roof mast is not

88

P.M. Klein, R. Coffman / Science of the Total Environment 512–513 (2015) 82–93

Fig. 4. Vegetative community change over 26 months by type; succulent (blue), grasses and herbs (green) and volunteers (red).

directly downwind of the green roof mast) and the siting of the instruments relative to the masts (the sonic anemometers were mounted facing south, i.e. for northerly wind directions wake effects of the masts may affect the data quality). Additionally, wind speeds and wind directions observed at both masts were compared. For the chosen sector, the observations at the two sites agreed well, while clear differences in the flow patterns were noted for other wind directions. The daily maximum buoyancy flux values at the green-roof site measured at 1.5 m (LL) and 3 m (TL) above the surface are both lower and higher than at the corresponding measurement levels above the

concrete roof (Fig. 10a). Despite the large scatter, a trend can be noted that the green roof observations are lower during days with high buoyancy fluxes. This trend can be more clearly seen if the ratio of buoyancy flux to net radiation H/R⁎, which describes how much of the available energy is partitioned into buoyancy, is plotted (Fig. 10b). On average, the buoyancy flux over the green roof at the 1.5 m measurement level accounts for 26% of the available net radiation while over the concrete roof at the same measurement height 46% of the available energy is partitioned into buoyancy (Table 4). Depending on atmospheric stability and moisture content, the buoyancy fluxes may overestimate the

Fig. 5. Surface coverage of materials showing a decreasing area of bare soils and increasing native grass cover during the initial 26 month establishment period.

P.M. Klein, R. Coffman / Science of the Total Environment 512–513 (2015) 82–93

89

Fig. 6. Comparison of monthly averages of daily (a, c) maximum and (b, d) minimum (a, b) temperature and (c, d) relative humidity measured over the green and concrete roof at a height of 1.5 m above the respective roof surface.

sensible heat fluxes by 10–30% (Liu et al., 2001). A higher moisture contribution at the green roof due to higher ET rates would cause the differences between sensible heat fluxes at the two sites to be even more pronounced than the reported differences in the buoyancy fluxes. The lack of latent and ground heat flux measurements prevents a detailed analysis of the surface energy balance for the two roof surfaces, but

the data provide evidence that higher ET rates at the green roof compensate for the higher net radiation such that small cooling effects can be noted in the lowest 2 m above the green roof (Table 4). The average statistics of various meteorological parameters computed using the records for the chosen wind direction sector (Table 4) confirm the general trends seen for the entire data set: at the lower measurement height,

Fig. 7. Histograms of (a) air temperature (in °C) and (b) relative humidity differences (in %) between the green and concrete roof stations during the 2011 study period.

90

P.M. Klein, R. Coffman / Science of the Total Environment 512–513 (2015) 82–93

Fig. 8. Monthly averages of daily maximum incoming solar and net radiation in W m−2. The net radiation values were measured over the concrete roof (CNR1 and NRLite sensors) and over the green roof (2 NRLite sensors). Differences in net radiation over the green vs. concrete roof are also plotted.

daily maximum temperature/relative humidity are slightly lower/ higher over the green roof than over the concrete roof while net radiation is ≈80 W m−2 higher over the green roof. 4. Discussion The overseeding of succulents with wildflowers and grasses was aimed at increasing survivorship, vegetative cover, and cooling effect. The success of prairie species is not surprising. However, the extremely high mortality of succulents was unexpected as roofs containing a mix of succulents, grasses and wildflowers are fairly common and typically employed to produce a broader array of ecological services. Butler and Orians (2009) found that Sedum album, a succulent, nursed wildflower plants during establishment. That result was not necessarily observed here as Sedums struggled to survive. However, prairie species richness surpassed succulents. Furthermore, it was anticipated that planting multiple species possessing varying growth forms could help regulate water loss and optimize roof surface cooling as Wolf and Lundholm (2008) suggest. Both the difficulties of establishment for proven plants and the effect on cooling benefits suggest the importance of plant selection and installation method. One contributing factor to plant mortality may be the physiological limitations of the succulents. Dvorak and Volder (2010) and Williams et al. (2010) recommend that planted roofs reflect their ecoregional context and argue that exotic succulents, such as temperate Sedums may experience limited fit in hot dryland locations. Like most succulents, Sedums rely on Crassulacean Acid Metabolism (CAM) photosynthesis for water-use efficiency, however, some species can alter their photosynthetic pathway. Williams et al. (2010) explain that some members of the genus can adapt to water and temperature changes by changing their photosynthetic pathway but many Sedums are limited in that regard. These findings indicate that several Sedum species, common to the trade, are unfit for warm climate green roofs. Although, S. kamtschaticum and S. alba ‘Coral Carpet’ both proved unfit for establishment, the limited survival of S. tetractinum ‘Vera Jamison’, S. mexicanum and S. spurium ‘Voodoo’ suggests that Sedum species differ

in their tolerance to extreme establishment conditions. These findings can be explained in part by Starry et al. (2014) who found, in a controlled drought experiment, that S. kamtschaticum exhibited C3 photosynthesis, CAM cycling, and greater water use efficiency than S. alba and concluded that Sedum species should be treated independently in green roof design. To understand physiological limitations more studies are needed comparing Sedums to grasses and herbs at the group and species level. It should also be further investigated how establishment affects community composition and its influence on the community's collective water efficiency. Irrigation of the roof could have favored prairie species and adaptive succulents, while compromising Sedum establishment. Observations of Sedum mortality in warm climates due to overwatering have occurred near a study site in Texas (Hopman, personal communication). Similar to these findings, Hopman (2014) found high mortality in the establishment of Sedum acre, S. reflexium ‘Blue Spruce’, S. mexicanum and S. ‘Vera Jameson’ in irrigated plots. Equally, Dvorak and Volder (2013a) found D. cooperi to be highly drought resistant in survivorship studies without irrigation. D. cooperi were resilient to conditions in which they resprouted after assumed mortality. S. kamtschaticum and S. moranense survived to lesser degrees in those same drought conditions. This reflects an interest in the prairie ecosystem as highly robust for the dynamic environments of a rooftop proposed by Sutton et al. (2012). Yet, our germination from seed was marginal, Sutton (2013) has shown that with improved techniques using drill seeding the 24 month 80% coverage threshold set by the FLL (Forschungsgesellschaft Landschaftsentwicklung Landschaftsbau) could be met. He found a cultivar of B. gracilis, ‘Bad River’ to be the leading performer of grasses. This indicates that B. gracilis and its cultivars are robust germinators, while other grasses including Buchloe dactyloides may be more sensitive to seeding method. Other prairie species. O. lambertii, D. purpurea, T. filifolium, E. capitatum, C. tinctoria, L. punctata, A. polyanthemos and R. columnifera were present but did not constitute much of the community. Although Ksiazek et al. (2014) have shown in a lab setting that flowering roof natives germinate at similar or higher rates to ground level plants, it is important to know what plants should constitute green

P.M. Klein, R. Coffman / Science of the Total Environment 512–513 (2015) 82–93

91

Fig. 9. Scatter plots between daily (a) maximum and (b) minimum net radiation values measured over the green and concrete roofs.

roof seed mixes and what the best protocols for establishing flowering herbaceous plants and grasses on a roof are. The atmospheric studies confirmed that green roof systems can have a lower albedo than light colored roof materials. The albedo values estimated for the green roof are slightly lower than the ones found in other studies (Lazzarin et al., 2005) which can be explained by the poor vegetation cover and dark substrate used in the green roof containers. It will be interesting to see how these results change once better plant coverage is established. However, the albedo of traditional, dark roofing materials can be as low as 0.1, i.e. the green-roof albedo observed in our study was still nearly twice as high (Lazzarin et al., 2005; Santamouris, 2014). The observations also show that the albedo of the light-colored concrete roof tiles is less than 0.3, which is lower than the value of 0.37 observed for concrete roofs by Takebayashi and Moriyama (2007). Albedo values of 0.7 or higher, which have been used in modeling studies investigating different heat island mitigation strategies (e.g.

Li et al., 2014), can be realized with highly reflective white membranes. However, such high albedo values can cause issues of glare at neighbor buildings (Akbari et al., 2001; Synnefa et al., 2008) and the longterm performance of such roofs is still investigated (Gaffin et al., 2012; Sleiman et al., 2011, 2014). As the materials weather and pollution settles on the membranes, the albedo is often reduced, but cool-roofing performance standards that require a three-year aged albedo above 0.50 have been met (Gaffin et al., 2012). Despite the lower albedo, air temperature/relative humidity tended to be lower/higher at the green-roof station. However, the differences between the green roof and control station are quite small and limited to the lowest couple of meters above the roof as it was also reported in previous experimental studies (Wong et al., 2003). The reduction in buoyancy fluxes indicates that some cooling effects may be attributed to higher ET rates (Lazzarin et al., 2005). Grass and wildflower species have higher ET rates than succulents, which can positively impact

92

P.M. Klein, R. Coffman / Science of the Total Environment 512–513 (2015) 82–93

et al., 2010). It is possible that B. gracilis was particularly effective in soil moisture capture, which could be due to advantageous root development and its ability to cycle in and out of dormancy under drought stress during the growing season (Sala and Lauenroth, 1982). ET rates can differ by green roof species in controlled settings (Ouldboukhitine et al., 2014b) and directing planting selection, establishment, and irrigation may optimize thermal performance. The concrete roof tiles dried off quickly after a rain event and were completely dry for most of the study period. The green roof was irrigated two-three times per week uniformly by hand wand unless the substrate was saturated by N1/4 in. rain event in the 2 days prior (Fig. 3). The irrigation periods included 04/12/10 to 12/03/10 and 03/03/11 to 12/04/11 and resumed 05/15/12. The supplemental irrigation provided an additional equivalent 17 in. of rain during the 2011 study period, which also contributed to the evaporative cooling effect of the green roof. However, given that the differences in the recorded temperatures were in the range of the instrument accuracies it remains unclear if the studied green roof had a significant effect on the air temperatures.

5. Conclusions and outlook

Fig. 10. Scatter plots between daily maximum (a) buoyancy flux H and (b) ratio of buoyancy flux to net radiation H/R⁎ at the green and concrete roof measurement sites. Only hourly records for which the wind direction at the 3-m level (TL) of the concrete site was between 200°–250° were included when computing the daily statistics.

cooling. The presence of B. gracilis, may explain some cooling effect. In drought stressed conditions, Kim (1983) reported ET rates for B. gracilis of 5.69 mm/day (1.57 in./wk). In green roofs, S. mexicanum has recorded more than twice lower rates less of 2.19 mm/day (Voyde

Table 4 Comparison of daily mean values of wind speed WS and wind direction WD, and daily maximum values of air temperature T, relative humidity RH, buoyancy flux H, net radiation R⁎, and ratio of buoyancy flux to net radiation H/R⁎ at the lower (LL) and top (TL) levels of the two meteorological stations. Only hourly records for which the wind direction at the TL of the concrete site was between 200°–250° were included when computing the daily statistics. Variable (unit)

Roof site

LL: 1.5 m

TL: 3 m

WSmean (m s−1) WDmean (°) Tmax (°C) RHmax (%) Hmax (W m−2) Rmax⁎ (W m−2) (H/R⁎)max (–)

Concrete Green Concrete Green Concrete Green Concrete Green Concrete Green Concrete Green Concrete Green

N/A N/A N/A N/A 33.38 32.84 51.00 51.93 199.62 102.85 457.37 543.77 0.48 0.26

1.93 1.71 222.96 242.61 33.46 33.22 52.00 53.49 207.87 152.29 N/A N/A 0.48 0.46

Both native prairie and non-native succulent vegetation can be established under extreme conditions but their establishment is species specific and high mortality is likely. B. gracilis and D. cooperi showed good survivability and adaptability in harsh climatic conditions. On the other hand, mortality was high in many Sedum plugs and all Sedums struggled to establish and cover. With this in mind the broad application Sedums should be avoided in warm climates and employed with caution when designing green roofs for climatic resiliency. Future investigations of Sedum should be aimed at species specific success. Sowing seeds is a viable alternative to plugs for establishing vegetative cover in roofs. Establishment of native prairie plants from seed was proven for some species, but germination was low in many others. Nine prairie species sown from seed were observed as germinating and growing into plants from the 24 seeded species. B. gracilis was the dominate species. An enhanced understanding of seeding techniques for geographic location would improve germination in other species. Despite the extreme drought conditions and poor vegetation coverage, air temperatures and buoyancy fluxes tended to be lower over the green roof. Thus, higher values in net radiation, caused by a reduction in albedo relative to the original, light-colored roof, were compensated by higher ET rates of the vegetated roof. The albedo of the green roof was still about twice higher than the albedos found for dark roofing surfaces and the albedo of the gray, concrete roof tiles was significantly lower than the albedo often used when simulating the effects of cool roofs. Albedo values as high as 0.7 may be hard to maintain and the evaporational cooling effects of vegetated roofs may be more beneficial than mitigation strategies that focus on increasing albedo only. So far, the observed temperature differences were small and within the range of the instrument uncertainties. We expect to see stronger effects once the vegetation coverage is higher. To improve the performance of the green roof system, drip irrigation system was installed in 2013, the substrate of the green roof containers was replaced and different plants and seeds were used to re-establish vegetation. We plan to report on the results from this second phase of the project in the future.

Acknowledgments The NWC Experimental Green Roof was supported by project ORF09-0031-CW of the Oklahoma Water Resources Board and Oklahoma Conservation Commission. The first author was also supported through the NSF Career award ILREUM (NSF ATM 0547882). The authors would also like to thank Lee Fithian, Jason Vogel, and Thomas Woodfin for their input and collaboration in establishing the experimental site.

P.M. Klein, R. Coffman / Science of the Total Environment 512–513 (2015) 82–93

References Akbari, H., Pomerantz, M., Taha, H., 2001. Cool surfaces and shade trees to reduce energy use and improve air quality in urban areas. Sol. Energy 70, 295–310. Alexandri, E., Jones, P., 2007. Developing a one-dimensional heat and mass transfer algorithm for describing the effect of green roofs on the built environment: comparison with experimental results. Build. Environ. 42, 2835–2849. Arnfield, A.J., 2003. Two decades of urban climate research: a review of turbulence, exchanges of energy and water, and the urban heat island. Int. J. Climatol. 23, 1–26. Baik, J.-J., Kwak, K.-H., Park, S.-B., Ryu, Y.-H., 2012. Effects of building roof greening on air quality in street canyons. Atmos. Environ. 61, 48–55. Berardi, U., GhaffarianHoseini, A., 2014. State-of-the-art analysis of the environmental benefits of green roofs. Appl. Energy 115, 411–428. Berndtsson, J.C., 2010. Green roof performance towards management of runoff water quantity and quality: a review. Ecol. Eng. 36, 351–360. Bowler, D.E., Buyung-Ali, L., Knight, T.M., Pullin, A.S., 2010. Urban greening to cool towns and cities: a systematic review of the empirical evidence. Landsc. Urban Plan. 97, 147–155. Brenneisen, S., 2005. Green roofs recapturing urban spaces for wildlife—a challenge for urban planning and environmental education. Greening Rooftops for Sustainable Communities, Washington D.C. Butler, C., Orians, C., 2009. Sedum facilitates the growth of neighboring plants of green roof under water limited conditions. Greening Rooftops for Sustainable Communities, Atlanta. Coutts, A.M., Daly, E., Beringer, J., Tapper, N.J., 2013. Assessing practical measures to reduce urban heat: green and cool roofs. Build. Environ. 70, 266–276. D'Orazio, M., Di Perna, C., Di Giuseppe, E., 2012. Green roof yearly performance: a case study in a highly insulated building under temperate climate. Energy Build. 55, 439–451. Dunnett, N., Kingsbury, N., 2004. Planting Green Roofs and Living Walls. Timber Press, Portland, OR. Dvorak, B., Volder, A., 2010. Green roof vegetation for North American ecoregions: a literature review. Landsc. Urban Plan. 96, 197–213. Dvorak, B., Volder, A., 2013a. Plant establishment on unirrigated green roof modules in a subtropical climate. AoB Plants 5. Dvorak, B., Volder, A., 2013b. Rooftop temperature reduction from unirrigated modular green roofs in south-central Texas. Urban For. Urban Green. 12, 28–35. Farrell, C., Mitchell, R.E., Szota, C., Rayner, J.P., Williams, N.S.G., 2012. Green roofs for hot and dry climates: Interacting effects of plant water use, succulence and substrate. Ecol. Eng. 49, 270–276. Fioretti, R., Palla, A., Lanza, L.G., Principi, P., 2010. Green roof energy and water related performance in the Mediterranean climate. Build. Environ. 45, 1890–1904. Gaffin, S.R., Khanbilvardi, R., Rosenzweig, C., 2009. Development of a green roof environmental monitoring and meteorological network in New York City. Sensors 9, 2647–2660. Gaffin, S., Imhoff, M., Rosenzweig, C., Khanbilvardi, R., Pasqualini, A., Kong, A., et al., 2012. Bright is the new black-multi-year performance of high-albedo roofs in an urban climate. Environ. Res. Lett. 7. Gago, E.J., Roldan, J., Pacheco-Torres, R., Ordonez, J., 2013. The city and urban heat islands: a review of strategies to mitigate adverse effects. Renew. Sustain. Energy Rev. 25, 749–758. Grimmond, C.S.B., Roth, M., Oke, T.R., Au, Y.C., Best, M., Betts, R., et al., 2010. Climate and more sustainable cities: climate information for improved planning and management of cities (producers/capabilities perspective). Procedia Environ. Sci. 1, 247–274. Hien, W.N., Yok, T.P., Yu, C., 2007. Study of thermal performance of extensive rooftop greenery systems in the tropical climate. Build. Environ. 42, 25–54. Hopman, D., 2014. The University of Texas at Arlington Extensive Green Roof. The University of Texas at Arlington School of Architecture. http://www.uta.edu/sustainability/_ downloads/Green%20Roof%20Report.pdf. Jaffal, I., Ouldboukhitine, S., Belarbi, R., 2012. A comprehensive study of the impact of green roofs on building energy performance. Renew. Energy 43, 157–164. Kim, K.S., 1983. Comparative evapotranspiration rates of thirteen turfgrasses grown under both non-limiting soil moisture and progressive water stress conditions. M.S. Texas A & M, p. 64. Ksiazek, K., Fant, J., Skogen, K., 2014. Native forbs produce high quality seeds on Chicago green roofs. Journal of Living Architecture 1, 21–33. Lazzarin, R., Castellotti, F., Busato, F., 2005. Experimental measurements and numerical modelling of a green roof. Energy Build. 37, 1260–1267. Li, D., Bou-Zeid, E., Oppenheimer, M., 2014. The effectiveness of cool and green roofs as urban heat island mitigation strategies. Environ. Res. Lett. 9. Lin, B.-S., Yu, C.-C., Su, A.-T., Lin, Y.-J., 2013. Impact of climatic conditions on the thermal effectiveness of an extensive green roof. Build. Environ. 67, 26–33. Liu, H., Peters, G., Foken, T., 2001. New equations for sonic temperature variance and buoyancy heat flux with an omnidirectional sonic anemometer. Bound.-Layer Meteorol. 100, 459–468.

93

Monterusso, M.A., Rowe, D.B., Rugh, C.L., 2005. Establishment and persistence of Sedum spp. and native taxa for green roof applications. Hortscience 40, 391–396. Oberndorfer, E., Lundholm, J., Bass, B., Coffman, R.R., Doshi, H., Dunnett, N., et al., 2007. Green roofs as urban ecosystems: ecological structures, functions, and services. Bioscience 57, 823–833. Oke, T.R., 1982. The energetic basis of the urban heat-island. Q. J. R. Meteorol. Soc. 108, 1–24. Oleson, K.W., Bonan, G.B., Feddema, J., 2010. Effects of white roofs on urban temperature in a global climate model. Geophys. Res. Lett. 37. Olivieri, F., Di Perna, C., D'Orazio, M., Olivieri, L., Neila, J., 2013. Experimental measurements and numerical model for the summer performance assessment of extensive green roofs in a Mediterranean coastal climate. Energy Build. 63, 1–14. Ouldboukhitine, S., Belarbi, R., Djedjig, R., 2012. Characterization of green roof components: measurements of thermal and hydrological properties. Build. Environ. 56, 78–85. Ouldboukhitine, S., Belarbi, R., Sailor, D., 2014a. Experimental and numerical investigation of urban street canyons to evaluate the impact of green roof inside and outside buildings. Appl. Energy 114, 273–282. Ouldboukhitine, S., Spolek, G., Belarbi, R., 2014b. Impact of plants transpiration, grey and clean water irrigation on the thermal resistance of green roofs. Ecol. Eng. 67, 60–66. Patz, J.A., Campbell-Lendrum, D., Holloway, T., Foley, J.A., 2005. Impact of regional climate change on human health. Nature 438, 310–317. Peng, L.L.H., Jim, C.Y., 2013. Green-roof effects on neighborhood microclimate and human thermal sensation. Energies 6, 598–618. Pincetl, S., 2010. Implementing municipal tree planting: Los Angeles million-tree initiative. Environ. Manag. 45, 227–238. Rowe, D.B., 2011. Green roofs as a means of pollution abatement. Environ. Pollut. 159, 2100–2110. Saadatian, O., Sopian, K., Salleh, E., Lim, C.H., Riffat, S., Saadatian, E., et al., 2013. A review of energy aspects of green roofs. Renew. Sustain. Energy Rev. 23, 155–168. Sala, O.E., Lauenroth, W.K., 1982. Small Rainfall Events: An Ecological Role in Semiarid Regions. Oecologia 53, 301–304. Santamouris, M., 2014. Cooling the cities — a review of reflective and green roof mitigation technologies to fight heat island and improve comfort in urban environments. Sol. Energy 103, 682–703. Scherba, A., Sailor, D.J., Rosenstiel, T.N., Wamser, C.C., 2011. Modeling impacts of roof reflectivity, integrated photovoltaic panels and green roof systems on sensible heat flux into the urban environment. Build. Environ. 46, 2542–2551. Sleiman, M., Ban-Weiss, G., Gilbert, H., Francois, D., Berdahl, P., Kirchstetter, T., et al., 2011. Soiling of building envelope surfaces and its effect on solar reflectance—part I: analysis of roofing product databases. Sol. Energy Mater. Sol. Cells 95, 3385–3399. Sleiman, M., Kirchstetter, T., Berdahl, P., Gilbert, H., Quelen, S., Marlot, L., et al., 2014. Soiling of building envelope surfaces and its effect on solar reflectance — part II: development of an accelerated aging method for roofing materials. Sol. Energy Mater. Sol. Cells 122, 271–281. Snodgrass, E., Snodgrass, L., 2006. Green Roof Plants: A Resource and Planting Guide. Timber Press, Portland, OR. Starry, O., Lea-Cox, J.D., Kim, J., van Iersel, M.W., November 2014. Photosynthesis and water use by two Sedum species in green roof substrate. Environ. Exp. Bot. 107, 105–112. http://dx.doi.org/10.1016/j.envexpbot.2014.05.014 ISSN 0098-8472. Susca, T., Gaffin, S.R., Dell'Osso, G.R., 2011. Positive effects of vegetation: urban heat island and green roofs. Environ. Pollut. 159, 2119–2126. Sutton, R., 2013. Seeding green roofs with native grasses. J. Living Archit. 1, 20. Sutton, R.K., Harrington, J.A., Skabelund, L., MacDonagh, P., Coffman, R.R., Koch, G., 2012. Prairie-based green roofs: literature, templates, and analogs. J. Green Build. 7, 143–172. Synnefa, A., Dandou, A., Santamouris, M., Tombrou, M., Soulakellis, N., 2008. On the use of cool materials as a heat island mitigation strategy. J. Appl. Meteorol. Climatol. 47, 2846–2856. Takebayashi, H., Moriyama, M., 2007. Surface heat budget on green roof and high reflection roof for mitigation of urban heat island. Build. Environ. 42, 2971–2979. Voyde, E., Fassman, E., Simcock, R., Wells, J., 2010. Quantifying evapotranspiration rates for New Zealand green roofs. J. Hydrol. Eng. 15, 395–403. Williams, N.S.G., Rayner, J.P., Raynor, K.J., 2010. Green roofs for a wide brown land: opportunities and barriers for rooftop greening in Australia. Urban For. Urban Green. 9, 245–251. Wolf, D., Lundholm, J.T., 2008. Water uptake in green roof microcosms: effects of plant species and water availability. Ecol. Eng. 33, 179–186. Wong, N.H., Tay, S.F., Wong, R., Ong, C.L., Sia, A., 2003. Life cycle cost analysis of rooftop gardens in Singapore. Build. Environ. 38, 499–509. Zinzi, M., Agnoli, S., 2012. Cool and green roofs. An energy and comfort comparison between passive cooling and mitigation urban heat island techniques for residential buildings in the Mediterranean region. Energy Build. 55, 66–76.