Keywords

4.1 Green Roofs as Hydrological Systems

The vast majority of the water that lands on a conventional roof quickly flows off. In sharp contrast, water that lands on a green roof enters a complex hydrological system (Fig. 4.1). Stocks of water are held on and within plants, in substrate, and in various layered materials such as drainage and water retention fabrics. Water exits the system through evaporation from the substrate and plant surfaces, transpiration, and runoff. The magnitude of the various stocks of water as well as the flux of water between stocks and out of the system is governed by complex interactions among green roof components and the physical environment. This complexity makes green roof performance inherently dynamic and contingent on the details of system design and local conditions (Berndtsson 2010).

Fig. 4.1
figure 1

Stocks and flows of water through a typical extensive green roof. The hypothetical scenario depicted is for an extensive green roof that has reached maximum water storage capacity during a spring rainstorm in Corvallis, OR U.S.A. The roof design is assumed to consist of 70 mm of pumice-based substrate, an 8 mm drainage layer, and Sedum sp. vegetation. Stock and flow estimates (in parentheses) are derived from published values for similar systems: precipitation (Schroll et al. 2011a), evaporation and transpiration (Voyde 2011), runoff as a function of % retention (Spolek 2008), substrate water storage capacity (Fassman and Simcock 2012), drainage layer water storage capacity (VanWoert et al. 2005a, Fassman and Simcock 2012), Sedum sp. water storage capacity (Berghage et al. 2007; Fassman and Simcock 2012)

Full size image

Nevertheless, most extensive green roofs share broadly similar hydrological characteristics. The bulk of the standing stock of water on a green roof is held in the substrate. The amount of water intercepted and held by vegetation is comparatively small, at least on Sedum dominated roofs. Many studies have reported no significant difference in stormwater storage between vegetated and un-vegetated (substrate only) roofs, although results vary with storm size and season (Monterusso et al. 2004; VanWoert et al. 2005a; Dunnett et al. 2008; Lundholm et al. 2010; Buccola and Spolek 2011; Starry 2013). While the amount of water captured by most green roof vegetation is small relative to the substrate, differences in plant architecture and vegetation structure have been shown to significantly influence water capture. Roofs planted with grasses and forbs as well as roofs that combine different growth forms capture and retain significantly more water than sedum only roofs (Lundholm et al. 2010; Nagase and Dunnett 2012). Also, some potential green roof plant choices have exceptional water capturing properties. For example, the sponge like physical structure of many mosses allows them to hold 8–10 times their dry weight in water. In contrast to other vegetation types, moss covered roofs can often retain significantly more water than substrate only roofs (Anderson et al. 2010). The stock of water held within plant tissues can also be sizeable for some vegetation types. For example, Sedum and other succulent species can be 80–90 % water by weight under well-watered conditions (Berghage et al. 2007). While this stock of water does not directly influence broad hydrological properties such as stormwater retention to a significant degree, it is an important component of the drought tolerance mechanisms for many species.

The substrates used in extensive green roofs are designed to be highly permeable, but also have relatively high water holding capacity for their weight. Extensive substrates typically have maximum densities of around 1 g/cm-3 (62 lbs/ft3) and target maximum water holding capacities (water storage at field capacity) that range from 25 to 65 % by volume (FLL 2008). However, under field conditions, the actual maximum storage capacity of substrates is typically less than that estimated from laboratory techniques, likely as a result of structural changes to the substrate caused by plant root development and evapotranspiration (Fassman and Simcock 2012). In addition to substrate, extensive designs usually incorporate one or more layers designed to facilitate drainage, minimize erosion, or retain water for plant use. Few studies explicitly report water-holding capacities for these layers, but those that do report values less than 20 % by volume (Miller 2003; VanWoert et al. 2005a).

Under most conditions, the water that enters an extensive green roof has a short residence time. Substrate profiles are only 20–150 mm (5–37 in.) deep and constrained by an impervious membrane. This limits the absolute storage capacity of extensive designs. Once maximum storage capacity is reached, the porous substrate and even more porous drainage layers quickly channel runoff off of the roof. In most designs runoff enters the municipal wastewater stream, but runoff can be captured by a gray water system and returned to the roof as irrigation (Compton and Whitlow 2006; Chang et al. 2011). In many studies the stormwater storage capacity of green roofs is defined in terms of the plant available water held in the substrate (storage at field capacity-storage at the wilting point of vegetation). This value represents the theoretical long-term average capacity of the roof to retain intercepted rainfall. Storage capacity varies considerably with substrate composition and age. Martin (2008) reported storage capacity values from eleven studies that ranged from 2 to 53 %. When conditions are favorable for photosynthesis, green roof vegetation quickly depletes the stock of water in the substrate, restoring the storage capacity of the roof. A number of studies report that under ideal conditions the water storage capacity of vegetated roofs can be completely restored within about a week (VanWoert et al. 2005b; Durhman et al. 2006; Berghage et al. 2007; Voyde et al. 2010a).

The restoration of storage capacity through evapotranspiration is a key component of roof function, but it is a process that is strongly dependent on climatic conditions, the water status of the roof, and the composition of the vegetation. Evapotranspiration is extremely sensitive to a number of climatic drivers, principally temperature, humidity, and wind (Allen et al. 1998). In addition, the transpiration component of evapotranspiration is strongly tied to water availability . Plants maximize transpiration when water is readily available. Under well-watered conditions, succulent green roof vegetation contributed more than 50 % of the evapotranspiration from extensive green roofs (Berghage et al. 2007; Voyde et al. 2010b). Transpiration declines as water availability declines until water stress reaches a plant specific threshold, at which point transpiration ceases. Many of the drought-adapted species commonly used in extensive green roof designs have relatively high transpiration rates when water is available. Sedum roofs have reported maximum evapotranspiration rates that range from 5 to 6 mm day−1 (VanWoert et al. 2005b; Durhman et al. 2006; Voyde 2011; Sherrard and Jacobs 2012; Chap. 2). Roofs planted with non-xerophytic species such as Spartina alterniflora and Solidago canadensis can attain evapotranspiration rates that are an order of magnitude greater (Compton and Whitlow 2006), and roofs planted with a combination of growth forms can exhibit significantly greater rates than monocultures (Lundholm et al. 2010).

4.2 Climatic Influence on Green Roof Water Dynamics

Extensive green roofs balance two inherently conflicting goals: lightweight and water storage. They achieve their relatively lightweight (even when saturated with water) by being shallow and using light porous substrates. Coarse substrate texture also minimizes the risk of ponding in the shallow profile even during extreme rain events. Despite the limitation on total water storage capacity set by this substrate design, the presence of substrate alone significantly retards the timing and reduces the amount of runoff compared to conventional roof designs (VanWoert et al. 2005a; Schroll et al. 2011a). The presence of plants can potentially significantly enhance this stormwater function by dynamically restoring storage capacity through transpiration. However, the typical substrate design creates a unique and often severe water environment for plants that constrains plant choice. The rapid flow of water through the system via runoff or evapotranspiration can quickly put plants under water stress. As a consequence, shallow-rooted, drought tolerant species have typically been used on extensive green roofs even in mesic climates. The most common growth form used are succulents , with a distinct fondness for members of the Crassulaceae, although a broader range of regionally matched species and growth forms are increasingly being used (Dvorak and Volder 2010; Sutton et al. 2012). This system can be remarkably effective at both attenuating runoff and maintaining healthy vegetation if individual rain events are small in total volume and spaced at moderate (1–2 week) intervals between periods during which conditions are favorable for photosynthesis (Berghage et al. 2007). Many temperate climates such as those in northern Europe and in northeastern North America meet these criteria for substantial parts of the year. Average yearly stormwater retention values for roofs in these climates typically range from 30 to 60 %, broadly approximating mass balance estimates of watershed evapotranspiration for their regions (Gregoire and Clausen 2011; Carson et al. 2013).

However, a range of regional climates have conditions that are challenging for green roofs (Fig. 4.2). Climates with low or strongly seasonal precipitation may not provide enough water in a green roof context to sustain many otherwise drought adapted species. Even if a green roof plant assemblage can survive under a particular regional water regime it can still suffer reduced plant cover, be constrained in species or growth form diversity, or suffer reduced aesthetics (Nagase and Dunnett 2010; Schroll et al. 2011b; MacIvor et al. 2013). Other aspects of regional climate such as temperature extremes can also restrict plant choices. For example, many sedum species are intolerant of hard frosts or high temperatures (Chap. 3) (Boivin et al. 2001; Livingston et al. 2004; Simmons et al. 2008; Williams et al. 2010; Rowe et al. 2012).

Fig. 4.2
figure 2

Seasonal precipitation and temperature values for four North American cities. Values are thirty-year (1981–2010) averages for total monthly precipitation (bars), average monthly minimum temperature (solid circles) and average monthly maximum temperature (open circles). Data from NOAA National Climatic Data Center (NOAA 2014)

Full size image

The pattern of water availability imposed by regional climate can also influence other aspects of green roof performance. The thermal benefits of green roofs are influenced by the water content of the substrate and by evapotranspiration, and have been shown to vary by building type and location (Sailor et al. 2012). During extended dry periods, low levels of soil moisture and reduced evapotranspiration can lower the thermal benefit of a green roof (Sun et al. 2014). In the humid tropics, green roofs can become large heat sinks, partly because extreme daytime temperature can cause plants with crassulacean acid metabolism (CAM) physiology to stop transpiring. Much of the stored heat can later be transmitted into the building at rates greater than a conventional roof (Jim 2014). Green roof stormwater attenuation performance is also strongly influenced by climatic conditions. In general, the proportion of precipitation retained by green roofs declines as storm volume and frequency increase and during winter when the potential evapotranspiration is limited (Mentens et al. 2006; Villarreal 2007; DiGiovanni et al. 2010; Carson et al. 2013).

A general, lack of published long-term monitoring data (Chaps. 2 and 13) as well was design differences among roofs make it difficult to assess the impact that regional climate has on overall green roof performance. The few data that are available principally report stormwater performance . In the Pacific Northwest of North America the majority of the yearly precipitation falls during the winter when the potential for evapotranspiration is small. In addition, winter storms can come closely spaced in time creating prolonged periods of precipitation. Spolek (2008) reports that roofs in Portland, OR monitored over 2–3 years had total rainfall retention that ranged from 12 to 25 %. A Seattle roof monitored for a year had a total retention of 31 % (Berkompas et al. 2009). Both of those values are among the lowest reported from full-scale monitored green roofs (Carson et al. 2013). The overall, long-term retention seems to be driven by reduced retention performance during the winter. On the Portland roofs monitored by Spolek (2008), total rainfall retention was only 12 % during the winter compared with 42 % during the spring and summer. Similarly, test-bed scale roofs in Corvallis, OR retained less than 28 % of intercepted rainfall during the winter, which was less than half their retention capacity during the summer (Schroll et al. 2011a). In the Corvallis study, the seasonal differences were partly attributable to the higher frequency and volume of storm events during the winter as well as to the reduced recharge capacity provided by the vegetation.

In other regions, such as sub-tropical Florida, the timing and intensity of individual storm events can influence stormwater performance even if the bulk of rainfall occurs during favorable evapotranspiration conditions. Using a field validated mass balance model Hardin et al. (2012) predicted green roof stormwater retention values for several Florida locations ranging from 33 to 51 %. Locations with moderate predicted retention efficiencies seem to reflect the large, overall volume of annual precipitation at those sites. Similarly, green roofs in climates that experience large yearly variation in conditions can exhibit large variation in stormwater performance. The southern California climate is notable for its extreme inter-year variation in total precipitation. There Bennett et al. (2008) report that modeled rainfall retention efficiency for a typical extensive design varied from 21 to 64 % depending on yearly precipitation patterns. It is important to note that total stormwater retention may not be an appropriate performance metric for some goals. Delay in runoff as well as reductions in peak flow for individual storm events can be more relevant to stormwater management, and these aspects of attenuation performance can be high relative to conventional roofs across a range of storm size and timing (Fioretti et al. 2010). In general, high variability in rainfall patterns and weather conditions make it difficult to predict the hydrological performance of green roofs based on storm characteristics. Instead, mechanistic models of water flux through the green roof system provide the best predictive descriptor of stormwater performance (Stovin et al. 2012).

4.3 Plant Water Use Strategies for Green Roofs

Ideally, green roof vegetation should combine high transpiration capacity with the ability to tolerate extended periods of water deficit. This is perhaps not as conflicted a goal as it might seem. Plants possess an incredible diversity of water use strategies , many of which are appropriate in green roof contexts across a range of regional climatic conditions. Farrell et al. (2013a) have developed a conceptual model to screen potential plants for green roof applications based on their water use under mesic and xeric conditions as well as their ability to minimize water stress during periods of water deficit. Here I categorize a slightly broader (although still not comprehensive) list of water use strategies into syndromes that relate to green roof performance.

4.3.1 Water Loss Minimizers

Many species from xeric or seasonally dry climates are exceptionally good at conserving water. Structural adaptations in these species include waxy or hairy leaf coverings, leaf orientations that reduce insolation and heating, fewer or smaller leaves, and reduced stomatal density. Many species also have succulent stems or leaves and use internally stored water to buffer the effects of soil water deficit. Another principal physiological adaptation is crassulacean acid metabolism (CAM) photosynthesis . Plants with one or more of these adaptations can survive extended periods of water deficit. Drought tolerance screening experiments have identified a number of species that can survive on typical extensive substrates without water for more than 130 days (Durhman et al. 2006; Bousselot et al. 2011; Farrell et al. 2012). Many water loss minimizing species used in green roof applications are low-growing perennial succulents , particularly those in the family Crassulaceae, but perennial forbs (Fig. 4.3a) and woody shallow rooted shrubs are also common.

Fig. 4.3
figure 3

Species representing different water use syndromes useful in green roof contexts. a Species like Erigeron linearis can withstand long periods of drought even on thin mineral soils, because they have consistently low transpiration rates. b Many species in the genus Sedum facultatively switch between C3 and CAM photosynthetic pathways allowing them to have exceptional drought tolerance but also achieve moderate transpiration rates when water is available. c Geophytes like Camassia quamash avoid water stress through dormancy. d Bryophytes like Racomitrium canescens tolerate long periods of desiccation. e Although not exceptionally drought tolerant, many warm season grasses like Zoysia sp. can withstand extreme temperature and light conditions. (Photograph credits: (a) Richard Martinson, (b, c), Erin Schroll, (d) John Lambrinos, (e) Alec Kowalewski)

Full size image

An inherent tradeoff of many of the adaptations that minimize water loss is reduced photosynthesis. Consequently many drought tolerant species have comparatively low photosynthetic rates per leaf area or biomass (Körner et al. 1979). Water loss minimizers emphasize consistent (although relatively low) photosynthetic capacity across a range of water conditions. Consequently, extreme water loss minimizers might not be the most appropriate plant choices for green roofs that experience modest periods of water deficit. In these cases, species with a greater peak capacity to transpire would be more desirable. Interestingly, great variation exists in the peak transpiration capacity among succulent species commonly used on green roofs, even within the same genus (Voyde et al. 2010b; Farrell et al. 2012; Starry 2013). Some of this variation potentially reflects dynamic responses to water availability in some species (see Sect. 4.3.2 Water loss adapters).

On the other hand, some water loss minimizers are robust choices in climates where green roof substrates are likely to experience more prolonged water deficit, or in climates where water availability is highly variable. Some xerophytic perennials have pronounced water loss adaptations, but are poor choices for extensive green roofs because they minimize water stress by accessing water stores from deep or spatially complex soil profiles (Ehleringer and Mooney 1983); the process of hydraulic redistribution is a tactic wholly unavailable on shallow, extensive green roofs.

4.3.2 Water Loss Adapters

Some species are particularly adept at adjusting their water use to the amount of water available, enabling them to have higher photosynthesis rates when water conditions are favorable. This is a desirable trait for a green roof plant. Woody perennials often adjust water use through gross morphological changes such as replacing photosynthetically efficient leaves with more water use efficient ones, or by shedding leaves altogether and entering a period of drought induced dormancy (Westman 1981). Many perennial temperate grasses achieve exceptional drought dormancy by combining leaf senescence with physiological dehydration tolerance mechanisms (Volaire and Norton 2006). Not all drought dimorphic species are desirable for extensive green roof applications. Many of these species (both woody and herbaceous) have deep or extensive root systems. In addition, drought induced changes can create undesirable aesthetics or pose fire safety concerns from the accumulation of dry biomass.

In some species, photosynthetic pathways are highly plastic and plants facultatively adjust photosynthetic metabolism based on water availability and other environmental conditions (Andrade et al. 2009). Many succulent species are known to switch between C3 and CAM photosynthetic pathways or to adjust the diurnal timing of gas exchange and CO2 fixation in relation to water availability (Sayed 2001; Fig. 4.3b). These adjustments allow some succulent species to achieve moderate transpiration rates during periods of high water availability. However, the details of photosynthetic plasticity and its influence on water use patterns are highly species specific, nuanced, and dependent on a number of environmental factors (Herrera 2009). For example, although Phedimus albus (syn. Sedum album) and Phedimus kamtschaticus (syn. S. kamtschaticum) are both broadly known to switch between C3 and CAM metabolism, they display markedly different physiological performance in a green roof context (Starry et al. 2014). Starry found that S. kamtschaticum had significantly higher transpiration, higher daily carbon assimilation, and switched from C3 to CAM metabolism at a lower substrate water availability compared to S. album. As a consequence it used 35 % more water than S. album. However, perhaps partly because of its more parsimonious water use S. album was more drought tolerant than S. kamtschaticum. Rowe et al. (2014) found similar results with syn. S. kamtschaticum var. floriferum (trade name S. floriferum). In a greenhouse study using experimental roof modules, S. album survived 84 days without water, but S. floriferum did not.

4.3.3 Water Stress Avoiders

Some species take drought dormancy to the extreme and either complete their entire life cycles before water availability declines , or exist for extended periods as highly specialized drought survival structures. Many of these species have ruderal life history strategies or evolved under strongly seasonal water availability such as deserts or seasonal wetlands. Desert annuals that complete their life cycle within a brief few weeks are prime examples. Many annuals from a range of different habitats could be suitable for green roof contexts, although they have not often been used to date (Chap. 10). Nagase and Dunnett (2013) report that a diverse annual meadow can be easily and economically established on an extensive green roof in the central UK climate. A diverse assemblage of species provided abundant flowers throughout the summer and fall even without irrigation, and the system required very little maintenance apart from annual mowing. However, the long-term performance of annual plant based systems has not been investigated. Annual systems will likely require a tolerance for large dynamic changes in the species composition of the roof over time (Chaps. 12 and 13).

Species with perennating organs such as geophytes are another category of drought stress avoiders that are potentially suitable in a green roof context (Schroll et al. 2011b; Nagase and Dunnett 2013; Van Mechelen et al. 2014; Benvenuti 2014; Fig. 4.3c). Like annuals, these species often produce strikingly attractive flowers, however bloom times can be relatively brief and all above ground structures typically die back completely. However, in contrast to many annuals, the amount of senescent biomass is relatively small and mowing management is not required. Many geophytes have particularly early or late bloom times that can be a valuable trait in terms of pollinator resources as well as expanded aesthetics (Benvenuti 2014) .

4.3.4 Water Loss Tolerators

Some species lack well-developed adaptations for conserving water, but instead have a remarkable ability to withstand desiccation (Hoekstra et al. 2001). Many of these species are non-vascular bryophytes and lichens , but some vascular resurrection plants have this ability as well (Gaff 1989). Mosses have most commonly been used on green roofs (usually in combination with sedum) in northern temperate climates (e.g. Bengtsson et al. 2005; Oberndorfer et al. 2007). However, their drought tolerance properties make them potentially suitable choices for a number of climates with extended periods of water scarcity (Anderson et al. 2010). Also, their lack of roots and prodigious water retention capacity also suggest that they could be used to develop extremely lightweight but still highly functional systems.

4.3.5 Water Loss Sensitive

Plants that do not have well developed drought tolerance mechanism can still be suitable choices for extensive and semi-intensive green roofs in some contexts. Because of their tolerance of extreme temperature and light conditions, warm season turf grasses have been used on green roofs in sub-tropical and tropical climates, although typically supplemental irrigation is provided (Jim 2012; Ju et al. 2012; Sutton et al. 2012; Chen 2013). Some wetland species have broad habitat tolerances or some ability to withstand short periods of dry conditions. They may be good choices for relatively wet climates. MacIvor et al. (2011) found that several wetland species were able to survive condition on an extensive green roof in maritime Nova Scotia over two growing seasons, although their overall cover was less than that of more dryland-adapted species.

It is important to note that the performance oriented syndromes described above don’t necessary reflect evolutionary or ecological tradeoffs. Species can combine aspects and traits that span categories. Indeed, most plants exhibit some characteristics of each syndrome to varying degrees. Still, the syndromes provide a useful way of relating dominant species traits to functional green roof goals as well as to the abiotic constraints imposed by different regional or situational contexts. Closely related species or species that share similar life forms and life history can often have very different overall water use patterns (Wolf and Lundholm 2008). This makes it suspect to use growth form or simple morphological traits as screening tools for green roof plant assembly. Plant choice decisions need to be based on detailed species-specific functional traits that are evaluated in the context of specific climate profiles and performance goals (Chaps. 9 and 11).

4.4 Modifications to Green Roof Water Dynamics

The typical extensive green roof design can be modified in a number of ways that alter the dynamics of water through the system. These changes are often made to better match a particular green roof to local climatic conditions, or to enhance particular green roof functions such as aesthetics or habitat quality.

4.4.1 Plant Assemblage Design

The species and growth form composition of green roof vegetation can have a strong influence on water capture and retention, as well as the rate at which storage capacity is restored following a rain event (see Sect. 4.1). For many functional goals such as stormwater management and building thermal load reduction, vegetation designs must balance a tradeoff between high transpiration capacity and drought tolerance. The optimal balance between these two functional traits depends strongly on the specific climatic context of the roof. However, designing systems that are inflexibly tailored to a narrow regime is unwise. Climatic conditions vary within and between years, and spectacularly so in some climates. Incorporating species with an ability to facultatively adjust water use depending on water availability is one strategy for improving performance under variable conditions (Chap. 11). The water use plasticity of many Sedum and other succulent species make them good choices for extensive green roofs. However, there is considerable variation in water use patterns as well as climatic tolerances among succulent species (Voyde et al. 2010b; Farrell et al. 2012; Starry 2013). In addition, species from other growth forms and taxonomic groups can also exhibit high degrees of water use plasticity (Farrell et al. 2013a). Such species-specific functional traits are too rarely taken into account when making green roof plant selections, partly reflecting lack of accessible data on the functional traits of candidate green roof species.

Another strategy for designing functionally resilient green roof vegetation is to combine species with complementary water use patterns or functional traits. Several studies have reported a positive relationship between the species or growth form richness of green roof vegetation and water management performance as well as other functions (Lundholm et al. 2010; Nagase and Dunnett 2010; Cook-Patton and Bauerle 2012; Chap. 9). Although the mechanistic causes of these relationships are not well understood, one likely reason is trait complementarity . Specifically exploiting complementarity could be an effective design strategy. For example, in the Pacific Northwest of North America most of the precipitation falls during the cool winter when potential evapotranspiration is low. During these periods mosses can significantly increase water storage capacity above that of the substrate itself through water held in their complex physical structure (Anderson et al. 2010). During the spring when conditions are more favorable for photosynthesis, vascular plants can contribute significantly to recharge capacity through transpiration (Schroll et al. 2011a). Preliminary results suggest that combining both moss and sedum can significantly improve yearly stormwater retention over single species vegetation types (Van Hoosen pers. comm.).

Another potential cause of the positive relationship between performance and vegetation diversity is that some species facilitate the growth and survival of other species in the assemblage. One broad way that facilitation can happen is that species modify abiotic conditions, making them more favorable for themselves or other species (Hastings et al. 2007). Butler and Orians (2011) showed that the growth and overall health of the perennial forbs Agastache rupestris and Asclepias verticillata on green roofs were decreased by the presence of Sedum species during favorable conditions but were increased during summer water deficit. Similarly, Heim (2013) found that the presence of the moss Polytrichum commune increased the growth of the perennial forb Solidago bicolor under experimental green roof conditions. In both studies the cause of the observed facilitation is equivocal, but both the Sedum and moss decreased temperature and increased water availability in the substrate. Plant-microbial symbioses are another broad mechanism for facilitation. For example, the symbiotic relationship between most plants and mycorrhizae fungi can directly increase their ability to acquire and uptake water, particularly under drought conditions (Augé 2001; Chap. 7). Most members of the Crassulaceae do not form arbuscular mycorrhizal associations (Wang and Qiu 2006). However, the near absence of arbuscular mycorrhizal fungi within newly installed engineered substrates could be a significant factor limiting the range of plant species that are suitable for green roofs as well as the drought tolerance of diverse green roof systems (John 2013).

As with complementarity , facilitation could potentially be exploited to improve functions associated with plant water use. For example, moss could be used to facilitate the establishment of vascular species, reduce the frequency of extreme drought stress , or reduce the overall need for supplemental water. However, manipulating species interactions to achieve specific functional goals is complicated by the inherent dynamism of green roof vegetation (Chap. 12). Although few published long-term studies exist, those that do suggest that the composition and relative species abundance of green roof vegetation can change dramatically in the years following establishment (Chap. 12). In some cases, vegetation changes seem to reflect consistent successional trajectories that are constrained by substrate type and depth, and by water availability (Köhler 2006; Rowe et al. 2012; Bates et al. 2013; Thuring and Dunnett 2014). However, there can also be considerable year-to-year variability in community structure much of which correlates with variation in drought stress (Bates et al. 2013). Indeed, one aspect of developing drought resilient vegetation may be accepting a degree of un-planned variation in its composition, including the natural colonization of species from the regional species pool (Chaps. 10, 11 and 12).

4.4.2 Substrate Design

Since substrate is the largest single store of water in a green roof system unsurprisingly its composition, depth, and slope strongly influence patterns of water flow through the system (Li and Babcock 2014). Substrate design is therefore an important way of optimizing extensive green roofs to particular environmental constraints or functional goals. In some cases substituting natural soil profiles for highly engineered substrates may be a productive approach (Chap. 6). Also, some authors have argued that wetland-like systems could be a practical alternative under some conditions (Song et al. 2013). More commonly, a number of adjustments have been made to the basic engineered substrate design in order to attain specific performance goals or in response to environmental constrains. However, developing appropriate design criteria is complicated by the complex interactions between substrate, plants, and environmental conditions (Chap. 5). For example, in isolation the influence of substrate characteristics on water retention and runoff dynamics are straightforward to predict using existing mass balance and more mechanistic hydrological models . However, because hydrological performance reflects strong interactions between a number of highly variable system components and environmental conditions it is necessary to calibrate and validate models to each specific case, limiting their usefulness as design tools (Li and Babcock 2014).

Another example of this contextual complexity occurs with the relationship between substrate depth and plant performance. Increasing substrate depth increases the store of water available to plants (VanWoert et al. 2005a), buffers plant roots from cold stress (Boivin et al. 2001; Rowe et al. 2012), and can accommodate species with deeper rooting profiles (Sutton et al. 2012). A number of studies have documented a positive relationship between growth and survival and substrate depth for a number of potential green roof species. Dvorak and Volder (2010) reviewed this literature and concluded that without irrigation only the most drought adapted succulent species are able to tolerate the water stress conditions imposed by the shallowest (< 10 cm) substrate profiles across a range of climates. Increasing substrate depth or providing supplemental irrigation greatly expands the diversity of species and functional types that a roof can support. However, the results are highly species specific and can vary over time (Dunnett et al. 2008; Getter and Rowe 2009; Rowe et al. 2012). For example, in the Rowe et al. (2012) study seven species performed well on 2.5–7.5 cm substrate depths when they were evaluated 2 years after establishment; yet by year seven only two of these species were still present on any media depth .

Most extensive green roof substrate designs are based on guidelines established by the German Landscape Research, Development, and Construction Society (FLL). The guidelines set performance criteria for key parameters such as permeability, water storage capacity, and maximum load. Performance targets vary for different green roof configurations and contexts, but broadly specify that substrates should have high permeability (saturated water flow ≥ 0.001 cm s−1), hold 35–65 % v/v water at field capacity , contain ≤ 15 % w/w of fine (< 63 μm) particles, and contain 10–20 % v/v organic matter (FLL 2008). Typically designers have achieved these performance targets using substrates composed primarily of course inorganic aggregates such as expanded clay, pumice, and a number of recycled materials such as crushed brick. A number of modifications to substrate composition have been proposed to increase the amount of water available to plants or to dampen fluctuations in water availability. These include increasing the organic matter content or incorporating other water retention additives such as polymer gels. As with substrate depth, the efficacy of these strategies appears to be highly context dependent. For example, in a short-term greenhouse experiment Nagase and Dunnett (2011) evaluated the influence of substrate organic matter content on the growth of four forbs and grass species. All four species responded differently to the level of organic matter, and results depended on the watering regime. Under a dry regime increasing organic matter content above 10 % by volume did not have any significant effect on plant growth . However, under a well-watered regime some species increased growth considerably with higher organic matter content. Nagase and Dunnett speculate that the lush growth might be a disadvantage under more natural conditions that include periodic drought. Papafotiou et al. (2013) tested the influence of substrate depth, organic matter type and content, and irrigation frequency on the growth of several drought adapted Mediterranean species. Similar to the Nagase and Dunnett (2011) study, they found significant interactions between treatments. Notably, however, some of the best plant performance was observed on the shallow (15 cm) compost amended substrate even under minimal irrigation. Other water retention amendments such as hydrophilic polymers (hydrogels) and silicate granules can increase overall as well as plant available water holding capacity of the substrate, although the magnitude of the effect depends on the type of additive and substrate (Farrell et al. 2013b). The incorporation of hydrogels into green roof media has been found to increase the growth of grasses and non-succulent forbs (Oschmann et al. 2007; Sutton 2008). Biochar is another potential amendment that could increase plant available water. Beck et al. (2011) report that green roof substrate amended with 7 % biochar had significantly greater water retention than non-amended substrate. However, no published studies have evaluated the effect of biochar on plant available water or plant performance in a green roof context .

In addition to the substrate itself, most extensive green roof profiles include a number of plastic or woven geotextile layers, some of which are explicitly designed to retain and increase plant available water. However, few studies have directly evaluated how these layers influence the plant water relations of the system (Chap. 3). In the only experimental study that has been reported to date, Savi et al. (2013) report that 90 % of the water retained by these layers is potentially available to plants, compared to 34 % for the substrate itself. The presence of the layers also had a significantly positive effect on plant water status and survival. However, the transfer of water through the roof profile was strongly influenced by diurnal temperature patterns and the details of system design, suggesting that designs could be optimized to enhance plant water availability under specific environmental conditions.

The wide variety of roof designs, the complex interactions between design parameters and environmental conditions, and the high degree of species-specific responses make it nearly impossible to establish universal design prescriptions for green roof substrate. Instead, a promising approach is to develop regionalized and function specific design (Chap. 9) criteria based on local field testing. Fassman and Simcock (2012) used this approach to develop design specifications for extensive green roof media for Auckland, NZ that will maintain plants without irrigation under typical conditions and capture 100 % of runoff from storms that have less than 25 mm (1 in.) of precipitation .

Despite the complications described above, incorporating heterogeneous substrate depths as well as improving the plant available water capacity of the substrate are promising strategies for maintaining species or growth from diverse vegetation on extensive green roofs. In climates that experience extreme water deficit, adjustments within extensive design constraints may not be sufficient if aesthetics or plant diversity are important design goals. An example of such a system is the green roof installed on the Oregon Dental Service (ODS) building in Bend, OR (Fig. 4.4). The average annual precipitation in Bend is only 29.5 cm (NOAA National Climatic Data Center) and since the roof was designed as an accessible recreation area of the building, diverse and aesthetically pleasing vegetation was a design criterion. To support the vegetation substrate was composed of 50 % mushroom compost and 50 % pumice and varied in depth from 20 to 81 cm. The roof was planted with 29 native western U.S. plant species that were matched to the specific substrate microhabitats .

Fig. 4.4
figure 4

The green roof on top of the Oregon Dental Service (ODS) Building in Bend, OR. In extreme water environments like Bend, designs may need to incorporate deeper substrate depths if diverse vegetation is a goal. Substrate depths here vary from 20 to 81 cm (8–32 in.). (Photo: Richard Martinson)

Full size image

The initial plantings established well. However, over the next several years maintenance crews removed the majority of the perennial grasses and forbs that were part of the initial installation. The crews were unfamiliar with the plant palette and removed as weeds anything they did not recognize. The company that initially designed and installed the roof has recently been hired to re-establish some of these plantings and to provide ongoing maintenance. This experience highlights the need for comprehensive management plans and properly trained maintenance personnel in order to ensure the long-term success of green roof systems (Chap. 13).

4.4.3 Irrigation

One of the fundamental appeals of extensive green roofs is that they can help solve a number of problems associated with urbanization at a low expenditure of resources such as energy, nutrients, or water . The use of resource inputs in their management therefore often receives skepticism. More practically, present and predicted freshwater scarcity will put increasing financial as well as legislative restrictions on commercial and residential water use (Falkenmark and Xia 2013). Although largely developed in northern Europe, the modern extensive design that combines shallow well-drained substrates with mats of low growing succulents adapts well across a range of climates even with no or minimal irrigation . Recent and ongoing research has identified regionally adapted vegetation and substrate designs (including increased depth) that can be used to develop more regionally tuned versions of this basic low input design. Examples include arid Australia (Razzaghmanesh et al. 2014; Farrell et al. 2012), subtropical New Zealand (Voyde et al. 2010b), Mediterranean Europe (Van Mechelon et al. 2014), and North America (Dvorak and Volder 2010; Sutton et al. 2012; Chaps. 3 and 11).

However, a number of potentially appropriate uses exist for irrigation on extensive green roofs depending on the context and functional goals. During the establishment period the growth and survivorship of even highly drought tolerant species is increased by supplemental water (Dunnett and Nolan 2004; Thuring et al. 2010; Sutton 2013). The development of high plant cover and health during establishment can reduce weed pressure and potentially influence other aspects of long term performance. After establishment, climatic variation can periodically create periods of extended water deficit even in relatively mesic climates. These periodic stresses may act as a stochastic species filter contributing to observed long term declines in species and growth form richness on un-irrigated green roofs (Köhler 2006; Rowe et al. 2012) or dramatic changes in species dominance (Chap. 12). Overall, irrigation can greatly expand the pool of plant species, growth forms, and functional types that are suitable for a particular green roof context, particularly on shallow substrates or in water limited climates (Monterusso et al. 2005; Price et al. 2011; Schroll et al. 2011b; MacIvor et al. 2013). Even if plants can survive a particular green roof water environment without irrigation, their growth and traits related to aesthetics such as flowering and canopy cover can be improved with irrigation (Nagase and Dunnett 2010; Schroll et al. 2011) .

Irrigation can potentially indirectly enhance a number of other green roof functions via its effects on the diversity and composition of vegetation. As described above (plant assemblage design), a number of studies have documented a positive relationship between green roof vegetation diversity and stormwater management performance. In aggregate, diverse green roof species assemblages can be more resilient to environmental perturbations such as drought stress compared to less diverse assemblages (Nagase and Dunnett 2010; Chap. 10). In addition, some aspects of aesthetic preference are related to functional and structural diversity. In a survey of Australian office workers, the most preferred living roof type had taller, grassy, and flowering vegetation while low growing succulent vegetation was least preferred (Lee et al. 2014).

The degree to which green roofs moderate internal building temperature is partly influenced by the water content of the system and by rates of evapotranspiration (Barrio 1998). Irrigation could potentially increase evapotranspirative cooling and be a tool for reducing building cooling costs. Sun et al. (2014) modeled thermal performance of a green roof in Beijing, China and estimated that the value of the avoided building cooling costs related to irrigation was greater than the monetary costs of the irrigation itself. However, other studies in a Mediterranean and a sub-tropical climate have reported low cooling efficiencies associated with green roofs and minimal or no contribution to building cooling associated with irrigation (Jim and Peng 2012; Schweitzer and Erell 2014).

Given the wide range of functional benefits associated with irrigation it seems reasonable to expect that its use can be justified to meet a range of performance goals in a number of contexts. In these cases irrigation systems should be optimized to maximize water use efficiency relative to the functional goals. Unfortunately, few published studies have evaluated irrigation system design in a green roof context. In one of the few, Rowe et al. (2014) compared the performance of overhead , drip, and sub-irrigation systems. They found that the overhead system resulted in the highest substrate water content and wasted the least amount of water in the form of runoff . The overhead system also produced a more even distribution of water through the three-dimensional substrate profile. Sub-surface and drip systems created more heterogeneous distributions, likely because vertical and lateral capillary movement of water was limited in the porous substrate. As a result, plant growth and health were greatest under overhead irrigation . However, optimum system design is likely dependent on other system elements such as substrate composition and depth, water retention layers, the specific species composition of the vegetation, environmental conditions (e.g., diurnal winds, relative humidity, and shade patterns), and cost constraints. In addition, the use of irrigation can directly decrease substrate stormwater storage capacity under some climatic conditions (Schroll et al. 2011a). The design of the irrigation system as well as other roof components such as substrate water holding capacity will need to reflect a balance between potential tradeoffs such this. Irrigation management will also likely need to be periodically adjusted over time as the roof ages .

The development of expert irrigation systems for green roofs that match the timing and amount of applied water to actual plant water needs is still in its infancy. Mass balance equations can be used to estimate the water status of the substrate and therefore predict the need for irrigation if the relationship between water status and plant water stress is known. The most difficult parameter of water mass balance to directly measure is usually evapotranspiration (Chap. 2). Several predictive evapotranspiration models have been developed for green roofs based on empirically derived regression models as well as a range of more mechanistic standard agricultural models (Kasmin et al. 2010; Rezaei and Jarrett 2006; Voyde et al. 2010b; DiGiovanni et al. 2012; Sherrard and Jacobs 2011; Karanam et al. 2013; Starry 2013). Estimates of evapotranspiration for green roofs based on reference evapotranspiration estimates require the application of attenuating factors (DiGiovanni et al. 2012) or water stress coefficients for water-limited conditions. Furthermore, there are few published values for crop coefficients or plant factors specific for green roof plant species that are necessary to adapt physically based reference evapotranspiration estimates to green roof vegetation types (DiGiovanni et al. 2012; Sherrard and Jacobs 2011; Starry 2013). Alternatively, empirically derived estimates of evapotranspiration for green roof systems were found to be robust across water status for specified vegetation types (Voyde 2011). Voyde et al. (2010b) use such an approach to develop irrigation guidelines for succulent planted roofs in Auckland, New Zealand. Empirical approaches require the calibration and validation of transpiration models for the variety of roof designs and environmental contexts. Currently no consensus exists on the most appropriate modeling approach for green roof contexts. Monitoring approaches including the development of inexpensive wireless sensor networks could provide a practical way of directly measuring the real-time water status of a roof in high spatial and temporal resolution (Starry et al. 2011; Lea-Cox et al. 2013; Chap. 2) .

Diverting runoff from green roofs into a gray-water system or using secondarily treated municipal wastewater for irrigation are two other approaches to minimizing the impact of green roof irrigation on regional water demand. In addition, the storage capacity of a gray water system can also greatly improve the stormwater performance of roofs in some climates. In central Florida, a green roof system that incorporated a cistern and irrigation retained 87 % of yearly runoff compared with only 43 % retention for a system without irrigation and a cistern (Hardin 2006; Wanielista and Hardin 2006). A number of potential problems arise with the use of gray water for irrigation (Maimon et al. 2010). Only one published study has evaluated gray water use in a green roof context. Moritani et al. (2013) exposed S. kamtschaticum to periodic irrigation water with elevated salinities typical of some (but not all) gray water. The irrigation water increased salt stress for the plants, which reduced evapotranspiration. This could have a positive impact on the overall irrigation requirements for the system, but of course could detrimentally influence long-term plant growth and survival as well as stormwater performance .

4.5 Future Research Needs and Questions

A great deal of progress has been made in understanding how green roofs use, store, and regulate the flow of water. However, there are still a number of areas in need of further research as well as a number of issues for which the broader community of green roof designers, managers, and users are still developing optimal solutions.

4.5.1 More Integrated and Regionalized System Designs

There has been progress in developing green roof designs that are better tuned to local climates or specific user goals such as native wildlife habitat. However, there is still great potential to develop more fully integrated system designs that coordinate substrate composition , vegetation composition, water management, and long term maintenance plans. Ideally, sets of region-specific design and management specifications should be developed to guide designers as well as those charged with maintaining the long-term functioning of green roofs. These specifications should be flexible enough to guide the design of roofs that vary in functional performance goals. For example, what is the best design for a roof in the U.S. Pacific Northwest whose main functional goal is stormwater management? Does the design need to change if wildlife habitat or aesthetics are prime performance goals? What if maintenance budgets or expertise are limited?

4.5.2 More Automated Systems for Tracking Water Status and Broader Performance

As green roofs are increasingly used in climates that might necessitate the use of irrigation there is a need for high spatial and temporal resolution data describing their water status. Automated systems that track water status coupled with efficient irrigation designs could significantly minimize the water use impact of green roofs. Even when green roofs are not irrigated, automated systems that track performance metrics such as runoff retention or thermal performance would facilitate adaptive management and could potentially be used to provide performance based credits or incentives within a payment for ecosystem services framework. The technical hurdles to the development of such systems are beginning to be overcome, but costs need to decline and more commercialized off-the-shelf systems need to be developed.

4.5.3 Policy and Design Guidelines for the Appropriate Use of Supplemental Water

Depending on the functional goal and the local climate, the use of irrigation as well as other inputs such as fertilizer may be unavoidable. We need better frameworks for assessing whether that resource use is appropriate, or whether alternative strategies such as cool roofs (Millstein and Menon 2011) might be more appropriate in some contexts. In regions such as the western U.S. specific policy guidelines and incentives regarding the use of water for green roofs might be helpful. For example, the City of Portland, OR U.S.A. offers building permit incentives for developers that incorporate green roofs into building designs. Importantly, the city also has set water use standards and provides technical assistance to meeting the standard (City of Portland 2009).