Abstract
One of the primary functions of green roofs in urban areas is to moderate rainwater runoff, and one of the major impediments to the survival of plants on an extensive green roof (EGR) is a lack of available water during dry periods. Runoff moderation and water storage are both influenced by the composition of the growing media. Here we present a framework for evaluating the hydrologic performance of EGR growing media and also provide hydrologic attribute data for several commonly used EGR media constituents. In this three-phase study, we: 1) measured hydrologic attributes of individual EGR media constituents, 2) predicted attributes of media mixtures using individual constituent data, and 3) tested the seven top-ranking mixtures to evaluate hydrologic performance. Hydrologic attributes included wet weight and water held at maximum retentive capacity, long-term water retention, and hydraulic conductivity. Because perlite was light in weight yet held the greatest amount of water both at its maximum retentive capacity and in the long term, media mixtures dominated by perlite were predicted to have the best overall hydrologic performance. Mixtures dominated by pumice were also predicted to perform relatively well but were heavier. Despite the slightly greater weight and slightly lower performance, pumice may be a preferred alternative to perlite because perlite is a processed constituent with greater estimated embodied energy. Results indicate that performance of mixtures can be adequately predicted using performance of individual constituents for wet weight, water held, and long-term water retention. Hydraulic conductivity was less predictable because the pore volume in mixtures can be unrelated to the pore volume of the individual constituents. The framework presented here can be used to evaluate the performance of other EGR media, and the media attribute data can be used in formulating EGR media mixtures for specific applications. In addition, the attribute data can serve as a benchmark for evaluating other EGR media. Our results underscore the need for standardization of methods for more effective comparisons of EGR substrates, and also reinforce the need to evaluate EGR components using real-world scenarios.
Keywords: green infrastructure, extensive green roof, plant growing media, stormwater runoff, hydrologic performance
1.0. Introduction
Green infrastructure has been defined as “the network of green spaces and water systems that delivers multiple environmental, social, and economic values and services to urban communities” (Pitman et al. 2015; USEPA 2008). Within this broad definition, green roofs are one of the most frequently installed structures because roofs in urban areas often afford the greatest amount of under-utilized space (Carter and Jackson 2007). The primary benefits of green roofs in urban areas include: (1) reducing or delaying rainwater runoff, (2) mitigating the urban heat island effect, (3) reducing energy requirements for building heating and cooling, and (4) improving building aesthetics (Nawaz et al. 2015; Oberndorfer et al. 2007). Green roof designs can be categorized as intensive green roofs (IGRs) or extensive green roofs (EGRs). IGRs typically have deeper soils with edible crops or landscaped gardens that require irrigation and relatively high maintenance. In contrast, EGRs typically have shallow soils, are intended to function with minimal irrigation and maintenance, and use a narrow set of plant species that are suited to the water stress created by this environment (Berndtsson 2010; Oberndorfer et al. 2007). IGRs are more often installed on new buildings, where structural support to accommodate the additional weight can be incorporated into the building design, whereas EGRs are typically installed on existing buildings. In this paper, we focus on EGRs.
EGR hydrologic performance can be strongly influenced by substrate (i.e., growing media) and vegetation composition (Graceson et al. 2013; Griffin et al. 2017; Nagase and Dunnett 2012; Stovin et al. 2015), but the effect of these design elements depends on the climate where the EGR is located. In temperate climates vegetation has the greatest influence on hydrologic performance during the summer months when vegetation is most abundant, media moisture content is sufficient for plant growth, and transpiration is greater due to higher temperatures (Berndtsson 2010; Berretta et al. 2014; Metselaar 2012). In regions where most of the precipitation falls when plants are dormant, EGR media composition may play a more dominant role than vegetation composition in reducing peak runoff when urban streams are at the greatest risk of flooding (Berghage et al. 2009; Schroll et al. 2011). Runoff retention is directly related to the storage capacity of the EGR media, so when the media is at or near field capacity, the ability to absorb additional moisture will be negligible (Spolek 2008). However, additional precipitation must still flow through the media before it runs off the building, so while the total volume of runoff through media at its maximum retentive capacity may not be significantly reduced, the rate of runoff can be slowed. This increases lag time to peak flow and thus may ameliorate urban flooding (Berghage et al. 2009). For green roofs, the hydraulic conductivity (HC) of media approximates how quickly additional rainfall will flow through the green roof once the media has reached its maximum retentive capacity. Optimal HC may vary depending on climatic conditions where the green roof is located. HC should be high enough to prevent standing water formation given site-specific maximum rainfall rates, but no higher so runoff delay time is maximized.
In addition to moderating runoff, an important consideration for EGR media composition is the ability of the media to retain water over time, which may be critical to plant survival during periods of drought. The typical EGR environment is challenging to most plants due to shallow soils, lack of irrigation, and direct exposure to sunlight and wind, which cause dramatic temperature fluctuations and sporadic moisture regimes. As a result, hardy, drought-tolerant species (e.g., Sedum spp.), often not native to the area, are most commonly planted (Emilsson 2008; Nagase and Dunnett 2010; Thuring and Dunnett 2014). Maximizing EGR water retention over time may facilitate sustainability of native plant communities on EGRs in temperate climates.
Due to load limitations of existing roofs, EGRs are typically designed with light, shallow (< 10 cm deep) substrates that will capture and hold rainwater but are also permeable enough for rainwater to pass through them during heavy rains. EGR media mixtures should contain no more than about 20% organic matter (e.g., peat moss) by volume to reduce fire risk, minimize media shrinkage through decomposition, and avoid leaching of excess nutrients (e.g., nitrogen and phosphorus) in runoff (Fassman and Simcock 2012; Sailor and Hagos 2011). Inorganic EGR constituents can be divided into three production categories: (1) recycled (e.g., crushed brick), (2) unprocessed (e.g., pumice, red cinder, and sand), and (3) processed (e.g., perlite and vermiculite) (Ampim et al. 2010). Perlite and vermiculite are both mined minerals that undergo expansion through high-temperature heating (> 540 °C for vermiculite and > 850 °C for perlite). When expanded, both absorb large amounts of water relative to their weights and therefore are desirable EGR media constituents. The energy requirements for constituent production are often considered when formulating green roof media mixtures. From a life cycle perspective, green roofs with lower embodied energy, fewer maintenance requirements, and longer functional lifespans, will be the most environmentally beneficial, all else being equal (Carter and Keeler 2008; Getter et al. 2009; Kosareo and Ries 2007; Saiz et al. 2006). Consequently, if performance is substantially equal among different prospective EGR media, the media with the lowest embodied energy would be the preferred alternative.
EGR installations often use commercially available media with either proprietary or undefined formulations (Nagase and Dunnett 2011). A wide variety of substrate designs have been used on EGRs, and while performance data are available for a few specific formulations and cases (e.g., (Estrella 2016), there are no published comprehensive evaluations of how the hydrologic performance profiles of substrates vary. Similarly, there are no published systematic methods for selecting or designing EGR media for different applications. Recognizing the range of EGR media constituents available for aiding in stormwater management, the objectives of this study were to develop a framework to identify EGR media mixtures with optimal hydrologic performance, and to provide attribute data for these mixtures and for several commonly used EGR media constituents. We considered an EGR media with optimal hydrologic performance to be one that did not exceed accepted EGR weight standards, had moderate hydraulic conductivity, and absorbed and retained the maximum amount of water. Our study was intended to facilitate creating optimal designs for green roof construction in urban ecosystems that are economical to construct, efficiently engineered, and ecologically functional for controlling stormwater runoff and maintaining long-term plant viability.
2.0. Materials and Methods
In this three-phase study, we: 1) measured the dry weight, wet weight and amount of water held at maximum retentive capacity, long-term water retention, and hydraulic conductivity for individual EGR media constituents; 2) predicted those attributes for candidate media mixtures using data for the individual constituents and ranked mixtures based on their predicted hydrologic performance and their estimated embodied energy; and 3) tested the seven top-ranking mixtures and compared the test results to our predictions.
2.1. Phase 1: Attributes of individual media constituents
2.1.1. Media constituents
The six EGR media constituents tested in this study were peat moss, perlite, pumice, red cinder, river sand, and vermiculite. The peat moss was produced by Sun Gro Horticulture (Agawam, MA), and the perlite and vermiculite (Uni-Gro brand) were obtained from L&L Nursery Supply (San Bernardino, CA). The pumice and red cinder were obtained from a local landscape material supplier (The Bark Place Inc., Corvallis, OR), and the river sand was from a local sand and gravel company (Green & White Rock Products, Corvallis, OR). A natural sandy loam reference soil (“Newberg”, Green & White Rock Products) and a reference potting media (Sunshine Mix #1, Sun Gro Horticulture), were also included in the first phase of the study.
2.1.2. Wet weight and water held at maximum retentive capacity
Five replicate 4-inch (10.16 cm) diameter standard plastic pots (Anderson Die & Mfg. Co., Portland, OR) were used for each media. Media volume in each pot was 350 cm3. Pots had drainage holes in the bottom and were lined with a geotextile fabric (“weed cloth”) to contain the media. To fully wet the media, pots were placed in a large tub (Figure 1), the water level was brought up to just above the media surface, and pots were soaked for 7 days. Fully wetting the peat moss, perlite, Sunshine mix, and vermiculite was difficult because they were hydrophobic and tended to float. After soaking, each pot was removed from the water tub and allowed to drain for one minute and weighed to obtain the wet weight of the media at maximum retentive capacity. Water held was calculated as wet weight minus dry weight.
Figure 1.
Pots of individual extensive green roof growing media constituents and reference materials, from left to right: red cinder, Sunshine Mix #1, peat moss, “Newberg” sandy loam soil, river sand, vermiculite, pumice, and perlite.
2.1.3. Long-term water retention
Water retention over 25 days of drying was evaluated in a growth chamber (Conviron Model No. PGW36, Controlled Environments Limited, Winnipeg, Canada) set to a 16-hour light cycle at 20 °C, an 8-hour dark cycle at 10 °C, and 70% relative humidity. Light cycle photosynthetically active radiation, measured as photosynthetic photon flux density (LI-190 Quantum Sensor, LI-COR Inc., Lincoln, NE) was approximately 500 μmol s−1 m−2. After obtaining the wet weight at maximum retentive capacity, pots were moved to the growth chamber and re-weighed daily for 25 days. We selected 14 days as an appropriate time point at which to statistically compare long-term water retention among the media because water retention for all media began to converge toward zero with additional drying time.
2.1.4. Dry weight and water content
After 25 days in the growth chamber the pots of media were dried for 72 hours at 105 °C in a Blue M drying oven (Model POM-326E, Thermal Product Solutions, New Columbia, PA) and weighed. For each media, the dry weight, wet weight (WW), water held (WH), and daily water content weights were expressed on a standard weight per volume basis (gm/cm3). Values reported for each media are the average of the five replicate pots.
Following the wetting and drying cycle the peat moss, Sunshine mix, and vermiculite had notably less volume than at the beginning of the study. To assess the effects of these volume changes on the hydrologic attributes the pots were again soaked and weighed after draining for 1 minute. The volume of peat moss expanded with re-wetting to near the original wet volume and the volume of Sunshine mix partially expanded with re-wetting, but the volume of vermiculite did not change upon re-wetting and remained at about 50% of its original wet volume. To determine how this irreversible volume change for vermiculite affected 14-day water retention (WR14), the re-wetted pots were returned to the growth chamber and weighed after 14 days. From the first to the second wetting vermiculite WW decreased from 0.89 to 0.58 g/cm3, WH decreased from 0.73 to 0.42 g/cm3, and WR14 decreased from 0.17 to 0.006 g/cm3. Because WH and WR14 for the first wetting were transient conditions for vermiculite, we used data from the second wetting for those attributes. Although WW for the first wetting may also be a transient condition, it could still exceed the load limitations of the roof, so we used the greatest WW between the two wettings.
2.1.5. Hydraulic conductivity
The hydraulic conductivity (HC) of media was measured using a Gilson model HM-832 ASTM/AASHTO permeameter (Gilson Company, Inc., Lewis Center, OH) following ASTM method D 2434–68 (ASTM 2006). HC was measured on three different aliquots of media, where five repeat measurements were made on each aliquot and averaged. To facilitate interpretation of HC measurements we measured particle size distribution using standard sieves. Particle size categories were < 0.25 mm, 0.25 to 0.5 mm, 0.5 to 1 mm, 1 to 2 mm, 2 to 4 mm, 4 to 6.3 mm, 6.3 to 12.5 mm (¼ to ½ inch), and > 12.5 mm (> ½ inch).
2.2. Phase 2: Predicting media mixture attributes and evaluating performance
We predicted the performance of candidate EGR media mixtures by multiplying the proportion of the individual constituent in the mixture by the value for that constituent’s hydrologic attribute, and then summing those products for the mixture. We then evaluated the candidate mixtures for the following performance attributes: 1) wet weight (WW), 2) water held (WH), 3) water retained after 14 days of drying (WR14), and 4) proportion of processed inorganics (PPI). We estimated that the processed inorganic constituents we used (perlite and vermiculite) both had greater embodied energy than any of the unprocessed inorganic constituents we used (pumice, red cinder, and sand), so we considered PPI to be a surrogate for the embodied energy associated with the candidate mixtures. PPI was calculated by summing the proportions of perlite and/or vermiculite, if any, in the candidate mixtures. Dry weight (DW) and hydraulic conductivity (HC) were not used as evaluation factors when selecting mixtures for testing because DW will always be less than WW, and optimal HC may differ depending on the climatic conditions where the green roof is located. To generate an overall ranking, the 15 candidate mixtures with the greatest WH and WR14 were ranked in descending order. Mixtures that were in the top 15 for both WH and WR14 were then ranked in order of decreasing PPI, and a mean rank was calculated from the WH, WR14, and PPI rankings. Following screening for excessively high WW, the seven top-ranking mixtures were selected for testing.
To screen candidate mixtures for excessively high WW, predicted WW was calculated by simulating a realistic deployment scenario of modular trays containing media and plants that would be placed directly on a roof. The weight per area of a 45 kg weight capacity HDPE horticultural tray (Kadon Corp., Dayton, OH) lined with wetted weed cloth (0.81 gm/cm2) was added to the WW of the media (gm/cm3) multiplied by the media depth in the tray (9 cm) to get the total weight per area (gm/cm2) of the installation. Although the maximum EGR weight is dependent on the structural capability of any particular roof, a typical recommended load is approximately 9.76 g/cm2 (20 lbs/ft2), not including snow loads (BES 2013). To provide a conservative maximum weight limit for this study, mixtures with a calculated total installation WW of 9.5 g/cm2 (19.5 lbs/ft2) or greater were excluded from further testing.
2.3. Phase 3: Media mixture testing
Testing of media mixtures followed the same procedures as for the individual constituents, with exceptions. Media mixtures were pre-moistened before the 7-day tub soaking to facilitate wetting, and weighted plastic disks were used to keep the hydrophobic constituents from floating during soaking. We also conducted a seedling survival test to identify the permanent wilting point for each media and thus determine the lower limit of plant-available water. Gravimetric soil water content (wet weight minus dry weight) measures the total water present in the media, but it does not provide insight into the portion of the total water contained in the media that is available to plants (Cao et al. 2014). The most commonly installed green roof plants are Sedum spp., which are slow-growing drought-adapted succulents that store large amounts of water in their tissues and thus are insensitive to changes in soil moisture and would be poor test plants to use to predict wilting for most vascular plants. We selected lettuce (Lactuca sativa L.) as the test plant due to its rapid growth and uniform germination, size, and morphology. We prepared 10 replicate pots for each of the 7 top-ranking mixtures, with 5 replicates containing only media, and 5 replicates planted with 3 lettuce seeds each. The date that a seedling wilted without recovering was recorded for each of the lettuce seedlings, and the mean number of days to wilting (wilting date minus starting date) was calculated for each of the test mixtures. Water retention and wilting were evaluated in a growth chamber under the same environmental conditions as for the previous experiment using the individual constituents. All pots were weighed daily as before, but for 35 days because seedlings were still alive after 25 days. In addition to evaluating water retention at 14 days to compare to predicted values, we also evaluated water retention at 28 days to compare to seedling wilting data.
2.4. Statistical analysis
SAS/STAT software Version 9.4 of the SAS system for Windows was used to analyze the data. Analysis of variance (ANOVA) was used to fit a general linear model (GLM procedure) to the data. Least squares means (LS-means) were computed to evaluate effects among individual media constituents and among mixtures, and p-values were produced to determine significant differences between LS-means (a = 0.01). In evaluating interactions between independent variables (lettuce/non-lettuce) and dependent variables (mixtures), a significance level of 10% (a = 0.1) was used to be conservative. Because interactions were not significant, lettuce and non-lettuce data were pooled to evaluate mixture effects.
3.0. Results
3.1. Attributes of individual media constituents
3.1.1. Dry weight, wet weight, and water held
Of the individual media constituents, peat moss had the lowest dry weight (DW), followed by Sunshine mix, perlite, and vermiculite (Figure 2, Appendix A in Supplementary Materials). DW was significantly different among all constituents (p < 0.01) except that peat moss, perlite, and Sunshine mix were not significantly different from one another. Sand had the highest DW, followed by Newberg soil, red cinder, and pumice. Sand also had the highest wet weight (WW), followed by Newberg soil and pumice. WW was significantly different among all constituents except that peat moss was not significantly different from vermiculite, and red cinder was not significantly different from Sunshine mix. Perlite had the lowest WW because it held less water than peat moss. Water held was significantly different among all constituents except that pumice was not significantly different from Newberg soil, and sand was not significantly different from vermiculite. Red cinder held the least amount of water and, as a result, had the second lowest WW.
Figure 2.
Attributes of individual extensive green roof media constituents and reference materials (mean and standard error, n = 5). WR14 = water retained after 14 days of drying.
3.1.2. Long-term water retention
The media constituents with the greatest 14-day water retention (WR14) were perlite and pumice, respectively (Figure 2, Appendix A in Supplementary Materials). WR14 was significantly different among all constituents (p < 0.01), except that peat moss, Newberg sandy loam soil, and sand were not significantly different from one another, and vermiculite was not significantly different from red cinder. Daily water loss for all media constituents began to level off after about 8 days and approached zero after 25 days of drying in the growth chamber (Figure 3). Perlite retained the greatest amount of water from Day 8 onward. Pumice had a similar retention behavior to perlite, but with slightly less water. Peat moss, which had the greatest initial moisture content, lost moisture more quickly than most of the other constituents. Sunshine mix, composed primarily of peat moss (~80%), behaved similarly to peat moss but retained more water over time, probably due to the inclusion of perlite in the mix. The Newberg soil and the sand also behaved similarly to each other, and both retained less water after 3 days than all the other constituents except red cinder. The coarsely fragmented red cinder (see Figure 1) retained the least water and was depleted of water after 7 days. Red cinder is often used as an inorganic mulch over another type of growing media, and these results indicate that this may be the most appropriate use for this material in EGR installations.
Figure 3.
Mean 25-day water retention for individual extensive green roof media constituents and reference materials (n = 5). Vermiculite is not included in this comparison because water retention when first wetted was a transient condition (see Section 2.1.4). Error bars are omitted to improve visual clarity.
3.1.3. Hydraulic conductivity
Hydraulic conductivity (HC) varied from 0.08 to 30.15 cm/sec for the individual media constituents (Table 1), although only red cinder was greater than 0.5 cm/sec due to its large particle size (Figures 1 and 4, Appendix B in Supplementary Materials). After red cinder, perlite and pumice had the highest HC (0.46 and 0.37 cm/sec respectively), followed by vermiculite and peat moss (0.24 and 0.22 cm/sec respectively), and sand (0.08 cm/sec). HC was significantly different among all constituents (p < 0.01) except that perlite was not significantly different from pumice, and peat moss was not significantly different from vermiculite.
Table 1.
Hydraulic conductivity (HC) for individual extensive green roof media constituents (mean and standard error, n = 3). Standard errors are in brackets (). Media with different letters (a, b, c) are significantly different (p < 0.01).
| Media | HC (cm/sec) |
|---|---|
| Peat Moss | 0.22 (0.003)a |
| Perlite | 0.46 (0.062)b |
| Pumice | 0.37 (0.009)b |
| Red Cinder | 30.15 (2.776)c |
| Sand | 0.08 (0.002)d |
| Vermiculite | 0.24 (0.014)a |
Figure 4.
Particle size distribution for individual extensive green roof media constituents (mean and standard error, n = 3).
Particle size distribution (PSD) was measured on constituents as obtained, except for peat moss (obtained in a compressed bale), which was broken up though a 2 mm sieve before testing. HC was generally aligned with PSD in that media constituents with a higher percentage of large particles (> 1 mm) usually had higher HC (Table 1, Figure 4, Appendix B in Supplementary Materials). Vermiculite, however, did not follow this trend. For example, vermiculite had a higher percentage of large particles than pumice but had significantly lower HC.
3.1.4. Overall performance of individual constituents
To facilitate evaluation of the overall performance of the various media constituents the relative performance of the constituents for each of the attributes were compared using a radar graph (Figure 5). For example, the polygons for peat moss and perlite indicate more desirable overall performance given that lower dry and wet weights, lower hydraulic conductivity, more water held, and more water retained after 14 days of drying are desirable EGR media attributes.
Figure 5.
Radar graph of attributes of individual extensive green roof constituents. Data values are normalized on the radii for each of the attributes.
3.2. Media mixture composition and predicted performance ranking
Red cinder was not considered for candidate media mixture formulations because its water retention was so poor relative to its weight, and sand was not considered due to its high weight. Because organic matter is important for plant nutrition, peat moss was included at 20% by volume in all candidate mixture formulations. Consequently, all candidate mixtures were 20% peat moss and some combination of perlite, pumice, and vermiculite (Appendix C in Supplementary Materials). Water held (WH) and water retained after 14 days (WR14) were highly correlated among the candidate mixtures because perlite, pumice, and vermiculite had the same ranking (1st, 2nd, and 3rd, respectively) for both WH and WR14. Due to the poor performance of vermiculite relative to perlite and pumice following the second wetting, all seven top-ranked mixtures were 20% peat moss and combinations of perlite and pumice (Table 2). Mixtures with more perlite had higher WH and WR14 because perlite had the highest WH and WR14. Furthermore, mixtures with more perlite had the highest proportion of processed inorganics (PPI) because perlite is a processed inorganic and pumice is not. The PPI of all seven top-ranking mixtures was the same as the proportion of perlite because perlite was the only processed inorganic constituent in the mixtures. The seven mixtures had the same ranking for predicted WH and WR14, with an inverse ranking for PPI.
Table 2.
Composition and predicted ranking of the seven tested extensive green roof media mixtures. WR14 = water retained after 14 days of drying, PPI = proportion of processed inorganics.
| Mix | Composition (%) | Rank | ||||
|---|---|---|---|---|---|---|
| Peat Moss | Perlite | Pumice | Water Held | WR14 | PPI | |
| #9 | 20 | 80 | 0 | 1 | 1 | 7 |
| I | 20 | 70 | 10 | 2 | 2 | 6 |
| J | 20 | 60 | 20 | 3 | 3 | 5 |
| K | 20 | 50 | 30 | 4 | 4 | 4 |
| L | 20 | 40 | 40 | 5 | 5 | 3 |
| M | 20 | 30 | 50 | 6 | 6 | 2 |
| N | 20 | 20 | 60 | 7 | 7 | 1 |
3.3. Performance of tested media mixtures
3.3.1. Dry weight, wet weight, and water held
Consistent with calculated predictions, observed dry weight (DW) and wet weight (WW) for the tested mixtures increased with a greater proportion of pumice in the mixture (Figure 6, Appendix D in Supplementary Materials). Observed DW was significantly different among all mixtures (p < 0.001) and was slightly less than predicted. Observed WW was significantly different (p < 0.01) among all mixes except for #9 vs I, J vs K, K vs L, L vs M, and M vs N. Observed WW was somewhat greater than predicted, which resulted in observed water held (WH) also being greater than predicted. Although WH was predicted to drop slightly with a decreasing proportion of perlite in the mixture, there was no discernable trend in the observed values (Figure 6), in part because the range was so narrow (0.634 to 0.675 g/cm3). There were no significant differences in observed WH among any of the mixtures.
Figure 6.
Observed (O; mean and standard error, n = 10) and predicted (P) attributes of the seven tested extensive green roof media mixtures. DW = dry weight, WW= wet weight and WH = water held.
3.3.2. 14-day water retention
Across all seven tested mixtures, the non-lettuce pots had significantly greater 14-day water retention (WR14) than the lettuce pots (p < 0.001), likely due to transpiration of the lettuce seedlings (Table 3). There was no significant interaction between mixtures and lettuce vs. non-lettuce for WR14 (p = 0.472), so the lettuce and non-lettuce data were pooled for statistical comparison of the mixtures. Across both lettuce/non-lettuce groups, Mix #9 (most perlite, 80%) had significantly greater WR14 than all mixtures except Mix I (p < 0.001). Mix I (2nd most perlite, 70%) retained more water than Mix N, J, or M (p < 0.001). The observed WR14 rankings were generally consistent with the predicted ranking, although actual WR14 was consistently less than predicted, suggesting that all the media dried out more quickly in the second experiment.
Table 3.
Observed (mean and standard error, n = 5, n = 10 pooled) and predicted (P) 14-day water retention (WR14) for the seven tested extensive green roof media mixtures. Standard errors are in brackets (). Mixtures with different letters (a, b, c) are significantly different (p < 0.01). Mixtures are listed in descending order of their observed 14-day water retention for pooled lettuce and non-lettuce data.
| Mix | 14-Day Water Retention (g/cm3) | WR14 Rank | ||||||
|---|---|---|---|---|---|---|---|---|
| Observed | P | Observed | P | |||||
| Non-lettuce (NL) | Lettuce (L) | NL+L (pooled) | NL | L | NL+L | |||
| #9 | 0.094 (0.0043) | 0.083 (0.0021) | 0.089 (0.0030)a | 0.124 | 1 | 1 | 1 | 1 |
| I | 0.087 (0.0039) | 0.080 (0.0029) | 0.083 (0.0026)a,b | 0.122 | 2 | 2 | 2 | 2 |
| K | 0.077 (0.0034) | 0.075 (0.0027) | 0.076 (0.0021)b,c | 0.117 | 5 | 3 | 3 | 4 |
| L | 0.078 (0.0016) | 0.073 (0.0034) | 0.076 (0.0019)b,c | 0.115 | 4 | 4 | 4 | 5 |
| N | 0.079 (0.0025) | 0.065 (0.0022) | 0.074 (0.0029)c | 0.110 | 3 | 7 | 5 | 7 |
| J | 0.077 (0.0025) | 0.069 (0.0032) | 0.073 (0.0023)c | 0.119 | 6 | 5 | 6 | 3 |
| M | 0.076 (0.0019) | 0.068 (0.0018) | 0.072 (0.0019)c | 0.112 | 7 | 6 | 7 | 6 |
3.3.3. 28-day water retention and days to wilting
As with 14-day water retention (WR14), the non-lettuce pots across all seven tested mixtures had significantly greater 28-day water retention (WR28) than the lettuce pots (Table 4; p < 0.001), and there was no significant interaction between mixtures and lettuce vs. non-lettuce (p = 0.348). Mix #9 and Mix I had significantly greater WR28 than all the other mixtures (p < 0.01), similar to WR14. Lettuce wilted significantly sooner in Mix N (least perlite, 20%; most pumice, 60%) than in Mix I (p = 0.004), but there were no other statistically significant differences in days to wilting among any of the mixtures. The days to wilting were aligned with WR28 with respect to the generally poorer performance of Mix N relative to Mix I for the lettuce pots but were not consistent with the significantly greater WR28 observed for Mix #9 (Table 4). The average days to wilting for the mixtures other than Mix N only ranged from 29 to 31 days, suggesting that among these mixtures, composition was not a significant influence on plant survival despite statistical differences in water retention.
Table 4.
Observed (mean and standard error, n = 5, n = 10 pooled) 28-day water retention (WR28) and days to wilting (DtW) for the seven tested extensive green roof media mixtures. Standard errors are in brackets (). Mixtures with different letters (a, b) are significantly different (p < 0.01). Mixtures are listed in descending order of their observed 28-day water retention for lettuce pots.
| Mix | 28-Day Water Retention (g/cm3) | WR28 Rank | Days to Wilting | DtW Rank | ||||
|---|---|---|---|---|---|---|---|---|
| Non-lettuce (NL) | Lettuce (L) | NL+L (pooled) | NL | L | NL+L | |||
| #9 | 0.0233 (0.00128) | 0.0115 (0.00175) | 0.0174 (0.00222)a | 1 | 1 | 1 | 30 (0.7)a,b | 5 |
| I | 0.0218 (0.00285) | 0.0099 (0.00166) | 0.0158 (0.00253)a | 2 | 2 | 2 | 31 (0.5)a | 1 |
| K | 0.0143 (0.00199) | 0.0079 (0.00117) | 0.0111 (0.00152)b | 5 | 3 | 4 | 30 (0.7)a,b | 2 |
| J | 0.0144 (0.00130) | 0.0063 (0.00068) | 0.0103 (0.00152)b | 4 | 4 | 6 | 30 (0.5)a,b | 4 |
| L | 0.0140 (0.00067) | 0.0061 (0.00052) | 0.0105 (0.00144)b | 7 | 5 | 5 | 30 (0.9)a,b | 3 |
| N | 0.0155 (0.00168) | 0.0060 (0.00035) | 0.0117 (0.00183)b | 3 | 6 | 3 | 28 (0.7)b | 7 |
| M | 0.0140 (0.00109) | 0.0057 (0.00029) | 0.0099 (0.00147)b | 6 | 7 | 7 | 29 (0.5)a,b | 6 |
3.3.4. Hydraulic conductivity
Observed hydraulic conductivity (HC) among the tested mixtures varied narrowly from 0.166 to 0.198 cm/sec (Table 5), and there were no significant differences in HC among any of the tested mixtures (p < 0.01). Observed HC in mixtures was only about half of what was predicted using the HC of the individual constituents, and the HCs of all the tested mixtures were lower than the HC of any of the individual constituents that made up the mixtures, suggesting that mixture HC cannot be calculated as a simple proportion of the constituent HCs. Also, mixtures with more perlite than pumice were predicted to have higher HC because perlite had higher HC than pumice, but the two mixtures with the most pumice (M and N) had the highest observed HC. Mixtures with more pumice had a more even particle size distribution (PSD) because pumice had a more even PSD than perlite.
Table 5.
Observed (mean and standard error, n = 3) and predicted (P) hydraulic conductivity (HC) and particle size distribution (PSD, mean and standard error, n = 3) for the seven tested extensive green roof media mixtures. Standard errors are in brackets (). There were no significant differences in HC among any of the tested mixtures (p > 0.01). Mixtures are listed in descending order of their predicted HC. Mixture PSD was calculated from the PSD of individual constituents.
| Mix | HC (cm/sec) | Particle Size Distribution (% by weight, classes in mm) | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Observed | P | < 0.25 | 0.25–0.5 | 0.5–1 | 1–2 | 2–4 | 4–6.3 | 6.3–12.5 | |
| #9 | 0.185 (0.0203) | 0.413 | 7 (1.0) | 11 (0.7) | 8 (1.0) | 30 (1.5) | 35 (2.1) | 8 (1.2) | 1 (0.0) |
| I | 0.169 (0.0191) | 0.403 | 7 (1.0) | 12 (0.7) | 10 (1.0) | 29 (1.3) | 33 (1.9) | 9 (1.1) | 1 (0.1) |
| J | 0.166 (0.0272) | 0.394 | 8 (1.0) | 13 (0.8) | 11 (1.0) | 27 (1.1) | 30 (1.7) | 9 (1.1) | 2 (0.2) |
| K | 0.172 (0.0154) | 0.385 | 8 (1.0) | 14 (0.8) | 13 (1.0) | 26 (0.9) | 28 (1.5) | 9 (1.1) | 2 (0.3) |
| L | 0.166 (0.0318) | 0.376 | 9 (1.0) | 15 (0.9) | 14 (1.0) | 25 (0.8) | 26 (1.4) | 9 (1.0) | 2 (0.4) |
| M | 0.198 (0.0317) | 0.367 | 10 (1.0) | 16 (1.0) | 15 (0.9) | 24 (0.6) | 24 (1.3) | 9 (1.0) | 3 (0.6) |
| N | 0.196 (0.0309) | 0.357 | 10 (1.0) | 17 (1.0) | 17 (0.9) | 22 (0.4) | 22 (1.4) | 9 (1.0) | 3 (0.7) |
3.3.5. Overall performance of tested media mixtures
The wet weight, water held, water retained after 14 days of drying, and hydraulic conductivity were similar among the tested mixtures (Figure 7). Although mixtures with more perlite than pumice had slightly lower wet weight and slightly greater 14-day water retention, the mixtures were nonetheless all relatively similar for these metrics. The major differences among the mixtures were that low dry weight and low percentage of processed inorganics were inversely proportional, depending of the relative amounts on perlite versus pumice in the mixtures.
Figure 7.
Radar graph of attributes of the seven tested extensive green roof media mixtures. Data values are normalized on the radii for each of the attributes.
4.0. Discussion
Our study elucidates characteristics of common EGR media constituents that are important criteria in EGR media selection and design. The attribute data for the media constituents can be used in formulating EGR media mixtures for specific applications. For example, regions with different climate regimes may require media with different hydrologic characteristics to accommodate seasonal temperature and/or precipitation extremes (Georgescua et al. 2014; Kazemi and Mohorko 2017; Lundholm et al. 2014). In addition, the attribute data for the media constituents and mixtures can be used as benchmarks against which to evaluate other EGR media.
We found that EGR media mixtures dominated by perlite had the best overall hydrologic performance because perlite held and retained the greatest amount of water among the inorganic media constituents considered. However, days to wilting showed minimal variation among the tested mixtures, indicating that the slightly greater water retention by perlite did not translate into a significantly longer time to wilting. Mixtures with up to 60% pumice also performed well, although they were slightly heavier than mixtures with more perlite. Nonetheless, EGR installations using pumice as described in this study would still be within typical roof load limitation guidelines. Despite the slightly lower hydrologic performance and slightly greater weight, pumice may be a preferred alternative to perlite because perlite is a processed constituent with greater embodied energy. EGR media constituents with greater embodied energy should be avoided if doing so will not detract substantially from media performance. In addition, the tendency of perlite to float when wetted after being in a dry condition may limit its utility as an EGR constituent. Both peat moss and perlite were hydrophobic and tended to float when dry, although once fully wetted they held the greatest amount of water. However, the success of hydrophobic materials as EGR growing media constituents will depend on the ability to fully wet them despite their hydrophobic tendencies and will likely require designs that will keep them contained throughout wetting/drying cycles.
Tested mixtures generally performed as predicted for wet weight and water held at maximum retentive capacity, and water retained after 14 days of drying, indicating performance of mixtures can be adequately predicted for these attributes using values calculated from data for individual constituents. Hydraulic conductivity (HC) was less predictable, suggesting that media mixture attributes to be predicted from individual constituent attributes should be evaluated by testing actual performance to determine the suitability of this approach for any specific attribute. One explanation for the inability to accurately predict the HC of a mixture using the HC of the individual constituents is that the total volume of pores (voids between particles) in mixtures can be less than the total volume of pores in any of the individual constituents making up the mixture. A mixture of large and small particles will have lower porosity than the large particles alone because the pores among the large particles will be filled with small particles, reducing the pore volume. The mixture will also have lower porosity than the small particles alone because there are no pores within the large particles themselves, also reducing the pore volume.
HC was not well correlated with particle size distribution (PSD) for our media mixtures because the two mixtures with the highest HC had the highest percentages of small (< 1 mm) particles). However, these two mixtures also had slightly higher percentages of the largest particles. Similarly, for the individual constituents, vermiculite had a higher percentage of large particles than pumice but had significantly lower HC. A lack of correlation between HC and PSD has been reported previously for some of these materials, where particle size ranges for vermiculite (0.5–2 mm), perlite (0.25–1 mm), and sand (0.25–1 mm) were not proportional to HCs of 0.06, 0.02, and 0.11 cm/sec, respectively (Vijayaraghavan and Raja 2014).
All our HC values were much greater than reported values for agricultural soils. For example, a sandy soil with 88% sand, 7% silt, and 5% clay had a HC of about 0.003 cm/sec, more than an order of magnitude less than the river sand we tested (HC = 0.08 cm/sec) and was the soil type with the highest reported HC (Saxton and Rawls 2006). Among other soil materials, gravels have HCs of between 10 and 0.1 cm/sec, coarse sand between 0.1 and 0.01 cm/sec, and fine sand between 0.01 and 0.0001 cm/sec (Klute 1986). Moderate EGR media HCs (e.g., between 1.0 and 0.01 cm/sec) may provide the best balance between preventing standing water formation and maximizing runoff delay time.
Our research approach was intended to provide a framework to evaluate the performance of other EGR media. Designing EGR substrates that provide optimal plant available water while increasing overall annual stormwater retention will help to maximize the ecological benefits provided by EGRs and are thus key objectives for many green infrastructure projects. Although the framework described here helps to meet those objectives by informing EGR media selection and design, some limitations exist. For example, the inability to accurately predict the HC of a mixture using the HC of the individual constituents identifies a potential weakness with predicting mixture behavior using data for individual constituents. Attributes to be predicted in this manner should be evaluated by testing actual performance to determine the suitability of this approach for any specific attribute. Also, the relatively poor performance of vermiculite due to the irreversible volume reduction following a wetting and drying cycle underscores the need for standardization of methods for more effective comparisons of EGR substrates and reinforces the need to evaluate EGR components using real-world scenarios that consider, for example, the effects of repeated wet/dry or freeze/thaw cycles (Bates et al. 2015; Lundholm et al. 2014). The framework could also be modified to address additional EGR design objectives by including other media evaluation factors, such as albedo (Santamouris 2014), thermal properties (Sailor and Hagos 2011; Sailor et al. 2008), nutrient and chemical capture or leaching (Gregoire and Clausen 2011; Harada et al. 2018; Solano et al. 2012), and suitability for growing native plants (Dvorak and Volder 2010). Similarly, there are many diverse EGR media types, including natural and man-made products and recycled or waste materials such as biochar, fly ash, crushed porcelain, and foamed glass (Ampim et al. 2010; Cao et al. 2014; Matlock and Rowe 2016; Molineux et al. 2009), that were beyond the scope of this study to include.
Evaluating trade-offs between EGRs and other uses of roof space that might reduce the footprint of the built environment should be a principal objective of EGR design. One of the primary threats to ecosystem integrity is the loss of natural areas to residential, agricultural, and other human development. Using roof space for placement of solar panels, wind turbines, agricultural crops, greenhouses, horticultural nurseries, and other uses could reduce the amount of land required for those developments. On a small scale, preserving or restoring functioning native ecosystems on the ground is likely to produce a greater net ecological benefit than attempting to construct such ecosystems on building roofs where the harsh environmental conditions could hamper or preclude success. However, the stormwater retention potential of EGRs on a large scale may provide significant ecological benefits to downstream native ecosystems (Carter and Jackson 2007) and EGRs in heavily urbanized areas may provide critical migration corridors and refugia for birds and insects (Williams et al. 2014). Identifying characteristics of various EGR substrate components, separately and in mixtures, may clarify trade-offs between physical and hydraulic properties and ecological benefits, and associated embodied energy, carbon footprints, and construction costs. With such information, we can more effectively tailor EGR designs to achieve desired outcomes with optimal cost-benefit ratios.
5.0. Acknowledgements
The views expressed in this article are those of the author(s) and do not necessarily represent the views or policies of the U.S. Environmental Protection Agency. Any mention of trade names, products, or services does not imply an endorsement by the U.S. Government or the U.S. Environmental Protection Agency. The EPA does not endorse any commercial products, services, or enterprises. DeSantis, DuChanois, and Etten-Bohm were supported by EPA Greater Research Opportunity Fellowships. Mark G. Johnson provided invaluable advice regarding soil physics, and Olyssa Starry provided constructive comments on manuscript preparation.
Supplementary Materials
Appendix A.
Attributes of the six individual extensive green roof media constituents and two reference materials (mean and standard error, n = 5). Standard errors are in brackets (). DW = dry weight, WW = wet weight, WH = water held, and WR14 = water retained after 14 days of drying. Media with different letters within a column (a, b, c) are significantly different (p < 0.01).
| Media | DW (g/cm3) | WW (g/cm3) | WH (g/cm3) | WR14 (g/cm3) |
|---|---|---|---|---|
| Peat Moss | 0.08 (0.001)a | 0.91 (0.010)a | 0.83 (0.010)a | 0.055 (0.0013)a |
| Perlite | 0.11 (0.003)a | 0.69 (0.008)b | 0.58 (0.006)b | 0.141 (0.0031)b |
| Pumice | 0.46 (0.012)b | 0.99 (0.010)c | 0.53 (0.011)c | 0.117 (0.0049)c |
| Red Cinder | 0.59 (0.021)c | 0.80 (0.019)d | 0.20 (0.006)d | 0.002 (0.0003)d |
| Sand | 1.33 (0.005)d | 1.77 (0.006)e | 0.44 (0.003)e | 0.047 (0.0008)a |
| Vermiculite | 0.16 (0.003)e | 0.89 (0.008)a | 0.42 (0.005)e | 0.006 (0.0003)d |
| Newberg Soil | 1.11 (0.008)f | 1.65 (0.010)f | 0.54 (0.004)c | 0.053 (0.0021)a |
| Sunshine Mix | 0.09 (0.001)a | 0.83 (0.009)d | 0.74 (0.008)f | 0.093 (0.0022)e |
Appendix B.
Particle size distribution for individual extensive green roof media constituents (mean and standard error, n = 3). Standard errors are in brackets ().
| Particle Size Distribution (% by weight, size classes in mm) | |||||||
|---|---|---|---|---|---|---|---|
| < 0.25 | 0.25–0.5 | 0.5–1 | 1–2 | 2–4 | 4–6.3 | 6.3–12.5 | > 12.5 |
| 9 (0.3) | 38 (1.2) | 24 (0.6) | 29 (2.1) | 0 NA | 0 NA | 0 NA | 0 NA |
| 6 (1.2) | 4 (0.7) | 4 (1.1) | 30 (2.2) | 44 (2.6) | 11 (1.4) | 1 (0.0) | 0 NA |
| 12 (1.2) | 14 (1.2) | 18 (0.9) | 18 (0.1) | 22 (2.0) | 11 (1.1 | 5 (1.1) | 0 NA |
| 0 NA | 0 NA | 0 NA | 0 NA | 0 NA | 0 NA | 0 NA | 100 (0.0) |
| 3 (0.3) | 41 (1.0) | 31 (0.5) | 15 (0.7) | 9 (0.7) | 0 NA | 0 NA | 0 NA |
| 0.5 (0.002) | 9 (2.1) | 12 (1.6) | 24 (0.5) | 37 (2.1) | 14 (1.6) | 4 (0.6) | 0 NA |
Appendix C.
Predicted candidate extensive green roof media mixture attributes and performance rankings. PM = peat moss, Pum = pumice, Per = perlite, Ver = vermiculite, PPI = proportion of processed inorganics (perlite + vermiculite), DW = dry weight, WW = wet weight, WH = water held, WR14 = water retained after 14 days of drying, HC = hydraulic conductivity. *Organic matter should be less than about 0.2 (20%) (Fassman et al 2012). #TI WW = Total installation wet weight (see Materials and Methods). Bold Type = Selected for further testing.
| Mix | Constituent Fraction | PPI | g/cm3 | Rank | TI WW# (g/cm2) | HC (cm/sec) | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PM* | Pum | Per | Ver | DW | WW | WH | WR14 | WH | WR14 | PPI | ||||
| #8 | 0.2 | 0 | 0 | 0.8 | 0.8 | 0.142 | 0.890 | 0.502 | 0.016 | 8.82 | 0.233 | |||
| A | 0.2 | 0.1 | 0 | 0.7 | 0.7 | 0.172 | 0.901 | 0.513 | 0.027 | 8.92 | 0.246 | |||
| B | 0.2 | 0.2 | 0 | 0.6 | 0.6 | 0.203 | 0.911 | 0.524 | 0.038 | 9.02 | 0.259 | |||
| C | 0.2 | 0.3 | 0 | 0.5 | 0.5 | 0.233 | 0.922 | 0.535 | 0.049 | 9.11 | 0.273 | |||
| D | 0.2 | 0.4 | 0 | 0.4 | 0.4 | 0.263 | 0.933 | 0.546 | 0.060 | 9.21 | 0.286 | |||
| E | 0.2 | 0.5 | 0 | 0.3 | 0.3 | 0.294 | 0.944 | 0.558 | 0.072 | 9.31 | 0.299 | |||
| F | 0.2 | 0.6 | 0 | 0.2 | 0.2 | 0.324 | 0.954 | 0.56870 | 0.083 | 15 | 9.40 | 0.312 | ||
| G | 0.2 | 0.7 | 0 | 0.1 | 0.1 | 0.354 | 0.965 | 0.57989 | 0.094 | 14 | 12 | 2 | 9.50 | 0.326 |
| H | 0.2 | 0.8 | 0 | 0 | 0 | 0.385 | 0.976 | 0.591 | 0.105 | 11 | 10 | 1 | 9.60 | 0.339 |
| #9 | 0.2 | 0 | 0.8 | 0 | 0.8 | 0.103 | 0.730 | 0.627 | 0.124 | 1 | 1 | 9 | 7.38 | 0.413 |
| I | 0.2 | 0.1 | 0.7 | 0 | 0.7 | 0.138 | 0.760 | 0.622 | 0.122 | 2 | 2 | 8 | 7.66 | 0.403 |
| J | 0.2 | 0.2 | 0.6 | 0 | 0.6 | 0.173 | 0.791 | 0.618 | 0.119 | 3 | 3 | 7 | 7.93 | 0.394 |
| K | 0.2 | 0.3 | 0.5 | 0 | 0.5 | 0.208 | 0.822 | 0.613 | 0.117 | 4 | 4 | 6 | 8.21 | 0.385 |
| L | 0.2 | 0.4 | 0.4 | 0 | 0.4 | 0.244 | 0.853 | 0.609 | 0.115 | 6 | 5 | 5 | 8.49 | 0.376 |
| M | 0.2 | 0.5 | 0.3 | 0 | 0.3 | 0.279 | 0.884 | 0.605 | 0.112 | 7 | 6 | 4 | 8.77 | 0.367 |
| N | 0.2 | 0.6 | 0.2 | 0 | 0.2 | 0.314 | 0.914 | 0.600 | 0.110 | 8 | 8 | 3 | 9.04 | 0.357 |
| O | 0.2 | 0.7 | 0.1 | 0 | 0.1 | 0.350 | 0.945 | 0.59555 | 0.107 | 10 | 9 | 2 | 9.32 | 0.348 |
| P | 0.2 | 0 | 0.7 | 0.1 | 0.8 | 0.108 | 0.750 | 0.611 | 0.111 | 5 | 7 | 9 | 7.56 | 0.390 |
| Q | 0.2 | 0 | 0.6 | 0.2 | 0.8 | 0.113 | 0.770 | 0.59557 | 0.097 | 9 | 11 | 9 | 7.74 | 0.368 |
| R | 0.2 | 0 | 0.5 | 0.3 | 0.8 | 0.117 | 0.790 | 0.57991 | 0.084 | 13 | 14 | 9 | 7.92 | 0.345 |
| #7 | 0.2 | 0 | 0.4 | 0.4 | 0.8 | 0.122 | 0.810 | 0.564 | 0.070 | 8.10 | 0.323 | |||
| S | 0.2 | 0 | 0.3 | 0.5 | 0.8 | 0.127 | 0.830 | 0.549 | 0.057 | 8.28 | 0.300 | |||
| T | 0.2 | 0 | 0.2 | 0.6 | 0.8 | 0.132 | 0.850 | 0.533 | 0.043 | 8.46 | 0.278 | |||
| U | 0.2 | 0 | 0.1 | 0.7 | 0.8 | 0.137 | 0.870 | 0.517 | 0.029 | 8.64 | 0.255 | |||
| V | 0.2 | 0.1 | 0.1 | 0.6 | 0.7 | 0.167 | 0.881 | 0.528 | 0.041 | 8.74 | 0.268 | |||
| W | 0.2 | 0.15 | 0.15 | 0.5 | 0.65 | 0.180 | 0.876 | 0.542 | 0.053 | 8.70 | 0.286 | |||
| X | 0.2 | 0.2 | 0.2 | 0.4 | 0.6 | 0.193 | 0.871 | 0.555 | 0.065 | 8.66 | 0.304 | |||
| Y | 0.2 | 0.25 | 0.25 | 0.3 | 0.55 | 0.206 | 0.867 | 0.56871 | 0.078 | 15 | 8.61 | 0.322 | ||
| Z | 0.2 | 0.3 | 0.3 | 0.2 | 0.5 | 0.218 | 0.862 | 0.582 | 0.090 | 12 | 13 | 6 | 8.57 | 0.340 |
Appendix D.
Observed (O; mean and standard error, n = 10) and predicted (P) attributes of the seven tested extensive green roof media mixtures. Standard errors are in brackets (). WH = water held. Mixtures with different letters (a, b, c) are significantly different (p < 0.01). Dry weight was significantly different among all mixtures, and there were no significant differences in WH among any of the mixtures. Mixtures are listed in ascending order of their dry/wet weights.
| Mix | Dry Weight (g/cm3) | Wet Weight (g/cm3) | Water Held (g/cm3) | WH Rank | ||||
|---|---|---|---|---|---|---|---|---|
| O | P | O | P | O | P | O | P | |
| #9 | 0.100 (0.0017) | 0.103 | 0.743 (0.0124)a | 0.730 | 0.644 (0.0123) | 0.627 | 6 | 1 |
| I | 0.135 (0.0038) | 0.138 | 0.768 (0.0129)a | 0.760 | 0.634 (0.0129) | 0.622 | 7 | 2 |
| J | 0.165 (0.0035) | 0.173 | 0.838 (0.0144)b | 0.791 | 0.674 (0.0129) | 0.618 | 2 | 3 |
| K | 0.199 (0.0034) | 0.208 | 0.873 (0.0165)b,c | 0.822 | 0.675 (0.0152) | 0.613 | 1 | 4 |
| L | 0.231 (0.0028) | 0.244 | 0.900 (0.0151)c,d | 0.853 | 0.670 (0.0162) | 0.609 | 3 | 5 |
| M | 0.277 (0.0017) | 0.279 | 0.945 (0.0099)d,e | 0.884 | 0.668 (0.0101) | 0.605 | 4 | 6 |
| N | 0.302 (0.0029) | 0.314 | 0.967 (0.0138)e | 0.914 | 0.665 (0.0159) | 0.600 | 5 | 7 |
Footnotes
Declarations of interest: none
References
- Ampim PAY; Sloan JJ; Cabrera RI; Harp DA; Jaber FH Green roof growing substrates: Types, ingredients, composition and properties. Journal of Environmental Horticulture. 28:244–252; 2010 [Google Scholar]
- ASTM. Standard Test Method for Permeability of Granular Soils (Constant Head). West Conshohocken, PA: American Society for Testing and Materials (ASTM) International; 2006 [Google Scholar]
- Bates AJ; Sadler JP; Greswell RB; Mackay R Effects of recycled aggregate growth substrate on green roof vegetation development: A six year experiment. Landscape and Urban Planning. 135:22–31; 2015 [Google Scholar]
- Berghage RD; Beattie D; Jarrett AR; Thuring C; Razaei F; O’Connor TP Green Roofs for Stormwater Runoff Control. Cincinnati, OH: U.S. Environmental Protection Agency, Office Of Research And Development, National Risk Management Research Laboratory; 2009 [Google Scholar]
- Berndtsson JC Green roof performance towards management of runoff water quantity and quality: A review. Ecological Engineering. 36:351–360; 2010 [Google Scholar]
- Berretta C; Poë S; Stovin V Moisture content behaviour in extensive green roofs during dry periods: The influence of vegetation and substrate characteristics. Journal of Hydrology. 511:374–386; 2014 [Google Scholar]
- BES. BES Red Cinder Ecoroof Design Guidelines. Portland, OR; 2013 [Google Scholar]
- Cao CTN; Farrell C; Kristiansen PE; Rayner JP Biochar makes green roof substrates lighter and improves water supply to plants. Ecological Engineering. 71:368–374; 2014 [Google Scholar]
- Carter T; Jackson CR Vegetated roofs for stormwater management at multiple spatial scales. Landscape and Urban Planning. 80:84–94; 2007 [Google Scholar]
- Carter T; Keeler A Life-cycle cost–benefit analysis of extensive vegetated roof systems. Journal of Environmental Management. 87:350–363; 2008 [DOI] [PubMed] [Google Scholar]
- Dvorak B; Volder A Green roof vegetation for North American ecoregions: A literature review. Landscape and Urban Planning. 96:197–213; 2010 [Google Scholar]
- Emilsson T Vegetation development on extensive vegetated green roofs: Influence of substrate composition, establishment method and species mix. Ecological Engineering. 33:265–277; 2008 [Google Scholar]
- Estrella M Variation in Green Roof Storage Capacity, Associated Drivers, and Implications for Stormwater Management in Portland, Oregon Civil and Environmental Engineering. Portland, OR: Portland State University; 2016 [Google Scholar]
- Fassman E; Simcock R Moisture measurements as performance criteria for extensive living roof substrates. Journal of Environmental Engineering. 138:841–851; 2012 [Google Scholar]
- Georgescua M; Morefield PE; Bierwagen BG; Weaver CP Urban adaptation can roll back warming of emerging megapolitan regions. Proceedings of the National Academy of Sciences of the United States of America. 111:2909–2914; 2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Getter KL; Rowe DB; Robertson GP; Cregg BM; Andresen JA Carbon sequestration potential of extensive green roofs. Environmental Science & Technology. 43:7564–7570; 2009 [DOI] [PubMed] [Google Scholar]
- Graceson A; Hare M; Monaghan J; Hall N The water retention capabilities of growing media for green roofs. Ecological Engineering. 61, Part A:328–334; 2013 [Google Scholar]
- Gregoire BG; Clausen JC Effect of a modular extensive green roof on stormwater runoff and water quality. Ecological Engineering. 37:963–969; 2011 [Google Scholar]
- Griffin WN; Cohan SM; Lea-Cox JD; Ristvey AG Green roof substrate composition affects Phedimus Kamtschaticus growth and substrate water content under controlled environmental conditions. HortScience. 52:320–325; 2017 [Google Scholar]
- Harada Y; Whitlow TH; Templer PH; Howarth RW; Walter MT; Bassuk NL; Russell-Anelli J Nitrogen biogeochemistry of an urban rooftop farm. Frontiers in Ecology and Evolution. 6; 2018 [Google Scholar]
- Kazemi F; Mohorko R Review on the roles and effects of growing media on plant performance in green roofs in world climates. Urban Forestry & Urban Greening. 23:13–26; 2017 [Google Scholar]
- Klute A ed. Methods of Soil Analysis. Part 1. Physical and Mineralogical Methods. Madison, WI: American Society of Agronony, Soil Science Society of America; 1986 [Google Scholar]
- Kosareo L; Ries R Comparative environmental life cycle assessment of green roofs. Building and Environment. 42:2606–2613; 2007 [Google Scholar]
- Lundholm JT; Weddle BM; MacIvor JS Snow depth and vegetation type affect green roof thermal performance in winter. Energy and Buildings. 84:299–307; 2014 [Google Scholar]
- Matlock JM; Rowe DB The suitability of crushed porcelain and foamed glass as alternatives to heat-expanded shale in green roof substrates: An assessment of plant growth, substrate moisture, and thermal regulation. Ecological Engineering. 94:244–254; 2016 [Google Scholar]
- Metselaar K Water retention and evapotranspiration of green roofs and possible natural vegetation types. Resources, Conservation and Recycling. 64:49–55; 2012 [Google Scholar]
- Molineux CJ; Fentiman CH; Gange AC Characterising alternative recycled waste materials for use as green roof growing media in the U.K. Ecological Engineering. 35:1507–1513; 2009 [Google Scholar]
- Nagase A; Dunnett N Drought tolerance in different vegetation types for extensive green roofs: Effects of watering and diversity. Landscape and Urban Planning. 97:318–327; 2010 [Google Scholar]
- Nagase A; Dunnett N The relationship between percentage of organic matter in substrate and plant growth in extensive green roofs. Landscape and Urban Planning. 103:230–236; 2011 [Google Scholar]
- Nagase A; Dunnett N Amount of water runoff from different vegetation types on extensive green roofs: Effects of plant species, diversity and plant structure. Landscape and Urban Planning. 104:356–363; 2012 [Google Scholar]
- Nawaz R; McDonald A; Postoyko S Hydrological performance of a full-scale extensive green roof located in a temperate climate. Ecological Engineering. 82:66–80; 2015 [Google Scholar]
- Oberndorfer E; Lundholm J; Bass B; Coffman RR; Doshi H; Dunnett N; Gaffin S; Köhler M; Liu KKY; Rowe B Green roofs as urban ecosystems: Ecological structures, functions, and services. BioScience. 57:823–833; 2007 [Google Scholar]
- Pitman SD; Daniels CB; Ely ME Green infrastructure as life support: Urban nature and climate change. Transactions of the Royal Society of South Australia. 139:97–112; 2015 [Google Scholar]
- Sailor DJ; Hagos M An updated and expanded set of thermal property data for green roof growing media. Energy and Buildings. 43:2298–2303; 2011 [Google Scholar]
- Sailor DJ; Hutchinson D; Bokovoy L Thermal property measurements for ecoroof soils common in the western U.S. Energy and Buildings. 40:1246–1251; 2008 [Google Scholar]
- Saiz S; Kennedy C; Bass B; Pressnail K Comparative life cycle assessment of standard and green roofs. Environmental Science & Technology. 40:4312–4316; 2006 [DOI] [PubMed] [Google Scholar]
- Santamouris M Cooling the cities – A review of reflective and green roof mitigation technologies to fight heat island and improve comfort in urban environments. Solar Energy. 103:682–703; 2014 [Google Scholar]
- Saxton KE; Rawls WJ Soil water characteristic estimates by texture and organic matter for hydrologic solutions. Soil Science Society of America Journal. 70:1569–1578; 2006 [Google Scholar]
- Schroll E; Lambrinos J; Righetti T; Sandrock D The role of vegetation in regulating stormwater runoff from green roofs in a winter rainfall climate. Ecological Engineering. 37:595–600; 2011 [Google Scholar]
- Solano L; Ristvey AG; Lea-Cox JD; Cohan SM Sequestering zinc from recycled crumb rubber in extensive green roof media. Ecological Engineering. 47:284–290; 2012 [Google Scholar]
- Spolek G Performance monitoring of three ecoroofs in Portland, Oregon. Urban Ecosyst. 11:349–359; 2008 [Google Scholar]
- Stovin V; Poë S; De-Ville S; Berretta C The influence of substrate and vegetation configuration on green roof hydrological performance. Ecological Engineering. 85:159–172; 2015 [Google Scholar]
- Thuring CE; Dunnett N Vegetation composition of old extensive green roofs (from 1980s Germany). Ecological Processes. 3:1–11; 2014 [Google Scholar]
- USEPA. Managing Wet Weather with Green Infrastructure: Action Strategy 2008. USEPA, Office of Water, National Pollutant Discharge Elimination System (NPDES); 2008 [Google Scholar]
- Vijayaraghavan K; Raja FD Design and development of green roof substrate to improve runoff water quality: Plant growth experiments and adsorption. Water Research. 63:94–101; 2014 [DOI] [PubMed] [Google Scholar]
- Williams NSG; Lundholm J; Scott MacIvor J FORUM: Do green roofs help urban biodiversity conservation? Journal of Applied Ecology. 51:1643–1649; 2014 [Google Scholar]