CO2 Payoff of Extensive Green Roofs with Different Vegetation Species

Green roofs are considered effective in the reduction of atmospheric CO2 because of their ability to reduce energy consumption of buildings and sequester carbon in plants and substrates. However, green roof system components (substrate, water proofing membrane, etc.) may cause CO2 emissions during their life cycle. Therefore, to assess the CO2-payoff for extensive green roofs, we calculated CO2 payback time it takes their CO2 sequestration and reduction to offset the CO2 emitted during its production process and maintenance practices. The amount of CO2 emitted during the production of a modular green roof system was found to be 25.2 kg-CO2·m. The annual CO2 emission from the maintenance of green roofs was 0.33 kg-CO2·m·yr. Annual CO2 sequestration by three grass species with irrigation treatment was about 2.5 kg-CO2·m·yr, which was higher than that of Sedum aizoon. In the hypothetical green roofs, annual CO2 reduction due to saved energy was between 1.703 and 1.889 kg-CO2·m·yr. From these results, we concluded that the CO2 payback time of the extensive green roofs was between 5.8 and 15.9 years, which indicates that extensive green roofs contribute to CO2 reduction within their lifespan.


Introduction
Global warming caused by the increase of greenhouse gases in the atmosphere has been identified as one of the most important environmental issues currently faced by human civilization. Carbon dioxide (CO 2 ) is known to be a primary anthropogenic greenhouse gas. In urban areas, according to the Intergovernmental Panel on Climate Change [1], climate change is projected to increase environmental risks for people, assets, economies and ecosystems, including risks from heat stress, storms and extreme precipitation, inland and coastal flooding, landslides, air pollution, drought, water scarcity, sea level rise and storm surges. Increasing urban greenspaces is one proposed method of mitigating these problems through urban planning [2][3][4][5][6].
Greenspace in urban areas reduces atmospheric CO 2 through sequestration, shading and evapotranspiration. Shading and evapotranspiration reduce atmospheric CO 2 indirectly by reducing the necessity for air conditioning [7][8][9][10][11], which decreases CO 2 emissions from electric power generation. CO 2 sequestration is the direct removal of CO 2 from the atmosphere through photosynthesis and the fixation of carbon in plant litter and root exudates. The capacity of urban greenspaces (urban Figure 1. A schematic drawing of the investigation scope in this study to estimate the CO2 payback time of a modular green roof.

Materials and Methods
In order to estimate CO2 emissions and energy savings from a modular green roof, we set a hypothetical average for green roofs in Japan to a greening area of 200 m 2 and a substrate depth of 5 cm. We conducted partial LCA for the hypothetical green roofs. The functional unit studied was 1 m 2 of the modular green roof with a service lifetime of 45 years on a flat concrete roof. We therefore converted our results for CO2 emissions and reductions into values per m 2 .

Definition of Goal and Scope
In this section, we calculated the CO2 emitted during the production and maintenance of a typical modular green roof. We defined the system boundaries as the production processes for the system components (substrate, substrate containers, water reservoir trays, water proofing membrane, edge dividers, irrigation tubes, irrigation pipes and automatic watering device) and the maintenance practices (irrigation and fertilization) of a modular green roof system. The life cycle system included the extraction and refinement of raw materials and the consumption of natural resources. The production process for vegetation mats in a farm and transportation of the components were not taken into account because of a lack of relevant information.

Inventory Analysis
We collected data from companies and experts about the components and maintenance practices of a typical modular green roof system. We calculated the CO2 emission factors for each component and maintenance practice using the MiLCA LCA software developed by the Japan Environmental Management Association for Industry. We used inter-industrial relation analysis to calculate the CO2 emission factors of the irrigation tube and automatic watering device. This analytical method is a form of economic analysis based on the interdependencies between economic and environmental sectors. It enables us to estimate the environmental impacts of products throughout their costs. For all of the other components and maintenance practices, we used a bottom-up approach for building the inventory. We calculated the amount of CO2 emitted by each component or maintenance practice by multiplying its CO2 emission factor by the quantity of that component used in the hypothetical green roof.
We set the dimensions of a typical green roof at 12.5 m by 16 m (200 m 2 ) and we calculated the use of each green roof system component in this size. A schematic drawing of a typical modular green roof system (with the irrigation pipes and automatic watering device excluded) is shown in Figure 2. The main raw materials used to produce each component and the quantity of each component used in the hypothetical green roof, are shown in Table 1.

Materials and Methods
In order to estimate CO 2 emissions and energy savings from a modular green roof, we set a hypothetical average for green roofs in Japan to a greening area of 200 m 2 and a substrate depth of 5 cm. We conducted partial LCA for the hypothetical green roofs. The functional unit studied was 1 m 2 of the modular green roof with a service lifetime of 45 years on a flat concrete roof. We therefore converted our results for CO 2 emissions and reductions into values per m 2 .

Definition of Goal and Scope
In this section, we calculated the CO 2 emitted during the production and maintenance of a typical modular green roof. We defined the system boundaries as the production processes for the system components (substrate, substrate containers, water reservoir trays, water proofing membrane, edge dividers, irrigation tubes, irrigation pipes and automatic watering device) and the maintenance practices (irrigation and fertilization) of a modular green roof system. The life cycle system included the extraction and refinement of raw materials and the consumption of natural resources. The production process for vegetation mats in a farm and transportation of the components were not taken into account because of a lack of relevant information.

Inventory Analysis
We collected data from companies and experts about the components and maintenance practices of a typical modular green roof system. We calculated the CO 2 emission factors for each component and maintenance practice using the MiLCA LCA software developed by the Japan Environmental Management Association for Industry. We used inter-industrial relation analysis to calculate the CO 2 emission factors of the irrigation tube and automatic watering device. This analytical method is a form of economic analysis based on the interdependencies between economic and environmental sectors. It enables us to estimate the environmental impacts of products throughout their costs. For all of the other components and maintenance practices, we used a bottom-up approach for building the inventory. We calculated the amount of CO 2 emitted by each component or maintenance practice by multiplying its CO 2 emission factor by the quantity of that component used in the hypothetical green roof.
We set the dimensions of a typical green roof at 12.5 m by 16 m (200 m 2 ) and we calculated the use of each green roof system component in this size. A schematic drawing of a typical modular green roof system (with the irrigation pipes and automatic watering device excluded) is shown in Figure 2. The main raw materials used to produce each component and the quantity of each component used in the hypothetical green roof, are shown in Table 1.

Figure 2.
A schematic drawing of a typical modular green roof system that was calculated the CO2 emissions from the production process in this experiment (with the irrigation pipes and automatic watering device excluded). The substrate contains more than 50% perlite, along with compost and zeolite. We calculated the CO2 emission factors of each substrate component using MiLCA and weighed their individual content ratios (kg･kg −1 ). We used the results to calculate a CO2 emission factor for the substrate.
The substrate containers were made of polypropylene, with numerous small holes for drainage. The substrate containers were 50 cm long by 50 cm wide and 6.5 cm deep. Vegetation mats were planted in the containers after filling with substrate to a depth of 5 cm. Each container was connected to adjoining containers to prevent wind uplift.
Drainage from the containers was collected in the water reservoir trays, which were 50 cm long by 50 cm wide by 1.5 cm deep and made of polyvinyl chloride. Each reservoir tray was also connected to adjoining trays.
The irrigation pipes connected the automatic watering device to the irrigation tubes, which were aligned under every second substrate container (13 lines × 16 m). The irrigation pipe needed to be aligned with the edge of the green roof, so we set the length of the irrigation pipe to 15 m. Irrigation pipe consists mainly of polyvinyl chloride.
The water proofing membrane was also made of polyvinyl chloride and was 0.3 mm thick. This membrane was the final layer of the modular green roof system and served as water proofing and a root barrier. It was intended to protect the building from penetration by water and roots.
The edge dividers were made of aluminum and used for sealing the edge of the green roof (57 m). The edge dividers play a crucial role in locking the green roof components in place and preventing wind uplift.
According to the inter-industry relations analysis implemented in MiLCA, the prices of the irrigation tubes and automatic watering device for the hypothetical green roof were $142 and $231, respectively, at an exchange rate of 110 yen to the U.S. dollar.
The use of water and fertilizer during maintenance of the hypothetical green roof are shown in Figure 2. A schematic drawing of a typical modular green roof system that was calculated the CO 2 emissions from the production process in this experiment (with the irrigation pipes and automatic watering device excluded). The substrate contains more than 50% perlite, along with compost and zeolite. We calculated the CO 2 emission factors of each substrate component using MiLCA and weighed their individual content ratios (kg·kg −1 ). We used the results to calculate a CO 2 emission factor for the substrate.
The substrate containers were made of polypropylene, with numerous small holes for drainage. The substrate containers were 50 cm long by 50 cm wide and 6.5 cm deep. Vegetation mats were planted in the containers after filling with substrate to a depth of 5 cm. Each container was connected to adjoining containers to prevent wind uplift.
Drainage from the containers was collected in the water reservoir trays, which were 50 cm long by 50 cm wide by 1.5 cm deep and made of polyvinyl chloride. Each reservoir tray was also connected to adjoining trays.
The irrigation pipes connected the automatic watering device to the irrigation tubes, which were aligned under every second substrate container (13 lines × 16 m). The irrigation pipe needed to be aligned with the edge of the green roof, so we set the length of the irrigation pipe to 15 m. Irrigation pipe consists mainly of polyvinyl chloride.
The water proofing membrane was also made of polyvinyl chloride and was 0.3 mm thick. This membrane was the final layer of the modular green roof system and served as water proofing and a root barrier. It was intended to protect the building from penetration by water and roots.
The edge dividers were made of aluminum and used for sealing the edge of the green roof (57 m). The edge dividers play a crucial role in locking the green roof components in place and preventing wind uplift.
According to the inter-industry relations analysis implemented in MiLCA, the prices of the irrigation tubes and automatic watering device for the hypothetical green roof were $142 and $231, respectively, at an exchange rate of 110 yen to the U.S. dollar.
The use of water and fertilizer during maintenance of the hypothetical green roof are shown in Table 1. According to an interview with a relevant company, a green roof is irrigated 101 times annually, with 8 L·m −2 of water used for each irrigation. Fertilizer is applied twice annually, with 20 g·m −2 used each time.

CO 2 Sequestration by Several Green Roof Plants
In order to quantify CO 2 sequestration in green roofs, we investigated the annual CO 2 sequestration by three grass species-Cynodon dactylon Pers., Festuca arundinacea Schreb. and Zoysia matrella L. "Himekourai-shiba"-and a flowering plant, Sedum aizoon L. Grasses and Sedum species are the most common green roof vegetation in Japan.
This experiment was conducted at the Center for Environment, Health and Field Sciences at Chiba University over one year. All plants in this experiment were propagated as cuttings in plug flats (128 cells·tray −1 ) filled with seedling propagation soil (Metro Mix; Sun Gro Horticulture, Agawam, USA). After approximately one month, the plugs were planted in 0.2 L polyethylene pots (44 cm 2 ) filled to a depth of 5 cm with commercial artificial soil for green roofs (the same as the substrate mentioned above) and grown in a greenhouse for two months. They were then placed on the rooftop and acclimated for three weeks. For a more accurate simulation in this experiment, we should have used the same modular green roof system. However, we used polyethylene pots and irrigated by hand sowing because we had to test a number of experimental plants using a limited roof area.
At the start of the experiment, on October 20, 2014, we sampled all species from a total of 15 pots over a period of about 10 days. Green roofs composed of grasses are generally fitted with irrigation systems to prevent drought stress. In contrast, irrigation systems are less common in Sedum green roofs because Sedum uses the CAM photosynthetic pathway and is thus better adapted to drought conditions. The S. aizoon pots were therefore assigned randomly to irrigation and non-irrigation treatments after the first sampling. Plants in the irrigation treatment group were watered once a week from January to March, once every two days from April to June, every day from July to September and once every two days from October to December. The non-irrigation treatment group was never irrigated. The three grass species received only the irrigation treatment, in keeping with general cultivation practice. All treatments received about 20 g·m −2 (0.1 g·pot −1 ) of controlled-release fertilizer (8N-8P-8K) on June 13 and August 13, 2015. As the end of the experiment, on 20 October 2015, we harvested 15 pots, including all species and treatments.
All plants were dried at 70 • C for 72 h and then divided into plant and substrate matter. Plant and substrate carbon concentrations were measured using an organic elemental analyzer (2400 Series II CHNS/O System; PerkinElmer, Waltham, MA, USA). Carbon content was quantified by multiplying carbon concentration by the dry weight. Annual CO 2 sequestration was calculated by subtracting the total carbon content in October 2014 from the total carbon content in October 2015.
In order to calculate the leaf area index (LAI) of the four species in summer, 15 pots, including all species and treatments, were sampled on August 20, 2015 and divided into leaves and non-leaf parts. Leaves were scanned (LP-A500; EPSON, Nagano, Japan) and image analysis software was used to measure the leaf area (ImageJ [35]). The LAI was calculated as leaf area per 44 cm 2 (the area of each polyethylene pot).
Data were analyzed using IBM SPSS Statistics version 22.0 (IBM Japan, Tokyo, Japan). Differences in mean values were assessed with a Student's t-test.

Estimation of the Energy Savings Amount
We used the Green Roof Energy Calculator to estimate the energy saved by a building covered with the hypothetical green roof (200 m 2 greening area of rooftop, 5 cm substrate depth). This web tool was developed from the Energyplus-based green roof model [24] and requests the following information: state and city, surface area of the roof, building type (old or new, office or apartment), substrate depth (limited to 5 cm < depth < 30 cm), leaf area index (limited to 0.5 < LAI < 5), irrigation flag (yes or no), percentage of roof covered by the green roof system and roofing type for the non-green roof area [black (albedo: 0.15) or white (albedo: 0.65)]. Because this tool requires LAI values, we ran separate simulations for the C. dactylon, F. arundinacea, Z. matrella and S. aizoon green roofs, as was done in the CO 2 sequestration experiment. The growing media characteristics for all green roof simulations were set as follows: thermal conductivity 0.35 W·mK −1 , density 1100 kg·m −3 , specific heat 1200 J·kgK −1 , saturation volumetric moisture 0.3, residual volumetric moisture 0.01, initial volumetric moisture 0.1. See the article for further details of GREC [24].
The location was entered as Houston, Texas, because the meteorological conditions in that city resemble those in Tokyo, Japan. The percentage of roof covered by the green roof system was set at 50%, which set the surface area of the roof at 400 m 2 . In Japan, about 90% of green roofs are located on new building, so the building type was set as "new office". In order to embed the LAI values in summer into this model, the results in August 2015 was added. The irrigation flag was set as "yes" for all simulations, expect for that of the non-irrigated S. aizoon roof. The roofing type for the non-green roof area was set as "black".
The reduction in electricity and gas consumption was estimated in kWh and converted to CO 2 emissions reduction (kg-CO 2 ) using the CO 2 emission factor (0.505 kg-CO 2 ·kWh −1 ) obtained from the Ministry of the Environment in Japan.

CO 2 Payback Time of Modular Green Roofs
In non-irrigated green roof systems (the non-irrigation treatment), CO 2 emitted during the production of the irrigation system (irrigation tubes, irrigation pipes and automatic watering device) and associated with the annual water supply was excluded from the calculation of the total CO 2 emissions from the modular green roof. The amount of annual CO 2 sequestration by the green roof plants may decrease with their age because carbon in plants and soils eventually reaches a carbon equilibrium, with sequestration offset by decomposition [36,37]. Therefore, we calculated the CO 2 payback time based on two different scenarios. In Scenario 1, we hypothesized that the CO 2 sequestration by green roof plants occurs only during the first year after construction. CO 2 payback time for this scenario was defined as CO 2 payback time = (CO 2 e-p -CO 2 r-s ) / (CO 2 r-e -CO 2 e-m ) In Scenario 2, we hypothesized that the same amount of CO 2 sequestration by the green roof plants occurs every year. CO 2 payback time for this scenario was defined as CO 2 payback time = CO 2 e-p / (CO 2 r-s + CO 2 r-e -CO 2 e-m ) where CO 2 e-p is the amount of CO 2 emitted during the production of the modular green roof system, CO 2 r-s is the annual CO 2 reduction owing to CO 2 sequestration, CO 2 r-e is the annual CO 2 reduction owing to energy savings and CO 2 e-m is the annual CO 2 emission from maintenance of the hypothetical green roof.

CO 2 Emissions from a Modular Green Roof
The CO 2 emission factors and total CO 2 emissions from each component are shown in Table 2. The aluminum edge divider had the highest CO 2 emission factor out of all of the components analyzed using a bottom-up approach. The water reservoir tray, water proofing membrane and irrigation pipe had similar CO 2 emission factors because they are all made from the same raw material (Tables 1  and 2).
Aluminum is a lightweight construction material and is therefore suitable for use in green roofs, whose weights are subject to architectural constraints. However, Bribián et al. [38] found that primary energy demand and global warming potential are higher for aluminum production than for the production of several other common building products. Using materials other than aluminum for edge dividers may therefore decrease the environmental load, including CO 2 emissions, associated with the production of green roof systems.
The amount of CO 2 emitted during the production of substrate was higher than that from any of the other components (Table 2). This was due mainly to the relatively large quantify of substrate used in green roof systems ( Table 1). The automatic watering device exhibited the second lowest CO 2 emissions out of all of the components. However, this result is the amount of CO 2 emitted during the production of one machine and, unlike the results for the other components, is independent of the area of the green roof. It is therefore clear that the environmental load per m 2 of the automatic watering device increases as the green roof area decreases.
The CO 2 emission factors and total CO 2 emissions for each maintenance practice are shown in Table 3. The total annual CO 2 emissions from maintenance practices were 0.33 kg-CO 2 ·m −2 ·yr −1 . The CO 2 emissions from the maintenance of a non-irrigated green roof were generated by fertilizer alone, at 0.04 kg-CO 2 ·m −2 ·yr −1 . Total CO 2 emissions from the production of a modular green roof were 25.2 kg-CO 2 ·m −2 ( Table 2). This result is similar to the results of previous research quantifying the amount of CO 2 emitted during the manufacturing phase of layered green roofs [27]. For a green roof without an irrigation system (i.e., a non-irrigated green roof), the total CO 2 emissions from production process were 24.6 kg-CO 2 ·m −2 .

CO 2 Sequestration by Several Green Roof Plant
The plant and substrate dry weight, carbon concentration and carbon content of the different plant species, as recorded on 20 October 2014 and 20 October 2015, are shown in Table 4. For all species and treatments, plant dry weight was significantly higher at the end of the experiment than at the start. There were no significant differences in plant carbon content between October 2014 and October 2015, except for that of S. aizoon with irrigation treatment. F. arundinacea was the only species for which substrate dry weight at the end of the experiment was significantly different from that at the beginning. The substrate carbon concentration of all species and treatments increased during the experimental period and C. dactylon was the only species for which it did not increase significantly. In addition, the plant and substrate carbon contents for all species and treatments increased significantly during the experimental period. These results suggest that some carbon from the plant litter and root exudate was fixed in the substrates. Table 4. Mean values (n = 15) for plant and substrate dry weight, carbon concentration, carbon content and total annual carbon sequestration by four green roof plants (C. dactylon, F. arundinacea, Z. matrella and S. aizoon). The total annual carbon sequestration by F. arundinacea was higher than that of any other species and was similar to those of the other grasses (Table 4). S. aizoon exhibited higher total annual carbon sequestration with irrigation treatment than with non-irrigation treatment. The results for S. aizoon were similar to results from previous research investigating Sedum green roofs [25]. With the assumption that carbon sequestration resulted only from CO 2 uptake, total annual carbon sequestration was converted to annual CO 2 sequestration. C. dactylon sequestered 2.530 kg-CO 2 ·m −2 ·yr −1 , F. arundinacea sequestered 2.754 kg-CO 2 ·m −2 ·yr −1 , Z. matrella sequestered 2.459 kg-CO 2 ·m −2 ·yr −1 , S. aizoon with irrigation treatment sequestered 1.684 kg-CO 2 ·m −2 ·yr −1 and S. aizoon with non-irrigation treatment sequestered 1.232 kg-CO 2 ·m −2 ·yr −1 . This experiment thus supports the CO 2 sequestration capacity of several green roof plants. Kuronuma and Watanabe [39] suggested that competence for carbon sequestration of Sedum under wet and increased nutrient conditions are equivalent to those of other green roof plants (Zoysia matrella and Ophiopogon japonicus). However, it was disaccorded with the present results. This is probably because the present experiment conditions keeping with general cultivation practice was nutrient-poor for Sedum species. From this reason, it is suggested that carbon sequestration of Sedum aizoon was lower than those of the grasses. For the carbon sequestration of green roofs, therefore, more suitable design and maintenance practice of vegetation should be studied further.

Species and Treatments
The LAI of each specie and treatment, as recorded in August 2015, is shown in Table 5. S. aizoon with irrigation treatment exhibited the highest LAI out of all of the species and treatments and C. dactylon exhibited the lowest.

Estimation of the Energy Savings Amount
The annual energy savings from the hypothetical green roofs are shown in Table 5. Of the irrigated green roofs (with irrigation treatment), the annual energy saving from the S. aizoon green roof was the highest because the LAI of S. aizoon was higher than that of the three grass species. In contrast, although S. aizoon with non-irrigation treatment exhibited the median LAI of all species and treatments, its annual energy saving was the lowest because its irrigation flag was set as "no". However, annual energy savings did not differ much between species and treatments and annual CO 2 reductions due to energy savings were between 1.703 and 1.889 kg-CO 2 ·m −2 ·yr −1 (Table 5), being similar to those in previous research [22]. Table 5. Mean values (n = 15) for leaf area index (LAI), energy saved annually and annual CO 2 reduction due to saved energy of modular green roofs with different species (C. dactylon, F. arundinacea, Z. matrella and S. aizoon) and treatments (irrigation and non-irrigation).

CO 2 Payback Time of Modular Green Roofs
The CO 2 payback time calculated from the Scenario 1, which hypothesized the CO 2 sequestration by the green roof plants occurs only first year after construction, was in the range of 14.0 to 15.9 years (Table 6). On the other hand, Scenario 2 was hypothesized that the same amount of annual CO 2 sequestration by the green roof plants occur every year. Thus, the CO 2 payback time calculated from the Scenario 2 was in the rage of 5.8 to 8.5 years and shorter than that calculated from the Scenario 1. For the S. aizoon green roofs, the amount of CO 2 emitted during the production of a non-irrigated green roof was lower than that from an irrigated green roof. Therefore, the CO 2 payback time calculated from Scenario 1 was longer for the irrigated green roof than for the non-irrigated green roof. However, the CO 2 payback time calculated from Scenario 2 was longer for a non-irrigated green roof than for an irrigated one because the increase through in CO 2 reduction from irrigation is greater than the amount of CO 2 emitted by the addition of an irrigation system and annual water supply (Tables 4 and 6). However, these results indicate that there were not significant differences among the CO 2 payback time of modular green roofs with different species and treatments in this experiment. Our results for the hypothetical green roofs indicate that CO 2 reduction from the combination of CO 2 sequestration and energy savings by a green roof can offset the CO 2 emitted during its production and maintenance after 5.8 to 15.9 years ( Table 6). The lifespans of green roof components are generally thought to be between 40 and 50 years [28][29][30]40], so it is clear that the lifetime CO 2 reduction of modular green roofs offsets the CO 2 emitted during their production and maintenance. Accordingly, our results suggest that modular green roofs are an effective way to reduce atmospheric CO 2 and mitigate global warming. In addition, Tripanagnostopoulos et al. [34] clarified CO 2 payback time of solar photovoltaic systems which is between 1.3 and 4.1 years. Thus, CO 2 payback time will be one of the index which could compare the carbon balance of green roofs with that of rooftop solar photovoltaic systems [41].
In this study, we focused on the production processes and maintenance practices for green roofs. However, CO 2 sequestration in plants and substrates seems to be greatly influenced by their disposal phase. In addition, the period for which green roofs serve as carbon sinks may be shorter than those of other urban greenspaces (e.g., urban forests, parks and trees) because of the more frequent repair and demolition of buildings. Therefore, the disposal phase also needs to be discussed to ensure the sustainable design of green roofs although the CO 2 payback time calculated from energy savings alone was shorter than the system component's lifespan (17.6 years for C. dactylon, 16.2 years for F. arundinacea, 17.1 years for Z. matrella, 16.2 years for S. aizoon with irrigation treatment and 14.7 years for S. aizoon with non-irrigation treatment).
This study did not cover all green roof types (e.g., different green roof systems, system components, vegetation plants, substrate depth and locations) that could influence the CO 2 payback time results. Thus, the CO 2 emissions from a green roof and CO 2 reductions from secondly environmental benefits should be studied further. Our results will serve as a baseline for future research on the CO 2 payoff of green roofs. Table 6. CO 2 payback time of modular green roofs with different species (C. dactylon, F. arundinacea, Z. matrella and S. aizoon) and treatments (irrigation and non-irrigation).

Conclusions
This study quantified the CO 2 emitted during the production and maintenance of a hypothetical modular green roof and estimated the CO 2 reduction from energy savings and CO 2 sequestration. The results of the study show that CO 2 emissions are offset through CO 2 sequestration and energy savings after 5.8 to 15.9 years, which indicates that extensive modular green roofs contribute to atmospheric CO 2 reduction and global warming mitigation within their lifespan. In addition, this study provided a method to assess the CO 2 payoff of green roofs. It is hoped that the findings presented in this paper will contribute to the development and design of green roofs more suitable for CO 2 reduction and global warming mitigation.