Following the integrated methodology framework (Fig. 1), we initially used GIS sampling method to estimate the potential area for green roof development in the case cities. Second, we applied the life cycle assessment method to calculate the environmental impacts, i.e., energy and water consumption and carbon emissions of green roofs in the entire life cycle. Third, the nested process-based models were used to simulate the direct FWE-related benefits of green roofs within city boundaries, including tomato yield, direct energy saving, direct water saving, and direct carbon capture. Fourth, we used the multi-regional input-output model to account for the avoided trans-boundary environmental footprints, i.e., energy, water, and carbon footprints, embodied in global supply chains. We then adapted the Nexus Indexes to indicate the total potential of avoided environmental footprints to visualize the FWE supply chain risks. In addition, a series of scenario analyses were conducted to evaluate the performance of city-wide green roof implementations.
Step 1: GIS sampling
The GIS sampling method was used to estimate the available building area in case cities. For SJC and Johannesburg, we respectively took 10% and 5% of their urban land area (we considered urban administrative area in this study) as the samples and found that the building footprints in SJC covered 7.5% of the sampled area and the footprints in Johannesburg occupied 14.7%. Thus, we assume that 5–10% of the total area in SJC and 15–20% in Johannesburg can be regarded as buildings.
Step 2: Life cycle assessment
To determine the environmental impacts of green roofs in the whole life cycle within a specific system boundary, i.e., impacts at the city scale, a cradle-to-grave life cycle assessment was conducted. In this study, three environmental impact categories considered were energy consumption, water consumption, and carbon emissions, the coefficients were taken from the corresponding primary energy demand (PED) including renewable and non-renewable resources, blue water consumption (BW) including hydropower, and greenhouse gases (GHG, IPCC AR5 GWP100, except for biogenic carbon) in Gabi database. Supplementary Table 3 outlines the detailed scope and boundary of energy, water consumption, and carbon emissions over the life cycle of green roofs. Given that vegetable growing often requires deeper mediums56 and tomato is one of the most widely consumed and grown vegetable crops around the world, a selection of intensive green roofs and tomato growing were regarded as representative of rooftop open-air farming. Due to the limitations of local site data, we accounted for the life cycle impacts of green roofs based on general data and assumed that SJC and Johannesburg implemented a similar green roof system with green roof structures, open-air farming processes, and rainwater harvesting system. Specifically, we included materials for green roof layers and materials for open-air farming in the infrastructure material production phase. The rainwater harvesting system was included in the operation and maintenance phase, this inclusion was based on our assumption that the harvested rainwater could be used for tomato irrigation in the operation and maintenance phase. The entire life cycle included the stages of infrastructure material production, installation and construction, operation and maintenance, demolition, and disposal. The basic formula is as below:
$${F}_{i}={f}_{i}\times {c}_{i}$$
1
where F refers to the total environmental impact embodied in each life cycle process; i refers to PED, BW, GHG; c is the coefficient for the environmental impacts embodied in each life cycle process, calculated based on the Gabi database; f is the activity data, i.e., the input quantity of each life cycle process in one year.
Step 3: FWE-related benefits evaluation
Local food production. The process-based biogeochemical model Denitrification-Decomposition (DNDC) model (http://www.dndc.sr.unh.edu/) was used to estimate the tomato yield on green roofs in the case cities. According to the default plant parameters of the maximum biomass yield, biomass fraction, carbon to nitrogen ratio for tomato crops, and the management strategies of precision fertilization and auto irrigation, the plant growth and carbon and nitrogen cycles can be simulated by the DNDC model, in response to environmental variables and management strategies in each location.
Energy saving. The Green Roof Energy Calculator (v2.0) (https://sustainability-innovation.asu.edu/urban-climate/green-roof-calculator/) within Eco-roof module was applied to evaluate the direct energy saving potential of green roofs. Given the lack of specific data on the case cities, we selected two proxy cities in the same climate zones as SJC and Johannesburg (Supplementary Table 4) to obtain their detailed climate information to achieve the calculations. Due to the difference in building characteristics as well as heating and cooling needs, the simulation results vary significantly between different building types. To extrapolate the urban-scale direct energy saving, the results must be weighed by taking the ratio of different building types. Given that the ratio data of specific building types were not available, we assumed that the ratio of new office buildings, new residential buildings, old office buildings, and old residential buildings in both cities was 1:1:1:1 to realize an up-scaling.
Water saving. The study considered a rainwater harvesting system to collect the rainwater for irrigation. The rainwater harvesting potential was essentially based on the average rainfall of each location and the characteristics of specific green roofs. The following Eq. 57 was used to estimate the amount of rainwater harvested by the harvesting system.
$$RH=A\times P\times C$$
2
where RH is the amount of rainwater harvested, in L; A is the unit area in m2; P is the actual local precipitation in mm; C refers to the harvesting efficiency of green roofs, often indicated as the runoff coefficient. We considered a conservative value of 15% efficiency for the catchment area to compensate for the effects of leaks, wind, and rainfall rates58.
Particularly, if the harvested rainwater was greater than the irrigation demand for tomatoes, the harvested rainwater was fully used to irrigate tomatoes and offset their irrigation water consumption; in that case, we defined the irrigation demand as direct water saving. Conversely, if the harvested rainwater was less than the irrigation demand for tomatoes, the irrigation demand was partially satisfied by the harvested rainwater, we defined the harvested rainwater as direct water saving.
Carbon capture. During the growing period, the average daily carbon capture of tomatoes59 is 18.56 g of CO2·m− 2. According to FAO60, tomatoes have a growing period of 90 to 150 days. Thus, we assumed that the growing period for tomato crops on green roofs in our case cities was 150 days, and there could be two growth cycles each year. Thereby, the annual carbon capture per unit area (i.e., m²) of the studied green roofs could be estimated.
Step 4: Avoided trans-boundary environmental footprints and nexus indexes
Avoided environmental footprint. The multi-regional input-output model and GTAP database were combined to estimate the avoided environmental footprints for each sector61 by green roofs per meter square. The Input-Output model is a widely used economic method that helps analyze the interdependence of sectors in an economy and shows how the total output of a given sector becomes an input to another sector, on a national or regional level62. This study considered that the entire food, water, and energy resources of the case cities were provided from other domestic regions inside the national territory. Based on this, the Brazil- and South Africa-based EIO-LCA (Economic Input-Output based Life Cycle Assessment) models were structured and used to account for the detailed flows of trans-boundary environmental impacts, including energy, water, and carbon footprints, embedded in global food, water, and energy supply chains63. The corresponding sectors in the EIO are “Vegetable, fruit, nuts (food)”, “Electricity (energy)”, “Gas manufacture, distribution (energy)”, and “Water (water)”. We additionally used the current prices of products, e.g., water, electricity, gas, tomato, and the consumer price index for 2019 to convert physical products into economic values for each sector.
Food-Water-Energy-Carbon Nexus Indexes. To determine the structure of the avoided trans-boundary environmental footprints induced by FWE-related benefits, i.e., local food production, direct energy saving, and direct water saving, of green roofs per meter square, we adapted the Nexus Index12 to indicate the total potential of avoided environmental footprints based on FWE nexus, including the avoided energy footprint (Nexus Energy Index, NEI), the avoided water footprint (Nexus Water Index, NWI), and the avoided carbon footprint (Nexus Carbon Index, NCI). The specific calculation formulas are as follows:
$$NEI={AE}_{f}+{AE}_{e}+{AE}_{w}$$
3
$$NWI={AW}_{f}+{AW}_{e}+{AW}_{w}$$
4
$$NCI={AC}_{f}+{AC}_{e}+{AC}_{w}$$
5
where AE is the avoided trans-boundary energy footprint, in MJ; AW is the avoided trans-boundary water footprint, in L; AC is the avoided trans-boundary carbon footprint, in kg CO2e; f, e, and w refer to the local food production, direct energy saving, and direct water saving, respectively; AEf refers to the avoided trans-boundary energy footprint by local food production, the same is true for the others.
Scenario analysis
For the performance evaluation of city-wide green roof implementations, we assumed that the rooftop area of a building was the same as the building footprint. As for the available rooftop, the building categories, rooftop structural resistance and slope, rooftops shaded by neighboring buildings were considered for the available rooftop identification. We assumed that the ratio of an available rooftop to the entire roof area in the cities was 20% (A), 30% (B), 40% (C), and 50% (D). Three sub-scenarios were set within the A-D scenarios (Table 1), which respectively refer to 20% (S1), 50% (S2), and 100% (S3) of green roof conversion. Scenario B (30%) was taken as an example to explain the performance of city-wide green roof implementations, including scenarios B-S1, B-S2, and B-S3. More details for the scenarios of 20% (A), 40% (C), and 50% (D) of available rooftop ratios can be found in Supplementary Fig. 3–5 and Supplementary Table 4. We first accounted for the implications of city-wide green roofs on food, water, energy, and carbon, then we compared these impacts with urban food, water, and energy demand and carbon emissions. Specifically, regarding food, the global average per capita consumption of tomatoes was 20.5 kg, constituting 15% of the total average vegetable intake64; thus, the urban vegetable demand per capita and the total vegetable demand of SJC and Johannesburg could be estimated. In addition, we regarded tomatoes as a proxy for the vegetable product from rooftop farming, this is based on the proxy reference from Gondhalekar and Ramsauer65; therefore, the modelling results for tomato yield could be used to evaluate the self-sufficiency of vegetables in the cities.
Table 1
Assumed scenario settings.
Available rooftop ratio | Scenario | Description |
20% (A) | A-S1 | 20% green roof conversion |
A-S2 | 50% green roof conversion |
A-S3 | 100% green roof conversion |
30% (B) | B-S1 | 20% green roof conversion |
B-S2 | 50% green roof conversion |
B-S3 | 100% green roof conversion |
40% (C) | C-S1 | 20% green roof conversion |
C-S2 | 50% green roof conversion |
C-S3 | 100% green roof conversion |
50% (D) | D-S1 | 20% green roof conversion |
D-S2 | 50% green roof conversion |
D-S3 | 100% green roof conversion |
Study areas
Globally, the population in urban areas is growing rapidly, and the mean annual growth rate of population in the global south cities is four times the rate in the global north66. In light of the unprecedented rate of urbanization and its negative implications for urban FWE resource management especially in global south countries, we need a strategy for sustainable urban FWE resources within urban areas. To our knowledge, although cities in developing countries, particularly large countries, share socio-economic characteristics and vulnerabilities, their predominant solutions vary widely across the regions, which is related to the demographic and development characteristics. On the basis of support of the IFWEN project (https://ifwen.org/), we selected two global south cities as the case cities, São José dos Campos (SJC) in Brazil and Johannesburg in South Africa, these cities provided an excellent basis for this study due to regional factors. SJC covers 1,099 km2 and represents the core municipality in the metropolitan region of the Paraíba River Valley in the southeast of São Paulo State, Brazil (Supplementary Fig. 1a). SJC is an important aircraft manufacturing city and scientific research center in Latin America, with a 2B climate type (Hot-Dry) in the ASHRAE (American Society of Heating, Refrigerating and Air-Conditioning Engineers, https://www.ashrae.org/) climate zone. The city is also one of the most economically dynamic in the state between the two most active production and consumption areas in the country, i.e., the megalopolis of São Paulo and Rio de Janeiro, with about 713,943 urban residents in 2018. Johannesburg is the capital of Gauteng Province in the northeast of South Africa (Supplementary Fig. 1b) covering 1,645 km2 with a 3A climate type (Warm-Humid) in the ASHRAE climate zone. In Johannesburg, the main industrial sectors are agriculture, tanning, and textiles. Johannesburg is the most developed and richest city in South Africa, and the urban population of Johannesburg was 5,674,824 in 2018. In particular, Johannesburg has clearly set its goals for a climate resilient city by 2050—by 2050, the city has 30% green cover (including green roofs)22. For São Paulo, resilience measures and nature-based solutions for climate change mitigation and adaptation are also favored21. Based on the above, the exploration of the impacts of green roofs on FWE nexus will provide an innovative solution for SJC and Johannesburg to align their priorities for food, water, and energy sustainability.