The food-water-energy nexus and green roofs in Sao Jose dos Campos, Brazil, and Johannesburg, South Africa

The study addresses how urban green roofs interact with the food-water-energy (FWE) nexus at city scale, especially in data-sparse contexts. Rapid urbanization intensifies challenges in food security, flooding, water scarcity, and energy demand. Green roofs link directly to FWE through rooftop farming (food), rainwater harvesting and water quality (water), and building energy savings via temperature modulation (energy). While green roofs can localize resource provision and reduce upstream footprints, they also incur life cycle impacts from materials, construction, maintenance, and end-of-life. The research question is how to systematically quantify trade-offs and transboundary effects of green roofs on the FWE nexus, capturing both direct operational benefits and indirect avoided environmental footprints across city boundaries. The purpose is to guide city-scale planning and governance to leverage green roofs for sustainability, using an urban systems perspective and applicable in data-sparse Global South cities.

Prior studies often assess green roofs in silos, focusing on life cycle impacts or single benefits. Examples include rooftop food production potential (Boston), PV-green roof integrations (Toronto), LCA of rooftop greenhouses vs. linear food systems (Barcelona), and city-level ecological/economic impacts (Corvallis). Multi-trade-off evaluations exist but are often neighborhood-scale (e.g., Barcelona roof mosaics assessing self-sufficiency and avoided CO2; Brazilian tech park Nexus Emission Index). A gap remains for systemic, city-scale analyses integrating life cycle burdens, multiple FWE-related operational benefits, and upstream transboundary effects, particularly in data-sparse cities. This study addresses that gap with an integrated GIS and urban metabolism-based framework.

The authors propose a four-step, generalizable framework integrating GIS and urban metabolism approaches to quantify green roofs’ life cycle impacts, direct FWE-related benefits, and avoided transboundary environmental footprints, with city-wide scenario scaling. - Step 1: Urban rooftop area extraction (GIS sampling). For SJC and Johannesburg, random samples of 10% and 5% of administrative areas were used to estimate building footprints. Extrapolated building footprint shares: SJC 5–10%; Johannesburg 15–20%; rooftop area assumed equal to building footprint area. - Step 2: Life cycle environmental impact assessment (LCA). Cradle-to-grave assessment per m²-year for energy consumption (Primary Energy Demand), water consumption (Blue water consumption), and GHG emissions (IPCC AR5 GWP100, excluding biogenic carbon). System includes: green roof structures, open-air farming system, and rainwater harvesting system. Life cycle stages: infrastructure material production (IMP), installation and construction (I&C), operation and maintenance (O&M), demolition (De), disposal (Di). Background data from GaBi database. Material contributors disaggregated into materials for green roof structures (MGR: root barrier, growing medium, water retention, drainage/filter) and materials for open-air farming and fertilizers (MOAF). Waste treatment impacts accounted in disposal. - Step 3: Direct FWEC-related benefits (process-based models). Food: Tomato as representative crop; simulated yields using the DNDC biogeochemical model with local meteorology (temperature, precipitation), soil, and cropping parameters. Energy: Building energy savings from green roofs using the Sailor & Bass green roof energy model, selecting proxy U.S. cities in matching ASHRAE climate zones to SJC and Johannesburg and interpolating by building type, LAI, media depth, and irrigation. Water: Rainwater harvesting potential RH = A × P × C (A=1 m²; P=local precipitation; C=runoff coefficient; 15% efficiency used). Direct water saving (DWS) equals harvested rainwater if RH ≤ irrigation demand (ID), otherwise equals ID; ID determined by crop blue water intensity and simulated tomato yield. Carbon: Direct plant carbon capture estimated from average daily CO2 uptake (18.56 g CO2 m⁻² during growth), with two 150-day growth cycles per year. - Step 4: Avoided transboundary environmental footprints and Nexus Index series (EIO-LCA). Constructed Brazil- and South Africa-specific economic input-output LCA models using GTAP v10 to map avoided upstream energy, water, and carbon footprints attributable to direct benefits: local food production (f), direct energy saving (e), direct water saving (w). Physical products converted to economic values using local prices and CPI. Defined Nexus Energy Index (NEI) = sum of avoided energy footprints (AE), Nexus Water Index (NWI) = sum of avoided water footprints (AW), and Nexus Carbon Index (NCI) = sum of avoided carbon footprints (AC) across f, e, w drivers. - Scenario analysis: Defined available rooftop ratios A–D = 20%, 30%, 40%, 50% of total rooftop area, each with conversion sub-scenarios S1=20%, S2=50%, S3=100% of available roofs converted to green roofs. Scaled per-m² results to city-wide implications for energy, water, carbon, and vegetable self-sufficiency. Scenario B (30% available roofs) highlighted in main text; full scenario results in Supplementary Information.

- Direct operational benefits per m²-year: • Food (tomato yield): SJC 6.83 kg; Johannesburg 5.68 kg. • Direct water saving: SJC 95.62 L; Johannesburg 101.94 L. • Direct energy saving: SJC 45.40 MJ; Johannesburg 43.82 MJ. • Direct carbon capture: both cities 5.57 kg CO2. - Life cycle environmental burdens per m²-year: • Energy consumption: SJC 88.59 MJ; Johannesburg 89.32 MJ (about 2× direct energy savings). • Carbon emissions: SJC 5.55 kg CO2; Johannesburg 5.59 kg CO2 (nearly offset by carbon capture → essentially carbon neutral). • Water consumption: SJC 9.44 L; Johannesburg 186.28 L (SJC net water beneficiary; Johannesburg net water consumer when compared to direct water savings). - Dominant life cycle stages/materials: • 88–89% of energy and 81–82% of carbon occur in IMP stage; MGR accounts for 93% of IMP energy and 95% of IMP carbon. Within MGR, drainage/filter layer dominates energy (54%); growing medium dominates carbon (56%). Disposal stage is the second contributor for life cycle carbon due to waste treatment. - Rainwater harvesting vs tap water (life cycle): rainwater system requires 1.2× energy and 1.5× carbon of tap water; 84–91% of these burdens come from IMP, with tanks contributing 86% within IMP. - Avoided transboundary footprints exceed burdens: • In SJC and Johannesburg, avoided upstream energy, water, and carbon footprints induced by direct benefits are 1.6–384× and 1.4–7.8× their respective life cycle impacts. • Drivers differ: In SJC, local food production drives 68% of avoided energy, 95% of avoided carbon, and 59% of avoided water. In Johannesburg, direct energy savings drive 51% of avoided energy and 55% of avoided carbon; local food production drives 80% of avoided water. • NEI in SJC is 56% of Johannesburg’s; NCI in SJC is 1.2× Johannesburg’s, reflecting carbon-intensive upstream supply chains for SJC’s imports and major role of rooftop farming there. - City-wide scenarios (30% available roofs): • Vegetable self-sufficiency: SJC can meet 100% (and more) under B-S3; Johannesburg up to 72.37%. • Water: In Johannesburg, green roofs’ direct water savings contribute 5.3–7.1% toward the goal of 25% alternative water sources by 2050; harvested rainwater meets ~37% of tomato irrigation demand (high irrigation demand leads to life cycle water consumption 1.8× direct water savings). In SJC, harvested rainwater can fully meet tomato irrigation demand; avoided upstream water footprints could be 1.2–2.4× the city’s annual water demand. • Energy: Within-city, green roofs are net energy consumers in both cities; avoided transboundary energy footprints can exceed 30% of city energy demand. Worst-case, SJC life cycle energy consumption could reach 45% of annual city demand. • Carbon: Within-city carbon neutral; avoided transboundary carbon footprints contribute 0.04–0.07% (SJC→Brazil 2030 target) and 0.14–0.2% (Johannesburg→South Africa 2030 target).

The integrated framework quantifies both direct and indirect (transboundary) implications of green roofs on the FWE nexus, addressing the need for a systemic, city-scale perspective. Findings show that although green roofs are essentially carbon neutral and net energy consumers over their life cycle, their indirect avoided footprints in upstream supply chains can outweigh life cycle burdens, thereby contributing meaningfully to regional sustainability goals. City-specific trade structures shape which benefits matter most: SJC’s rooftop farming strongly reduces upstream carbon and water footprints and can meet local vegetable demand, indicating food-focused planning (e.g., rooftop agriculture) offers large transboundary gains. In contrast, Johannesburg’s strongest transboundary effects come from energy savings; despite net life cycle energy consumption, avoided upstream energy footprints from direct savings can compensate for net consumption, suggesting prioritization of energy-saving design and potentially renewable generation on roofs. Rainwater harvesting yields direct water savings but currently has higher life cycle energy and carbon than tap water; optimizing material choices (especially tanks) and system design could improve performance. Governance implications include aligning green roof strategies with city FWE goals, supply chain risks, and trade agreements, and recognizing the broader spatial scale of impacts when forming policies on urban land teleconnections and resource sustainability.

This study presents a generalizable, integrated methodology combining GIS sampling, LCA, process-based modeling, and EIO-LCA to assess green roofs’ life cycle burdens, operational FWE-related benefits, and avoided transboundary footprints at city scale, tailored for data-sparse contexts. Applied to SJC and Johannesburg, the approach reveals that green roofs are carbon neutral yet net energy consumers, with divergent water outcomes (SJC net beneficiary; Johannesburg net consumer). Indirect avoided footprints can exceed life cycle impacts, with SJC best served by prioritizing rooftop food production and Johannesburg by maximizing energy savings. The Nexus Index series (NEI, NWI, NCI) provides decision-support for prioritizing FWE-oriented green roof strategies. Future research should build local data systems to refine rooftop suitability, life cycle parameters, and process model inputs; expand assessment dimensions (e.g., education, health, stormwater mitigation, water purification, PV integration); and extend the framework to other green and blue infrastructures and diverse Global South cities.

- Rooftop availability estimated via sampling and assumptions; lack of building-level structural and shading data prevents precise suitability mapping. - Use of generic life cycle parameters from the GaBi database due to limited local data; empirical coefficients and proxy data used in process-based models introduce uncertainty. - EIO-LCA relies on sectoral data to map embodied resource flows; assumes city demands supplied domestically, which may not fully capture international trade effects. - Scenario analyses proportionally scale per-m² results; uncertainties in rooftop area assumptions propagate to city-wide estimates. - The rainwater harvesting assessment indicates higher life cycle burdens than tap water for current designs; results sensitive to material choices (e.g., storage tanks) and system configurations.