Urban oasis living lab: a hybrid approach to climate-resilient urban agriculture in Beirut

As cities face increasing resource constraints, environmental pressures, and food-system vulnerabilities, urban agriculture is gaining recognition as a practical approach to enhancing climate resilience and sustainability. However, many urban agriculture initiatives remain fragmented, often focusing on isolated ecological or technical interventions without fully integrating resource management, stakeholder engagement, and long-term resilience mechanisms. To address these gaps, the Urban Oasis was established by the Environment and Sustainable Development Unit (ESDU) at the American University of Beirut (AUB), with support from the Khalifa International Award for Date Palm and Agricultural Innovation. An abandoned animal house was transformed into a multifunctional living lab for research, innovation, education, and community engagement through a structured five-phase process comprising participatory design, site rehabilitation, technology activation, engagement and applied learning, and continuous monitoring and evaluation. The resulting living lab combines sustainable agriculture practices, circular economy principles, low-impact infrastructure, smart sensing systems, artificial intelligence, and automated controls. Over 3 years of operation, the Urban Oasis produced approximately 8,500 kg of food, recovers approximately 480,000 liters of water annually, and avoids an estimated 1.26 tons of carbon dioxide emissions each year. More than 5,000 individuals have engaged with the testbed, and 62.7% of participants reported adopting or willingness to adopt at least one demonstrated practice in their daily lives. The initiative has also supported academic research, student projects, and scientific publications. The Urban Oasis demonstrates how urban agriculture can evolve beyond food production into an integrated resilience framework connecting resource efficiency, community engagement, education, and technological innovation. As a scalable and adaptable model, it offers insights for advancing sustainable and climate-resilient urban development across the Middle East and North Africa region and beyond.

The Middle East and North Africa (MENA) region is among the most climate-vulnerable regions globally due to limited freshwater availability, reliance on food imports, rapid urbanization, and growing pressure on natural resources. Urban areas are particularly exposed to these challenges, as population growth continues to increase demands for food, water, and energy while reducing access to productive land and ecosystem services. Rising temperatures, prolonged droughts, water scarcity, and environmental degradation are further contributing to the deterioration of urban food systems (IOM, 2023).

Among MENA countries, Lebanon exemplifies many of these interconnected challenges. The country faces mounting environmental pressures compounded by recurring political instability, economic crises, infrastructure deterioration, and periods of conflict. Climate projections indicate increasing heat stress, declining precipitation, groundwater depletion, and heightened risks to food security in the coming decades, particularly in urban areas (FAO, 2025). These overlapping stresses have exposed the vulnerability of conventional food supply systems and highlighted the urgent need for localized, resource-efficient, and climate-resilient food production approaches (ILO, 2025).

In this context, urban agriculture (UA) has emerged as a promising strategy for enhancing climate adaptation and resilience. Beyond food production, UA can improve resource efficiency, strengthen food security, enhance urban biodiversity, reduce food miles, and foster community engagement when integrated within broader urban sustainability frameworks (Gao et al., 2025). However, despite its potential, many UA initiatives continue to face challenges related to long-term impact, scalability, and sustainability.

Three key limitations are frequently observed in existing UA interventions. First, stakeholder engagement is frequently concentrated during implementation, with limited mechanisms to ensure continuous participation, local ownership, knowledge exchange, and long-term stewardship (Owen and Drake, 2023). Second, UA initiatives are often designed around a single objective, such as food production, water conservation, or technological innovation, without adequately addressing the interconnections among water, energy, and food systems, as well as ecosystem health. This limits their long-term sustainability and often results in challenges in maintaining and scaling interventions (Fox-Kämper et al., 2023). Third, many projects lack robust monitoring and evaluation mechanisms, making it difficult to quantify their environmental, social, and economic impacts over time (Berberi et al., 2021; Teoh et al., 2024). This diminishes their attractiveness and eventually their proper functioning.

To address these three challenges UA projects have been coupled with several solutions. First, researchers have increasingly explored participatory approaches that place stakeholders at the center of UA innovation processes. Among these, living labs have gained prominence as dynamic, open innovation ecosystems where research and innovation unfold in a real-life environment rather than in isolated laboratories, thereby strengthening engagement and improving the long-term sustainability of interventions (Yan et al., 2022). At the same time, growing attention has been given to the Water–Energy–Food–Ecosystems (WEFE) Nexus, which recognizes the interconnections among water resources, energy systems, food production, and ecosystem services to improve resource efficiency and sustainability (Al-Saidi and Hussein, 2021). Finally, recent advances in digital agriculture provide additional opportunities to support integrated resource management. Artificial intelligence (AI), sensor-based monitoring systems, and data-driven decision-support tools can generate real-time insights into system performance, enabling more efficient irrigation, energy management, crop production, and adaptive resource allocation. These technologies can also provide the evidence needed to evaluate impacts and support continuous improvement (Snehalatha and Patil, 2026).

Despite the growing body of literature on the integration of UA with living labs, WEFE-informed resource management, and digital agriculture, these research streams have largely evolved independently. Considering that each solves a specific aspect of the problem, it is interesting to understand how stakeholder engagement, integrated resource management, and digital monitoring technologies can be combined within a single framework to create UA systems that are resilient, resource-efficient, and capable of delivering sustained long-term impacts.

This study evaluates the Urban Oasis (UO) at the American University of Beirut as an integrated climate-responsive urban agriculture model operating within a resource-constrained urban environment. The UO combines a WEFE-informed resource-management framework, AI-enabled monitoring, circular economy practices, climate-responsive agricultural systems, and participatory Living Lab principles. The study addresses the following research questions:

RQ1: To what extent does the integrated UO model improve social, economic, and environmental performance in urban settings?

RQ2: How effectively does the Living Lab approach promote the adoption or intended adoption of UA practices?

RQ3: What features of the UO model support or constrain its potential replication in other resource-constrained urban contexts?

This study is guided by two conceptual hypotheses, used to bound the scope of interpretation rather than for formal statistical testing: successful implementation of integrated urban agriculture systems depends on supportive governance, regulation, and stakeholder collaboration (H1), while their long-term sustainability and scalability depend on institutional support, financial viability, and sustained stakeholder engagement (H2). These hypotheses frame governance and regulation as enabling conditions rather than objects of direct investigation.

Study site description

The study was conducted at the UO, an urban agriculture living lab located on the campus of the AUB in Beirut, Lebanon. The UO occupies an approximately 250 m 2 site that formerly served as the Faculty of Agricultural and Food Sciences (FAFS) Animal House before becoming abandoned and underutilized. Through a low-impact adaptive reuse approach, the facility was transformed into the UO.

To operationalize this framework, UO was built through a 5-phase process. The first phase, Participatory Design and Planning, engaged stakeholders in the conceptualization and design of the space. This phase ensured that the UO responded to user needs and expectations while fostering ownership, relevance, and long-term commitment to the intervention. The second phase, Site Rehabilitation and Resource Integration, translated WEFE Nexus principles into practice through the establishment of interconnected water, energy, food, and environmental systems. By optimizing resource flows, promoting circularity, reducing dependence on external inputs, and integrating climate-responsive design measures, this phase aimed to strengthen the environmental and economic sustainability of the UO while enhancing its long-term resilience and operational viability. The third phase, Smart System Deployment and Artificial Intelligence Integration, enhanced the management of these resource systems through real-time monitoring, automation, and data visualization tools. This increased the transparency of system performance, improved resource-use efficiency, and generated evidence to support informed decision-making and continuous improvement. The fourth phase, Stakeholder Engagement and Applied Learning, activated the Living Lab through workshops, volunteering activities, demonstrations, experimentation, and knowledge exchange, enabling participants to directly interact with the implemented systems and experience the practical benefits of UA. By transforming users from passive observers into active participants, this phase sought to build awareness, strengthen technical capacity, increase confidence in the demonstrated practices, and ultimately encourage their adoption and replication beyond the UO. Finally, the fifth phase, Monitoring and Evaluation, assessed environmental, economic, social, and operational performance indicators to continuously refine the living lab while evaluating the effectiveness of the implemented interventions.

The following sections detail the implementation of each phase (Figure 1).

Living lab framework

Participatory design and planning

As a living lab, the UO was designed through a participatory approach that positioned users as co-creators in shaping the site’s vision and functionality. To establish this foundation, a design competition was launched in collaboration with the Landscape Design and Ecosystem Management Department (LDEM) at AUB, inviting students and alumni to propose concepts for transforming the former FAFS Animal House into an UA living lab.

The site was selected due to its enclosed and underutilized character, reflecting the symbolism of an oasis as a protected and productive space within a resource-constrained environment (Figure 2). To encourage interdisciplinary thinking, participating teams combined expertise in landscape architecture, architecture, and agricultural engineering.

Proposals were evaluated by an independent jury based on their integration of urban agricultural principles, alignment with circularity and resource-efficiency objectives, spatial and ecological coherence, innovation, and implementation feasibility. Fifteen teams registered for the competition, of which eight submitted complete proposals. The competition generated valuable design inputs that informed the final development of the UO, including the selection of heritage-inspired features, spatial organization, and educational components integrated into the living lab.

Site rehabilitation and circular resource integration

Following the participatory design phase, the UO was rehabilitated through a low-impact adaptive reuse approach guided by the WEFE Nexus. The WEFE Nexus was applied as the planning framework to systematically identify, map, and integrate the interactions among water, energy, food, and environmental systems. During the design stage, resource flows, inputs, outputs, and feedback loops were mapped to identify opportunities for resource recovery, reuse, and efficiency while minimizing potential trade-offs between systems. By mapping the interactions among these domains during the design stage, the UO was configured as a circular system where resources could be recovered, reused, and optimized to support environmental and financial sustainability (Figure 3).

Waste minimization

Waste being an interrelated pillar among the water, energy, and food sectors, resource circularity was incorporated throughout the rehabilitation process to minimize waste generation and maximize material reuse. Existing wood residues available on campus were repurposed into furniture and architectural elements, while recycled plastic materials were incorporated into selected structural components. Composting infrastructure was installed to convert organic waste into soil amendments for agricultural production, and construction residues were diverted for reuse whenever possible. These interventions reduced material consumption while reinforcing the circular economy principles underpinning the UO (Figure 4).

Energy efficiency measure integration

To reduce reliance on conventional energy sources, the UO was designed to operate primarily on renewable energy. Electricity demands associated with lighting, monitoring systems, and operational activities were supplied through a photovoltaic system installed on a nearby facility. Energy consumption was further reduced through passive design measures, including the installation of a traditional mud wall for thermal insulation, natural ventilation strategies, shading elements, and the use of energy-efficient equipment (Figure 5).

Water harvesting and resource management

Water management was designed around the recovery, collection, and efficient use of available water resources. The UO functions as a collection and routing point for portions of the AUB Upper Campus, capturing rainfall and surface runoff for reuse in irrigation and landscape maintenance. Additional interventions, including soil improvement, compost application, and green infrastructure, were implemented to enhance water retention and reduce irrigation requirements (Figure 6).

Community food production systems

Being an UA engagement center, food production forms the central component of the UO and serves as the primary interface through which water, energy, and environmental resources interact. Hence, the site incorporated a range of agricultural features designed to demonstrate climate-responsive local UA practices, including a green roof, community gardening areas, traditional food displays, plots dedicated to local and heritage crop varieties, and small-scale animal integration (Figure 7).

Smart system deployment and artificial intelligence integration

Following the rehabilitation phase, a digital infrastructure was deployed to transform the UO into an intelligent living lab capable of monitoring environmental conditions, collecting user-interaction data, and supporting automated resource management. The system was designed to enhance the operational efficiency of the Oasis while providing a real-world platform for testing the application of AI within climate-smart UA.

A multi-layer Internet of Things (IoT) architecture was established using distributed sensors, smart controllers, and edge-computing devices. Environmental and operational parameters, including occupancy, light intensity, temperature, humidity, and appliance energy consumption, were continuously monitored through a network of sensors connected via the MQTT communication protocol. Data acquisition and system coordination were managed through a Raspberry Pi-based edge-computing platform that hosted the local database, communication services, and user dashboard (Figure 8). Specifically, a Raspberry Pi 5 (8GB) served as the central edge-computing hub, running a YOLOv5 object-detection model restricted to the ‘person’ class for occupancy sensing via a USB webcam. Door status and ambient light intensity were monitored through an ESP32-based circuit incorporating a hall-effect sensor (KY-024) and a photoresistor (LDR), while a human-presence millimeter-wave radar sensor (HLK-LD2410) filtered minor movements such as breathing to improve detection reliability. Automated switching of lighting and other loads was performed through 5 V relays and Wi-Fi-enabled smart plugs (TP-Link Tapo P110, rated 3.68 kW) supporting scheduled activation and real-time power monitoring. Communication between the Raspberry Pi and ESP32 devices was handled through a local MQTT broker (Eclipse Mosquitto), with an independent Wi-Fi access point established to work around restrictions on the campus network.

The monitoring system generated a continuous stream of environmental and behavioral data that enabled real-time visualization of resource flows within the UO. These datasets supported both educational objectives by increasing user awareness of sustainability processes and operational objectives through the development of data-driven management strategies.

To optimize resource utilization, an AI-based occupancy detection system was integrated into the platform. Using computer vision techniques and machine learning models, the system identified user presence and generated occupancy patterns that informed operational decision-making. To reduce the influence of individual missed or spurious frame-level detections on downstream control decisions, raw detections were logged and averaged over one-minute windows, with occupancy confirmed only through sustained detection over that window rather than single-frame output. The resulting occupancy predictions were combined with environmental and energy-consumption data to support adaptive control strategies aimed at reducing unnecessary energy use while maintaining user comfort.

Automated environmental control was implemented through smart relays and connected devices that regulated selected electrical loads based on occupancy and environmental conditions. This enabled the dynamic management of lighting and other energy-consuming equipment, improving overall resource efficiency and aligning energy demand with actual usage patterns.

Prior to full deployment, all hardware and software components underwent calibration and validation to ensure reliable operation under real-world conditions. Continuous system refinement was subsequently performed through iterative testing and adjustment of sensor thresholds, communication protocols, and control logic. This adaptive process improved system performance while maintaining the flexibility required for experimentation and learning within the living lab environment.

To facilitate stakeholder engagement, a web-based user interface was developed to provide access to real-time environmental data, system controls, and performance indicators. The platform enabled students, researchers, and visitors to interact directly with the smart infrastructure while observing the relationships between resource use, environmental conditions, and system behavior. Appropriate cybersecurity measures were incorporated to ensure secure operation and protect system integrity throughout the study period. User feedback on the dashboard and control interface was gathered informally through the workshops and focus-group discussions described in Section 2.2.4, and is being used to guide iterative refinements to the platform, including simplifying the display of performance indicators and adding requested controls; these refinements will be incorporated into subsequent phases through the Living Lab’s monitoring and evaluation cycle.

Stakeholder engagement and applied learning

The effectiveness of a living lab extends beyond its physical and technological infrastructure and depends on its ability to facilitate user interaction, knowledge exchange, and the adoption of sustainable practices. Consequently, the UO was designed as an experiential learning environment that engages stakeholders across local, regional, and international scales.

At the local level, the Oasis was integrated into academic courses, practical workshops, volunteering activities, and research projects at the American University of Beirut. Students engaged directly with the agricultural systems, resource-management practices, and digital monitoring technologies through hands-on learning experiences. The site also served as a platform for student research, community events, and a weekly farmers’ market that strengthened connections between urban consumers and rural producers (Figure 9).

Beyond the university campus, the UO functioned as a regional learning hub for schools, community groups, non-governmental organizations, and development actors. Educational visits, training activities, and awareness events were organized to promote UA, environmental stewardship, and the preservation of agricultural heritage. These activities facilitated knowledge transfer while expanding access to practical demonstrations of climate-responsive UA. The weekly farmers’ market further connected the Oasis directly to the surrounding neighborhood’s food system, giving nearby residents recurring, low-barrier access to fresh, locally produced goods and providing a concrete link between the Living Lab and community-based food security rather than positioning it solely as a university-run demonstration site.

At the international level, the UO contributed to research, knowledge dissemination, and partnership development. The living lab supported academic publications, conference presentations, and exchanges with international institutions, development agencies, and diplomatic delegations. These engagements enhanced the visibility of the UO while enabling the sharing of lessons learned and the exploration of opportunities for replication in other resource-constrained and climate-vulnerable contexts.

Through these multi-level engagement mechanisms, the UO extended its role beyond a demonstration site to become a platform for learning, experimentation, and the adoption of sustainable practices, supporting both capacity development and the broader dissemination of climate-responsive UA approaches.

Data collection and analysis

A monitoring and evaluation framework based on the Triple Bottom Line (TBL) approach was established to assess the environmental, economic, and social performance of the UO. Data were collected through the smart monitoring infrastructure, agricultural production records, operational logs, participation records, surveys, and project documentation. These data were used to evaluate the effectiveness of the UO as an integrated UA Living Lab and to support continuous system improvement.

Operational performance assessment

The AI model performance metrics reported in this section were collected over a six-month monitoring period. The ‘person’ class (class_id = 0) was selected for evaluation because person detection is the basis of the occupancy-detection system described in Section 2.2.3: it directly determines whether the automated controls correctly link lighting and equipment operation to actual occupancy. As a result, the precision, recall, mAP@0.5, and inference speed reported below serve as direct indicators of the reliability of the energy-saving control logic itself, rather than as general-purpose benchmarks of the underlying detection model. The ‘person’ class is also among the most extensively studied categories in the object-detection literature, spanning applications such as surveillance, human activity recognition, and autonomous systems, which allows these results to be interpreted against a well-established performance baseline. The model’s performance for the ‘person’ class is evaluated using the following metrics:

Precision: Measures the accuracy of positive predictions, calculated as: Where P (True Positives) and FP (False Positives) represent correct and incorrect detections, respectively.

mAP@0.5: The mean Average Precision at an Intersection over Union (IoU) threshold of 0.5, computed as: Where AP (Average Precision) is the area under the precision-recall curve for a specific IoU threshold.

Average inference Speed: The total processing time per image, measured as:

Environmental performance assessment

Environmental performance was assessed using indicators of resource conservation and local food production. These included annual carbon-emission reductions, energy consumption and usage, water recovered through harvesting systems, and food produced within the UO and through replicated practices adopted by users.

Energy consumption was calculated by summing the energy usage of each appliance, using platform results for LED lights and values reported by the smart plugs. For the remaining appliances, energy usage was estimated based on their rated power.

The total energy saved (kWh) demonstrated a significant reduction in consumption with the implementation of the smart system. The baseline energy consumption, when the system was running on a generator, was recorded at a mean of 57.95 kWh/week. After transitioning to solar energy and implementing the smart system, the energy consumption reduced to an average of 29.08 kWh/week. The total weekly energy savings were calculated as:

The carbon footprint reduction (kg/week) was another key metric that demonstrated the environmental impact of the energy transition. Considering that most pollution came from the energy pillar, the reduced CO₂ emissions were calculated by applying the standard emission factor for generator fuel consumption, which is typically around 0.84 kg of CO₂ per kWh of energy used (based on typical diesel generators), through the following:

It is important to note that these calculations were based on average usage patterns and system efficiency, with a few assumptions made for the sake of estimating savings. Variations in real-world usage, environmental conditions, and system calibration may introduce a margin of error, particularly due to potential sensor and automation variability. Nonetheless, these results offer a clear picture of the smart system’s impact on both energy consumption and carbon footprint reduction. It should be noted that the carbon footprint calculation presented here is limited to emissions associated with energy use, in line with the study’s focus on the impact of the smart energy management system; it does not constitute a full life-cycle carbon assessment of the Urban Oasis. Other environmental performance dimensions relevant to urban agriculture, such as water quality and soil health, require dedicated sampling and laboratory analysis that fell outside the scope of the current monitoring infrastructure, which was designed primarily around energy, occupancy, and environmental sensing. These dimensions were not quantitatively assessed in this study and are identified as priorities for future monitoring and evaluation.

Water recovery and food production were quantified using direct measurements and production records maintained throughout the monitoring period.

Economic performance assessment

Economic performance was analyzed to determine the savings achieved through smart automation. With an initial investment of $3,000 for the components and system’s installation, the return on investment (ROI) can be calculated as follows:

It is good to note that additional revenue generated through produce sales was incorporated into the final ROI calculation.

Social performance assessment

Social performance was assessed through pre and post-tests and focus group discussions, estimating stakeholder participation, engagement, knowledge generation, and practice adoption. Indicators included the total number of users interacting with the Oasis, participant retention rates, adoption rate of sustainable practices, and academic outputs such as publications, conference presentations, and graduate research.

The performance of the UO was evaluated over a three-year period using the monitoring and evaluation framework described in Section 2.2.5.

Resource usage

Before presenting the operational, environmental, economic, and social performance results in the following sections, this section describes the results of the WEFE mapping that was done at the design phase (Figure 10). It details how the UO’s water, energy, and food systems are physically and operationally connected. This is important considering that the results that follow are outcomes of this interconnected system rather than independent measures of isolated components.

Water and food production are linked through both direct supply and soil-mediated pathways. Rainwater harvesting transforms rainfall and surface runoff into a productive resource rather than allowing it to leave the site as unmanaged stormwater; recovered water is stored and redistributed to support crop cultivation, green roofs, heritage crops, and landscaped areas. Composting reinforces this relationship from the soil side; organic residues returned to cultivated soils increase organic matter and improve water-holding capacity, allowing irrigation water to remain available to crops for longer periods. This relationship also carries a trade-off: increasing agricultural production raises water demand, particularly during dry periods when rainfall is limited, and expanding hydroponic systems, introducing new crops, or increasing cultivated area would require additional water, reducing system flexibility under prolonged drought. Storage capacity itself is constrained by available space and infrastructure costs, so periods of high rainfall may exceed storage capacity, while dry periods require prioritizing water allocation among food production, biodiversity planting, landscape maintenance, and educational activities.

Water and energy are connected through the infrastructure that operates both: water recovery, storage, monitoring, and distribution rely on energy to run pumps, sensors, communication systems, and control equipment, while the transition from generator-based to solar electricity reduces the operational emissions associated with these processes and improves their reliability. This relationship also introduces a trade-off, since technologically intensive water-management infrastructure requires greater dependence on maintenance, technical expertise, and reliable electricity supply than simpler systems would.

Energy and food are connected through the equipment that sustains production: renewable energy powers the monitoring systems, hydroponic equipment, communication devices, and environmental sensors that support the continuity and efficiency of food production; a relationship that matters particularly given Lebanon’s unreliable electricity supply, where interruptions could otherwise compromise irrigation schedules or controlled-environment agriculture. Expanding automation or monitoring equipment to strengthen this relationship further increases capital investment and maintenance requirements; the Urban Oasis addresses part of this trade-off by prioritizing passive cooling, natural ventilation, and efficient building design to reduce energy demand before renewable energy is introduced, rather than relying on renewable capacity alone.

These water-energy, energy-food, and water-food linkages mean that a change in one resource system produces measurable effects on the others; the performance results presented in Sections 3.2 to 3.5 should be read with this in mind.

Operational performance

The AI-enabled monitoring and automation platform operated reliably throughout the monitoring period, demonstrating its suitability for supporting adaptive resource management within the Urban Oasis. The occupancy detection model achieved a precision of 87.5%, indicating a high proportion of correctly identified users, while the mean Average Precision (mAP@0.5) reached 70.2%, confirming satisfactory overall detection performance. The model’s recall for the ‘person’ class was 58.3%, corresponding to a false-negative rate of 41.7%, indicating that a substantial share of individual detections was missed on any single frame. The system maintained an average inference speed of approximately 258 ms, enabling near real-time occupancy detection and automated operational control (Figure 11).

These results demonstrate that the digital platform was capable of continuously monitoring user activity and supporting data-driven management decisions while maintaining response times appropriate for real-time operation within the Living Lab. The comparatively lower recall indicates that not every individual was detected within a single frame; however, because occupancy was confirmed through the one-minute averaging window described in Section 2.2.3 rather than single-frame detections, the effect of these missed frames on the reliability of the automated energy controls remained limited.

Environmental performance

The integration of renewable energy, circular resource management, AI-assisted monitoring, and climate-smart agricultural practices generated measurable environmental benefits across multiple components of the WEFE system.

Implementation of the smart energy management platform, combined with the transition from generator-based electricity to solar power, reduced weekly electricity consumption by 28.87 kWh, resulting in an estimated reduction of 24.25 kg CO₂ per week, equivalent to approximately 1.26 tCO₂e annually. These results demonstrate the contribution of intelligent energy management to reducing operational emissions while improving energy efficiency.

Water management interventions, including rainwater harvesting and redistribution systems, enabled the recovery of approximately 480,000 L of water annually, providing an alternative water source for irrigation and landscape maintenance while reducing dependence on external freshwater supplies.

Food production represented another major environmental outcome. During the three-year assessment period, the Urban Oasis produced approximately 8,500 kg of fresh produce within the site. In addition, participants’ replication of demonstrated agricultural practices generated an estimated 1,200 kg of food production outside the Urban Oasis, indicating that the project’s environmental benefits extended beyond the Living Lab’s physical boundaries.

Economic performance

The Urban Oasis generated measurable economic benefits through reductions in operational costs associated with energy management and resource optimization. Based on the initial installation costs and the estimated annual savings from the integrated smart management system, the project is projected to achieve a 11.4% return on investment (ROI).

When income generated through the sale of agricultural produce was incorporated into the analysis, the estimated ROI increased to approximately 13%, demonstrating that integrating climate-smart agriculture with resource-efficient technologies can contribute to the long-term financial sustainability of urban Living Lab initiatives.

Social performance and knowledge transfer

The Urban Oasis demonstrated substantial social impact through stakeholder engagement, education, and knowledge dissemination. During the three-year monitoring period, more than 5,000 participants engaged with the Living Lab through university courses, practical workshops, volunteer activities, research projects, community events, and educational visits.

Stakeholder engagement remained consistently high throughout the implementation period, with a participant retention rate of 72%, indicating sustained interaction with the Living Lab and continued participation in its activities. Survey findings further revealed that 62.7% of participants adopted, or expressed their intention to adopt, at least one sustainable practice after engaging with the Urban Oasis, demonstrating the project’s contribution to behavioral change and knowledge transfer beyond the demonstration site.

The Urban Oasis also served as a platform for research and innovation, contributing to the production of two scientific manuscripts, five conference presentations, and one graduate thesis. Collectively, these academic outputs, together with the observed adoption of sustainable practices and the replication of agricultural production beyond the site, provide evidence of the Urban Oasis’ potential to support wider dissemination and replication of integrated climate-smart urban agriculture approaches.

Contribution of the integrated framework

The results indicate that the UO’s performance stems not from any single intervention but from the integration of three complementary pillars, each addressing a key limitation commonly observed in UA initiatives (Owen and Drake, 2023; Fox-Kämper et al., 2023; Teoh et al., 2024): (i) WEFE-informed resource integration, addressing the tendency of UA projects to target a single objective in isolation; (ii) AI-enabled monitoring, addressing the frequent absence of robust monitoring and evaluation mechanisms; and (iii) Living Lab-based stakeholder engagement, addressing the concentration of participation during implementation rather than throughout the project’s lifecycle. In relation to RQ1, the results presented in Sections 3.1–3.3 demonstrate that integrating these three pillars significantly enhanced the UO’s overall performance. Specifically, the WEFE mapping presented in Section 3.1 illustrates how water, energy, and food systems were intentionally designed to reinforce one another rather than function as siloed interventions, consistent with the principle that these resources should be managed as interdependent systems rather than separate sectors (Al-Saidi and Hussein, 2021; Pokhrel et al., 2022; Simpson and Jewitt, 2019). Furthermore, Sections 3.2 and 3.3 demonstrate, through a TBL assessment, that this integrated framework generated substantial environmental, social, and economic improvements, highlighting the added value of the combined model for adopting UA.

Strengths, trade-offs, and replication potential of the three pillars

While the three pillars collectively underpin the UO’s performance, each can also be evaluated individually, revealing both distinct strengths and inherent limitations. As seen in section 3.1, the resource integration, for instance, involves trade-offs rather than only uniform gains. Between water and food, increasing agricultural production raises water demand, particularly during dry periods, and storage capacity remains constrained by available space and infrastructure costs, so that high-rainfall periods may exceed storage while dry periods require prioritizing allocation among production, biodiversity, and educational uses; a tension that echoes findings elsewhere that rainwater harvesting strengthens urban agriculture’s resilience only when paired with adequate storage and distribution planning (Fox-Kämper et al., 2023; Al-Saidi and Hussein, 2021). Between water and energy, and energy and food, technologically intensive infrastructure requires greater dependence on maintenance, technical expertise, and reliable electricity supply than simpler systems would; the UO addresses part of this trade-off by prioritizing passive cooling, natural ventilation, and efficient building design to reduce demand before renewable capacity is introduced, rather than relying on renewable energy alone. These trade-offs indicate that the benefits documented here depend on active, continuous balancing of interacting resource demands rather than being an automatic outcome of integration itself; a distinction directly relevant to RQ3, discussed further below.

AI-enabled monitoring functioned as the second pillar, operating not as an independent resource but as an adaptive management layer that made the WEFE interactions above legible and actionable in real time, consistent with the broader shift toward data-driven decision support in digital agriculture (Lv and Sarker, 2024; Fei et al., 2025; Rojek et al., 2026; Snehalatha and Patil, 2026). The occupancy-detection model’s precision of 87.5% and mAP@0.5 of 70.2% indicate that the system reliably linked automated lighting and equipment control to actual occupancy, directly supporting the energy savings reported above; the comparatively lower recall (58.3%) shows a genuine limitation in single-frame detection, but because control decisions relied on the one-minute averaging window described in Section 2.2.3 rather than individual frames, this limitation had a bounded effect on the reliability of the energy-management outcomes it was meant to support. This also points to a constraint relevant to RQ3: replicating this pillar elsewhere requires reliable connectivity, sensor calibration, and technical maintenance capacity, which may not be available in all resource-constrained settings.

The third pillar, Living Lab-based stakeholder engagement, addresses RQ2 directly and reflects the growing recognition of living labs as open innovation ecosystems that strengthen engagement by situating research and innovation in real-life settings rather than isolated demonstrations (Yan et al., 2022; Gibert and Plantec, 2026; Sofronievska et al., 2026). The 72% participant retention rate and the finding that 62.7% of participants adopted or intended to adopt at least one demonstrated sustainable practice indicate that experiential, hands-on engagement, rather than one-off exposure, was associated with sustained participation and self-reported behavioral intent. This suggests the Living Lab model was effective at promoting adoption intention, though intended adoption does not guarantee sustained real-world practice change, and the study did not track participants’ behavior after they left the UO to confirm this translation.

Taken together, these results speak to RQ3 and to the two hypotheses framing this study. Consistent with H1, the UO’s implementation depended on supportive institutional governance and on stakeholder collaboration across the participatory design and engagement phases, echoing evidence that continuous participation and local ownership, rather than one-off consultation, are what sustain UA interventions over time (Owen and Drake, 2023); the study did not evaluate regulatory frameworks directly, so this remains an enabling condition inferred from context rather than directly tested. Consistent with H2, the sustainability of the model to date has depended on institutional support (continued AUB backing) and on the financial viability demonstrated by the projected 11.4–13% ROI, alongside sustained stakeholder engagement evidenced by the retention and adoption figures above. Replication elsewhere would likely depend on the presence of similar enabling conditions, including institutional backing, initial capital investment, and a stakeholder base willing to engage iteratively, rather than on directly replicating the specific technologies deployed at the UO. This finding is consistent with prior observations that single-objective UA interventions often struggle to scale because they neglect these enabling, cross-cutting conditions (Fox-Kämper et al., 2023).

Broader implications for food security and sustainable development

More broadly, the integrated outcomes of the UO map onto several Sustainable Development Goals (SDGs), consistent with previous evidence that Urban Living Labs can contribute simultaneously to multiple SDGs (Brzeziecka et al., 2024), while also strengthening the four pillars of food security: availability, access, utilization, and stability. These outcomes are particularly relevant to Mediterranean and other climate-vulnerable regions where similar UA-based Living Lab demonstrations are increasingly being piloted (Hoff et al., 2019; MedECC, 2024). Localized food production and the promotion of climate-smart agricultural practices support SDG 2 (Zero Hunger) by strengthening food-system resilience rather than food production alone. Rainwater harvesting, water reuse, and AI-assisted irrigation management contribute to SDG 6 (Clean Water and Sanitation); renewable energy integration and smart energy management support SDG 7 (Affordable and Clean Energy); the adaptive reuse of an abandoned urban space into a Living Lab, combined with community engagement, advances SDG 11 (Sustainable Cities and Communities); composting and circular resource management, consistent with findings elsewhere on composting’s role in closing urban agriculture’s resource loops (Fox-Kämper et al., 2023; Nikolaidis et al., 2025), contribute to SDG 12 (Responsible Consumption and Production); and the integration of climate adaptation and resource efficiency supports SDG 13 (Climate Action). Through localized food production, community engagement, sustainable resource management, and climate-resilient practices, the UO contributes to food availability, improves access to locally produced food, supports utilization through sustainable production systems and environmental quality, and enhances the stability of food systems by reducing vulnerability to resource scarcity and climate-related shocks.

Positioning the urban oasis within regional and international research

The findings of this study align with a growing body of regional initiatives demonstrating the potential of integrated approaches to strengthen climate resilience and sustainable agriculture across the MENA region. On resource integration, previous implementations in Lebanon and Jordan have demonstrated that combining renewable energy, wastewater reuse, sustainable irrigation, and circular resource management can improve agricultural productivity while reducing pressure on limited natural resources (Hoff et al., 2019), and more recently, regional WEFE demonstration projects and Living Labs have promoted participatory approaches that integrate sustainable irrigation, renewable energy, hydroponics, governance, and stakeholder engagement across Mediterranean countries (MedECC, 2024). On participatory engagement specifically, living labs have gained prominence internationally as open innovation ecosystems that strengthen long-term stewardship by situating research in real-life settings rather than isolated demonstrations (Yan et al., 2022), while on digital agriculture, AI and sensor-based monitoring are increasingly recognized as tools for generating the real-time evidence needed to evaluate and continuously improve resource management (Snehalatha and Patil, 2026).

While these initiatives demonstrate growing regional adoption of each pillar individually, they largely do so in isolation. In contrast, the Urban Oasis expands this perspective by integrating all three within a single operational urban Living Lab, enabling the simultaneous assessment of environmental, social, and economic outcomes.

Study limitations and future research

This study has several limitations that should be considered when interpreting the findings. First, although the Urban Oasis demonstrates the operational integration of water, energy, food, and ecosystem resources, not all WEFE interactions could be quantified with equal precision. Resource allocation, crop-specific water productivity, and nutrient recovery were assessed only partially, resulting in some interactions being interpreted qualitatively. Second, while the AI-enabled monitoring system effectively supported operational management, its performance depended on reliable connectivity, sensor calibration, and technical maintenance, which may limit replication in resource-constrained settings. Third, stakeholder outcomes were measured primarily through self-reported adoption intentions rather than longitudinal behavioral change. Fourth, the study represents a single Living Lab within a specific institutional and environmental context, limiting the direct generalizability of the quantitative findings to other settings. Finally, future research should focus on developing and validating a unified monitoring and evaluation framework for the proposed three-pillar model, addressing a gap that has also been identified in the evaluation of environmental and agricultural Living Labs (Berberi et al., 2021). Such a framework should establish standardized indicators across the WEFE resource integration, AI-enabled monitoring, and Living Lab engagement pillars to enable consistent performance assessment, facilitate comparisons across different implementation contexts, and support evidence-based replication and scaling.

This study evaluated the UO at the American University of Beirut as an integrated UA framework combining Living Lab principles, WEFE-informed resource management, and AI-enabled technologies. The framework was assessed using the TBL approach, evaluating its environmental, economic, and social sustainability performance. The results demonstrated positive outcomes across all three dimensions, including improved resource efficiency, measurable environmental benefits, operational savings, and high levels of stakeholder engagement and adoption.

The findings suggest that the effectiveness of the UO stemmed not from any individual component, but from the interaction among them. Stakeholder participation facilitated adoption, integrated resource management supported environmental and economic sustainability, and digital technologies enhanced optimization, transparency, and evidence generation. Together, these components created a reinforcing system capable of simultaneously addressing adoption, sustainability, and performance improvement.

This study, therefore, provides empirical evidence supporting an integrated framework for climate-resilient UA. As cities continue to face increasing pressures related to climate change, resource scarcity, and food-system vulnerability, such integrated approaches may offer a practical pathway for developing resilient, sustainable, and scalable UA systems in the MENA region and beyond.

Future research should evaluate the applicability of this framework across different urban contexts, scales, and socioeconomic settings. Particular attention should be given to assessing long-term behavioral change, city-scale impacts, and the potential integration of advanced predictive analytics and decision-support tools to enhance further UA planning, resource management, and climate resilience.

Overall, the findings demonstrate that integrating WEFE-informed resource management, AI-enabled monitoring, and Living Lab-based stakeholder engagement provides a practical and transferable framework for advancing UA. By emphasizing the interaction among these three pillars rather than isolated interventions, the proposed model offers a foundation for guiding future UA initiatives in the MENA region and other resource-constrained settings.