A living lab approach to co-designing climate adaptation strategies

1. INTRODUCTION

Climate variability poses significant challenges to the livelihoods of people in urban, peri-urban interfaces and rural areas in semi-arid tropics of southern India (Zewdu et al. 2025). It affects highly vulnerable informal dwellings in rapidly urbanising Sub-Saharan Africa due to lack of preparedness, making effective climate mitigation and adaptation strategies imperative (Hugo 2023). Concerns over civic resilience have extended to Mexico City’s peri-urban interface, where including the food production industry in urban climate policies can foster adaptation initiatives and significantly enhance their effectiveness (Bausch et al. 2018). Civic resilience refers to the collective capacity of communities to anticipate, absorb and adapt to climatic stresses through processes of learning, experimentation and collaborative adjustment, aligning with Folke’s (2016) framing of resilience as a transformative, rather than a purely reactive, capacity. The United Nations’ Sustainable Development Goals (SDGs) 9, 11 and 13 advocate for climate mitigation through sustainable infrastructure, urban planning, ecological transition and climate policies (Karakadzai & Chirisa 2024), whereas SDG 2 underscores the need for climate adaptation in food production to ensure long-term sustainability, civic resilience and food security (Abed et al. 2025; Barati et al. 2024).

Despite recent advances in climate adaptation strategies through technological interventions, evidence shows that their effectiveness remains limited due to a persistent disconnect between technical design and community context (Ensor & Harvey 2015). This gap has contributed to unresolved challenges, including the digital divide, high up-front costs, complexity and environmental footprints, that hinder stakeholders’ adoption of adaptation measures (Parra-López et al. 2024). A key factor is that technological innovations have historically ignored or only partially involved communities in design, implementation and dissemination (Habanyati et al. 2024, 2025), leading to resistance, limited adoption and reduced impact of adaptation strategies (Preet & Chahal 2024: 47–68). Although technological advancements are essential, they remain insufficient to ensure successful and long-term ecological transitions (Axon & Morrissey 2020), as they often lack meaningful pathways for citizen engagement and overlook Indigenous knowledge and socio-economic realities within the adaptation process (Kangana et al. 2024). This limitation stems not merely from implementation challenges but from a deeper epistemological conflict in which top-down technological solutions fail to account for the situated forms of knowledge that ultimately shape local adoption (Olazabal et al. 2024). Integrating Indigenous knowledge with scientific research has been shown to enhance the effectiveness of adaptation solutions, yet this integration is best achieved through participatory approaches that position communities as active contributors to decision-making (Onyango et al. 2023). Despite broad recognition of this need (Parsons et al. 2025), most existing frameworks remain trapped in a consultative rather than co-creative paradigm, collecting local input without enabling genuine collaborative design. This critical synthesis highlights the lack of structured methodologies that can systematically bridge technical feasibility with socio-cultural legitimacy. Living labs have increasingly emerged as one approach capable of addressing this gap by embedding collaborative knowledge production within real-world contexts (Belfield 2025).

Existing research on participatory approaches and experiential learning has advanced understanding of how co-creation can support adaptation (Bhatta et al. 2025a; Lacombe et al. 2018), yet critical tensions persist. As socio-technical systems of learning and experimentation, living labs aim to foster iterative learning and co-creation (Hernández et al. 2009). Although experiential learning forms a theoretical foundation of living labs frameworks, its principles are applied inconsistently in practice, leaving their contribution to adaptive knowledge generation and co-creation underexamined (Bhatta et al. 2025a). Previous studies highlight persistent gaps between technological innovation and local needs, often arising from mismatches between expert-driven solutions and farmers’ situated knowledge (Eastwood et al. 2022; Hölting et al. 2022). Despite their emphasis on inclusivity (Rădulescu et al. 2022), many living labs’ implementations struggle to translate participatory insights into actionable interventions when analytical tools or policy frameworks do not align with on-farm realities (Prost 2021). Moreover, while participatory methods promote dialogue and learning, they frequently lack mechanisms for systematically linking experiential knowledge with scientific evaluation, limiting their ability to generate robust adaptation strategies (Banerjee et al. 2019; Ditzler et al. 2018). These limitations highlight a broader challenge: participatory efforts often remain consultative rather than genuinely co-creative, particularly in attempts to integrate Indigenous knowledge, where local insights are collected but rarely incorporated into design decisions (Filho et al. 2025).

Although living labs have been widely applied across sustainability research, the specific role of modelling tools within participatory frameworks remains underexplored (Jamieson & Martin 2022). Much of the existing literature continues to position modelling as a primarily technical and expert-driven activity, rather than as an interactive medium capable of facilitating dialogue, shared understanding and joint problem exploration (Hedelin et al. 2017; Voinov et al. 2016). This limitation is especially significant in climate adaptation, where analytical assessments must be integrated with local experiential knowledge and context-specific insights. By foregrounding participatory modelling within a living lab setting, this study addresses this overlooked area and illustrates how modelling can function as a boundary object that bridges experiential learning with scientific assessment (Grogan 2021), prevents incompatibilities during co-design (Yunyun & Broenink 2014) and aligns shared insights with evidence-based assessment (Banerjee et al. 2019). Yet most participatory frameworks still use modelling in minimal or isolated ways, limiting its contribution to co-design and decision-making (Moallemi et al. 2021). Recent assessments further show that many participatory modelling initiatives remain technically complex and only partially collaborative, with stakeholders rarely engaged across the full modelling cycle or supported through iterative reflection and uncertainty exploration (Hedelin et al. 2017; Klemm 2024). By translating complex biophysical processes into visual and interpretable outputs, process-based modelling offers a more accessible and cognitively manageable way for farmers to engage with scenario exploration and decision-making, thereby addressing several of these limitations. This constraint is especially consequential in climate-vulnerable regions such as semi-arid tropics, where Indigenous communities are among the first to detect shifts in weather patterns and resource availability (Barati et al. 2022). Collectively, these tensions frame the core research problem addressed in this study: the need for participatory frameworks that not only facilitate co-creation but also meaningfully integrate modelling-based evidence to produce context-specific, locally validated, and scalable adaptation strategies.

Building on these limitations, this study advances ongoing debates on co-design and experiential learning by operationalising Kolb’s experiential learning cycle within a living lab setting (Kolb 1984). A process-based crop model is employed as an interactive boundary object that enables farmers to iteratively interpret, test and refine adaptation options through both experiential and analytical forms of learning. By linking simulation-based assessment with experiential learning and farmer-led reflection, the study demonstrates how participatory modelling can move beyond consultation toward genuinely co-constructed knowledge production.

Collaborative design (co-design), an essential component of living labs, refers to a collaborative and iterative process in which farmers and researchers jointly diagnose problems, generate knowledge and evaluate adaptation options. This understanding aligns with critical participatory design theory (CPDT), which emphasises shared agency, reflexive learning and social innovation (Sanders & Stappers 2008). Integrating Kolb’s experiential learning theory (ELT) within this co-design approach positions learning as both an outcome and a driver of resilience (Kolb 1984). This co-design approach facilitates active engagement from community members, ensuring that solutions are relevant, contextually appropriate and sustainable. It employs participatory tools, such as storytelling (Rodriguez Vega 2022), mind mapping, sketch-assisted brainstorming (Varma et al. 2021), modelling (Banerjee et al. 2019), scenario-based planning and implementation (Habanyati et al. 2025), to enhance ecological transitions (Mallam et al. 2021). Co-design practices within a living labs platform contribute to the transparency of policymaking processes (Rogers et al. 2021). The effectiveness of these practices ultimately depends on the extent to which they are integrated with analytical feedback and iterative learning cycles, a contribution that modelling can uniquely strengthen within participatory settings.

This study is positioned within the transdisciplinary domain of living labs as agents of socio-ecological change, examining how participatory frameworks mediate between diverse knowledge systems to foster civic resilience. It is guided by four interconnected research questions that investigate both the procedural dynamics and the transformative outcomes of such collaborations. First, it asks what participatory processes and mediating mechanisms within a living lab most effectively integrate Indigenous knowledge with scientific modelling to jointly diagnose climate vulnerabilities and co-produce contextually robust adaptation strategies. Second, it examines how iterative cycles of experiential learning, concrete experience, reflective observation, abstract conceptualisation, and active experimentation shape the refinement and local legitimisation of co-designed solutions. Third, it investigates the transformative effects of participatory modelling on farmers’ agency, trust in scientific tools and perceived adaptive capacity. Finally, it explores what configurations of socio-ecological resilience emerge through the co-design process and how these manifest across biophysical (yield and water stress) and socio-cognitive (knowledge and agency) dimensions. To address these questions empirically, the study pursues four specific objectives: (1) to develop and implement a living lab-based participatory framework for climate adaptation; (2) to assess the effectiveness of co-designed strategies in mitigating water stress through crop modelling; (3) to evaluate how participatory modelling shapes adoption through pre- and post-intervention assessments; and (4) to examine how experiential learning contributes to long-term resilience-building through implementation feedback and reflective practice. By integrating scientific modelling with Indigenous knowledge, the study enhances the practical relevance of participatory approaches and provides a foundation for developing policy-oriented frameworks for climate-resilient agriculture.

2. METHODOLOGY

This case study was conducted in Sadivayal village of Coimbatore district, in the foothills of the Western Ghats in southern India, using a living lab approach to learn traditional knowledge, experience real-life challenges, co-create and implement sustainable solutions with the subsistence farming community. The living lab was implemented between 2023 and 2025 through a structured sequence of field engagement, participatory diagnosis, co-design workshops, iterative co-modelling cycles and post-implementation feedback sessions. Kolb’s ELT provided the overarching structure for the participatory methodology, framing learning as a cyclical process of concrete experience, reflective observation, abstract conceptualisation and active experimentation. Within this framework, the Live-in-Labs experiential learning platform translated field experiences into actionable design insights through five sequential steps (Figure 1). Workshops and engagement activities were conducted in three main locations: the village school, which served as the primary meeting space; small field shelters adjacent to the farms, where several modelling-related discussions were conducted; and open areas next to the fields, where participatory exercises took place (Figure 2).

Figure 1

Figure 2

The living labs process began in early 2023 with the initial diagnostic stage aligned with the concrete experience phase, which involved gathering firsthand community data using participatory rural appraisal (PRA) tools such as transect walks, resource maps and seasonal calendars. These activities engaged all 20 farming households and enabled the joint identification of climate vulnerabilities, livelihood challenges and adaptive capacities, including locally available resources and skills. All 20 farming households in the village were invited to participate, and involvement in the workshops and modelling sessions was entirely voluntary. This was followed by reflective observation activities conducted in June and September, where insights from field journals, problem-tree analysis and farm surveys were synthesised to clarify the main climate-related challenges faced by the community. During the abstract conceptualisation phase in late 2023, potential solutions were developed by integrating community insights, analysing long-term rainfall patterns and refining ideas through focus group discussions. A structured brainstorming session with farmers was conducted to prioritise feasible adaptation options.

Building on this foundation, three co-design workshops were held within the active experimentation phase between late 2023 and early 2024, using sketching, mind mapping and storytelling methods to capture farmers’ experiential knowledge and translate it into design-relevant inputs. Each technique generated distinct types of qualitative data, which were subsequently analysed and integrated into the participatory modelling workflow.

Participatory mind mapping was conducted with 12 farmers, organised into four groups of three, each supported by trained Live-in-Labs facilitators. After short introductory interactions to create a comfortable environment, farmers were presented with the central theme of ‘rice cultivation under water stress’. This theme was placed at the centre of each map, from which primary branches such as soil, irrigation, rainfall and river/canal flows were extended. Farmers were encouraged to freely express their observations and experiences in response to guiding questions such as: What makes rice cultivation difficult?; What happens when water becomes scarce?; How long are the intervals between irrigations under normal conditions?; and How influential is each rainfall event? Their inputs, examples and explanations were recorded visually on the maps, reflecting the interconnected relationships they perceived (e.g. rainfall timing to river flow to irrigation decisions). Once group maps were completed, they were reviewed collectively with participants to validate linkages. The final maps were then coded and synthesised into thematic clusters that represented shared constraints and priorities, which subsequently informed parameter adjustments in the model and the development of scenario designs.

In the sketching session, farmers visually articulated their mental models of cultivation challenges and water-related stress. Building on the energy created through earlier ice-breaker interactions and the shared understanding developed during mind mapping, facilitators supported farmers as they worked in small groups to translate their lived experiences into drawings. Farmers were invited to sketch situations reflecting their current cultivation challenges, such as river flow timing, water accessibility, erratic monsoon onset, drying field channels, empty wells or the struggle to protect young seedlings during dry spells, while facilitators encouraged discussion and ensured that each group member contributed. Once sketches were completed, groups presented them and described the experiences behind each drawing, allowing the research team to capture shared practical concerns across groups.

Storytelling was used in the co-design workshops to capture farmers’ lived experiences with water scarcity and rainfall uncertainty. After the mind mapping and sketching activities, farmers shared short personal accounts describing how erratic monsoons, drying streams or sudden rainfall shifts affected their sowing decisions and water security. Several participants recounted incidents from recent growing seasons, explaining how unexpected dry spells had directly harmed their crops. These narratives also revealed field-specific situations, allowing the team to understand water stress, crop loss and climate impacts on a case-by-case basis that had not been fully articulated in earlier workshops.

The analysis revealed several patterns that directly shaped the modelling framework. Insights from the mind mapping clusters clarified how farmers linked rainfall timing, soil moisture and irrigation constraints, which helped refine the Food and Agriculture Organization’s (FAO) AquaCrop inputs such as planting dates, irrigation intervals and initial soil-water content. Storytelling outputs were reviewed case by case to understand field-specific water stress, losses experienced under erratic monsoon behaviour and the practical limits shaping farmers’ decisions. When these strands of evidence were brought together, ‘adjusting sowing time’ consistently emerged as the most feasible and widely supported adaptation pathway. This collective preference guided the prioritisation of sowing-date scenarios selected for simulation, ensuring that the modelling reflected both biophysical realities and everyday farming experiences.

The process then transitioned to simulation-supported learning through two participatory modelling workshops conducted between May and August 2024, each spaced one month apart to allow farmers time to review model outputs, reflect on scenario implications and provide informed feedback. Twelve farmers from the 16 climate-affected households participated in these sessions. In this step, crop modelling combined with Coupled Model Intercomparison Project (CMIP6) climate projections (Döscher et al. 2022) was employed to design and evaluate sustainable strategies under future scenarios. AquaCrop, a user-friendly process-based model with intuitive interfaces, served as an interactive boundary object during these workshops, enabling participants to iteratively refine parameters, make real-time modifications, respond effectively to changing conditions, receive actionable insights, interpret scenario outcomes and adjust the proposed sowing strategy.

Farmers actively contributed to the model set-up by providing data on crop characteristics, soil conditions and farm management practices, ensuring the farming systems were accurately represented (Figure 3). The AquaCrop model simulated rice yield under different sustainable solutions, namely 19th, 23rd, 27th, 38th, 40th and 42nd Standard Meteorological Weeks (SMWs) as sowing dates through a participatory approach under historical (Barati et al. 2024) and future climate scenarios (Barati & Soundharajan 2025) derived from CMIP6 models (Huang et al. 2024). Daily temperature and precipitation datasets (with a high resolution of 0.25^o), downscaled for India by Mishra et al. (2020), were used under two shared socio-economic pathways (SSPs), i.e. SSP245 and SSP585 for the period 2015–59. A simplified temperature-based Penman–Monteith model (TPM) derived from the full Penman–Monteith (PM) equation was used to calculate the daily reference crop evapotranspiration (ET_o) values as input for simulation model (Allen et al. 1998).

Figure 3

The maximum attainable yield under no stress (Y_m) and actual (under stressful conditions, Y_a) rice yield were simulated under each SSP (Raes et al. 2023) and leveraged to compute resilience indicators and water stress index (WSI) using the following formulas:

where ET_a and ET_m are actual and maximum crop evapotranspiration, respectively, calculated by AquaCrop; and σ is the standard deviation of yield-reduction ratios from 2015 to 2059. To ensure that model evaluation responded to farmers’ priorities, the WSI and resilience indicator were selected as the primary performance metrics based on explicit participant input. A second policy-relevant rationale guided this choice by translating farmers’ experiential knowledge into standardised and comparable metrics that policymaking and extension systems can use to evaluate, prioritise and scale adaptation options across contexts. During participatory activities farmers repeatedly identified water stress, timing of rainfall events and yield instability as their main concerns; these experiential priorities guided the interpretation of indicator. Provisional WSI and resilience outputs were iteratively reviewed with participants during the modelling workshops so that the final indicators reflected locally meaningful conditions rather than purely technical conventions.

Between modelling sessions, individual follow-up visits were conducted to maintain engagement, particularly for farmers who worked off-farm or resided in neighbouring hamlets. After the modelling cycle concluded, a final reflection workshop was held to review participants’ experiences, consolidate insights from the active experimentation phase and assess readiness to implement the co-designed early sowing strategy. Additional meetings during the 2025 sowing period and early crop growth stages facilitated continuous feedback and local validation of model predictions.

In the last step, the effectiveness of participatory modelling framework was assessed by conducting a structured interview, using a combination of score comparison methods for knowledge assessment and Likert-type scales for measuring participants’ interest (Figure 4a). The survey, comprising five to 10 questions, targeted their grasp of the challenges posed by water and rainfall variability and their perceptions of how co-designed adaptation measures could address these challenges effectively.

Figure 4

The interview focused on understanding farmers’ awareness and perceptions about critical issues such as water availability, water scarcity and rainfall variability. Additionally, it explored their insights into potential solutions and the importance of sustainable measures, such as the adaptation strategies co-designed during the intervention. Furthermore, in order to bridge the gap between design and practical application, the co-designed solution was implemented by farmers. Several field schools were conducted to gather feedback on the challenges faced by farmers and to obtain ideas for making the design more tailored and context-specific (Figure 4b). This iterative stepwise framework forms the backbone of Live-in-Labs programme interventions, allowing students, researchers and farmers to collaboratively address agricultural sustainability challenges.

3. RESULTS

Using PRA tools during the first two steps of the living lab approach, extensive experiential research was conducted to experience the community issues and identify the main challenges faced by the farmers (Barati et al. 2023). In addition, drawing on findings from an earlier baseline farm survey was conducted by Abed et al. (2025), the sustainability status of farming systems and contributing factors were understood. Resource mapping demonstrated both the resources that exist and those that are lacking but essential. For example, the presence of water sources, such as a man-made canal, underscored the accessibility to water. The vital observations from the transect walk with farmers highlighted the traditional farming practices and dependence on a seasonal river and its flow patterns. Any signs of water scarcity or potential issues related to changing weather patterns were identified. Subsequently, during brainstorming sessions, several challenges were identified, including yield loss due to rainfall variability, poor water accessibility due to river flow fluctuations, animal attacks, leading to unproductive farming and financial issues. To identify the key challenges, the community actively engaged in analysing a problem tree and exploring agricultural limitations, financial constraints and total yield loss. Finally, climate change-induced erratic monsoon rainfall was identified as the prime challenge, leading to inconsistent water availability and intermittent dry spells during the rice-growing period, which underscores the need for a context-specific adaptation measure (Table 1).

The insights gained through PRA enable the formulation of targeted solutions, as shown by Sriharsha et al. (2021), where integrated watershed development and integrated pest management technologies were co-designed with local experts and farmers to address water scarcity effectively. Following community interactions in the third step by PRA and co-design tools such as focus group discussions, and rainfall analysis, three solutions were proposed, including on-farm trenches, gate irrigation technique and shifting sowing time (Figure 5). The brainstorming session ultimately highlighted shifting sowing time as the most effective climate adaptation strategy for resilient rice farming in Sadivayal. Although two other alternatives initially appeared affordable and technically feasible, farmers emphasised that both faced substantial administrative hurdles due to the village’s location within a forest-governed zone, where any structural intervention requires lengthy approvals from the Forest Department.

Figure 5

Farmers also clarified that despite rainwater entering the village from surrounding hills through waterfalls, flowing in seasonal streams and man-made canals, the highly variable flow intensity makes these sources unreliable. Water often arrives intermittently and at very low discharge, meaning that on-farm trenches cannot be consistently filled or maintained. This practical challenge, which surfaced through collective discussion, reinforced the farmers’ preference for a low-risk, non-structural measure—adjusting sowing time—over options dependent on uncertain water flows or requiring regulatory permissions.

Insights from the qualitative data were used to ensure the AquaCrop model settings accurately reflected local conditions. The qualitative analysis helped identify and prioritise the adaptation strategies that were most meaningful and feasible from the community’s perspective. This ensured that the subsequent quantitative modelling focused on evaluating solutions that had a high potential for local adoption.

Analysis of the participatory mind mapping exercises revealed several consistent patterns that directly informed the formulation of the co-designed adaptation strategy (Figure 6). Across all groups, farmers emphasised that two weeks of early monsoon rainfall were sufficient for transplanting because the clay-loam soils in Sadivayal retain received rainfall for longer. This insight shaped the modelling of initial soil–water conditions and helped identify the 38th SMW as the most viable window for early sowing. The maps also highlighted the soil’s low infiltration, which aligned with farmers’ observations of standing water following heavy rainfall, informing the adjustment of AquaCrop parameters. Synthesised through coding and thematic analysis, these restrictions, everyday realities and experiential understandings converged into four core themes: optimal early season rainfall, soil-moisture retention, low infiltration and adjusted supplemental irrigation needs, which collectively guided the development of the final co-designed solution: adjusting the sowing time.

Figure 6

Sketching exercises highlighted how water reaches the fields and how flow conditions change during the season. Farmers’ drawings illustrated inconsistent river–canal discharge, frequent channel drying and damage to irrigation pathways by wildlife (Figure 7). These visual insights clarified when and where water access becomes unreliable, helping identify periods of likely supplemental irrigation need and showing that declining flows often coincide with sensitive crop stages.

Figure 7

Storytelling exercises provided case-specific accounts of how farmers experienced water stress across different field locations, revealing micro-environmental factors that had not surfaced in earlier workshops. These narratives clarified that the phenological duration of long-duration local varieties, such as black rice, often exceeded locally assumed norms in these areas. This insight was incorporated into the modelling step by adjusting growth-stage durations and initial soil-moisture conditions for affected farm clusters. Storytelling also clarified the perceived benefits and risks associated with different sowing windows, informing the co-design decision to prioritise 38th SMW sowing. Farmers explained that low river flow during the Kharif season makes the 19th SMW sowing option impractical, even though the model indicated relatively high resilience for that week.

In addition, the narratives revealed management practices not captured through mapping or sketching, particularly the use of farmyard manure to improve soil structure and enhance moisture retention. Farmers described how regular organic manure application increased soil softness and prolonged wetness after rainfall, enabling better seedling establishment under early or uncertain monsoon conditions. These accounts demonstrated that several plots possessed higher functional water-holding capacity than initially assumed, which informed the modelling process by refining initial soil-water content settings and adjusting early season stress assumptions. The narratives further illustrated how plot-level capacities shaped farmers’ willingness to adopt earlier sowing, strengthening both the parameterisation of AquaCrop and the co-design rationale behind the 38th SMW as a context-appropriate adaptation measure.

In the active experimentation phase, the participatory modelling approach used AquaCrop to evaluate whether the community-preferred the early sowing date (38th SMW), identified during co-design, would genuinely perform well under local climatic and soil conditions. In this sense, the simulations functioned as a virtual field experiment, allowing farmers to test the feasibility of early sowing without the risks of real-world trial and error. The model simulations indicated that the 38th SMW, combined with life-saving supplemental irrigation adjusted to rainfall events and field-specific soil-moisture retention identified through mind mapping, sketching and storytelling, could significantly enhance rice yields. Through iterative workshops, farmers validated the model outputs by comparing simulated yields and water-stress indices with their field experiences, noting that these evaluations were comparative and not derived from a formal parameter sensitivity analysis.

The effectiveness of this solution was evaluated through the resilience indicator and WSI computed under different SSPs. Figure 8 shows the results of the resilience indicator under SSP245 and SSP585. Week 38 clearly exhibits the highest resilience in both scenarios, with SSP585 showing especially excellent performance (> 2.6) in comparison with SSP245 (2.3). Farmers’ storytelling during participatory modelling highlighted the fact that the low flow of river water during the Kharif season makes the 19th SMW an ineffective solution, although it expresses good resilience compared with other sowing weeks.

Figure 8

Although weeks 40 and 42 exhibit a considerable level of resilience, they are nevertheless less successful than week 38 which showed higher resilience even under more severe climate projections (SSP585). Figure 9 illustrates the WSI for two climate scenarios (SSP245 and SSP585) throughout the course of several sowing weeks (19–42). The result of WSI assessment across the proposed sowing weeks strengthened the effectiveness of the 38th SMW in mitigating climate variability by reducing water stress on rice yield. Reduced water stress is indicated by lower WSI readings, which is advantageous for crop performance. Interestingly, under SSP585, sowing week 38 continually shows the lowest median WSI, indicating that it is the best window for sowing to reduce water stress in future climates. Conversely, the 19th and 23rd SMWs exhibit the highest WSI values, indicating that early sowing exposes the crops to severe water deficits. Greater variability in water stress in the high-emission scenario (SSP585) is further highlighted by the plots’ range of values, particularly for sowing weeks 42 and 40.

Figure 9

In the last step, the assessment of the living labs approach demonstrated an increase in farmers’ knowledge and a positive shift in their attitudes towards the co-designed adaptation measures. The community was empowered to actively participate in the solutions’ implementation, and the participatory modelling technique increased the solutions’ acceptability and relevance. The survey results highlight the effectiveness of the participatory modelling approach in enhancing both farmers’ knowledge and interest in adopting new practices (Table 2). A majority of farmers reported a significant increase in their knowledge post-intervention. For instance, farmers 1’s and 2’s knowledge rose dramatically by 70 percentage points (from 10% to 80%, and from 20% to 90%, respectively). However, not all farmers experienced improvement, e.g. farmer 12’s knowledge remained at 10% both pre- and post-intervention. A critical analysis of the cases with little to no knowledge change reveals important barriers that must be addressed.

Similarly, the programme successfully boosted farmers’ interest in adopting interventions. Many farmers showed marked improvements, such as farmer 6, whose interest rose from 2/5 to 5/5, and farmer 1, whose interest increased from 1/5 to 4/5. Yet, there were exceptions, such as farmer 9, who, despite a knowledge gain from 50% to 90%, remained uninterested in adoption. Additionally, farmer 12 showed no change in their interest, remaining at 1/5. Overall, the data suggest that while the participatory approach was effective in most cases, a small subset of farmers did not benefit, which indicates a need for more tailored strategies to address individual barriers to knowledge and adoption.

Nine farmers adopted the early sowing adaptation measure, shifting their planting schedules to the 38th SMW. During the growing season, field schools were organised to provide hands-on training to farmers, equipping them with the knowledge and skills to implement the measures effectively. Peer-to-peer learning was encouraged, fostering a sense of collaboration and trust among community members. As the pioneer farmers shared their successes and challenges, other farmers became increasingly motivated to explore the benefits of these adaptations, amplifying the initiative’s impact. Finally, the assessment of the co-design process highlighted the transformative outcomes of this participatory approach. Surveys and feedback revealed a marked improvement in farmers’ knowledge and a shift in their attitudes toward climate adaptation measures.

The nine farmers’ implementation of early sowing practices served as a demonstration of the feasibility and effectiveness of the strategies, inspiring others in the community (Figure 10). Field schools further refined the adaptation measures by addressing local nuances, enabling farmers to optimise their practices. The iterative learning process empowered the community to take ownership of the solutions, fostering long-term sustainability and resilience. The success of this approach in Sadivayal underscores its potential applicability in other regions facing similar climate-related challenges, offering a replicable model for inclusive and effective climate adaptation in agriculture.

Figure 10

Trained farmers’ role as change agents facilitated peer-to-peer learning, wherein knowledge and experiences were shared informally among other farmers in the community. Farmer field schools were also organised to provide structured learning environments, where trained farmers demonstrated the implementation of the solutions and their outcomes. The farmer field school approach supports climate-resilient agricultural practices and enhances the institutional capabilities of farming communities (Campbell & Lester 2023). These schools fostered interactive discussions and practical demonstrations, significantly influencing the interest and willingness of other farmers to adopt the implemented practices. This multilayered, farmer-led implementation strategy ensured the widespread dissemination of sustainable practices and created a strong foundation for long-term adoption and scalability of the co-designed solutions within the broader farming community. Continuous feedback from the community played a pivotal role in refining the proposed solution, contributing to its ongoing improvement. Farmer-led implementation ensured alignment with farmers’ priorities, boosting acceptance and long-term sustainability.

4. DISCUSSION

The living lab process demonstrated how experiential and participatory learning cycles can structure and support climate adaptation by strengthening feedback loops between experience, reflection, modelling and behavioural change (Tschakert & Dietrich 2010). Similar to these studies, the living lab process here translated tacit knowledge into actionable strategies, showing that farmer-led reflection can refine technical recommendations derived from process-based models. Each stage of Kolb’s ELT cycle was clearly reflected in the intervention. The transition to participatory modelling represented abstract conceptualisation, where experiential insights from mind mapping, sketching and storytelling sessions were translated into formalised scenarios to evaluate the feasibility of adaptation options. Finally, the adoption of early sowing (38th SMW) by nine farmers exemplified active experimentation, where the co-designed solution was tested under real field conditions and generated new learning for the wider community.

These findings show that integrating qualitative insights was essential: storytelling, sketching and mind mapping corrected several AquaCrop assumptions that would otherwise have generated recommendations inconsistent with real farming conditions. For example, despite AquaCrop indicating high resilience for the 19th SMW, farmers rejected it because the river and canal flows are extremely low in the early Kharif season. This situated insight aligns with CPDT, which argues that design must be grounded in socio-ecological realities rather than abstract optimisation (Brandau & Alirezabeigi 2023). Similar mismatches between model outputs and lived realities have been noted in other participatory modelling studies, where contextual knowledge was required to avoid technically optimal but practically unworkable solutions (Hedelin et al. 2017). Similarly, variations in field-level microclimates, such as delayed flowering in shaded border plots, surfaced only through storytelling and led to adjusted phenological parameters in the modelling workflow. These examples illustrate CPDT’s insistence that local experiential knowledge must override model outputs when the two diverge.

Another key contribution relates to the persistent critique that participatory methods often generate dialogue without producing tangible outcomes (Wacnik et al. 2025). The living lab process here counteracts this critique by producing a concrete, widely adopted intervention. The co-designed early sowing strategy (38th SMW) was not only discussed but also adopted by 60% of farmers, and model outputs confirmed its consistently higher resilience under both SSP245 and SSP585. This shows that integrating qualitative insights with iterative modelling can move participation from discussion to real implementation, increasing the chances of sustained adoption and improving climate resilience. Qualitative insights also enriched model parameterisation. Farmers’ observations of slow soil drying, prolonged moisture retention after early rainfall and field-specific delays in flowering for long-duration varieties informed adjustments to initial soil-water content, supplemental irrigation requirements and phenological durations. Similar examples of qualitative-to-quantitative refinement have been documented in participatory agro-hydrological modelling (Basco-Carrera et al. 2017), confirming that community-derived knowledge improves representational accuracy when formal datasets are limited.

At the same time, the process revealed important limitations related to participation dynamics, an issue central to CPDT. While most farmers exhibited substantial gains in knowledge and interest, a subset (farmers 3, 9, 11, 12) showed minimal or no improvement. Follow-up discussions showed that non-participation stemmed from practical constraints and personal hesitations rather than disagreement with the intervention itself. For instance, farmer 11’s off-farm employment in ecotourism reduced his availability for field schools, limiting both exposure and comprehension. Farmer 9 perceived early sowing as a financial risk despite understanding its benefits, reflecting risk aversion that is common among smallholders with unstable incomes. Others, such as farmers 3 and 12, relied strongly on generational practices, a form of tacit knowledge that shaped their risk perception and lowered their willingness to experiment. These cases underline that adoption is influenced not only by agronomic performance but also by socio-economic constraints, identity-based knowledge systems and competing livelihood commitments.

Consistent with Smith & Iversen (2018), this study finds that the likelihood of adopting sustainable behaviours and knowledge gains is substantially correlated with high engagement intensity. Conversely, sporadic attendance, high opportunity costs and reliance on traditional knowledge systems can hinder learning and participation. To achieve equitable participation, this study recognises that future living labs must incorporate more personalised engagement pathways, such as flexible scheduling, targeted support for risk-averse households and diversified communication tools. Although AquaCrop offered useful comparative insights into alternative sowing windows, its outputs inevitably reflect the assumptions embedded in climate inputs, soil parameters and management settings. In line with standard practice in participatory modelling, the simulations were interpreted comparatively rather than as precise forecasts (Voinov & Bousquet 2010). This approach appropriately aligns the model’s role with the co-design process, while leaving room for future work to incorporate more formal sensitivity or uncertainty analyses if required.

The study advances debates on co-design and experiential learning by showing how participatory modelling can operate simultaneously as an analytical tool and a pedagogical device within living labs. Using AquaCrop as a boundary object enabled shared interpretation, iterative refinement and transparent adjustment of model parameters, addressing longstanding critiques of expert-driven modelling and demonstrating how qualitative insights can be systematically integrated with comparative simulation outputs. For practitioners, the findings provide a practical framework for embedding participatory modelling into routine extension systems through coordinated facilitation across universities, extension services and local organisations, supported by iterative field trials, feedback loops and regular scenario updates. For farming communities, the co-modelling process enhanced adaptive capacity by allowing participants to test, validate and refine adaptation options under their own conditions, strengthening trust, ownership and agency. Sustaining these gains will require continued facilitation support and periodic model updates that enable communities to adjust practices as climatic conditions evolve.

Overall, the study demonstrates that a modelling-enabled living lab, grounded in experiential learning and critical participatory design principles, can strengthen civic resilience in smallholder systems by enhancing communities’ collective capacity to interpret climatic risks, co-design feasible responses and implement context-specific adaptation measures. The combination of qualitative insight, model-supported exploration and farmer-led experimentation produced a context-specific intervention with strong local legitimacy, high adoption potential and measurable agronomic benefits. Together, these outcomes illustrate a scalable pathway for inclusive and effective climate adaptation in comparable socio-ecological settings.

5. CONCLUSIONS

This study demonstrates how a modelling-enabled living lab can support the development of climate adaptation measures that are both scientifically grounded and locally feasible for resilient agriculture in Sadivayal. Grounded in Kolb’s experiential learning theory (ELT), a participatory modelling framework translated community-driven insights into AquaCrop through an iterative and interactive process of designing and testing sustainable practices. This ensured that simulations reflected lived farming realities such as soil moisture dynamics, river–canal flow variability and field-level microclimates, which in turn confirmed the 38th Standard Meteorological Week (SMW) as the most viable and community-validated sowing strategy.

Farmer-led implementation of the co-designed solution generated insights that improved the relevance of modelling outputs and strengthened validity of the findings. The experiential knowledge gained during implementation enabled farmers to refine the solution and make better informed decisions. Active engagement in the modelling workshops enhanced transparency in how sowing options were evaluated, strengthening trust in both the process and its recommendations. The observed performance of early sown fields supported peer learning and collective confidence-building, reinforcing social cohesion within the community. The participatory modelling framework also increased farmer knowledge and adoption interest, as demonstrated through the field-level implementation of early sowing and supplemental irrigation practices. These on-farm trials validated the effectiveness of the proposed strategies and fostered a sense of ownership and sustained trust among participants. While climate data uncertainty constrained the simulations to comparative use, the integrated approach proved relevant and useful for adaptation planning.

Importantly, the findings also carry broader implications for planning and agricultural policy. The integration of iterative participatory modelling within extension services offers a practical pathway for incorporating farmer-derived experiential knowledge into formal adaptation planning. Expanding agricultural departments’ fundamental modelling capabilities and establishing transparent data-sharing protocols would encourage broader adoption. The use of comparative water stress and resilience indicators enhances transparency and legitimacy in decision-making, demonstrating how participatory modelling can provide actionable and scalable support for adaptation planning in semi-arid systems.

AUTHOR CONTRIBUTIONS

MKB: writing – original draft, review & editing; visualization; methodology; investigation; formal analysis; data curation; conceptualization. SB-S: writing – review & editing; supervision; methodology.

COMPETING INTERESTS

The authors have no competing interests to declare.

DATA ACCESSIBILITY

The data supporting this study are available from the corresponding author upon reasonable request.

ETHICAL APPROVAL

The study received ethical approval from Amrita Vishwa Vidyapeetham. All participants provided informed consent before participation, and all responses were anonymised.