Twinning living heritage: human-centered digital twins coupling CADA and HUL for arid cities

1 Introduction

Public spaces in hot-arid cities are where three otherwise separate policy problems converge. Thermally, remote-sensing analyses of Gulf cities have reported rising land-surface temperatures and worsening outdoor discomfort over the past 2 decades, with land-cover change implicated in the loss of comfortable outdoor conditions (Patel et al., 2025; Yasmin et al., 2026). Morphologically, compact, shaded, courtyard-based fabrics that once moderated radiant exposure have given way to dispersed, paved, and car-oriented layouts, leaving residual spaces between road corridors and historic cores (Trancik, 1986; Othman and Alshboul, 2020; Abedrabboh et al., 2025). Heritage districts are increasingly managed as monuments or façades, while the everyday practices and gatherings that give them meaning remain exposed to redevelopment and climatic stress (Taylor, 2016; Moscatelli, 2023). While governed individually, these issues can be tackled together with one intervention, that is, a canopy, a planted threshold, or a repaved route—that simultaneously influences all three through the interplay of shade, greenery, and the feel of the ground beneath.

Each framework applied to this setting addresses only a part of it. Climate-responsive design has robust thermal comfort metrics but treats heritage as a morphological constraint rather than as significance held by identifiable communities (Ali-Toudert and Mayer, 2006; Abedrabboh et al., 2025). Values-based heritage management—chiefly UNESCO’s (2011) Historic Urban Landscape (HUL) Recommendation—articulates why places matter and to whom, but implementation has concentrated on value identification and is rarely coupled to scenario testing or climatic evaluation (Ginzarly et al., 2019a; Rey-Pérez and Pereira Roders, 2020). Urban digital twins (UDTs) integrate data and simulations for monitoring (Batty, 2018; Ferré-Bigorra et al., 2022); however, a growing body of critical literature finds them technocratic and infrastructure-oriented, with participation confined to visualization and consultation (Nochta et al., 2021; Argota Sánchez-Vaquerizo, 2024; Azadi et al., 2025; Knopf et al., 2025) and insensitive to the sociopolitical contexts in which they operate (Charitonidou, 2022; Milan and Treré, 2019).

Therefore, the gap is specific and not blanket. Heritage is not absent from digital environments: HUL values have been mapped from social media data (Ginzarly et al., 2019b), participatory GIS records place values (Brown and Kyttä, 2014), and a heritage digital twin ontology has been proposed (Niccolucci et al., 2022). However, reviews find the field dominated by geometric documentation and condition monitoring, with intangible and social values weakly represented and loosely coupled to the technical core (Tootoonchi and Hess, 2026), and urban-scale twins seldom represent heritage beyond static 3D forms (Jeddoub et al., 2023; Al-Thani et al., 2026). To our knowledge, no existing framework combines three capabilities: encoding plural, time-varying, contested significance as data; co-registering it with climatic performance at the scale of individual public spaces; and using the resulting trade-offs to discipline option generation and negotiation. Without them, a twin that correctly predicts a canopy will lower radiant temperature cannot register that it severs a valued sightline and will present a culturally damaging option as the thermal optimum—treating heat adaptation as a technical problem separate from heritage governance and equitable access (Kitchin, 2014; Melis et al., 2023).

This study develops a theoretical framework to close that gap, coupling the Convergent Appreciation/Divergent Activation (CADA) design model (Azhar et al., 2026) with HUL through a human-centered digital twin as an operational intermediary. Four questions organize the argument: (RQ1) what limits the human-centeredness of current UDTs regarding heritage values and distributive equity; (RQ2) how a named design epistemology couples with a values-based heritage approach into one implementable framework; (RQ3) what such a twin can represent, simulate, and negotiate that a conventional twin cannot; and (RQ4) under what conditions the framework can be tested and transferred beyond hot-arid heritage public spaces. The contribution is a novel coupling of existing components, not new components: a CADA–HUL crosswalk, a value-to-layer mechanism representing significance as plural, provenance-tagged, versioned data, a five-layer twin architecture, and six disconfirmable propositions (Table 1). No empirical validation or software implementation is claimed; the Doha example (Section 3.5) is an illustrative parameterized walkthrough, and the empirical program is specified in Sections 3.6 and 4.3.

TABLE 1. Bounded statement of contributions and their evidence status.

Two scope clarifications follow, with the terms fixed in Table 2. First, the framework analyzes the hot-arid heritage public space—the precinct-scale realm combining thermal exposure and heritage values—not the city. Second, “Gulf” names the empirical setting of the worked example and is not equated with “arid”: not every Gulf city is climatically arid, so the argument is bounded by climate rather than region and extends to rapidly urbanizing hot-arid cities generally. Heritage values are the reasons a place matters—historical, aesthetic, social, spiritual, or use-based significance attributed by communities and experts (Mason, 2002; Australia ICOMOS, 2013); attributes are the tangible and intangible features that carry those values (Veldpaus, 2015); and the values-based approach establishes significance before deriving policy and design from it (Avrami et al., 2000). Tangible fabric is distinguished from intangible practice only when that distinction is operationally consequential.

TABLE 2. Operational definitions of heritage and framework terms used in this study.

The concept of residual public space, often referred to as “lost space” by Trancik (1986) and identified in planning as space left over after planning (SLOAP), is a recurring theme in this study. These spaces, frequently found in areas where road expansion and rapid development disrupt historic districts, are introduced here because of their dual characteristics: they are both thermally challenging and readily available for climate-responsive and heritage-sensitive reactivation in hot-arid regions. An example of such a precinct is provided in Section 3.5.

Section 2 synthesizes the four constituent literatures, examines the principal critiques of HUL, positions the framework against rival integrative frameworks, and identifies the integration gap. Section 3 develops the framework (premises, crosswalk, operationalization mechanism, five-layer architecture, a worked example, and six propositions). Section 4 discusses the theoretical and governance implications, empirical testing pathways, transferability, counterarguments, and limitations. Section 5 concludes.

1.1 Methodology: a desk-based theoretical synthesis

This desk-based contribution uses structured theoretical synthesis and abductive inference rather than empirical data collection; it is a conceptual synthesis, not a systematic review, and does not claim to be PRISMA-level complete. The full protocol (databases, search strings, eligibility criteria, screening, and the rubric behind Table 3’s ratings) is reported in Supplementary Tables S1–S3 to keep the derivation contestable. Four bodies of literature (Supplementary Table S2) were searched in Scopus, Web of Science, and Google Scholar using combinations of terms including “urban digital twin,” “heritage digital twin,” “Historic Urban Landscape,” “heritage values,” and “spatial equity,” emphasizing work from 2019 onward while retaining foundational studies. The corpus was updated in 2026 for recent Gulf, HUL, and Heritage Digital Twin scholarship. This literature was appraised against three axes—human-centeredness, heritage-value integration (Table 2), and climate-equity operationalization—to identify the integration gap (Table 3), from which the framework was derived abductively as the most economical architecture satisfying all three simultaneously and rendered testable as propositions with comparison conditions, disconfirmation criteria, and empirical methods (Sections 3.6, 4.3). Because several propositions concern deliberative and institutional processes, they are disconfirmed under stated conditions rather than strictly falsified. The study progresses from the examination of issues related to arid cities (Sections 1–2), through the theoretical underpinnings (Sections 2.2–2.4) and the development of a framework (Section 3), culminating in a testable agenda (Sections 3.6, 4.3). Although it delineates the framework and validation process, it does not address software implementation. Section 4.6 evaluates the feasibility of this approach.

TABLE 3. Appraisal of the four constituent literatures against the three integration axes, with evidence supporting each rating.

Ratings (weak, partial, moderate, and strong) follow the rubric in Supplementary Table S3 and describe each literature’s dominant tendency as qualitative judgments over representative, not exhaustive, evidence. Heritage value integration is scored in the four sub-dimensions in Table 2. The bottom row shows the target capabilities of the proposed framework, not literature-based scores.

2 Conceptual background and the integration gap

This section synthesizes the four bodies of literature integrated into this framework, examines the principal critiques of HUL, and positions the proposed coupling against rival integrative frameworks.

2.1 Urban digital twins and AI in city governance

The digital twin (DT) concept, initially a dynamic virtual replica of a physical asset within product lifecycle engineering, has been extended to urban environments (Grieves and Vickers, 2017; Batty, 2018) as a near-real-time digital representation of the physical urban environment built from multisource data and advanced technologies for the simulation, monitoring, and enhancement of urban systems (Lei et al., 2023; Ferré-Bigorra et al., 2022). Applications span energy, transport, water and wastewater, environmental factors such as noise, floods, and climatology, and AI, the Internet of Things, and big-data analytics (Bibri and Krogstie, 2017; Shahat et al., 2021). In practice, the prevailing paradigm emphasizes technical efficiency, using twins for control and optimization to improve system performance with minimal human effort.

Critical assessments (Argota Sánchez-Vaquerizo, 2024; Elsehrawy et al., 2024) have identified lasting conceptual ambiguity in the “urban twin,” spanning passive 3D models to active cyber-physical systems, with most implementations falling short of the two-way, automated linkage that a strict definition implies (Argota Sánchez-Vaquerizo, 2024; Jeddoub et al., 2023). This immaturity is not treated here as a flaw requiring greater technical specificity but as an opening for purpose- and value-oriented intervention. Technocratic, as used in this study, denotes a specific, criticizable arrangement rather than a term of disapproval: decisions about public space are framed as technical problems, delegated to experts and quantified criteria, and settled by optimizing measurable objectives, so choices about whose interests prevail appear as findings rather than contestable political judgments (Kitchin, 2014; Charitonidou, 2022; Knopf et al., 2025). Three features follow, each referred to later: the questions asked are restricted to what available data can answer; the objective function is set by whoever commissions the system; and outputs carry the authority of computation.

Charitonidou (2022) argues that “digital universalism,” which reduces culturally varied settings to a single, ostensibly neutral framework, poses a real threat, echoing earlier critiques of datafication and the smart city (Kitchin, 2014; Milan and Treré, 2019). Knopf et al. (2025) examine UDTs’ political dimensions of UDTs (privacy, surveillance, representativeness, and entrenched power), and Argota Sánchez-Vaquerizo (2024) argues that UDTs can deepen social division by producing disparate urban experiences. Systematic and sociotechnical analyses consistently find human-centricity emergent but superficial, equity rarely implemented, and heritage regarded mainly in fixed spatial form (Jeddoub et al., 2023; Nochta et al., 2021; Azadi et al., 2025). Where participatory functions exist, as in citizen-feedback twins, they typically collect reactions to predefined scenarios rather than letting citizens set the values by which the scenarios are judged (White et al., 2021). Dawkins and Kitchin (2025) propose repositioning twins for “participatory steering,” implying a need for human-centered reform, but without integrating heritage methodologies or design frameworks.

AI intensifies both the benefits and risks of digital twins; predictive and generative models can monitor systems, synthesize configurations, predict microclimatic outcomes, and rank interventions rapidly. This is precisely where the human-centeredness deficit becomes perilous, rather than merely insufficient. An AI that cannot register sightline preservation, informal economic impacts, or disparities in data quality operates with undue confidence, and algorithmic governance scholarship warns that such systems reflect the values of whoever sets their objectives, often at the expense of distributive and cultural considerations that resist quantification (Kitchin, 2014; Knopf et al., 2025). Therefore, the question is not whether AI should be integrated into twins, but which design epistemology governs their generative capacity. Section 2.2 introduces CADA as one such epistemology.

2.2 The convergent appreciation/divergent activation (CADA) model

CADA is a human-centered design model with two interconnected processes: convergent appreciation, methodically identifying and harmonizing the collective value of a place for users and communities, and divergent activation, which generates context-aware design solutions built on that shared understanding. Value is not treated as a subject for documentation but as a deliberate contribution that regulates subsequent design divergence (Azhar et al., 2026). CADA’s convergent moment resonates with and departs from the literature on place attachment and sense of place (Low and Altman, 1992) and Lynch’s (1960) focus on legibility and imageability. Whereas Lynch’s work addresses design descriptively, CADA’s convergent moment is procedural and generative, aligning with the place-value emphasis underpinning methodologies for heritage significance, such as the comprehension-to-policy sequence of the Burra Charter (Australia ICOMOS, 2013).

Lynch (1960) and the Burra Charter are cited here not as CADA’s predecessors but as the established traditions against which its distinctiveness is defined. Lynch’s account of legibility and imageability is descriptive, and the Burra Charter codifies an understand-then-act sequence within values-based conservation. Neither is generative in CADA’s sense, and CADA’s divergent moment extends beyond conservation management to design generation. Its contribution is to render the convergent, value-identifying moment procedural and constrain the divergent generation that follows. The framework requires only that some processes move from a shared, deliberated statement of value to a bounded set of generated options. Value-sensitive design (Friedman et al., 2006) and participatory co-design are the most direct alternatives, and comparative testing against them is a future research requirement rather than a claim that CADA is uniquely necessary. Because CADA was developed by the first author, its inclusion is justified functionally rather than by precedence, and Proposition P2 was designed to test whether it adds value beyond divergent generation alone. Convergence here does not imply stakeholder consensus but rather the structured identification of shared, divergent, and contested place values before design alternatives are generated.

Recent participatory heritage research supports this appreciation-before-activation sequence. Liu et al. (2023) staged the community redesign of the vacant Politiebureau Groningen Centrum into distinct convergent and divergent phases, eliciting residents’ views through cognitive mapping and interviews before generating scenarios, and found substantial shared value. Djabarouti (2024) found conservation architects involving participants from the earliest stages, and Cunha Ferreira et al. (2024) showed that combining expert and non-expert input enriches significance assessment, as in Siza’s Ocean Swimming Pool. Park et al. (2025) found that value-centered framing improved comprehension of collective sentiment and willingness to negotiate without added cognitive load. This appreciation-then-activation logic makes CADA a suitable basis for a human-centered method, though one that presupposes that place value can be articulated and acted upon; where values remain wholly implicit or power imbalances prevent genuine appreciation, CADA cannot represent them adequately.

2.3 The historic urban landscape (HUL) approach

The 2011 UNESCO Recommendation on the Historic Urban Landscape treats urban heritage not as isolated monuments but as a layered landscape of cultural and natural values, integrating conservation into urban development through six steps: surveying the city’s natural, cultural, and human resources; reaching a consensus on values and attributes to protect; assessing their vulnerability to socio-economic stress and climate change; integrating them into city development; prioritizing conservation and development actions; and establishing partnerships and local management frameworks (UNESCO, 2011; Bandarin and van Oers, 2012). The approach rests on a values-based reading in which significance is distributed across historical, social, aesthetic, and spiritual dimensions rather than being fixed to monumental fabric (Ginzarly et al., 2019a; Rey-Pérez et al., 2020), operationalized by distinguishing values (why a place matters) from attributes (what conveys that significance) and recording which stakeholders hold each value (Veldpaus, 2015; Pereira Roders and Bandarin, 2019). Two properties matter here: significance is temporal, accruing and being reinterpreted over time rather than settled at designation, and it is frequently contested, held differently by different communities, a condition the contested-heritage literature treats as constitutive rather than incidental (Liu et al., 2021; Wu, 2020). A values-based twin must therefore represent significance as time-aware and plural.

HUL has attracted several distinct critiques, and scholars do not agree on a single flaw; therefore, this framework must state which it engages. The implementation gap is the most documented: HUL literature and practice concentrate on mapping and value identification, while integration into planning instruments, institutional coordination, and change management lag far behind (Ginzarly et al., 2019a; Rey-Pérez and Pereira Roders, 2020; Karagkouni et al., 2026), as in Latin American and Caribbean cities (Astudillo-Cordero et al., 2026), in Al Fahidi and the Heart of Sharjah, which stall at planning integration and partnership (Marthya et al., 2026), and in Manama’s leftover heritage buildings (Hanif et al., 2026). Second, there is an absence of indicators, evaluation tools, and financing instruments for monitoring outcomes (Angrisano et al., 2016; Rey-Pérez and Pereira Roders, 2020). Third, conceptually, landscape framing is powerful but open to divergent institutional interpretations (Taylor, 2016). Fourth, institutionally: HUL is a Recommendation, not a Convention, so no state is legally bound to adopt it, and even UNESCO’s own World Heritage Operational Guidelines, which govern how World Heritage Sites are managed, have been slow to incorporate it (Gong et al., 2025). Fifth, value-based approaches can reproduce an authorized discourse in which expert values prevail over community meanings, with participation remaining tokenistic (Smith, 2006; Labadi and Logan, 2016; Jones, 2017). Sixth, climate change entered HUL practice late, and its coupling with climate evaluation remains underdeveloped (Sesana et al., 2019).

These critiques call for different remedies. The framework directly addresses implementation, indicators, and climate critiques, moving from a statement of significance to a spatially referenced record scored against design alternatives (Section 3.3) and co-registering significance with climatic performance. It partially answers the authorized-discourse and participation critiques through plural, provenance-tagged value sets rather than a single expert weighting, but not HUL’s legal status, inter-agency coordination, or financing, which require institutional remedies (Sections 4.2, 4.7). HUL-oriented digital tools already exist, notably values mapped from social media data (Ginzarly et al., 2019b), but remain confined to value identification, uncoupled from scenario generation or simulation; the twin supplies that missing link rather than being HUL’s first computational tool.

HUL’s features are amenable to computational operationalization, and the framework adopts them as its foundational element. The layered, staged architecture developed below is the authors’ computational interpretation of HUL for digital twin implementation, not a formal UNESCO data model: UNESCO (2011) recommends a values-based, layered approach and a six-step process, but the specific layer ontology (natural, built, infrastructural, social, symbolic, congruent with digital twin data models) and its workflow mapping are the authors’ own. The HUL process (mapping, consensus on values, vulnerability assessment, integration, prioritization, partnership, and management) corresponds to the sense, represent, reason, negotiate, and intervene phases intrinsic to digital twinning; most concerns about this correspondence are rooted in execution rather than in theoretical understanding, lying in the translation of values, their adaptability to change, and assimilation into planning frameworks. Because the literature offers limited means to bridge mapped values with quantitative climate and morphological decisions, such values are often neglected during planning. Digital twinning can provide a computational environment in which heritage values sit alongside other planning criteria, supporting HUL’s integrative objective of HUL rather than burdening it.

2.4 Climate-responsive design and spatial equity in arid public spaces

In exposed hot-arid public spaces, radiant temperature usually dominates daytime thermal stress (its weight varies with season, time, humidity, wind, and activity) and is chiefly governed by morphology and materials, including street-canyon density, courtyard form, shade, and tree placement. Simulation studies have shown that deeper street canyons and appropriate orientation substantially reduce daytime pedestrian heat stress, principally by lowering the mean radiant temperature (Ali-Toudert and Mayer, 2006), and that shaded, compact, courtyard-based fabrics can outperform dispersed layouts under specified summer conditions, subject to orientation, aspect ratio, materials, and wind exposure (Othman and Alshboul, 2020; Abedrabboh et al., 2025). Tall trees and blue-green infrastructure achieve substantial reductions, but aggregate “green coverage” can obscure water scarcity and uneven vegetation distribution (Yasmin et al., 2026).

Arid conditions are theoretically load-bearing, not merely contextual: shade is a contested, finite resource, greening carries water scarcity trade-offs absent in temperate cities, and public spaces are directly exposed to radiation loads determining whether they remain usable. Equity therefore enters as a distributive question, who can reach cool, usable space and when, which heat-vulnerability studies increasingly frame as a justice issue, showing thermal burden and access to cooling unevenly distributed and mitigation prioritization benefiting from an explicit equity lens (Melis et al., 2023; Hsu et al., 2021; Hoffman et al., 2020), though most evidence is North American and European, with comparable Gulf analyses still scarce. The arid city’s residual spaces, Trancik’s “lost space,” are simultaneously its most thermally hostile and most available for reactivation (Trancik, 1986). Climate equity spans thermal benefits and burdens, accessibility across mobility and vulnerability profiles, temporal access, and the social, ecological, and maintenance burdens created by interventions. A design that improves mean radiant temperature while restricting informal commerce, displacing practices, consuming scarce water, or excluding less powerful users is not equitable merely because its thermal metric improves.

Heritage and climate are deeply linked in arid environments, where heritage typologies often outperform modern ones thermally. Compact courtyard residences, narrow shaded pathways (sikkak), wind catchers, and mashrabiya screens are among the most effective passive cooling strategies in the region. Alnaim and Alnaim (2026b) show climatic resilience embedded in urban morphology at the district scale in AlUla Old Town, providing evidence that heritage typologies can function as climate instruments. Therefore, a twin must carry morphology as significance rather than inert geometry; such morphology can outperform modern dispersed structures under specified summer radiation where site geometry, materials, and maintenance allow it (Abedrabboh et al., 2025; Othman and Alshboul, 2020). Climate-responsive interventions and heritage are therefore often variations on one objective, a relationship that a disconnected energy-versus-heritage approach cannot recognize. Conversely, poorly managed heat mitigation can degrade heritage, as when a reflective canopy over a historic threshing floor diminishes its integrity or water-intensive landscaping displaces traditional majlis seating (Abedrabboh et al., 2025; Sesana et al., 2019). Viewed through a values-based lens (Ginzarly et al., 2019a), this erosion concerns mainly intangible use values and authenticity (circulation routes, evening gatherings, and informal commerce) rather than the material fabric. Thus, the arid city is the framework’s primary, though not exclusive, empirical setting.

2.5 Synthesis: the integration gap made visible

Table 3 appraises the four studies against the three integration axes using the rubric in Supplementary Table S3; each rating is accompanied by the evidence on which it rests rather than being asserted. Heritage-value integration is scored on the four sub-dimensions defined in Table 2: breadth of value types, intangible and social values, plurality of value holders, and temporality. A consistent pattern emerges: each study is strong on at most one axis. Urban digital twin research excels at data integration and simulation but is weak on heritage-value integration, since heritage enters mainly as geometry or asset records; CADA and value-sensitive design are strong on human-centeredness but lack a climatic or computational layer; HUL scholarship is strong on heritage values by definition but weak on operational and climatic tooling; and climate-responsive research is strong on thermal and distributive metrics but treats heritage as morphology rather than attributed significance. These ratings characterize each literature’s dominant tendency, not every study within it. The digital twin can connect all three axes (Figure 1) but has rarely been theorized as doing so, which is the gap this framework addresses.

2.6 Why this coupling? positioning against rival integrative frameworks

Selecting CADA and HUL over other integrative approaches requires justification, and each alternative contributes something worth retaining. The capability approach evaluates places by the real freedoms they afford residents (Sen, 1999; Nussbaum, 2011; Frediani, 2010), providing a strong normative basis for the distributive axis but no account of heritage significance and no computational form. Social–ecological systems frameworks integrate ecological and social dynamics across scales (Ostrom, 2009; Folke, 2006) but treat heritage mainly as a resource and omit design generation. Participatory GIS records spatially explicit place values with considerable reach (Brown and Kyttä, 2014) but is descriptive, carrying neither a staged design epistemology nor a dynamic model of changing significance. Value-sensitive design embeds stakeholder values in technology (Friedman et al., 2006), but is not place- or heritage-specific. Human-centered UDT proposals bring citizens into twins (White et al., 2021; Dawkins and Kitchin, 2025) without drawing on heritage-management practice, so heritage remains a geometry or incidental context.

The case for this pairing is functional and not exclusive. CADA offers a procedural, appreciation-first model for generating divergent options, and HUL supplies an internationally recognized, values-based model of the urban heritage continuum; together, their application becomes computable, continuous, and spatially explicit. The contribution is this specific combination—a formalized design epistemology coupled with values-based heritage management within a dynamic representation of hot-arid public space—rather than the general concept of human-centered digital twins, a claim stated at this resolution to remain both modest and refutable.

Two recent Gulf frameworks address overlapping problems and are the most relevant comparators (Table 4). The Heritage–Modernization–Sustainability Nexus (HMSN; Alnaim and Alnaim, 2026a) is derived from a qualitative comparison of Saudi heritage policy, international benchmarks, and two contrasting cases—the monumental Al-Turaif district in Diriyah and the living landscape of Al-Ahsa Oasis—across governance, authenticity, participation, environmental integration, and economic activation, framing heritage as both a product and catalyst of sustainable modernization. It finds governance shifting toward a hybrid of centralized and distributed arrangements, while community participation, intangible-practice transmission, and inter-agency coordination lag behind physical conservation—supporting two premises here: that intangible use-values are the most vulnerable component of significance and that participation requires structured facilitation rather than assumption. HMSN remains interpretive; it identifies which dimensions matter but offers no representation in which competing claims can be recorded, compared across options, or revisited as conditions change. Thus, the present framework can be read as an operational layer for HMSN’s dimensions of HMSNs rather than as a rival.

TABLE 4. Comparison of the CADA × HUL twin with recent Gulf integrative heritage–climate frameworks.

The Comprehensive Waterfront Development Index (CWDI; Hegazy and Imam, 2026) takes the opposite, quantitative route: it operationalizes six dimensions—accessibility, public space and amenities, economic vitality, environmental sustainability, cultural and social integration, and resilience and climate adaptation—through global and context-specific indicators weighted by the Analytic Hierarchy Process, calibrated with residents, business operators, and planning professionals, and applied to three Jeddah waterfront zones via a 312-respondent survey. Its findings—new zones scoring well on amenities and access while historic Al-Balad performs poorly on resilience and environmental measures—are an empirical instance of the heritage–climate–equity coupling theorized here. The CWDI has two strengths that this framework lacks: empirical application and stakeholder-calibrated weights offering a tested precedent for the negotiation layer (Layer 4). Its limitation is aggregation: a single weight vector and composite score resolve stakeholder disagreement before reporting, and resilience is rated rather than simulated. The CADA × HUL twin instead retains multiple provenance-tagged value sets whose conflicts stay visible, co-registers them with microclimatic layers at the scale of individual public spaces, and uses them to constrain option generation, suggesting a practical sequence in which a district-scale index, such as the CWDI, identifies underperforming zones for the twin to negotiate. In a third study, Alnaim and Alnaim (2026b) argue that traditional AlUla morphology embodies transferable climate knowledge, supporting the premise that heritage typologies are climate instruments (Section 2.4). However, similar to HMSN, it stops short of weighing the performance against other values.

2.7 Positioning against existing human-centered and participatory digital twins

Because a claim of novelty depends on what already exists, this subsection positions the framework against current human-centered and participatory UDT practice, so the proposed twin reads as the next computational step rather than the first attempt. Three strands are relevant to this study. Citizen-centric and participatory UDTs embed engagement through workshops, feedback, crowdsourcing, and immersive visualization; however, reviews find participation concentrated at the visualization and consultation stages rather than in structuring values (White et al., 2021; Argota Sánchez-Vaquerizo, 2024; Azadi et al., 2025), with few documented applications in the corpus reviewed here (Supplementary Table S2) concerning hot-arid cities. Co-creation and serious game platforms elicit preferences well but treat heritage as a static backdrop, with no layered account of significance. GIS, 3D-model, and heritage decision-support systems, including HUL-aligned mapping, make significance spatially explicit—from social-media-mapped HUL values to formal heritage digital-twin ontologies—but leave it descriptive, uncoupled from generative design or climate-equity simulation (Ginzarly et al., 2019b; Rey-Pérez and Pereira Roders, 2020; Niccolucci et al., 2022).

Relative to these strands, the framework adds a design epistemology (CADA) disciplining, not merely visualizing, the move from appreciation to option generation; an internationally recognized, layered heritage framework (HUL) in place of heritage as a backdrop; a focus on hot-arid heritage public space, where participatory twins are least developed; and a significance layer retaining disagreement as data rather than aggregating it (Table 4) — combining citizen-twin participation, serious-game scenario generation, and GIS-based spatial explicitness in one value-aware instrument.

3 The CADA × HUL human-centered digital twin (DT)

The framework is constructed in a layered manner, commencing with premises, progressing to crosswalks, then to the operationalization mechanism, followed by the architecture, a detailed scenario, and concluding with empirically testable propositions with pre-specified disconfirmation criteria.

3.1 Foundational premises and the meaning of a living twin

Three premises underpin the model. First, a human-centered twin is not a high-fidelity 3D replica but a dynamic, value-aware representation that jointly interrogates the physical, climatic, and cultural layers. Second, representation is a design decision—which layers, weights, and values are encoded—with distributional consequences (Kitchin, 2014). Third, climate response and value-based heritage management are not distinct optimization problems but rather a negotiation over the evolution of a thermally stressed public realm. The negotiation layer operates in a hybrid manner: bottom-up elicitation of heritage values feeds authorized, facilitated deliberation—neither top-down administration nor unstructured consultation—so affected communities set objectives while accountability for the final decision stays explicit.

Because the representation is not neutral, the framework claims no objectivity and does not attempt to do so. It states which values it carries, at what confidence, and on whose authority, rather than concealing them in a single optimized outcome—an inversion of typical twin epistemics, where outcomes are presented as found rather than constructed. A values-aware twin organizes dissenting viewpoints for negotiation rather than asserting solutions. The premises also bound the framework: decisions are not automated, the cultural layer shapes design rather than archiving artifacts, and thermal performance is one input to negotiation and not an overriding factor.

The twin is human-centered across three dimensions simultaneously, and this combination distinguishes it from a conventional UDT. At the data level, a temporal cultural significance component carries a weight equal to the physical and climatic dimensions. At the procedural level, engagement shifts from post-design consultation to value-setting and trade-off adjudication beforehand, so communities define objectives rather than reacting to them. At the epistemic level, values are contested, weighted, and provenance-tagged rather than a single optimized surface, allowing disagreement to persist through computation. A conventional UDT may realize one dimension—typically participation—while remaining technocratic in the other two; here, all three operate together, bound by the CADA × HUL coupling. This specifies design criteria, not evidence of human-centered results.

“Living” is a design property, not a slogan. Unlike conventional twins, which treat heritage as static records of scanned structures and designated boundaries (Zarri, 2022), the cultural layer carries temporal awareness and version control, capturing intangible practices, cyclical use, and ongoing reinterpretation: heritage as a continuously updated process rather than a BIM snapshot. “Living” here denotes temporal updating, changing practices, versioned significance, and ongoing.

3.2 Reconciling CADA and HUL: a crosswalk

Table 5 illustrates the correlation between CADA’s design movements of CADA, HUL’s operational layers of HUL, and the computational structures of the twin, which form the conceptual core of the framework. CADA’s convergent appreciation aligns with HUL’s value-mapping/statement-of-significance phase as a cultural-significance layer co-registered with the physical and climatic layers; divergent activation aligns with HUL’s management-of-change and intervention phases via AI-assisted scenario generation in the reasoning layer; human-centered deliberation aligns with HUL’s stakeholder engagement as a negotiation layer where values are debated and set rather than accepted; and place-responsive intervention aligns with HUL’s sustainable-development and monitoring phase as an intervention layer targeting public and residual spaces and feeding back to the twin.

TABLE 5. CADA–HUL–digital twin crosswalk (conceptual core of the framework with the problems addressed by each coupling).

3.3 The operationalization mechanism: from statement of significance to computable layer

Two attributes distinguish a crosswalk from a mere analogy. First, the correspondence is bidirectional and regulatory, not ornamental: CADA’s convergent appreciation not only resembles HUL’s value-mapping step but supplies a procedural sequence, moving from mapped values to design options, that HUL’s mapping step does not itself specify, while HUL’s layered ontology gives CADA’s previously place-general appreciation a heritage-structured content. Thus, the coupling addresses a limitation of each approach.

The rightmost column of Table 5 moves beyond analogy by naming the data structure and decision rule that would realize each construct. No running implementation is claimed; rather, each construct is shown to be translatable, in principle, into an implementable data schema (Supplementary Table S4a) and thereby exposed to counterexamples. Thus, a row that cannot be implemented is eliminated. Figure 2 illustrates one such transformation, from qualitative value to computability, without collapsing its nuances into false agreement.

• Elicitation. A HUL statement of significance for a place, such as a shaded courtyard used for evening gatherings, is decomposed into structured attributes: value type (social, historical, aesthetic, spiritual) and the attribute that carries it, bearers (which communities hold it), spatial extent (geometry and threshold zones), and temporal rhythm (hours, seasons, occasions of use).
• Encoding. Each attribute is a field in the geolocated significance layer. Significance is represented as one or more weighted value sets rather than a single scalar: where communities disagree, the layer carries multiple coexisting weightings with their provenance; thus, contestation is preserved as data rather than averaged away. Uncertainty is stored explicitly (confidence, source, and recency).
• Co-registration. The significance layer is spatially aligned with the climatic layer (simulated mean radiant temperature, shade hours) and the morphological layer, so a single location holds both its thermal performance and its cultural meaning.
• Scoring and feedback. Candidate interventions from divergent activation (a new canopy, a repaved threshold) were simulated for the thermal effect and scored against each value set. Thus, convergent appreciation constrains divergent activation in a computable loop: options that improve comfort while degrading high-confidence, widely held values are flagged, and trade-offs are passed to the negotiation layer rather than resolved automatically.

This addresses a documented implementation gap in HUL scholarship: existing literature offers limited operational guidance for moving from a statement of significance to a form that integrates tangible and intangible assets into planning (Angrisano et al., 2016; Karagkouni et al., 2026; Hanif et al., 2026). The framework’s principal contribution to heritage is this value-to-layer transformation, which requires empirical testing; because part of the significance is incommensurable, the mechanism represents disagreement rather than resolving it (Section 4.7).

The mechanism is data-based (Supplementary Table S4). Physical and climatic layers rely on mature technologies such as remote sensing and land surface temperature retrieval, LiDAR and photogrammetry, microclimate simulation, and sensor networks. Cultural-significance sources are deliberately heterogeneous: participatory GIS and community mapping (Brown and Kyttä, 2014), oral history and ethnography, archival records and inventories, and potentially crowdsourced social media traces (Ginzarly et al., 2019b), all contributing geolocated data on usage and attachment.​

The principal difficulty lies with attributes carrying intangible and social values, which have no stable geometric extent. Four kinds recur: practices (evening gathering, informal trading) occupying a place only at certain hours or seasons; routes and thresholds whose social boundaries rarely match property lines; perceptual relations, such as a valued sightline, belonging to a line of sight rather than a surface; and ambient or associative qualities, such as a market’s sound or a corner’s remembered meaning. Each has a spatial referent but no polygon fully containing it. Rather than assigning false numerical proxies, the model attaches structured descriptors to locations and marks what is uncertain or contested, keeping the qualitative source intact and addressing critiques of datafication (Kitchin, 2014). In practice: a practice is recorded with its temporal condition, not as permanent land use; a sightline as geometry plus the value it carries for named holders; and an associative quality as a verbatim, provenanced statement, keeping the record auditable and its limits visible (Supplementary Table S4a).

3.4 Framework architecture

The framework is structured into five interacting layers (Figure 3, precisely defined in Table 6 to prevent conceptual overlap).

• Layer 1 (sensing) gathers climate, morphology, mobility, and significance data using sensors and participatory and ethnographic methods.
• Layer 2 (representation) co-registers physical, climatic and time-aware cultural data.
• Layer 3 (reasoning) combines CADA convergence analytics with AI and microclimate simulation.
• Layer 4 (negotiation) is where weights are set and trade-offs are adjudicated by human judgment rather than an algorithm.
• Layer 5, the Intervention Layer, facilitates the implementation of selected alterations in the public and residual domains. It subsequently determines the resultant thermal and social consequences, which are then communicated to guarantee the “twin’s” continued viability. Monitoring is separated from value revision: recorded outcomes may prompt a review of the representation layer but never automatically change cultural values or weights. Significance updates require provenance, version control, and human authorization via the negotiation layer, after which the reasoning layer is re-run on the revised state, preserving the distinction between measured change, interpreted significance, and authorized value revision.

TABLE 6. Definitions and boundaries of the five layers (input, function, output, and the question each answers).

3.5 Worked example: a hot-arid residual heritage precinct

3.5.1 Conceptual walkthrough

Consider a residual, road-widening remnant (a SLOAP) adjacent to a historic district in a Gulf city—exposed and unshaded yet used as an informal evening route to a heritage market with a valued sightline toward it. The sensing layer records its intense summer radiant load alongside ethnographic evidence of that valued route and sightline; the representation layer holds both its thermal exposure and its significance as a place of movement and gathering; and the reasoning layer consolidates this high-confidence, widely held value into a statement of collective appreciation. Divergent activation then generates three climatically responsive options and estimates each’s reduction in mean radiant temperature: a solid canopy, offering the greatest thermal gain but obstructing the sightline; a mashrabiya-screened re-enclosure, offering a smaller gain while preserving the sightline and reinforcing the threshold as a gathering edge; and a tree-and-water threshold, offering the smallest gain at a recurring water cost. The negotiation layer surfaces these trade-offs to residents and authorities, who set the weights by which the options are compared—each layer doing something a conventional twin cannot. Section 3.5.2 gives illustrative numerical values.

How would a twin, adhering to the traditional notion of fidelity, manage an identical plot, and what strategies would it employ? In the absence of a substantial layer, the site is reduced to mere geometry, materials, and radiant load. Consequently, a twin focused solely on optimization prioritizes the solid canopy, lacking the means to document the disrupted sightline, displaced gathering, or compromised route. The twins do not differ in terms of data or accuracy; rather, they differ in their structural ability to acknowledge cultural and distributive costs. This enables the practical significance of negotiability: the CADA × HUL twin can diverge from its thermal optimum to preserve a heritage value, a capability that the conventional twin lacks. Although the example is constructed, the risk that climate mitigation and adaptation measures will degrade heritage significance is recognized in the heritage–climate literature (Sesana et al., 2019).

3.5.2 Illustrative parameterized proof-of-concept

The walkthrough was parameterized for a single precinct—a road-widening SLOAP adjacent to Souq Waqif, Doha—to demonstrate the value-to-layer mechanism of Section 3.3 (elicitation, encoding, co-registration, scoring) with numbers rather than prose; all inputs and calculations appear in Supplementary Table S6. No microclimate simulations were run and no participants were consulted: climatic values were assigned from the order of magnitude reported for comparable interventions in hot-arid cities (Ali-Toudert and Mayer, 2006; Othman and Alshboul, 2020; Abedrabboh et al., 2025), and the two value sets are synthetic records built to exercise the significance-layer schema (Supplementary Table S4a). The exercise therefore tests the internal logic of the scoring loop, not the framework’s empirical claims.

Three layers were defined. The physical layer describes the remnant’s geometry and high-albedo paved surface. The climatic layer assumes a clear-sky July afternoon (∼14:00) with a baseline mean radiant temperature (MRT) of ∼62 °C at the crossing. The significance layer holds two weighted, provenance-tagged value sets: older residents (evening-gathering and sightline value 0.9; confidence high; tag “oral history”) and younger users (movement value 0.6; confidence moderate; tag “survey”), co-registered with the climatic surface (Figure 2) — tags indicating which fields elicited data would populate, not data collected here.

Assigned midday MRT reductions were ∼18 °C for the solid canopy (MRT ≈44 °C), ∼14 °C for the mashrabiya re-enclosure (≈48 °C), and ∼11 °C for the tree-and-water threshold (≈51 °C). A thermal-only ranking gives canopy > re-enclosure > threshold. The values-aware ranking applies a lexicographic rule: any option dropping a high-confidence significance score below a preset threshold is flagged and ranked below non-conflicting options, which are then ordered by thermal gain. The canopy, severing the older residents’ sightline, is flagged, reordering the ranking to re-enclosure > threshold > canopy (Kendall’s τ = −0.33 between the two orderings, i.e.,​ descriptive only at n = 3, and one of several defensible rules the negotiation layer could adopt). The result is structural: once significance is represented, the same thermal data support a different priority order, with the underlying conflict recorded for human adjudication rather than resolved algorithmically—the logic Proposition P1 specifies for empirical testing. The existing condition and the three options are schematically in Figure 4. The subsequent step involves a calibrated simulation (such as ENVI-met or SOLWEIG) validated by field measurements, with value sets obtained from genuine stakeholders under ethical approval; Propositions P2–P6 (Figure 5) outline the broader programme.

3.6 Propositions and their empirical strategies

The framework yielded six propositions (Supplementary Table S5). For each, this study provides the construct, comparison, or null, the disconfirming result, and a test strategy (measures in Section 4.3; summary in Figure 5). Together, they answer the research questions (Table 7): P1 and P3 address RQ3 (what the coupled twins make representable and protectable); P2 and P4 address RQ2 and the human-centered aim; and P5 and P6 address governance and transferability.

• P1 (representation changes prioritization): activating a time-aware significance layer changes the priority order of candidate interventions for the same precinct. Outcome: rank-order similarity between the two lists (Kendall’s τ), assessed by a paired permutation or bootstrap procedure against a pre-registered effect-size threshold; the intervention set must be large enough (e.g., eight or more options) for inference to be informative. Disconfirmed if the difference falls below the threshold. Test: paired computational experiments with and without the layer.
• P2 (CADA broadens the equitable option space): split into two outcomes, since more diverse options need not be more equitable. P2a: coupling convergent appreciation with divergent activation increases option-set dispersion relative to divergent generation alone. P2b: coupling improves distributive balance, measured as group-specific change in population-weighted access to spaces meeting a predefined thermal-comfort threshold, by vulnerability group and time period, alongside the distribution of intervention burdens; no single inequality index is treated as proof of justice. Disconfirmed if dispersion (P2a) or distributive balance (P2b) is unchanged. Test: controlled design-generation trial.
• P3 (HUL layering reduces heritage erosion): at matched thermal gain, a HUL-informed representation yields fewer severe, high-confidence heritage conflicts than a non-heritage-aware one. Erosion at stake needs specification, since the term is used loosely in the heritage-climate literature: it is neither the physical decay of heritage structures, which climate retrofits in public space rarely cause directly, nor a loss of tourism revenue, an economic outcome mediated by factors beyond design. Thermal retrofits typically reallocate shade, sightlines, thresholds, and circulation, so loss falls on the attributes carrying intangible and social values, such as a gathering displaced by re-paving, a route interrupted, a sightline severed, or informal commerce excluded from a redesigned edge (Australia ICOMOS, 2013; Jones, 2017; Sesana et al., 2019). Heritage erosion is therefore a measurable reduction in a place’s capacity to sustain the practices and relations through which its significance is held, reported as a multidimensional profile (value type, holder, extent, timing, confidence, severity, reversibility, disagreement), with intangible use values primary and tangible fabric secondary. Reversibility carries weight, since a removable canopy suspending a gathering for one season is not equivalent to a permanent re-enclosure, and the profile records that difference rather than summing it into one count. Disconfirmed if the HUL-informed representation produces as many or more severe, high-confidence conflicts than the comparator at matched thermal gain. Test: comparative scenario analysis at constant ΔMRT, scoring both representations against the same significance records so the comparison isolates representation from data.
• P4 (structured negotiation changes distributive priorities): facilitated negotiation produces different distributive priorities than automated selection; an equity improvement is not assumed. Because workshops can be captured by dominant groups, facilitation controls are pre-specified and group dominance is measured as a confound. Disconfirmed if population-weighted access to usable cooled space is unchanged or worse. Test: participatory workshop with pre/post access analysis.
• P5 (contestation preserved, not flattened): a multi-value-set representation surfaces trade-offs a single-score representation hides, measured by the number of distinct value conflicts and the divergence of ranked intervention sets. Disconfirmed if both are identical across representations. Test: within-participant comparison of the two representations.
• P6 (conditional transferability): once P1–P3 are supported in a hot-arid precinct, its three core functions—contested-value representation, explicit trade-off identification, human-authorized revision—remain implementable in a non-arid precinct after climate, governance, cultural-practice, and Layer one data conditions are re-specified as boundary conditions. This concerns the feasibility of implementation, not performance. Disconfirmed if core functions cannot be instantiated or depend on arid conditions. Test: sequential cross-site replication.

TABLE 7. Mapping research questions to sections, framework components and propositions.

4 Discussion

4.1 Theoretical implications

The framework makes two contributions to UDT literature. First, it reconfigures the digital twin from a technical representation into a value-aware instrument in which cultural significance and distributive outcomes are primary rather than supplementary. Second, it gives the HUL approach a computational form that carries tangible and intangible values across the design cycle, addressing the documented gap between UNESCO’s principles and the tools available to implement them (Angrisano et al., 2016; Ginzarly et al., 2019a) — though, as Section 2.3 sets out, it answers the implementation, indicator, and climate critiques of HUL, not the legal and institutional ones. It also contributes to design theory for AI-assisted urban modeling, by using CADA’s convergent–divergent sequence to govern generative tools within a spatially defined context and treating the outcome of appreciation as a computable constraint on activation.

At the core of both lies an alternative definition of the urban twin, as shown in Table 8. In the engineering lineage a twin is judged by fidelity: how precisely it reproduces its physical counterpart. Here it is judged by negotiability: its capacity to hold multiple value positions, expose their consequences under alternative interventions, and support authorized decision-making without collapsing disagreement into a single objective function. Under fidelity, whatever resists measurement—cultural meaning, but equally perceived safety, informal livelihoods, or the fairness of a distribution—is proxied or omitted, and the twin’s silence reads as unimportance; under negotiability, the same considerations enter as claims held by identifiable people. The shift therefore applies wherever contested public values meet a computational decision instrument, with heritage the clearest case rather than the only one. So construed, the critiques reviewed in Section 2.1 are objections to the fidelity interpretation, not to the instrument, and a negotiable-first conception answers them from within. The CADA × HUL coupling is one such implementation; others are possible.

TABLE 8. Key concepts, their definitions in this study, and their distinctions.

4.2 Practical and governance implications

For arid cities, the framework treats heat adaptation and heritage conservation as one design problem, cutting across the customary separation of climate and heritage departments. Because values are set before generation rather than consulted after design, the negotiation layer exposes the politics of weighting that default parameters otherwise conceal, retaining provenance, confidence, and disagreement instead of deferring to one optimized response (Knopf et al., 2025).

The governance implications are substantial. Urban twins are ordinarily acquired by engineering, IT, or planning-technology units, on the implicit understanding that their outputs are accurate. A negotiability-first twin instead requires a custodial arrangement giving heritage authorities, community representatives, and climate engineers standing to contest its weights—closer to a contested public budget than a control room and demanding for procurement cultures built on the latter model. The alternative is worse: a twin whose value commitments are set implicitly by whoever defines its defaults presents debatable political choices as neutral computation.

Timing matters, for a general reason. Technology is easiest to alter when its consequences are least understood and hardest once entrenched (Collingridge, 1980), a dilemma still framing the governance of emerging technologies (Genus and Stirling, 2018). Urban twins sit early on that curve, since most documented implementations remain partial (Ferré-Bigorra et al., 2022; Jeddoub et al., 2023), so what a twin represents, and who may contest its weights, remains open. Once a twin has supplied the evidence base for a sequence of planning decisions, its representational commitments are embedded in those decisions as well as the software and revising them means revising the decisions too—an argument about entrenchment, not cost. What the twin cannot supply is the legal standing, inter-agency coordination, or financing HUL implementation requires (Section 2.3); it can make that need visible, but these remain institutional tasks.

4.3 Pathways to empirical testing

Each proposition can be verified through practical research. The methods are summarized in Figure 5 and are based on established techniques, including microclimate simulation, GIS-based accessibility analysis, and participatory workshops. In brief, P1 and P3 use paired arid precinct scenarios (with and without the significance layer) compared to matched thermal performance; P2 and P5 use controlled design-generation trials; P4 uses participatory workshops measuring pre/post population-weighted access to cooled space; and P6 reparameterizes climate, governance, cultural practice, and data conditions for a non-arid precinct and re-executes P1–P3. Figure 5 summarizes this testing program as a proposition-by-method matrix: for each proposition (P1–P6), it shows the comparison or manipulation, the outcome measure, the pre-registered disconfirmation criterion, and the method—paired computational experiment (P1, P3), controlled design generation trial (P2, P5), participatory workshop (P4), and reparameterized cross-site replication (P6). The parameterized walkthrough in Section 3.5.2 illustrates the logic of the P1 comparison; however, because its inputs are assigned rather than measured, it does not test it.

Metrics were deliberately selected from pre-existing measurement tools to facilitate reliable and reproducible tests. The thermal environment can be quantified using metrics such as the mean radiant temperature (MRT), Universal Thermal Climate Index (UTCI), and Physiological Equivalent Temperature (PET), all of which are commonly employed in the analysis of microclimates in arid regions (Abedrabboh et al., 2025). Distributive access can be implemented via weighted two-step floating catchment area or least-cost path methodologies to quantify accessibility to desirable functional public spaces. The weighting incorporates both population density and heat vulnerability, contingent on data availability for such detailed operationalization, in accordance with the common practices in spatial equity research (Melis et al., 2023).

Value erosion is operationalized as the proportion of top-ranked interventions that drive a high-confidence significance score below a pre-registered threshold. The variety of options can be quantified using conventional dispersion metrics applied to a set of generated solutions. Concurrently, the prominence of trade-offs is determined by the number of conflicts presented to stakeholders for each representation type. Fixing disconfirmation thresholds before comparative tests guard against post hoc rationalization, presenting the propositions as a cost-effective, pre-registered agenda that can be achieved in a single adequately funded precinct study.

4.4 Transferability beyond arid contexts

The framework’s logic is general, even with the elevated stakes in the arid public realm, where shade is a contested resource, greening necessitates additional water, and heritage sites are exposed to severe solar radiation. In humid urban environments, the climate is associated with ventilation and moisture. In contrast, for postindustrial heritage cities, this aspect may highlight soil contamination and the strategic reuse of existing structures. The integration of CADA with HUL and its operationalization are not tied to a particular climate, but transfer requires more than changing climatic parameters: governance arrangements, cultural practices, and the availability of Layer one data differ between contexts and must also be re-specified. Because mean radiant temperature drivers, humidity, vegetation–water trade-offs, and heritage-governance regimes differ substantially across hot-arid and semi-arid contexts (Doha, Riyadh, Cairo, Alexandria, Siwa, and Marrakech), the climatic and governance layers must be reparameterized per site before the ranking loop is re-run, which is what Proposition P6 formalizes.

4.5 Counterarguments and ethical limitations: is the digital twin an inappropriate instrument

The strongest objection is that the digital twin is inherently technocratic in the sense defined in Section 2.1, and therefore incompatible with the grassroots, values-driven approaches of HUL and human-centered design. The charge has three components, which are not equally answerable. The first is epistemic: twin reasons only over what has been encoded, so encoding sets the boundary of what can count as consideration, and whatever is not entered is not underweighted but invisible (Kitchin, 2014). The second is procedural: authority to specify layers, weights, and thresholds rests with technical units, and a ranked output invites its recipients to treat the ranking as a finding rather than a proposal (Knopf et al., 2025). The third concerns collection: representing the use of public space requires observing residents, raising questions of consent and surveillance that participation does not by itself resolve (Charitonidou, 2022; Argota Sánchez-Vaquerizo, 2024; Dawkins and Kitchin, 2025).

The framework answers the second component and only partly the first and third. Against the procedural charge, the negotiation layer withholds the authority to select weights that are set and revised by identified people, revisions require provenance and authorization, and conflicts are surfaced rather than settled. Against the epistemic charge, plural value sets, explicit uncertainty, and retained verbatim statements keep the encoded boundary visible and auditable, but do not extend it. Against the collection charge, precinct-scale elicitation under ethical approval, aggregation, and the avoidance of individual-level tracking reduce exposure without eliminating it.

Which values are most likely to be lost can be stated rather than left open. Three kinds are at risk: values that are tacit or embodied, carried in habit rather than articulated when asked; values held by groups with least standing in a facilitated setting—in Gulf public space, typically migrant workers, women, and children, whose use of a place may be extensive but whose testimony is least often sought; and values with no stable spatial referent, such as associative or spiritual meanings attached to a practice rather than a location. Documented heritage digital twins, oriented to geometry and condition, represent none of these well (Tootoonchi and Hess, 2026). The framework mitigates but does not eliminate the risk, and elicitation designs should report which groups were reached and which were not.

4.6 Implementation realism and pilot pathway

This data-intensive framework invites charges of infeasibility; therefore, it is designed for incremental, precinct-scale application rather than immediate citywide rollout: significance elicitation is manageable at that scale, and residual, heritage-adjacent spaces (public spaces bordering a designated or informally recognized heritage precinct) benefit the most. A basic implementation requires only an established microclimate simulation, a GIS integration layer, and a structured significance layer for one precinct, gathered using conventional participatory and ethnographic methods. The negotiation layer can begin as a facilitated workshop reviewing the twin’s trade-offs before any dedicated software, so the Section 3.6 propositions become testable in a single precinct by a modestly resourced team rather than awaiting a city-scale twin that may be years from completion. Most documented UDTs remain in the developmental stages with limited operational deployment (Jeddoub et al., 2023; Lei et al., 2023); therefore, the CADA × HUL twin is not presented as an advancement beyond existing maturity (the operational-deployment stage of a UDT, from a static 3D model to a live, bidirectional cyber-physical system; Jeddoub et al., 2023) but as a reallocation of ongoing efforts, feasible with the current technological capabilities. The tension with the data appetite limitation (Section 4.7) is one of scale: the framework is feasible at the precinct scale with a bounded significance layer, and exhaustive city-scale capture is not required.

4.7 Limitations

This study has five principal limitations.

• The propositions form a research agenda lacking empirical substantiation and should not be construed as results.
• The proof-of-concept is illustrative: its climatic inputs are assigned from published orders of magnitude rather than simulated, its value sets are synthetic, and its three-option comparison cannot support statistical inference. It demonstrates the mechanism’s logic, not its validity.
• The framework is data-intensive, especially in eliciting significance information, which is costly, unevenly available, and politically contested. This demand is scale-dependent: a bounded, precinct-scale significance layer is tractable, but exhaustive city-scale capture is not, so the feasibility claimed earlier holds only on a scale, and validity is constrained by the quality, provenance, and inclusivity of the significance data obtained.
• Representational justice is the crux: which values are evoked, by whom, and with what authority to object determines whether the twin is democratized or entrenched. The multivalue-set design mitigates, but does not eliminate, the risk of domination by the most vocal participants, and urban twins raise privacy and surveillance concerns requiring explicit governance (Kitchin, 2014). The framework is offered for reflexive application, with explicit recognition of what computation cannot capture and what it cannot resolve: HUL’s legal, coordination, and financing constraints (Section 2.3).
• CADA’s role in the framework is justified functionally (Section 2.2), and Proposition P2 tests whether it adds value beyond divergent generation alone; comparison with alternative value-sensitive design processes remain necessary.

5 Conclusion

Urban digital twins are now the primary instruments for city governance, but their prevailing approach targets heat management and infrastructure in arid simulations while disregarding living heritage and equitable thermal comfort. The missing element, as argued here, is conceptual rather than a matter of data. This study proposes a theoretical framework coupling CADA’s convergent–divergent logic with the Historic Urban Landscape approach in a human-centered digital twin, enabling the representation, simulation, and negotiation of cultural meaning, climate performance, and equitable access. The operationalization mechanism addresses HUL’s implementation gap by adding a time-aware cultural layer, so heritage values are represented as changing rather than fixed. Six propositions are offered for empirical testing with pre-specified disconfirmation criteria. The framework is not a definitive optimization tool but a reflective instrument, meant to reveal trade-offs, retain dissenting viewpoints, and support constructive debate toward climate-resilient, culturally rooted, equitable arid-urban futures.

The broader principle the CADA × HUL twin exemplifies is that the urban digital twin stands at a definitional turning point: one path optimizes measurable elements into streamlined cities; the other defines the twin through negotiability, admitting disputes over shared space, more demanding of institutions and less reassuring, but keeping people and heritage central. This choice is sharpest in hot-arid cities, where heat and heritage press hardest, but the framework’s logic, comprising CADA-HUL coupling, contestable salience, and negotiability, is not bound to that setting. Transfer to humid, temperate, or post-industrial contexts requires re-specifying climatic, governance, cultural, and data conditions, as Proposition P6 formalizes. The negotiability-first approach is practical with current technology and framed for empirical scrutiny and peer assessment: if it succeeds, that is through a shift in the twin’s purpose rather than its computation; if it fails, it does so by disconfirming testable propositions, and either outcome advances the field beyond debates over technical efficiency.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. The authors used ProWritingAid for language refinement purposes. All outputs were reviewed, edited and verified by the authors. The authors take full responsibility for the content of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbuil.2026.1977948/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Nazish Abid, University of Bahrain, Bahrain

Reviewed by: Mohamed Ahmed F. Mahdy, Alexandria University, Egypt

Saad Hanif, University of Bahrain, Bahrain