IntroductionThe Anthropocene, characterised by the human-carbon nexus, has led to heat-related climate hazards, causing approximately 489,000 deaths annually since 2010, with about 15,500 in China1. In response to these critical human-induced environmental changes, global initiatives such as the 2015 Paris Agreement, the Sendai Framework, and the 2030 Agenda for Sustainable Development aim to limit global warming to below 2 °C2. Aligned with Climate Change Action, countries have been regularly raising the ambition of their Nationally Determined Contributions and Disaster Risk Reduction. China, for instance, foresees peak carbon emissions by 2030, followed by achieving carbon neutrality by 2060. These changes will entail significant socio-economic changes impacting vulnerable communities, often with low-quality housing and inadequate infrastructure, accommodating large, high-risk populations, and bearing the dual burden of extreme heat and the challenges associated with transitioning to carbon neutrality (such as chronic disease, unemployment, increased energy costs, and potential displacement risks).The importance of quality settlements and resilient infrastructure is particularly framed under Agenda 2030’s SDG11 (Make cities and human settlements inclusive, safe, resilient, and sustainable). However, SDG11 risks being the least connected to other SDGs due to its broad scope and potential trade-offs3. Also, reaching SDG11 targets in vulnerable communities requires additional effort because of resource constraints, capacity limitations, disconnected functional departments, and institutional deficiencies, which complicate the integration of climate adaptation and mitigation efforts4. To remedy these shortcomings, one of the most cost-effective ways for vulnerable communities’ climate change adaptation is potentially through community-based approaches4,5, including when addressing increasing urban heat, which is primarily considered in this study.Community-based approaches consist of supporting residents and community organisations working in partnership with local institutions to implement an ‘upgrading’ plan, integrating multiple measures adapted to local needs, capacities, and circumstances for urban heat adaptation. Different forms of urban upgrading, including comprehensive community-based upgrading, can be aligned with SDG11 sub-targets to enhance the resilience of vulnerable communities. However, coordination is essential when situating the SDG11 sub-targets in specific contexts, and differences between national frameworks and governance should be considered.In China, numerous assessment studies have employed top-down approaches to evaluate and identify metrics for SDG indicators, drawing on statistical yearbooks, government reports, local archives, or emerging big data sources (e.g., APIs, remote sensing satellites6). Despite these efforts, indicator sensitivity varies significantly based on selected datasets, particularly in the Global South, where data is often lacking, not publicly available, or not shared across sectors7. This issue is especially pronounced at the community level, which extrapolates higher-level statistical trends to local conditions, struggling to measure who is served and who contributes to decision-making4. The bottom-up approach, using local data (e.g., community surveys & participatory mapping) is necessary to capture the perceived exposure, hazards, and vulnerability associated with increasing heat and community-based solutions. By enhancing the sensitivity and applicability of SDG indicators, this approach helps bridge the action gap in SDG11.Our empirical study is conducted in Xi’an, a mature historical and cultural Chinese city that experiences increasingly high temperatures. The city’s urban upgrading strategy has officially shifted its focus from massive demolition towards offering higher standards of living, although this is yet to be fully materialised. This paper explores vulnerable communities’ perceptions and coping with increasing urban heat, their available resources, and how to align SDG11 with future high-temperature scenarios. It aims to reveal barriers to action and inform future community-based upgrading policy steps under the SDG11 framework.Theoretical lens to assess vulnerability to increasing urban heatVulnerability is a contested concept with multiple definitions8,9, for which maintaining an analytical perspective is crucial to avoiding unproductive intervention efforts10. The attempts to introduce a perfect disaster mitigation and adaptation model failed to characterise the relationship and co-evolution between risks, hazards, vulnerability, exposure, and coping capacities, as this is unachievable in an uncertain world. However, two main theoretical streams that address human-environment vulnerability are identified11: i) vulnerable livelihoods to poverty; and ii) the vulnerability of social-ecological systems (SESs) sitting within natural hazards (risk-hazard) and human ecology research and including resilience studies. This study is framed under the latter stream, which synthesises concepts of resilience and vulnerability - seemingly opposing characteristics of SESs12. Under this theoretical stream, the Pressure-and-Release (PAR) model was developed by Wisner13 to more effectively elucidate vulnerability in developing countries11,12, serving as the foundational framework for this research.The PAR model explains disasters by tracing the linkages between the impact of hazards on people and the sets of social factors and processes that generate vulnerability. In the case of extreme heat, disasters occur at the intersection of exposure to urban heat hazards and pre-existing vulnerability. Under the PAR perspective, vulnerable communities are highly sensitive to increasing urban heat due to a lack of community capital resulting from historical conditions14, with susceptibility further heightened by pre-existing vulnerabilities and exposure factors related to their physical living conditions and demographic composition. When they are troubled by hazards, exposures, and structural vulnerabilities, the communities have few resources and coping capacities, making absorbing, adapting, and recovering from the hazard difficult, often leading to severely damaging or even disastrous consequences. Hence urban heat is sometimes considered a ‘slow-onset hazard’, an ‘invisible’ or ‘neglected’ disaster since affected vulnerable communities are often forced to overcome difficulties independently because they do not know when/where the hazard/disaster occurs15,16. Relatedly, heat stress represents a set of conditions in which the body is stressed by overheating, exacerbating these challenges. All these disproportionate urban heat-related effects are caused by physical, social-environmental, and individual awareness factor(s)17.The PAR model encompasses three types of vulnerability factors13: i) the root causes as economic, demographic, and political processes reflect the distribution of power and affect resource distribution among different groups, for example, the lack of influence on cooling resource allocation decisions due to participatory and political rights; ii) dynamic pressures translate root causes in time and space into unsafe conditions (see next), such as the lack of heat adaptation/mitigation resources or increased heat due to urban development; iii) the unsafe conditions are specific manifestations of group vulnerability in relation to hazards, depending on the initial level of well-being of the population and how that level varies between regions, microregions, households, and individuals, for example, the lack of cooling infrastructure or fragile housing conditions.In the Chinese context, the physical environment of vulnerable communities is characterised by ageing, inadequate maintenance, and poorly planned and designed housing and infrastructure. Secondly, from a social perspective, the vulnerability is exacerbated by the erosion and disappearance of the work-units system (i.e., 单位, Danwei), as well as informality (see specific examples in the background), which brings disarray to grassroots community governance. Thirdly, given China’s significant rural-to-urban migration, economic filtering processes render these communities in China an alternative to affordable/social housing for mobile populations due to low rents, which accommodate young labourers moving their families, including the elderly and children18. This is intrinsically tied to the informal economy sectors like street vending, scavenging, cleaning, and construction19, which expose community members to greater heat stress.Community-based adaptation and upgrading pathways under SDG11Under the 2030 Agenda for Sustainable Development, the sustainability challenge has shifted from protecting natural resources to adaptation because of the uncertain future20. As the community’s available resources significantly impact its adaptation approach21, this shift requires moving beyond exogenous paradigms primarily based on risk and disengagement, which fail to capture the localised aspects of communities. In resource-rich communities, large and expensive public projects could facilitate adaptation to climate change, however, in resource-poor vulnerable communities, community-based experiences are often viewed as substitutes for government support or external funding22.Utilising local experience to assist communities in implementing upgrading programmes by involving different social groups and local governments is a cost-effective way of building resilience. Such community-based adaptation involves residents in identifying the underlying vulnerabilities, followed by co-developing climate change adaptation scenarios that consider upgrading pathways22. This can ensure actions are locally adapted, and knowledge channelled upward for incorporation into sustainable policy development23.Seeking measures beyond the community scale to the city-regional scale requires strong community-local government partnerships for infrastructure improvements and future risk assessments to address climate-related hazards4,24. The ability to seek internal and external support depends on the socio-economic context and pre-existing vulnerabilities, with successful community-based adaptation often influenced by governance networks25. Relatedly, community-based governance uses community-based approaches to establishing goals, defining rules, and controlling the outcomes from the respective rules26. However, its effectiveness depends on local communities’ ability to respond to environmental priorities (e.g., addressing high temperatures) and to engage community personnel to implement related measures27,28. SDG11’s broad goals can potentially be better addressed through this localised approach to tackle vulnerabilities and foster synergies within its sub-targets for effective and manageable adaptation.ResultsDemographic characteristicsThe respondents were predominantly Xi’an natives (76.19%), of Han ethnicity (85.71%), and males (61.9%). A balanced age distribution of the sample was obtained, and most respondents had secondary education or above (Fig. 1, and Supplementary Table 1). Family sizes ranged from 1 to 12 people, with 3- or 4-person households being the most common (55.18%). The dominant income brackets were 2200–3400 RMB/month (302.82–468 USD, 21.57%) and 3400–6700 RMB/month (468–922.23 USD, 21.85%), coinciding with those at the national level. Only a marginal 0.84% (3 respondents) indicated an income of less than 650 RMB/month (89.47 USD), representing the lowest-income group according to quintiles defined by the National Bureau of Statistics in China.Fig. 1: Overview of eight key demographic dimensions.Total sample size is of 357 survey participants. Full question prompts and category definitions are provided in Supplementary Table 1.Full size imageRespondents most commonly lived in urban villages (brick and concrete housing, 33.33%), gated commercial housing (29.97%), and Danwei housing or affiliated (sub-) communities (14.01%). 80.67% of respondents owned their homes, 17.65% rented, and the remainder lived with relatives or in collective housing. Rent ranged from 200 RMB/month (~28 USD) in shantytowns to 4000 RMB/month (~550 USD) in gated commercial housing (Fig. 1; see Supplementary Table 1 for further details).Descriptive statistics of residents’ urban heat stress perceptionBased on the survey data, we iteratively refine the Pressure-and-Release (PAR)13 framework-formulated variables (Table 1 & Fig. 9) by aggregating responses to multiple questions. These variables capture unsafe conditions (Exposure and cooling demand), and combined dynamic pressures (Awareness, Community-level services, Socio-economic pressure) and governance quality (Community-based governance effectiveness) as the root causes.Table 1 Definition and theoretical framing for latent variablesFull size tableExposure and cooling demandMany residents perceived the surge in road vehicles (N = 220) as a primary contributor to increased urban heat, followed by population influx (N = 201). Those who perceived their community as hotter than others (29.97%) associated it with factors such as rapid population growth and urbanization (e.g., R173), exposure-prone locations (e.g., R289), or unprotected buildings and infrastructure (e.g., R324). Those who considered their homes hotter than others (14.85%) similarly attributed it to high population density (e.g., R265), aged buildings (e.g., R7), poor insulation (e.g., R40), inadequate ventilation (e.g., R331), living on upper floors (e.g., R16), inefficient heat dissipation (e.g., R106), or west-facing rooms (e.g., R137). Most respondents (94.4%) had air conditioning, with 61.9% using it for over six hours daily.Satisfaction with heat mitigation infrastructure was low, with 42% of respondents expressing dissatisfaction due to inadequate cooling facilities and restricted hours. Severe heat events, described as ‘burning’ or ‘furnace-like’ experiences, with heatstroke among outdoor workers and fatalities among the homeless (reported by R31, R14, R343, & R94), were observed by 45.1% of respondents. These were experienced especially when midday temperatures frequently surpassed 40 °C and reached over 45 °C in July and August of 2023, 2022, or 2021 (as noted by R136, R141, & R189). During these peak summer months, 31.37% of respondents reported spending 2–6 h outdoors daily, and 17.93% spent over 6 h due to necessary activities (see Supplementary Table 1).In open-ended responses on cooling measures, outdoor strategies were dominated by natural and traditional methods: using fans (24.93%), seeking shade (22.97%), sun protection (umbrellas, sunscreen, protective clothing) (19.33%), and water-related cooling (hydration, swimming, water spray) (19.05%). Indoor cooling showed greater uniformity, with most respondents relying on air conditioning (81.51%), followed by electric fans (28.85%).The relationship between employment and heat impact varied across groups. Indoor workers largely felt high temperatures had ‘no/minor impact’ (28 mentions), attributed to indoor air conditioning (AC), although some indicated ‘health and well-being risks’ (24 mentions) like fatigue and ‘increased difficulty in work/study’ (23 mentions). These issues were more severe for outdoor workers, who reported dizziness and heatstroke. Self-employed individuals noted reduced business due to heat-deterring customers (e.g., R191 & R201), while a cold drinks vendor saw increased sales (R42). ‘Economic impacts’ were linked to ‘Behaviour changes or willingness to go out’ (16 mentions) and work difficulties, while the unemployed experienced additional job-search difficulties or decreased job-seeking motivation. Reluctance to go outdoors was also encountered among retirees and students (Fig. 2).Fig. 2: High-temperature effects and occupational status.The number of respondents with different occupational statuses who report different impacts of high temperatures is presented in each stacked bar in the chart. Coloured segments and associated numbers correspond to different types of heat-related impacts.Full size imageAwarenessRespondents primarily relied on digital and electronic media for information on urban heat, with Chinese social media (250 mentions) and television (195 mentions) as the most cited sources. However, less than half (40.9%) were aware of the government’s daily allowance (1.5–2 USD) for employees working during the hot season from June 15 to September 15, and only 17.65% had received it (e.g., R15, R78, R79, R197).Severe heat-related issues affected 22.13% of respondents or their acquaintances, especially those in driving, cleaning, construction, security, and courier roles, with reported conditions such as heat stroke, dehydration, cognitive impairment, and fatalities. Additionally, 31.09% were aware of broader urban heat risks, including health impacts, reduced productivity (e.g., R73), increased AC costs (e.g., R52 & R250), fire hazards (e.g., R27 & R183), vehicle damage (e.g., tyre bursts, R170, R341), economic losses (R231), environmental effects (e.g., R230, R232, & R260), and infrastructure strain, particularly road damage (e.g., R134 & R331), see Supplementary Table 1.Community-level servicesSuch services for heat exposure were acknowledged by 28.01% of respondents, including free blood pressure, glucose, and electrocardiogram tests, as well as community hospital visits (R4 & R7), health insurance (R98 & R251), heat stroke assistance (R8 & R133), and cooling supplies (R133, R267, & R282). However, only 8.4% were aware of local heat-preparation measures such as heat allowances (e.g., R60 & R85), water sprinkling (e.g., R10 & R170), cooling consumables (e.g., R109 & R297), AC-equipped community centres (R138), and sun shelters (R204). Furthermore, just 3.08% had received training on urban heat mitigation, which was offered through emergency operations (R63), community programs (R85, R86, R89, & R170), workplace sessions (R124 & R212), seminars (R109), or school courses (R342).About 26.61% reported access to home upgrades for urban heat resilience, with 68 emphasizing AC installation or extended usage. Additional methods included blackout curtains (R202 & R338), electrical upgrades (R8), structural adjustments (R6), greenery planting (R306), and enhanced insulation on facades or roofs (R19, R197, & R301). A few mentioned financial or livelihood improvement efforts (R213, R233, & R267) and housing replacement (R43, R146, & R302). They all underscored the self-reliant nature of these interventions and not the support from the community/government.Socio-economic pressure7.84% of respondents reported experiencing discrimination, inequality, or social exclusion due to factors like living in informal settlements (R141), lacking the ability to install AC (R4, R7, & R8), being part of the mobile population (R201), socio-economic disparities (R187 & R188), or limited community representation (R40, R242, & R243). A higher percentage of 35.29% of respondents identified socio-economic factors as amplifying their heat vulnerability, citing high electricity costs limiting AC use (54 mentions), along with livelihood challenges, such as low income, minimal retirement benefits, and inconsistent earnings (42 mentions). High housing costs (R42, R60, R163, & R292) and, for some, the need for seasonal migration (R198) further contributed to their susceptibility. Additionally, 21% reported financial barriers to accessing heat mitigation products or services, with many expressing a desire for improved housing, such as homes in green environments (R104 & R158) or those featuring green (R229) or cool roofs (e.g., R16, R256, R286, & R34).29.97% had a debt repayment plan for housing, cars, or tuition fees, and only 23 respondents reported no significant financial impact. Others experienced notable effects, such as lacking discretionary funds and reduced spending on heat mitigation strategies (21 mentions), decreasing AC usage due to high electricity costs (24 mentions) and substituting it with economical alternatives like fans and showers (R29), or cutting other daily expenses (R141).Community-based governance effectivenessRespondents largely acknowledged the importance of heat mitigation policies, with 19.61% viewing them as highly urgent and 31.37% as urgent but not paramount, while only 2.24% did not consider urban heat a concern.Most respondents rated the effectiveness of local institutions in managing heat issues as average or below average, with 37.25% rating it negatively due to the absence of programs specifically protecting residents from extreme or rising urban heat. A further 11.20% rated it as very poor, attributing diminished adaptive capacities to inadequate institutional management.In terms of community engagement, most respondents reported minimal engagement in decisions related to projects, mitigation plans, and urban planning. A large segment, 36.97% (see Supplementary Table 1) felt entirely excluded, with no opportunity to contribute or influence outcomes, while 30.81% experienced limited engagement, being consulted only for feedback on pre-made decisions. In contrast, a small minority, 1.68% (6 respondents), reported significant influence, with their input directly shaping implemented decisions.Statistical modelling of vulnerable community perceptions on urban heat stressThe Kolmogorov–Smirnov test (sample = 357) showed a significant departure from normality for all variables (see Table 2, more details in the Supplementary Table 3), indicating the necessity of non-parametric methods for subsequent correlation analyses (applied below).The internal dynamics of the effectiveness of community-based governance (Q12, Q15, & Q16, see Supplementary Table 1 for all questions’ corresponding meaning) and the impact of satisfaction with cooling infrastructure (Q5) on these dynamics were examined to understand how external factors influence internal management dynamics (Fig. 3).Table 2 Variables employed in statistical analysesFull size tableFig. 3: The Spearman correlation analysis for cooling infrastructure and community-based governance.Four survey items (Q5, Q12, Q15, and Q16) were assessed. Stronger positive correlations are shown in darker red shades, weaker correlations appear in lighter shades of red, and negative correlations appear in purple.Full size imageThe Spearman correlation revealed a moderately positive correlation between perceptions of local institution effectiveness (Q12) and engagement in community decision-making (Q15) (ρ = 0.53, p