Socio-Spatial Disparities in Heatwave Risk Perception and Cooling Shelter Utilization in Gwangju, South Korea

Abstract

Heatwaves are increasing in frequency and intensity owing to climate change, posing severe health risks to urban populations, particularly vulnerable groups. This study investigates public perceptions, adaptive behavior, and policy awareness regarding extreme heat in Gwangju Metropolitan City, South Korea, a heat-prone urban area. Using a mixed-methods approach, we analyzed primary survey data from 814 residents and secondary data from the 2020 Gwangju Citizen Heatwave Awareness Survey. Statistical analyses, including chi-squared and t-tests, examined differences across socioeconomic age groups. Results indicate that while general awareness of heatwave risks is high, low-income residents exhibit lower perceived severity, limited access to mechanical cooling, and greater reliance on passive avoidance behaviors. Awareness and use of municipal cooling shelters were low, with satisfaction hindered by concerns over accessibility, cleanliness, and operational hours. Television and emergency text alerts were the main information channels; however, trust and perceived usefulness were limited. Policy recommendations include spatially targeted shelter placement informed by vulnerability mapping, improved operational standards, diversified risk communication, and enhanced community engagement. This study underscores the importance of equity-driven adaptation strategies and provides practical insights for global municipalities facing similar climate-related heat risks.

1. Introduction

As global temperatures continue to rise, extreme heat events have emerged as critical challenges for urban resilience and public health. South Korea has witnessed record-breaking heatwaves, with the 2018 event resulting in over 44,000 heat-related illnesses, prompting the government to classify heatwaves as a natural disaster [1]. These events disproportionately affect vulnerable populations, such as older adults, low-income households, and those residing in poorly ventilated or heat-prone neighborhoods. Gwangju Metropolitan City, located in the southwestern basin of Korea, exemplifies this vulnerability due to its topographical enclosure, limited wind circulation, and rapid urbanization. Between 2005 and 2018, the number of heatwave days in Gwangju more than doubled, and the city’s average summer temperature has risen by approximately 2.7 °C over the past 80 years. In response, local authorities have implemented various heat mitigation policies, including the designation of public “cool shelters”, expansion of green infrastructure, and urban design interventions. However, the effectiveness of these interventions heavily depends on public awareness, behavioral compliance, and satisfaction with these measures.

Despite the growing urgency of this issue, empirical research examining heatwave risk perception, adaptive behavior, and public policy engagement at the municipal level remains limited—particularly in the Korean context. Most existing studies either focus on health impacts or emphasize the physical availability of cooling infrastructure, with relatively little attention to citizens’ subjective experiences, behavioral responses, or policy evaluations.

This study addresses this research gap by analyzing primary and secondary municipal survey results to evaluate citizens’ awareness, behavioral adaptation, and policy receptivity to heatwaves. By incorporating data from the 2020 Gwangju Heatwave Awareness Survey, this study provides a comprehensive understanding of how citizens across different socioeconomic groups perceive heat risks and respond to government intervention.

To guide this analysis, the study addresses the following research questions:

(1)

How do residents in Gwangju perceive the risks associated with extreme heat events, and how do these perceptions differ across socioeconomic groups?

(2)

What adaptive behaviors are employed in response to heatwaves, and what barriers limit the capacity to adapt among vulnerable populations?

(3)

How do citizens evaluate the accessibility and effectiveness of, and their satisfaction with, municipal cooling shelters?

(4)

What implications can be inferred in relation to policy changes to improve socially equitable heat adaptation strategies in mid-sized urban contexts?

By situating these questions within a mixed-methods empirical design, this study aims to offer actionable insights for enhancing local climate resilience. The findings may provide valuable guidance for urban policymakers in similarly vulnerable cities seeking to implement more inclusive and equity-orientated climate adaptation measures.

2. Literature Review

Extreme heat events have emerged as some of urban environments’ most pressing climate-related health threats. A growing body of literature has identified the socio-demographic vulnerabilities associated with heatwave exposure and mortality. Elderly individuals, low-income populations, and residents of high-density or poorly ventilated housing are particularly susceptible to heat-related health risks [2,3]. These groups often lack access to active cooling strategies, such as air conditioning, and instead rely on passive behaviors, such as staying indoors or increasing water intake [2,3].

Recent studies have underscored the importance of tailoring risk communication strategies to different demographic groups. For example, Schoessow et al. [4] found that low-income, non-white, and disabled individuals perceive themselves to be more vulnerable to heat waves, which aligns with observed health disparities. Sex and age also play significant roles: while women generally report higher risk perceptions, men exhibit greater heat-related mortality. Similarly, older adults, particularly those over 75, often underestimate their personal risk [4,5]. Beyond these variables, Rhomberg et al. [6] and Schoessow et al. [4] report that higher-income individuals tend to perceive greater vulnerability to, and severity of, heatwave risks, whereas those with lower educational attainment show less awareness. Farman et al. [7] further note that individuals living with others and those who have previously heard about heatwaves are more likely to perceive risks than those living alone. These findings collectively highlight the need for communication and intervention strategies that account for socioeconomic status, education level, living arrangements, and prior exposure to risk information. Geographic variations further complicate public perceptions and behavior. Individuals residing in regions with historically bitter climates tend to report greater risk awareness, whereas populations in cooler areas may experience heightened vulnerability owing to lower perceived threat levels [8]. Local susceptibility also influences perceptions, as those living in heat-vulnerable neighborhoods are more likely to express concerns during heatwaves [9].

Despite growing public awareness, behavioral adaptation remains limited among vulnerable populations. Erens et al. [5] reported that many individuals do not adopt recommended protective actions, often because of a lack of information or capacity to act. While generally aware of heatwave risks, health professionals frequently lack adequate knowledge or institutional guidelines to respond proactively [10]. Furthermore, increased exposure to risk communication does not necessarily translate into behavioral change, highlighting the importance of message clarity, perceived efficacy, and demographic targeting [11]. Socioeconomic status is consistently associated with disparities in heatwave responses. O’Neill et al. [12] observed that low-income urban residents in the United States are less likely to receive or trust official heat warnings, resulting in slower and less effective responses. Similarly, Reid et al. [13] found that although cooling centers were often available, their utilization was hindered by physical barriers and a general lack of awareness, particularly among older adults. Cooling shelters have proven to be effective in mitigating heat-related morbidity and mortality. Foster [14] reported that shelter use among older adults reduces the odds of mortality by up to 66%. However, the benefits of these facilities may be limited by poor accessibility and underutilization among high-risk populations [13,14]. Various shelter allocation models have been proposed to address these shortcomings. Woo et al. [15] introduced an integer programing model to maximize shelter coverage while minimizing operational costs. Yoon et al. [16] further developed a multi-depot location-routing model, demonstrating logistical improvements and potential cost savings of up to USD 49,000 compared with traditional approaches. Alternative approaches are essential in regions that lack infrastructure. Glicksman and Nelson [17] emphasized the role of passive cooling techniques in building design, especially in under-resourced or developing contexts where access to air conditioning is limited.

In Korea, thermal vulnerability is linked to income and aging. Kim et al. [18] and Eum [19] developed a vulnerability index across metropolitan regions to identify the disparities in thermal perceptions based on income and age. Kim et al. [20] stated that vulnerability indicators serve as essential tools for estimating climate change impacts and mapping vulnerability distributions across South Korea. However, few studies in Korea and other Asian cities have integrated public perception, behavior, and policy satisfaction into a unified empirical framework. Severtson and Henriques [11] highlighted that the efficacy of risk communication depends on frequency, perceived relevance, and clarity. The present study combines behavioral survey data with satisfaction metrics to address this gap and provide a holistic approach to equitable heatwave adaptation. Although numerous studies have examined individual vulnerability and shelter availability [13,14,15], few have integrated public awareness, behavioral adaptation, and policy evaluation within the same analytical framework, particularly in rapidly urbanizing Asian cities.

Building on the prior literature highlighting heatwave-related vulnerabilities and behavioral gaps, this study adopted a mixed-method survey approach to empirically assess public perceptions, behavioral responses, and satisfaction with municipal heat adaptation policies in Gwangju, South Korea. Furthermore, drawing on Kim et al. [18] and Kim et al. [20], this study emphasizes localized demographic stratification (e.g., income level, age, and district), which enables cross-sectional comparisons across vulnerability dimensions. Thus, the research aims to assess perceived and actual exposure and the equity and effectiveness of public service delivery during extreme heat events.

3. Materials and Methods

3.1. Study Area and Data Collection

This study was conducted in Gwangju Metropolitan City, located in the southwestern inland basin region of South Korea, with a population of approximately 1.5 million. The city’s enclosed topography and growing urban density have increased its vulnerability to persistent and intense heat waves, particularly with the accelerating effects of global warming. Figure 1 illustrates the geographical context and urban transformation of Gwangju Metropolitan City, the study site for this research. Panel (b) reveals substantial urbanization between 1988 and 2019, particularly concentrated in the central and eastern districts. These are closely aligned with the regions of rapid land-use change. Figure 2 showed examples of “cool shelter” facilities and field survey photographs.

Table 1. Details of the survey outline.

Contents	Detail
Attributes of the respondents	Gwangju citizens (over 20 years old): 1,170,046 people
Data	11–22 May 2020
Survey target	814 individuals (0.07% of the population). 514 ordinary citizens, 300 people with low income
Valid sample	Interview and online survey
Sampling	Extraction considering the district, sex, age, etc.

presents the survey details. A structured survey was conducted between 11 May and 22 May 2020, targeting residents aged 20 years and older in Gwangju Metropolitan City. A total of 814 valid responses were collected. Demographic representation was considered with respect to gender, age group, and administrative district. The sampling procedure did not involve probabilistic randomization. A purposive, non-probability sampling method with stratified quota allocation was used to achieve demographic balance across the selected variables. The survey was administered through face-to-face interviews and online questionnaires. No predefined sampling frame or invitation list was used, and response rates were not calculated. Within the total sample, 300 respondents identified as low-income were intentionally included for subgroup analysis. Low-income status was determined by self-reported household income. Vulnerable groups in the study included older adults (65+ years), low-income residents, and people living alone. Individuals with chronic illness and young children were excluded. The survey instrument included 24 categorical (nominal and ordinal) and scaled questions. The topics covered included the perceived severity of heatwaves, health-related symptoms, adaptive behaviors, and awareness or experience with local policy tools, such as public cooling shelters and early warning systems. Ethical approval for the study was obtained from the local research ethics board, and informed consent was obtained from all participants before data collection.

To supplement the primary dataset, additional data from the 2020 Gwangju Citizen Heatwave Awareness Survey was obtained by the Gwangju Institute of Science and Technology (GIST). This complementary dataset consists of responses from 1000 citizens and includes detailed satisfaction ratings for cooling shelter operations (e.g., indoor temperature, cleanliness, and accessibility), preferred information delivery channels, and policy suggestions. Including this external dataset allowed for triangulation and extended the insights into public service evaluation and citizen demands.

All data were analyzed using SPSS 19.0. Descriptive statistics (frequencies and means) were used to summarize the perception levels and behavioral responses. Chi-squared tests were used to examine associations between demographic variables (age, sex, and income level) and selected outcome variables. No formal inferential tests were conducted beyond this, given the non-probabilistic nature of the sample.

Table 2. Demographic characteristics of survey respondents.

-	Category	Subcategory	N	Percentage (%)
1	Sex	Men	390	47.9
2		Women	424	52.1
	Total		814	100
3	Age group	20–39 years	310	38.1
4		40–59 years	208	25.6
5		60 years and over	296	36.4
	Total		814	100
6	Income level	General	514	63.1
7		Low income	300	36.9
	Total		814	100

presents the demographic characteristics of respondents. A total of 814 respondents, comprising the general population and low-income groups, participated in the survey. As shown in Table 1, the sample included a relatively balanced sex distribution (47.9% men and 52.1% women) and diverse age representations. Notably, 36.9% of the respondents were classified as low-income based on their self-reported household income levels.

3.2. Survey Instrument

The structured questionnaire employed in this study comprised 24 items grouped into four thematic domains designed to capture respondents’ perceptions, experiences, and policy awareness related to heatwaves. The instrument was developed in consultation with urban climate and risk communication experts and pretested with a pilot group to ensure clarity, reliability, and contextual appropriateness. The first section assessed the respondents’ perceptions of heatwave risks, including their subjective understanding of heat-related threats and the primary sources through which they obtained information. This section captures individual-level cognitive orientation and informational exposure to climate-induced extreme heat events. The second section focuses on heatwave impact experiences, addressing both physical symptoms and behavioral responses during heatwave periods. It includes items related to health discomfort, such as fatigue or dizziness, and self-reported coping behaviors (e.g., increased water intake, modified activity schedules, or use of cooling devices). The third section evaluated the discomfort and functional constraints experienced during heatwaves, such as avoidance of certain places or periods and restrictions on daily routines. These items were intended to illuminate the spatial and temporal patterns of heat-related vulnerability from the residents’ perspectives.

The final section addresses institutional policy awareness and civic expectations. Specifically, it focuses on public cooling shelters (commonly referred to as “cool shelters”), which have been widely implemented by local governments in South Korea as a central heat adaptation measure [21]. This module assessed respondents’ awareness of, past use of, and satisfaction with shelter services, as well as broader public suggestions for improving municipal heatwave policies. The questionnaire utilized a combination of nominal, ordinal, and Likert-scale response formats, facilitating descriptive statistics and inferential analyses across demographic and spatial regions.

All participants were adults aged 20 years or older residing in Gwangju Metropolitan City. No personally identifiable information (e.g., names, contact details) was collected, and all survey responses were anonymized at the point of data entry. Participation in the survey was entirely voluntary, and participants provided informed consent prior to beginning the questionnaire.

4. Results

4.1. Perceived Severity and Risk Awareness

Among general citizens, 70.4% rated heatwaves as “very serious”, whereas only 56.3% of low-income respondents reported the same, suggesting a discrepancy in perceived environmental vulnerability. Sex and age did not yield statistically significant differences in perceptions.

Table 3. Perceived severity of heatwaves by socio-demographic group.

	Group	Perceived as Very Serious (n)
1	General population	70.4% (362)
2	Low-income group	56.3% (169)
3	Men	31.9% (260)
4	Women	68.1% (554)
5	20–39 years	38.1% (310)
6	40–59 years	25.6% (208)
7	60 years and over	36.4% (296)

illustrates the differences in the perception of heatwave severity across demographic groups. Although 70.4% of the general population considered heatwaves “very serious”, only 56.3% of the low-income group reported the same concern. Among the age groups, the highest proportion perceiving heatwaves as very serious was observed in the 20–39-year cohort (38.1%), followed by adults aged 60 years and over (36.4%). In contrast, the 40–59-year group reported the lowest perceived severity (25.6%), suggesting a potential underestimation of heat-related risks in this middle-aged population. This pattern may reflect differences in daily exposure, occupational conditions, or perceived personal resilience across age categories.

4.2. Behavioral Responses to Heatwaves

The comparison of heatwave response behaviors between income groups (Figure 3) presents self-reported adaptive actions taken by low-income and general-income respondents during heatwave periods. Key responses include increased water intake, use of fans or air conditioners, altered daily routines, and visitation of public cooling shelters. The figure highlights notable disparities, particularly in access to mechanical cooling and utilization of institutional resources, with low-income groups showing lower engagement with cooling centers despite higher reported heat stress.

The most common coping behavior was increased use of air conditioners (58.6%), followed by minimizing outdoor activities (51.4%). Notably, low-income households relied more heavily on behavioral avoidance (e.g., staying indoors without cooling) owing to limited access to cooling appliances, according to self-reported behavioral responses to extreme heat among general and low-income residents in Gwangju. Air conditioning use was more prevalent among the general population, whereas low-income groups were more likely to rely on avoidance behaviors, such as minimizing outdoor activities.

4.3. Policy Awareness and Shelter Usage

The mean satisfaction scores reported by respondents for various aspects of public cool shelter services (Figure 4) include thermal comfort, cleanliness, accessibility, informational visibility, and operating hours. Responses were measured on a five-point Likert scale (1 = Very Dissatisfied, 5 = Very Satisfied). Overall, cleanliness and accessibility received relatively higher satisfaction ratings, while informational signage and perceived thermal comfort were rated lower. These results suggest targeted areas for improvement in public cooling infrastructure design and communication. Only 36.8% of respondents were aware of the city’s cool shelter program, and actual usage was reported by just 17.2%. Shelter satisfaction varied significantly, with temperature control and spatial accessibility emerging as major concerns. According to secondary survey data, cleanliness (2.94/5) and distance from home (2.76/5) were the least satisfactory factors, particularly for older residents.

The proportion of respondents who were aware of and had used designated public cool shelters during heatwave events (

Table 4. Awareness and usage of cool shelters.

	Group	Awareness of Shelters (n)	Actual Usage (n)
1	General population	36.8% (189)	17.2% (88)
2	Low-income group	28.5% (86)	10.7% (32)

) is presented, showing levels of awareness and actual usage among the surveyed population. Results are disaggregated by income group (low-income vs. general-income), showing notable disparities in awareness and utilization rates. While general-income respondents demonstrated higher levels of awareness, the actual usage rates remained low in both groups, suggesting gaps in information dissemination and behavioral engagement. Awareness and usage rates of government-designated cool shelters were generally low (Table 4). Only 36.8% of the general population and 28.5% of the low-income group were aware of such facilities, with actual usage rates falling below 20%. These findings suggest limited reach and penetration of current shelter programs.

Table 5. Within-sample associations between demographic variables and shelter usage (chi-squared tests with effect sizes).

	Variable	Grouping Criteria	Test Type	df	Test Statistic	p-Value	Effect Size
1	Shelter usage × Income level	Low, middle, high income	X2 test	2	15.82	<0.001	Cramér’s V = 0.14
2	Shelter usage × Age group	20–39, 40–64, 65+ years	X2 test	2	12.47	0.002	Cramér’s V = 0.14

Notes: df, degrees of freedom. Effect sizes are reported as Cramér’s V. All analyses are within-sample and exploratory; results are not generalized to the population. Variable coding: age (20–39, 40–64, 65+), income (self-reported: low [bottom 30%], middle, high). Sample size for specific tests may vary due to item nonresponse. No multiple-comparison correction was applied.

shows within-sample associations between demographic variables and shelter usage. Chi-squared tests were used to examine within-sample differences between demographic groups in heatwave perception, adaptive behaviors, and cooling shelter use. Independent samples t-tests were not included in the final analysis, as no such results were reported in the original manuscript. Given the non-probabilistic sampling design, the chi-squared results are presented solely for exploratory purposes and are not intended for population-level inference. Effect size measures were calculated for each test: Cramér’s V was used for chi-squared tests. Cramér’s V value for both tests was 0.14, indicating a small effect size. This suggests that, while these socio-demographic factors are related to shelter utilization, their explanatory power is limited. Consequently, other determinants—such as shelter proximity, operating hours, quality of facilities, and the effectiveness of information dissemination—may play a more substantial role in shaping behavioral responses. These findings imply that policy interventions should extend beyond demographic targeting to address broader structural and environmental barriers to cooling shelter use.

4.4. Information Channels and Trust

Television was the most common source of heatwave warnings among the respondents (36.8%), followed by mobile phone emergency alerts (28.7%) (Figure 5). Online channels such as municipal websites and social media were used far less frequently, particularly by older or lower-income participants. These results underscore the necessity of maintaining multiple redundant communication pathways to ensure equitable access to early warning information. Regarding the primary sources through which residents received heatwave-related information, television and emergency text alerts were the most commonly reported channels, followed by the internet and social networks.

4.5. Policy Improvement Demands

Regarding citizens’ suggestions for policy improvement (Figure 6), a substantial proportion (66.9%) of the respondents suggested extending shelter operating hours, and 59.4% recommended establishing more shelters within residential areas. These findings suggest strong public support for policy expansion and customization, particularly among vulnerable groups. Regarding public preference for heatwave policy enhancements, the most frequent suggestions include extending shelter operating hours, increasing the number of facilities, and improving cleanliness and outreach. These findings provide direct feedback for urban planners and public health authorities aiming to adapt to local strategies under increasing heat stress conditions.

5. Discussion

The findings of this study highlight the increasing public awareness of extreme heat events in Gwangju and reveal significant gaps in adaptive capacity, particularly among vulnerable populations. Although most residents recognized the severity of the heatwaves, substantial disparities in risk perception, behavioral adaptation, and policy engagement were observed across socioeconomic lines.

First, the lower reported perception of heatwave severity among low-income respondents compared to the general population suggests a potential association between economic status and environmental risk sensitivity. Previous studies have reported similar patterns, where limited adaptation options correlate with reduced reported concern [2,18]. This suggests the importance of risk communication strategies tailored to socially disadvantaged groups.

Second, behavioral responses to heatwaves were significantly influenced by economic constraints. Low-income respondents’ high reliance on behavioral avoidance strategies, such as staying indoors without cooling devices, highlights their limited access to technological adaptations such as air conditioning. These findings are consistent with international studies linking income inequality to varying access to urban cooling resources [12]. Policy measures should therefore be designed to address both general needs and the structural limitations faced by vulnerable groups.

Third, despite the city’s investment in “cool shelters”, awareness and utilization of these facilities remain low. According to the primary and secondary surveys, only a minority of respondents reported being aware of or using these shelters. Furthermore, satisfaction ratings for shelter facilities were below average, with key concerns including inadequate cooling, inconvenient locations, and lack of cleanliness. These results suggest that physical access alone is insufficient for successful policy implementation and that user experience and perceived legitimacy are equally critical. The effectiveness of information dissemination has emerged as a vital area for improvement. Although television and emergency text alerts were the most common sources of heatwave information, many respondents expressed low trust or felt that the messages lacked specificity. This indicates a “potential information–action gap”, where residents receive warnings but are not adequately guided toward actionable behavior. In line with studies on environmental risk communication [11], our findings advocate for a more interactive, localized, and behaviorally informed communication strategy.

Several clear recommendations have emerged from a policy perspective. First, cities should consider the spatially optimized placement of shelters based on population vulnerability mapping using GIS tools. Second, shelter design and operational standards must prioritize user comfort, cleanliness, and accessibility. Third, risk communication should be diversified through mobile applications, social media, and community-level engagement, particularly targeting digitally excluded groups such as older adults.

This study has limitations. The cross-sectional survey design does not permit causal inference, and certain vulnerable groups (e.g., older adults in institutional care) were not included. Additionally, physiological and meteorological measures such as the wet-bulb globe temperature and humidity index were not incorporated, which may have strengthened vulnerability assessments. Future studies should apply longitudinal and spatial analyses to examine changes in behavior and policy performance under evolving climatic conditions.

6. Conclusions

This study examined public perceptions, behavioral responses, and policy awareness regarding extreme heat events in Gwangju Metropolitan City, South Korea, using both primary survey data and a secondary municipal dataset. The results show that, while heatwaves are widely recognized as a serious risk, disparities exist in perception and adaptive capacity between the general population and vulnerable groups, particularly older adults and low-income residents. These groups reported greater environmental discomfort and reduced access to technological cooling resources.

Behavioral coping among vulnerable groups was largely avoidance-based, with lower use of proactive cooling strategies. Awareness and utilization of public cooling shelters were limited; approximately one third of respondents knew about the program, and fewer than 20% used a shelter during extreme heat. Satisfaction levels were modest, and concerns focused on temperature control, accessibility, and cleanliness. Shelter usage was associated with both income level and age. Information about heatwaves was primarily obtained through television and text alerts, but respondents expressed low confidence in the quality and timeliness of these messages. Reported preferences included longer shelter operating hours and greater proximity to residential areas.

Policy recommendations include the following:

(1)

Municipal disaster management departments, public health centers, social welfare offices, and urban planning divisions should jointly address shelter provision, operational standards, and accessibility.

(2)

Community organizations and non-governmental groups should be engaged to support outreach and service delivery, especially for older adults, low-income residents, and those with mobility constraints.

(3)

Risk communication channels should be diversified to include mobile applications, social media, and in-person community outreach.

While this study was limited to one city, the methodology and findings can inform adaptation planning in other urban areas with similar climatic and demographic profiles. Future studies should incorporate longitudinal and physiological data to assess changes in behavior and policy effectiveness under evolving heat risk conditions.