Spatiotemporal Dynamics and Impact Mechanism of Heatwave Exposure in the Urban Elderly Population Across China

Abstract

Heatwaves are intensifying across China under global warming. Although previous SSP-RCP studies project more frequent and intense events, systematic evaluations of exposure mechanisms among the elderly in China remain limited. The purpose of the paper is to reveal the spatiotemporal dynamics and inequality of heatwave exposure among China’s urban elderly and to disentangle the driving influences of climate change, ageing, and urbanization. Historical and future heatwaves across China are identified and analyzed, exposure inequality is evaluated using the Gini coefficient, and the relative contributions of key drivers are quantified through factor separation. Results showed that heatwave frequency and duration increased from 2000 to 2019, with high-risk provinces clustering in the Yangtze River Basin, North China Plain, and Sichuan Basin. Future projections indicate substantial growth in elderly exposure to heatwaves, while under the SSP3-70 scenario, inter-provincial inequality in exposure tends to alleviate rather than intensify. Climate change was identified as the dominant driver, while ageing amplified risks and urbanization partly mitigated growth. These findings highlighted the urgent need for place-based adaptation and health protection strategies, aligned with climate mitigation, demographic transition, and sustainable urban planning.

1. Introduction

Global warming has significantly increased the frequency and duration of heatwaves [1,2,3,4], contributing to rising rates of heat-related diseases and mortality [5,6]. Extreme heat has become a major challenge to human health and sustainable development [7,8]. According to the 2025 Lancet Countdown report, rising temperatures combined with increasing population vulnerability have resulted in a 63% increase in global heat-related mortality since the 1990s, reaching an estimated 546,000 deaths per year on average during 2012–2021 [9]. Historical heatwave events further illustrate this hazard. The 2003 European heatwave, one of the most severe climatic disasters in recent history, resulted in more than 70,000 excess deaths across Europe [10]. Subsequent analyses confirmed that anthropogenic warming greatly increased the likelihood and intensity of this event [11]. Similarly, the 2010 Russian (Siberian) heatwave, characterized by persistent high-temperature anomalies and widespread wildfires, led to approximately 55,000 fatalities, with air pollution compounding its health impacts [12]. Such staggering human losses highlight the profound and escalating global health burden imposed by extreme heat.

Meanwhile, the population of the elderly is accelerating worldwide, driven by increasing life expectancy and declining fertility [13,14,15]. Heatwaves are typically characterized as episodes of unusually high temperatures lasting for several consecutive days [16]. Such events markedly increase the risks of cardiovascular and cerebrovascular diseases, and elderly individuals as well as those with chronic conditions, including diabetes and hypertension, are particularly susceptible owing to impaired thermoregulation and reduced adaptive capacity [17,18]. Recent studies indicated that, compared with the 1986–2005 average, people aged 65 years and older and infants under one year experienced 3.7 billion more heatwave days in 2021, while heat-related deaths increased by 68% between 2000–2004 and 2017–2021 [19]. Furthermore, the mortality risk of the elderly during heatwaves was significantly higher than that of other groups [20]. Against the dual pressures of global warming and population ageing, there is an urgent need to refine assessments of elderly heat-related health risks and to provide scientific evidence for targeted health interventions and risk management strategies.

Extreme heatwaves, as critical indicators of climate change, are attracting increasing attention for their impacts on human health [21,22]. In health risk assessments, population exposure serves as a key mediator linking climatic factors with health outcomes and has become a central point for understanding mechanisms of heat-related risks [23]. Although growing evidence highlights the impacts of heatwaves on elderly health [24,25,26], systematic studies in China remain limited, particularly in the context of rapid population ageing. As the world’s largest developing country, China faces both a massive population base and an accelerating ageing process [27,28]. According to the latest data from the National Bureau of Statistics of China (NBSC), by the end of 2024 the population aged 65 and above had reached nearly 220 million, accounting for 15.6% of the total and far exceeding the UN threshold for an “Aged Society” [29]. By 2050, this share is estimated to rise to 25%, with the elderly population (aged 65 and above) expected to exceed 330 million [30]. Notably, the accelerated urbanization process is interacting with ageing, intensifying the spatiotemporal heterogeneity of elderly heatwave exposure through mechanisms such as urban heat island effects and increasing residential density [31,32,33,34]. The expansion of impervious surfaces has amplified surface heat storage, while the reduction of green space and ventilation corridors has weakened urban heat dissipation, creating local heat islands [35]. At the same time, the elderly are disproportionately concentrated in densely built areas, where restricted public space and insufficient age-friendly infrastructure are inadequate to meet their cooling demands [36,37,38]. These challenges are even more pronounced in developing countries, where rapid urban growth often coincides with lagging infrastructure, and old residential communities lack heat-adaptive design, thereby intertwining thermal exposure with health risks and straining medical resources [39]. Therefore, it is essential to analyze systematically the multidimensional drivers of climate change, population ageing, and urban expansion to reveal the spatiotemporal dynamics of elderly heatwave exposure.

Developing effective management measures for adapting to heatwaves in China requires the construction of future environmental scenarios that account for potential regional changes across multiple dimensions. Therefore, referring to the Shared Socioeconomic Pathways (SSPs) and Representative Concentration Pathways (RCPs), which are widely applied in climate change research, provides a solid scientific basis [40,41,42]. SSPs describe alternative global socioeconomic development trajectories until 2100, reflecting the impacts of climate change and related policies on socioeconomic structures, while RCPs depict possible pathways of atmospheric greenhouse gas concentrations and associated radiative forcing. With advances in climate modeling, the Coupled Model Intercomparison Project Phase 6 (CMIP6), which incorporates SSP-RCP combined scenarios, has been widely applied to project future heatwave trends and assess their potential impacts on health and socioeconomic systems [43,44,45]. Studies have shown that under anthropogenic climate change, heatwaves will become more frequent and intense in the coming decades [46]. Wang et al. used CMIP6 simulations to assess systematically global heatwaves and their socioeconomic risks, identifying a strong correlation between economic growth and heatwave exposure [47]. Similarly, Tejedor et al. evaluated the influence of anthropogenic climate change on the frequency of extreme heatwave events in the western Mediterranean, finding that summer temperatures in 2022 and 2023 exceeded the range of natural variability over the past millennium, with unprecedented anomalies of 3.6 °C and 2.9 °C, respectively [48].

Despite progress in revealing changes in heatwave indices and future trends, quantitative analyses of the contribution and mechanisms of underlying drivers remain insufficient. In China, most existing studies focus on spatial distribution patterns of heatwaves or estimates of population exposure, while systematic evaluations of the multiple mechanisms driving dynamic changes of the elderly exposure during rapid urbanization are lacking. This limitation hinders a deeper understanding of the future evolution of heatwave risks and weakens the basis for developing evidence-based and targeted risk management strategies. Therefore, a systematic analysis of the spatiotemporal dynamics of elderly heatwave exposure in urban China under different climate scenarios, and the identification of key influencing factors, is of great theoretical and policy significance.

The objective of this paper is to reveal the spatiotemporal dynamics and inequality of elderly exposure to heatwaves in the context of China’s urbanization and to disentangle the underlying driving mechanisms. To this end, heatwave events were identified and their historical (2000–2019) and projected (2020–2099) spatiotemporal patterns were analyzed across China. The dynamic evolution of exposure risks for the urban elderly was examined under the Shared Socioeconomic Pathway framework, and exposure inequality among provinces was evaluated using the Gini coefficient. The relative contributions of climate change, population ageing, and urbanization were further quantified through factor separation. This study highlights pronounced spatial disparities in heatwave exposure of the elderly and provides insights into the evolving nature of heat-related health risks under concurrent climate and demographic changes.

2. Materials and Methods

2.1. Study Area

This study focuses on China (Figure 1). Once the world’s most populous country and currently experiencing an accelerating ageing process, China has undergone significant warming and increasingly frequent extreme heatwaves over the past few decades, while simultaneously facing a stabilizing urbanization process and intensified demographic transitions [49]. The combined effects of the urban heat island phenomenon and demographic ageing have made elderly populations increasingly vulnerable to health risks under heatwave conditions, with relevant spatiotemporal patterns of climate, ageing, and urbanization illustrated in Figure A1. In the coming decades, China will continue to advance new-type urbanization and pursue carbon neutrality strategies, leading to an even more complex interplay between demographic and climatic factors [50]. Therefore, selecting China as the study area not only facilitates the examination of spatiotemporal dynamics of elderly urban heatwave exposure under different climate scenarios but also provides a scientific basis for addressing heat-related health risks and for strengthening the adaptation capacity of the urban elderly population.

2.2. Research Design

This study consists of four main steps (Figure 2): data collection and processing, analysis of the spatiotemporal characteristics of historical and future heatwaves, assessment of future population exposure to heatwaves, and evaluation of the drivers of elderly heatwave exposure.

2.3. Data Collection

Historical meteorological data were obtained from the ERA5 global climate reanalysis product, developed by the European Centre for Medium-Range Weather Forecasts (ECMWF) for the Copernicus Climate Change Service, covering the period from 2000 to 2019. For future scenarios, temperature data were sourced from the HiCPC dataset (High-resolution CMIP6 downscaled daily Climate Projections over China) provided by the National Tibetan Plateau Data Center of China (https://www.scidb.cn/detail?dataSetId=73c1ddbd79e54638bd0ca2a6bd48e3ff, accessed on 25 August 2025) [51]. This dataset was derived from CMIP6 global model outputs through downscaling and bias correction. The study employed daily maximum near-surface air temperature (Tmax, K) derived from the MRI-ESM2-0 model, developed by the Meteorological Research Institute of Japan. Previous studies have demonstrated that this model can effectively simulate East Asian monsoon processes [52,53,54], and therefore, it was considered highly reliable in reproducing key climatic features of China. All meteorological data were resampled to a uniform spatial resolution of 1 km and used to construct Tmax series for 2020–2099. Although reanalysis products are available up to the early 2020s, all datasets in this study were organized within a consistent framework by combining reanalysis-based historical data (2000–2019) with model-based projections (2020–2099). To ensure temporal consistency across the entire analysis period, only ERA5 data from 2000 to 2019 were used to represent the historical baseline, whereas data for 2020–2024 were obtained from model simulations. Comparative evaluations indicate that differences between the simulated data for 2020–2024 and the corresponding ERA5 reanalysis outputs are minimal, thereby not affecting the robustness of the analysis. Data processing was conducted using ArcGIS Pro (version 3.4) and its built-in Arcpy package.

Data on the elderly population, required for exposure measurement, were obtained from NBSC (https://www.stats.gov.cn/sj/ndsj/, accessed on 25 August 2025). Future population projections were derived from gridded population datasets constructed under the SSPs framework. Based on localized population and economic parameters, a Population Development Environment (PDE) model was employed to generate gridded SSP population data for 2020–2099 [55]. These datasets are consistent with the ScenarioMIP core experiment (Tier-1) within the CMIP6 framework [56]. In the present study, four typical future scenarios—SSP1-26, SSP2-45, SSP3-70, and SSP5-85—were selected. These correspond to the low, medium, medium-to-high, and high ends of the future forcing pathways, respectively, and update the original RCP2.6, RCP4.5, RCP6.0, and RCP8.5 pathways. The dataset provides globally gridded population distributions with a spatial resolution of 0.5°, offering essential support for assessing the long-term evolution of population size and structure in China under various socioeconomic and climate change combinations. The elderly population data originally had a five-year temporal resolution. A bilinear interpolation method was therefore applied to generate annual estimates, and all datasets were reprojected and resampled to a uniform spatial resolution of 1 km to meet the requirements of fine-scale analysis.

At the grid-cell level, historical urbanization proportions were derived from the HYBMAP dataset (Global hybrid land-cover maps from 2000 to 2020 with a resolution of approximately 1 km) [57]. This dataset includes 17 land-use and land-cover types such as Grasslands, Permanent Wetlands, Croplands, Urban and Built-up Lands, Barren, and Water Bodies. The Urban and Built-up Lands category was extracted through binarization to determine the urbanization proportion of each grid cell. For the future period, the harmonized land-use dataset from the annual Global Carbon Budget (LUH2) [58] was employed. Values from the Urban land band in the NetCDF files were directly read and used to calculate the proportion of urbanization at each grid cell.

This study focused on 31 mainland province-level administrative units—provinces, autonomous regions, and municipalities directly under the central government—hereafter collectively referred to as “provinces”. The four municipalities (Beijing, Shanghai, Tianjin, and Chongqing), although named as cities, are administratively equivalent to provinces and therefore included. The province-level scale served as the primary unit of analysis, while national summaries and aggregations at broader regional or basin scales were occasionally employed to facilitate interpretation.

2.4. Definition of Heatwave Indices and Urban Elderly Population Heatwave Exposure

A heatwave is generally understood as an abnormal high-temperature event lasting for several consecutive days, although there is no single accepted definition [16]. In the present study, a heatwave event is defined at each grid cell when daily Tmax exceeds its relative threshold for at least three consecutive days [59,60,61,62,63]. The relative threshold is defined as the 90th percentile of the daily Tmax distribution at each grid cell during the period 2000–2019. Two heatwave indices are introduced to characterize the spatiotemporal properties of heatwave events: heatwave frequency (HWF) and heatwave duration (HWD). HWF is defined as the total number of heatwave events occurring within a year, while HWD represents the average duration of heatwaves within a year.

Exposure of the elderly population to heatwaves is typically determined by the number of elderly who are affected by heatwave events in urban settings. This exposure metric is computed by multiplying the heatwave index (including HWF and HWD) by the elderly population and the proportion of urbanized grids.

$${\mathrm{Exp}}_{\mathrm{elderly}(\mathrm{i})}^{\mathrm{T}}={\mathrm{P}}_{\mathrm{elderly}(\mathrm{i})}^{\mathrm{T}}\times {\mathrm{H}}_{(\mathrm{i})}^{\mathrm{T}}$$

(1)

where ${\mathrm{Exp}}_{\mathrm{elderly}(\mathrm{i})}^{\mathrm{T}}$ represents the elderly population’s exposure to heatwaves at the ith grid in year T; ${\mathrm{P}}_{\mathrm{elderly}(\mathrm{i})}^{\mathrm{T}}$ represents the elderly population at the ith grid in year T; ${\mathrm{H}}_{(\mathrm{i})}^{\mathrm{T}}$ is the heatwave index at the ith grid in year T.

2.5. Modified Mann-Kendall Test

The Modified Mann-Kendall (MMK) test was employed to detect long-term monotonic trends in the studied variables [64]. Unlike the classical Mann-Kendall test, which may overestimate trend significance when data exhibit autocorrelation, the MMK test adjusted the variance of the statistic to account for serial dependence [64,65]. This modification enhanced its reliability and made it particularly suitable for climate datasets, where persistence was common. In this research, the MMK method was employed separately to detect monotonic trends in the spatiotemporal variation of heatwave indices (including HWF and HWD), ensuring robustness in the preliminary trend analyses.

2.6. Inequality in Elderly Heatwave Exposure

The Gini index is a commonly used metric to quantify inequality across different regions [66,67,68,69]. In this study, the Gini index was applied to measure the spatial inequality of elderly population exposure to heatwaves across provinces in China under different SSP scenarios and years. For each year from 2020 to 2099, the provincial mean exposure values were first calculated by overlaying the provincial boundaries with the gridded exposure dataset. Then, the Gini index was derived based on these provincial mean values according to the following formula:

$$\mathrm{Gini}={\displaystyle \frac{{\displaystyle {\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\displaystyle {\sum }_{\mathrm{j}=1}^{\mathrm{i}}|{\mathrm{e}}_{\mathrm{i}} - {\mathrm{e}}_{\mathrm{j}}|}}}{2\mathrm{n}\times {\displaystyle {\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{e}}_{\mathrm{i}}}}}$$

(2)

where ${\mathrm{e}}_{\mathrm{i}}$ is the average exposure value of the ith province, and $\mathrm{n}$ is the total number of provinces. The Gini index ranges from 0 (absolute equality) to 1 (absolute inequality). A lower Gini index indicates that elderly exposure to heatwaves is more evenly distributed across provinces, whereas a higher Gini index indicates larger spatial disparities. For analytical purposes, the Gini index is further classified into five categories: very low (0.0–0.2), low (0.2–0.4), medium (0.4–0.6), high (0.6–0.8), and very high inequality (0.8–1.0).

To capture long-term inequality trends, exposure values were aggregated into four time periods (2020–2039, 2040–2059, 2060–2079, and 2080–2099), and the corresponding Gini indices were calculated based on the multi-year mean exposure values for each province.

2.7. Contribution of Drivers to Exposure Estimates

Following the factor separation method of Stein and Alpert [70], the changes in heatwave exposure of the elderly population are decomposed into the effects of ageing population, climate change, urban expansion, and their interactions. To estimate the contributions of these driving factors quantitatively, a control scenario (Control) is first established, representing baseline climate conditions, population size, and urbanization level. On this basis, different experimental scenarios are constructed by replacing individual or combined factors (

Table A1. Description of conventions used in calculating contributions to the exposure of each driver.

Name	Description
Expcontrol	2020–2029 extreme heatwave events with 2029 urban area and 2029 elderly population
Exppop	2020–2029 extreme heatwave events with 2029 urban area and 2099 elderly population
Expclim	2090–2099 extreme heatwave events with 2029 urban area and 2029 elderly population
Expurban	2020–2029 extreme heatwave events with 2099 urban area and 2029 elderly population
Exppop+clim	2090–2099 extreme heatwave events with 2029 urban area and 2099 elderly population
Exppop+urban	2020–2029 extreme heatwave events with 2099 urban area and 2099 elderly population
Expclim+urban	2090–2099 extreme heatwave events with 2099 urban area and 2029 elderly population
Exppop+clim+urban	2090–2099 extreme heatwave events with 2099 urban area and 2099 elderly population

). The independent and interactive effects of each driver, as defined in

Table A2. Description of drivers of heatwave population exposure.

Driver	Description
Population ageing effect	The direct effect of population ageing from 2020 to 2099 levels with other variables held constant.
Climate effect	The direct effect of climate warming from 2020 to 2099 levels with other variables held constant.
Urbanization effect	The direct effect of increasing urbanization expansion from 2020 to 2099 levels with other variables held constant.
Population ageing and climate interaction effect	The interaction effect that occurred when climate warming and population ageing are simultaneously increased from 2020 to 2099 levels.
Population ageing and urbanization interaction effect	The interaction effect that occurred when population ageing and urbanization are simultaneously increased from 2020 to 2099 levels.
Climate and urbanization interaction effect	The interaction effect that occurred when climate warming and urbanization are simultaneously increased from 2020 to 2099 levels.
Population ageing and climate and urbanization	The interaction effect that occurred when all three variables are simultaneously increased from 2020 to 2099 levels.

, are then calculated using the following formulae.

$${\mathrm{Effect}}_{\mathrm{Population} \mathrm{ageing}}={\mathrm{Exp}}_{\mathrm{pop}} - {\mathrm{Exp}}_{\mathrm{Control}}$$

(3)

$${\mathrm{Effect}}_{\mathrm{climate}}{=\mathrm{Exp}}_{\mathrm{clim}} - {\mathrm{Exp}}_{\mathrm{Control}}$$

(4)

$${\mathrm{Effect}}_{\mathrm{urbanization}}{=\mathrm{Exp}}_{\mathrm{urban}} - {\mathrm{Exp}}_{\mathrm{Control}}$$

(5)

$${\mathrm{Effect}}_{\mathrm{Population} \mathrm{ageing}\&\mathrm{climate}}{=\mathrm{Exp}}_{\mathrm{pop}+\mathrm{clim}} - \left({\mathrm{Exp}}_{\mathrm{pop}}{+\mathrm{Exp}}_{\mathrm{clim}}\right){+\mathrm{Exp}}_{\mathrm{Control}}$$

(6)

$${\mathrm{Effect}}_{\mathrm{Population} \mathrm{ageing}\&\mathrm{urbanization}}{=\mathrm{Exp}}_{\mathrm{pop}+\mathrm{urban}} - \left({\mathrm{Exp}}_{\mathrm{pop}}{+\mathrm{Exp}}_{\mathrm{urban}}\right){+\mathrm{Exp}}_{\mathrm{Control}}$$

(7)

$${\mathrm{Effect}}_{\mathrm{climate}\&\mathrm{urbanization}}{=\mathrm{Exp}}_{\mathrm{clim}+\mathrm{urban}} - \left({\mathrm{Exp}}_{\mathrm{clim}}{+\mathrm{Exp}}_{\mathrm{urban}}\right){+\mathrm{Exp}}_{\mathrm{Control}}$$

(8)

$$\begin{array}{l}{\mathrm{Effect}}_{\mathrm{Population} \mathrm{ageing}\&\mathrm{climate}\&\mathrm{urbanization}}{=\mathrm{Exp}}_{\mathrm{pop}+\mathrm{clim}+\mathrm{urban}}\\ - \left({\mathrm{Exp}}_{\mathrm{pop}+\mathrm{clim}}{+\mathrm{Exp}}_{\mathrm{pop}+\mathrm{urban}}{+\mathrm{Exp}}_{\mathrm{clim}+\mathrm{urban}}\right)+\left({\mathrm{Exp}}_{\mathrm{pop}}{+\mathrm{Exp}}_{\mathrm{clim}}{+\mathrm{Exp}}_{\mathrm{urban}}\right){-\mathrm{Exp}}_{\mathrm{Control}}\end{array}$$

(9)

where ${\mathrm{Exp}}_{\mathrm{Control}}$ denotes the baseline exposure; ${\mathrm{Exp}}_{\mathrm{pop}}$, ${\mathrm{Exp}}_{\mathrm{clim}}$, ${\mathrm{Exp}}_{\mathrm{urban}}$ represent the exposures driven solely by population, climate, and urbanization, respectively (C1, C2, C3); ${\mathrm{Exp}}_{\mathrm{pop}+\mathrm{clim}}$, ${\mathrm{Exp}}_{\mathrm{pop}+\mathrm{urban}}$, ${\mathrm{Exp}}_{\mathrm{clim}+\mathrm{urban}}$ represent exposures under pairwise combinations of drivers (C4, C5, C6); and ${\mathrm{Exp}}_{\mathrm{pop}+\mathrm{clim}+\mathrm{urban}}$ represents the exposure under the combined influence of all three drivers (C7).

3. Results

3.1. Historical Heatwave Events

The spatial and temporal distributions and changing trends of HWF and HWD in China during 2000–2019 are illustrated in Figure 3. Overall, heatwaves occurred widely across the country but showed clear regional differences. Specifically, HWF was generally higher in both northern and southern China, whereas markedly lower values were observed in Southwest China, the Tibetan Plateau, and the areas around the Yangtze River region, where the annual averages were mostly below 4 heatwaves (Figure 3a). In contrast, HWD showed markedly higher values in the western Tibetan Plateau and the Yangtze River Basin of the heartland of China, with average durations locally exceeding 6.7 days, whereas relatively low values were observed in northern and southern China (Figure 3b). In terms of trends, both HWF and HWD generally increased across most regions of China during 2000–2019 (Figure 3c,d). The increase in HWF was mainly concentrated in the North China Plain, the eastern coastal region, and parts of Southwest China, with annual growth close to 0.21 heatwaves. The increase in HWD was more substantial in western Xinjiang, northeast Hunan, southeast Hubei, and northern Jiangxi, where the annual growth rate exceeded 0.53 days. Time-series analysis further supported these findings: although the annual mean HWF fluctuated, its overall change was not significant (slope = 0.03 heatwaves·yr⁻¹, p > 0.30), whereas the annual mean HWD showed a significant upward trend (slope = 0.06 days·yr⁻¹, p < 0.01). This indicated that over the past two decades, the intensification of heatwaves in China was mainly due to prolonged duration (Figure 3e,f).

3.2. Future Heatwave Events

The projected trajectories of Tmax, elderly population size, and urbanization rates under different SSPs are presented in Figure A2, providing the socioeconomic and climatic background for subsequent analyses. Building on this, the long-term evolution and spatial distributions of both HWF and HWD across SSP pathways are illustrated in Figure A3, serving as the basis for the detailed provincial-level assessments that followed.

Figure 4 shows the evolution of HWF in China from 2020 to 2099 under different SSP-RCP scenarios. Overall, HWF exhibited a pronounced upward trend in the future, becoming more evident over time and with the intensity of emission scenarios (Figure 4a). At the national average level, the differences in HWF among the four scenarios were small during 2020–2039, with an average of approximately 5–6 heatwaves per year. However, as climate warming intensified by the mid-21st century, the gaps among scenarios gradually widened. By the end of the century, HWF under the low-emission scenario SSP1-26 increased only slightly to roughly 5.5–6.5 heatwaves per year, whereas under the high-emission scenario SSP5-85 mean value exceeded 7.5 heatwaves, nearly 26% higher than in the baseline period, with a growth rate much faster than in the other scenarios. The boxplot results (Figure 4b) further illustrate the differences of HWF among scenarios and periods within China, rather than spatial variations. The significance of the long-term evolution of national-scale HWF is assessed using the MMK trend test applied to the annual mean HWF series (2020–2099) for each emission scenario. As shown in

Table A3. MMK test results for heatwave indices (HWF, HWD) and population exposure in China under different SSP-RCP scenarios.

Driver	Z_Score	p_Value	Sen’s	Significance	Trend Direction
MMK test for HWF in China under SSP1-26	0.83	0.408	200	Not significant	No trend
MMK test for HWF in China under SSP2-45	7.11	0.000 *	1712	Highly significant	Upward
MMK test for HWF in China under SSP3-70	9.56	0.000 *	2302	Highly significant	Upward
MMK test for HWF in China under SSP585	10.16	0.000 *	2446	Highly significant	Upward
MMK test for HWD in China under SSP126	2.35	0.019	566	Significant	Upward
MMK test for HWD in China under SSP245	7.72	0.000 *	1858	Highly significant	Upward
MMK test for HWD in China under SSP370	9.33	0.000 *	2246	Highly significant	Upward
MMK test for HWD in China under SSP585	10.23	0.000 *	2464	Highly significant	Upward
MMK test for Population exposure to HWF in China under SSP126	3.16	0.002	762	Significant	Upward
MMK test for Population exposure to HWF in China under SSP245	6.10	0.000 *	1470	Highly significant	Upward
MMK test for Population exposure to HWF in China under SSP370	8.25	0.000 *	1986	Highly significant	Upward
MMK test for Population exposure to HWF in China under SSP585	7.66	0.000 *	1844	Highly significant	Upward
MMK test for Population exposure to HWD in China under SSP126	3.43	0.001	826	Significant	Upward
MMK test for Population exposure to HWD in China under SSP245	5.49	0.000 *	1322	Highly significant	Upward
MMK test for Population exposure to HWD in China under SSP370	6.22	0.000 *	1498	Highly significant	Upward
MMK test for Population exposure to HWD in China under SSP585	9.23	0.000 *	2222	Highly significant	Upward

Notes: Z is the standardized Mann-Kendall statistic, p is the significance level, and Sen’s slope indicates the median rate of change. Significance levels are denoted as: p < 0.05 (significant), p < 0.01 (*, highly significant). Positive Sen’s slope values indicate upward trends, while negative values indicate downward trends.

, HWF shows no significant trend in the low-emission scenario (SSP1-26, p = 0.408), but displays significant upward trends in higher-emission scenarios. Meanwhile, HWD and population exposure exhibit consistent and increasingly strong upward trends across all scenarios. The spatial distribution of HWF at the provincial level (Figure 4c) showed marked regional clustering. Under the SSP3-70 and SSP5-85 scenarios, pronounced increases were concentrated in eastern and southern provinces, such as Jiangsu, Zhejiang, Anhui, Guangdong, and Guangxi, where the annual mean HWF exceeded 7.49 heatwaves by the end of the 21st Century, including areas surrounding the Yangtze River. By contrast, provinces in Northwest China and on the Tibetan Plateau maintained relatively low HWF values under the SSP1-26 scenario, though still exhibited a gradual upward trend. Overall, the provincial pattern revealed a clear southeast–northwest gradient, which became increasingly distinct over time.

The projected changes in the spatiotemporal distribution of HWD in China based on province-level data are depicted in Figure 5. Consistent with HWF, HWD showed a significant upward trend in all scenarios and periods (Table A3), with the increase becoming more pronounced over time and under higher emission scenarios (Figure 5a). During 2020–2039, the differences among the four scenarios are minor, with averages remaining around 7 days. However, the divergence among scenarios gradually widened in the mid and late 21st century. During 2080–2099, the national average HWD reached about 6 days under the SSP1-26 scenario, whereas it approached 8.6 days under the SSP5-85 scenario, representing an approximately 43% increase, which is substantially greater than that under the low-emission scenario. The boxplot results (Figure 5b) show clear variations of HWD across different emission scenarios and time periods, with larger differences under high-emission scenarios and in later periods. In some provincial units (Hainan Province, Taiwan Province, and the Tibet Autonomous Region), extreme values exceeded 15 days, indicating longer durations of heatwave events. The spatial distribution (Figure 5c) indicates that, under the SSP5-85 scenario, the most pronounced increases in HWD occur in southern and southwestern China, particularly in Sichuan, Yunnan, and Guizhou, as well as in the middle and lower reaches of the Yangtze River, including Hubei, Hunan, Jiangxi, and Anhui. Under this high-emission pathway, HWD values in many southwestern provinces rise sharply, with some regions reaching nearly 16 days during 2080–2099.

In Figure 4 and Figure 5, highly consistent upward trends in both HWF and HWD across temporal and spatial scales are shown, with marked scenario dependence. At the national scale, increases in HWF and HWD under SSP1-26 were relatively modest and more manageable, while under SSP5-85 the increases are the greatest, with national averages by the period 2080–2099 rising by approximately 46% and 66%, respectively, compared with those in 2020–2039. These results highlight a substantial intensification and persistence of heatwave activity across China under higher-emission scenarios.

3.3. Future Urban Elderly Population Exposure to Heatwaves

The mean exposure of the urban elderly population to HWF in China from 2020 to 2099 is illustrated in Figure 6. Nationally (Figure 6a), exposure increases under all emission scenarios, with the largest magnitude and most rapid growth occurring under SSP5-85, which peaks during 2060–2079 and slightly declines thereafter while remaining substantially higher than in low-emission scenarios. In contrast, the trajectories of SSP1-26, SSP2-45, and SSP3-70 show relatively similar but more moderate increases. The boxplot (Figure 6b) reveals considerable inter-provincial variability, which becomes increasingly pronounced under higher-emission scenarios as several provinces exhibit exceptionally high exposure values. Spatially (Figure 6c), HWF exposure shows a clear southeast-to-northwest gradient, with the southeastern coastal and middle-lower Yangtze River regions (e.g., Jiangsu, Shanghai, Zhejiang, Anhui, Jiangxi, Hubei, Hunan, Guangdong, and Guangxi) consistently exhibiting the highest exposure levels. In contrast, exposure remains relatively low in northwestern China and on the Tibetan Plateau, though these areas also experience a gradual increase over time.

Figure 7 further reveals the mean exposure of the urban elderly population to HWD. Overall, the temporal evolution of HWD exposure is broadly consistent with that of HWF but still exhibits several distinct differences. Nationally (Figure 7a), mean HWD exposure generally increases across all scenarios, reflecting a continued lengthening of heatwave duration in the future. However, under SSP2-45 and SSP5-85, exposure peaks during 2060–2079 and declines slightly during 2080–2099, which may be associated with the combined effects of population dynamics and spatial exposure patterns. At the provincial level (Figure 7b), the boxplots reveal substantial disparities among provinces, with a right-skewed distribution in which some provinces exhibit notable deviations from the national mean. Over time, the median exposure values in all scenarios gradually increase, indicating sustained growth in elderly population exposure across most regions, albeit with clear provincial differences in magnitude. Spatially (Figure 7c), high-exposure zones expand nationwide, most prominently across eastern and central China, highlighting a further spatial enlargement of long-duration heatwave risks in the future.

Across both HWF and HWD, the exposure of the urban elderly population in China generally increases over the twenty-first century, although several scenarios show a slight decline in the late period, reflecting differences among emission pathways. Under high-emission pathways(SSP5-85), exposure increases rapidly in the early decades but shows signs of stabilization or slight decline after mid-century, whereas under low-emission scenarios, the growth remains weaker yet more gradual. Figure A4 provides additional details on the year-to-year variations in heatwave exposure of the urban elderly population under different SSP scenarios, serving as a supplement to the main results. Figure A5 and Figure A6 further illustrate these temporal and spatial variation patterns, highlighting the changing dynamics and regional heterogeneity of heatwave exposure across scenarios and periods. These findings indicate that the combined effects of population ageing and climate warming are more pronounced in the near and mid-term but tend to be alleviated toward the end of the century.

3.4. Inequality in Heatwave Exposure in Provinces of China

The Gini index of heatwave exposure among the elderly population across Chinese provinces under different scenarios is shown in Figure 8. The results indicate that exposure differences among provinces remain generally large, with Gini index values ranging from 0.78 to 0.89, representing high to very high levels of inequality. The Gini index for HWF is consistently lower than that for HWD and gradually declines over time, from about 0.85 to around 0.82, suggesting a slight weakening in province-level disparities of frequency-related exposure. In contrast, the Gini index for HWD remains persistently very high between 0.87 and 0.89, with very limited changes, indicating that inequalities in duration-related exposure stay pronounced without noticeable improvement. Marked differences are also observed among emission scenarios. Under SSP3-70, both HWF and HWD Gini indices display continuous declines, with interprovincial inequality easing steadily over time. Under SSP5-85, the indices are relatively higher, reflecting more uneven distributions of exposure. These findings highlight that the distribution of exposure across provinces is characterized by significant inequality for both HWF and HWD, with disparities in HWD particularly pronounced.

3.5. Drivers of Heatwave Exposure

Based on the independent effects of elderly population, climate, and urbanization rate, together with their interactions, the contributions of seven driving factors to the urban elderly heatwave exposure are quantified (Figure 9). C1 (population ageing) consistently dominates across all scenarios, accounting for roughly 30–50% of total changes and representing the main source of exposure increases. Under SSP1-26, it contributes about 9000 person·yr⁻¹ to HWF, while under SSP5-85, its impact rises to more than 12,000 person·yr⁻¹. The influence of C2 (climate effect) becomes more evident under medium- to high-emission scenarios, and its contribution exceeds 40% for both HWF and HWD in SSP3-70. In contrast, C3 (urbanization) generally exhibits negative values, indicating that urbanization alone reduces net exposure slightly. Among interaction terms, C4 (population ageing and climate) makes stable and considerable positive contributions, ranking just behind the independent effects of population and climate. C5 (population ageing and urbanization) is mostly negative, while C6 and C7 have minor influences. Overall, population ageing was the dominant factor in low- and high-emission scenarios, whereas climate effects are equally important under the medium-emission pathway (SSP3-70). Interaction effects play supplementary but non-negligible roles in shaping exposure patterns across the provinces.

4. Discussion

4.1. Spatial Heterogeneity of Heatwave Dynamics and Elderly Population Exposure

Comparisons between historical and future results demonstrated that both the frequency and duration of heatwaves in China increased significantly, with their spatial patterns exhibiting strong clustering. Provinces in the middle and lower reaches of the Yangtze River, the North China Plain, and southwestern China consistently emerged as high-risk regions, a pattern closely tied to the background of climate dynamics and regional topography. Previous studies have suggested that global warming, coupled with enhanced variability of the East Asian summer monsoon, will make eastern and southern China more prone to the control of subtropical high-pressure systems, thereby leading to more frequent and persistent heatwave events [71,72,73]. At the same time, the enclosed topography of the Sichuan Basin and the humid environment of the middle-lower Yangtze River Basin favor the accumulation and amplification of heat, substantially increasing the probability of heatwave occurrence [74]. The spatial patterns identified in this study are highly consistent with recent CMIP6-based modeling results, which also highlight eastern and central China as global hotspots for heatwave risk [75,76]. This convergence not only confirms the robustness of the conclusions but also emphasizes the critical role of joint influences from climate dynamics and topographic conditions.

Nevertheless, the spatial heterogeneity of heatwaves is not shaped solely by atmospheric and climatic processes. Population structures and urbanization trajectories act as equally important drivers. Growing evidence shows that the intensity of the urban heat island effect is largely modulated by both regional climatic conditions and population size, and that high population densities and rapidly expanding urban morphologies tend to magnify disparities in extreme heat exposure [77]. Within this context, population ageing adds further complexity. Several eastern and southern provinces with dense populations and advanced urbanization levels are subject to sustained strong heat island effects while also accommodating large elderly populations, resulting in persistently high exposure risks [33,78,79]. By contrast, some inland provinces with basin-like topography and high population densities illustrate another mechanism: restricted air circulation, combined with dense settlement patterns, aggravates heat retention and exposure risks for older residents [80,81]. In addition, rapid urbanization reduces the availability of green spaces and wind corridors and increases surface imperviousness, thereby intensifying both the persistence and the spatial imbalance of urban heat islands [35,82]. Taken together, these findings suggest that the spatial heterogeneity of elderly heatwave exposure is the outcome of multiple interacting processes, including climate dynamics, demographic changes, and urban spatial configurations. Such an understanding underscores the need for province-specific adaptation priorities in public health planning across diverse urban contexts, reflecting both regional climatic settings and socio-demographic characteristics.

4.2. Inequality of Exposure and Driving Mechanisms

The Gini index results indicate that, under all future climate and development pathways, the exposure of the elderly population to heatwaves in China remains at a relatively high level of inequality, with spatial disparities in HWD exposure consistently larger than those in HWF. Provinces along the eastern coast and in the central-eastern parts of the Yangtze River Plain generally exhibit higher exposure levels due to both the concentration of elderly populations and hot-humid climatic conditions. In contrast, many western and northeastern provinces experience lower exposures, either because of relatively small population bases or milder climates. This pattern aligns well with previous studies on the uneven spatial distribution of vulnerable groups facing heatwave exposure [23,83,84], suggesting that future health risks from extreme heat are not only characterized by nationwide increases but also by pronounced spatial concentration [85]. The factor separation analysis further reveals the underlying mechanisms: population ageing consistently emerges as the dominant driver, accounting for nearly half of the increases, while climate change also plays a substantial but secondary role. The influence of climate change varies more substantially across scenarios—its contribution becomes markedly stronger under SSP3-70, where intensified warming leads to an effect nearly comparable to that of population ageing. Among the interaction factors, the combined effect of population ageing, climate, and urbanization (C7) is generally small across most scenarios, becoming slightly more pronounced under SSP5-85. In contrast, the interaction effect between population ageing and climate (C4) exerts a stronger influence in most cases [86,87]. Although population change consistently played the leading role in driving future heatwave exposure of the elderly population, the results also indicated that differences among provinces would persist and could not be resolved through demographic change alone. The frequent occurrence of heatwaves and the complex impacts of urban spatial development could either amplify or partly offset the influence of population in some scenarios. Therefore, future governance should emphasize coordinated strategies at the provincial level, while also prioritizing sustained global climate mitigation and carefully guiding urbanization, in order to prevent rapid and uncontrollable increases in elderly exposure during extreme heatwaves.

The future trajectories of exposure inequality diverge under different development pathways. Under SSP1-26, although the overall exposure burden remains relatively low, the Gini index shows a gradual increase and reaches a higher level by the end of the century. This suggests that low-emission pathways might paradoxically intensify the concentration of risks, as the growth of heatwaves was more pronounced in densely populated provinces of eastern and central China, thereby widening the gap with western provinces [88,89]. By contrast, SSP3-70 is characterized by a rapid rise in total exposure but the most pronounced alleviation of inequality. The Gini index drops steadily from 0.84 to 0.78 approximately, ultimately becoming lower than in most other scenarios by 2099. This trend is closely related to the spatial expansion of extreme heatwaves under high emissions, which increases exposure even in previously low-risk western and northeastern provinces, thereby narrowing interprovincial disparities [90]. SSP2-45 and SSP5-85 showed limited change, with the Gini index fluctuating narrowly between 0.83 and 0.84, without significant improvements or deteriorations. These results indicate that, under these two pathways, the combined effects of climate change and population distribution do not trigger a clear process of spatial rebalancing, and inequality largely persists [91]. Taken together, exposure inequality under different SSP trajectories exhibits divergent characteristics: it is likely to intensify under low-emission scenarios due to increasing concentration of risks, while under high-emission scenarios the expansion of heatwave coverage helps mitigate inequality to some extent. At the same time, the role of urbanization in modulating heatwave characteristics should not be overlooked. Previous findings show that urbanization partly alleviates the amplification of summer humid heatwaves [92], implying that in the future it might influence the spatial evolution of exposure inequality among the elderly by altering local thermal environments and redistributing population.

4.3. Limitations

This study employed multi-source data and scenario-based simulations to assess the spatiotemporal patterns of future heatwave exposure among the elderly population in Chinese cities. Nevertheless, the findings are inevitably constrained by data availability and methodological choices. Future projections of population and urbanization inherently involve uncertainties, and variations in scenario settings and model pathways may affect the precision of exposure assessments. In addition, although the single climate model adopted here performs relatively well in East Asia, it cannot capture the full range of uncertainties arising from inter-model differences. Another limitation lies in the assumption that all elderly people respond to heatwaves equally. While this homogeneous response assumption simplifies the modeling process, it overlooks potential differences in vulnerability due to socioeconomic and health factors. However, given that the Chinese population is predominantly Han nationality, with relatively limited genetic and physiological diversity, this assumption remains a reasonable approximation for large-scale national assessments. It should also be emphasized that the present study relies on scenario-based trend analyses, aiming to reveal potential risk patterns and policy implications rather than to provide deterministic forecasts.

5. Conclusions

Drawing upon high-resolution meteorological observations and reanalysis data, this study systematically assessed the historical evolution of urban heatwave events in China during 2000–2019. Building on this foundation, future climate scenario simulations were combined with demographic and urbanization projections to reveal the spatiotemporal dynamics, inequality patterns, and driving mechanisms of urban elderly population exposure to heatwaves across China during 2020–2099. The main conclusions were as follows:

• Heatwave characteristics: between 2000 and 2019, both HWF and HWD increased significantly across China. For 2020–2099, all SSP scenarios projected increases in HWF and HWD, with stronger trends and larger regional disparities under high-emission pathways, particularly in eastern and southern provinces.
• Elderly exposure: the exposure of the urban elderly population to heatwaves increased most sharply during the early and middle projection periods, followed by a gradual decline toward the end of the century.
• Spatial inequality: considerable interprovincial inequality in elderly exposure was observed, with Gini index values remaining high across the century. Spatial disparities in HWD exposure were consistently larger than those in HWF. Inequality slightly eases under SSP3-70, whereas high-emission pathways maintain relatively high heterogeneity.
• Contributions to exposure: population ageing emerged as the dominant driver of increasing exposure, accounting for nearly half of the contribution. Climate change was the second-largest factor, with its effect strengthened markedly under SSP3-70. Urbanization exerted a negative independent effect, yet its interactions with population and climate modulated localized risks.