Urban Heatwave Resilience and Spatial Drivers: A 20-Year Geo-Spatial Analysis in Chengdu, China

Urban Heatwave Resilience and Spatial Drivers: A 20-Year Geo-Spatial Analysis in Chengdu, China | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Urban Heatwave Resilience and Spatial Drivers: A 20-Year Geo-Spatial Analysis in Chengdu, China shichen fan, hongyong liu, Zhihui lai, yi xu This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7374786/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 4 You are reading this latest preprint version Abstract Amidst global climate change and rapid urbanization, urban heatwaves pose a severe threat to thermal environmental resilience, impacting public health and socioeconomic stability. Accurately quantifying the spatiotemporal dynamics of heatwave risks and resilience is a critical prerequisite for developing effective urban climate adaptation strategies. However, previous research is often limited by discrete station-based data and lacks standardized thermal environment metrics, failing to fully capture the complex, non-linear, and spatially non-stationary characteristics of heatwave drivers. To address these gaps, this study introduces new thermal environment indicators and employs an integrated analytical framework to analyze a 20-year high-resolution dataset for Chengdu, China. The results reveal that while land use patterns (PLAND) are the dominant factor shaping the thermal environment, the impacts of various drivers exhibit significant spatial non-stationarity; for instance, the cooling effect of green spaces diminishes in highly urbanized cores. ‘Contextual reversal’ of Aerosol Optical Depth (AOD)’s effect is shown in the result: AOD provides a cooling ‘parasol effect’ under normal conditions but can reverse its role to a warming ‘blanket effect’ during extreme heat and drought, thereby exacerbating heatwaves. Based on these findings, this study challenges ‘one-size-fits-all’ approaches and provides a robust scientific foundation for developing precise, adaptive governance strategies, such as climate-adaptive zoning, dynamic risk monitoring, and synergistic pollution-heat control policies. Urban Heatwave Thermal resilience Climate Change sustainable urban development Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 1. Introduction The global climate system is undergoing significant changes characterized by warming, with human activities, particularly greenhouse gas emissions, identified as the primary drivers. This warming trend, in turn, directly contributes to an increased frequency, intensity, and duration of extreme weather events, particularly heatwaves(Perkins-Kirkpatrick et al., 2020). Concurrently, rapid and often unplanned urbanization—resulting in increased impervious surfaces, reduced vegetation, and greater anthropogenic heat emissions—creates a distinct urban heat island (UHI) effect. This localized warming is superimposed on global climate change, further exacerbating heatwave risks in urban areas (Cheval et al., 2024). The negative impacts of urban heat waves are multifaceted. First, for urban ecosystems, high temperature stress affects the physiological and ecological processes of urban vegetation, reduces its ability to sequester carbon and release oxygen, and even leads to death, destroying urban biodiversity (Esperon-Rodriguez et al., 2022). High temperatures also heat urban water bodies, affecting water quality, exacerbating energy consumption (e.g. air conditioning and refrigeration), and indirectly increasing greenhouse gas emissions. Secondly, urban heat waves have particularly direct and severe impacts on the health of the population (Gasparrini et al., 2015).They also have a significant impact on socio-economic activities, as they can exacerbate the pressure on the city's energy supply, and pose a threat to the normal functioning of infrastructure such as transportation and water supply (Li et al., 2024a). In the face of increasingly severe urban natural disasters, scientifically assessing urban resilience has become a critical issue. Within urban resilience theory, ‘stability’ and ‘resistance’ are core dimensions for evaluating how urban systems respond to disturbances, and they are typically used to measure the performance of critical infrastructure (e.g., power grids, transportation networks) or the continuity of essential urban services (Lv et al., 2024). However, in the specific field of the urban thermal environment , the quantification of these resilience components is still in its nascent stages. The concept of using surface temperature differences to measure urban thermal resilience was primarily proposed in the work of Xi(Xi et al., 2023), but a standardized and replicable set of calculation criteria for ‘thermal stability" and ‘thermal resistance’ has not yet been established. To fill this gap, this study draws upon resilience theory to formally introduce and define these two key indicators, aiming to provide a clear and quantifiable analytical framework for assessing urban thermal environmental resilience. Despite the growing research on urban heat waves, several challenges remain. Traditional studies mostly rely on weather station data, which point out observation characteristics make it difficult to comprehensively capture the spatial continuity of heat waves on complex urban surfaces, limiting the accuracy of refined analyses. Furthermore, the definition of heat waves itself has a ‘definitional dilemma’: Traditional definitions rely on fixed thresholds, whereas percentile-based methods with inappropriate sliding windows may introduce systematic biases, leading to underestimation of frequency and misclassification of trends (Brunner L et al., 2024) and lack of globally harmonized standards harmonized standards (Abunyewah et al., 2025). There is also lack of recognized metrics and methodologies to quantify the resilience of urban thermal environments, and it is still difficult to construct comprehensive indicators that can fully reflect the thermal characteristics of cities and their response to perturbations. All these factors limit the accuracy of risk assessment and the effectiveness of adaptation strategies. To address research gaps, this study introduces several methodological innovations. First, we utilize the TRIMS LST daily dataset, whose high spatiotemporal resolution (1 km, daily) and ‘all-weather’ characteristics provide a continuous and reliable data foundation. Second, drawing on ecological resilience theory (Huang et al., 2021; Yi C et al., 2021; Allen et al., 2019), we construct two novel indicators to assess urban thermal environmental resilience: (1) Stability (ST), which quantifies LST fluctuations during non-heatwave periods, and (2) Resistance (RS), which evaluates the system's ability to mitigate temperature increases during heatwave events. Third, we employ an optimized heatwave identification methodology by dynamically constructing pixel-level thresholds for each day of the year (DOY) based on the 90th percentile within a 3-day sliding window, combined with a 3-day minimum duration criterion. This allows for the precise calculation of key heatwave characteristics: Frequency (HWF), Duration (HWD), and Cumulative Intensity (HWC). The integration of these methodologies aims to enhance the accuracy and depth of the study, laying the foundation for a more accurate understanding of the spatial and temporal dynamics of urban heat waves and their complex relationship with the resilience of the thermal environment. Building upon this framework, this study systematically investigates the spatiotemporal evolution of heatwave risk and thermal resilience in Chengdu and elucidates its complex driving mechanisms. The analytical process involves a two-stage approach. First, the GeoDetector model is utilized to identify the dominant driving forces from a pool of 14 multi-source factors. Subsequently, a Geographically Weighted Random Forest model with SHAP analysis (GWRF+SHAP) is applied to these dominant drivers to dissect the spatial non-stationarity and dynamic evolution of their impacts. 2. Data Selection 2.1 Study area Chengdu, as the capital of Sichuan Province, is a typical example of rapid urbanization in western China. With the permanent resident population surging to over 21 million, its construction land has expanded dramatically, and a large amount of natural land surface has been replaced by impervious surfaces (Li et al., 2024b). This drastic change in land use has significantly exacerbated the urban heat island (UHI) effect. Studies have shown that Chengdu's heat island pattern has evolved into a multi-centered, suburban-spreading trend, resulting in highly spatially heterogeneous heatwave risks (Shi et al., 2023). The non-coordinated development model where urban expansion outpaces population growth may exacerbate the heat island effect in newly developed areas, thereby increasing the exposure risks for incoming populations (Lyu et al., 2023). In the summer of 2022, a historic heatwave had a profound impact on Chengdu. The event caused Sichuan Province to experience its strongest heatwave and drought since meteorological records began in 1961, with hydropower capacity plummeting by over 50%. The resulting severe power shortages forced large-scale production cuts and shutdowns in Chengdu's industrial sector, highlighting the urgency and importance of conducting heatwave risk research in the region. 2.2 Data and Preprocessing 2.2.1 TRIMS LST daily dataset Most current studies on urban spatio-temporal variability of heat waves rely on weather station data (Zhan et al., 2019) . Weather station data are point-based and only measure temperatures within a specific radius around the station. Therefore, they cannot fully capture the spatial continuity of heatwaves. Some scholars have pointed out that LST is an important parameter in urban thermal environment and dynamics research, providing a continuous spatial view that temperature data cannot achieve, and is crucial for detailed investigations of urban surface climate (Almeida et al., 2021). Considering the challenges of data collection and the importance of maintaining uninterrupted time series in the dataset, this study utilized the TRIMS LST daily dataset (Tang et al., 2024) to investigate the spatiotemporal variations of heatwaves in Chengdu from May to September 2004-2023. By considering the influence of various factors, this study conducted a comprehensive analysis of the factors affecting the spatiotemporal variations of heatwaves in Chengdu over the past 20 years using multi-source climate and socio-economic data. 2.2.2. Identification and Definition of Heatwaves Accurate identification and quantification of the spatiotemporal characteristics of heatwaves are prerequisites for assessing heat risks and developing effective response strategies. Currently, the academic community has developed various heatwave identification methods (Perkins et al., 2013), laying the foundation for research on heatwave risk mitigation and adaptation. To accurately quantify the spatiotemporal characteristics of heatwaves, this study adopted a dynamic threshold-based identification method, primarily referencing the research by Brunner & Voigt (2024) (Brunner L et al., 2024). This method first calculates land surface temperature (LST) anomalies to remove seasonal cycles. Then, for each day of the year, a 3-day sliding window centered on that day is used to aggregate LST anomaly data from all years (2004-2023), and the 90th percentile is taken as the dynamic threshold for that day. When the LST anomaly value of a pixel exceeds its corresponding threshold on a given day and this condition persists for at least three consecutive days, it is defined as a heatwave event. Based on the identified heatwave events, this study focuses on the summer season (May–September) and calculates the following three key indicators: heatwave frequency (HWF), representing the number of heatwave events per year; heatwave duration (HWD), indicating the total number of heatwave days in a year; and heatwave cumulative intensity (HWC), measuring the cumulative excess temperature exceeding the threshold during heatwave periods. 2.2.3. Urban heat environment resilience indicators Our approach to defining urban heat resilience indicators is grounded in established climatological principles and resilience theory. The World Meteorological Organization (WMO) recognizes statistical measures like standard deviation as effective for describing climate variability (World Meteorological Organization, 2018), providing a foundation for our ‘Stability’ metric. This study operates core concepts from resilience science. Although 'resistance' and 'stability' are often considered two key perspectives in resilience science for evaluating system responses (Bruneau et al., 2003; Cutter, 2016), their application to the urban thermal environment has been largely conceptual, lacking standardized, LST-based quantitative metrics to assess the system's intrinsic thermal performance. To fill this gap, this study proposes and calculates two specific thermal resilience indicators: (1) Stability (ST): This indicator measures the inherent variability of the system during non-heatwave periods. It is calculated as the standard deviation of the deviation between the daily land surface temperature (LST) and its multi-year average baseline (i.e., the average LST of non-heatwave days over the same period) during all non-heatwave days in summer. A smaller ST value indicates greater thermal stability of the system under normal conditions. (2) Resistance (RS): This indicator assesses the system's ability to suppress extreme temperature increases during heatwave impacts. It is calculated as the average increase in peak LST relative to the baseline LST on the same day during all heatwave events in summer. A smaller RS value indicates stronger resistance to heatwave impacts. 2.2.4. Selection of influencing factors To comprehensively and scientifically analyze the influencing factors of heatwaves, this study systematically selected and integrated influencing factors from three dimensions: key climate factors, urban land surface characteristics, and human socio-economic activities. Climate factors : Air temperature (Ta), precipitation (Pre), and surface solar radiation (SSR) serve as the basic metrics for heatwave intensity, regulators of surface energy balance, and the primary energy source for surface warming, respectively (Nairn et al., 2014). Urban land surface characteristics : To characterize their moderating effects on the local thermal environment, the following factors were included: normalized difference vegetation index (NDVI) and impervious surface ratio, which reflect the cooling and warming effects of land cover (De Razza et al., 2024);Digital Elevation Model (DEM) to characterize terrain effects (WANG M et al., 2018); Landscape Pattern Indices (PLAND, PD, LSI) to quantify the area, fragmentation, and shape complexity of blue-green spaces (Kong et al., 2025); and Aerosol Optical Depth (AOD) and near-surface wind speed (WS) to represent atmospheric turbidity (Gil-Díaz et al., 2025)and local ventilation capacity (Luo et al., 2023), respectively. Socioeconomic factors : To characterize human activity impacts, nighttime lighting (NL) and carbon emissions (CE) were selected as proxy indicators for economic activity and anthropogenic heat release (Debnath et al., 2025), and population density (PopDen) was chosen to indicate potential anthropogenic heat source intensity and population exposure risk (Wang et al., 2021). The specific sources, spatio-temporal resolution, and processing methods for each influencing factor are detailed in Appendix 1. 3. Methodology This study aims to investigate the spatial and temporal characteristics of heat waves and urban heat resilience in Chengdu, as well as their influencing factors. The study adopts the framework diagram shown in Figure 2. Initially, the data was preprocessed, and the heat wave index was extracted and calculated. Then, the temporal, spatial and spatial characteristics of heat waves were analyzed using various methods. Finally, the influencing factors of heat waves and urban heat resilience were analyzed. 3.1 Temporal and spatial characteristics and resilience index analysis 3.1.1 Trend analysis To investigate the trends and significance of various heatwave indicators during the study period, this study combined Theil-Sen Median trend analysis with the Mann-Kendall (MK) test. This combined method has no special requirements for data distribution, is computationally robust, and can effectively resist the interference of outliers, making it suitable for trend analysis of long-term time series(Shao et al., 2024). By calculating the trend slope ( ) and significance statistic ( ) and setting the significance level ( ), this method identifies the spatial patterns of significant increases, significant decreases, or no significant changes in each indicator. 3.1.2 Spatial autocorrelation analysis To explore the spatial distribution patterns of the trends of the indicators, spatial autocorrelation analysis was further introduced in this study. Moran's index (Moran's I) was used to examine the spatial distribution patterns of heat wave indicators and their trends. The global Moran's index measures the overall spatial aggregation, while the localized Moran's index (LISA) further identifies the hot (HH) and cold (LL) regions, which are used to reveal local anomalous regions. The formula for calculating the Moran index is as follows: (1) In the formula, is the deviation between the attribute of element and its average value , is the spatial weight between elements and , is equal to the total number of elements, and is the aggregation of all spatial weights. 3.2. Factor Analysis of Heat Wave and Toughness Indicators 3.2.1. GeoDetector model To investigate the driving forces behind the spatial differentiation of heatwave characteristics, this study employs the Geodetector model. This model is particularly adept at analyzing interactions between factors and does not require the numerous assumptions of traditional linear regression. The Geodetector model has significant advantages in handling spatial data and complex variable interactions. The specific applications in this study include: (1) Factor Detector, which uses the statistic to measure the explanatory power of individual factors; (2) Interaction Detector, which identifies the type (e.g., enhancing, nonlinear) and strength of two-factor interactions by comparing changes in values. 3.3.2. GWRF+shap model To further reveal the spatial heterogeneity of ecological resilience determinants, a geographically weighted random forest (GWRF) model was constructed, integrating the nonparametric modeling capabilities of random forests with the local regression concept of GWR. The model establishes local random forests using spatially weighted samples, with Gaussian kernel functions defining the weights, thereby enhancing adaptability to non-stationary relationships (Fotheringham et al., 2017).The bandwidth is automatically selected using pseudo leave-one-out cross-validation to balance accuracy and generalization ability (Wu et al., 2024). To explain the local prediction results of the GWRF “black box” model, this study integrates the SHAP (SHapley Additive exPlanations) method. This method quantifies the marginal contribution of each driver to the model output at each spatial location, thereby identifying the dominant influencing factors in different regions and their direction of influence. 4 Results 4.1 Trend patterns of heat wave indicators This study employs Theil-Sen Median analysis and Mann-Kendall tests to conduct trend analysis and statistical analysis of heatwaves and heat resilience indicators. The results indicate that over the past two decades, both heatwaves and resilience indicators have exhibited a ‘central peripheral’ spatial pattern, with trends peaking in urban centers and decreasing toward the periphery. More critically, heatwave intensification zones and resilience degradation zones exhibit high spatial overlap, particularly in urban centers and the southeastern regions, forming distinct ‘dual vulnerability’ zones. This transformation has rendered urban core areas into ‘hotspots’ characterized by the concentrated risks of thermal environmental hazards and ecological fragility. To illustrate this composite risk, Figure 3 and Figure 4 highlight the spatial trends of indicators representing the comprehensive severity of heatwaves (HWC) and indicators directly reflecting the ability to resist impacts (RS). Areas with significant increases in HWC and significant decreases in RS are highly consistent, directly confirming the spatial coupling relationship. Overall, these phenomena reveal that Chengdu is currently facing a negative feedback loop of ‘h eatwave intensification-resilience decline ’ driven by urbanization: urban development has amplified the urban heat island effect, leading to more severe heatwaves, while simultaneously eroding the ecosystem's inherent regulatory capacity, further exacerbating the accumulation of risks. (For detailed trends and spatial distributions of other indicators, please refer to Appendix Figures S1-S3.) 4.2 Spatial autocorrelation analysis Spatial autocorrelation analysis (Table 1) revealed that all indicators exhibited extremely strong positive spatial autocorrelation (Moran's I: 0.466-0.658, p <0.01), confirming that the deterioration of the thermal environment and the decline in resilience are not randomly distributed in space but have formed a significant clustering pattern. Local spatial autocorrelation (LISA) analysis further revealed the specific spatial distribution characteristics of this clustering pattern (Figure 5): a contiguous high-risk zone composed of “hotspots” (H-H clustering) precisely covers the city center and southern expansion zone. This directly reflects the profound impact of high-intensity urbanization on local climate and has formed a negative feedback core of “heatwave intensification-resilience decline” in this region. In contrast, “cold spots” (L-L clustering) are primarily distributed in the northwestern Longmen Mountain ecological conservation area, where the stable thermal environment serves as a natural barrier against external high temperatures. Notably, within the “hot spot” core area, multiple “low-high” (L-H) outliers corresponding to large parks and water bodies are clearly visible, indicating that urban blue-green spaces continue to play a critical local cooling “cool island” effect even under overall deteriorating conditions. However, a deeper analysis of the LISA pattern reveals that the regulatory roles of different geographical elements exhibit significant indicator specificity. The Longmen Mountains in the northwest exhibit unique advantages in maintaining temperature stability, with only the stability (ST) indicator forming a large-scale L-L cold spot cluster (Figure 5). This is primarily due to the synergistic regulatory effects of its high-altitude terrain and dense forest vegetation. However, its mitigating effects on other indicators such as heatwave frequency and duration are not significant, indicating that its ecological functions should not be generalized. In contrast, the Longquan Mountains in the eastern part of the city, although not forming a large-scale cold spot core, serve as an important ecological corridor, with surrounding areas showing L-L clustering or relatively stable trends in multiple indicators. Under the backdrop of high-intensity urbanization, it effectively blocks the uncontrolled spread of the urban heat island effect, providing critical ecological support to break the negative feedback loop in the urban core. Based on this spatial differentiation, policymaking should adopt differentiated strategies: implement high-intensity ecological restoration in the “hotspot” areas of the urban center, strictly protect the Longquan Mountains and leverage their ecological corridor functions, and prioritize maintaining the temperature stabilization capabilities of the transition zone in the northwestern Longmen Mountains. Table 1 Results of global spatial autocorrelation analysis of heat wave indicators and thermal resilience indicators Indicator Moran index P value Z value HWC 0.527946 0.000000 88.262037 HWF 0.465567 0.000000 77.836641 ST 0.658267 0.000000 110.050951 RS 0.449707 0.000000 75.186666 HWD 0.514865 0.000000 86.075111 4.3 Geodetector analysis The factor analysis results have shown in Table 2. The result indicates that landscape composition (PLAND) is the dominant factor shaping the spatial variation of all heatwave and resilience indicators. Its explanatory power (q-values ranging from 0.564 to 0.761) is shown in Table 2. This confirms that the macro-scale pattern of the urban thermal environment is primarily determined by land use types (e.g., the ratio of impervious surfaces to blue-green spaces). In addition to PLAND, the secondary driving factors for each indicator exhibit significant differences, revealing that the formation mechanisms of heatwaves and the resilience mechanisms of the thermal environment are regulated by different factors. Specifically, the direct characteristics of heatwaves (such as frequency HWF, intensity HWC, and duration HWD) are largely influenced by immediate atmospheric and meteorological factors, such as aerosol (AOD), wind speed (WS), and solar radiation (SR), which exhibit strong explanatory power. This indicates that the formation and dissipation of heatwaves are direct responses to short-term energy and atmospheric changes based on surface patterns. However, when assessing the inherent thermal environmental resilience of the system, the importance ranking of driving factors is notably restructured. System stability (ST), referring to the degree of fluctuation under normal conditions, is primarily constrained by relatively stable factors such as the background climate (ambient temperature, AT), local topography (digital elevation model, DEM), and baseline ventilation conditions (wind speed, WS). In contrast, system resistance (RS), defined as the ability to withstand external shocks, exhibits greater sensitivity to atmospheric pollution (aerosol optical depth, AOD) and hydrological conditions (precipitation, Pre). This differentiation in driving mechanisms highlights that the formation of heatwave events and the maintenance of urban thermal environmental resilience are two related but distinct urban eco-climatic processes. Table 2 Explanatory power of various driving factors on the spatial differentiation of heat waves and resilience indicators ( q statistic) Indicator Variable CE Ta NDVI Pre NL DEM ISP AOD AOD PopDen SSR Pland PD LSI HWF 0.009 0.022 0.020 0.027 0.031 0.002 0.010 0.085 0.064 0.010 0.054 0.717 0.021 0.013 HWC 0.006 0.015 0.022 0.014 0.023 0.001 0.015 0.046 0.032 0.005 0.027 0.564 0.010 0.012 HWD 0.004 0.010 0.063 0.021 0.022 0.005 0.010 0.082 0.057 0.010 0.032 0.748 0.014 0.013 Stability 0.029 0.154 0.045 0.015 0.048 0.105 0.019 0.066 0.103 0.020 0.026 0.761 0.043 0.015 Resistance 0.014 0.05 0.040 0.138 0.040 0.031 0.018 0.217 0.079 0.016 0.096 0.735 0.025 0.015 Interaction detection results (Figure 6) indicate that the interaction between any two driving factors exhibits either a two-factor enhancement or a nonlinear enhancement, with their explanatory power for the dependent variable typically far exceeding the sum of the individual factors, highlighting the complex synergistic nature of urban thermal environment driving mechanisms (Wang et al., 2022). Among these, the interaction effect of landscape composition (PLAND) is particularly prominent. The interactions between PLAND and any other factor (such as AOD, NTL, etc.) almost always form the strongest combination in explaining the dependent variable. This indicates that PLAND is not only a dominant direct driver but also a foundational “amplifier”, determining the extent to which other factors (such as anthropogenic heat and air pollution) can influence the local thermal environment. Specific interaction combinations reveal several key driving pathways: First, the synergistic intensification effect of urbanization factors is significant. For example, the strong interaction between nighttime lighting (NL) and impervious surfaces (ISP) (NL∩ISP) precisely points to the process by which high-density built-up areas, due to building heat storage and high-intensity anthropogenic heat emissions, jointly exacerbate thermal environment degradation at night. Second, the composite regulatory effects of natural geography and meteorological conditions are complex. For example, the interaction between wind speed and terrain (WS∩DEM) reveals the key role of ventilation corridors in heat dissipation (Lu et al., 2020); while the interaction between aerosols and solar radiation (AOD∩SSR) demonstrates the bidirectional regulatory mechanism of atmospheric transparency on surface energy balance—high AOD reduces daytime solar radiation input but may also exacerbate nighttime insulation effects by absorbing long-wave radiation, especially under weak wind conditions (AOD∩WS) (Singh et al., 2024). From the perspective of thermal environmental resilience, the interaction mechanism is even more unique. The system's resistance (RS), i.e., its ability to withstand heatwave impacts, has been proven to be primarily dominated by the nonlinear synergistic effects of terrain (DEM), aerosols (AOD), and solar radiation (SSR). This finding may be related to Chengdu's unique basin topography. This topography acts as a natural “container”, suppressing near-surface atmospheric advection and convection, significantly amplifying the radiative and thermodynamic effects of aerosols, causing polluted air masses and heat to linger over the city for extended periods, thereby systematically weakening the city's overall resistance. 4.4 GWRF+SHAP Analysis This section selects the years 2004, 2015, and 2023 as temporal snapshots for analysis, each representing the decadal trend of change. This section focuses on the spatiotemporal differentiation of the two dominant factors, PLAND and AOD, identified by the GeoDetector model as key drivers of cumulative heatwave intensity in Chengdu. As shown in Figure 7, the influence of PLAND on HWC underwent a fundamental reversal during the study period. In 2004, PLAND exhibited a significant cooling effect (SHAP values were negative) across the entire study area. However, as the city expanded rapidly toward the south and southeast, by 2023, its positive warming effect had formed large contiguous areas in the urban core and suburban regions, while the original cooling effect had retreated to the peripheral mountainous areas. This reversal of effects reveals the paradox of urbanization: although green spaces are traditional cooling measures, in high-density built-up areas, green space fragmentation, soil drying, and urban canyon effects collectively weaken their evapotranspiration cooling function (Li et al., 2022), transforming them from “cooling sources” into potential “heat accumulation zones”. This effect is further exacerbated by the unfavorable low wind speeds in basins, intensifying local heat risks. The regulatory effect of AOD on HWC shown in Figure 8, which exhibits significant context dependence, transitioning from a “shading cooling” (umbrella effect) to a “heating warming” (blanket effect) under different spatiotemporal conditions. In most regions in 2004 and 2015, AOD primarily reflected shortwave solar radiation, resulting in a cooling effect. However, by 2023, under extreme high-temperature drought conditions and specific atmospheric circulation patterns (such as the westward extension of the subtropical high-pressure system), the absorption of longwave radiation by absorbent aerosols and their obstruction of heat dissipation became prominent, triggering AOD's warming “blanket” effect, particularly in the northwestern foothills of Chengdu, where heat and pollutants tend to linger. Notably, since 2015, Chengdu's sustained pollution control efforts have significantly reduced AOD concentrations. This may explain why the cooling effect of AOD was maintained or even reappeared in the eastern region in 2023, highlighting the complex role of pollution control in regulating heatwave risks. In summary, the impact of AOD on heatwaves is not static or consistent but is subject to the coupled regulation of multiple factors, including the macro-thermal environmental background, atmospheric circulation patterns, and the effectiveness of pollution control measures. Future urban heatwave management should fully consider the context-dependent characteristics of aerosols, identify their response pathways under different conditions, and avoid making blanket judgments about their regulatory effects to achieve more precise pollution control and heatwave adaptation strategies. 5 Discussion and Conclusion Based on high-resolution remote sensing data from 2004 to 2023 and a newly developed evaluation index system for thermal environmental resilience, this study systematically analyzes the spatiotemporal evolution of heatwaves and their complex driving mechanisms in Chengdu. The core findings reveal that the city is undergoing a simultaneous deterioration process characterized by intensified heatwaves and declining thermal environmental resilience. These two phenomena are highly spatially coupled, giving rise to “dual-risk zones” centered around the urban core and expansion areas. The pattern of land surface landscape (PLAND) is identified as the absolute dominant factor determining the spatial heterogeneity of the thermal environment. However, the formation of heatwaves and the maintenance of resilience are governed by different combinations of factors, reflecting a divergence in driving mechanisms. Notably, this study uncovers the non-stationarity and scenario dependency of key driving factors. For instance, under extreme hot and arid conditions, atmospheric aerosols (AOD) can shift from acting as a cooling ‘parasol’ to a warming ‘blanket’, thereby significantly amplifying local thermal risks. These findings underscore the need to move beyond a ‘one-size-fits-all’ approach in urban thermal management and instead implement spatially differentiated, precision-based strategies. Specific policy recommendations include: Develop precision adaptation plans based on spatial heterogeneity. Integrate identified high-risk ‘hotspot areas’ into climate-adaptive land use zoning frameworks. Enforce the construction of highly connected blue-green ecological networks, prioritize the protection of urban ventilation corridors, and promote localized cooling technologies in urban renewal initiatives. Establish a dynamic resilience monitoring and early warning system. Incorporate the stability and resistance indicators proposed in this study into routine urban health assessments. Build an integrated warning platform that couples physical risk with social vulnerability, with special alerts targeting extreme scenarios such as the reversal effects of AOD. Implement adaptive co-regulation of air pollution and urban heat islands. Formulate ‘scenario-based’ air pollution control plans that consider the context-dependent effects of AOD. Under extreme hot and dry conditions, priority should be given to limiting emissions of highly absorptive aerosols, such as black carbon, to achieve synergistic benefits in pollution control and heatwave adaptation. Despite the methodological innovations and novel findings of this study, certain limitations remain. Specifically, the research falls short in aspects such as data scale matching, three-dimensional urban spatial analysis, and the deep integration of socio-economic dimensions. Future studies should focus on leveraging multi-source data with higher spatial and temporal resolution and incorporate urban canopy models to enable three-dimensional thermal simulations. In addition, there is an urgent need to promote interdisciplinary integration by coupling physical risk assessments with public health and socio-economic data. This would support the development of a comprehensive urban heatwave vulnerability and risk assessment framework, thereby providing a more solid scientific basis for designing climate adaptation strategies that are both equitable and targeted. Declarations CRediT authorship contribution statement Shichen Fan: Conceptualization, Methodology, Data Curation, Formal Analysis, Investigation, Writing – Original Draft, Writing – Review & Editing. Hongyong Liu: Supervision, Funding Acquisition, Project Administration, Conceptualization, Writing – Review & Editing. ZhiHui Lai: Writing – Review & Editing, Visualization. Yi Xu: Software, Writing – Review & Editing. Acknowledgments We are grateful to Xiaoping Wang, a PhD candidate at Deakin University, for his insightful suggestions and assistance in revising this manuscript. The authors also acknowledge the data support from the following sources: the National Earth System Science Data Center, National Science & Technology Infrastructure of China; the National Tibetan Plateau / Third Pole Environment Data Center (http://data.tpdc.ac.cn); and the DPECv1.0-v1.2 emission data provided by the MEIC model (http://meicmodel.org.cn). Declaration of generative AI and AI-assisted technologies in the writing process Statement: During the preparation of this work the author(s) used Gemini (a large language model developed by Google) in order to assist with language refinement, translation between English and Chinese. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the published article. References Abunyewah M, Gajendran T, Erdiaw-Kwasie M O, et al., 2025. The multidimensional impacts of heatwaves on human ecosystems: A systematic literature review and future research direction[J/OL]. Environmental Science & Policy, 165: 104024. DOI:10.1016/j.envsci.2025.104024. Allen C R, Angeler D G, Chaffin B C, et al., 2019. Resilience reconciled[J/OL]. Nature Sustainability, 2(10): 898-900. DOI:10.1038/s41893-019-0401-4. Almeida C R D, Teodoro A C, Gonçalves A, 2021. Study of the Urban Heat Island (UHI) Using Remote Sensing Data/Techniques: A Systematic Review[J/OL]. Environments, 8(10): 105. DOI:10.3390/environments8100105. Bai K, Li Ke, Ma M, et al., 2022. LGHAP: the Long-term Gap-free High-resolution Air Pollutant concentration dataset, derived via tensor-flow-based multimodal data fusion[J/OL]. Earth System Science Data, 14(2): 907-927. DOI:10.5194/essd-14-907-2022. Bruneau M, Chang S E, Eguchi R T, et al., 2003. A Framework to Quantitatively Assess and Enhance the Seismic Resilience of Communities[J/OL]. Earthquake Spectra, 19(4): 733-752. DOI:10.1193/1.1623497. Brunner L, Voigt A, 2024. Pitfalls in diagnosing temperature extremes[J/OL]. Nature Communications, 15(1): 2087. DOI:10.1038/s41467-024-46349-x. Cao C, Lee X, Liu S, et al., 2016. Urban heat islands in China enhanced by haze pollution[J/OL]. Nature Communications, 7(1): 12509. DOI:10.1038/ncomms12509. Chen Z, Yu B, Yang C, et al., 2021. An extended time series (2000–2018) of global NPP-VIIRS-like nighttime light data from a cross-sensor calibration[J/OL]. Earth System Science Data, 13(3): 889-906. DOI:10.5194/essd-13-889-2021. Cheng J, Tong D, Liu Y, et al., 2023. A synergistic approach to air pollution control and carbon neutrality in China can avoid millions of premature deaths annually by 2060[J/OL]. One Earth, 6(8): 978-989. DOI:10.1016/j.oneear.2023.07.007. Cheval S, Amihăesei V A, Chitu Z, et al., 2024. A systematic review of urban heat island and heat waves research (1991–2022)[J/OL]. Climate Risk Management, 44: 100603. DOI:https://doi.org/10.1016/j.crm.2024.100603. Cutter S L, 2016. The landscape of disaster resilience indicators in the USA[J/OL]. Natural Hazards, 80(2): 741-758. DOI:10.1007/s11069-015-1993-2. De Razza S, Zanetti C, De Marchi M, et al., 2024. Mapping urban heatwaves and islands: the reverse effect of Salento’s “white cities”[J/OL]. Frontiers in Earth Science, 12[2025-06-03]. https://www.frontiersin.org/journals/earth-science/articles/10.3389/feart.2024.1375827/full. DOI:10.3389/feart.2024.1375827. Debnath R, Chandel T, Han F, et al., 2025. Heatwave increases nighttime light intensity in hyperdense cities of the Global South: A double machine learning study[A/OL]. arXiv[2025-06-03]. https://arxiv.org/abs/2503.00557. DOI:10.48550/ARXIV.2503.00557. Didan K, 2015. MOD13A3 MODIS/Terra Vegetation Indices Monthly L3 Global 1km SIN Grid V006[DS/OL]. NASA Land Processes Distributed Active Archive Center[2025-06-21]. https://lpdaac.usgs.gov/products/mod13a3v006/. DOI:10.5067/MODIS/MOD13A3.006. Esperon-Rodriguez M, Tjoelker M G, Lenoir J, et al., 2022. Climate change increases global risk to urban forests[J/OL]. Nature Climate Change, 12(10): 950-955. DOI:10.1038/s41558-022-01465-8. Fotheringham A S, Yang W, Kang W, 2017. Multiscale Geographically Weighted Regression (MGWR)[J/OL]. Annals of the American Association of Geographers, 107(6): 1247-1265. DOI:10.1080/24694452.2017.1352480. Gasparrini A, Guo Y, Hashizume M, et al., 2015. Mortality risk attributable to high and low ambient temperature: a multicountry observational study[J/OL]. The Lancet, 386(9991): 369-375. DOI:10.1016/S0140-6736(14)62114-0. Gil-Díaz C, Sicard M, Nabat P, et al., 2025. Dust Aerosol Radiative Effects During a Dust Event and Heatwave in Summer 2019 Simulated with a Regional Climate Atmospheric Model over the Iberian Peninsula[J/OL]. Remote Sensing, 17(11): 1817. DOI:10.3390/rs17111817. Guo A, Yang J, Xiao X, et al., 2020. Influences of urban spatial form on urban heat island effects at the community level in China[J/OL]. Sustainable Cities and Society, 53: 101972. DOI:10.1016/j.scs.2019.101972. Huang Z, Liu X, Yang Q, et al., 2021. Quantifying the spatiotemporal characteristics of multi-dimensional karst ecosystem stability with Landsat time series in southwest China[J/OL]. International Journal of Applied Earth Observation and Geoinformation, 104: 102575. DOI:https://doi.org/10.1016/j.jag.2021.102575. jianfang K. Data set of 1 km resolution DEM in China[Z/OL]//National Cryosphere Desert Data Center. National Cryosphere Desert Data Center(www.ncdc.ac.cn). http://www.ncdc.ac.cn/portal/metadata/ba0e4b6b-8fef-4eec-9c84-81646f266282. Kong W, Liu Y, Dong S, et al., 2025. Research on the Influence of Evolution of Landscape Patterns of Blue-Green Space on the Cooling Effect in the Central Urban Area of Xi’an[J/OL]. Landscape Architecture, 32(5): 78-85. DOI:10.3724/j.fjyl.202408210473. Lebakula V, Epting J, Moehl J, et al., 2024. LandScan Silver Edition[DS/OL]. Oak Ridge, TN: Oak Ridge National Laboratory. DOI:10.48690/1531770. Li C, Lu L, Fu Z, et al., 2022. Diverse cooling effects of green space on urban heat island in tropical megacities[J/OL]. Frontiers in Environmental Science, 10: 1073914. DOI:10.3389/fenvs.2022.1073914. Li D, Bou-Zeid E, 2013. Synergistic Interactions between Urban Heat Islands and Heat Waves: The Impact in Cities Is Larger than the Sum of Its Parts[J/OL]. Journal of Applied Meteorology and Climatology, 52(9): 2051-2064. DOI:10.1175/JAMC-D-13-02.1. Li Y, Schubert S, Kropp J P, et al., 2020. On the influence of density and morphology on the Urban Heat Island intensity[J/OL]. Nature Communications, 11(1): 2647. DOI:10.1038/s41467-020-16461-9. Li Yuanxi, Luo X, Wang M, et al., 2024a. Spatiotemporal variations and influencing factors of heatwaves in Chengdu, China[J/OL]. Ecological Informatics, 84: 102924. DOI:10.1016/j.ecoinf.2024.102924. Li Yuanxi, Luo X, Wang M, et al., 2024b. Spatiotemporal variations and influencing factors of heatwaves in Chengdu, China[J/OL]. Ecological Informatics, 84: 102924. DOI:https://doi.org/10.1016/j.ecoinf.2024.102924. Liao W, Liu X, Wang D, et al., 2017. The Impact of Energy Consumption on the Surface Urban Heat Island in China’s 32 Major Cities[J/OL]. Remote Sensing, 9(3): 250. DOI:10.3390/rs9030250. Lu L, Weng Q, Xiao D, et al., 2020. Spatiotemporal Variation of Surface Urban Heat Islands in Relation to Land Cover Composition and Configuration: A Multi-Scale Case Study of Xi’an, China[J/OL]. Remote Sensing, 12(17). https://www.mdpi.com/2072-4292/12/17/2713. DOI:10.3390/rs12172713. Luo F, Yang Y, Zong L, et al., 2023. The interactions between urban heat island and heat waves amplify urban warming in Guangzhou, China: Roles of urban ventilation and local climate zones[J/OL]. Frontiers in Environmental Science, 11: 1084473. DOI:10.3389/fenvs.2023.1084473. Lv Y, Sarker M N I, 2024. Integrative approaches to urban resilience: Evaluating the efficacy of resilience strategies in mitigating climate change vulnerabilities[J/OL]. Heliyon, 10(6): e28191. DOI:10.1016/j.heliyon.2024.e28191. Lyu Z Z, Gao H, Gao R, et al., 2023. Extreme characteristics and causes of the drought event in the whole Yangtze River Basin in the midsummer of 2022[J/OL]. Advances in Climate Change Research, 14(5): 642-650. DOI:10.1016/j.accre.2023.09.007. Miao L, Ju L, Sun S, et al., 2024. Unveiling the dynamics of sequential extreme precipitation-heatwave compounds in China[J/OL]. npj Climate and Atmospheric Science, 7(1): 67. DOI:10.1038/s41612-024-00613-5. Nairn J, Fawcett R, 2014. The Excess Heat Factor: A Metric for Heatwave Intensity and Its Use in Classifying Heatwave Severity[J/OL]. International Journal of Environmental Research and Public Health, 12(1): 227-253. DOI:10.3390/ijerph120100227. PENG Shouzhang, 2024. 1-km monthly precipitation dataset for China (1901-2023)[DS/OL]. National Tibetan Plateau Data Center. https://dx.doi.org/10.5281/zenodo.3114194. DOI:10.5281/zenodo.3114194. Perkins S E, Alexander L V, 2013. On the Measurement of Heat Waves[J/OL]. Journal of Climate, 26(13): 4500-4517. DOI:10.1175/JCLI-D-12-00383.1. Perkins-Kirkpatrick S E, Lewis S C, 2020. Increasing trends in regional heatwaves[J/OL]. Nature Communications, 11(1): 3357. DOI:10.1038/s41467-020-16970-7. Shao Y, Zhu Q, Feng Z, et al., 2024. Temporal and Spatial Assessment of Carbon Flux Dynamics: Evaluating Emissions and Sequestration in the Three Northern Protection Forest Project Areas Supported by Google Earth Engine[J/OL]. Remote Sensing, 16(5): 777. DOI:10.3390/rs16050777. Shi H, Song X, Zeng S, 2023. Impact of the Urban Heat Island Effect on Ozone Pollution in Chengdu City, China[J/OL]. Chinese Geographical Science, 33(6): 1017-1032. DOI:10.1007/s11769-023-1391-8. Singh S P, Rafiuddin M, Banerjee S, et al., 2024. Investigation of the Thermal Structure in the Atmospheric Boundary Layer During Evening Transition and the Impact of Aerosols on Radiative Cooling[A/OL]. arXiv[2025-06-27]. https://arxiv.org/abs/2403.06656. DOI:10.48550/ARXIV.2403.06656. Tang W, Zhou J, Ma J, et al., 2024. TRIMS LST: a daily 1 km all-weather land surface temperature dataset for China’s landmass and surrounding areas (2000–2022)[J/OL]. Earth System Science Data, 16(1): 387-419. DOI:10.5194/essd-16-387-2024. Wang J, Chen Y, Liao W, et al., 2021. Anthropogenic emissions and urbanization increase risk of compound hot extremes in cities[J/OL]. Nature Climate Change, 11(12): 1084-1089. DOI:10.1038/s41558-021-01196-2. WANG M, XU H, 2018. Analyzing the Influence of Urban Forms on Surface Urban Heat Islands Intensity in Chinese Mega Cities[J/OL]. Journal of Geo-information Science, 20(12): 1787-1798. DOI:10.12082/dqxxkx.2018.180257. Wang R, Wang M, Zhang Z, et al., 2022. Geographical Detection of Urban Thermal Environment Based on the Local Climate Zones: A Case Study in Wuhan, China[J/OL]. Remote Sensing, 14(5): 1067. DOI:10.3390/rs14051067. Weng Q, Lu D, Schubring J, 2004. Estimation of land surface temperature–vegetation abundance relationship for urban heat island studies[J/OL]. Remote Sensing of Environment, 89(4): 467-483. DOI:https://doi.org/10.1016/j.rse.2003.11.005. World Meteorological Organization, 2018. Guide to Climatological Practices[M]. 3rd ed. Geneva: WMO. Wu D, Zhang Y, Xiang Q, 2024. Geographically weighted random forests for macro-level crash frequency prediction[J/OL]. Accident Analysis & Prevention, 194: 107370. DOI:10.1016/j.aap.2023.107370. Xi Z, Li C, Zhou L, et al., 2023. Built environment influences on urban climate resilience: Evidence from extreme heat events in Macau[J/OL]. Science of The Total Environment, 859: 160270. DOI:https://doi.org/10.1016/j.scitotenv.2022.160270. Xie Y, Weng Q, Fu P, 2019. Temporal variations of artificial nighttime lights and their implications for urbanization in the conterminous United States, 2013–2017[J/OL]. Remote Sensing of Environment, 225: 160-174. DOI:https://doi.org/10.1016/j.rse.2019.03.008. Yang M, Ren C, Wang H, et al., 2024. Mitigating urban heat island through neighboring rural land cover[J/OL]. Nature Cities, 1(8): 522-532. DOI:10.1038/s44284-024-00091-z. Yi C, Jackson N, 2021. A review of measuring ecosystem resilience to disturbance[J/OL]. Environmental Research Letters, 16(5): 053008. DOI:10.1088/1748-9326/abdf09. Zhan L F, Wang Y, Sun H, et al., 2019. Study on the Change Characteristics of and Population Exposure to Heatwave Events on the North China Plain[J/OL]. Advances in Meteorology, 2019: 1-10. DOI:10.1155/2019/7069195. Zhang P, Imhoff M L, Bounoua L, et al., 2012. Exploring the influence of impervious surface density and shape on urban heat islands in the northeast United States using MODIS and Landsat[J/OL]. Canadian Journal of Remote Sensing, 38(4): 441-451. DOI:10.5589/m12-036. Zhang Q, Jia B, Li T, et al., 2025. Dynamic changes and drives of surface urban heat islands in China[J/OL]. City and Environment Interactions, 27: 100203. DOI:https://doi.org/10.1016/j.cacint.2025.100203. Additional Declarations No competing interests reported. Supplementary Files Highlights.docx GraphicalAbstract.png Appendix.docx FigureS1.png FigureS2.png FigureS3.png Cite Share Download PDF Status: Under Review Version 1 posted Reviewers invited by journal 23 Oct, 2025 Editor assigned by journal 15 Aug, 2025 Submission checks completed at journal 15 Aug, 2025 First submitted to journal 14 Aug, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7374786","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":538880245,"identity":"63c89af7-3c4e-4586-828b-a57b5d925cb8","order_by":0,"name":"shichen fan","email":"","orcid":"","institution":"Southwest Petroleum University","correspondingAuthor":false,"prefix":"","firstName":"shichen","middleName":"","lastName":"fan","suffix":""},{"id":538880246,"identity":"bb5fff6c-1095-4387-9b8f-0a4e8e86023d","order_by":1,"name":"hongyong liu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAu0lEQVRIiWNgGAWjYBACA2YGNgaGCiiPh3gtZ0jSwgDUwthGkhZ2HrPHvPPqErdLJDA+eNvGIG9O2GE85sa82w4n7pyRwGw4t43BcGcDYS1m0rzbDuRuuJHAJs3bxpBgcIAoLXPqQFrYf5OgpYEZbAszkVrYyiTnHDtcv7PnYbPknHMShhsIabHvP7xN4k1NnbE5e/LBD2/KbOQJ2gICTKDoMGBgbABSEkSoBwLGH2Ato2AUjIJRMApwAADsWjffmcD6+QAAAABJRU5ErkJggg==","orcid":"","institution":"Southwest Petroleum University","correspondingAuthor":true,"prefix":"","firstName":"hongyong","middleName":"","lastName":"liu","suffix":""},{"id":538880247,"identity":"fbbdecb3-91f0-40e0-9e38-60e3b86284d3","order_by":2,"name":"Zhihui lai","email":"","orcid":"","institution":"Southwestern University of Finance and Economics","correspondingAuthor":false,"prefix":"","firstName":"Zhihui","middleName":"","lastName":"lai","suffix":""},{"id":538880248,"identity":"fbed619f-5735-46c2-8bb8-d93ad9853586","order_by":3,"name":"yi xu","email":"","orcid":"","institution":"Southwest University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"yi","middleName":"","lastName":"xu","suffix":""}],"badges":[],"createdAt":"2025-08-14 14:23:38","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7374786/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7374786/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":95228256,"identity":"62118395-f977-43a5-ae57-296804767fd5","added_by":"auto","created_at":"2025-11-05 16:33:32","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":853816,"visible":true,"origin":"","legend":"\u003cp\u003eResearch Area\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7374786/v1/1d118a0509200fdcdc40267e.png"},{"id":95171747,"identity":"4e0372c8-4cdf-4766-aed5-cfa910a0c18d","added_by":"auto","created_at":"2025-11-05 06:35:44","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":270138,"visible":true,"origin":"","legend":"\u003cp\u003eResearch framework\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7374786/v1/0a66071d570cc1bc03660f36.png"},{"id":95171753,"identity":"9c567fbd-c38f-49bb-8d1d-f1709bed50be","added_by":"auto","created_at":"2025-11-05 06:35:44","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":372003,"visible":true,"origin":"","legend":"\u003cp\u003eRS spatial change trend and area distribution map\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7374786/v1/522b14b5e7c0b0b50bbaa4bc.png"},{"id":95171754,"identity":"e3e2caac-5e87-4fc3-bcfa-8787657ca1c9","added_by":"auto","created_at":"2025-11-05 06:35:44","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":372035,"visible":true,"origin":"","legend":"\u003cp\u003eHWC spatial change trend and area distribution map\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7374786/v1/6eb5e43bfc2b86868d2be440.png"},{"id":95171752,"identity":"6450d470-67b8-4636-bcd3-f89b98f00d11","added_by":"auto","created_at":"2025-11-05 06:35:44","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":590247,"visible":true,"origin":"","legend":"\u003cp\u003eLISA cluster diagram of heatwave indicators and heat resilience indicators trends\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-7374786/v1/2dcd0b2ff3339726733fcffb.png"},{"id":95227624,"identity":"8aae8ad6-310c-4a77-b5f3-b7f7f6899d6d","added_by":"auto","created_at":"2025-11-05 16:32:40","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":586039,"visible":true,"origin":"","legend":"\u003cp\u003eHeat map of the explanatory power (\u003cem\u003eq\u003c/em\u003estatistic) of the interaction between various driving factors on different indicators\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-7374786/v1/ec5b99a1da9722c76cd09304.png"},{"id":95171750,"identity":"1691a709-7a26-41c5-9225-0690f0715699","added_by":"auto","created_at":"2025-11-05 06:35:44","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":208299,"visible":true,"origin":"","legend":"\u003cp\u003eSpatial distribution of SHAP values of PLAND on heat wave cumulative intensity (HWC) in Chengdu (2004, 2015 and 2023)\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-7374786/v1/188bb48352ad33d381d9c981.png"},{"id":95171758,"identity":"26e79073-b6fb-4c79-972a-17ba216b8d79","added_by":"auto","created_at":"2025-11-05 06:35:44","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":252501,"visible":true,"origin":"","legend":"\u003cp\u003eSpatial distribution of SHAP values of AOD on cumulative intensity of heat waves (HWC) in Chengdu (2004, 2015 and 2023)\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-7374786/v1/7c53f25ca655be6c4fb0003a.png"},{"id":95230652,"identity":"a0a49b62-1f99-4042-9304-82d15b942a03","added_by":"auto","created_at":"2025-11-05 16:38:09","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4127557,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7374786/v1/1ca90eb2-e625-40a3-a31c-cc1a83b19788.pdf"},{"id":95171748,"identity":"4e23c0be-005a-410c-953f-242502d1fb31","added_by":"auto","created_at":"2025-11-05 06:35:44","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":14401,"visible":true,"origin":"","legend":"","description":"","filename":"Highlights.docx","url":"https://assets-eu.researchsquare.com/files/rs-7374786/v1/5dd96ab673fcce3ed26d2dd3.docx"},{"id":95171749,"identity":"6a07788e-3d7a-4a51-ba68-2f6e90649d16","added_by":"auto","created_at":"2025-11-05 06:35:44","extension":"png","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":1384919,"visible":true,"origin":"","legend":"","description":"","filename":"GraphicalAbstract.png","url":"https://assets-eu.researchsquare.com/files/rs-7374786/v1/fade92ebb7d62e73bd26e316.png"},{"id":95171744,"identity":"a52bcf22-5893-4731-abd0-22f00dee3727","added_by":"auto","created_at":"2025-11-05 06:35:44","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":33558,"visible":true,"origin":"","legend":"","description":"","filename":"Appendix.docx","url":"https://assets-eu.researchsquare.com/files/rs-7374786/v1/396b274512b4e5b4541fa5ed.docx"},{"id":95171782,"identity":"39567d7c-a74a-413c-af13-1568a2ace999","added_by":"auto","created_at":"2025-11-05 06:35:45","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":12802972,"visible":true,"origin":"","legend":"","description":"","filename":"FigureS1.png","url":"https://assets-eu.researchsquare.com/files/rs-7374786/v1/b59e1ce11eb33047cc905f52.png"},{"id":95171785,"identity":"bbd3ed58-eaee-4c15-bfe1-f899638fd206","added_by":"auto","created_at":"2025-11-05 06:35:45","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":14276497,"visible":true,"origin":"","legend":"","description":"","filename":"FigureS2.png","url":"https://assets-eu.researchsquare.com/files/rs-7374786/v1/d81a561f8cc19c781e8d656f.png"},{"id":95171786,"identity":"c9aa5d25-ae39-44d6-a10a-678df4ba5593","added_by":"auto","created_at":"2025-11-05 06:35:45","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":13565549,"visible":true,"origin":"","legend":"","description":"","filename":"FigureS3.png","url":"https://assets-eu.researchsquare.com/files/rs-7374786/v1/b1a5dd68fb4e04792709548c.png"}],"financialInterests":"No competing interests reported.","formattedTitle":"Urban Heatwave Resilience and Spatial Drivers: A 20-Year Geo-Spatial Analysis in Chengdu, China","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eThe global climate system is undergoing significant changes characterized by warming, with human activities, particularly greenhouse gas emissions, identified as the primary drivers. This warming trend, in turn, directly contributes to an increased frequency, intensity, and duration of extreme weather events, particularly heatwaves(Perkins-Kirkpatrick et al., 2020). Concurrently, rapid and often unplanned urbanization—resulting in increased impervious surfaces, reduced vegetation, and greater anthropogenic heat emissions—creates a distinct urban heat island (UHI) effect. This localized warming is superimposed on global climate change, further exacerbating heatwave risks in urban areas (Cheval et al., 2024).\u003c/p\u003e\n\u003cp\u003eThe negative impacts of urban heat waves are multifaceted. First, for urban ecosystems, high temperature stress affects the physiological and ecological processes of urban vegetation, reduces its ability to sequester carbon and release oxygen, and even leads to death, destroying urban biodiversity (Esperon-Rodriguez et al., 2022). High temperatures also heat urban water bodies, affecting water quality, exacerbating energy consumption (e.g. air conditioning and refrigeration), and indirectly increasing greenhouse gas emissions. Secondly, urban heat waves have particularly direct and severe impacts on the health of the population (Gasparrini et al., 2015).They also have a significant impact on socio-economic activities, as they can exacerbate the pressure on the city's energy supply, and pose a threat to the normal functioning of infrastructure such as transportation and water supply (Li et al., 2024a).\u003c/p\u003e\n\u003cp\u003eIn the face of increasingly severe urban natural disasters, scientifically assessing urban resilience has become a critical issue. Within urban resilience theory, ‘stability’ and ‘resistance’ are core dimensions for evaluating how urban systems respond to disturbances, and they are typically used to measure the performance of critical infrastructure (e.g., power grids, transportation networks) or the continuity of essential urban services (Lv et al., 2024). However, in the specific field of the urban \u003cem\u003ethermal environment\u003c/em\u003e, the quantification of these resilience components is still in its nascent stages. The concept of using surface temperature differences to measure urban thermal resilience was primarily proposed in the work of Xi(Xi et al., 2023), but a standardized and replicable set of calculation criteria for ‘thermal stability\" and ‘thermal resistance’ has not yet been established. To fill this gap, this study draws upon resilience theory to formally introduce and define these two key indicators, aiming to provide a clear and quantifiable analytical framework for assessing urban thermal environmental resilience.\u003c/p\u003e\n\u003cp\u003eDespite the growing research on urban heat waves, several challenges remain. Traditional studies mostly rely on weather station data, which point out observation characteristics make it difficult to comprehensively capture the spatial continuity of heat waves on complex urban surfaces, limiting the accuracy of refined analyses. Furthermore, the definition of heat waves itself has a ‘definitional dilemma’: Traditional definitions rely on fixed thresholds, whereas percentile-based methods with inappropriate sliding windows may introduce systematic biases, leading to underestimation of frequency and misclassification of trends (Brunner L et al., 2024) and lack of globally harmonized standards harmonized standards (Abunyewah et al., 2025). There is also lack of recognized metrics and methodologies to quantify the resilience of urban thermal environments, and it is still difficult to construct comprehensive indicators that can fully reflect the thermal characteristics of cities and their response to perturbations. All these factors limit the accuracy of risk assessment and the effectiveness of adaptation strategies.\u003c/p\u003e\n\u003cp\u003eTo address research gaps, this study introduces several methodological innovations. First, we utilize the TRIMS LST daily dataset, whose high spatiotemporal resolution (1 km, daily) and ‘all-weather’ characteristics provide a continuous and reliable data foundation. Second, drawing on ecological resilience theory (Huang et al., 2021; Yi C et al., 2021; Allen et al., 2019), we construct two novel indicators to assess urban thermal environmental resilience: (1) Stability (ST), which quantifies LST fluctuations during non-heatwave periods, and (2) Resistance (RS), which evaluates the system's ability to mitigate temperature increases during heatwave events. Third, we employ an optimized heatwave identification methodology by dynamically constructing pixel-level thresholds for each day of the year (DOY) based on the 90th percentile within a 3-day sliding window, combined with a 3-day minimum duration criterion. This allows for the precise calculation of key heatwave characteristics: Frequency (HWF), Duration (HWD), and Cumulative Intensity (HWC). The integration of these methodologies aims to enhance the accuracy and depth of the study, laying the foundation for a more accurate understanding of the spatial and temporal dynamics of urban heat waves and their complex relationship with the resilience of the thermal environment.\u003c/p\u003e\n\u003cp\u003eBuilding upon this framework, this study systematically investigates the spatiotemporal evolution of heatwave risk and thermal resilience in Chengdu and elucidates its complex driving mechanisms. The analytical process involves a two-stage approach. First, the GeoDetector model is utilized to identify the dominant driving forces from a pool of 14 multi-source factors. Subsequently, a Geographically Weighted Random Forest model with SHAP analysis (GWRF+SHAP) is applied to these dominant drivers to dissect the spatial non-stationarity and dynamic evolution of their impacts.\u003c/p\u003e"},{"header":"2. Data Selection","content":"\u003ch2\u003e\u003cem\u003e2.1 Study area\u003c/em\u003e\u003c/h2\u003e\n\u003cp\u003eChengdu, as the capital of Sichuan Province, is a typical example of rapid urbanization in western China. With the permanent resident population surging to over 21 million, its construction land has expanded dramatically, and a large amount of natural land surface has been replaced by impervious surfaces (Li et al., 2024b). This drastic change in land use has significantly exacerbated the urban heat island (UHI) effect. Studies have shown that Chengdu\u0026apos;s heat island pattern has evolved into a multi-centered, suburban-spreading trend, resulting in highly spatially heterogeneous heatwave risks (Shi et al., 2023). The non-coordinated development model where urban expansion outpaces population growth may exacerbate the heat island effect in newly developed areas, thereby increasing the exposure risks for incoming populations (Lyu et al., 2023). In the summer of 2022, a historic heatwave had a profound impact on Chengdu. The event caused Sichuan Province to experience its strongest heatwave and drought since meteorological records began in 1961, with hydropower capacity plummeting by over 50%. The resulting severe power shortages forced large-scale production cuts and shutdowns in Chengdu\u0026apos;s industrial sector, highlighting the urgency and importance of conducting heatwave risk research in the region.\u003c/p\u003e\n\u003ch2\u003e\u003cem\u003e2.2 Data and Preprocessing\u003c/em\u003e\u003c/h2\u003e\n\u003ch3\u003e2.2.1 TRIMS LST daily dataset\u0026nbsp;\u003c/h3\u003e\n\u003cp\u003eMost current studies on urban spatio-temporal variability of heat waves rely on weather station data (Zhan et al., 2019) . Weather station data are point-based and only measure temperatures within a specific radius around the station. Therefore, they cannot fully capture the spatial continuity of heatwaves. Some scholars have pointed out that LST is an important parameter in urban thermal environment and dynamics research, providing a continuous spatial view that temperature data cannot achieve, and is crucial for detailed investigations of urban surface climate (Almeida et al., 2021). Considering the challenges of data collection and the importance of maintaining uninterrupted time series in the dataset, this study utilized the TRIMS LST daily dataset (Tang et al., 2024) to investigate the spatiotemporal variations of heatwaves in Chengdu from May to September 2004-2023. By considering the influence of various factors, this study conducted a comprehensive analysis of the factors affecting the spatiotemporal variations of heatwaves in Chengdu over the past 20 years using multi-source climate and socio-economic data.\u003c/p\u003e\n\u003ch3\u003e2.2.2. Identification and Definition of Heatwaves\u0026nbsp;\u003c/h3\u003e\n\u003cp\u003eAccurate identification and quantification of the spatiotemporal characteristics of heatwaves are prerequisites for assessing heat risks and developing effective response strategies. Currently, the academic community has developed various heatwave identification methods\u0026nbsp;(Perkins et al., 2013), laying the foundation for research on heatwave risk mitigation and adaptation. To accurately quantify the spatiotemporal characteristics of heatwaves, this study adopted a dynamic threshold-based identification method, primarily referencing the research by Brunner \u0026amp; Voigt (2024)\u0026nbsp;(Brunner L et al., 2024). This method first calculates land surface temperature (LST) anomalies to remove seasonal cycles. Then, for each day of the year, a 3-day sliding window centered on that day is used to aggregate LST anomaly data from all years (2004-2023), and the 90th percentile is taken as the dynamic threshold for that day. When the LST anomaly value of a pixel exceeds its corresponding threshold on a given day and this condition persists for at least three consecutive days, it is defined as a heatwave event. Based on the identified heatwave events, this study focuses on the summer season (May\u0026ndash;September) and calculates the following three key indicators: heatwave frequency (HWF), representing the number of heatwave events per year; heatwave duration (HWD), indicating the total number of heatwave days in a year; and heatwave cumulative intensity (HWC), measuring the cumulative excess temperature exceeding the threshold during heatwave periods.\u0026nbsp;\u003c/p\u003e\n\u003ch3\u003e2.2.3. Urban heat environment resilience indicators\u003c/h3\u003e\n\u003cp\u003eOur approach to defining urban heat resilience indicators is grounded in established climatological principles and resilience theory. The World Meteorological Organization (WMO) recognizes statistical measures like standard deviation as effective for describing climate variability (World Meteorological Organization, 2018), providing a foundation for our \u0026lsquo;Stability\u0026rsquo; metric. This study operates core concepts from resilience science. Although \u0026apos;resistance\u0026apos; and \u0026apos;stability\u0026apos; are often considered two key perspectives in resilience science for evaluating system responses (Bruneau et al., 2003; Cutter, 2016), their application to the urban thermal environment has been largely conceptual, lacking standardized, LST-based quantitative metrics to assess the system\u0026apos;s intrinsic thermal performance. To fill this gap, this study proposes and calculates two specific thermal resilience indicators: (1) \u003cem\u003eStability\u003c/em\u003e (ST): This indicator measures the inherent variability of the system during non-heatwave periods. It is calculated as the standard deviation of the deviation between the daily land surface temperature (LST) and its multi-year average baseline (i.e., the average LST of non-heatwave days over the same period) during all non-heatwave days in summer. A smaller ST value indicates greater thermal stability of the system under normal conditions. (2) \u003cem\u003eResistance\u003c/em\u003e (RS): This indicator assesses the system\u0026apos;s ability to suppress extreme temperature increases during heatwave impacts. It is calculated as the average increase in peak LST relative to the baseline LST on the same day during all heatwave events in summer. A smaller RS value indicates stronger resistance to heatwave impacts.\u003c/p\u003e\n\u003ch3\u003e2.2.4. Selection of influencing factors\u003c/h3\u003e\n\u003cp\u003eTo comprehensively and scientifically analyze the influencing factors of heatwaves, this study systematically selected and integrated influencing factors from three dimensions: key climate factors, urban land surface characteristics, and human socio-economic activities.\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003e\u003cstrong\u003eClimate factors\u003c/strong\u003e: Air temperature (Ta), precipitation (Pre), and surface solar radiation (SSR) serve as the basic metrics for heatwave intensity, regulators of surface energy balance, and the primary energy source for surface warming, respectively (Nairn et al., 2014).\u003c/li\u003e\n \u003cli\u003e\u003cstrong\u003eUrban land surface characteristics\u003c/strong\u003e: To characterize their moderating effects on the local thermal environment, the following factors were included: normalized difference vegetation index (NDVI) and impervious surface ratio, which reflect the cooling and warming effects of land cover (De Razza et al., 2024);Digital Elevation Model (DEM) to characterize terrain effects\u0026nbsp;(WANG M et al., 2018); Landscape Pattern Indices (PLAND, PD, LSI) to quantify the area, fragmentation, and shape complexity of blue-green spaces\u0026nbsp;(Kong et al., 2025); and Aerosol Optical Depth (AOD) and near-surface wind speed (WS) to represent atmospheric turbidity\u0026nbsp;(Gil-D\u0026iacute;az et al., 2025)and local ventilation capacity\u0026nbsp;(Luo et al., 2023), respectively.\u003c/li\u003e\n \u003cli\u003e\u003cstrong\u003eSocioeconomic factors\u003c/strong\u003e: To characterize human activity impacts, nighttime lighting (NL) and carbon emissions (CE) were selected as proxy indicators for economic activity and anthropogenic heat release (Debnath et al., 2025), and population density (PopDen) was chosen to indicate potential anthropogenic heat source intensity and population exposure risk (Wang et al., 2021). The specific sources, spatio-temporal resolution, and processing methods for each influencing factor are detailed in Appendix 1.\u003c/li\u003e\n\u003c/ul\u003e"},{"header":"3. Methodology","content":"\u003cp\u003eThis study aims to investigate the spatial and temporal characteristics of heat waves and urban heat resilience in Chengdu, as well as their influencing factors. The study adopts the framework diagram shown in Figure 2. Initially, the data was preprocessed, and the heat wave index was extracted and calculated. Then, the temporal, spatial and spatial characteristics of heat waves were analyzed using various methods. Finally, the influencing factors of heat waves and urban heat resilience were analyzed.\u003c/p\u003e\n\u003ch2\u003e\u003cem\u003e3.1 Temporal and spatial characteristics and resilience index analysis\u003c/em\u003e\u003c/h2\u003e\n\u003ch3\u003e3.1.1 Trend analysis\u003c/h3\u003e\n\u003cp\u003eTo investigate the trends and significance of various heatwave indicators during the study period, this study combined Theil-Sen Median trend analysis with the Mann-Kendall (MK) test. This combined method has no special requirements for data distribution, is computationally robust, and can effectively resist the interference of outliers, making it suitable for trend analysis of long-term time series(Shao et al., 2024). By calculating the trend slope (\u003cimg width=\"9\" height=\"19\" src=\"data:image/png;base64,R0lGODlhDQAcAHcAMSH+GlNvZnR3YXJlOiBNaWNyb3NvZnQgT2ZmaWNlACH5BAEAAAAALAEABgAMABYAhAAAAAAAAAAAOgAAZgA6ZgA6kABmkABmtjoAADpmkDpmtjqQtjqQ22YAAGY6Oma2/5A6AJC225Db/7ZmALbbtrbb/7b//9uQOtu2Ztu2kNvbttv///+2Zv/bkP//tv//2wVuICBmThAIyiaKV7CoWFCoQIfM6xQwbPCsAE7gAPhABBKgkGhEAnU/T2NgWdlwUqoIgyBUAVmLLWCorMIiDSSQUKFzQ/D0K7JR3yIpEl9D3BtOK0IzUicRIwhOWRQlJwlmclpAZ3OTlJKWfECaKyEAOw==\" alt=\"image\"\u003e) and significance statistic (\u003cimg width=\"10\" height=\"19\" src=\"data:image/png;base64,R0lGODlhDwAcAHcAMSH+GlNvZnR3YXJlOiBNaWNyb3NvZnQgT2ZmaWNlACH5BAEAAAAALAAABwAOABAAhAAAAAAAAAAAOgAAZgA6kABmtjoAADqQtjqQ22YAAGYAOma2/5A6AJDb/7ZmALb//9uQOtu2Ztv///+2Zv/bkP//tv//2wECAwECAwECAwECAwECAwECAwECAwECAwECAwVPICBOQWmahQg4wSGpUDA8osUQr0gZwaJWisYvEUCojioWDnmMCYRMHc8XBVSIqeoqsKySntpdT2sLZKus2ZAK+EJFEDVAbFRFuDnGaX8OAQA7\" alt=\"image\"\u003e) and setting the significance level\u0026nbsp;\u003cimg width=\"64\" height=\"19\" src=\"data:image/png;base64,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\" alt=\"image\"\u003e\u0026nbsp;(\u003cimg width=\"78\" height=\"19\" src=\"data:image/png;base64,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\" alt=\"image\"\u003e), this method identifies the spatial patterns of significant increases, significant decreases, or no significant changes in each indicator.\u003c/p\u003e\n\u003ch3\u003e3.1.2 Spatial autocorrelation analysis\u003c/h3\u003e\n\u003cp\u003eTo explore the spatial distribution patterns of the trends of the indicators, spatial autocorrelation analysis was further introduced in this study. Moran's index (Moran's I) was used to examine the spatial distribution patterns of heat wave indicators and their trends. The global Moran's index measures the overall spatial aggregation, while the localized Moran's index (LISA) further identifies the hot (HH) and cold (LL) regions, which are used to reveal local anomalous regions. The formula for calculating the Moran index is as follows:\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp; \u003cimg width=\"133\" height=\"35\" src=\"data:image/png;base64,R0lGODlhyAA1AHcAMSH+GlNvZnR3YXJlOiBNaWNyb3NvZnQgT2ZmaWNlACH5BAEAAAAALAAABADIADEAhQAAAAAAAAAAOgAAZgA6ZgA6kABmtjoAADoAOjoAZjo6ZjpmkDpmtjqQkDqQ22YAAGYAOmYAZmY6AGY6OmY6kGZmZmZmtmaQkGaQtmaQ22a222a2/5A6AJA6OpBmAJBmOpDb/7ZmALZmOrZmZraQZrbb27bb/7b/27b//9uQOtuQZtu2Ztu2ttvbttvb29v/ttv///+2Zv/bkP/btv//tv//2wECAwECAwECAwECAwECAwECAwECAwECAwECAwECAwb/QIBwSCwaj8ikcomsiQKG2idQgDGvRCdUSrViv+CweEwuM1eaGKFiiglA5iNazXbD4/i8fs8HpAYoADGAfUJ/gYOBhYuMjVgpAZGSklUANRwGQiGVYJCTk5WXmQCbXo6nqI6XARtEMgeVNA+tlw5iq61Dr7Gzlhy2qcHCfK+EQymVdgCyG2hhxYqGyW/Ls87D2NlikKNCMZWHvgHAYdxE31bhq+Ta7e5NHKyq8bmLkMbKr/Jgsp/dRP0m/XtnBpojg4xCUCO1L1w1droO5JKBAM4lTrKAUbTIgZOQV/WyqThAAMUMCV2YhIByauVAPjHk9csVghzII4m8ZbpUMmKu/xg7OfQ8Z0xkhlcXLMBQtkRW0SEZr2xM4jRakZtLphaJ2SpFgwe2ZCSwyk+CRyQ0zJoiuJWln4VKmB3BqiQmXCNykeSsG+DuxwMOZCg4gYlUvZVnxbFLexaxKcZrFRNMsTDEUyQhXoqhAeHOkcx5OHuG+sBAh1aZUzDI8utI3mVq/QCrBTV2isW93F20Ikvz1suaUjIRrRe4kHVXiBdZtdqPcCGvXVWEHntG7ryQAVgPqbVd778QjdBd3lrcp33Kr0qM22vV+VzpiVhWRHm0IL+GjK2ctJDpfkl3OeTNJCExgpUdK4R3XHlQRRBIdHF1ZgRtADmoy3TDSTgGMpFp4v9bERxiAVoRK0SiQSr1UScABh1KE1k4TGVonx+JCShghDOKaMAMFrygYVQq7ahUU2AZEVNibAlyHiXpcOLeJ/BpOERMSz434pMEQidlOQEQYAINFgIwXhKQeMnEmM59qFs8VVaSGY8tFhEflp9gVGRyW5JBgmc35tEnQwomyUSZJpx5QCQFSrWeoYjucZKaYzw6IQf4CWrpn8GstECcZGjaIUJJUFklgDlaWoakpjoSE6SpZuNpq42sFCistNZKxiWV8jFqlUTs6uuvwAYr7LDEFvsrdA8YZ+uyzDJxJKdTAptrs9SiOE612JJBJauN4JLtt1/IsNors56ywqHlgqv/LgAseBEiNqJeu+4wWkQxBZJ4oDPvviCOWsUca7QxbRm38RsuSn0tQBZb3kbECYzKlkEDBdAaHG0BJbQgQcTugEqjFaJogq8ZIyxs8Wf4jDyZW97wQguDeqSQ6MmfqWxpw1tRw8w1WAQkySgxABPDzDQPWEChy3qcXyDIgTHVbs5JwjHNIwlw4oLPJenSIkD5MlTRYcxAjya2pMCtMFUxkh3YY3wn1oNhJgkhsumKiaGcsQF0J9tb2ZmJvpcQrc2I4i0aal8zZlfwX4IXDYkGLrjgATVmL9j4MHtJnGd22/F9xgSSKIB05b5cHgyawdmspX3/RVJZ1p7XFQoHpqNCUmGDTLdGZ6PL5KnE7bFnNZaYw09WIyFzu+Y7VbkFr1LZdaMSb51WxIjjF9Y7n8QqZ8/jb5Mgswnl6l+8qz3Nbw55vY5wnn8yoVkdus+gXSKtTRAAOw==\" alt=\"image\"\u003e\u0026nbsp;\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;(1)\u003c/p\u003e\n\u003cp\u003eIn the formula,\u0026nbsp;\u003cimg width=\"13\" height=\"19\" src=\"data:image/png;base64,R0lGODlhEwAcAHcAMSH+GlNvZnR3YXJlOiBNaWNyb3NvZnQgT2ZmaWNlACH5BAEAAAAALAAADAARABAAhAAAAAAAAAAAOgAAZgA6kABmtjoAADpmtjqQ22YAAGa222a2/5A6AJC225Db/7ZmALZmOrb/27b//9uQOtu2Ztv///+2Zv/bkP//tv//2wECAwECAwECAwECAwECAwECAwVaIABYQWkGgiOu4iM06xMg7JoxSxwcdS9OgYKvByRUao+BhEUy1m5O0cWgHLKmKesKk8i2hABgIHcT5ESU8WpSFZ9QqlZUC+DS6FLDGT/x4pMREHtWFgYCCgAhADs=\" alt=\"image\"\u003e\u0026nbsp;is the deviation between the attribute of element\u0026nbsp;\u003cimg width=\"5\" height=\"19\" src=\"data:image/png;base64,R0lGODlhCAAcAHcAMSH+GlNvZnR3YXJlOiBNaWNyb3NvZnQgT2ZmaWNlACH5BAEAAAAALAEACAAGAA8AhAAAAAAAAAAAOgAAZgA6kABmtjoAADqQ22YAAGa222a2/5A6AJDb/7ZmALaQOrb//9v///+2Zv/bkP//tv//2wECAwECAwECAwECAwECAwECAwECAwECAwECAwECAwECAwUuICA2wSFSi8CIbMtOSFCKsEojwyNKBgGJkdkooDgtcrveDxAssJoUxwEVECQgIQA7\" alt=\"image\"\u003e\u0026nbsp;and its average value\u0026nbsp;\u003cimg width=\"43\" height=\"19\" src=\"data:image/png;base64,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\" alt=\"image\"\u003e,\u0026nbsp;\u003cimg width=\"25\" height=\"21\" src=\"data:image/png;base64,R0lGODlhJQAfAHcAMSH+GlNvZnR3YXJlOiBNaWNyb3NvZnQgT2ZmaWNlACH5BAEAAAAALAAADAAjABMAhAAAAAAAAAAAOgAAZgA6ZgA6kABmtjoAADoAOjpmZjqQkDqQ22YAAGYAOmZmOma222a2/5A6AJDb/7ZmALZmOrb/27b//9uQOtu2Ztv///+2Zv/bkP//tv//2wECAwECAwW6oBYERQacV7CcHDNCZyzPMhYMVjyp8cbgtOCsBQS0AobeYSVsGhkCyUmEjGmKzmAnEgVsFYySN8I8TbA0EQywg10rEbFGfNrSg+ccOwBpQbY4fllZKQsTYhNRE0mDTiIJCFIAFwIOd5MvQlRlmzEpXTMXaDNqSnciZTKHJkKieo0yLak0q7AzGwdrQbK2oaBBuLq9bAMVFH9kMpSSwwAaBwIPY6m1zTQcDcwtjNYzxzEdFL/deAEEzAAhADs=\" alt=\"image\"\u003e\u0026nbsp;is the spatial weight between elements\u0026nbsp;\u003cimg width=\"5\" height=\"19\" src=\"data:image/png;base64,R0lGODlhCAAcAHcAMSH+GlNvZnR3YXJlOiBNaWNyb3NvZnQgT2ZmaWNlACH5BAEAAAAALAEACAAGAA8AhAAAAAAAAAAAOgAAZgA6kABmtjoAADqQ22YAAGa222a2/5A6AJDb/7ZmALaQOrb//9v///+2Zv/bkP//tv//2wECAwECAwECAwECAwECAwECAwECAwECAwECAwECAwECAwUuICA2wSFSi8CIbMtOSFCKsEojwyNKBgGJkdkooDgtcrveDxAssJoUxwEVECQgIQA7\" alt=\"image\"\u003e\u0026nbsp;and\u003cimg width=\"9\" height=\"19\" src=\"data:image/png;base64,R0lGODlhDgAcAHcAMSH+GlNvZnR3YXJlOiBNaWNyb3NvZnQgT2ZmaWNlACH5BAEAAAAALAMACAAKABQAhAAAAAAAAAAAOgAAZgA6kABmtjoAADqQ22YAAGaQtma2/5A6AJDb/7ZmALb//9uQOtu2kNv///+2Zv/bkP//tv//2wECAwECAwECAwECAwECAwECAwECAwECAwECAwECAwU+ICCOgBQU5NgESuq+brUEZzy3qSwwLoUMDtfEgHI9AofXCkeSAYUGQsRlKqZWydxiB5UakS5IdKoKCBLkUQgAOw==\" alt=\"image\"\u003e,\u0026nbsp;\u003cimg width=\"9\" height=\"19\" src=\"data:image/png;base64,R0lGODlhDgAcAHcAMSH+GlNvZnR3YXJlOiBNaWNyb3NvZnQgT2ZmaWNlACH5BAEAAAAALAAADAAOAAsAhAAAAAAAAAAAOgAAZgA6ZgA6kABmtjoAADo6ADo6OjpmtjqQ22YAAGa222a2/5A6AJA6OpC225Db/7ZmALZmOrb//9uQOtuQZtu2Ztv///+2Zv/bkP/btv//tv//2wECAwVKIIAhQZGJR0BUAHApmRU4VAPIRqtPQuK0mkBO53kIdcFhq8MYsFqTmQ6wOSiLAslUttBVTdPoD3jUMZ07qfcAhs44kF9yWlVFACEAOw==\" alt=\"image\"\u003e\u0026nbsp;is equal to the total number of elements, and\u0026nbsp;\u003cimg width=\"15\" height=\"19\" src=\"data:image/png;base64,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\" alt=\"image\"\u003e\u0026nbsp;is the aggregation of all spatial weights.\u003c/p\u003e\n\u003ch2\u003e\u003cem\u003e3.2. Factor Analysis of Heat Wave and Toughness Indicators\u003c/em\u003e\u003c/h2\u003e\n\u003ch3\u003e3.2.1. GeoDetector model\u003c/h3\u003e\n\u003cp\u003eTo investigate the driving forces behind the spatial differentiation of heatwave characteristics, this study employs the Geodetector model. This model is particularly adept at analyzing interactions between factors and does not require the numerous assumptions of traditional linear regression. The Geodetector model has significant advantages in handling spatial data and complex variable interactions. The specific applications in this study include: (1) Factor Detector, which uses the statistic to measure the explanatory power of individual factors; (2) Interaction Detector, which identifies the type (e.g., enhancing, nonlinear) and strength of two-factor interactions by comparing changes in\u0026nbsp;\u003cimg width=\"9\" height=\"19\" src=\"data:image/png;base64,R0lGODlhDgAcAHcAMSH+GlNvZnR3YXJlOiBNaWNyb3NvZnQgT2ZmaWNlACH5BAEAAAAALAAADAANABAAhAAAAAAAAAAAOgAAZgA6kABmtjoAADo6ADpmtjqQ22YAAGY6OmaQtma2/5A6AJA6OpBmZpDbtpDb/7ZmALZmOraQOrbb/7b//9uQOtu2Ztv///+2Zv/bkP/btv//tv//2wVcICAC3RIEAlMV2pgFhAVwRpCMNNECnxM0op5AMuoNLqINbAfwKHQijG0EUBZGkx81C+T5ujPDEKvlPQ7HkVXTgUQUVyrlhNAob9T8ZJwvOtJ9IjlMfVaBZAJ1IiEAOw==\" alt=\"image\"\u003evalues.\u003c/p\u003e\n\u003ch3\u003e3.3.2. GWRF+shap model\u003c/h3\u003e\n\u003cp\u003eTo further reveal the spatial heterogeneity of ecological resilience determinants, a geographically weighted random forest (GWRF) model was constructed, integrating the nonparametric modeling capabilities of random forests with the local regression concept of GWR. The model establishes local random forests using spatially weighted samples, with Gaussian kernel functions defining the weights, thereby enhancing adaptability to non-stationary relationships (Fotheringham et al., 2017).The bandwidth is automatically selected using pseudo leave-one-out cross-validation to balance accuracy and generalization ability (Wu et al., 2024). To explain the local prediction results of the GWRF “black box” model, this study integrates the SHAP (SHapley Additive exPlanations) method. This method quantifies the marginal contribution of each driver to the model output at each spatial location, thereby identifying the dominant influencing factors in different regions and their direction of influence.\u003c/p\u003e"},{"header":"4 Results","content":"\u003ch2\u003e\u003cem\u003e4.1 Trend patterns of heat wave indicators\u003c/em\u003e\u003c/h2\u003e\n\u003cp\u003eThis study employs Theil-Sen Median analysis and Mann-Kendall tests to conduct trend analysis and statistical analysis of heatwaves and heat resilience indicators. The results indicate that over the past two decades, both heatwaves and resilience indicators have exhibited a \u0026lsquo;central peripheral\u0026rsquo; spatial pattern, with trends peaking in urban centers and decreasing toward the periphery. More critically, heatwave intensification zones and resilience degradation zones exhibit high spatial overlap, particularly in urban centers and the southeastern regions, forming distinct \u0026lsquo;dual vulnerability\u0026rsquo; zones. This transformation has rendered urban core areas into \u0026lsquo;hotspots\u0026rsquo; characterized by the concentrated risks of thermal environmental hazards and ecological fragility.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo illustrate this composite risk, Figure 3\u0026nbsp;and\u0026nbsp;Figure 4 highlight the spatial trends of indicators representing the comprehensive severity of heatwaves (HWC) and indicators directly reflecting the ability to resist impacts (RS). Areas with significant increases in HWC and significant decreases in RS are highly consistent, directly confirming the spatial coupling relationship. Overall, these phenomena reveal that Chengdu is currently facing a negative feedback loop of \u0026lsquo;h\u003cem\u003eeatwave intensification-resilience decline\u003c/em\u003e\u0026rsquo; driven by urbanization: urban development has amplified the urban heat island effect, leading to more severe heatwaves, while simultaneously eroding the ecosystem\u0026apos;s inherent regulatory capacity, further exacerbating the accumulation of risks. (For detailed trends and spatial distributions of other indicators, please refer to Appendix Figures S1-S3.)\u003c/p\u003e\n\u003ch2\u003e\u003cem\u003e4.2 Spatial autocorrelation analysis\u003c/em\u003e\u003c/h2\u003e\n\u003cp\u003eSpatial autocorrelation analysis (Table 1) revealed that all indicators exhibited extremely strong positive spatial autocorrelation (Moran\u0026apos;s I: 0.466-0.658, p \u0026lt;0.01), confirming that the deterioration of the thermal environment and the decline in resilience are not randomly distributed in space but have formed a significant clustering pattern.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eLocal spatial autocorrelation (LISA) analysis further revealed the specific spatial distribution characteristics of this clustering pattern (Figure 5): a contiguous high-risk zone composed of \u0026ldquo;hotspots\u0026rdquo; (H-H clustering) precisely covers the city center and southern expansion zone. This directly reflects the profound impact of high-intensity urbanization on local climate and has formed a negative feedback core of \u0026ldquo;heatwave intensification-resilience decline\u0026rdquo; in this region. In contrast, \u0026ldquo;cold spots\u0026rdquo; (L-L clustering) are primarily distributed in the northwestern Longmen Mountain ecological conservation area, where the stable thermal environment serves as a natural barrier against external high temperatures. Notably, within the \u0026ldquo;hot spot\u0026rdquo; core area, multiple \u0026ldquo;low-high\u0026rdquo; (L-H) outliers corresponding to large parks and water bodies are clearly visible, indicating that urban blue-green spaces continue to play a critical local cooling \u0026ldquo;cool island\u0026rdquo; effect even under overall deteriorating conditions. However, a deeper analysis of the LISA pattern reveals that the regulatory roles of different geographical elements exhibit significant indicator specificity. The Longmen Mountains in the northwest exhibit unique advantages in maintaining temperature stability, with only the stability (ST) indicator forming a large-scale L-L cold spot cluster (Figure 5). This is primarily due to the synergistic regulatory effects of its high-altitude terrain and dense forest vegetation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHowever, its mitigating effects on other indicators such as heatwave frequency and duration are not significant, indicating that its ecological functions should not be generalized. In contrast, the Longquan Mountains in the eastern part of the city, although not forming a large-scale cold spot core, serve as an important ecological corridor, with surrounding areas showing L-L clustering or relatively stable trends in multiple indicators. Under the backdrop of high-intensity urbanization, it effectively blocks the uncontrolled spread of the urban heat island effect, providing critical ecological support to break the negative feedback loop in the urban core. Based on this spatial differentiation, policymaking should adopt differentiated strategies: implement high-intensity ecological restoration in the \u0026ldquo;hotspot\u0026rdquo; areas of the urban center, strictly protect the Longquan Mountains and leverage their ecological corridor functions, and prioritize maintaining the temperature stabilization capabilities of the transition zone in the northwestern Longmen Mountains.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026nbsp;Table 1\u003c/strong\u003e Results of global spatial autocorrelation analysis of heat wave indicators and thermal resilience indicators\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"99%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.449%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003cstrong\u003eIndicator\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28.5714%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;Moran index\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.449%;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003cstrong\u003e\u003cem\u003eP\u003c/em\u003e value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 26.5306%;\"\u003e\n \u003cp\u003e\u003cem\u003e\u0026nbsp;\u003cstrong\u003eZ\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e\u0026nbsp;value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.449%;\"\u003e\n \u003cp\u003e\u0026nbsp;HWC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28.5714%;\"\u003e\n \u003cp\u003e\u0026nbsp;0.527946\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.449%;\"\u003e\n \u003cp\u003e\u0026nbsp;0.000000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 26.5306%;\"\u003e\n \u003cp\u003e\u0026nbsp;88.262037\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.449%;\"\u003e\n \u003cp\u003e\u0026nbsp;HWF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28.5714%;\"\u003e\n \u003cp\u003e\u0026nbsp;0.465567\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.449%;\"\u003e\n \u003cp\u003e\u0026nbsp;0.000000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 26.5306%;\"\u003e\n \u003cp\u003e\u0026nbsp;77.836641\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.449%;\"\u003e\n \u003cp\u003e\u0026nbsp;ST\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28.5714%;\"\u003e\n \u003cp\u003e\u0026nbsp;0.658267\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.449%;\"\u003e\n \u003cp\u003e\u0026nbsp;0.000000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 26.5306%;\"\u003e\n \u003cp\u003e\u0026nbsp;110.050951\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.449%;\"\u003e\n \u003cp\u003e\u0026nbsp;RS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28.5714%;\"\u003e\n \u003cp\u003e\u0026nbsp;0.449707\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.449%;\"\u003e\n \u003cp\u003e\u0026nbsp;0.000000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 26.5306%;\"\u003e\n \u003cp\u003e\u0026nbsp;75.186666\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 22.449%;\"\u003e\n \u003cp\u003e\u0026nbsp;HWD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28.5714%;\"\u003e\n \u003cp\u003e\u0026nbsp;0.514865\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 22.449%;\"\u003e\n \u003cp\u003e\u0026nbsp;0.000000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 26.5306%;\"\u003e\n \u003cp\u003e\u0026nbsp;86.075111\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003e\u003cem\u003e4.3 Geodetector analysis\u003c/em\u003e\u003c/h2\u003e\n\u003cp\u003eThe factor analysis results have shown in Table 2. The result indicates that landscape composition (PLAND) is the dominant factor shaping the spatial variation of all heatwave and resilience indicators. Its explanatory power (q-values ranging from 0.564 to 0.761) is shown in Table 2. This confirms that the macro-scale pattern of the urban thermal environment is primarily determined by land use types (e.g., the ratio of impervious surfaces to blue-green spaces). In addition to PLAND, the secondary driving factors for each indicator exhibit significant differences, revealing that the formation mechanisms of heatwaves and the resilience mechanisms of the thermal environment are regulated by different factors. Specifically, the direct characteristics of heatwaves (such as frequency HWF, intensity HWC, and duration HWD) are largely influenced by immediate atmospheric and meteorological factors, such as aerosol (AOD), wind speed (WS), and solar radiation (SR), which exhibit strong explanatory power. This indicates that the formation and dissipation of heatwaves are direct responses to short-term energy and atmospheric changes based on surface patterns. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHowever, when assessing the inherent thermal environmental resilience of the system, the importance ranking of driving factors is notably restructured. System stability (ST), referring to the degree of fluctuation under normal conditions, is primarily constrained by relatively stable factors such as the background climate (ambient temperature, AT), local topography (digital elevation model, DEM), and baseline ventilation conditions (wind speed, WS). In contrast, system resistance (RS), defined as the ability to withstand external shocks, exhibits greater sensitivity to atmospheric pollution (aerosol optical depth, AOD) and hydrological conditions (precipitation, Pre). This differentiation in driving mechanisms highlights that the formation of heatwave events and the maintenance of urban thermal environmental resilience are two related but distinct urban eco-climatic processes.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2\u003c/strong\u003e Explanatory power of various driving factors on the spatial differentiation of heat waves and resilience indicators (\u003cem\u003eq\u003c/em\u003e statistic)\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"633\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 9.62145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eIndicator Variable\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5.83596%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTa\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.46688%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNDVI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePre\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5.83596%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNL\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.62461%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDEM\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eISP\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.62461%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAOD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.46688%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAOD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8.35962%;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePopDen\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSSR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.78233%;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePland\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLSI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 9.62145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eHWF\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e0.009\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5.83596%;\"\u003e\n \u003cp\u003e0.022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.46688%;\"\u003e\n \u003cp\u003e0.020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e0.027\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5.83596%;\"\u003e\n \u003cp\u003e0.031\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.62461%;\"\u003e\n \u003cp\u003e0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e0.010\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.62461%;\"\u003e\n \u003cp\u003e0.085\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.46688%;\"\u003e\n \u003cp\u003e0.064\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8.35962%;\"\u003e\n \u003cp\u003e0.010\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e0.054\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.78233%;\"\u003e\n \u003cp\u003e0.717\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e0.021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e0.013\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 9.62145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eHWC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e0.006\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5.83596%;\"\u003e\n \u003cp\u003e0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.46688%;\"\u003e\n \u003cp\u003e0.022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e0.014\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5.83596%;\"\u003e\n \u003cp\u003e0.023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.62461%;\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.62461%;\"\u003e\n \u003cp\u003e0.046\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.46688%;\"\u003e\n \u003cp\u003e0.032\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8.35962%;\"\u003e\n \u003cp\u003e0.005\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e0.027\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.78233%;\"\u003e\n \u003cp\u003e0.564\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e0.010\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e0.012\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 9.62145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eHWD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e0.004\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5.83596%;\"\u003e\n \u003cp\u003e0.010\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.46688%;\"\u003e\n \u003cp\u003e0.063\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e0.021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5.83596%;\"\u003e\n \u003cp\u003e0.022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.62461%;\"\u003e\n \u003cp\u003e0.005\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e0.010\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.62461%;\"\u003e\n \u003cp\u003e0.082\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.46688%;\"\u003e\n \u003cp\u003e0.057\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8.35962%;\"\u003e\n \u003cp\u003e0.010\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e0.032\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.78233%;\"\u003e\n \u003cp\u003e0.748\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e0.014\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e0.013\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 9.62145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eStability\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e0.029\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5.83596%;\"\u003e\n \u003cp\u003e0.154\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.46688%;\"\u003e\n \u003cp\u003e0.045\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5.83596%;\"\u003e\n \u003cp\u003e0.048\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.62461%;\"\u003e\n \u003cp\u003e0.105\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e0.019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.62461%;\"\u003e\n \u003cp\u003e0.066\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.46688%;\"\u003e\n \u003cp\u003e0.103\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8.35962%;\"\u003e\n \u003cp\u003e0.020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e0.026\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.78233%;\"\u003e\n \u003cp\u003e0.761\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e0.043\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 9.62145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eResistance\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e0.014\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5.83596%;\"\u003e\n \u003cp\u003e0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.46688%;\"\u003e\n \u003cp\u003e0.040\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e0.138\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 5.83596%;\"\u003e\n \u003cp\u003e0.040\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.62461%;\"\u003e\n \u003cp\u003e0.031\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e0.018\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.62461%;\"\u003e\n \u003cp\u003e0.217\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.46688%;\"\u003e\n \u003cp\u003e0.079\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 8.35962%;\"\u003e\n \u003cp\u003e0.016\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.15142%;\"\u003e\n \u003cp\u003e0.096\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.78233%;\"\u003e\n \u003cp\u003e0.735\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e0.025\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.30915%;\"\u003e\n \u003cp\u003e0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eInteraction detection results (Figure 6) indicate that the interaction between any two driving factors exhibits either a two-factor enhancement or a nonlinear enhancement, with their explanatory power for the dependent variable typically far exceeding the sum of the individual factors, highlighting the complex synergistic nature of urban thermal environment driving mechanisms (Wang et al., 2022). Among these, the interaction effect of landscape composition (PLAND) is particularly prominent. The interactions between PLAND and any other factor (such as AOD, NTL, etc.) almost always form the strongest combination in explaining the dependent variable. This indicates that PLAND is not only a dominant direct driver but also a foundational \u0026ldquo;amplifier\u0026rdquo;, determining the extent to which other factors (such as anthropogenic heat and air pollution) can influence the local thermal environment. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSpecific interaction combinations reveal several key driving pathways: First, the synergistic intensification effect of urbanization factors is significant. For example, the strong interaction between nighttime lighting (NL) and impervious surfaces (ISP) (NL\u0026cap;ISP) precisely points to the process by which high-density built-up areas, due to building heat storage and high-intensity anthropogenic heat emissions, jointly exacerbate thermal environment degradation at night. Second, the composite regulatory effects of natural geography and meteorological conditions are complex. For example, the interaction between wind speed and terrain (WS\u0026cap;DEM) reveals the key role of ventilation corridors in heat dissipation (Lu et al., 2020); while the interaction between aerosols and solar radiation (AOD\u0026cap;SSR) demonstrates the bidirectional regulatory mechanism of atmospheric transparency on surface energy balance\u0026mdash;high AOD reduces daytime solar radiation input but may also exacerbate nighttime insulation effects by absorbing long-wave radiation, especially under weak wind conditions (AOD\u0026cap;WS) (Singh et al., 2024). \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFrom the perspective of thermal environmental resilience, the interaction mechanism is even more unique. The system\u0026apos;s resistance (RS), i.e., its ability to withstand heatwave impacts, has been proven to be primarily dominated by the nonlinear synergistic effects of terrain (DEM), aerosols (AOD), and solar radiation (SSR). This finding may be related to Chengdu\u0026apos;s unique basin topography. This topography acts as a natural \u0026ldquo;container\u0026rdquo;, suppressing near-surface atmospheric advection and convection, significantly amplifying the radiative and thermodynamic effects of aerosols, causing polluted air masses and heat to linger over the city for extended periods, thereby systematically weakening the city\u0026apos;s overall resistance.\u003c/p\u003e\n\u003ch2\u003e\u003cem\u003e4.4 GWRF+SHAP Analysis\u003c/em\u003e\u003c/h2\u003e\n\u003cp\u003eThis section selects the years 2004, 2015, and 2023 as temporal snapshots for analysis, each representing the decadal trend of change. This section focuses on the spatiotemporal differentiation of the two dominant factors, PLAND and AOD, identified by the GeoDetector model as key drivers of cumulative heatwave intensity in Chengdu.\u003c/p\u003e\n\u003cp\u003eAs shown in Figure 7, the influence of PLAND on HWC underwent a fundamental reversal during the study period. In 2004, PLAND exhibited a significant cooling effect (SHAP values were negative) across the entire study area. However, as the city expanded rapidly toward the south and southeast, by 2023, its positive warming effect had formed large contiguous areas in the urban core and suburban regions, while the original cooling effect had retreated to the peripheral mountainous areas. This reversal of effects reveals the paradox of urbanization: although green spaces are traditional cooling measures, in high-density built-up areas, green space fragmentation, soil drying, and urban canyon effects collectively weaken their evapotranspiration cooling function (Li et al., 2022), transforming them from \u0026ldquo;cooling sources\u0026rdquo; into potential \u0026ldquo;heat accumulation zones\u0026rdquo;. This effect is further exacerbated by the unfavorable low wind speeds in basins, intensifying local heat risks.\u003c/p\u003e\n\u003cp\u003eThe regulatory effect of AOD on HWC shown in Figure 8, which exhibits significant context dependence, transitioning from a \u0026ldquo;shading cooling\u0026rdquo; (umbrella effect) to a \u0026ldquo;heating warming\u0026rdquo; (blanket effect) under different spatiotemporal conditions. In most regions in 2004 and 2015, AOD primarily reflected shortwave solar radiation, resulting in a cooling effect. However, by 2023, under extreme high-temperature drought conditions and specific atmospheric circulation patterns (such as the westward extension of the subtropical high-pressure system), the absorption of longwave radiation by absorbent aerosols and their obstruction of heat dissipation became prominent, triggering AOD\u0026apos;s warming \u0026ldquo;blanket\u0026rdquo; effect, particularly in the northwestern foothills of Chengdu, where heat and pollutants tend to linger. Notably, since 2015, Chengdu\u0026apos;s sustained pollution control efforts have significantly reduced AOD concentrations. This may explain why the cooling effect of AOD was maintained or even reappeared in the eastern region in 2023, highlighting the complex role of pollution control in regulating heatwave risks.\u003c/p\u003e\n\u003cp\u003eIn summary, the impact of AOD on heatwaves is not static or consistent but is subject to the coupled regulation of multiple factors, including the macro-thermal environmental background, atmospheric circulation patterns, and the effectiveness of pollution control measures. Future urban heatwave management should fully consider the context-dependent characteristics of aerosols, identify their response pathways under different conditions, and avoid making blanket judgments about their regulatory effects to achieve more precise pollution control and heatwave adaptation strategies.\u003c/p\u003e"},{"header":"5 Discussion and Conclusion","content":"\u003cp\u003eBased on high-resolution remote sensing data from 2004 to 2023 and a newly developed evaluation index system for thermal environmental resilience, this study systematically analyzes the spatiotemporal evolution of heatwaves and their complex driving mechanisms in Chengdu. The core findings reveal that the city is undergoing a simultaneous deterioration process characterized by intensified heatwaves and declining thermal environmental resilience. These two phenomena are highly spatially coupled, giving rise to \u0026ldquo;dual-risk zones\u0026rdquo; centered around the urban core and expansion areas.\u003c/p\u003e\n\u003cp\u003eThe pattern of land surface landscape (PLAND) is identified as the absolute dominant factor determining the spatial heterogeneity of the thermal environment. However, the formation of heatwaves and the maintenance of resilience are governed by different combinations of factors, reflecting a divergence in driving mechanisms. Notably, this study uncovers the non-stationarity and scenario dependency of key driving factors. For instance, under extreme hot and arid conditions, atmospheric aerosols (AOD) can shift from acting as a cooling \u0026lsquo;parasol\u0026rsquo; to a warming \u0026lsquo;blanket\u0026rsquo;, thereby significantly amplifying local thermal risks. These findings underscore the need to move beyond a \u0026lsquo;one-size-fits-all\u0026rsquo; approach in urban thermal management and instead implement spatially differentiated, precision-based strategies. Specific policy recommendations include:\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003eDevelop precision adaptation plans based on spatial heterogeneity. Integrate identified high-risk \u0026lsquo;hotspot areas\u0026rsquo; into climate-adaptive land use zoning frameworks. Enforce the construction of highly connected blue-green ecological networks, prioritize the protection of urban ventilation corridors, and promote localized cooling technologies in urban renewal initiatives.\u003c/li\u003e\n \u003cli\u003eEstablish a dynamic resilience monitoring and early warning system. Incorporate the stability and resistance indicators proposed in this study into routine urban health assessments. Build an integrated warning platform that couples physical risk with social vulnerability, with special alerts targeting extreme scenarios such as the reversal effects of AOD.\u003c/li\u003e\n \u003cli\u003eImplement adaptive co-regulation of air pollution and urban heat islands. Formulate \u0026lsquo;scenario-based\u0026rsquo; air pollution control plans that consider the context-dependent effects of AOD. Under extreme hot and dry conditions, priority should be given to limiting emissions of highly absorptive aerosols, such as black carbon, to achieve synergistic benefits in pollution control and heatwave adaptation.\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eDespite the methodological innovations and novel findings of this study, certain limitations remain. Specifically, the research falls short in aspects such as data scale matching, three-dimensional urban spatial analysis, and the deep integration of socio-economic dimensions. Future studies should focus on leveraging multi-source data with higher spatial and temporal resolution and incorporate urban canopy models to enable three-dimensional thermal simulations. In addition, there is an urgent need to promote interdisciplinary integration by coupling physical risk assessments with public health and socio-economic data. This would support the development of a comprehensive urban heatwave vulnerability and risk assessment framework, thereby providing a more solid scientific basis for designing climate adaptation strategies that are both equitable and targeted.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eCRediT authorship contribution statement\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eShichen Fan:\u0026nbsp;\u003c/strong\u003eConceptualization, Methodology, Data Curation, Formal Analysis, Investigation, Writing \u0026ndash; Original Draft, Writing \u0026ndash; Review \u0026amp; Editing. \u003cstrong\u003eHongyong Liu:\u003c/strong\u003e Supervision, Funding Acquisition, Project Administration, Conceptualization, Writing \u0026ndash; Review \u0026amp; Editing.\u003cstrong\u003e\u0026nbsp;ZhiHui Lai:\u003c/strong\u003e Writing \u0026ndash; Review \u0026amp; Editing, Visualization.\u003cstrong\u003e\u0026nbsp;Yi Xu:\u003c/strong\u003e Software, Writing \u0026ndash; Review \u0026amp; Editing.\u003c/p\u003e\n\u003cp\u003eAcknowledgments\u003c/p\u003e\n\u003cp\u003eWe are grateful to Xiaoping Wang, a PhD candidate at Deakin University, for his insightful suggestions and assistance in revising this manuscript. The authors also acknowledge the data support from the following sources: the National Earth System Science Data Center, National Science \u0026amp; Technology Infrastructure of China; the National Tibetan Plateau / Third Pole Environment Data Center (http://data.tpdc.ac.cn); and the DPECv1.0-v1.2 emission data provided by the MEIC model (http://meicmodel.org.cn).\u003c/p\u003e\n\u003cp\u003eDeclaration of generative AI and AI-assisted technologies in the writing process\u003c/p\u003e\n\u003cp\u003eStatement: During the preparation of this work the author(s) used Gemini (a large language model developed by Google) in order to assist with language refinement, translation between English and Chinese. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the published article.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eAbunyewah M, Gajendran T, Erdiaw-Kwasie M O, et al., 2025. The multidimensional impacts of heatwaves on human ecosystems: A systematic literature review and future research direction[J/OL]. Environmental Science \u0026amp; Policy, 165: 104024. DOI:10.1016/j.envsci.2025.104024.\u003c/li\u003e\n \u003cli\u003eAllen C R, Angeler D G, Chaffin B C, et al., 2019. Resilience reconciled[J/OL]. Nature Sustainability, 2(10): 898-900. DOI:10.1038/s41893-019-0401-4.\u003c/li\u003e\n \u003cli\u003eAlmeida C R D, Teodoro A C, Gon\u0026ccedil;alves A, 2021. Study of the Urban Heat Island (UHI) Using Remote Sensing Data/Techniques: A Systematic Review[J/OL]. Environments, 8(10): 105. DOI:10.3390/environments8100105.\u003c/li\u003e\n \u003cli\u003eBai K, Li Ke, Ma M, et al., 2022. LGHAP: the Long-term Gap-free High-resolution Air Pollutant concentration dataset, derived via tensor-flow-based multimodal data fusion[J/OL]. Earth System Science Data, 14(2): 907-927. DOI:10.5194/essd-14-907-2022.\u003c/li\u003e\n \u003cli\u003eBruneau M, Chang S E, Eguchi R T, et al., 2003. A Framework to Quantitatively Assess and Enhance the Seismic Resilience of Communities[J/OL]. Earthquake Spectra, 19(4): 733-752. DOI:10.1193/1.1623497.\u003c/li\u003e\n \u003cli\u003eBrunner L, Voigt A, 2024. Pitfalls in diagnosing temperature extremes[J/OL]. Nature Communications, 15(1): 2087. DOI:10.1038/s41467-024-46349-x.\u003c/li\u003e\n \u003cli\u003eCao C, Lee X, Liu S, et al., 2016. Urban heat islands in China enhanced by haze pollution[J/OL]. Nature Communications, 7(1): 12509. DOI:10.1038/ncomms12509.\u003c/li\u003e\n \u003cli\u003eChen Z, Yu B, Yang C, et al., 2021. An extended time series (2000\u0026ndash;2018) of global NPP-VIIRS-like nighttime light data from a cross-sensor calibration[J/OL]. Earth System Science Data, 13(3): 889-906. DOI:10.5194/essd-13-889-2021.\u003c/li\u003e\n \u003cli\u003eCheng J, Tong D, Liu Y, et al., 2023. A synergistic approach to air pollution control and carbon neutrality in China can avoid millions of premature deaths annually by 2060[J/OL]. One Earth, 6(8): 978-989. DOI:10.1016/j.oneear.2023.07.007.\u003c/li\u003e\n \u003cli\u003eCheval S, Amihăesei V A, Chitu Z, et al., 2024. A systematic review of urban heat island and heat waves research (1991\u0026ndash;2022)[J/OL]. Climate Risk Management, 44: 100603. DOI:https://doi.org/10.1016/j.crm.2024.100603.\u003c/li\u003e\n \u003cli\u003eCutter S L, 2016. The landscape of disaster resilience indicators in the USA[J/OL]. Natural Hazards, 80(2): 741-758. DOI:10.1007/s11069-015-1993-2.\u003c/li\u003e\n \u003cli\u003eDe Razza S, Zanetti C, De Marchi M, et al., 2024. Mapping urban heatwaves and islands: the reverse effect of Salento\u0026rsquo;s \u0026ldquo;white cities\u0026rdquo;[J/OL]. Frontiers in Earth Science, 12[2025-06-03]. https://www.frontiersin.org/journals/earth-science/articles/10.3389/feart.2024.1375827/full. DOI:10.3389/feart.2024.1375827.\u003c/li\u003e\n \u003cli\u003eDebnath R, Chandel T, Han F, et al., 2025. Heatwave increases nighttime light intensity in hyperdense cities of the Global South: A double machine learning study[A/OL]. arXiv[2025-06-03]. https://arxiv.org/abs/2503.00557. DOI:10.48550/ARXIV.2503.00557.\u003c/li\u003e\n \u003cli\u003eDidan K, 2015. MOD13A3 MODIS/Terra Vegetation Indices Monthly L3 Global 1km SIN Grid V006[DS/OL]. NASA Land Processes Distributed Active Archive Center[2025-06-21]. https://lpdaac.usgs.gov/products/mod13a3v006/. DOI:10.5067/MODIS/MOD13A3.006.\u003c/li\u003e\n \u003cli\u003eEsperon-Rodriguez M, Tjoelker M G, Lenoir J, et al., 2022. Climate change increases global risk to urban forests[J/OL]. Nature Climate Change, 12(10): 950-955. DOI:10.1038/s41558-022-01465-8.\u003c/li\u003e\n \u003cli\u003eFotheringham A S, Yang W, Kang W, 2017. Multiscale Geographically Weighted Regression (MGWR)[J/OL]. Annals of the American Association of Geographers, 107(6): 1247-1265. DOI:10.1080/24694452.2017.1352480.\u003c/li\u003e\n \u003cli\u003eGasparrini A, Guo Y, Hashizume M, et al., 2015. Mortality risk attributable to high and low ambient temperature: a multicountry observational study[J/OL]. The Lancet, 386(9991): 369-375. DOI:10.1016/S0140-6736(14)62114-0.\u003c/li\u003e\n \u003cli\u003eGil-D\u0026iacute;az C, Sicard M, Nabat P, et al., 2025. Dust Aerosol Radiative Effects During a Dust Event and Heatwave in Summer 2019 Simulated with a Regional Climate Atmospheric Model over the Iberian Peninsula[J/OL]. Remote Sensing, 17(11): 1817. DOI:10.3390/rs17111817.\u003c/li\u003e\n \u003cli\u003eGuo A, Yang J, Xiao X, et al., 2020. Influences of urban spatial form on urban heat island effects at the community level in China[J/OL]. Sustainable Cities and Society, 53: 101972. DOI:10.1016/j.scs.2019.101972.\u003c/li\u003e\n \u003cli\u003eHuang Z, Liu X, Yang Q, et al., 2021. Quantifying the spatiotemporal characteristics of multi-dimensional karst ecosystem stability with Landsat time series in southwest China[J/OL]. International Journal of Applied Earth Observation and Geoinformation, 104: 102575. DOI:https://doi.org/10.1016/j.jag.2021.102575.\u003c/li\u003e\n \u003cli\u003ejianfang K. Data set of 1 km resolution DEM in China[Z/OL]//National Cryosphere Desert Data Center. National Cryosphere Desert Data Center(www.ncdc.ac.cn). http://www.ncdc.ac.cn/portal/metadata/ba0e4b6b-8fef-4eec-9c84-81646f266282.\u003c/li\u003e\n \u003cli\u003eKong W, Liu Y, Dong S, et al., 2025. Research on the Influence of Evolution of Landscape Patterns of Blue-Green Space on the Cooling Effect in the Central Urban Area of Xi\u0026rsquo;an[J/OL]. Landscape Architecture, 32(5): 78-85. DOI:10.3724/j.fjyl.202408210473.\u003c/li\u003e\n \u003cli\u003eLebakula V, Epting J, Moehl J, et al., 2024. LandScan Silver Edition[DS/OL]. Oak Ridge, TN: Oak Ridge National Laboratory. DOI:10.48690/1531770.\u003c/li\u003e\n \u003cli\u003eLi C, Lu L, Fu Z, et al., 2022. Diverse cooling effects of green space on urban heat island in tropical megacities[J/OL]. Frontiers in Environmental Science, 10: 1073914. DOI:10.3389/fenvs.2022.1073914.\u003c/li\u003e\n \u003cli\u003eLi D, Bou-Zeid E, 2013. Synergistic Interactions between Urban Heat Islands and Heat Waves: The Impact in Cities Is Larger than the Sum of Its Parts[J/OL]. Journal of Applied Meteorology and Climatology, 52(9): 2051-2064. DOI:10.1175/JAMC-D-13-02.1.\u003c/li\u003e\n \u003cli\u003eLi Y, Schubert S, Kropp J P, et al., 2020. On the influence of density and morphology on the Urban Heat Island intensity[J/OL]. Nature Communications, 11(1): 2647. DOI:10.1038/s41467-020-16461-9.\u003c/li\u003e\n \u003cli\u003eLi Yuanxi, Luo X, Wang M, et al., 2024a. Spatiotemporal variations and influencing factors of heatwaves in Chengdu, China[J/OL]. Ecological Informatics, 84: 102924. DOI:10.1016/j.ecoinf.2024.102924.\u003c/li\u003e\n \u003cli\u003eLi Yuanxi, Luo X, Wang M, et al., 2024b. Spatiotemporal variations and influencing factors of heatwaves in Chengdu, China[J/OL]. Ecological Informatics, 84: 102924. DOI:https://doi.org/10.1016/j.ecoinf.2024.102924.\u003c/li\u003e\n \u003cli\u003eLiao W, Liu X, Wang D, et al., 2017. The Impact of Energy Consumption on the Surface Urban Heat Island in China\u0026rsquo;s 32 Major Cities[J/OL]. Remote Sensing, 9(3): 250. DOI:10.3390/rs9030250.\u003c/li\u003e\n \u003cli\u003eLu L, Weng Q, Xiao D, et al., 2020. Spatiotemporal Variation of Surface Urban Heat Islands in Relation to Land Cover Composition and Configuration: A Multi-Scale Case Study of Xi\u0026rsquo;an, China[J/OL]. Remote Sensing, 12(17). https://www.mdpi.com/2072-4292/12/17/2713. DOI:10.3390/rs12172713.\u003c/li\u003e\n \u003cli\u003eLuo F, Yang Y, Zong L, et al., 2023. The interactions between urban heat island and heat waves amplify urban warming in Guangzhou, China: Roles of urban ventilation and local climate zones[J/OL]. Frontiers in Environmental Science, 11: 1084473. DOI:10.3389/fenvs.2023.1084473.\u003c/li\u003e\n \u003cli\u003eLv Y, Sarker M N I, 2024. Integrative approaches to urban resilience: Evaluating the efficacy of resilience strategies in mitigating climate change vulnerabilities[J/OL]. Heliyon, 10(6): e28191. DOI:10.1016/j.heliyon.2024.e28191.\u003c/li\u003e\n \u003cli\u003eLyu Z Z, Gao H, Gao R, et al., 2023. Extreme characteristics and causes of the drought event in the whole Yangtze River Basin in the midsummer of 2022[J/OL]. Advances in Climate Change Research, 14(5): 642-650. DOI:10.1016/j.accre.2023.09.007.\u003c/li\u003e\n \u003cli\u003eMiao L, Ju L, Sun S, et al., 2024. Unveiling the dynamics of sequential extreme precipitation-heatwave compounds in China[J/OL]. npj Climate and Atmospheric Science, 7(1): 67. DOI:10.1038/s41612-024-00613-5.\u003c/li\u003e\n \u003cli\u003eNairn J, Fawcett R, 2014. The Excess Heat Factor: A Metric for Heatwave Intensity and Its Use in Classifying Heatwave Severity[J/OL]. International Journal of Environmental Research and Public Health, 12(1): 227-253. DOI:10.3390/ijerph120100227.\u003c/li\u003e\n \u003cli\u003ePENG Shouzhang, 2024. 1-km monthly precipitation dataset for China (1901-2023)[DS/OL]. National Tibetan Plateau Data Center. https://dx.doi.org/10.5281/zenodo.3114194. DOI:10.5281/zenodo.3114194.\u003c/li\u003e\n \u003cli\u003ePerkins S E, Alexander L V, 2013. On the Measurement of Heat Waves[J/OL]. Journal of Climate, 26(13): 4500-4517. DOI:10.1175/JCLI-D-12-00383.1.\u003c/li\u003e\n \u003cli\u003ePerkins-Kirkpatrick S E, Lewis S C, 2020. Increasing trends in regional heatwaves[J/OL]. Nature Communications, 11(1): 3357. DOI:10.1038/s41467-020-16970-7.\u003c/li\u003e\n \u003cli\u003eShao Y, Zhu Q, Feng Z, et al., 2024. Temporal and Spatial Assessment of Carbon Flux Dynamics: Evaluating Emissions and Sequestration in the Three Northern Protection Forest Project Areas Supported by Google Earth Engine[J/OL]. Remote Sensing, 16(5): 777. DOI:10.3390/rs16050777.\u003c/li\u003e\n \u003cli\u003eShi H, Song X, Zeng S, 2023. Impact of the Urban Heat Island Effect on Ozone Pollution in Chengdu City, China[J/OL]. Chinese Geographical Science, 33(6): 1017-1032. DOI:10.1007/s11769-023-1391-8.\u003c/li\u003e\n \u003cli\u003eSingh S P, Rafiuddin M, Banerjee S, et al., 2024. Investigation of the Thermal Structure in the Atmospheric Boundary Layer During Evening Transition and the Impact of Aerosols on Radiative Cooling[A/OL]. arXiv[2025-06-27]. https://arxiv.org/abs/2403.06656. DOI:10.48550/ARXIV.2403.06656.\u003c/li\u003e\n \u003cli\u003eTang W, Zhou J, Ma J, et al., 2024. TRIMS LST: a daily 1 km all-weather land surface temperature dataset for China\u0026rsquo;s landmass and surrounding areas (2000\u0026ndash;2022)[J/OL]. Earth System Science Data, 16(1): 387-419. DOI:10.5194/essd-16-387-2024.\u003c/li\u003e\n \u003cli\u003eWang J, Chen Y, Liao W, et al., 2021. Anthropogenic emissions and urbanization increase risk of compound hot extremes in cities[J/OL]. Nature Climate Change, 11(12): 1084-1089. DOI:10.1038/s41558-021-01196-2.\u003c/li\u003e\n \u003cli\u003eWANG M, XU H, 2018. Analyzing the Influence of Urban Forms on Surface Urban Heat Islands Intensity in Chinese Mega Cities[J/OL]. Journal of Geo-information Science, 20(12): 1787-1798. DOI:10.12082/dqxxkx.2018.180257.\u003c/li\u003e\n \u003cli\u003eWang R, Wang M, Zhang Z, et al., 2022. Geographical Detection of Urban Thermal Environment Based on the Local Climate Zones: A Case Study in Wuhan, China[J/OL]. Remote Sensing, 14(5): 1067. DOI:10.3390/rs14051067.\u003c/li\u003e\n \u003cli\u003eWeng Q, Lu D, Schubring J, 2004. Estimation of land surface temperature\u0026ndash;vegetation abundance relationship for urban heat island studies[J/OL]. Remote Sensing of Environment, 89(4): 467-483. DOI:https://doi.org/10.1016/j.rse.2003.11.005.\u003c/li\u003e\n \u003cli\u003eWorld Meteorological Organization, 2018. Guide to Climatological Practices[M]. 3rd ed. Geneva: WMO.\u003c/li\u003e\n \u003cli\u003eWu D, Zhang Y, Xiang Q, 2024. Geographically weighted random forests for macro-level crash frequency prediction[J/OL]. Accident Analysis \u0026amp; Prevention, 194: 107370. DOI:10.1016/j.aap.2023.107370.\u003c/li\u003e\n \u003cli\u003eXi Z, Li C, Zhou L, et al., 2023. Built environment influences on urban climate resilience: Evidence from extreme heat events in Macau[J/OL]. Science of The Total Environment, 859: 160270. DOI:https://doi.org/10.1016/j.scitotenv.2022.160270.\u003c/li\u003e\n \u003cli\u003eXie Y, Weng Q, Fu P, 2019. Temporal variations of artificial nighttime lights and their implications for urbanization in the conterminous United States, 2013\u0026ndash;2017[J/OL]. Remote Sensing of Environment, 225: 160-174. DOI:https://doi.org/10.1016/j.rse.2019.03.008.\u003c/li\u003e\n \u003cli\u003eYang M, Ren C, Wang H, et al., 2024. Mitigating urban heat island through neighboring rural land cover[J/OL]. Nature Cities, 1(8): 522-532. DOI:10.1038/s44284-024-00091-z.\u003c/li\u003e\n \u003cli\u003eYi C, Jackson N, 2021. A review of measuring ecosystem resilience to disturbance[J/OL]. Environmental Research Letters, 16(5): 053008. DOI:10.1088/1748-9326/abdf09.\u003c/li\u003e\n \u003cli\u003eZhan L F, Wang Y, Sun H, et al., 2019. Study on the Change Characteristics of and Population Exposure to Heatwave Events on the North China Plain[J/OL]. Advances in Meteorology, 2019: 1-10. DOI:10.1155/2019/7069195.\u003c/li\u003e\n \u003cli\u003eZhang P, Imhoff M L, Bounoua L, et al., 2012. Exploring the influence of impervious surface density and shape on urban heat islands in the northeast United States using MODIS and Landsat[J/OL]. Canadian Journal of Remote Sensing, 38(4): 441-451. DOI:10.5589/m12-036.\u003c/li\u003e\n \u003cli\u003eZhang Q, Jia B, Li T, et al., 2025. Dynamic changes and drives of surface urban heat islands in China[J/OL]. City and Environment Interactions, 27: 100203. DOI:https://doi.org/10.1016/j.cacint.2025.100203.\u003cstrong\u003e\u003c/strong\u003e\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"theoretical-and-applied-climatology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"taac","sideBox":"Learn more about [Theoretical and Applied Climatology](https://www.springer.com/journal/704)","snPcode":"704","submissionUrl":"https://submission.nature.com/new-submission/704/3","title":"Theoretical and Applied Climatology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Urban Heatwave, Thermal resilience, Climate Change, sustainable urban development","lastPublishedDoi":"10.21203/rs.3.rs-7374786/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7374786/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Amidst global climate change and rapid urbanization, urban heatwaves pose a severe threat to thermal environmental resilience, impacting public health and socioeconomic stability. Accurately quantifying the spatiotemporal dynamics of heatwave risks and resilience is a critical prerequisite for developing effective urban climate adaptation strategies. However, previous research is often limited by discrete station-based data and lacks standardized thermal environment metrics, failing to fully capture the complex, non-linear, and spatially non-stationary characteristics of heatwave drivers. To address these gaps, this study introduces new thermal environment indicators and employs an integrated analytical framework to analyze a 20-year high-resolution dataset for Chengdu, China. The results reveal that while land use patterns (PLAND) are the dominant factor shaping the thermal environment, the impacts of various drivers exhibit significant spatial non-stationarity; for instance, the cooling effect of green spaces diminishes in highly urbanized cores. ‘Contextual reversal’ of Aerosol Optical Depth (AOD)’s effect is shown in the result: AOD provides a cooling ‘parasol effect’ under normal conditions but can reverse its role to a warming ‘blanket effect’ during extreme heat and drought, thereby exacerbating heatwaves. Based on these findings, this study challenges ‘one-size-fits-all’ approaches and provides a robust scientific foundation for developing precise, adaptive governance strategies, such as climate-adaptive zoning, dynamic risk monitoring, and synergistic pollution-heat control policies.","manuscriptTitle":"Urban Heatwave Resilience and Spatial Drivers: A 20-Year Geo-Spatial Analysis in Chengdu, China","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-05 06:35:39","doi":"10.21203/rs.3.rs-7374786/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewersInvited","content":"","date":"2025-10-24T03:21:51+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-08-15T06:02:10+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-08-15T06:01:56+00:00","index":"","fulltext":""},{"type":"submitted","content":"Theoretical and Applied Climatology","date":"2025-08-14T14:20:49+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"theoretical-and-applied-climatology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"taac","sideBox":"Learn more about [Theoretical and Applied Climatology](https://www.springer.com/journal/704)","snPcode":"704","submissionUrl":"https://submission.nature.com/new-submission/704/3","title":"Theoretical and Applied Climatology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"8e4e83ef-52bd-4256-b6cb-6b2db13479a5","owner":[],"postedDate":"November 5th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-11-05T06:35:39+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-05 06:35:39","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7374786","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7374786","identity":"rs-7374786","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}