Resilience Assessment and Evolution Characteristics of Urban Earthquakes in the Sichuan–Yunnan Region Based on the DPSIR Model

Abstract

The Sichuan–Yunnan region, a primary seismic-prone zone on the Qinghai–Tibet Plateau, has experienced heightened seismic exposure due to rapid urbanisation. In order to address the issue of disaster risks and to promote sustainable urban development, this study establishes an integrated urban seismic resilience evaluation framework based on the DPSIR (Driving–Pressure–State–Impact–Response) model. The CRITIC–AHP combined weighting method was utilised to determine indicator weights, and data from 37 prefecture-level cities (2010, 2015, 2020) were analysed to reveal spatial–temporal evolution patterns and correlations. The results demonstrate a consistent improvement in regional seismic resilience, with the overall index increasing from 0.501 in 2010 to 0.526 in 2020. Sichuan exhibited a “decline-then-rise” trend (0.570 to 0.566 to 0.585), while Yunnan demonstrated continuous growth (0.517 to 0.557). The spatial pattern underwent an evolution from “west–low, central–eastern–high” to “south–high, north–low”, with over half of the cities attaining relatively high resilience by 2020. Chengdu and Kunming have been identified as dual high-resilience cores, diffusing resilience outward to neighbouring regions. In contrast, mountainous areas such as Garze and Aba have been found to exhibit low resilience levels, primarily due to high seismic stress and limited socioeconomic capacity. Subsystem analysis has revealed divergent resilience pathways across provinces, while spatial autocorrelation has demonstrated fluctuating global Moran’s I values and temporary local clustering. This research provides a scientific foundation for seismic disaster mitigation and offers a transferable analytical framework for enhancing urban resilience in earthquake-prone regions globally.

1. Introduction

Earthquakes are a common occurrence worldwide. The Circum-Pacific Volcanic Belt has been determined to be a region of particular seismic activity [1], with approximately 80% of the world’s shallow-focus earthquakes and 90% of its intermediate- and deep-focus events occurring there. A substantial portion of China is situated within this seismic belt, encompassing five major seismic zones, including the Qinghai–Tibet Plateau and North China Seismic Zones [2]. It is evident that China experiences a high frequency and magnitude of earthquake-induced losses, which is a global trend [3,4]. Of these, the Qinghai–Tibet Plateau Seismic Zone, located at the leading edge of the Indian–Eurasian Plate collision [5], constitutes the core of China’s “North–South Seismic Belt” and exhibits the most intense and frequent seismic activity [6].

The Sichuan–Yunnan (S–Y) region, located at the heart of this seismic zone and serving as a major economic and cultural hub in Southwest China, faces the dual challenges of complex geological structures and rapid urbanisation. Consequently, enhancing urban seismic resilience has become a critical issue for disaster prevention and sustainable development. The S–Y region is considered to be one of the most seismically active regions in China, and it has been documented as a leading contributor to seismic energy release on a national scale [7]. The 2008 Wenchuan (Mw7.9) and 2013 Lushan (Mw6.6) [8] earthquakes resulted in over 69,000 fatalities and economic losses exceeding 600 billion yuan [9,10], underscoring the strong spatial coupling between seismic risk and urbanisation. In the context of ongoing urban expansion and population concentration, seismic risks in this region are becoming increasingly intricate. Conventional post-disaster rescue and reconstruction strategies are inadequate for the promotion of sustainable development. The concept of urban seismic resilience, which integrates resistance, recovery, and adaptive capacity across the entire disaster cycle [11], provides a new paradigm for seismic risk management and pre-disaster planning.

Urban resilience has been the focus of extensive global research [12,13,14]. The term was introduced into the field of ecology in 1973, when it was defined as the ability of an ecosystem to maintain stability and recover function following disturbance [15]. The concept originates from the Latin resilio, meaning “to rebound”. The field has since expanded into economics, sociology, and engineering [16,17,18]. In the context of seismic events, the term “urban resilience” is employed to denote a city’s capacity to either maintain or expeditiously restore critical functions in the aftermath of an earthquake [19,20]. This concept encompasses a range of aspects, including pre-disaster resistance, in-disaster absorption, post-disaster recovery, and long-term adaptation [21,22].

With regard to the focus and methodology of research, studies have predominantly followed two directions. One strand of research emphasises physical resilience, examining the seismic performance of buildings and infrastructure, and recovery capabilities [23,24,25]. The other salient points pertain to multidimensional resilience, incorporating social, economic, and ecological factors such as population density, economic development, and community cohesion [26,27,28]. In terms of methodology, researchers have combined mathematical and spatial analyses, employing approaches such as the Analytic Hierarchy Process (AHP), Entropy Weight Method, Principal Component Analysis (PCA), and TOPSIS [29,30,31]. Comparative analyses, such as Civera et al., have refined multi-criteria decision-making frameworks. GIS-based spatial overlay techniques further reveal the spatial and temporal evolution of regional resilience [32], enabling a shift from static evaluation to dynamic modelling [23,27].

However, the majority of extant studies have focused on the resilience of urban “hardware” and “software” systems [13,17,33], while neglecting the integrated analysis of seismic losses and regional heterogeneity. In order to undertake a comprehensive evaluation of the seismic resilience of a region, it is necessary to consider three factors concurrently: the frequency of seismic events, the intensity of the associated hazard, and the socio-ecological context. This issue constitutes a complex multi-attribute decision-making problem, and research in this area remains insufficient at present.

In order to address this gap, the present study proposes a regional seismic resilience assessment framework based on the DPSIR model. The aim of this study is to evaluate and characterise the spatiotemporal evolution characteristics of resilience in the Sichuan–Yunnan (S–Y) region. The “Driving Force–Pressure–State–Impact–Response” (DPSIR) model, proposed by the European Environment Agency (EEA) [34], is a theoretical framework developed from the traditional PSR (Pressure-State-Response) model, with two additional dimensions of “Driving Force” and “Impact”. This integration facilitates dynamic analysis of interactions between humans and the environment. The model has been widely applied in the fields of environmental management, sustainable development, and disaster risk reduction [35,36,37], due to the clarity of its concepts and the systematic logic of its structure. However, the utilisation of this approach within the domain of urban seismic resilience remains limited, underscoring the significant research contributions of this study.

A pivotal stage in the implementation of the DPSIR framework pertains to the determination of indicator weights. Subjective methods, such as AHP [38], reflect expert judgement, while objective methods, such as CRITIC [39], derive weights from data variability. In order to minimise bias and enhance robustness, this study adopts a combined CRITIC–AHP weighting method [40,41], integrating expert insight with data-driven rigour. This framework, tailored to the seismic and urban characteristics of the S–Y region, provides a robust scientific foundation for the management of seismic risk, urban planning, and sustainable development in seismically active areas.

2. Materials and Methods

2.1. Overview of the Study Area

The S–Y region is situated on the southeastern margin of the Qinghai–Tibet Plateau, a tectonically active zone formed by the collision and compression of the Eurasian and Indian plates. This region is characterised by intense deformation, complex geological structures, and intersecting fault systems, including the Xianshuihe-Xiaojiang [42], Anninghe-Zemuhe [43,44], and Longmenshan fault zones [45]. Frequent seismic activity, driven by active plate movements and fault zone interactions, has established the area as one of China’s most earthquake-prone regions [46,47]. In 2021, the population affected by earthquakes in Sichuan and Yunnan provinces accounted for 59.9% of the total population affected by earthquakes nationwide, and the direct economic losses caused by earthquakes in these two provinces accounted for 63.2% of the total direct economic losses from earthquakes across the country [48].

For the purpose of analysis, in accordance with the “Classification and Coding Rules for Geomorphological Types” (GB/T 44060-2024) [49], the Sichuan and Yunnan regions are first divided into three basic geomorphological zones: plains, hills, and mountain-based on regional elevation data. Considering that the proportion of mountains in Sichuan was 79.52% and in Yunnan was 88.64% according to the first national geographic conditions census in 2017, and given the significant altitude differences within the mountainous areas of the two provinces, the mountains are further subdivided into low-altitude mountains, mid-altitude mountains, and high-altitude mountains. This results in a total of five major divisions: plains, hills, low-altitude mountains, mid-altitude mountains, and high-altitude mountains. This accurately reflects the differences in altitude, earthquake disaster risks, and socio-economic conditions among different regions (see Figure 1).

From a demographic perspective, the Sichuan–Yunnan region in southwestern China is a significant cultural and seismic hub. Sichuan Province, located within the southwestern region of China, encompasses an area of approximately 486,000 square kilometres. According to the latest census data, the province had a population of around 83.64 million at the end of 2024, comprising 50.27 million urban residents and 33.37 million rural residents. The mean population density is approximately 172 people per km², and the resident population urbanisation rate is 60.10%. At the close of 2024 [50], Yunnan Province’s total population stood at approximately 46.55 million, with 25.19 million residing in urban areas and 21.36 million inhabiting rural regions. The mean population density of the area is approximately 96 people per square kilometre. The resident population urbanisation rate is 54.11% [51]. Over the past two decades, rapid urbanisation has increased social and economic activity. However, significant disparities remain in terms of education levels, awareness of disaster preparedness, and self-rescue capabilities, particularly in rural and mountainous communities, which are more vulnerable to seismic hazards.

Economically, the region is rich in natural resources and ecological diversity, supporting an industrial structure dominated by agriculture, tourism, and modern services. Despite its role as a key economic development pole in southwestern China, challenges such as uneven infrastructure development and insufficient disaster mitigation capacity hinder progress toward sustainable urbanisation. Enhancing urban seismic resilience in this region is critical to achieving the goals of Chinese-style modernization, ensuring not only disaster resistance but also long-term socioeconomic stability and ecological sustainability.

2.2. Methodology

2.2.1. Theoretical Framework of the DPSIR Model

The DPSIR model represents an optimisation and further development of the Pressure-State-Response (PSR) model and the Driving Force–State–Response (DSR) model. The PSR model was proposed by Canadian statisticians and has been widely applied in evaluating the stability of ecosystems and ecological security [52]. Its aim is reveal the relationships among environmental pressure, environmental state, and social response. The DSR model, an in-depth expansion of the PSR model by the United Nations Commission on Sustainable Development (UNCSD) in 2007, transforms the pressure element in the PSR model into a driving force element [53,54]. The driving force element is more comprehensive, significantly enhancing the model’s adaptability. The DPSIR model, based on causal relationships, closely links the five aspects of “Driving force”, “Pressure”, “State”, “Impact”, and “Response”, jointly constructing the internal mechanism and evolutionary process of urban seismic resilience.

Urban seismic resilience not only considers the ability of a city to maintain its functions during an earthquake but also includes the characteristics of resistance, absorption, recovery during an earthquake, and post-disaster integration and learning. The DPSIR model has good applicability under this concept of seismic resilience. The “Driving force (D)” encompasses natural and human factors that affect urban seismic resilience. The “Pressure (P)” reflects the various pressures exerted on the city by an earthquake and is the key factor that urban seismic resilience needs to address. The “State (S)” reflects the basic conditions and state of the city before an earthquake, determining the initial resilience level. The “Impact (I)” describes the multi-faceted impacts of an earthquake disaster on the city, and urban seismic resilience aims to mitigate these adverse impacts to achieve sustainable development. The “Response (R)” corresponds to the city’s response measures and recovery capabilities after an earthquake, which is in line with the characteristics of urban seismic resilience during and after an earthquake. In summary, each link of the DPSIR model is closely related to the definition and connotation of urban seismic resilience, enabling it to be comprehensively and effectively applied to urban seismic resilience research.

The DPSIR (Driving force–Pressure–State–Impact–Response) framework has been widely applied in urban resilience research to integrate social, economic, and infrastructural indicators. However, most existing studies focus on single cities or static assessments and rarely consider evolutionary resilience over multiple time periods. This study applies the DPSIR framework to urban seismic resilience for the first time, constructing an evaluation index system and measuring resilience from an evolutionary perspective.

Compared with previous studies on urban seismic resilience [55,56], based on the essential connotation of urban seismic resilience and supported by the DPSIR model, constructing an evaluation index system and conducting resilience measurement from the perspective of evolutionary resilience are more in line with current research trends. This approach helps to promote the transformation of resilience research from stable equilibrium to dynamic development.

2.2.2. Indicator Selection and Theoretical Basis

The selection of specific indicators under each dimension of the DPSIR framework, as shown in Figure 2, was guided by the principles of scientific relevance, data availability and theoretical linkage with urban seismic resilience. These indicators were identified through a review of previous studies on disasters and urban resilience to ensure that each one reflects a key aspect of a city’s ability to withstand, absorb, recover from and adapt to the impact of earthquakes.

The Driving Force (D) subsystem captures the socio-economic dynamics underlying urban resilience. Indicators such as regional GDP, disposable income per capita and the urbanisation rate reflect the economic development and demographic patterns that influence the availability of resources, social capital and the city’s adaptive capacity. All of these are key preconditions for seismic resilience [28,57].

The Pressure (P) subsystem quantifies the external stresses imposed by seismic hazards, including such metrics as earthquake frequency, casualties, property damage, and the fault length per unit area within the region. These indicators are designed to measure the intensity of natural shocks and have been shown to pose a direct threat to the city’s functional stability, revealing its exposure and vulnerability levels [58,59].

The State (S) subsystem reflects pre-disaster conditions and the natural resilience of urban systems. Indicators such as the intensity of seismic fortification, representing the earthquake resistance of buildings and infrastructure, shelter capacity, reflecting the ability to accommodate displaced populations, and green space area, which supports evacuation and post-disaster recovery, characterise the baseline state of resilience. These variables largely determine the city’s inherent resistance and absorption capacity prior to seismic events [58,60].

The Impact (I) subsystem evaluates the multifaceted consequences of earthquakes on urban systems. Indicators include the employment rate, vegetation coverage, and the rate at which sewage is treated. These variables reflect the economic, social, and environmental disruptions that resilient cities aim to mitigate and recover from. This emphasises the importance of maintaining essential functions during and after disasters [58,60].

The Response (R) subsystem focuses on post-disaster governance, public service provision, and recovery potential. Indicators such as public budget expenditure, medical resources (including hospital beds and medical staff), and the length of the road network capture the city’s capacity to respond to hazards, restore critical infrastructure, and adapt to future risks. These indicators reflect the dynamic and evolving nature of urban seismic resilience [60].

Overall, the selected indicators closely correspond to the core dimensions of the resilience framework: resistance, absorption, recovery and adaptation. This provides a comprehensive, measurable basis for evaluating urban seismic resilience.

This study employs the DPSIR model to construct an evaluation index system for urban seismic resilience in the Sichuan–Yunnan region, based on the principles of data availability, scientific reliability, and systematic integrity. The observation periods are set as 2010, 2015 and 2020, which correspond to the final years of China’s 11th, 12th and 13th Five-Year Plans, respectively. These years effectively capture evolutionary trends in urban seismic resilience at different stages of regional socio-economic development and seismic risk management.

The research data were primarily obtained from the Sichuan Statistical Yearbook (2011, 2016, 2021) and the Yunnan Statistical Yearbook (2011, 2016, 2021), and were supplemented by the statistical bulletins on the national economic and social development of each city for the years 2010, 2015 and 2020. Additional supporting data on seismic hazard intensity, population density and economic indicators were obtained from the National Bureau of Statistics of China and the China Earthquake Networks Centre. These datasets ensure consistency, comparability and representativeness across the study period [61,62,63,64,65,66].

During the data processing of this study, it was found that some indicators had missing values, mainly including Sewage Treatment Rate, Solid Waste Utilisation Rate, Employment Rate, and Natural Population Growth Rate. The proportion of missing data for each indicator ranged from 2% to 7%. To ensure the continuity and comparability of the data, this study used linear interpolation to fill in the missing data [67]. That is, based on the changing trends of the known data before and after the missing years, the data were estimated and filled in chronological order. For consecutive years of missing data, the interpolation was uniformly estimated based on the known data at both ends. The supplemented data were checked for trend consistency and outliers [68]. The results showed that the interpolated data were consistent with the changing trends of the original data and there were no obvious outliers, ensuring the reliability of the standardisation and weight calculation.

2.2.3. Research Methods

• Calculation of Urban Seismic Resilience

This study employs the AHP method to determine the subjective weights of indicators based on expert judgments, reflecting the understanding of domain experts regarding the significance of indicators related to urban seismic resilience. Then, by combining the objective weights obtained through the CRITIC method, a comprehensive weight system (CRITIC–AHP) is constructed to enhance the scientificity and rationality of the final weight determination.

Data Normalisation Processing: To eliminate dimensional discrepancies among indicators, raw data were normalised using min-max scaling. For positive indicators (where higher values indicate better resilience), normalisation was performed as:

$$y_{ij}={\displaystyle \frac{x_{ij}-x_{min}}{x_{max}-x_{min}}}$$

(1)

For negative indicators (where lower values indicate better resilience):

$$y_{ij}={\displaystyle \frac{x_{max}-x_{ij}}{x_{max}-x_{min}}}$$

(2)

where $y_{ij}$ is the normalised index value, In this study, the letter $i$ is used to denote the 37 cities in question, while the letter $j$ is used to denote the 26 indicators; $x_{ij}$ is the original value of the seismic resilience index; $x_{max}$ and $x_{min}$ are the maximum and minimum values of the index, respectively.

Analytic Hierarchy Process (AHP): Based on the constructed urban seismic resilience assessment system, the Analytic Hierarchy Process (AHP) is adopted to determine the subjective weights of the indicators. This paper did not conduct a formal expert questionnaire survey. The pairwise comparison relationships among the indicators were mainly determined through a systematic literature review and internal discussions within the research team, referring to existing research results on urban seismic resilience and disaster risk assessment [37,69,70], and in combination with the seismic geological characteristics and actual urban development in the Sichuan–Yunnan region, The scale comparison method was used to construct the judgement matrix, a judgement matrix $A={\left(a_{ij}\right)}_{m\times n}$ is constructed using the scale comparison method, where $a_{ij}$ represents the importance of indicator $i$ relative to indicator $j$, all judgement matrices passed the consistency test (CR < 0.1) to ensure the logical consistency and reliability of the weight results. The values and interpretations of the scale are based on

Table 1. Grade and interpretations.

Grade	Interpretations
2, 4, 6, 8	Intermediate values of the above scales
1	Factors $i$ and $j$ are equally important
3	Factor $i$ is slightly more important than $j$
5	Factor $i$ is significantly more important than $j$
7	Factor $i$ is strongly more important than $j$
9	Factor $i$ is extremely more important than $j$

.

The weights of the indicators are calculated based on the judgement matrix using the square root method, as shown in Formula (3):

$${M_i}^{\prime }=\sqrt[n]{\prod _{j=1}^n{a}_{ij}}\hspace{1em}(i=1,2,\dots ,n)$$

(3)

Among them, ${M_i}^{\prime }$ represents the initial weight of the $i$-th element, $a_{ij}$ is the element of the judgement matrix, and $n$ is the order of the matrix.

$$M_j={\displaystyle \frac{{M_i}^{\prime }}{{\sum }_{k=1}^n{M_k}^{\prime }}}\hspace{1em}(i=1,2,\dots ,n)$$

(4)

Here, $W_i$ represents the final weight of the $i$th element, and ${\sum }_{k=1}^n{W_k}^{\prime }$ is the total sum of all the initial weights.

CRITIC Weighting Method: The CRITIC (Criteria Importance Through Intercriteria Correlation) method assigns weights by integrating variability (contrast intensity) and conflict (correlation between indicators) [38].

• Indicator Variability

Measured by the standard deviation ($S_j$) of normalised values:

$$S_j=\sqrt{{\displaystyle \frac{{\sum }_{i=1}^n{(y_{ij}-\overline{y_j})}^2}{n-1}}}$$

(5)

Here, $S_j$ represents the standard deviation of the $j$-th indicator; $i$ is the sample serial number, ranging from 1 to the total sample size $n$; $y_{ij}$ is the observed value of the ith sample on the $j$-th indicator; $\overline{y_j}$ is the sample mean of the $j$-th indicator; $n$ is the total number of samples involved in the evaluation; $n-1$ is the degree of freedom when calculating the sample standard deviation, used for an unbiased estimation of the population standard deviation.

• Indicator Conflict

Calculated using the correlation coefficient:

$$R_j=\sum _{i=1}^p(1-r_{ij})$$

(6)

Let $R_j$ denote the degree of conflict of the $j$-th indicator, which reflects the degree of non-correlation between this indicator and other indicators. The greater the conflict, the more unique information this indicator provides in the evaluation system. $p$ represents the total number of indicators in the evaluation system, meaning there are $p$ indicators involved in the evaluation. $i$ is the serial number of the indicator, ranging from 1 to $p$. $r_{ij}$ represents the correlation coefficient between the $i$-th indicator and the $j$-th indicator, used to measure the degree of linear correlation between the two indicators, with a value range of [−1, 1]. $1-r_{ij}$ then represents the “degree of non-correlation” between the two indicators, which is a local manifestation of the conflict.

• Information Quantity

By applying the indicator variability and indicator conflict, the final information value of the indicator on urban resilience can be obtained, which comprehensively represents the importance of the indicator in the evaluation index system. The formula is as follows:

$$C_j=S_j\sum _{i=1}^p(1-r_{ij})=S_j\ast R_j$$

(7)

$C_j$ represents the comprehensive information coefficient of the $j$-th indicator, which is jointly determined by the variability and conflict of the indicator; $S_j$ is the standard deviation of the $j$-th indicator, reflecting the degree of variability of the indicator’s data. The greater the variability, the larger $S_j$. $p$ is the total number of indicators in the evaluation system, and $i$ is the indicator sequence number ranging from 1 to $p$. $r_{ij}$ is the correlation coefficient between the $i$-th indicator and the $j$-th indicator usually the Pearson correlation coefficient, measuring the degree of linear correlation between the two indicators (with a range of [−1, 1]). $1-r_{ij}$ represents the “degree of non-correlation” between the two indicators. The sum of these values gives $R_j$, which is the conflict of the j-th indicator, reflecting the degree of non-correlation between this indicator and other indicators. The greater the conflict, the more unique information there is.

• Indicator Weight Value:

By calculating the normalisation of the indicator information value and obtaining its proportion in the total information value, the final weight of the indicator on urban resilience can be obtained. The formula is as follows:

$$W_j={\displaystyle \frac{C_j}{{\sum }_{j=1}^p{C}_j}}$$

(8)

$W_j$ represents the weight of the $j$-th indicator, which is used to measure the importance of the indicator in the evaluation system; $C_j$ is the comprehensive information coefficient of the $j$-th indicator, jointly determined by the variability and conflict of the indicator, and it is the core intermediate variable of the CRITIC method.

Comprehensive Evaluation Method

Based on the combined CRITIC-AHP method, the comprehensive indicator weights are obtained. These weights are then used in a comprehensive evaluation framework to assess urban seismic resilience in the S–Y region, enabling an integrated and quantitative analysis of resilience across different dimensions.

Calculate the indicator weights by the CRITIC-AHP method. The formula is as follows:

$$W_{jk}={\displaystyle \frac{M_j{W}_j}{\sum M_j{W}_j}}$$

(9)

where $W_k$ is the comprehensive weight value, and $M_j$ is the weight value of the analytic hierarchy process, $W_j$ is the Weight for CRITIC Method.

Evaluate the urban seismic resilience in the S–Y region by the comprehensive evaluation method. The formula is as follows:

$$Y_i={\sum }_{j=1}^n{W}_{jk}{y}_{ij}(i=1,2,\dots ,n)$$

(10)

$$ERI=a{Y}_1+b{Y}_2+c{Y}_3+d{Y}_4+e{Y}_5$$

(11)

where $Y_1$, $Y_2, Y_3, Y_4, Y_5$ respectively represent the driving force, pressure, state, influence, and response resilience indices, with a value range of 0~1. $y_{ij}$ is the normalised value of the indicator, and $a$, $b$, $c$, $d$, $e$ are the weights of the indicator layer, and their values are uniformly 0.2. $ERI$ stands for Earthquake Resilience Index, which is the comprehensive value of regional earthquake resilience.

• Resilience Spatial Correlation Analysis

This paper uses the global spatial autocorrelation index and the local spatial autocorrelation index to explore the agglomeration characteristics and difference characteristics of spatial data, to reveal the spatial distribution law of urban seismic resilience.

(1)

The purpose of global spatial autocorrelation analysis is to evaluate from an overall perspective whether the distribution of spatial data shows significant spatial dependence or randomness. Spatial Autocorrelation Tools are a category of statistical methods used to quantitatively describe the interrelationships among spatial objects in geographic space. Among them, Moran’s I is the most commonly used global spatial autocorrelation measure.

The Moran’s I index is used to reflect the aggregation or dispersion characteristics of the research object in space, and its calculation formula is as follows:

$$\text{Moran’s I}={\displaystyle \frac{n{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^n{w}_{ij}(x_i-\overline{x})(x_j-\overline{x})}}}{({\displaystyle {\sum }_{i=1}^n{(x_i-\overline{x})}^2}){\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^n{w}_{ij}}}}}$$

(12)

where $n$ is the number of spatial units, $x_i$ and $x_j$ are the attribute values of spatial units $i$ and $j$, $\overline{x}$ is the mean value of the attribute values, which represents the mean value of urban seismic resilience in this paper, and $w_{ij}$ is an element in the spatial weight matrix.

The Moran’s I index is in the range of [−1, 1]. When the Moran’s I index = 0, the space is randomly distributed; when the Moran’s I index > 0 and the p value is significant, it is a positive correlation, and the change in this attribute in the region shows the same trend on adjacent spatial units, and there is a spatial clustering phenomenon. The larger the Moran’s I index, the more obvious the spatial positive correlation; conversely, when the Moran’s I index < 0 and the p value is significant, it is a negative correlation, and the change in this attribute in the region shows a different trend on adjacent spatial units, and there is a spatial dispersion phenomenon. The smaller the Moran’s I index, the more obvious the spatial negative correlation.

The p-value represents the probability of observing the calculated Moran’s I under the null hypothesis of spatial randomness. A small p-value (e.g., p < 0.05) indicates that the observed spatial autocorrelation is statistically significant.

(2)

Local Spatial Autocorrelation: Local spatial autocorrelation aims to reveal the similarity between a spatial unit and its neighbouring units. The local Moran’s I is commonly used, and the formula is as follows.

$$\mathit{\text{Local Moran’s I}}={\displaystyle \frac{Z_i}{S^2}}{\displaystyle \sum _{j\ne i}^n{w}_{ij}{Z}_j}$$

(13)

$$Z_i=x_i-\overline{x}, Z_j=x_j-\overline{x}, S^2={\displaystyle \frac{1}{n}}{\displaystyle \sum {(x_i-\overline{x})}^2}$$

(14)

where $w_{ij}$ is the spatial weight value, $n$ is the sum of all regions in the study area, and $local Moran’s I$ is the local Moran’s index.

The local spatial association patterns can be divided into four types: low–low (L-L) clustering, low-high (L-H) clustering, high–low (H-L) clustering, and high–high (H-H) clustering. Among them, low–low clustering means that a low-value spatial unit is surrounded by low-value spatial units; low-high clustering means that a low-value spatial unit is surrounded by high-value spatial units; high–low clustering means that a high-value spatial unit is surrounded by low-value spatial units; high–high clustering means that a high-value spatial unit is surrounded by other high-value spatial units.

3. Results

Based on regional characteristics and related research, this paper combines the DPSIR model to construct an urban seismic resilience evaluation index system for the S–Y region. Using the index data, the index weights at three time points were calculated through Formulas (1)–(10). Taking 2010 as an example, the urban seismic resilience evaluation index system is shown in

Table 2. Urban seismic resilience index system in 2010.

Criterion Layer	Criterion Weight	Indicator Layer	CRITIC Weight	AHP Weight	Comprehensive Weight
Driving Force	0.14	Regional GDP	0.10	0.19	0.13
		Per Capita Disposable Income	0.15	0.14	0.14
		Population Density	0.18	0.25	0.30
		Urbanisation Rate	0.18	0.24	0.28
		Proportion of Tertiary Industry	0.28	0.09	0.09
		Retail Sales of Consumer Goods	0.11	0.10	0.07
Pressure	0.36	Frequency of Earthquakes (Past 5 Years)	0.39	0.20	0.33
		Earthquake Casualties (Past 5 Years)	0.19	0.40	0.33
		Property Loss from Earthquakes (Past 5 Years)	0.19	0.30	0.25
		Fault length per unit area	0.23	0.10	0.10
State	0.20	Seismic Fortification Intensity	0.41	0.20	0.31
		Green Space Area	0.23	0.30	0.26
		Shelter Capacity	0.26	0.40	0.39
		Urban Built-up Area	0.11	0.10	0.00
Impact	0.06	Sewage Treatment Rate	0.16	0.10	0.10
		Solid Waste Utilisation Rate	0.20	0.05	0.06
		Vegetation Coverage Rate	0.21	0.14	0.20
		Employment Rate	0.08	0.24	0.13
		Nighttime Light Density	0.09	0.29	0.18
		Natural Population Growth Rate	0.25	0.19	0.33
Response	0.24	Total Public Budget Expenditure	0.13	0.29	0.23
		Road Network Length	0.29	0.14	0.26
		Postal Service Revenue	0.17	0.05	0.05
		Number of College Students	0.13	0.10	0.08
		Hospital Bed Capacity	0.13	0.19	0.15
		Number of Medical Staff	0.15	0.24	0.23

.

3.1. Analysis of Urban Seismic Resilience in S–Y Region

3.1.1. Analysis of the Time-Series Evolution Characteristics of Urban Seismic Resilience

The six sub-graphs in the figure depict the standardised values of overall resilience, driving resilience, pressure resilience, state resilience, impact resilience and response resilience in the Sichuan–Yunnan (S–Y) region, Sichuan and Yunnan, from 2010 to 2020 (as Figure 3). They offer insights into seismic resilience and related subsystem changes in this area.

Overall resilience (a): Sichuan’s overall resilience decreased slightly at first, then increased significantly, rising from 0.570 in 2010 to 0.566 in 2015 and reaching 0.585 in 2020. The region demonstrated a strong comprehensive performance. Yunnan’s resilience continued to rise, increasing from 0.517 in 2010 to 0.527 in 2015 and reaching 0.557 in 2020. The region exhibited clear growth momentum. The S–Y region’s resilience improved steadily, rising from 0.501 in 2010 to 0.503 in 2015 and reaching 0.526 in 2020. This reflects the region’s continuous optimisation of its overall seismic resilience.

Driving Resilience (b): Sichuan’s Driving Resilience Index shows an initial decrease followed by an increase: it fell from 0.343 in 2010 to 0.330 in 2015, before surging to 0.369 in 2020, marking significant growth in the latter period. This may be related to intensified investment in economic development and disaster prevention capacity building. In contrast, Yunnan’s driving resilience index shows an ‘increase-then-decrease’ pattern, rising from 0.294 in 2010 to 0.322 in 2015 before falling to 0.280 in 2020, which may be due to adjustments in regional development priorities. Meanwhile, the S–Y region has shown overall stability, with a slight increase from around 0.294 in 2010 to 0.300 in 2015 and 0.301 in 2020, reflecting its fundamental role in supporting resilience.

Pressure resilience (c): Its trend is identical to that of driving resilience and its data overlaps with the latter’s. For Sichuan, the figures are 0.343 in 2010, 0.330 in 2015 and 0.369 in 2020. For Yunnan, the figures are 0.294 in 2010, 0.322 in 2015 and 0.280 in 2020. For the Sichuan–Yunnan region, the figures are approximately 0.294 in 2010, 0.300 in 2015 and 0.301 in 2020. These figures indicate a strong correlation between driving factors and pressure resilience. Sichuan’s ability to cope with disaster risk pressure improved significantly during the latter period, whereas Yunnan’s pressure resilience declined. This decline stems from adjustments to the region’s development strategies.

State resilience (d): Yunnan’s state resilience increased rapidly, rising from 0.405 in 2010 to 0.486 in 2015 before decreasing slightly to 0.484 in 2020. This reflects significant improvements in earthquake-related urban resilience and resource allocation, with minor fluctuations in the latter period possibly resulting from management pressures associated with urban expansion. Sichuan: Its state resilience steadily increased from 0.251 in 2010 to 0.275 in 2015 and has remained stable since then, indicating that its improvement is more sustainable. S–Y region: It rose from 0.297 in 2010 to 0.344 in 2015, and then decreased slightly to 0.343 in 2020. This shows an overall trend of increase followed by stability, reflecting the coordinated development of regional state resilience.

Impact resilience (e): Sichuan experienced a strong ‘V-shaped rebound’, with its impact resilience score decreasing from 0.533 in 2010 to 0.499 in 2015 before soaring to 0.603 in 2020. This reflects a substantial improvement in its ability to respond to secondary environmental risks and economic shocks from earthquakes. Yunnan: It decreased from 0.492 in 2010 to 0.451 in 2015, before rising to 0.501 in 2020, indicating a clear recovery trend. S–Y region: It fell from 0.472 in 2010 to 0.438 in 2015 before increasing to 0.510 in 2020. This achieved cross-stage growth in overall impact resilience and was possibly due to improvements in regional collaborative disaster reduction mechanisms.

Response resilience (f): All three areas exhibit a ‘V-shaped’ trend. Sichuan: Decreased from 0.271 in 2010 to 0.237 in 2015, before rebounding to 0.264 in 2020. Yunnan: Dropped from 0.207 in 2010 to 0.183 in 2015, then increased to 0.216 in 2020. S–Y region: Fell from 0.221 in 2010 to 0.194 in 2015, then increased to 0.222 in 2020. These figures suggest that significant investments were made in infrastructure, emergency information systems and medical rescue capabilities after 2015, driving a rapid recovery and improvement in response resilience in the three areas.

In summary, the seismic resilience and related subsystems in the S–Y region (Sichuan and Yunnan) each exhibit distinct characteristics and trends. Sichuan shows significant growth in certain subsystems, such as impact and overall resilience, during the later period. Meanwhile, Yunnan’s state resilience fluctuates at a high level and the S–Y region achieves stable improvement overall. Understanding these trends is crucial for formulating differentiated strategies to enhance seismic resilience in different regions.

3.1.2. Regional Evolution of Overall Urban Resilience

A comprehensive analysis of seismic resilience indicators for 37 prefecture-level cities in the S–Y region was conducted, and on the basis of this analysis and of previous research, a five-tier classification system was developed. The system utilised for the categorisation of cities is as follows: low resilience (0–0.3); relatively low resilience (0.3–0.4); moderate resilience (0.4–0.5); relatively high resilience (0.5–0.6); or high resilience (0.6–0.85). The creation of seismic resilience zoning maps for the Chuan-Dian region in 2010, 2015 and 2020 was conducted using ArcGIS (10.8) software, with the aforementioned framework being employed (Figure 4).

The results of the spatial analysis indicate a marked spatiotemporal heterogeneity in urban seismic resilience across the study area. In 2010, low-resilience zones were predominantly concentrated in northern Sichuan (indicated in red), while relatively low-resilience zones were distributed in southern Sichuan (indicated in orange), forming a spatial pattern of “low in the west and high in the central and eastern regions.” By 2015, the prevalence of these clusters had significantly diminished, with their presence remaining only in a few areas of central Sichuan, such as Ya’an. By 2020, low- and relatively low-resilience zones were primarily concentrated in Yibin and Aba, respectively, while the proportion of relatively high-resilience areas increased markedly, accounting for more than 51% of all cities. The collective resilience of the region has undergone a substantial enhancement, manifesting in a spatial configuration that can be delineated as “high in the south and low in the north.” Throughout the study period, urban seismic resilience demonstrated a consistent upward trend.

With regard to spatial cores, Chengdu (Sichuan Province) and Kunming (Yunnan Province) emerged as dual centres of resilience enhancement, demonstrating a radial diffusion of resilience towards surrounding cities. It is evident that both cities exhibited high levels of resilience (green) throughout the three-year period under consideration, indicative of consistent and sustainable seismic adaptability. Conversely, regions constrained by topography and developmental conditions–such as the elevated mountains in northern Sichuan and the mid-altitude mountainous regions in western Yunnan–exhibited reduced resilience levels, a spatial pattern consistent with the east–west elevation gradient.

Quantitative analysis further reveals a clear declining trend in the number of cities with medium or lower resilience (corresponding to red, orange, and yellow categories). Among the 37 cities analysed, 20 cities (54.05%) were at medium or below resilience levels in 2010; 21 cities (56.76%) in 2015, as low-resilience zones decreased but medium resilience remained widespread; and only 14 cities (37.84%) in 2020, indicating a substantial improvement in overall resilience.

Significant population growth has been observed in satellite cities surrounding Chengdu, such as Suining, Ziyang, and Neijiang, driven by rapid urbanisation and economic growth [71]. However, the limited availability of urban space and the persistent seismic pressure have constrained their resilience, resulting in a medium resilience level (yellow), indicating the urgent need for targeted enhancement strategies. Topographic variations have resulted in disparities in economic development, infrastructure, and public services, thereby further widening inter-city resilience gaps. In these gaps, low-altitude mountainous areas generally exhibit high resilience (green), while cities adjacent to plains tend to maintain medium resilience (yellow).

The plains demonstrate strong socioeconomic driving forces at the regional level, but relatively limited seismic resilience due to systemic vulnerabilities and amplified disaster effects. Conversely, low-altitude mountainous areas have been shown to display more balanced subsystem performance, resulting in higher comprehensive resilience (as illustrated by extensive green areas on the maps). This evident spatial heterogeneity underscores the necessity of developing region-specific and differentiated resilience optimisation strategies that align with local geophysical and socioeconomic characteristics.

3.1.3. Regional Evolution of Seismic Resilience Subsystems

As demonstrated in Figure 5, the spatial–temporal evolution of the driving force subsystem of urban seismic resilience exhibits marked heterogeneity. In 2010, low driving-force areas (red) emerged in southern Sichuan and Yunnan, while high values (green) were concentrated in the Chengdu and Kunming agglomerations, reflecting early development advantages. By 2015, low-value zones had expanded across northwestern Sichuan and southern Yunnan, suggesting weakened subsystem support for resilience. By 2020, high-value areas in Chengdu and Kunming had expanded significantly, indicating enhanced economic and demographic driving forces.

The overall increase in the driving force subsystem was steady, with economic growth, population expansion and urbanisation acting as primary drivers. The Chengdu and Kunming agglomerations demonstrated the most rapid growth, evolving into mature metropolitan economic circles. However, as population density approached saturation, development and seismic risks intensified. It is important to note that strong driving forces do not necessarily guarantee high levels of comprehensive resilience, as vulnerability and recovery capacity also play pivotal roles. Consequently, the conversion of driving-force advantages into actual resilience enhancement necessitates coordinated planning, infrastructure upgrading, and investment in disaster prevention.

The phenomenon of regional pressure is closely associated with active faults and plate tectonic movements. The present study calculates pressure-based seismic resilience by integrating seismic and fault data from the five years prior to each time node, as illustrated in Figure 6. In 2010, the prevalence of low resilience (red) was observed in western high-to-mid-altitude mountainous regions, while medium resilience (yellow), relatively high resilience (light green), and high resilience (dark green) were concentrated in eastern plains, hilly regions, and low-altitude mountainous areas. By 2015, the western low-resilience zone remained prominent, yet the extent of relatively high and high-resilience (green tones) in the east expanded moderately. In 2020, high resilience (dark green) in the eastern regions remained stable, while low resilience (red) persisted in the west.

Additionally, sporadic low-resilience patches emerged in parts of the east. The spatial pattern that emerges is one of “east–high, west–low”. Throughout the period under investigation, the eastern plains, hilly areas and low-altitude mountainous regions demonstrated high levels of resilience. Conversely, western high- to mid-altitude mountainous regions exhibited low levels of resilience. This discrepancy is consistent with the spatial heterogeneity of regional plate movement activity and fault distribution density.

The regional state of resilience is indicative of the current urban earthquake resistance level, which is influenced by the combined effect of driving forces and pressures. Associated factors include regional seismic fortification standards, the number of parks, and the number of emergency shelters, as illustrated in Figure 7. In 2010, low resilience (red) was concentrated in eastern Sichuan (e.g., Guang’an, Luzhou) and western Yunnan (Dehong). Medium resilience (yellow) covered areas such as Aba, Pu’er, and Wenshan. Resilience levels of relatively high (light green) and high (dark green) were mainly distributed in western Sichuan (Garze, Liangshan) and central Yunnan (around Kunming), initially forming a pattern of “higher in the south, lower in the north”. By 2015, the low resilience (red) area had diminished, becoming concentrated in certain regions of eastern Sichuan. The range of medium resilience (yellow) exhibited negligible change, while the areas of relatively high and high resilience (green tones) underwent substantial expansion in central Yunnan and western Sichuan, thereby accentuating the “higher in the south, lower in the north” pattern. In 2020, the area of low resilience (red) was further diminished, with the presence of this resilience only documented in a few areas of eastern Sichuan. Conversely, relatively high resilience (light green) exhibited an expansion around Chengdu, while high resilience (dark green) remained stable in western Sichuan and central Yunnan.

The overall spatial pattern of “higher in the south, lower in the north” was maintained, and the coherence of high-resilience areas in the south was enhanced. In conclusion, the state resilience displays a pattern of “higher in the south, lower in the north”. It is evident that, over time, the advantage of high-resilience areas in the south becomes more apparent, while low-resilience areas in the north gradually diminish. This phenomenon is closely related to the spatial–temporal optimisation of regional seismic facility configuration and emergency resource layout in urban planning.

The concept of regional impact resilience pertains to the influence of seismic disasters on diverse urban components. As demonstrated in Figure 8, there is an evident imbalance in the spatial pattern over time. In 2010, the areas of Garze in western Sichuan, and the districts of Aba and Luzhou in southern Sichuan, exhibited low resilience (red). Resilience levels of a comparatively lower order were evident in Aba, Panzhihua, and Dali (orange). Medium resilience (yellow) was observed in Guangyuan, Nanchong, and Wenshan. The regions of Xishuangbanna, Pu’er, and Lincang in southern Yunnan, as well as Liangshan in southern Sichuan, demonstrated relatively high and high resilience (green tones). By 2015, the phenomenon of low resilience persisted in western Sichuan’s Garze and emerged in central Yunnan’s Chuxiong; relatively low resilience spread in Aba, Honghe, and Wenshan; green-toned resilience areas were seen in Liangshan and Pu’er. In 2020, the areas of western Yunnan (Dali, Lijiang, and western Sichuan (Diqing and Nujiang) exhibited low resilience. Wenshan and Chuxiong demonstrated relatively low resilience, while Kunming and Nanchong exhibited medium resilience. In contrast, southern Yunnan (Xishuangbanna, Pu’er), parts of Garze in western Sichuan (light green), and northern Sichuan (Guangyuan and Bazhong) exhibited relatively high and high resilience. The imbalance can be attributed to a number of factors, including urban development and exposure to disasters. Cities in the vicinity of Chengdu, whether in plain or hilly areas, exhibit high economic levels and have undergone complete urban development. However, these cities experience significant vegetation coverage gaps when compared to other regions. Consequently, they are particularly vulnerable to earthquakes with respect to impact resilience. In certain periods, these cities demonstrate comparatively low regional impact resilience. Concurrently, regions characterised by active tectonics or underdeveloped urban areas in western Sichuan and western Yunnan demonstrated low resilience due to inadequate disaster mitigation capacity. This observation highlights the interactive effects of economic development, urbanisation and seismic disaster impacts on regional impact resilience.

As demonstrated in Figure 9, the urban response resilience demonstrates an uneven distribution pattern that evolves over time. In 2010, regions exhibiting low resilience (indicated in red) were predominantly located in Dehong, Nujiang, and Lijiang in western Yunnan, along with Panzhihua in Sichuan and Xishuangbanna. The topographical features of the region are such that there are areas with relatively low resilience (orange) which are surrounded by Chengdu, including Meishan and Ziyang. The medium-resilience zones that have been delineated are those encompassing Aba, Wenshan, and Pu’er. Concurrently, relatively high and high-resilience areas (green tones) were distributed across parts of western Sichuan, including Garze and Liangshan, as well as regions surrounding Kunming and Honghe in central Yunnan. By 2015, the area classified as low resilience (orange) around Chengdu had expanded, medium-resilience areas (green) in Pu’er and Wenshan had become more distinct, and the green-toned regions in Garze, Liangshan, and around Kunming had undergone changes. In 2020, the presence of low-resilience areas was observed in Dehong, Nujiang, Lijiang, Panzhihua, Xishuangbanna, and the vicinity of Ya’an in western Sichuan. The investigation revealed relatively low levels of resilience in Aba and parts of Chengdu, medium levels in southern Aba and Wenshan, and high levels of resilience in the green-toned areas of Garze, Liangshan, the vicinity of Kunming, and Honghe.

This uneven distribution can be attributed to significant regional disparities. Low-resilience areas are characterised by their location in mid- to high-altitude mountainous regions, a notable example of which can be found in western Yunnan and western Sichuan. These regions are characterised by underdeveloped economic infrastructure, constrained fiscal resources, and inadequate investment in healthcare and educational infrastructure. These factors collectively contribute to the regions’ limited capacity for disaster response and recovery. Conversely, high-resilience areas, such as those surrounding Kunming, have been shown to experience more favourable economic conditions. A substantial scale of earthquake-resistant fiscal expenditure, in conjunction with the ongoing enhancement of medical service capacity through a variety of initiatives, has been demonstrated to be efficacious in the augmentation of the regional response resilience level.

3.2. Spatial Correlation Analysis of Urban Seismic Resilience in S–Y Region

3.2.1. Global Spatial Autocorrelation Analysis

The software programme GeoDa (version 1.16.0.12) was utilised to calculate spatial weights [72], and the global autocorrelation coefficients (Moran’s I) of urban seismic resilience in the S–Y region for the years 2010, 2015, and 2020 were calculated. The Moran’s I values for the three five-year periods are −0.051, 0.028, and −0.020, respectively. It is evident that the year 2015 demonstrates a positive value, while 2010 and 2020 exhibit negative values. This finding indicates that, during the study period, urban seismic resilience in the S–Y region exhibited fluctuating spatial autocorrelation characteristics, with negative spatial autocorrelation in 2010 and 2020 and weak positive spatial autocorrelation in 2015.

It is evident from the change in Moran’s I value that the most significant increase occurred during the 12-th Five-Year Plan period, with a rise from −0.051 to 0.028. During this period, the spatial negative correlation of urban seismic resilience shifted to a weak positive correlation, spatial heterogeneity weakened significantly, spatial correlation enhanced, and the spatial agglomeration trend and homogeneity of earthquake resilience became more prominent. During the 13-th Five-Year Plan period, the Moran’s I value returned to negative (from 0.028 to −0.020), indicating that the positive spatial agglomeration pattern did not persist, and spatial randomness increased to some extent.

The urban seismic resilience in the S–Y region has been shown to demonstrate a general upward trend, accompanied by fluctuating spatial agglomeration characteristics. The spatial distribution of these settlements is primarily influenced by various factors, including policies, economics, and topography. Notable among these factors are the “Western Development Strategy”, the emergence of regional tourism, and the distribution of terrain. The implementation of national policies has precipitated the rapid development of the regional economy, the enhancement of the urban economic level, and the augmentation of investment in earthquake resistance and disaster reduction. This has engendered a temporary weakening of spatial heterogeneity and the short-term formation of positive spatial agglomeration during the 12-th Five-Year Plan period. However, the subsequent return to negative Moran’s I in 2020 reflects the complexity of spatial pattern evolution, which may be related to uneven regional development or changes in disaster prevention and mitigation strategies.

3.2.2. Local Spatial Autocorrelation Analysis

The LISA (Local Indicators of Spatial Association) agglomeration map of urban seismic resilience in the S–Y region has been created using GeoDa and ArcGIS (10.8), as illustrated in Figure 10. During the study period, the local spatial agglomeration of urban resilience in the S–Y region presents distinct phased characteristics.

In 2010, notable agglomeration types include H-L cluster areas (orange), concentrated around Chengdu, and L-H cluster areas (grey-blue), in Chuxiong, while most regions show “Not Significant” (light grey).

During the 12-th Five-Year Plan period (2015), H-H cluster areas (red) emerged prominently in southeastern Yunnan, covering cities such as Chuxiong, Yuxi, Honghe, and Wenshan, indicating a local small-scale high-value spatial agglomeration. Concurrently, L-H cluster areas (grey-blue) manifested in Pu’er, while other regions exhibited minimal significance.

In 2020, the distribution of H-L cluster areas (orange) was centred on Chengdu and Deyang. The L-H cluster areas (grey-blue) exhibited persistence in Chuxiong, Yuxi, and Wenshan, while a L-L cluster area (blue) emerged in Ya’an.

The local spatial agglomeration phenomenon is not pervasive, but rather shows a phased agglomeration trend. The “H-H” agglomeration was concentrated in southeastern Yunnan during the 12-th Five-Year Plan period, while the “H-L” agglomeration centred around Chengdu. “L-H” and “L-L” agglomerations are scattered in parts of Yunnan and western Sichuan, reflecting the spatial heterogeneity and phased agglomeration characteristics of urban seismic resilience in the S–Y region. These characteristics are closely related to regional economic development, policy implementation, and topographic conditions.

4. Discussion

4.1. Significance and Current Situation of Urban Seismic Resilience in the S–Y Region

The S–Y region occupies a pivotal role in the domain of earthquake resistance, both within China and on a global scale. In the field of geological research, the continuous exploration of complex geological structures has yielded invaluable evidence for earthquake prediction and seismic fortification. In the field of technological innovation, a suite of advanced seismic building technologies and earthquake monitoring equipment have been developed. The disaster relief system has undergone continuous improvement, with the establishment of an efficient emergency response mechanism and rescue force deployment plan. Furthermore, significant progress has been made in cultivating public awareness. It is evident that the public’s awareness of earthquake prevention and disaster mitigation has been significantly enhanced through the implementation of comprehensive publicity and educational initiatives. These endeavours have resulted in invaluable contributions to the realm of earthquake disaster management and the safeguarding of human life and property. However, the reality is that the losses caused by earthquake disasters remain at a relatively high level, highlighting the urgency and necessity of strengthening the construction of urban seismic resilience in the region and improving urban earthquake resistance capabilities.

4.2. Temporal Evolution Characteristics of Urban Seismic Resilience in the S–Y Region

From the perspective of the time dimension, during the period from 2010 to 2015, the resilience levels of the areas around Chengdu and Kunming underwent change. In certain regions initially exhibiting comparatively diminished resilience, an enhancement in their resilience levels was observed, a phenomenon that is predominantly ascribed to the development of infrastructural frameworks or the refinement of emergency management systems during the specified period. However, the resilience levels of some high-altitude and mid-altitude mountainous areas remained relatively unchanged, indicating that these areas face considerable challenges in enhancing seismic resilience. From 2015 to 2020, this dynamic change continued. The degree of regional resilience exhibited a continuous upward trend, with the geographical area encompassing high-resilience zones undergoing constant expansion. This phenomenon is indicative of the positive promotional effect of urban development on enhancing seismic resilience. However, the enhancement of seismic resilience in remote mountainous regions remains progressing at a markedly languid pace. This underscores the necessity for substantial resource investment and the development of meticulously targeted measures to augment seismic resilience in these areas [73,74].

4.3. Analysis of Urban Seismic Resilience Subsystems in the S–Y Region

The urban seismic resilience of the S–Y region comprises five distinct subsystems: driving force, pressure, state, impact, and response. The aforementioned subsystems are interconnected, and collectively determine the overall seismic resilience of the region.

In terms of driving force resilience, economic development and urbanisation have been shown to promote urban earthquake resistance. However, the unbalanced development and industrial structure adjustment have led to fluctuations or declines in the driving force resilience of some regions, affecting the resource acquisition and development momentum of other subsystems. With regard to pressure resilience, there are numerous active faults in the western part of the S–Y region, and the area is subject to frequent seismic activity. Despite the low population density typically observed in high-altitude mountainous regions, these areas are susceptible to significant seismic activity and subsequent secondary disasters, which possess a considerable destructive capacity. The high-pressure environment has been shown to have a detrimental effect on state resilience, to increase losses of impact resilience, and to impose considerable requirements on response resilience. Inadequate monitoring and planning will result in a reduction in the overall earthquake resistance capacity. The state’s capacity for resilience is reflected in fundamental conditions, including the implementation of seismic standards and the provision of emergency shelters. In Yunnan, the state resilience was enhanced in the early stage by means of raising standards and expanding shelters. However, this resilience declined later due to urban expansion. In the Sichuan region, there has been a consistent enhancement of state resilience through dynamic optimisation, thereby providing a foundation for overall resilience. Impact resilience is defined as the capacity to withstand the consequences of seismic disasters, and is found to be comprehensively influenced by other subsystems. In the early stages, Yunnan and the region as a whole were more susceptible to disasters. However, the implementation of relevant measures in subsequent periods has led to an enhancement in their resilience. It is evident that the province of Sichuan has been successful in mitigating the growth through the implementation of the emergency mechanism. This serves to emphasise the significance of subsystem collaboration in order to reduce the impact.

The concept of response resilience is integral to the entire process of earthquake response, and its development is contingent on the progression of other subsystems. Following 2015, Yunnan and Sichuan have made significant progress in terms of their response capabilities through the implementation of various measures. The experience gained from these measures has also led to the optimisation of other subsystems.

4.4. Applicability of the Proposed Methodological Framework to Global Earthquake-Prone Regions

Although the present study focuses on a specific region, the methodological framework that has been proposed–which integrates DPSIR subsystem evaluation, spatial autocorrelation analysis (Moran’s I), and the CRITIC–AHP combined weighting method–exhibits inherent potential for application in other earthquake-prone regions worldwide. However, the transferability of such models is predicated on the careful adaptation to regional geological, socioeconomic, and data-specific contexts.

4.4.1. Universality of the Core Methodological Framework

The core framework developed in this study aligns with internationally recognised resilience assessment principles, making it applicable across diverse seismic contexts. The DPSIR structure provides a systematic means of linking human activities, environmental pressures, and systemic responses, thereby enabling a holistic understanding of resilience formation. In a similar vein, spatial statistical techniques, including Moran’s I, have exhibited universal validity in identifying spatial clustering and dependence within resilience distributions. The collective efficacy of these analytical elements is such that the framework can be effectively extended to other regions, provided that local contextual adjustments are made.

4.4.2. Targeted Methodological Adaptations for Regional Specificity

Despite its universality, successful regional application of the framework requires targeted methodological adaptations.

It is imperative to emphasise the significance of indicator localization. The indicator system should be refined to reflect local hazard characteristics, socioeconomic structures, and institutional conditions, ensuring both relevance and comparability. This contextual adaptation enables the framework to accurately represent the multi-dimensional drivers and responses of seismic resilience in distinct environments.

Secondly, the weighting of indicators necessitates optimisation to achieve equilibrium between subjectivity and objectivity. The CRITIC–AHP combined weighting method adopted in this study effectively integrates objective data dispersion with expert judgment, thereby enhancing the credibility of the assessment. However, disparities in data quality and expert consistency across regions may potentially impact the stability of weighting. It is recommended that future studies explore the potential of adaptive weighting approaches, with a view to better reflecting regional data heterogeneity and interdependencies among indicators.

Finally, spatial weight configuration represents another critical methodological consideration. A key limitation of the study was the absence of calibration based on administrative adjacency or geological continuity. In future applications, region-specific spatial weighting strategies could be developed to more accurately capture the spatial diffusion of seismic impacts and the structural connectivity of built environments, thereby improving the robustness of spatial pattern analysis.

The proposed framework is universally applicable due to its systematic integration of multi-dimensional drivers and spatial dependencies of seismic resilience. In order to ensure effective transferability, it is essential that indicator systems be locally adapted, weighting schemes be refined, and spatial parameters be optimised in accordance with regional conditions. These adjustments are underpinned by the principle of "precision adaptation," which supports evidence-based and context-sensitive resilience enhancement in diverse global earthquake-prone regions.

4.5. Strategies for Enhancing Urban Seismic Resilience in the S–Y Region

In order to enhance urban seismic resilience in the S–Y region, it is necessary for all subsystems to collaborate. In the future, there is a necessity to coordinate overall development, strengthen the driving force, reduce pressure response, enhance the state, stabilise the impact, strengthen the response, comprehensively improve regional seismic resilience, and ensure safety and sustainable development.

It is evident from the analysis of the weights and compositions of the scientific evaluation model that aspects such as response-ability, driving force, and state exert a relatively high influence on the urban resilience level. In consideration of the aforementioned points, in conjunction with the spatial distribution phenomena and the conclusions drawn from the model analysis, the following targeted enhancement strategies are proposed (see Figure 11):

(1)

The enhancement of population factors is a key priority. It is imperative to optimise the internal structure of urban agglomerations and the industrial structure. It is imperative that big cities develop modern service industries in order to facilitate the transfer of the employed population. Similarly, medium and small cities should concentrate on the development of the manufacturing industry with a view to attracting the floating population to find employment in the immediate vicinity. Meanwhile, small towns should encourage the establishment of characteristic industries and increase policy support. It is imperative to establish a rational industrial distribution strategy to mitigate the escalating urban pressures precipitated by unregulated population mobility.

(2)

Enhancement of responsibility. It is imperative that measures are taken in the following areas: education, medical care, transportation, and budget expenditure. In the field of education, it is recommended that universities in Sichuan and Yunnan incorporate relevant courses into their curriculum, organise practical activities, and establish internship bases. In the field of medical care, the following measures are recommended: firstly, an increase in the number of doctors and hospital beds is to be effected; secondly, hospitals are to be built or expanded; thirdly, emergency plans are to be formulated; and fourthly, resource integration is to be strengthened. In the domain of transportation, it is imperative to optimise the planning process, augment investment, reinforce and renovate existing road infrastructure, and establish emergency channels as a contingency. In the context of budget expenditure, the establishment of special funds is imperative. These funds must be allocated and supervised with a reasonable degree of oversight, and the involvement of social capital is to be encouraged.

(3)

Economic improvement plan: It is recommended that the government consider ways in which it can increase its financial investment, establish special funds, and raise funds through multiple channels. The development of characteristic industries, the strengthening of regional cooperation, the encouragement of enterprises to participate, the optimisation of financial services, the improvement of the quality of the labour force, and the increase in investment in scientific and technological research and development are all recommended.

Figure 11. Urban seismic resilience Improvement Plan.

Figure 11. Urban seismic resilience Improvement Plan.

The utilisation of urban space must be rational. A comprehensive strategy is required to address the challenges posed by urban planning. This strategy should encompass the optimisation of functional areas, the augmentation of public spaces and green areas, the optimisation of the transportation network, the development of underground spaces, the renovation of old urban areas, the promotion of prefabricated buildings, the reservation of land for emergency facilities, and the strengthening of the control of building spacing.

The enhancement of urban seismic resilience in the S–Y region is a systematic and complex project that requires comprehensive consideration of multiple aspects, including but not limited to population, responsibility, urban space utilisation, and infrastructure. Concurrently, cognisance of the characteristics and disparities among diverse regions is imperative, and the formulation of targeted strategies should be undertaken. The implementation of the aforementioned comprehensive measures is expected to effectively strengthen the urban resilience ability to deal with earthquake disasters, thereby laying a solid foundation for safeguarding people’s lives and property, as well as the sustainable development of cities. In the future, it is imperative to direct continuous attention to the development of seismic resilience in this region. Relevant strategies must be continuously adjusted and improved according to actual conditions in order to better meet the challenges posed by earthquake disasters.

5. Conclusions

(1)

The present study is based on the DPSIR model. An urban seismic resilience evaluation system for the Sichuan–Yunnan (S–Y) region was constructed, and indicator weights were determined using the combined weighting method of CRITIC and AHP (e.g., the weight of the Pressure criterion layer is 0.36, and that of the Response criterion layer is 0.24). The findings indicate that between 2010 and 2020, the overall regional seismic resilience index increased from 0.501 to 0.526. Sichuan’s overall resilience exhibited a “decline first, then rise” trend (0.570 → 0.566 → 0.585), while Yunnan’s demonstrated continuous growth (0.517 → 0.557), which is consistent with the direction of national relevant strategies.

(2)

The spatial distribution of resilience exhibits dynamic heterogeneity: in 2010, the pattern was “low in the west and high in the central and eastern regions”, and it shifted to “high in the south and low in the north” by 2020, with cities with relatively high resilience accounting for over 51%. Chengdu and Kunming have traditionally exhibited dual core characteristics, demonstrating high levels of resilience that extend to surrounding regions. However, high-altitude mountainous areas in western Sichuan and mid-altitude mountainous areas in western Yunnan exhibited low resilience, a phenomenon attributable to the presence of concentrated fault zones and pervasive economic backwardness. The satellite cities around Chengdu (e.g., Suining) exhibited medium resilience, attributable to the superposition of limited urban space and seismic pressure.

(3)

The five subsystems demonstrate distinct evolutionary characteristics. With regard to Driving Force resilience, Sichuan exhibited a “decline first, then rise” trend (0.343 → 0.369), while Yunnan exhibited a “rise first, then decline” trend (0.294 → 0.280). Pressure resilience followed a consistent trend with Driving Force resilience. With regard to State resilience, Yunnan exhibited a “rapid rise then stability” trend (0.405 → 0.484), and Sichuan exhibited “steady growth” (0.251 → 0.275). Impact resilience exhibited a “V-shaped” recovery (0.472 → 0.438 → 0.510), and Response resilience rebounded significantly after 2015.

(4)

Spatial correlation demonstrated fluctuating characteristics: the global Moran’s I index indicated a weak negative correlation in 2010 (−0.051) and 2020 (−0.020), and shifted to a weak positive correlation in 2015 (0.028). The process of local agglomeration was implemented in a phased manner. The H-H clusters emerged in southeastern Yunnan in 2015, the H-L clusters persisted around Chengdu, and the L-L clusters were newly added in Ya’an in 2020. These findings reflect the impact of economic gaps and terrain on the spatial distribution of resilience.