Analysis of the Spatio-Temporal Evolution and Driving Factors of Urban Cascading Disaster Resilience Based on Spatial Theory

Abstract

To address the increasing threat of urban cascading disasters to the resilient development of cities, this paper establishes an urban cascading disaster resilience evaluation system from the three dimensions of “physical space–social space–information space”. Based on the idea of game theory, the entropy weight–CRITIC method is used to obtain the optimal combination weight of the indicators. Taking Jiangsu Province as an example, the TOPSIS model is used to calculate the urban cascading disaster resilience values of each city from 2014 to 2023. Then, ArcGIS (10.8) spatial analysis tools, Kernel density analysis, spatial correlation analysis, and the geographical detector are used to explore the spatio-temporal pattern evolution, spatial correlation, and driving factors of urban cascading disaster resilience in Jiangsu Province. The results show that (1) there is obvious regional heterogeneity in the urban cascading disaster resilience of Jiangsu Province in space; (2) the overall urban cascading disaster resilience of Jiangsu Province shows an upward trend over time; (3) there is a significant positive correlation between the urban cascading disaster resilience of Jiangsu Province and geographical location; and (4) there are high-level driving factors in each space.

1. Introduction

In the era of rapid urbanization, cities have become the core areas of human, social, and economic activities. However, the high population density and complex systems of cities make them more vulnerable when facing natural and man-made disasters. Especially cascading disasters—a chain reaction where one disaster event triggers others—pose a significant threat to the stability and safety of urban systems [1]. The complexity of urban cascading disasters lies in their chain reactions. For instance, an earthquake may damage urban infrastructure, leading to secondary disasters such as fires, floods, and even chemical leaks. This multi-hazard coupling phenomenon causes the destructive power of disasters to increase exponentially, and the difficulty of urban recovery also rises significantly [2]. Globally, major disaster events in recent years have fully exposed the vulnerability of cities to cascading disasters. The 2011 Fukushima nuclear accident in Japan, triggered by an earthquake and followed by a tsunami, led to a nuclear power plant leak, which not only caused huge casualties and economic losses but also had a profound impact on global nuclear safety governance. In China, with the acceleration of urbanization, the disaster risks faced by cities are also increasing. For example, the extreme rainstorm in Zhengzhou, Henan Province in 2021, triggered a series of chain reactions such as urban flooding, subway suspension, and power outages, causing a huge impact on urban operations and residents’ lives.

Currently, the planning and construction of resilient cities represent a new trend and stage in urban disaster management, which have been widely recognized and applied globally [3]. It covers comprehensive disaster management from prevention to response and recovery [4]. Incorporating resilience theory into the research framework of cascading disasters can provide solid theoretical support and practical guidance for cities to build a defense system against cascading disasters. However, current research on cascading disasters mainly focuses on the mechanisms and processes of cascading disasters [5], studying the interactions and triggering mechanisms between disasters, such as exploring the characteristics and risk evolution patterns of compound disasters in the Yellow River Basin [6]. Some research is on the prediction and modeling of cascading disasters [7], such as using Bayesian methods to explore the evolution process of earthquake cascading disasters in chemical industrial parks [8] and identifying the main factors that lead to the expansion disaster. There are also studies on how to quantitatively assess the risks of cascading disasters [9], such as evaluating the impact of cascading disasters on urban infrastructure [10].

Most of these studies focus on the resilience measures at a “certain moment”, lacking a description of spatio-temporal evolution and ignoring the influencing factors that affect the resilience of cascading disasters. This makes it difficult to distill the evolution laws of resilience over a long time dimension and to construct resilience enhancement strategies with temporal continuity and spatial targeting based on this. Moreover, there is still insufficient research on systematically assessing the resilience of urban cascading disasters from a multi-dimensional perspective, and less consideration is given to the coordinated cooperation among cities. As a complex giant system, cities cover multiple aspects such as physical space, social structure, and information networks [11]. The “physical-social-information” multi-dimensional perspective can be introduced to assess the resilience of urban cascading disasters. This theory has been widely applied in resilience evaluation, such as earthquake safety resilience [12] and community building seismic resilience [13].

Based on the deficiencies of existing research, this paper proposes to explore the key factors influencing urban cascading disaster resilience from a multi-dimensional perspective of “physical-social-information” in spatial theory and establish an evaluation system. Then, by combining the application of the TOPSIS evaluation model, spatial correlation analysis, and the geographical detector method, it summarizes the spatio-temporal evolution laws of urban cascading disasters in the study area, explores the spatial distribution characteristics and driving factors, and draws relevant conclusions. Finally, it puts forward differentiated resilience improvement measures that break through administrative boundaries. This provides reference suggestions and decision support for the construction of a new paradigm of urban cascading disaster resilience governance with joint prevention and control in large urban agglomerations.

Our research focuses on urban resilience development, an area closely related to multiple sustainable development goals. By exploring the adaptability and recovery capabilities of cities in the face of natural disasters and socio-economic challenges, our work not only contributes to achieving sustainable cities and communities, but also enhances environmental and social resilience through optimizing urban planning and infrastructure construction. Moreover, it promotes good health and well-being by improving public health facilities and community layouts. In the future, we will continue to deepen interdisciplinary research, promote policy innovation and public participation, and provide scientific evidence and practical solutions for urban resilience development to support the realization of global sustainable development goals.

2. Theoretical Framework

2.1. Analysis of Urban Cascading Disasters Under Resilience Theory

2.1.1. Urban Cascading Disasters

Cascading disasters have significant uniqueness compared to traditional disasters. They occur in a complex environment where natural and social systems are intertwined. A single disaster event alone often fails to trigger a cascading effect [14]. Cascading disasters also exhibit a hierarchical difference through amplification at different stages, and this “amplification” is a key characteristic that distinguishes them from traditional disasters. Additionally, cascading disasters demonstrate the interaction among different risk factors [5]. In urban environments, this complexity is further magnified [15]. Urban systems are highly complex, interdependent, and highly coupled. An initial disaster event can trigger other disaster events through direct or indirect means, causing damage to multiple areas and spaces of the city and resulting in extensive losses. Therefore, urban cascading disasters are distinct from general disasters, featuring nonlinearity, chain reactions, interactivity, and amplification [16].

2.1.2. Urban Cascading Disaster Resilience

Resilience refers to the ability of a system to resist risks when disturbed. It encompasses three core capabilities: resistance, recovery, and adaptability [4]. In the context of urban cascading disaster resilience, resistance is mainly reflected in the physical space, through the robustness of infrastructure, the seismic and flood resistance of buildings, and the rationality of urban planning, effectively reducing the direct damage of the initial disaster to urban functions. Recovery is mainly reflected in the social and information spaces. Cities need to have strong information collection and analysis capabilities, using big data and artificial intelligence technologies to quickly analyze the evolution path and impact range of cascading disasters, and promptly provide medical assistance and living supplies. Adaptability is mainly reflected in the physical and social spaces. After a cascading disaster occurs, cities need to conduct in-depth analysis to identify the key points that lead to the escalation of the disaster and take targeted measures to strengthen and improve them, achieving a shift from passive response to active defense. The improvement of adaptability will, in turn, enhance the city’s resistance and recovery capabilities in the face of disturbances. Resistance, recovery, and adaptability keep the city in a state of dynamic evolution and continuous optimization.

2.2. Spatial Theoretical Model of Urban Cascading Disaster Resilience

Urban expansion, high density and agglomeration of population, and multiple economic structures make urban development face unprecedented uncertainties and unknown risks [17]. The new type of composite risk represented by the cascading disaster involves engineering, economic, social, information and other systems. Analyze the characteristics of urban cascading disasters, construct a resilience evaluation system for urban cascading disasters in three dimensions of physical, social and information based on spatial theory, and propose countermeasures to enhance the sustainable development of cities.

First, analyze the components of the three spaces in the city. The physical space mainly includes infrastructure, buildings, natural ecosystems, etc.; the social space mainly contains economy, culture, medical care, education and norms, etc.; the information space mainly includes communication networks, the Internet, big data and platforms, etc. When responding to urban cascading disasters, fully consider the characteristics of the disasters and improve the comprehensive construction level of each space: the physical space focuses on the evolution characteristics of disasters, and when designing infrastructure or buildings, take into account multiple risks such as earthquake resistance and fire prevention to reduce the probability of initial disasters escalating; the social space conducts multi-department and cross-border collaboration, efficiently integrates resources, and reduces the damage caused by disasters; the information space attaches importance to information collection and release, dynamically updates early warnings, and uses big data to predict the evolution path of urban cascading disasters.

After the occurrence of urban cascading disasters, the collaborative mechanism across three Spaces is started. Through the monitoring equipment in the physical space, the community network in the social space and the big data platform in the information space, the evolution path of the cascading disaster should be fully found out, the disaster dynamics should be accurately mastered, and the real-time situation of the disaster should be released in time through multiple platforms. At the same time, comprehensive consideration should be given to the impacts of disasters on urban transportation, energy supply, medical system, and economic activities to achieve effective allocation of emergency resources and efficient implementation of emergency measures.

Therefore, under the coordinated operation of the “physical—social—information” three Spaces, not only can the losses caused by disasters be minimized to the greatest extent, but also the safety and stability of the city can be guaranteed, promoting the city to develop in a more resilient and sustainable direction. The urban cascading disaster resilience model constructed based on spatial theory is shown in Figure 1.

This model demonstrates a high degree of comprehensiveness in its discussion of urban components. Compared with other existing models internationally (such as Resilience-based Earthquake Design Initiative REDi), it places greater emphasis on the role of information space. This focus on information space not only enriches the understanding of urban systems but also provides a more precise analytical perspective for cities when facing cascading disaster scenarios. Moreover, the model is closely integrated with resilience theory and can conduct targeted evaluations of urban cascading disaster resilience based on the specific characteristics and needs of different regions. This makes the model highly adaptable to regions with different geographical, social, and economic backgrounds. Additionally, the design of the model also takes into account the operability in practical applications. Its evaluation methods and indicator system have a certain degree of replicability, which can provide reference and inspiration for other cities when conducting similar evaluations, thereby promoting the scientific and standardized development of urban resilience construction.

3. Research Methods

3.1. Evaluation Process

The evaluation of urban cascading disaster resilience is a complex and rigorous task. During the evaluation process, the characteristics of spatial theory and resilience theory should be considered to construct a comprehensive urban cascading disaster resilience evaluation index system. To ensure the accuracy of the results, the study uses a combination weighting method based on game theory (entropy weight method and CRITIC method) to determine the index weights, and then applies the TOPSIS evaluation model to calculate the cascading disaster resilience values of each city. The calculation results are used for spatio-temporal evolution and spatial correlation analysis. The study also introduces the Geodetector to identify driving factors. Based on the analysis results, corresponding measures will be proposed and management strengthened to enhance resilience. The specific evaluation process is shown in Figure 2.

3.2. Construction of the Index System

Based on relevant literature [18,19,20,21,22,23] and drawing on existing research results [24], the urban cascading disaster resilience evaluation index system is established under the guidance of spatial theory and resilience theory, as shown in

Table 1. Evaluation index system of urban cascading disaster resilience.

Target Layer	Criterion Layer	Index Layer	Index Meaning	Attribute	Resilience Characteristics
Urban cascading disaster resilience C	Physical space C1	C11 Per capita road area (m2)	Urban traffic capacity	+	Resistance, Adaptability
		C12 Green coverage rate (%)	Urban ecological greening level	+
		C13 Urban water supply pipe density	Water resource supply capacity	+
		C14 Urban drainage pipe density	Urban seismic resistance capacity	+
		C15 Proportion of earthquake-resistant building structures (%)	Urban earthquake resistance capacity level	+
		C16 Per capita emergency shelter area (m2)	Urban disaster-bearing capacity	+
	Social space C2	C21 Per capita GDP	Social per capita economic volume	+	Recovery, Adaptability
		C22 Population density	Social population pressure	-
		C23 Number of hospital beds in medical and health institutions	Social medical security capacity	+
		C24 Proportion of higher education personnel (%)	Social education foundation level	+
		C25 Proportion of science and technology expenditure (%)	Innovation foundation environment	+
		C26 Research and development (R&D) expenditure	Scientific research foundation	+
		C27 Proportion of public management and social organization personnel	Disaster emergency organization capacity	+
		C28 Disaster publicity and drill level	Disaster escape drill level	+
		C29 Proportion of elderly population	Social population composition	-
	Information space C3	C31 Internet penetration rate (%)	Network infrastructure level	+	Adaptability
		C32 Mobile phone penetration rate	Basic communication penetration level	+
		C33 Proportion of software and related information service industry income (%)	Information service construction level	+
		C34 Radio and television coverage rate	Disaster information publicity level	+
		C35 Number of earthquake stations and networks per 10,000 people	Earthquake disaster early warning capacity	+
		C36 Number of meteorological stations per 10,000 people	Meteorological early warning capacity	+

.

Physical space encompasses a city’s transportation capacity, ecological greening level, and disaster resistance capacity, etc., mainly reflecting its resistance and adaptability. The indicators of the Per capita road area and the Urban water supply pipe density mainly reflect the distribution of infrastructure and measure the ability of cities to ensure smooth traffic and resource supply after disasters. The Urban drainage pipe density and the Proportion of earthquake-resistant building structures are designed to measure a city’s capacity to withstand disasters such as floods and earthquakes that are prone to trigger cascading effects. The Per capita emergency shelter area mainly measures the distribution of shelters and their capacity for rapid evacuation. The Green coverage rate reflects the investment intensity of urban ecological protection and proactive disaster prevention.

Social space encompasses economic, cultural, medical, educational and disaster organization capabilities, mainly reflecting recovery and adaptability. The Population density and the Proportion of the elderly population reflect the level of human resilience in the process of responding to urban cascading disasters. The Proportion of public management and social organization personnel and the Proportion of higher education personnel represent the advantages of technical and knowledge reserves, and demonstrate corresponding adaptability in the process of responding to and recovering from urban cascading disasters. The Per capita GDP, the Proportion of science and technology expenditure, and the Research and development (R&D) expenditure are important guarantees for recovery capacity. The higher the economic resilience, the better the environment for scientific and technological development, and the more abundant the resources for recovery and reconstruction. The Number of hospital beds in medical and health institutions measures the capacity to accommodate the injured and sick during disasters; the Disaster publicity and drill level reflects the speed of the public’s first response and the efficiency of collaboration in urban cascading disasters.

Information space includes communication networks, disaster information transmission services, and various disaster warning information platforms, mainly reflecting adaptability. The Proportion of software and related information service industry income reflects the level of information service construction. The Internet penetration rate meets the demand for various information transmission after disasters, ensuring the timeliness of information transmission; the number of earthquake stations and networks and number of meteorological stations, as disaster monitoring indicators, ensure the understanding of the evolution path of cascading disasters during their occurrence in cities; the Radio and television coverage rate and the Mobile phone penetration rate improve the overall response speed through rapid and accurate data feedback.

3.3. Combined Weighting Method

The study first uses the entropy weight method and CRITIC method to determine the weights of each index. The calculation process of the entropy weight method includes data standardization, calculation of entropy values and difference coefficients, and determination of weights. The main calculation steps of the CRITIC method are standardization processing, calculation of index variability, index conflict, index information volume, and weight calculation. The two methods form effective complementarity in the logic of weight determination: The entropy weight method captures the implicit information entropy value of the data from the perspective of information theory, focusing on the “intrinsic information content” of the data; The principle of CRITIC considers both the variation degree and the conflict characteristics of indicators from the perspective of statistics, and emphasizes the “independence and identification” of indicators. This dual consideration of “information measurement” and “statistical characteristics” can more comprehensively explore data features and avoid the one-sidedness of single-dimensional weight assignment. Finally, the combined weights are determined based on game theory [25]. Based on the idea of game theory, this method seeks the optimal balance point of two weighting results. It does not rely on subjective empirical judgment but minimizes the sum of the squares of the deviations between the combined weights and the weights of a single method to ensure that the combined results not only respect the independent judgment of each method but also achieve overall optimality mathematically, significantly enhancing the objectivity and robustness of weight determination.

3.3.1. Entropy Weight Method

(1)

Establish an initialization index system matrix. Construct the original matrix with $m$ indicators and $n$ objects:

$$M={\left[x_{ij}\right]}_{n\times m}$$

(1)

(2)

Standardize the original data

For positive indicators, the standardized processing formula is as follows:

$$x_{ij}^{\prime }={\displaystyle \frac{x_{ij}-min\left(x_{ij}\right)}{max\left(x_{ij}\right)-min\left(x_{ij}\right)}}$$

(2)

For negative indicators, the standardized processing formula is as follows:

$$x_{ij}^{\prime }={\displaystyle \frac{max\left(x_{ij}\right)-x_{ij}}{max\left(x_{ij}\right)-min\left(x_{ij}\right)}}$$

(3)

In the formula: $x_{ij}$ represents the original value of the $j$ indicator of the $i$ unit, and $x_{ij}^{\prime }$ represents the $j$ indicator of the $i$ unit after standardization processing.

(3)

Determine the information entropy value $e_j$ and the coefficient of difference $d_j$

$$e_j=-k\left[\sum _{i=1}^n{P}_{ij}ln\left(P_{ij}\right)\right]$$

(4)

$$d_j=1-e_j$$

(5)

In the formula: $P_{ij}={\displaystyle \frac{x_{ij}^{\prime }}{{\sum }_{i=1}^n{x}_{ij}^{\prime }}} , K=1/ln\left(n\right)$. It is the information entropy value of the $j$-th indicator. The larger the entropy value $e_j$ of a certain indicator, the smaller its role in the comprehensive evaluation and the smaller its weight. Conversely, the greater the weight. $P_{ij}$ represents the proportion of the $i$-th unit under the $j$-th indicator to this indicator.

(4)

Calculate the weight $w_1$:

$$w_1=\left(1-e_j\right)/\sum _{j=1}^m\left(1-e_j\right)$$

(6)

3.3.2. CRITIC Method

(1)

Calculate the variability $S_j$ of the $j$-th indicator:

$$S_j=\sqrt{{\displaystyle \frac{{\sum }_{i=1}^n{\left(x_{ij}^{\prime }-\overline{x_j^{\prime }}\right)}^2}{n-1}}}$$

(7)

(2)

Calculate the conflict $R_j$ of the $j$-th indicator:

$$R_j=\sum _{k=1}^m\left(1-x_{jk}\right)$$

(8)

(3)

Calculate the information volume $C_j$ of the $j$-th indicator:

$${C_j=S}_j\times R_j$$

(9)

(4)

Calculate the weight $w_2$:

$$w_2=C_j/\sum _{j=1}^m{C}_j$$

(10)

3.3.3. Determine the Combined Weighting

(1)

Linear combination of vectors. Suppose $L$ methods are adopted to determine the weights of $n$ indicators.

$$w=\sum _{l=1}^L{\mathsf{\alpha }}_l{w}_l^T ({\mathsf{\alpha }}_l>0)$$

(11)

In the formula, $w$ represents the combined weight vector, ${\mathsf{\alpha }}_l$ is the linear combination coefficient, and $w_l$ is the set of basic weight vectors constructed.

(2)

Based on the combination principle of game theory, the deviation between $w$ and ${\mathsf{\alpha }}_l$ is minimized, and the objective function is:

$$\min {‖\sum _{l=1}^L{\mathsf{\alpha }}_l{w}_l^T-w_p^T‖}_2, p=1,2,\dots ,L$$

(12)

(3)

According to the properties of matrix differentiation, the conditions for optimizing the first derivative are:

$$\left[\begin{array}{cccc}{w}_1{w}_1^T& w_1{w}_2^T& \dots & w_1{w}_L^T\\ w_2{w}_1^T& w_2{w}_2^T& \dots & w_2{w}_L^T\\ ⋮& ⋮& ⋮& ⋮\\ w_L{w}_1^T& w_L{w}_2^T& \dots & w_L{w}_L^T\end{array}\right]\left[\begin{array}{c}{\mathsf{\alpha }}_1\\ {\mathsf{\alpha }}_2\\ ⋮\\ {\mathsf{\alpha }}_L\end{array}\right]=\left[\begin{array}{c}{w}_1{w}_1^T\\ w_2{w}_2^T\\ ⋮\\ w_L{w}_L^T\end{array}\right]$$

(13)

(4)

Calculate the combined weight.

$$w^{*}=\sum _{l=1}^L{\mathsf{\alpha }}_l^{*}{w}_l^T,l=1,2,\dots ,L$$

(14)

Using Formulas (1)–(14) for calculation, the results are shown in

Table 2. Index weight of urban cascading disaster resilience evaluation.

Target Layer	Criterion Layer	Index Layer	Entropy Weight	CRITIC Weight	Combined Weight
C	C1	C11	0.033	0.054	0.037
		C12	0.013	0.045	0.019
		C13	0.044	0.056	0.046
		C14	0.037	0.048	0.039
		C15	0.035	0.044	0.037
		C16	0.047	0.032	0.044
	C2	C21	0.042	0.042	0.042
		C22	0.014	0.058	0.022
		C23	0.059	0.049	0.057
		C24	0.106	0.048	0.095
		C25	0.054	0.032	0.050
		C26	0.161	0.045	0.139
		C27	0.068	0.034	0.061
		C28	0.053	0.047	0.052
		C29	0.014	0.072	0.024
	C3	C31	0.049	0.045	0.048
		C32	0.046	0.036	0.044
		C33	0.033	0.060	0.038
		C34	0.006	0.042	0.012
		C35	0.044	0.062	0.048
		C36	0.044	0.048	0.044

.

3.4. TOPSIS Evaluation Model

The TOPSIS evaluation model assesses the superiority or inferiority of a scheme based on the distance from the ideal solution and the negative ideal solution. This method can more scientifically and comprehensively consider the relative superiority or inferiority of multiple indicators [26]. The study applies the TOPSIS model to the comprehensive evaluation of urban cascading disaster resilience, which can more accurately reflect the relative superiority or inferiority of urban cascading disaster resilience in various regions and provide a basis for subsequent analysis.

The calculation formula is as follows:

(1)

Determine the weighted normalized matrix $Z$

$$Z=M\times W_j$$

(15)

(2)

Determine the positive ideal solution $Z_j^{+}$ and the negative ideal solution ${ Z}_j^{-}$

$$Z_j^{+}=Max\left\{Z_{i1},Z_{i2},\dots ,Z_{im}\right\}$$

(16)

$$Z_j^{-}=Min\left\{Z_{i1},Z_{i2},\dots ,Z_{im}\right\}$$

(17)

(3)

Calculate the Euclidean distance ${ D}_j^{+}$ and ${ D}_j^{-}$

$${ D}_i^{+}=\sqrt{\sum _{j=1}^m{\left(Z_j^{+}-z_{ij}\right)}^2}$$

(18)

$${ D}_i^{-}=\sqrt{\sum _{j=1}^m{\left(Z_j^{-}-z_{ij}\right)}^2}$$

(19)

(4)

Calculate the relative progress of the posts $C_i$

$$C_i={\displaystyle \frac{D_i^{-}}{D_i^{+}+D_i^{-}}} , 0<C_i<1$$

(20)

In the formula, $i=1,2,3\dots n$, the larger the $C_i$ value, the better the cascading disaster resilience level of the city; conversely, the lower it is.

3.5. Kernel Density Estimation Model

Kernel density estimation is a commonly used method for studying unbalanced distributions. It mainly regards the distribution pattern of the object under investigation as a certain probability distribution and examines its characteristics and trends over time [27]. By using the Kernel density estimation method and through the smoothing method, the distribution pattern of random variables can be described with a continuous density curve, which can effectively demonstrate the dynamic evolution of the distribution of urban cascading disaster resilience in the study area.

3.6. Spatial Correlation Analysis

Spatial correlation analysis can detect and study the spatial agglomeration of urban cascading disaster resilience in the region, and calculate the Global Moran’s I and the Local Moran’s I, respectively [28]. It can effectively determine whether there is a spatial effect on the overall urban cascading disaster resilience of the research area and identify the spatial agglomeration characteristics among cities. The result not only avoids the inaccurate results caused by the neglect of spatial interaction in traditional models, but also provides direct references for the accurate allocation of disaster prevention resources and cross-regional collaboration mechanisms. Moreover, it can determine whether relevant measures truly weaken spatial polarization through temporal evolution, thereby achieving precise governance.

3.7. Geodetector

The factor detection module in the Geodetector is utilized to study the degree of explanation and changes of each influencing factor in the evolution process of cascading disaster resilience in cities of Jiangsu Province [29]. The factor detection module of the geographic detector measures the explanatory power of each driving factor on the evolution of urban cascading disaster resilience by q value. Through the ranking and variation range of q value, the importance changes of the factors can be determined, providing dynamic information for policymakers and avoiding the underestimation or overestimation of the true contribution of key variables due to the mis-setting of the function form in traditional regression.

3.8. Overview of the Study Area

Jiangsu Province is located in the eastern coastal area of China and is one of the most economically developed provinces in China. It has 13 prefecture-level cities, with Nanjing as its provincial capital. Jiangsu Province is usually divided into three regions: Southern Jiangsu, Central Jiangsu, and Northern Jiangsu. Southern Jiangsu includes Nanjing, Wuxi, Changzhou, Suzhou, and Zhenjiang. Central Jiangsu includes Nantong, Yangzhou, and Taizhou. Northern Jiangsu includes Xuzhou, Lianyungang, Suqian, Huai’an, and Yancheng. The terrain in Southern Jiangsu is low and flat, and areas such as the Taihu Lake Basin are prone to waterlogging. The Southern Jiangsu Plain also has the problem of ground fissures, a new type of geological disaster that has emerged in recent years, mainly occurring in the economically highly developed Suzhou–Wuxi–Changzhou region, causing significant harm and losses. Central Jiangsu is severely threatened by typhoon storm surges. Typhoon paths often land in Central Jiangsu and Southern Shandong and move northeastward. The areas along the north bank of the Yangtze River, such as Haimen and Tongzhou, are frequently affected by tides. Northern Jiangsu is a relatively high-incidence area for tornadoes, especially in the middle of the Northern Jiangsu Plain, the Li Xiahe region, which is low-lying and has a dense network of rivers, making it prone to strong convective weather and extreme meteorological disasters such as tornadoes.

In the development process of Jiangsu Province, disaster prevention is of extremely important significance. It is not only a key measure to ensure the safety of people’s lives and property but also a fundamental guarantee for maintaining economic stability, protecting the ecological environment, and enhancing urban resilience.

The data for this study are sourced from the statistical yearbooks of each city in Jiangsu Province from 2014 to 2023, the “China Statistical Yearbook”, the “China City Statistical Yearbook”, the “Statistical Communiqué of the People’s Republic of China on National Economic and Social Development”, and the data released by administrative institutions such as the statistics bureaus, emergency management bureaus, and ecological environment bureaus of each city. Missing data are completed using interpolation methods.

4. Results and Analysis

4.1. Spatial Distribution of Urban Cascading Disaster Resilience in Jiangsu Province

To explore the spatial distribution changes of urban cascading disaster resilience in various cities of Jiangsu Province at different times, the resilience values were calculated using Formulas (15)–(20) based on the TOPSIS model. The results are shown in

Table 3. Resilience values of urban cascading disasters in each region of Jiangsu Province from 2013 to 2024.

	2014	2015	2016	2017	2018	2019	2020	2021	2022	2023
Nanjing	0.696	0.711	0.698	0.666	0.685	0.671	0.676	0.668	0.646	0.684
Wuxi	0.552	0.550	0.591	0.590	0.585	0.565	0.554	0.583	0.598	0.627
Xuzhou	0.279	0.277	0.281	0.262	0.272	0.272	0.267	0.281	0.298	0.298
Changzhou	0.343	0.340	0.338	0.313	0.323	0.305	0.314	0.371	0.381	0.387
Suzhou	0.681	0.684	0.663	0.635	0.666	0.647	0.649	0.648	0.647	0.651
Nantong	0.268	0.255	0.247	0.265	0.284	0.281	0.275	0.294	0.304	0.311
Lianyungang	0.172	0.174	0.213	0.182	0.178	0.177	0.183	0.218	0.233	0.225
Huai’an	0.180	0.189	0.195	0.218	0.223	0.197	0.199	0.202	0.212	0.223
Yancheng	0.283	0.291	0.301	0.266	0.279	0.284	0.284	0.288	0.292	0.291
Yangzhou	0.282	0.279	0.264	0.267	0.285	0.257	0.264	0.274	0.275	0.268
Zhenjiang	0.293	0.304	0.309	0.283	0.293	0.311	0.328	0.338	0.340	0.350
Taizhou	0.276	0.270	0.258	0.256	0.265	0.251	0.268	0.275	0.288	0.290
Suqian	0.229	0.227	0.224	0.214	0.218	0.218	0.217	0.246	0.251	0.260

.

Some time points during the period from 2014 to 2023 (2014, 2016, 2019, 2021 and 2023) were selected. The results of urban cascading disaster resilience were classified into five levels using the natural breakpoint method. The spatial distribution of the results of urban cascading disaster resilience was visualized through ArcGIS (10.8), as shown in Figure 3.

From the perspective of spatial distribution characteristics, the urban cascading disaster resilience of cities in Jiangsu Province shows significant regional heterogeneity. The resilience of cities in the southern part of Jiangsu is generally higher than that of cities in the northern part. From the perspective of regional location, cities such as Nanjing, Wuxi, Changzhou, and Suzhou in the southern part of Jiangsu have maintained a relatively high level of cascading disaster resilience over the past decade and are among the top in the province. According to the index data, these cities are more economically developed and have better infrastructure than cities in the northern and central parts of Jiangsu, and thus have stronger disaster response capabilities. However, cities such as Lianyungang, Huai’an, and Suqian in the northern part of Jiangsu are relatively lagging behind in terms of urban cascading disaster resilience and still have room for improvement in education and medical care. The city of Nantong in the central part of Jiangsu has seen rapid development in urban cascading disaster resilience over the past decade, thanks to investment in science and technology and healthcare. From a temporal perspective, the resilience of cities such as Lianyungang, Huai’an, and Nantong has increased rapidly over the past decade, gradually narrowing the gap with economically developed cities like Nanjing and Suzhou. However, the resilience of Nanjing and Suzhou, two economically strong cities, has declined in the past two years. Nanjing, as the provincial capital of Jiangsu, has a strong economy and superior infrastructure conditions, but due to the continuous increase in population density, the urban disaster risk has been rising year by year. Suzhou, as an important economic city in Jiangsu and even the whole country, has a well-developed water system, and the risk of flood disasters has been increasing in recent years. Overall, the urban cascading disaster resilience of cities in Jiangsu Province has generally shown an upward trend over the past decade, but regional differences remain significant. Although the southern part of Jiangsu has achieved certain results in urban resilience construction, it still faces new risks and challenges; although the resilience of the northern part of Jiangsu has improved, there is still a gap compared with the southern part, and it needs to further strengthen urban resilience construction to cope with various natural disasters and socio-economic risks that may be faced in the future.

4.2. The Temporal Evolution Trend of Urban Cascading Disaster Resilience in Jiangsu Province

To analyze the temporal evolution characteristics of urban cascading disasters in Jiangsu Province, the kernel density of the resilience measurement results of urban cascading disasters in Jiangsu Province was calculated, and the kernel density estimation curves of resilience changes in different years were obtained, as shown in Figure 4.

By observing Figure 4, it can be found that the urban cascading disaster resilience kernel density distribution curve of Jiangsu Province as a whole has shown a trend of rightward shift from 2014 to 2023. This change indicates that with the passage of time, the overall urban cascading disaster resilience of Jiangsu Province has been continuously improving, showing a good development trend. From the perspective of waveform evolution, as time goes by, the height of the main peak gradually decreases, while the width gradually increases, indicating that low-resilience cities are catching up with high-resilience cities, and the development gap between regions is gradually narrowing. In addition, the overall distribution shows a multi-peak trend, but the side peaks are relatively low, indicating a slight multi-polarization phenomenon, reflecting the existence of certain differences in the development of urban cascading disaster resilience among different regions within Jiangsu Province. From the perspective of distribution extension, the curve shows a significant rightward tailing phenomenon, indicating that the urban cascading disaster resilience of some regions within Jiangsu Province is still significantly lower than that of other regions. This difference suggests that more attention should be paid to the radiation and driving effect of high-resilience regions on low-resilience regions, and through regional coordination and resource sharing, the balanced development of urban cascading disaster resilience throughout the province should be promoted to further enhance the overall disaster response capacity of Jiangsu Province.

4.3. Spatial Correlation of Urban Cascading Disaster Resilience in Jiangsu Province

4.3.1. Overall Spatial Correlation Analysis of Jiangsu Province

Based on the previous analysis, there are significant regional differentiation characteristics among different cities in Jiangsu Province, but the spatial correlation between regions still needs to be further examined. The study selected the global Moran index to test whether there is a spatial correlation among the cities in Jiangsu Province, as shown in

Table 4. The Global Moran Index of urban cascading disaster resilience in Jiangsu Province from 2014 to 2023.

Year	2014	2015	2016	2017	2018	2019	2020	2021	2022	2023
Global Moran Index	0.163	0.155	0.176	0.158	0.158	0.168	0.176	0.262	0.287	0.273
p value	0.184	0.199	0.166	0.198	0.197	0.180	0.165	0.067	0.050	0.059

.

By observing Table 4, it can be seen that the global Moran index is greater than zero and ranges from 0.15 to 0.29 during the period from 2014 to 2023, indicating that there is positive spatial autocorrelation in the urban cascading disaster resilience of cities in Jiangsu Province. From 2014 to 2020, the Moran index remained between 0.15 and 0.18, with a weak spatial aggregation effect and insignificant spatial autocorrelation, indicating that during this period, the development of urban cascading disaster resilience was relatively scattered and no clear high-resilience or low-resilience aggregation areas had formed. From 2021 to 2023, the Moran index significantly increased to between 0.26 and 0.29, and the value in 2022 was 0.05, which was verified, indicating that the spatial aggregation effect of urban cascading disaster resilience among cities in Jiangsu Province has strengthened. This trend reflects that high-resilience cities have gradually formed distinct high-value clusters, while low-resilience cities have also shown a low-value clustering effect. However, in 2023, the Moran index dropped to 0.273, indicating a weakened spatial autocorrelation. This suggests that as the impact of the pandemic eased, the policies and recovery measures across regions gradually balanced out, and the differences in cascading disaster resilience among cities gradually narrowed.

Although, from an overall perspective, there is no significant correlation between the development level of a city’s cascading disaster resilience and that of its neighboring regions, this does not mean that the development level of cascading disaster resilience of all cities is unrelated to that of their neighboring cities. The overall analysis may mask the particularities of local regions. Therefore, it is not comprehensive to rely solely on the results of the overall level analysis. It is also necessary to measure the local Moran index to detect the correlation between some regions and determine whether there is spatial correlation among some cities.

4.3.2. Local Spatial Correlation Analysis in Jiangsu Province

To further examine the correlation among cities in some regions of Jiangsu Province, the local Moran Index was calculated using the disaster resilience values of cities in Jiangsu Province from 2014 to 2023. The distribution status is shown in

Table 5. The distribution status of the local Moran Index in Jiangsu Province from 2014 to 2023.

Year	Aggregation Type	City
2014	Low-high anomaly (L-H)	Changzhou, Zhenjiang
	Low-low aggregation (L-L)	Suqian, Huai’an, Lianyungang
2021	High-high aggregation (H-H)	Changzhou, Zhenjiang, Wuxi, Nanjing
	Low-low aggregation (L-L)	Suqian, Huai’an, Lianyungang

.

The clustering results of urban cascading disaster resilience in Jiangsu Province from 2014 to 2023 were relatively stable, but there were significant differences among different categories. From 2014 to 2020, the “low-low aggregation (L-L)” type was mainly concentrated in cities such as Suqian, Huai’an, and Lianyungang in northern Jiangsu. According to the index data, the scientific research foundation level and innovation environment of these cities lagged behind those of cities in southern Jiangsu. Moreover, these cities had relatively few economic connections and cooperation with surrounding cities and lacked an effective regional coordinated development mechanism. The “low-high aggregation (L-H)” type was mainly concentrated in southern Jiangsu cities such as Changzhou and Zhenjiang, showing a significant reverse difference in development compared to the surrounding high-value areas. In 2021, the distribution pattern changed. The “high-high aggregation (H-H)” areas included Changzhou, Zhenjiang, Wuxi, and Nanjing. Changzhou is located in southern Jiangsu and borders economically developed cities such as Nanjing and Wuxi. In recent years, Changzhou’s economic strength has continuously improved, and its economic ties with surrounding cities have become increasingly close, forming a strong effect of experience sharing and risk response capabilities, and thus showing a clear “spatial club” convergence phenomenon, demonstrating a stable high-resilience development trend. The cities in the “low-low aggregation (L-L)” type remained unchanged. The data indicates that cities like Suqian still have considerable room for improvement in terms of per capita GDP, education level, and scientific research investment to enhance their cascading disaster resilience.

4.4. Detection of Spatio-Temporal Differentiation Driving Factors for Urban Cascading Disaster Resilience in Jiangsu Province

The enhancement and development of urban cascading disaster resilience is the result of the combined effects of three Spaces within the city. The geographic detector model was used to explore the explanatory power of various index factors on the spatial differentiation characteristics of urban cascading disaster resilience in Jiangsu Province. Only the top 10 main detection factors were listed, as shown in

Table 6. Spatio-temporal differentiation detection results of urban cascading disaster resilience in Jiangsu Province.

Year	1	2	3	4	5	6	7	8	9	10
2014	C32 (0.93)	C26 (0.93)	C28 (0.92)	C15 (0.85)	C16 (0.78)	C25 (0.76)	C31 (0.75)	C21 (0.69)	C24 (0.63)	C27 (0.58)
2016	C21 (0.96)	C26 (0.96)	C28 (0.95)	C32 (0.88)	C16 (0.81)	C31 (0.77)	C25 (0.68)	C24 (0.62)	C29 (0.59)	C27 (0.57)
2019	C31 (0.96)	C25 (0.95)	C26 (0.94)	C33 (0.87)	C16 (0.86)	C23 (0.80)	C21 (0.77)	C32 (0.76)	C27 (0.73)	C28 (0.67)
2021	C26 (0.94)	C32 (0.90)	C16 (0.88)	C33 (0.81)	C21 (0.80)	C24 (0.79)	C27 (0.79)	C23 (0.77)	C25 (0.69)	C31 (0.68)
2023	C33 (0.96)	C26 (0.95)	C32 (0.89)	C16 (0.87)	C25 (0.82)	C21 (0.80)	C28 (0.76)	C24 (0.76)	C23 (0.73)	C31 (0.63)

.

According to the calculation results, the significant q values at the five time points of 2014, 2016, 2019, 2021, and 2023 are: in the physical space, C16 (Urban disaster-bearing capacity); in the social space, C21 (Social per capita economic volume), C25 (Innovation foundation environment), and C26 (Scientific research foundation); in the information space, C31 (Network infrastructure level) and C32 (Basic communication penetration level). Each space has high-level driving factors, indicating that the improvement of urban cascading disaster resilience cannot rely solely on the efforts of one space but requires the collaboration of all spaces to face risks and challenges.

(1)

Driving factors in the physical space. In the physical space dimension, the q mean value of Urban disaster-bearing capacity has been significantly higher than other indicators in recent years, playing a significant positive driving role in the spatio-temporal differentiation of urban cascading disaster resilience in Jiangsu Province. Jiangsu Province has continuously promoted the construction of emergency shelters, with the number steadily increasing. During the occurrence of cascading disasters, evacuating the affected population and providing shelters are of paramount importance. Moreover, disasters often first impact infrastructure, and the construction of infrastructure must be strictly controlled, with emphasis on its multi-functional design. At the design stage, the potential impacts of various disasters should be considered to ensure that infrastructure can effectively respond to multi-disaster scenarios and avoid secondary injuries caused by facility damage.

(2)

Driving factors in the social space. Social space is the core among the three spaces for enhancing the resilience of urban cascading disasters in Jiangsu Province. The key factors driving the spatio-temporal differentiation of urban cascading disaster resilience in Jiangsu Province are the Scientific research foundation, Social per capita economic volume, and the Innovation foundation environment, with their q means being 0.95, 0.81, and 0.79, respectively. However, the population pressure and population composition need to be optimized, as their factor values have remained low in recent years. Jiangsu Province, as a leading province in China in terms of economic volume and technological innovation capabilities, has abundant financial resources and high-end technological elements, which provide solid support for the research on the mechanism of cascading disasters and their resilience governance, thus giving it more confidence in dealing with cascading disasters. However, in recent years, the population has rapidly concentrated in the central cities, the degree of aging has continued to deepen, and the proportion of vulnerable groups has increased accordingly, bringing new risk variables to emergency evacuation, medical rescue, and post-disaster recovery. It is necessary to optimize the population policy, through spatial diversion and age structure adjustment, to enhance the overall risk resistance capacity.

(3)

Driving factors in the information space. With the advent of the Internet era and the rapid advancement of the informatization process, not only has the speed of obtaining disaster information significantly increased, but also its accuracy has improved. Among them, the q means of Basic communication penetration level and Network infrastructure level construction are 0.88 and 0.76, respectively, indicating that Jiangsu Province has always placed the construction of information infrastructure at the top of the regional public safety system. Moreover, the explanatory power of the factor of Information service construction level has increased from 0.31 in 2014 to 0.96 in 2023. This shows that in recent years, Jiangsu Province has paid more attention to the collection and release of disaster information, and utilized new technologies to empower information development. When cascading disasters occur, it can make full use of relevant information technologies to accurately judge the situation and evolution trend of the disaster, and make scientific judgments based on this, thereby efficiently carrying out disaster relief tasks.

4.5. Suggestions for Improvement

Based on the analysis of the above evaluation results, the following suggestions for enhancing resilience are proposed from the perspective of the three spaces:

(1)

Physical space level. According to the data, the infrastructure shortcomings in the high-density areas of southern Jiangsu and the thin foundation in northern Jiangsu need to be addressed. In the flood-prone areas such as Suzhou, Wuxi, Changzhou, Zhenjiang, and Yangzhou, underground storage deep tunnels and other infrastructure should be constructed; at the same time, ecological buffer zones and high-position wetlands should be built to reduce the initial impact of cascading disasters [30]. In the axis of Suqian-Kaohsiung-Lianyungang in northern Jiangsu, the existing corridors of Xuzhou-Lianyungang high-speed railway and Lianyungang-Horgos Expressway should be utilized to reserve emergency channels and underground comprehensive pipe galleries, which can serve as logistics channels in normal times and rescue lifelines in disaster times.

(2)

Social space level. With the trilateral synergy of “government-market-society” and the dual drive of “southern Jiangsu experience-northern Jiangsu assistance”, the rapid response capacity at the grassroots level should be enhanced [31]. The governments of the three major metropolitan areas of Nanjing, Xuzhou, and Suzhou should unify the disaster chain scenario plans, material reserve directories, and rescue team numbers to achieve cross-city resource dispatch within 2 h. The five cities in northern Jiangsu (Xuzhou, Lianyungang, Suqian, Huai’an, and Yancheng) should establish “land port-sea port-airport” emergency logistics centers, and reserve cold chain, tents, generators and other materials in advance.

(3)

Information space level. Continuously upgrade the Jiangsu Urban Information Model (CIM) platform, integrating three-dimensional terrain, underground pipelines, building BIM, and real-time water conditions, meteorological, and tide level data. Establish smart cloud communities, issue early warning information, and coordinate relevant management departments, communities, and residents to take emergency measures based on the early warning information, achieving intelligent decision-making management [32].

(4)

Through the three-dimensional spatial synergy of “physical–social–information”, the resilience of cascading disasters in Jiangsu Province can achieve a leap from the traditional model of each fighting on its own to a modern model that is systematically absorbable, recoverable and adaptable. Other regions can draw on the corresponding paradigms, collect data of the corresponding regions based on the listed indicators, apply this evaluation method for evaluation, and propose corresponding improvement measures according to the results in different Spaces. Among them, strengthen infrastructure construction in the physical space; Always keep emergency supplies on hand in social Spaces, popularize disaster knowledge and respond promptly during disasters; Ensure accurate and rapid information exchange during disasters in the information space, accurately grasp the evolution path of urban cascading disasters, and enable the resilience of urban cascading disasters to grow in tandem with the process of urbanization.

5. Conclusions

This paper, based on the data of Jiangsu Province from 2014 to 2023, constructs an evaluation index system for urban cascading disaster resilience based on spatial theory, and determines the combined weights based on the entropy weight–CRITIC method and uses the TOPSIS method to calculate the resilience values. Through kernel density analysis, spatial correlation analysis, and geographic detector, it explores the spatial distribution, spatio-temporal evolution characteristics, spatial correlation, and influencing mechanisms of urban cascading disaster resilience in Jiangsu Province, and reaches the following conclusions:

(1)

In terms of spatial distribution, the cities in Jiangsu Province exhibit significant regional heterogeneity. The development of urban cascading disaster resilience shows that the resilience of cities in southern Jiangsu is generally higher than that of cities in northern Jiangsu. Although cities in northern Jiangsu, such as Suqian, have seen considerable improvement, regional differences remain quite obvious.

(2)

In terms of temporal evolution, the urban cascading disaster resilience in Jiangsu Province has generally increased from 2014 to 2023, but there is a slight multi-polarization phenomenon. Moreover, within the region, there are certain areas where the cascading disaster resilience is significantly higher than in other areas.

(3)

In terms of spatial correlation, there is a significant positive spatial correlation between the urban cascading disaster resilience and geographical location among different regions in Jiangsu Province. The degree of spatial agglomeration has been continuously increasing over time, with “high-high agglomeration” mainly existing in southern Jiangsu and “low-low agglomeration” mainly concentrated in northern Jiangsu.

(4)

In the detection of driving factors, each spatial area has a relatively high level of driving factors. Spatial differentiation is mainly caused by differences in Scientific research foundations, the Basic communication penetration level, the Urban disaster-bearing capacity, the Innovation foundation environment, the Social per capita economic volume, and the Network infrastructure level.

(5)

Although this study has comprehensively explored urban cascading disaster resilience using multi-dimensional indicators, it still mainly adopts a static framework. Future research could delve into the complex and dynamic nature of urban systems, focus on the coupling mechanisms of urban systems, predict the evolution trend of urban cascading disaster resilience, and achieve the enhancement of urban cascading disaster resilience.