Research on Resilience Evaluation and Prediction of Urban Ecosystems in Plateau and Mountainous Area: Case Study of Kunming City

Abstract

In the face of increasingly complex urban challenges, a critical question arises: can urban ecosystems maintain resilience, vitality, and sustainability when confronted with external threats and pressures? Taking Kunming—a plateau-mountainous city in China—as a case study, this research constructs an urban ecosystem resilience (UER) assessment model based on the DPSIR (Driving forces, Pressures, States, Impacts, and Responses) framework. A total of 25 indicators were selected via questionnaire surveys, covering five dimensions: driving forces such as natural population growth, annual GDP growth, urbanization level, urban population density, and resident consumption price growth; pressures including per capita farmland, per capita urban construction land, land reclamation and cultivation rate, proportion of natural disaster-stricken areas, and unit GDP energy consumption; states measured by Evenness Index (EI), Shannon Diversity Index (SHDI), Aggregation Index (AI), Interspersion and Juxtaposition Index (IJI), Landscape Shape Index (LSI), and Normalized Vegetation Index (NDVI); impacts involving per capita GDP, economic density, per capita disposable income growth, per capita green space area, and per capita water resources; and responses including proportion of natural reserve areas, proportion of environmental protection investment to GDP, overall utilization of industrial solid waste, and afforestation area. Based on remote sensing and other data, indicator values were calculated for 2006, 2011, and 2016. The entire-array polygon indicator method was used to visualize indicator interactions and derive composite resilience index values, all of which remained below 0.25—indicating a persistent low-resilience state, marked by sustained economic growth, frequent natural disasters, and declining ecological self-recovery capacity. Forecasting results suggest that, under current development trajectories, Kunming’s UER will remain low over the next decade. This study is the first to integrate the DPSIR framework, entire-array polygon indicator method, and Grey System Forecasting Model into the evaluation and prediction of urban ecosystem resilience in plateau-mountainous cities. The findings highlight the ecosystem’s inherent capacities for self-organization, adaptation, learning, and innovation and reveal its nested, multi-scalar resilience structure. The DPSIR-based framework not only reflects the complex human–nature interactions in urban systems but also identifies key drivers and enables the prediction of future resilience patterns—providing valuable insights for sustainable urban development.

1. Introduction

Compared with 20 years ago, 54% of the world’s population currently lives in urban areas, and urban ecosystems are facing tremendous challenges and transformations. The current model of urbanization is unsustainable in many respects and has led to a series of social, economic, and environmental issues, such as biodiversity loss, epidemic outbreaks, rising sea levels, ecological degradation, the increased frequency of natural disasters, urban heat island effects, smog, traffic congestion, severe soil erosion, widening income gaps, dysfunctional urban ecosystem services, and an unreasonable spatial allocation [1]. The uncertainty and spread of events—such as the record-breaking heatwaves in many parts of the world in June 2021 and the ongoing rise in sea levels—have further underscored the importance of urban ecosystem resilience (UER). The term “resilience” originates from the Latin word “resilio”, which means “to rebound or return to a previous state (after being damaged).” After being introduced by Holling, the concept began to be applied across various disciplines in the 1970s and remains relevant today [2].

The original definition of resilient cities is the ability of a city’s structure and functions to maintain their original state after changes to and reorganization of the internal and external driving forces of urban development [3]. Later, resilience was defined as “the ability of a city to absorb and adapt to external disturbances while preserving its main characteristics, structures, and key functions,” primarily reflected in four dimensions: governance, energy metabolism, spatial environment, and social dynamics [4]. Ahern [5] identified multifunctionality, redundancy, modularity, diversity, and networking as the five core components of resilient cities. Adger [6] argued that redundant loops within urban networks can help cities maintain spatial and functional connectivity in the face of various disturbances. UER refers to the capacity of urban ecosystems to withstand and recover from external shocks without undergoing fundamental changes to their essential life processes and structural integrity [7].

UER is multi-dimensional, complex, and dynamic; therefore, multiple factors must be considered when developing a resilience evaluation system. The assessment of urban ecosystem resilience should be conducted using qualitative, quantitative, and integrated approaches, tailored to the specific attributes of urban environments. At the regional scale, which refers to a landscape composed of a spatial mosaic of multiple ecosystems or administrative unit-based urban ecosystems, the scope of ecosystem research is further expanded [8]. Among current frameworks for regional-scale evaluations, the four-category indicator system proposed by the US Rockefeller Foundation is considered relatively mature. It includes the dimensions of health and well-being, economy and society, urban systems and services, and leadership and strategy [9]. In the context of climate change and disaster risk prevention, the Agency for International Development (AID) [10], the World Bank [11], and the United Nations Office for Disaster Risk Reduction (UNDRR) have each proposed indicator systems that cover socio-economic conditions, coastal resource management, early warning and evacuation, urban disaster governance structures, financial capacity, political commitment, and risk mitigation measures [12]. Urban scale, density, and form are key attributes that reflect the processes of urban development. A resilient urban form enables the urban system to function as a network of interconnected spatial and socio-ecological subsystems, characterized by dynamic and evolving spatiotemporal patterns. Such a system can effectively maintain its integrity, habitability, and functionality under changing conditions [13]. Influential models for evaluating urban resilience include the PSR model (Pressure–State–Response) developed by the Organization for Economic Co-operation and Development (OECD) [14], the DSR model (Driving force–State–Response) proposed by the United Nations Conference on Environment and Development (UNCED) [15], and the DPSIR model (Driving force–Pressure–State–Impact–Response) introduced by the European Environment Agency (EEA) [16].

Furthermore, other models including System Dynamics Models (SDMs) and Analytical Network Processes (ANPs) have also been applied in urban resilience evaluations [17]. By integrating probabilistic models with comprehensive indicators, urban resilience assessment tools have been developed to evaluate the multi-period, multi-dimensional, and dynamic evolution of urban systems [18]. Moreover, the sustainable development framework has been integrated with the concept of urban resilience to assess UER by analyzing and modeling issues related to the economic, social, environmental, and infrastructural dimensions of cities [19].

Currently, many scholars have devoted substantial efforts to enhancing overall urban resilience, leading to the emergence of a wide variety of assessment methods [20]. For instance, Feng et al. [21] proposed a “scale–density–morphology” elasticity framework and an exponential model grounded in landscape ecology and evolutionary resilience theory. Building on this foundation, a three-dimensional analytical framework was established in 2021 using landscape ecology principles. A corresponding resilience index system was constructed through the ECP framework—representing “Exposure”, “Connection”, and “Potential”—to comprehensively evaluate urban resilience across different stages of spatial evolution. Abebaw et al. [22] utilized land-cover changes detected by remote sensing and applied the DPSIR framework to analyze land use planning, sustainable resource utilization, and urban development in sub-Saharan Africa. Jieer et al. [23] developed an urban resilience model that quantitatively evaluates stormwater-related disasters by integrating the “Pressure–State–Response” theory, disaster theory, and ecological theory. This model incorporates rough set theory to unify urban economic, social, and ecological systems, and it leverages the CMIP5 model to conduct a multi-scenario dynamic trajectory analysis and spatial visualization of urban resilience. Xiao et al. [24] constructed a resilience evaluation system using the DPSIR framework and the entropy method. Using data from 115 counties and districts in Guangdong Province, they assessed rural human settlement resilience in 2020 and analyzed spatial differences and influencing factors using the Geographically Weighted Regression (GWR) model. Kees et al. [25] evaluated urban security and resilience in the context of urban water system security through the DPSIR model, arguing that urban water security fundamentally depends on the system’s ability to respond adequately to external pressures. The effectiveness of these responses determines how pressure evolves and affects a system’s functionality, ensuring that cities can remain operational even under high-stress conditions. In addition to technical and ecological perspectives, several scholars emphasize that urban resilience planning must incorporate principles of equity and social justice, particularly for low-income communities [26,27]. Shokry et al. [28] proposed a spatially explicit and quantitative method to evaluate whether “green resilient infrastructure” can effectively protect vulnerable social groups during climate-related disasters, contributing a new conceptual framework to urban adaptation research and practice.

Urban resilience embodies a city’s capacity to continuously evolve and adapt. Understanding future trends of urban resilience through predictive modeling is a prerequisite for promoting healthy and stable urban development. At present, related studies primarily focus on prediction methods such as System Dynamics Modeling (SDM) [29], the 3D spatiotemporal forecasting algorithm (STFA) based on multi-source transfer learning [30], and the dynamic prediction of urban resilience based on the Grey Forecasting Model (GM) [31]. Among them, the GM model forecasts future trends by accumulating or subtracting discrete time-series data and constructing continuous differential equations with time as the independent variable. It is a grey prediction method that calculates correlations based on the developmental similarity among internal factors of the ecosystem. Over the past decades, grey system theory has gradually developed into a comprehensive theoretical system built on grey fuzzy sets, an analytical framework supported by grey relational models, and a modeling system centered on the GM model. This modeling framework is particularly suitable for medium- and long-term scenario simulations regarding the coordinated development of urbanization and the ecological environment. It can identify key influencing factors in both historical and future urban–ecological systems and reveal the interrelationships between urban ecosystems and urban development. With its simple computational principles and high predictive accuracy, the GM model has been widely applied in simulation and forecasting across economic, social, and technological fields [31].

In summary, increasing scholarly attention has been devoted to the enhancement of UER, with a variety of assessment models and methods being proposed. These include urban resilience models developed based on perspective theory, disaster theory, and ecological theory, often in combination with CMIP5 climate models, to quantitatively evaluate rain and flood disasters and to support a multi-scenario dynamic trajectory analysis and spatial visualization. Other methods integrate System Dynamics Models (SDMs) and the Analytic Network Process (ANP) for comprehensive assessments and scenario simulations. Additionally, some studies employ intuitionistic fuzzy set theory and the TOPSIS method to construct urban resilience evaluation index systems from four key dimensions: ecological environment, municipal infrastructure, economic development, and social progress. Predictive modeling approaches based on the Grey Model (GM) are also used to simulate future trends in urban resilience. Furthermore, a modified DPSIR model, derived from the PSR framework, has been applied to reflect the complex interactions between human activities and the ecological environment. Compared to other methods, the DPSIR model is particularly effective in representing the integration of natural and socio-cultural indicators. It facilitates a clearer understanding of the elastic state of urban ecosystems under uncertain conditions and helps identify major sources of pressure, underlying causes, and appropriate countermeasures for an improvement in resilience.

Due to the ecological fragility inherent to high-altitude regions, plateau-mountainous cities often exhibit fragmented natural ecological patches within the urban fabric, including rivers, grasslands, and forest belts. During the urbanization process, these cities face complex and diverse challenges typical of urban ecosystems, such as an insufficient provision of public service resources, frequent natural disasters, and severe lake pollution. Analyzing the differentiation patterns of subsystems within the plateau-mountainous urban ecosystem reveals that, due to variations in the gradients and hierarchical levels of system components, as well as differences in element combinations, the study area constitutes an embedded composite system composed of multiple landscape types, essentially a plateau-mountainous urban ecosystem (Figure 1). This type of system is characterized by (1) a more pronounced three-dimensional spatial integrity and independence; (2) structural and functional diversity within ecosystems; (3) an inherent system vulnerability; and (4) the interdependence and synchronization of ecological and economic benefits [32]. Therefore, the primary objective of this study is to evaluate the resilience status of plateau-mountainous urban ecosystems by constructing a resilience assessment index system based on the principles of the DPSIR model. This evaluation aims to understand the overall health of regional ecosystems, identify potential ecological crises within urban systems, reveal the interactions between socio-economic activities and the ecological environment in plateau-mountainous cities, and forecast the future resilience status of these ecosystems. The ultimate goal is to provide informed recommendations for ecosystem protection strategies in similar urban regions.

2. Methods

2.1. Research Steps

This study integrates the DPSIR model into the assessment of UER by analyzing the dynamic changes in 25 specific indicators across five dimensions: driving forces, pressures, states, impacts, and responses. Based on the quantified model results, the entire-array polygon indicator method is applied to eliminate the influence of unit dimensions. Subsequently, standardized data are used to evaluate the overall resilience level of the urban ecosystem during the study period. By analyzing the dynamic variation in indicator values in conjunction with the actual conditions of the study area, the underlying causes of these changes are explored. Finally, the Grey System Prediction Model (GM) is employed to forecast the resilience status of the urban ecosystem in the study area over the next decade, forming the basis for proposing targeted restoration and protection strategies.

2.2. Data Sources

Remote sensing imagery data: The study area covers the six districts and eight counties of Kunming City, located between 102°10′–103°40′ E and 24°23′–26°22′ N, with altitudes ranging from 1500 to 2800 m and a total area of 21,012.54 km² (Figure 2). Landsat TM remote sensing data for the years 2006, 2011, and 2016 were obtained from the Geospatial Data Cloud. All images have a spatial resolution of 30 m and cloud cover of less than 10%. A land use classification was conducted using the remote sensing imagery, forming the foundation for calculating relevant indicators in this study. The interpretation results show that the overall classification accuracy exceeds 90%, with Kappa coefficients greater than 0.7, indicating a good classification quality that meets standard requirements (Figure 3). Other fundamental data were primarily sourced from the Kunming Statistical Yearbooks for 2007 and 2012 [33,34], the Kunming National Economic and Social Development Statistical Bulletin in 2016 [35]; the Yunnan Water Resources Bulletin 2011 and Kunming Water Resources Bulletin 2016 [36,37]; the Nature Reserve List of Yunnan Province 2011 [38]; the National Nature Reserve List 2011 and 2016 [39,40]; and the Environmental Status Bulletin of Kunming City 2016 [41].

2.3. The DPSIR Model Construction

The DPSIR evaluation model, originally proposed by the European Environment Agency (EEA) in 1998, serves as a comprehensive environmental information framework that effectively illustrates the complex interactions between human activities and the natural environment. It captures the influence of external drivers—such as economic, social, and cultural factors—on environmental conditions, as well as the societal responses to the resulting changes in habitat quality and ecological status. The model is widely recognized for its systematic structure and holistic analytical capacity, making it particularly well-suited for assessing the dynamics of coupled human–environment systems [35,42].

The primary logical structure of the DPSIR model can be expressed as follows: urban economic and social development (i.e., driving force, D) exerts pressure (P) on the ecosystem. Under the influence of these pressures, the ecological state (S) shifts from its previous stable condition. As the ecosystem undergoes changes in response to the pressure, it generates counter-effects on both the ecosystem itself and human society, resulting in various impacts (I). In turn, human society responds (R) to these changes and impacts through environmental, economic, and other policy decisions or management measures, thereby alleviating the pressure on the environment.

To ensure that each factor and its associated indicators accurately and comprehensively reflect the effects and feedback of social, economic, cultural, and natural conditions on the ecosystem within the study area, a questionnaire was designed comprising 43 indicators across the five components of the DPSIR framework (Appendix A). A total of 15 valid responses were collected from experts affiliated with Yunnan University, Central South University, Nanjing University, and other scholars based in the Kunming area. The Analytic Hierarchy Process (AHP) method was then applied, using specialized software to analyze and rank the relative importance of each indicator within its corresponding DPSIR component. Based on the results and taking data availability into consideration, 25 representative indicators were ultimately selected for constructing the DPSIR evaluation model (Figure 4) (

Table 1. List of Kunming ecosystem resilience evaluation indicators, calculation results, and standardized values.

Factor	No.	Indicator	2006		2011		2016
			Calculated Results	Standard Value	Calculated Results	Standard Value	Calculated Results	Standard Value
Driving force	1	Natural population growth (%)	6.47	−0.622	5.66	−0.664	6.21	−0.635
	2	Annual GDP growth (%)	12.07	−0.366	18.36	−0.136	8.32	−0.531
	3	Urbanization level (%)	58.99	0.662	66.00	0.739	71.05	0.788
	4	Urban population density (persons/hm2)	2.928	−0.817	3.087	−0.807	3.202	−0.801
	5	Resident consumption price growth (food) (%)	1.8	−0.885	10.8	−0.420	3.1	−0.807
Pressure	6	Per capita farmland (hm2/person)	0.077	−0.997	0.067	−0.998	0.065	−0.998
	7	Per capita urban construction land (hm2/person)	0.048	−0.999	0.053	−0.998	0.069	−0.997
	8	Land reclamation and cultivation rate (%)	22.54	−0.007	20.72	−0.061	20.89	−0.056
	9	Proportion of natural disaster-stricken area (%)	19.89	−0.087	32.8	0.247	5.23	−0.687
	10	Unit GDP energy consumption (ton standard coal/10,000 CNY)	1.143	−0.927	0.557	−0.965	0.547	−0.966
State	11	Evenness Index (EI)	0.7968	−0.949	0.7864	−0.950	0.7121	−0.955
	12	Shannon Diversity Index (SHDI)	1.5506	−0.901	1.5303	−0.902	1.3856	−0.911
	13	Aggregation Index (AI)	97.1776	0.986	97.2779	0.987	97.6256	0.989
	14	Interspersion and Juxtaposition Index (IJI)	69.5997	0.954	71.8750	0.932	53.2194	0.838
	15	Landscape Shape Index (LSI)	92.1164	0.774	88.8813	0.796	76.6895	0.590
	16	Normalized Vegetation Index (NDVI)	0.1184	−0.994	0.1542	−0.991	0.1572	−0.992
Impact	17	Per capita GDP	1.976	−0.874	3.869	−0.762	6.392	−0.626
	18	Economic density (10,000 CNY/hm2)	5.79	−0.657	11.94	−0.372	20.47	−0.069
	19	Per capita disposable income growth (%)	10.2	−0.446	11.0	−0.411	8.2	−0.537
	20	Per capita green space area (m2/person)	12.72	−0.340	25.36	0.070	23.63	0.024
	21	Per capita water resources (m3/person)	0.081	−0.996	0.035	−1.000	0.088	−0.995
Response	22	Proportion of natural reserve area (%)	14.03	−0.289	3.44	−0.787	2.84	−0.822
	23	Proportion of environmental protection investment to GDP (%)	2.82	−0.823	5.30	−0.683	3.71	−0.771
	24	Overall utilization of industrial solid waste (%)	92.62	0.958	71.35	0.791	99.36	1.000
	25	Afforestation area (10,000 mu)	0.53	−0.967	37.16	0.335	80.30	0.868

Note: Indicator sources are as follows: a. Indicators 1, 4, 10, 17, 20, and 21 were obtained from [43]. b. Indicators 2, 7, 23, and 24 were obtained from [44]. c. Indicator 3 was obtained from [45]. d. Indicator 6 was obtained from [46]. e. Indicator 11 was obtained from [47]. f. Indicator 12 was obtained from [48]. g. The remaining unmentioned indicators are derived from the indicator questionnaire survey conducted in “2.3 DPSIR Model Construction” and were finally determined based on the ecosystem characteristics of the study area.

).

In constructing the DPSIR model, various indicators representing the degree of UER are used to capture the complexity and dynamic characteristics of urban ecosystems, which are shaped by both natural conditions and socio-economic factors.

Driving forces (D) refers to the fundamental causes that promote changes in the existing ecological state or initiate new developments, namely, the potential factors that may drive changes in the ecological environment. In the context of urban development, the relocation of numerous industrial enterprises can stimulate economic growth, generate employment opportunities, and attract labor inflows, thereby promoting urban economic expansion and accelerating the urbanization process. However, such developments may also lead to increased population concentrations and intensified land development activities, such as the promotion of investment, which can result in the conversion of agricultural and ecological land into construction land. This in turn may cause increased emissions of waste gases and other pollutants, thereby contributing to ecological degradation. Based on existing research and the actual conditions of the study area, this study selects five indicators that most effectively reflect urban human activities and living standards as the driving force evaluation indicators: natural population growth rate, annual GDP growth rate, urbanization level, urban population density, and consumer price growth rate (food).

Pressure (P) refers to the direct causes of changes in the ecological state—that is, factors that directly affect the ecosystem. It reflects the manifestation of driving forces, primarily through human-induced and environmental stressors. Among these, land use change is one of the most direct indicators reflecting the pressure status of urban ecosystems. Per capita cultivated land area, per capita urban construction land area, and land reclamation rate can best illustrate the impact of population aggregation and land use intensity on the ecological environment. Additionally, energy consumption per unit of GDP reflects the energy demands of socio-economic development and the resulting pressure on the ecosystem. Considering that the study area is located in a plateau-mountainous region prone to natural disasters, the proportion of land affected by natural disasters is also included as an indicator to capture natural stress factors. These indicators aim to comprehensively assess the combined pressures exerted by both human activities and natural forces on the ecosystem.

State (S) refers to the condition and appearance of the ecological environment under the combined influence of driving forces and pressure factors. These indicators are designed to directly reflect the resilience characteristics of the ecosystem. The landscape patterns and structures based on current land use conditions can present the true appearance and characteristics of ecosystems under specific land use conditions. Based on the land use classification results from three time periods, this study selects the following landscape-level indicators to characterize the structural complexity and ecological diversity of the ecosystem: Shannon’s Diversity Index (SHDI), Evenness Index (EI), Aggregation Index (AI), Interspersion and Juxtaposition Index (IJI), and Landscape Shape Index (LSI). In addition, vegetation indices are used to reflect vegetation productivity, which is a key component of ecosystems’ vitality and resilience (refer to

Table 2. Table of specific definitions and calculation methods for “state” indicators.

Indicator Name	Indicator Definition	Calculation Method
Evenness Index (EI)	An index that reflects the uniformity of area distribution among different land use types.	$E={\displaystyle \frac{H}{H_{max}}}$ where $E \mathrm{is} \mathrm{the} \mathrm{Evenness} \mathrm{Index};$ $H \mathrm{is} \mathrm{the} \mathrm{Diversity} \mathrm{Index};$ $H_{max}$ is the Maximum Diversity Index.
Shannon Diversity Index (SHDI)	An index that reflects landscape diversity.	$H_T=-{\displaystyle \sum _{i=1}^m}{P}_i\ln P_i$ where $H_T \mathrm{is} \mathrm{the} \mathrm{Shannon} \mathrm{Diversity} \mathrm{Index};$ $P_i \mathrm{is} \mathrm{the} \mathrm{proportion} \mathrm{of} \mathrm{land} \mathrm{use} i;$ $m$ is the number of land use types.
Aggregation Index (AI)	An index that reflects the degree of clustering of a specific landscape type.	${AI}_i={\displaystyle \frac{e_{ii}}{\mathrm{m}\mathrm{a}\mathrm{x}\_e_{ii}}}$ where ${AI}_i \mathrm{is} \mathrm{the} \mathrm{Aggregation} \mathrm{Index} \mathrm{of} \mathrm{landscape} \mathrm{type} i;$ $e_{ii} \mathrm{is} \mathrm{the} \mathrm{number} \mathrm{of} \mathrm{shared} \mathrm{edges} \mathrm{between} \mathrm{adjacent} \mathrm{cells} \mathrm{of} \mathrm{type} i;$ $\mathrm{m}\mathrm{a}\mathrm{x}\_e_{ii} \mathrm{is} \mathrm{the} \mathrm{maximum} \mathrm{possible} \mathrm{number} \mathrm{of} \mathrm{shared} \mathrm{edges} \mathrm{of} \mathrm{type} i$.
Interspersion and Juxtaposition Index (IJI)	An index that reflects the overall interspersion and juxtaposition of different patch types in the landscape.	It measures the percentage probability that a given patch type is adjacent to other patch types in the landscape.
Landscape Shape Index (LSI)	An index that reflects the overall shape characteristics of the landscape. It not only considers the shape information of individual patches but also reveals the degree of aggregation or dispersion among different patches.	$LSI={\displaystyle \frac{0.25E}{\sqrt{A}}}$ where $E \mathrm{is} \mathrm{the} \mathrm{total} \mathrm{length} \mathrm{of} \mathrm{all} \mathrm{patch} \mathrm{boundaries} \mathrm{in} \mathrm{the} \mathrm{landscape};$ $A$ is the total area of the landscape.
Normalized Vegetation Index (NDVI)	An indicator that reflects the vegetation productivity of the ecosystem.	$NDVI={\displaystyle \frac{NIR - RED}{NIR+RED}}$ where $NIR \mathrm{is} \mathrm{the} \mathrm{reflectance} \mathrm{in} \mathrm{the} \mathrm{near}-\mathrm{infrared} \mathrm{band};$ $RED$ is the reflectance in the red band.

for detailed definitions of the indicators).

Impact (I) refers to the reaction of the ecological environment to human society under pressure [33]. Urban ecosystems are not only shaped by development activities but also feed back into the socio-economic system by influencing urban quality of life and sustainability. These impacts are manifested in three main ways: first, through effects on the level of economic development and residents’ livelihoods; second, through constraints on urban greening and ecological construction due to competing land uses; and third, through feedback effects on ecological systems’ performance. In this study, impact indicators are divided into two categories: socio-economic impacts (including per capita GDP, economic density, and per capita disposable income growth) and environmental impacts (including per capita public green space area and per capita water resources).

Response (R) refers to the measures and interventions implemented directly or indirectly to mitigate negative impacts on ecosystem resilience. These responses are typically initiated by government agencies or social organizations through policy adjustments, planning, and technical solutions aimed at improving environmental conditions. In this study, response indicators are categorized into socio-economic and environmental responses. The socio-economic response indicators reflect artificial interventions for improving UER, such as the proportion of natural reserve area, the share of environmental protection expenditure in GDP, and the overall utilization of industrial solid waste. Meanwhile, environmental response indicators capture natural restoration and protection efforts, with afforestation area used as a representative metric.

2.4. Entire-Array Polygon Indicator Method

The entire-array polygon indicator method, based on the urban ecosystem resilience evaluation index system, constructs a central n-sided polygon using the maximum standardized values of n indicators as the radius. A total of (n − 1)/2 distinct irregular central polygons can be formed from the n indicators. The composite index value is calculated based on the relationship between each standardized indicator value and its corresponding maximum and minimum values [49]. The calculation formula is as follows:

The indicator values are standardized using a dual-curve standardization function, expressed as

$$F\left(x\right)={\displaystyle \frac{a}{bx+c}} (a \ne 0, x \ge 0)$$

(1)

$F\left(x\right)$ satisfies the following conditions: $F\left(L\right)$= 1, $F\left(T\right)=0$, $F\left(U\right)=1$, where $U$ is the upper limit of indicator $x$, $L$ is the lower limit of indicator $x$, and $T$ is the threshold value [50].

Based on the above conditions, the following formula can be obtained:

$$F\left(x\right)={\displaystyle \frac{(U-L)(x-T)}{\left(U+L-2T\right)x+UT+LT-2UL}} (a \ne 0, x \ge 0)$$

(2)

According to the properties of $F\left(x\right)$, the standardized function maps indicator values within the interval [$L, U$] to the range [−1, 1]. This mapping alters the growth rate of the indicator: when the indicator value is below the threshold $T$, the growth rate of the standardized indicator gradually decreases; when the indicator value exceeds the threshold, the growth rate gradually increases. In other words, the indicator shifts from a linear growth along the x-axis before standardization to a nonlinear “fast–slow–fast” growth pattern after standardization, with the threshold value $T$ serving as the inflection point of the indicator’s growth rate [50].

Therefore, for the $i$-th indicator $x_i$, the standardization calculation formula is as follows:

$$S_i={\displaystyle \frac{(U_i-L_i)(x_i-T_i)}{\left(U_i+L_i-2T_i\right)x_i+U_i{T}_i+L_i{T}_i-2U_i{L}_i}}$$

(3)

where $L_i$, $T_i$, and $U_i$ represent the minimum, threshold, and maximum values of the indicator $x_i$, respectively [50].

The final composite index is then calculated by

$$S={\displaystyle \frac{\sum _{i=j}^{i,j}\left(S_i+1\right)\times (S_j+1)}{2n\times (n-1)}}$$

(4)

where $S_i$ and $S_j$ represent the standardized values of the $i$ and $j$ indicators, respectively; n represents the number of indicators; and $S$ represents the overall indicator value.

The evaluation grade standards for the resilience status of urban ecosystems are shown in

Table 3. Criteria for evaluating the resilience status of urban ecosystems.

Level	Low Resilience	Medium Low Resilience	Medium High Resilience	High Resilience
Interval	<0.25	0.25–0.5	0.5–0.75	>0.75

.

The entire-array polygon indicator method eliminates subjectivity in the process of determining indicator weights, enabling a more objective and quantitative assessment [51]. It has been widely applied with satisfactory outcomes in environmental and resource studies. However, this methodology has not yet been employed in the assessment of urban resilience.

2.5. Gray System Forecasting Method

As a composite economic–social–ecological system, the urban ecosystem exhibits typical characteristics of a gray system, including a degree of ambiguity [52]. To predict the level of UER in the study area, the gray system forecasting method was adopted, as it effectively accounts for the uncertainties inherent in urban ecosystems and quantitatively captures their dynamic fluctuations. Given that the original data were relatively dispersed, conventional modeling approaches often failed to meet the required accuracy tests, making precise prediction challenging. Therefore, interval forecasting was employed to estimate a plausible range of future outcomes, as presented in Equations (5)–(10).

Suppose $X^{\left(0\right)}=(x^{\left(0\right)}\left(1\right), x^{\left(0\right)}\left(2\right), \wedge ,x^{\left(0\right)}(n\left)\right)$ is the original sequence, and its 1-AGO sequence is $X^{\left(1\right)}=(x^{\left(1\right)}\left(1\right),x^{\left(1\right)}\left(2\right),\wedge ,x^{\left(1\right)}(n\left)\right)$, where n is the number of original data items.

$$\mathrm{Let} {\sigma }_{max}=\underset{1\le \mathrm{k}\le \mathrm{n}}{\max }\left\{x^{\left(0\right)}\right(k\left)\right\}, {\sigma }_{min}\{x^{\left(0\right)}(k)\}$$

(5)

Then, the upper-bound function $f_s(n+t)$ and lower-bound function $f_u(n+t)$ are expressed as follows:

$$f_s \left(n+t\right)=x^{\left(0\right)} \left(n\right)+{t\sigma }_{max}$$

(6)

$$f_u \left(n+t\right)=x^{\left(1\right)} \left(n\right)+t{\sigma }_{min}$$

(7)

where t = 1, 2 … m, and m is the predicted number of data items.

The inverse accumulated generating operation of $X_s^{\left(0\right)}$ is deduced as follows:

$${{\widehat{x}}_s}^{\left(0\right)} \left(n+t\right)=f_s \left(n+t+1\right)-f_s (n+t)$$

(8)

The inverse accumulated generating operation of $X_u^{\left(0\right)}$ is also deduced as follows:

$${{\widehat{x}}_u}^{\left(0\right)} \left(n+t\right)=f_u \left(n+t+1\right)-f_{u }(n+t)$$

(9)

$${∆}_{smax}=\underset{n=1\le k\le n=m}{\max }\{{{\widehat{x}}_s}^{\left(0\right)} (k\left)\right\}, {\Delta }_{umin}=\underset{n=1\le k\le n=m}{\min }\{{{\widehat{x}}_u}^{\left(0\right)} (k\left)\right\}$$

(10)

where ${∆}_{smax}$ and ${\Delta }_{umin}$ represent the upper and lower bounds, respectively, of the predicted values.

3. Evaluation Results of Urban Ecosystem Resilience in Kunming

3.1. Calculation Results of Ecosystem Resilience Evaluation Indicators in Kunming

Based on the above model and the relevant data for Kunming City from the years 2006, 2011, and 2016, the evaluation results for the ecosystem resilience indicators in the study area were obtained. To eliminate the effects of differing indicator dimensions, hyperbolic functions were applied, and all the indicators were standardized using the R programming environment [42]. The standardized results are presented in Table 1.

3.2. Construction and Analysis Results of Entire-Array Polygon Diagrams for Individual Indicators

The entire-array polygon indicator method was applied to evaluate the urban ecosystem resilience indicators of Kunming. Based on the standardized results of each indicator listed in Table 1, polygon diagrams can be constructed. The individual indicator polygons for each factor level are shown as follows (Figure 5).

Several observations can be drawn from Table 1 and Figure 5:

(1) The level of urban ecosystem resilience (UER) is significantly influenced by the “driver” factors related to urban development.

During the study period, the continuous expansion of urban areas was accompanied by a steady increase in urban population density, indicating isotropic spatial changes over time. The annual growth rate of urban GDP showed a downward trend, while both the natural growth rate of the urban population and the growth rate of consumer costs (food) first decreased and then increased. Meanwhile, the continuous rise in population density placed increasing pressure on the environmental carrying capacity. The fluctuations in food price growth may reflect a reduction in cultivated land—possibly leading to food shortages—or may indicate a rise in ecological costs, suggesting that improved living standards are achieved at the expense of ecological resources. Among the five selected driver indicators, the data over the past three years reveal that, except for the positive contributions of the rising natural population growth rate and the improvement in urbanization level, the remaining three factors—especially the decline in GDP growth—tend to weaken the overall resilience of the urban ecosystem.

(2) “Pressure” factors such as energy consumption and natural disasters are direct causes influencing UER.

During the study period, the per capita cultivated land area consistently declined, while the per capita urban construction land area steadily increased. This negative correlation between the two reflects the ongoing acceleration of urban construction land’s expansion driven by urban development, whereby cultivated land is increasingly converted into construction land—ultimately diminishing the resilience of the urban ecosystem. Meanwhile, the urban land reclamation rate showed a downward trend year by year, and energy consumption per unit of GDP also declined, which to some extent contributed positively to the improvement in UER. However, the proportion of land affected by natural disasters peaked in 2011, significantly weakening the urban ecosystem by increasing its fragility, reducing the environment’s self-recovery capacity, and intensifying the ecological pressure from disasters, thereby further decreasing the level of UER.

(3) Landscape pattern indices can characterize the “state” of urban ecosystem resilience at the landscape level.

Both the Diversity Index and the Evenness Index showed a gradual decline over the study period, suggesting that environmental pollution—particularly from the Dianchi Lake Basin—hindered normal ecological processes. This led to a reduction in ecosystem diversity within the study area and a decrease in plant species richness. The Aggregation Index (AI) increased over the previous three years, largely due to the clustering of artificial urban construction land. However, this coincided with the fragmentation of other landscape patches, resulting in a more pronounced aggregation of specific landscape types but with an increasingly uneven spatial distribution. The Interspersion and Juxtaposition Index (IJI) peaked in 2011, exhibiting characteristics typically found in arid regions. When considered alongside changes in the “proportion of land affected by natural disasters” among the pressure factors, it becomes evident that the study area—being a plateau-mountainous city—was affected by drought events. These disasters caused a spatial heterogeneity in soil moisture, leading to greater complexity in the landscape structure. Altogether, these changes suggest that natural landscape elements in the region have been disrupted by both natural and anthropogenic factors, contributing to a continued decline in the resilience of the urban ecosystem.

(4) “Impact” factors, such as socio-economic conditions, play a guiding role in shaping UER.

During the study period, both per capita GDP and economic density in the study area exhibited strong upward trends, indicating that with the support of ecosystem services, the urban economy has developed rapidly and economic benefits have steadily improved. The per capita disposable income of urban residents also increased consistently, though the growth rate first accelerated and then slowed—aligning with the characteristics described by the Environmental Kuznets Curve. Per capita water resources dropped sharply in 2011 but returned to normal levels by 2016. This fluctuation was primarily due to a prolonged drought affecting the study area from 2009 to 2011, after which precipitation increased, leading to higher water levels in rivers and lakes, as well as enhanced water storage capacity in reservoirs and ponds. Meanwhile, the per capita green space area saw significant improvement from 2011 to 2016, reflecting a rapid enhancement in urban ecological conditions. These changes in the ecosystem have supported sustained socio-economic development and, at the same time, helped to offset the loss of ecological resources, indicating a continuous improvement in the resilience of the urban ecosystem.

(5) Changes in “response” factor indicators reflect the disruptive effects of social and human activities on UER.

During the study period, the proportion of land designated as nature reserves declined significantly from 2006 to 2016. This was accompanied by issues such as the absence of formal establishment and lack of organizational management, resulting in ineffective conservation practices, reduced ecological responsiveness, and a year-on-year decline in biodiversity. The proportion of environmental protection investment relative to urban GDP increased from 2006 to 2011 but declined from 2011 to 2016. Meanwhile, the comprehensive utilization rate of industrial solid waste dropped sharply in 2011 but showed a marked improvement by 2016. Issues such as unstable waste treatment efficiency, the incomplete implementation of environmental policies, low environmental management effectiveness, and a “development first, governance later” approach have had a negative impact on the ecosystem’s ability to respond naturally, thereby weakening UER. However, afforestation efforts have steadily increased year by year, with numerous reforestation projects being implemented. Policies such as the “Grain-for-Green” (returning farmland to forest) initiative have shown tangible results, contributing positively to environmental recovery and helping to enhance the overall level of UER.

3.3. Calculation of the Comprehensive Index Values and Assessment of the Resilience Level

According to Equation (4), the standardized values of each indicator for each year were used to calculate the comprehensive index value. The final assessment results of UER in the study area showed that the urban ecosystem comprehensive index values in 2006, 2011, and 2016 were 0.1796, 0.2432, and 0.2186, respectively. These values all fall within the “low resilience” category. This indicates that the urban ecosystem resilience in the study area has not yet reached a state of coordination across the five dimensions of “Driving Forces–Pressure–State–Impact–Response.” Overall, the results reflect a growing tension between rapid urban development and the human–nature relationship in recent years. On the one hand, the study area has maintained strong and continuous economic growth; on the other hand, it has experienced frequent natural disasters, a declining ecological self-recovery capacity, and an incomplete implementation of environmental response measures. Therefore, it is urgent to explore strategies for enhancing urban ecosystem resilience to ensure the sustainable development of the study area in the future.

3.4. Prediction of Ecosystem Resilience Status in Kunming

Based on the comprehensive index values from the three study periods, the grey system forecasting method was applied to predict the status of ecosystem resilience over the next 10 years.

$$X^{\left(0\right)}=\left(x^{\left(0\right)} \left(1\right),{ x}^{\left(0\right)} \left(2\right),{ x}^{\left(0\right)} \left(3\right)\right)=(0.1799,0.2432,0.2186),$$

where $x^{\left(0\right)} \left(k\right)(k=1, 2, 3)$ represents the original sequence values, with $x^{\left(0\right)} \left(1\right)=0.1799$ obtained from 2006, $x^{\left(0\right)} \left(2\right)=0.2432$ from 2011, and $x^{\left(0\right)} \left(3\right)=0.2186$ from 2016.

The calculation was conducted using Equations (5)–(10), yielding the following results:

$$x^{\left(0\right)}\left(k\right)={\sum }_{i=1}^k{x}^{\left(0\right)} \left(1\right), \mathrm{and} \mathrm{got} \mathrm{the} 1-\mathrm{AGO} \mathrm{sequence} \mathrm{of} X^{\left(0\right)},$$

where $X^{\left(1\right)}=\left(x^{\left(1\right)},{ x}^{\left(1\right)}\left(2\right),{ x}^{\left(1\right)}\left(3\right)\right)=(0.1799,0.4231,0.6417)$,

$$\mathrm{so} f_s \left(3+k\right)=x^{\left(1\right)} \left(3\right)+k{\sigma }_{max}$$

$$f_u \left(3+k\right)=x^{\left(1\right)} \left(3\right)+k{\sigma }_{min},$$

when $k=1,2$, the highest predicted values are obtained,

$${\widehat{x}}_s^{\left(1\right)}\left(4\right)=f_s\left(3+1\right)=x^{\left(1\right)}\left(3\right)+1\times {\sigma }_{max}$$

$${\widehat{x}}_s^{\left(1\right)}\left(5\right)=f_s\left(3+2\right)=x^{\left(1\right)}\left(3\right)+2\times {\sigma }_{max}$$

From this we can get the cumulative reduction formula of ${X_s}^{\left(0\right)}$:

$${\widehat{x}}_s^{\left(0\right)}\left(4\right)=0.8849-0.6417=0.2432$$

$${\widehat{x}}_u^{\left(0\right)}\left(5\right)=1.1281-0.8849=0.2432$$

The lowest predicted value is as follows:

$${\widehat{x}}_u^{\left(1\right)} \left(4\right)=f_u \left(4+1\right)=x^{\left(1\right)} \left(3\right)+1\times {\sigma }_{min}=0.8216$$

$${\widehat{x}}_u^{\left(1\right)} \left(5\right)=f_u \left(3+2\right)=x^{\left(1\right)} \left(3\right)+2\times {\sigma }_{min}=1.0015$$

From this we can get the cumulative reduction formula of ${X_u}^{\left(0\right)}$:

$${{\widehat{x}}_u}^{\left(0\right)} \left(4\right)=0.8216-0.6417=0.1799$$

$${{\widehat{x}}_u}^{\left(0\right)} \left(5\right)=1.0015-0.8216=0.1799$$

At the same time, ${\sigma }_{max}=\underset{1\le k\le 3}{\max }\left\{x^{\left(0\right)}\right(k\left)\right\}=0.2432,{\sigma }_{min}=\underset{1\le k\le 3}{\min }\left\{x^{\left(0\right)}\right(k\left)\right\}=0.1799$.

And the calculation results can be obtained:

$${∆}_{smax}=0.2432, {∆}_{umin}=0.1799$$

The results suggest that, under the current trajectory, the projected resilience value for 2026 will remain within the range of [0.1799, 0.2432], indicating the continued low-resilience status of the urban ecosystem. In the absence of a decisive intervention, the resilience level is unlikely to improve and may even deteriorate further.

4. Discussion

4.1. Assessing the Resilience Status of Plateau-Mountainous Urban Ecosystems Based on the DPSIR Model

Given the complex challenges and uncertainties faced by urban ecosystems [53], the inherent intricacy of urban systems often makes it difficult to isolate the specific drivers and impacts affecting the provision of ecosystem services [54]. In constructing the DPSIR model, we integrated socio-economic data with geographical data to identify indicators that best capture the ecological fragility, frequent natural disasters, and limited ecological restoration capacity typical of plateau-mountainous cities. This approach enables a comprehensive assessment of how natural, anthropogenic, and social factors collectively influence the resilience of urban ecosystems.

Further research has shown that different cities possess distinct characteristics and face unique challenges. Urban ecosystems are dynamic and exhibit non-equilibrium behavior, continuously evolving over time. Consequently, even when applying the same modeling framework, the selection of specific indicators may vary across different types of cities. In this context, “response” measures—such as policy shifts and strategic adjustments—serve as adaptive mechanisms to address changes in urban ecosystems. These responses are often shaped by socio-economic and other “impact” factors, which in turn influence both the ecosystem service functions and the resilience “state” of urban ecosystems [55]. The resulting dynamics are reflected in the “pressure” exerted on the urban ecosystem and the “drivers” that trigger change. The system adapts and evolves under these interacting forces, demonstrating a degree of self-organization and adaptability—key attributes of ecosystem resilience.

4.2. Strategies to Improve the UER in Plateau-Mountainous Areas

(1) Rational allocation and exploration of potential “drivers” to enhance UER.

Building on the rising natural population growth rate, increasing annual GDP growth, and improving urbanization levels, it is essential to scientifically guide the rational allocation and balanced distribution of industries, population, employment, infrastructure, and resources between urban centers and surrounding areas. This should be carried out in alignment with the ecological environment’s carrying capacity. Efforts should also be made to reasonably control urban population density and the rate of consumer price increases, enhance the quality of development, and unlock the development potential of peripheral counties. These strategies collectively aim to strengthen the resilience of urban ecosystems in plateau-mountainous regions.

(2) Adjust the industrial structure and alleviate “pressure” on the urban ecosystem.

It is essential to maintain a stable per capita cultivated land area and control indicators such as per capita construction land, urban land reclamation rate, and energy consumption per unit of GDP. These measures should be accompanied by efforts to reasonably adjust the industrial structure and enhance the technological sophistication of industrial development. At the same time, the area affected by natural disasters should be reduced through heightened attention to ecological changes, strong disaster preparedness, and efficient emergency response systems. By lowering the frequency and duration of major disasters and strengthening urban disaster resilience, the overall resilience of the urban ecosystem can be progressively improved.

(3) Increase efforts to protect nature and continue to improve the resilience “state” of urban ecosystems.

Efforts should be made to enhance the diversity of urban ecosystems, and maintaining the urban ecological balance is essential to reducing the damage caused by resource extraction to intact and contiguous natural landscape patches. This helps to decrease the fragmentation and dispersion of natural ecological landscape patches, preserve the natural characteristics of landscape shapes, and promote the integrity and continuous distribution of ecological landscapes. Increasing vegetation coverage in urban areas and strengthening the construction of ecological corridors that connect green spaces within the city to surrounding natural patches effectively link ecological flows between urban and peri-urban areas. This connectivity provides food resources and water channels for flora and fauna, thereby promoting biodiversity. Additionally, ecological engineering restoration measures—such as wetland rehabilitation and vegetation restoration on slopes—can enhance the disaster resilience and ecological buffering capacity of urban ecosystems. These efforts ultimately improve the urban ecosystem’s capacity to respond to and recover from external disturbances [56].

(4) Regulate socio-economic and environmental “impact” indicators to enhance UER.

Improving socio-economic impact indicators such as per capita GDP, economic density, and the growth rate of residents’ disposable income, alongside environmental impact indicators, including per capita public green space and per capita water resources, can promote the intensive use of ecosystem resources and raise residents’ economic well-being. This, in turn, supports the expansion of green planting areas, increases vegetation survival rates, and enhances soil and water conservation capacity. Furthermore, strengthening the implementation of water pollution control measures can improve water conservation capabilities, mitigating negative effects common in plateau-mountainous cities such as poor soil and water retention and frequent droughts. These efforts help maintain a balanced relationship between urban development and ecological welfare, thereby improving the UER.

(5) Improve ecological protection efficiency and enhance the UER through “response” measures.

The urban development edge of Kunming City overlaps and borders, to varying degrees, with ecological functional areas such as the Dianchi Lake Basin Protection Zone and the Jiaozi Snow Mountain Nature Reserve. These nature reserves serve not only as critical barriers for regional ecological security but also directly influence key ecosystem services, including urban water supply, the maintenance of biodiversity, and the regulation of air quality. By delineating and connecting ecological redlines more effectively, increasing the proportion of nature reserves within the study area, and strengthening the ecological connectivity between urban green infrastructure and protected areas, spatial integration and functional complementarity between the city and nature reserves can be promoted. This integration enhances the overall ecosystem’s responsiveness and adaptive capacity. Additionally, increasing investment in environmental governance, improving the efficiency of environmental protection efforts, and enhancing the comprehensive utilization rate of industrial solid waste can improve the efficiency of natural resource conservation. “Response” measures such as afforestation and expanding plantation areas also contribute to improving the UER.

(6) Future research could fully utilize remote sensing data from geographic information systems (GISs) to analyze the spatial distribution of ecosystem resilience across counties and cities within the study area. By considering the characteristic elements of ecosystem states at multiple scales and selecting a range of indicators to represent urban ecosystem conditions, such studies can better reveal the interrelationships among adjacent smaller regions within the larger area. This approach would enable researchers to conduct a more detailed analysis of the cultural and socio-economic dynamics within the study region, thereby enhancing the precision of strategies aimed at improving ecosystem resilience. Moreover, it would facilitate a more intuitive and effective visual representation of the research findings through mapping.

5. Conclusions

(1) The UER assessment system based on the DPSIR model offers a comprehensive approach to evaluating interactions among “driving forces,” “pressures,” “state,” “impact,” and “response” in plateau-mountainous cities. Using Kunming as a case study, results from 2006 to 2016 show a consistently low resilience, with composite index values below 0.25. The urbanization-induced expansion of construction land, loss of cultivated land, and increased landscape fragmentation were key “pressure” factors. High proportions of disaster-affected areas and elevated energy use per GDP further undermined ecosystem self-regulation. To enhance resilience, it is essential to regulate “driving forces” at the source, mitigate “pressures,” improve the ecological “state,” and strengthen recovery pathways through targeted “response” policies. Within the DPSIR framework, each indicator category plays a distinct role: “driving forces” and “pressures” represent causes and direct stressors; “state” reflects ecosystem stability; “impact” indicates vulnerability under stress; and “response” captures policy and management interventions. This structure underscores the urban ecosystem’s capacity for self-organization, adaptation, learning, and innovation, forming nested, multi-scale resilience cycles [57].

(2) The UER in plateau-mountainous cities is shaped by development-related “driving forces” such as urbanization, population growth, and rising consumer prices. Construction land’s expansion and arable land’s decline challenge ecological carrying capacity, while “pressure” factors like energy consumption per GDP and disaster exposure directly impair resilience. Landscape pattern metrics such as the Shannon Diversity Index, Aggregation Index, and Landscape Shape Index reflect the diversity and fragmentation of ecological patches, thus representing the “state” of resilience at the landscape level. Socio-economic “impact” indicators—including per capita GDP, economic density, residents’ per capita disposable income, and per capita public green space—offer varying degrees of support to ecological services and can influence the direction of ecosystem resilience. Meanwhile, “response” indicators—such as the proportion of natural reserves, overall utilization of industrial solid waste, and afforestation area—reflect human interventions, including policies and regulations, that enhance future UER in plateau-mountainous areas. Under the combined influence of these factors, the composite resilience index values of the study area in 2006, 2011, and 2016 were 0.1796, 0.2432, and 0.2186, respectively, all indicating a low-resilience state. Therefore, restoring urban ecosystem resilience in plat-eau-mountainous cities requires a comprehensive strategy that includes regulating “driv-ing forces,” reducing “pressures,” improving the “state,” exerting constructive “impacts,” and enhancing “response” measures.

(3) By integrating the entire-array polygon indicator method with the DPSIR model, this study achieved a standardized treatment of indicators with different units and dimensions, ensuring comparability across all variables. The use of entire-array polygon diagrams allowed for an intuitive and visual representation of the dynamic changes among various indicators and DPSIR factors, clearly demonstrating the overall variation patterns across different time periods. This approach provided more accessible and interpretable quantitative insights into the temporal evolution of urban ecosystem resilience, facilitating dynamic comparisons of resilience levels over time. As a result, a more accurate and comprehensive assessment of the urban ecosystem’s overall resilience was obtained, demonstrating the method’s strong generalizability and practical applicability in broader ecological resilience evaluation contexts.

(4) Building upon the multi-period composite index values of urban ecosystem resilience, this study employed the Grey System Forecasting Method to predict future resilience levels under conditions of incomplete information. This approach enables the forecasting of future urban ecosystem resilience and allows for the formulation of targeted restoration strategies aimed at enhancing the system’s self-organizing capacity and its ability to adapt to external pressures. The method offers a high degree of scientific rigor and specificity, making it a valuable tool for guiding future urban ecological resilience planning and adaptive management.