关键词: SRP模型, AHP-熵权组合赋权, 生态脆弱性, 时空格局, 空间自相关分析

Abstract:

Understanding the spatio-temporal evolution of ecological vulnerability in coastal cities is of fundamental importance for revealing the dynamics of coupled human-environment systems and supporting sustainable urban development under the dual pressures of rapid urbanization and global environmental change. Coastal cities are often characterized by high population density, intensive land-use conversion, and complex interactions between natural ecosystems and human activities, which collectively exacerbate ecological stress and increase vulnerability risks. Against this background, a systematic assessment of ecological vulnerability patterns and their driving mechanisms is essential for improving ecological risk prevention, spatial planning, and management strategies. This study takes Qingdao, a representative coastal metropolis in northern China, as a case study to investigate the spatio-temporal evolution of ecological vulnerability from 2011 to 2021. Qingdao has experienced accelerated urban expansion, industrial restructuring, and coastal zone development during the past decade, making it an ideal region for examining vulnerability responses to natural and anthropogenic disturbances. Based on the Sensitivity-Resilience-Pressure (SRP) model, a multidimensional ecological vulnerability evaluation framework was constructed by integrating indicators from natural environmental conditions and socioeconomic development processes. The SRP model emphasizes the combined effects of ecosystem sensitivity to disturbances, intrinsic resilience capacity, and external pressures imposed by human activities, thereby providing a theoretical basis for vulnerability assessment in coastal systems. The ecological sensitivity dimension incorporated topographic, meteorological, and surface-related indicators. Specifically, topographic factors, including elevation, slope, aspect, and surface roughness, were derived from a digital elevation model (DEM) data to reflect terrain constraints and geomorphological heterogeneity. Meteorological factors included annual precipitation, extreme maximum temperature, and extreme minimum temperature, which were obtained from the WorldClim dataset to characterize climatic variability and extreme events influencing ecosystem stability. Surface factors consisted of land use types reclassified according to national land use standards and patch aggregation degree calculated using landscape pattern indices, capturing the effects of land-use structure and spatial configuration on ecological sensitivity. The ecological resilience dimension focused on vegetation-related indicators that represent ecosystem recovery capacity and biological productivity. These indicators included net primary productivity derived from NASA remote sensing products, a biological richness index constructed based on habitat quality equivalency, vegetation coverage estimated using a statistical quantile method applied to the normalized difference vegetation index (NDVI), and NDVI itself extracted from Landsat imagery. Together, these indicators reflect vegetation conditions, ecosystem functioning, and the ability to recover from disturbances. The ecological pressure dimension quantified anthropogenic stress exerted on ecosystems through socioeconomic activities. Population density data were obtained from the LandScan dataset to represent demographic pressure, GDP data were collected from municipal statistical yearbooks to reflect economic intensity, and nighttime light intensity was derived from simulated NPP-VIIRS data as a proxy for human activity intensity and urban development level. These indicators collectively capture the spatial heterogeneity and temporal evolution of human-induced pressure in Qingdao. All datasets were subjected to rigorous preprocessing procedures to ensure spatial consistency and comparability. Spatial reference systems were unified to WGS_1984_UTM_Zone_51N, and all raster datasets were resampled to a spatial resolution of 30 m. Standardization and normalization were performed according to indicator attributes to eliminate dimensional differences and directional effects. To enhance the robustness and objectivity of indicator weighting, a combined weighting approach was adopted. The subjective Analytic Hierarchy Process (AHP) was used to incorporate expert knowledge and theoretical understanding of ecological vulnerability, while the objective entropy weight method was applied to quantify information contribution based on data variability. Final integrated weights for the sixteen indicators were determined using a Lagrange multiplier-based combinatorial optimization model, achieving a balance between subjective judgment and objective data. The ecological vulnerability index was calculated annually using a linear weighted summation method. The resulting vulnerability values were classified into five levels, namely micro, mild, moderate, severe, and extreme, using the natural breaks classification method in ArcGIS, which minimizes intra-class variance and maximizes inter-class differences. To further explore spatial dependence and heterogeneity, spatial autocorrelation analysis was conducted. Global Moran’s I was employed to assess overall spatial clustering patterns of ecological vulnerability, while local indicators of spatial association (LISA) were used to identify localized clustering characteristics, including hotspot, cold spot, and spatial outlier patterns. The results reveal distinct spatio-temporal evolution characteristics of ecological vulnerability in Qingdao over the study period. From a temporal perspective, ecological vulnerability exhibited a significant nonlinear transformation. Slightly vulnerable areas remained relatively stable, accounting for approximately 5% of the total area, whereas extremely vulnerable areas expanded markedly from 7.62% in 2011 to 11.43% in 2021, with a peak value of 11.69% in 2020. In contrast, moderately vulnerable areas decreased by 4.16%, indicating a clear structural shift from a moderate-dominated vulnerability pattern toward the coexistence and expansion of medium- and high-vulnerability classes. This transformation was closely associated with cumulative urbanization effects, particularly large-scale coastal development initiatives such as the construction of the West Coast New Area and industrial agglomeration along Jiaozhou Bay. Spatially, a persistent coastal-inland gradient of ecological vulnerability was observed. High vulnerability zones were consistently concentrated in central urban districts, densely populated industrial corridors along Jiaozhou Bay, and rapidly developing coastal areas. Low vulnerability zones were mainly distributed in northern hilly regions, including Laixi City, northern Jimo District, and the Laoshan Mountain area, which benefit from higher vegetation coverage, complex terrain, and relatively low human disturbance. Moderately vulnerable areas were primarily located in urban-rural transitional zones, functioning as ecologically sensitive interfaces undergoing rapid land-use conversion and pressure intensification. Spatial autocorrelation analysis demonstrated that Global Moran’s I values exceeded 0.7 throughout the study period and increased from 0.704 in 2011 to 0.761 in 2021, indicating strengthened spatial clustering and polarization of ecological vulnerability. LISA results revealed persistent high-high clusters in core urbanized areas, stable low-low clusters in northern ecological conservation zones, and scattered spatial outliers in transitional and urban fringe areas. Overall, this study elucidates the spatio-temporal patterns, spatial aggregation, and driving mechanisms of ecological vulnerability in a coastal city. The findings provide valuable scientific support for ecological risk prevention, spatial pattern optimization, and differentiated zoning management in coastal urban regions. The results highlight the necessity of integrating ecological conservation objectives with sustainable urban planning, particularly in sensitive transitional zones, and offer practical implications for improving ecosystem governance in urbanizing coastal areas.

Key words: SRP model, AHP-entropy combined weighting, ecological vulnerability, spatio-temporal pattern, spatial autocorrelation analysis