Extreme precipitation events have increased globally in recent years, leading to more frequent and intense urban flooding that seriously threatens public safety. Reliable models for assessing storm waterlogging vulnerability and studies on spatial differentiation are essential for effective disaster prevention and mitigation. However, traditional models often face two main limitations. First, traditional models frequently overlook human adaptive responses to flooding and rely on single-weight calculation methods, reducing the accuracy of their insights. Second, traditional models generally apply to larger scales, such as cities or regions, and fail to capture smaller-scale spatial differences in vulnerability. This study introduces a refined storm waterlogging vulnerability assessment model that includes exposure, sensitivity, and coping capacity. The model allows researchers to reveal the spatial clustering of storm waterlogging vulnerability in more detail, providing more in-depth insights into areas most prone to flooding.

To establish a comprehensive assessment system, this study combined city-specific conditions with human adaptive responses. Nine key indicators, such as annual rainfall, were selected to capture urban-specific vulnerability. Subjective weights were assigned based on an improved expert scoring method, effectively incorporating expert insights. To enhance objectivity, the entropy method was used to calculate objective weights. Then, these subjective and objective weights were combined and optimized using the Nash equilibrium equation to achieve a balanced vulnerability evaluation. Multisource data and ArcGIS software enabled the visualization of storm waterlogging vulnerability on a 1 km grid scale in Xi'an. Global Moran's I and local indicators of spatial association (LISA) score clustering were used to analyze spatial patterns in storm waterlogging vulnerability, revealing clusters and trends across the city. In addition, a vulnerability triangle classified vulnerability levels into eight distinct types, highlighting dominant factors across regions and supporting targeted resilience planning.

The vulnerability assessment showed that areas with high and relatively high vulnerability to urban storm waterlogging mainly clustered in the central old city within the Third Ring Road. This area primarily consisted of various functional zones with hard-paved surfaces, dense construction, intensive development, and high population density. In contrast, areas with low and relatively low vulnerability were mainly clustered in Chang'an, Huyi, Lintong, Yanliang, Zhouzhi, and Lantian Districts, which consisted mostly of forest and agricultural lands, providing high ecological resilience. LISA clustering analysis revealed that storm waterlogging vulnerability had a clear spatial clustering pattern with a significant positive spatial correlation. Eight storm waterlogging vulnerability types are identified: strongly integrated vulnerability (ESC), sensitivity-dominated (S), exposure and sensitivity-dominated (ES), sensitivity and coping capability dominated (SC), coping capability-dominated (C), exposure and coping capability-dominated (EC), weakly integrated vulnerability (O), and exposure-dominated (E) types. The ESC type mainly appeared in the northwest and northern areas of Xi'an; the S and ES types were concentrated within the central urban area; the SC types appeared around the city center; the C types were found in the eastern regions; the EC types occupied much of the southern area of the city; and the O and E types were less common and appeared sporadically in various locations. In lower vulnerability areas, the C types were predominant. In lower to moderate vulnerability areas, the EC types were the most common. In higher vulnerability areas, the SC and ESC types were the most frequent.

The proposed method provides a reliable scientific approach to assessing storm waterlogging vulnerability. It effectively visualizes quantitative data and identifies key factors influencing vulnerability across regions. The findings support the creation of storm waterlogging risk maps and indicate targeted disaster prevention and mitigation strategies.