Entrance/Exit Characteristics-Driven Flood Risk Assessment of Urban Underground Garages Under Extreme Rainfall Scenarios

Abstract

Under the frequent occurrence of urban waterlogging disasters globally, underground spaces, due to their unique environmental conditions and structural vulnerabilities, are facing growing flood pressure, resulting in substantial economic losses that hinder sustainable urban development. This study focused on a high-density urban area in China, investigating surface waterlogging conditions under rainfall characteristics as the primary driver of flooding. Focusing on the main nodes—entrances and exits—within the waterlogging disaster chain of underground garages, a risk assessment framework was constructed that encompasses three key dimensions: the attributes of extreme rainfall, the structural characteristics of entrances/exits, and emergency response capacities. Subsequently, a waterlogging risk assessment was conducted for selected underground garages in the study area under a 100-year return period extreme rainfall scenario. The results revealed that the flood depth at entrances/exits and the structural height of entrances/exits are the primary factors influencing flood risk in urban underground garages. Under this simulation scenario, 37.5% of the entrances and exits exhibited varying degrees of flood risk. The assessment framework and indicator system developed in this study provide valuable insights for flood risk evaluation in underground garage systems and offer decision-makers a more scientific and robust foundation for formulating improvement measures.

1. Introduction

As a world-renowned developing country, China has experienced rapid urbanization and population growth, which has driven an increase in demand for land resource utilization. This trend has led most Chinese cities to exhibit high-density development patterns, especially in central urban areas. To address these pressures, large-scale underground space development has emerged as a strategic response. Official statistics indicate that China′s urban underground space construction area surpassed 2.9 billion square meters by 2023, with development and utilization rates steadily increasing [1]. Ariaratnam et al. [2] reviewed the development and utilization of underground spaces worldwide and found that, against the backdrop of population migration, the demand for underground space development is steadily increasing. Ensuring its sustainable use is therefore of critical importance. Amid global climate change, the intensity and frequency of extreme rainfall have shown an upward trajectory [3], while the interplay of multiple disaster-causing factors has intensified the urban flooding phenomenon. Underground spaces face heightened risk due to their confined geographic environments and structural vulnerabilities, and their inundation poses severe threats to public safety, property security, and urban mobility, as shown in

Table 1. Major urban underground flood events caused by extreme rainfall in recent years [4,5,6,7,8].

Date	Location	Impacts and Consequences
July 2017	Paris, France	Twenty metro stations were temporarily closed
July 2021	Zhengzhou, China	Flooding of Metro Line 5 led to 14 fatalities; more than half of residential underground spaces were inundated
July 2022	New York City, USA	Train station services were suspended
September 2023	Hongkong, China	Four subways were taken out of service
April 2024	Oman and the United Arab Emirates (UAE)	Subway flooding caused a service suspension

. Consequently, developing effective strategies to assess flood risk in urban underground spaces and enhance their resilience has become a critical challenge for sustainable urban governance.

A systematic review of the literature using the keywords such as “Underground Space” and “Waterlogging” revealed that scholars have employed indicator system methodologies to assess flood risks in underground spaces. Liu et al. [9] established a risk assessment model based on the Pressure–State–Response (PSR) framework, selecting Beijing metro stations as the research object. Their study incorporated critical indicators, including the topographic characteristics surrounding station entrances/exits and the frequency of extreme rainfall events. Wang et al. [10] applied Analytic Hierarchy Process (AHP) and Fuzzy AHP (FAHP) methods to evaluate 20 factors—including storm intensity, precipitation frequency, pedestrian density, and regional GDP—in order to develop a comprehensive flood risk assessment framework for Beijing’s subway system.

In addition to the indicator-based assessment methods discussed earlier, scholars have employed rainfall model simulations to evaluate flood risks in underground spaces. This approach is justified as data on waterlogging depth and duration remain critical indicators for risk quantification. Xu et al. [11] applied the InfoWorks ICM model to analyze flood risks at the entrances/exits of four subway stations along Beijing’s Line 11 under varying rainfall scenarios, identifying specific thresholds associated with inundation events. Furthermore, Lyu et al. [12] combined the SWMM model with ArcGIS 10.5 to simulate the inundation depth of the Shanghai metro system under extreme rainfall scenarios.

A review of the existing literature reveals that current flood risk assessments of urban underground spaces predominantly focus on subway systems, while underground garages have received relatively little scholarly attention. Although subway systems and building-attached underground garages share similar hydrological and hydraulic characteristics in regard to flood mechanisms, significant divergences exist in their structural configurations, design specifications, and disaster vulnerability levels. Consequently, findings from previous studies cannot be directly generalized to all underground space typologies. This gap in the literature highlights the need for targeted investigations into the flood risks specific to underground garages, which this study aims to address.

We propose a flood risk assessment framework specifically for underground garages, along with its methodological construction. This study analyzes the underlying mechanisms of flooding and identifies key influencing factors. Our approach advances understanding of the hydrological characteristics and behavioral patterns of underground garages during flood events, providing actionable insights for assessing flood risks in existing urban underground garages and developing targeted disaster prevention and mitigation strategies.

2. Materials and Methods

2.1. Study Area

The study area is situated in the central zone of Yantai Development District, Shandong Province, China (121°13′–121°16′ E, 37°33′–37°34′ N). Bordered by the sea to the north, the region spans 706 hectares and exhibits a highly urbanized landscape with minimal green space. The terrain is predominantly flat, sloping gently from west to east, with elevation ranging from 0 to 40 m above sea level. Rainwater drainage in this area primarily relies on roadside pipeline networks that discharge stormwater directly into the sea. However, the system’s design standards remain suboptimal, characterized by drainage capacities corresponding to ≤1-year return periods. Historical disaster records and reports demonstrate persistent waterlogging hotspots in the study area under extreme rainfall conditions. Through a comprehensive field survey, we systematically selected 32 entrances/exits of underground garages from 16 residential and commercial complexes located in areas with a history of urban flooding. The geospatial distribution and associated information of the study area and selected underground garages are illustrated in Figure 1.

2.2. Research Framework

Extensive empirical evidence from prior studies and documented flooding incidents in underground spaces indicates that surface water intrusion through entrances/exits constitutes a predominant driver of underground inundation [4,13,14], which significantly governs the spatial distribution of flood risks. Therefore, this study established a flood risk assessment framework by identifying underground garage entrances/exits as localized risk nodes. Critically, underground garages with multiple access points necessitate individual risk evaluations for each entrance/exit to ensure granular flood risk mapping.

To comprehensively characterize surface waterlogging conditions at underground garage entrances/exits in the study area, this study integrated hydrological modeling simulation results into the indicator system to identify multi-dimensional flood-inducing mechanisms [15]. The framework aims to systematically assess waterlogging risks for each underground garage through weighted indicator analysis, while elucidating the differential impacts of diverse drivers on inundation dynamics. For urban flood scenario simulations, the MIKE FLOOD model was employed due to its validated performance in urban hydrology studies, particularly thanks to its user-friendly interface and robust hydrodynamic coupling capabilities [16,17]. The research framework is shown in Figure 2.

2.3. Construction of the Assessment Indicator System

2.3.1. Flood Hazard Factors

In urban flooding events, hazard factors primarily stem from the hydrological attributes of rainfall [18]. Underground spaces often exacerbate or trigger waterlogging events under extreme rainfall scenarios; however, there is currently no unified standard for defining the return period of extreme rainfall. Considering that the flood control standard in the study area is established for a 100-year return period, characterized by high rainfall intensity, low frequency, and extensive impacts—features that are commonly recognized as indicative of extreme rainfall—this return period has been selected as the simulation scenario for MIKE FLOOD in this research.

Drainage Network Model

Drainage system data were obtained from the Construction and Transportation Bureau of Yantai Economic and Technological Development Area. After data cleaning using CAD 2018, the topological relationships between nodes and pipelines were established using ArcGIS 10.5. The resulting drainage network, comprising 80 pipes, 69 nodes, and 3 outfalls, was subsequently imported into the MIKE URBAN model. Then the subcatchment areas were divided using Thiessen polygons, and a total of 69 subcatchment areas were delineated. The urban drainage network is shown in Figure 3a.

Meshing Model

The surface elevation data for the study area model were obtained from the ALOS website, and had a resolution of 12.5 m, while land use data were provided by the Construction and Transportation Bureau of Yantai Economic and Technological Development Area. The land use distribution is shown in Figure 3b.

As this study focused solely on simulating surface water accumulation at the entrances/exits of underground garages, the elevation data of the building layer were left unchanged during the construction of the MIKE 21 two-dimensional surface model. Only the elevation of the road layer was reduced by 0.15 m to simulate the water-blocking effect of curbstones.

Designed Rainfall Events

The rainfall intensity formula used in this study is shown in Formula (1). For this simulation, a single-peak Chicago Design Storm boundary condition was generated with the following parameters: return period (p) = 100-year, rainfall duration (t) = 120 min, temporal resolution = 1 min, and peak position coefficient (r) = 0.4.

$$\mathrm{q}={\displaystyle \frac{1619.486(1+0.958\mathit{lgp})}{{(t+11.142)}^{0.698}}}$$

(1)

where q is the design rainstorm intensity, in mm·min⁻¹; p is the design return period, in a; and t is the design rainfall duration, in min.

Model Validation

This study employed the runoff coefficient method to validate the accuracy of the model [19,20]. The runoff coefficient method is one of the most commonly used validation techniques when no measured data are available. The actual comprehensive runoff coefficient of the study area was taken as the target value and compared with the comprehensive runoff coefficient obtained from the model simulation. The degree of agreement between the two was evaluated using the coefficient of variation (C_V). The calculated comprehensive runoff coefficient for the area was 0.68. The model validation was carried out using two simulation scenarios: a conventional rainfall scenario with a 1-year return period, and an extreme rainfall scenario with a 100-year return period. The validation results, shown in

Table 2. Model validation for the runoff coefficient method.

Rainfall Period/a	Rainfall/mm	Model Comprehensive Runoff Coefficient	Comprehensive Runoff Coefficient	CV/%
1	38.89	0.65	0.68	4.41
100	113.40	0.67		1.47

, indicate that the C_V is less than 10% in both scenarios, suggesting that the model has a certain degree of reliability.

Garage Station Generalization

Following field investigations, the locations of the garage entrances and exits were identified in ArcGIS 10.5 based on the satellite image model, in order to analyze the simulation results generated by MIKE FLOOD.

Selection of Indicators

(1) Flood Depth

The depth of waterlogging serves as a crucial indicator of flood severity [21]. Increased surface water depths at underground garage entrances/exits cause an elevated risk of interior inundation and subsequent vehicle damage.

(2) Flood Duration

Flood duration is a critical indicator for evaluating flood risk during extreme rainfall events. Extended flood durations significantly influence residents’ travel willingness and commute efficiency. Furthermore, prolonged inundation may compromise building structural integrity, impair vehicle functionality, and escalate economic losses.

2.3.2. Flood Exposure Factors

The disaster-forming environment of underground-space waterlogging encompasses the integrated conditions of structural design and surrounding topography that govern the initiation, propagation, and exacerbation of inundation-induced losses. This study specifically targeted the entrance/exit infrastructure of underground garages as critical flood risk nodes, with explicit exclusion of internal spatial configurations. Consequently, assessment indicators were selected based on two core dimensions: structural attributes of entrance/exit infrastructure and localized topographic features within their buffer zones.

Entrance/Exit Height

For underground garages, entrances/exits serve as critical interfaces between interior and exterior spaces, with their height acting as a key threshold for floodwater ingress. If the height of entrances/exits is lower than the accumulated water level on adjacent roadways, these structures may become a direct route for backflow inundation. Xiao et al. [22] evaluated flood vulnerability across 109 metro stations in Tianjin’s central district using 14 indicators, revealing that entrance/exit step height constitutes the primary determinant of station-level flood susceptibility. Toda et al. [23] quantitatively investigated step-installation mitigation strategies at Kobe’s underground facilities, demonstrating that elevated entrance/exit thresholds represent an effective engineering intervention for reducing subterranean flood risks.

Entrance/Exit Type

Underground garage entrances/exits are primarily categorized into open and closed structural designs. The key distinction lies in the incorporation of rainwater sheltering facilities (e.g., canopies) above internal passageways in closed entrances/exits, which are absent in open configurations. During rainfall events, open entrances/exits permit direct stormwater intrusion into garage interiors, leading to wetting of the garage surface. Under extreme precipitation scenarios, intensified rainfall exacerbates internal flood risk, particularly when compromised entrance/exit structures fail to mitigate external water influx. Consequently, open-design entrances/exits exhibit significantly higher flood vulnerability. Closed configurations, however, demonstrate enhanced resilience through protective infrastructure, such as shelter canopies and lateral barriers, that effectively intercepts rainwater ingress, thereby substantially reducing subsurface flood hazards [24].

Entrance/Exit-Adjacent Terrain

Surface elevation represents a critical determinant of flood susceptibility, as low-lying terrains exhibit significantly higher runoff accumulation potential compared to elevated zones [25]. Such hydrological convergence not only elevates surface flood risks, but also amplifies the likelihood of water ingress into subterranean structures.

2.3.3. Flood Resilience Factors

Disaster resilience constitutes a pivotal component in flood risk mitigation, with its operational capacity directly governing the magnitude of disaster propagation. The disaster prevention and mitigation capacity of hazard-bearing entities constitutes a critical determinant in urban waterlogging control, as its magnitude governs the spatial propagation of flood hazards. This capacity can be assessed through the availability of flood-resilient material stockpiles and the timeliness of emergency response coordination between municipal authorities and property management entities.

Flood-Resilient Material Stockpiles

The inundation risk of underground spaces is critically dependent on the operational efficacy of flood control infrastructure. These facilities serve as secondary defense mechanisms when external water depths approach the critical threshold of entrance/exit elevations. Flood control facilities can be categorized into two types based on hydraulic performance: water-retention structures [4] (e.g., sandbags, waterproof barriers) and stormwater drainage systems. Water-retention structures partially impede surface flow, though overflow inevitably occurs once water depths surpass the obstruction height. Drainage ditches are common drainage facilities that effectively collect and direct rainwater, with their spatial placement and hydraulic capacity significantly influencing localized flood risks.

Flood Emergency Response Protocol

Incorporating underground garages as critical vulnerabilities into official lists of key flood control targets facilitates coordinated efforts among emergency management agencies, property administrators, and stakeholders to enhance flood defense infrastructure and implement preemptive disaster drills. This proactive approach ensures systematic response execution and orderly flood mitigation during actual disasters, thereby curbing risk escalation—a demonstration of both the strategic value of underground garage protection and the imperative to strengthen early warning systems. The indicators selected in this study, along with their sources and definitions are shown in

Table 3. Indicator system for flood risk assessment of underground garages.

Standardized Layer	Indicator Layer	Define	Source
Hazard	Flood Depth	Depth of water at entrance/exit	Simulation results of MIKE FLOOD
	Flood Duration	Waterlogging time at entrance/exit
Exposure	Entrance/Exit Height	Relative height difference between highest point of entrance/exit and adjacent level ground	Field research
	Entrance/Exit Type	Open or closed
	Entrance/Exit-Adjacent Terrain	Elevation variations within 30 m buffer zone of entrance/exit	ArcGIS 10.5 elevation data analysis
Resilience	Flood-Resilient Material Stockpiles	Completeness of flood control materials	Field research
	Flood Emergency Response Protocol	Designation of key targets for official early warning and flood control	Urban management and law enforcement

.

2.4. Determination of Indicator Weights

To mitigate the influence of subjective biases in weight assignment, this study employed an objective weighting methodology. The Entropy Weight Method (EWM) is a widely adopted objective approach that quantifies weights based on data variability and information entropy, yet it neglects inter-indicator correlations [26]. In contrast, the Criteria Importance Through Intercriteria Correlation (CRITIC) method determines weights by evaluating both contrast intensity and conflict among indicators, explicitly accounting for interdependencies between variables—thereby compensating for EWM’s limitations and enhancing weight accuracy [27]. Consequently, this study integrated EWM and CRITIC into a hybrid objective framework, assigning combined weights through balanced consideration of both methodologies.

Due to the inclusion of both quantitative and qualitative indicators in the assessment metrics, it was challenging to uniformly calculate weights for the data; therefore, risk level classification was performed for each indicator, and values ranging from 1 to 5 were assigned. The specific classification details can be found in

Table 4. Basis for assigning risk to indicators.

Levels/Assignments	I/1	II/2	III/3	IV/4	V/5
Flood Depth/cm	(0, 5]	(5, 15]	(15, 30]	(30, 50]	50
Flood Duration/min	(0, 30]	(30, 60]	(60, 90]	(90, 120]	>120
Entrance/Exit Height/cm	>50	(30, 50]	(15, 30]	(5, 15]	[0, 5]
Entrance/Exit Type	/	Closed	/	Open	/
Entrance/Exit-Adjacent Terrain	Elevation (entrance/exit) = Elevation (buffer max)	/	Elevation (buffer min) < Elevation (entrance/exit) < Elevation (buffer max)	/	Elevation (entrance/exit) = Elevation (buffer min)
Flood-Resilient Material Stockpiles	Sandbags, waterproof barriers, drainage ditches	Sandbags/waterproof barriers, drainage ditches	Drainage ditches	Sandbags /waterproof barriers	No
Flood Emergency Response Protocol	/	Included	/	Not included	/

.

3. Results

3.1. Calculation Results of Indicator Weights

The EWM-CRITIC combined weighting method was applied to calculate indicator weights, with the results summarized in

Table 5. Results of indicator weight calculations.

Assessment Target	Standardized Layer	Indicator Layer	EWM Weights	CRITIC Weights	Combined Weights
Risk Assessment of Flooding in Underground Garages	Hazard	Flood Depth	0.2805	0.1408	0.2107
		Flood Duration	0.2221	0.1026	0.1624
	Exposure	Entrance/Exit Height	0.2669	0.1358	0.2014
		Entrance/Exit Type	0.0663	0.1806	0.1235
		Entrance/Exit-Adjacent Terrain	0.0806	0.1541	0.1174
	Resilience	Flood-Resilient Material Stockpiles	0.0378	0.1252	0.0815
		Flood Emergency Response Protocol	0.0458	0.1609	0.1034

. The three highest-weighted indicators are flood depth (0.2107), entrance/exit height (0.2014), and flood duration (0.1624), identifying these parameters as predominant drivers of flood risk in urban underground garages.

3.2. Risk Metric Characterization

3.2.1. Analysis of Flood Hazards

The flood depth and flood duration that were simulated based on the MIKE FLOOD model are shown in Figure 4. An analysis of the 32 entrance and exit points in the 16 selected underground garages revealed that only 12 points experience surface water accumulation; these are primarily distributed in the central area of the study region, with most exhibiting moderate-to-high risk levels. The area has high development intensity, low green coverage, and predominantly impermeable surfaces, which contribute to significant runoff. Additionally, the regional drainage network has a severe capacity deficit, and the design standards often fail to meet the once-in-a-year occurrence criteria, contributing to the susceptibility of the central section and surrounding areas to flooding.

A comparative analysis of flood depth and flood duration distributions at the underground garage entrances/exits revealed that greater ponding depths generally correlate with longer inundation durations, though no strict positive linear relationship exists. For instance, J3 exhibited the maximum flood depth (109.6 cm), but its flood duration (97.4 min) was not the longest, while K1 demonstrated the longest duration (107 min), despite attaining only a Level II flood depth risk. This nonlinear relationship likely stems from the combined influence of localized drainage capacity, topographic gradients, and spatiotemporal variations in rainfall intensity.

Overall, among the 12 entrances and exits with flood risks, O1, O2, and J3 exhibited higher risk levels in terms of flood depth and duration. Notably, the flood depth at O1 and O2 was classified as having the highest risk status, indicating that the O underground garage poses the most significant danger under this simulated scenario and should be prioritized in flood risk assessment.

3.2.2. Analysis of Flood Exposure

The risk value assignment for the flood exposure indicators is shown in Figure 5. A field survey of the 16 selected underground garages revealed that most entrance and exit constructions have low defensive thresholds, with many being flush or nearly flush with the surrounding ground, allowing surface water to easily flow into the garages. However, regarding entrance and exit types, 78% are closed structures, effectively blocking natural rainfall from directly entering the internal space and reducing the risk to the garage. Additionally, 56% of the entrances and exits are situated at elevations between the maximum and minimum values within the buffer zone, resulting in a flood risk level assignment of 3.

The construction type of entrances/exits exhibits minimal variability; however, significant spatial heterogeneity is observed in entrance/exit height and adjacent terrain. This disparity highlights that entrance/exit flood sensitivity mainly depends on height thresholds and terrain morphology. Table 5 further demonstrates that the entrance/exit height weights (0.2014) significantly exceed those of adjacent-terrain features (0.1174), identifying height thresholds as the paramount determinant of sensitivity to flood exposure.

3.2.3. Analysis of Flood Resilience

Field investigations revealed that underground garages in the study area primarily employ basic flood mitigation measures, including sandbags, waterproof barriers, and drainage ditches at entrances/exits. According to the district’s publicly released flood control list by the Urban Management And Law Enforcement, 13 underground garages affiliated with residential communities in this study area are designated as key flood control priorities. The specifics of the flood resilience capabilities of the underground garage entrances and exits are shown in Figure 6.

3.3. Flood Risk Assessment of Entrances/Exits

In this study, waterlogging risk assessments were conducted exclusively for the 12 entrances/exits exhibiting surface ponding under the simulated rainfall scenario. The remaining 20 entrances/exits were categorically excluded from risk classification due to the absence of observed surface water accumulation under this return-period rainfall intensity. The flood risk values at the entrances and exits were calculated using the following formula:

$$RV=\sum _i^n{W}_i\times R_i$$

(2)

where RV is the flood risk value; n is the number of indicators; W_i is the indicator weights; and R_i is the risk assignment of the indicator.

The flood risk values for each entrance and exit, shown in Figure 7, indicate that the comprehensive risk values for D1, D2, J3, K1, O1, and O2 are relatively high. Specifically, the flood depth risk levels at J3, O1, and O2 are classified as V, significantly exceeding the defensive capacity of their entrance heights. Meanwhile, the flood depth risk results for D1, D2, and K1 are rated as II, while their entrance heights are rated as V, indicating that the minimum flood depth still exceeds their corresponding entrance heights, thus posing a certain risk of water ingress into the garages. Based on this analysis, these six entrances and exits are defined as high-risk points for flooding, as the surface water depth at these locations exceeds the entrance/exit heights, creating a risk of accumulated water entering the garages.

For M2, J1, and J2, the ponding depths are within their respective height’s thresholds, indicating moderate flood exposure, despite localized water accumulation. In particular, J1 and J2 demonstrate enhanced resilience due to functional flood control infrastructure, effectively mitigating risks. In contrast, B1′s open structure allows for direct rainfall ingress, raising its risk ranking, despite its partial height protection.

Although K2 and L1 exhibit minor surface ponding, their depths are lower than their heights, making it difficult for water to flow inside. Among these, L1′s flood control facilities are weaker than K2′s and have not been designated as key flood control targets. Therefore, L1′s emergency response and early-warning capabilities for flooding are relatively weak, which is why its flood risk is slightly greater than that of K2′s.

Based on the above analysis, the entrances/exits of the selected underground garages in the study area were classified into four flood-risk tiers, with their geospatial distribution shown in Figure 8.

3.4. Overall Ingress Risk of Underground Garages

Based on the previous analysis of the risk levels at each entrance and exit, and a comparison of risky entrances and exits to the total number shown in Figure 9, an analysis of the overall ingress risk of the underground garage system was conducted. Both O1 and O2 exhibit high risk with extremely high-risk values. The J garage has J1, J2, and J3 at varying levels of flood risk, with J3 having the highest risk value, classified as high-risk, while J1 and J2 are classified as moderate-risk. Both D1 and D2 are high-risk entrances, but their risk values and water levels are significantly lower than those of O1, O2, and J3. K1 is classified as high-risk, while K2 is classified as low-risk. The M and B garages each have one entrance with moderate risk, with M2 has a slightly higher value than B1. The L garage has only L1 classified as low-risk, while L2 is considered to have no risk. In summary, according to a systematic ranking of the seven flood-prone underground garages in the study area, the severity is as follows: Garage O > J > D > K > M > B > L.

Based on the risk assessment results and the identification of key influencing factors in this study, the research provides a scientific basis for relevant authorities to develop more targeted flood prevention and mitigation strategies, such as improving entrance/exit designs and promoting the development of sponge city initiatives. This can effectively reduce the inundation risks of underground garages and contribute to the realization of sustainable urban development.

4. Discussion

4.1. Strategies for Flood Risk Management

Globally, flood risk management has shifted from traditional flood control to more integrated and sustainable approaches, emphasizing the role of green infrastructure and community participation. Similarly to China’s “Sponge City” concept, these strategies are being implemented in cities worldwide to enhance resilience against urban flooding [28,29].

China pioneered the “Sponge City” initiative in 2013, with its concept formally codified during a 2013 State Council executive meeting [11]. Diverging from conventional gray-infrastructure-dominated approaches to urban flooding, this paradigm integrates green–gray hybrid infrastructure through nature-based solutions (NBSs). By establishing stormwater management systems that leverage natural retention, infiltration, and purification processes, cities can effectively absorb, store, and regulate rainfall runoff, mitigating urban waterlogging through controlled delayed discharge. This approach has emerged as China’s primary strategy for addressing urban drainage challenges and enhancing climate resilience [30].

Therefore, this study proposes a retrofitting strategy to develop “sponge-garage” systems for underground facilities requiring urgent flood mitigation measures to protect against the impacts of surface runoff [31]. Implementation of cost-effective green infrastructure, such as rain gardens and sunken greenbelts, around retrofittable entrances/exits, combined with micro-topographic adjustments (e.g., slope optimization), can effectively redirect surface runoff into sponge infrastructure, thereby alleviating entrance/exit ponding.

Entrances and exits that lack construction measures can be equipped with automatic water barriers, which are activated to provide emergency flood control when the water level reaches a certain warning height. Additionally, enhancing the flood response capabilities of underground parking garages can be achieved by appropriately raising the construction height of the entrances and exits, which improves the flood control devices at these points, and by installing canopies to block rain.

For low-risk entrances and exits, it is recommended to improve flood control facilities and conduct regular flood response drills to enhance the emergency response and early-warning capabilities of the underground parking garages, thereby reducing the flood risk at optimal cost.

These solutions are scalable, allowing them to be adapted for various urban densities and environmental contexts, making them applicable in both newly constructed and retrofitted areas. Furthermore, their integration into existing urban infrastructure could be achieved with minimal disruption, reducing the cost and time of implementation.

4.2. Limitations of Research Perspectives

This study focused exclusively on external waterlogging risk assessment centered on underground garage entrances/exits. While the methodology successfully identified garages susceptible to surface water ingress, the hydrodynamic impacts of internal water propagation post-ingress remain unquantified. Consequently, the current study framework has intrinsic limitations in assessing the systemic flood risks of underground garages. Future research should integrate external and internal spatial interactions of underground garages with the complete disaster progression pathways of underground space waterlogging.

The authors posit that the waterlogging risks within the internal spaces of underground garages ought to be categorized according to the danger levels imposed on vehicles and individuals by the depth of water accumulation. Comprehensively considering surface water inflow volume, internal-space dimensions, and internal drainage capacity will contribute to a more accurate assessment of waterlogging risk in underground garages.

5. Conclusions

This study established a flood risk assessment framework for urban underground garages, anchored in the primary inundation pathway—entrances and exits—through a tripartite index system encompassing hazard, exposure, and resilience. The framework was applied to evaluate waterlogging risks across 32 entrances/exits of 16 underground garages within the study area under a 100-year-return-period extreme rainfall scenario. Diverging from prior methodologies, this study innovatively integrated MIKE FLOOD urban hydrodynamic modeling outputs into the indicator system, explicitly accounting for multi-factor interactions among urban underlying surface, terrain, and drainage network capacity in modulating extreme rainfall impacts. Compared to merely considering rainfall attributes such as storm intensity and annual average precipitation, this approach, which better reflects actual urban surface waterlogging phenomena, yields more accurate risk assessment results. The key findings are summarized as follows:

• The hybrid EWM-CRITIC combination weighting method was employed to calculate composite indicators, effectively mitigating subjective bias in risk quantification. The results of weight analysis demonstrated that surface flood depth at the entrance/exit (0.2107) and entrance/exit height (0.2014) emerged as dominant determinants of the flood vulnerability of underground garages.
• Under the 100-year-return-period extreme rainfall scenario, the 32 entrances/exits of the 16 underground garages exhibited the following risk stratification: 6 were high-risk, 4 were medium-risk, 2 were low-risk, and 20 had no risk. The seven flood-prone underground garages were hierarchized as follows: Garage O > J > D > K > M > B > L.
• This study elucidated the primary causal factors of underground garage inundation risks and proposed targeted mitigation strategies.