Street Community-Level Urban Flood Risk Assessment Based on Numerical Simulation

Abstract

Urban waterlogging is a serious urban disaster, which brings huge losses to the social economy and environment of the city. As an important means of urban rainfall inundation analysis, numerical simulation plays an important role in promoting the risk assessment of urban waterlogging. Scientific and accurate assessment of waterlogging disaster losses is of scientific significance for the formulation of disaster prevention and mitigation measures and the guidance of post-disaster recovery and reconstruction. In this study, the SCS-CN hydrological model and GIS coupling numerical simulation method were used to simulate the inundation of urban waterlogging under four different rainfall return periods and to realize the visualization of the inundation range and waterlogging depth in Zhengzhou. At the same time, based on the numerical simulation results, the building is used as the basic assessment unit to construct a refined assessment framework for urban waterlogging risk at the street community level based on hazard, exposure, and vulnerability analysis. The refined risk assessment results have an important reference value for optimizing the working ideas of waterlogging control and providing a reference for local management departments to effectively deal with waterlogging disasters.

1. Introduction

The vigorous development of China’s economy has significantly accelerated China’s urbanization process. The rapid development of urbanization has promoted the steady progress of China’s basic livelihood issues such as population, economy, education, and medical care, and has also brought great challenges to the safety management of cities [1]. Among the problems of urban safety management, the harm and impact of natural disasters on cities are the most lasting and huge [2]. Among the natural disaster events, urban waterlogging disasters are the most frequent, causing the most serious loss of personnel and economic property to the residents of the city.

Urban waterlogging disaster refers to the phenomenon of regional waterlogging caused by short-term heavy precipitation or continuous precipitation exceeding the urban drainage capacity and affecting low-lying terrain in cities [3]. Since the 1950s, with the intensification of global climate change and the rapid expansion of urbanization, the hydrological and climatic characteristics of the regional scale have changed, resulting in the abrupt occurrence of extreme climatic conditions such as extremely heavy rainfalls all over the world. Affected by the heat island effect, a large area of waterlogging disaster in the city has been induced, and the frequency and intensity of this waterlogging disaster have been increasing, and the scope of influence has gradually expanded [4].

Urban waterlogging risk assessment methods mainly include four kinds: historical disaster mathematical statistics method, scenario simulation evaluation method, remote sensing and GIS coupling method, multi-criteria index system method [5]. Among them, the historical disaster statistics method generally predicts the frequency of, depth of, and direct loss from flood disasters in different spatial locations through the statistics of historical flood disaster data released by the government or relevant departments [6]. Some researchers use historical rainstorm and waterlogging disaster data to construct vulnerability curves in the study area and conduct quantitative or qualitative waterlogging risk analysis [7]. On the basis of historical disasters, the spatial transformation and damage from the waterlogging risk can be assessed in more detail by combining land use type data [8] and hydrological models [9]. For example, in order to improve the risk mitigation policy for waterlogging disasters and further improve the public’s awareness of risk prevention, Denis et al. conducted a visual analysis of the historical disaster information obtained and, also, relevant research on waterlogging risk management [10]. The idea of the mathematical statistics method to predict historical disasters is relatively clear, the calculation process is relatively simple, and the risk assessment results obtained are very consistent with the actual disaster results because the original data are derived from the historical disaster information. However, the use of this method also has certain limitations because it has a strong dependence on the historical data of the study area and requires the use of relatively fine and rich data [11], and the quality of data spatial accuracy directly affects the accuracy of the evaluation results. In general, this method is only suitable for large-scale spatial areas such as cities and provinces [12] for a macroscopic risk assessment.

A remote sensing and GIS coupling method uses remote sensing technology to obtain information such as the waterlogged area and inundation duration in the disaster area [13]. Cui Chen used remote sensing images from Xinxiang City, Henan Province, to study the dynamic information of the water accumulation range and water accumulation time in the waterlogging area of the Zhaogu No. 2 Mine from 2014 to 2020 in the GIS environment and carried out the risk assessment of the dynamic changes [14]. Cheng Xianfu et al. selected different indicators from three aspects: disaster-causing factors, disaster-pregnant environment, and disaster-bearing bodies to establish a waterlogging disaster risk assessment system in the Chaohu Lake Basin and comprehensively evaluated the waterlogging disaster risk by combining scenarios from various indicators through the scenario analysis technology [15]. Jiang focused on the risk and vulnerability of Zhejiang Province, selected suitable index factors, used geographic information processing technology to spatialize each factor, employed an analytic hierarchy process to determine the weight of each index, and obtained a waterlogging disaster risk distribution map [16].

Multi-criteria indicator system method (MCSM) is a widely used method of flood risk assessment, which has low data requirements, is economically feasible and can be applied flexibly according to the characteristics of the study area and the data availability. Most researchers select different index factors, such as elevation, rainfall, GDP, and road density to construct the index system; use a principal component analysis to determine the weights of each index [17]; and conduct a risk assessment of the study area [18]. For the first time, Gilbert F. White took the people’s response to waterlogging as one of the indicators in the risk assessment of waterlogging disasters and included the characteristics of the population distribution and the building structure of houses into the evaluation system. San selected 13 indicators, including elevation, slope, rainstorm rainfall, and rainstorm proportion, constructed a multi-index system model of waterlogging based on the four-factor theory, and comprehensively evaluated the risk of waterlogging disasters. The determination of index weights is an important research issue when using the multi-criteria indicator system method for risk assessment, so researchers will use different methods, such as subjective and objective weighting, to determine the weights of the selected finger parameters [5]. For example, Cabrera uses the secondary analysis method and the maximum entropy model method to assign weights to the indicators [19]. This method can be used to select indicators according to the characteristics of the study area and the availability of data; a flexible use of the quantitative method for evaluation can also intuitively reflect the family system between each indicator and waterlogging risk [20], which is suitable for regions with different spatial scales—the difficulty lies in the selection of indicators and the determination of weights [21].

The scenario simulation assessment method can dynamically simulate and evaluate the disaster process in the study area and provide the accurate spatial distribution characteristics of urban flood risk [22]. The one-dimensional pipe network hydraulic model is the first widely used numerical model because the required data are relatively simple and the model results are more accurate—for example, the SWMM model and SIPSON model [23]. With the advancement of research, some researchers have coupled one-dimensional land surface and river channel models and two-dimensional land surface models and proposed a two-dimensional waterlogging simulation model based on empirical formulas and artificial intelligence, which can accurately simulate the evolution of surface waterlogging during the whole rainfall process—examples include commercial models such as MIKE and Info Works ICM [24]. Huang used Info Works ICM software to construct a stormwater model in the Donghao Chung River Basin of Guangzhou and combined it with eight indicators to form an index system to evaluate the waterlogging risk in the study area under the 1-year, 5-year, and 50-year rainstorm scenarios, respectively. The scenario simulation method can intuitively reflect the degree and scope of a disaster in a certain area, but it has high requirements for the geographic data of the study area, and the model is complex and not universal.

As we all know, once the city suffers from a waterlogging disaster, buildings are the most direct disaster-bearing bodies and they are the most vulnerable to economic losses. As the basic unit of disaster risk management, communities play a key role in flood control and disaster reduction. Disaster risk analysis at the street community scale is a basic activity for optimizing risk management, which can provide a basis for the community to take adaptation measures and build resilient communities. However, most researchers focus on the district and county scale, and there are not many studies on the refined risk assessment of the street community level. Mei Chao proposed a multi-scale evaluation framework of ‘city-block-facilities’ and took Beijing as an example for analysis [25]. Jiao Sheng proposed a multi-scale evaluation framework of ‘main pipe-branch pipe-site’ and evaluated it with Changsha as an example [26]. However, their research has not been combined with the inundation depth of waterlogging. In order to make a more refined risk assessment of community-level areas, this study conducted a more detailed comprehensive assessment based on land use type, building distribution, and inundation depth results of numerical simulation. Recent research has further expanded the understanding of urban waterlogging risk. For example, the empirical study of Shenzhen shows that the vertical urban form has a significant impact on the scale effect of waterlogging, and the spatial distribution of waterlogging risk can be better understood by assessing the inundation conditions under different building heights and densities [27]. In addition, a study in the Alaphuzha district of Kerala, South India, explored the impact of the built environment on flood resilience and proposed flood resilience building techniques suitable for local geography, climate, and socio-economic conditions [28].

Taking Zhengzhou City as an example, this paper uses the numerical simulation method to simulate the rainfall of four different recurrence periods, and based on the simulation results, it carries out a study on the risk assessment of a rainstorm waterlogging disaster at the street community scale. First of all, the local equal volume method is used to simulate inundation, which is more accurate than the traditional method. By keeping the volume invariance of a water body, the water flow diffusion and water depth during waterlogging are simulated more accurately, and the reliability of the simulation results is improved. Second, this study refines the risk assessment to the level of the street community, using a smaller and more detailed assessment unit, which can more accurately reflect the risk level of different areas within the community and provide a scientific basis for identifying high-risk areas and formulating accurate flood prevention and mitigation measures. In addition, this study takes into account the vulnerability and potential economic losses of buildings, not only focusing on the direct impact of waterlogging on buildings but also assessing economic losses, thus providing more comprehensive risk assessment results. The research results can provide a basis for community flood risk assessment and rescue decision-making, and the comprehensive analysis method can also provide a reference for community disaster risk analysis.

2. Materials and Methods

2.1. The Study Area

Zhengzhou is located in Central China, in the lower reaches of the Yellow River and the hinterland of the Central Plains, and is the capital city of Henan Province. In terms of its geographical location, Zhengzhou is located between 112°42′ E and 114°13′ E, and between 34°16′ N and 34°58′ N. The overall terrain is high in the southwest and low in the northeast, with a stepwise decline. From the western and southwestern tectonic erosion in the low mountains, the gradual decline transitions to tectonic erosion hills, loess hills, tilted plains, and alluvial plains, forming a relatively complete geomorphological sequence.

Zhengzhou belongs to the north temperate continental monsoon climate. The four seasons are distinct. It is dry and rainless in spring, hot and rainy in summer, sunny with long periods of sunshine in autumn, and cold with little snow in winter. Zhengzhou City has the longest winter, followed by summer, and spring is shorter, with an annual average of 14.3 °C. The average annual rainfall in Zhengzhou is 640.9 mm, and the highest rainfall in 2016 was 833 mm.

On 20 July 2021, a heavy rainstorm disaster occurred in Zhengzhou, Henan Province. The extreme rainstorm caused serious urban waterlogging, river floods, flash floods, landslides, and other disasters, resulting in particularly heavy casualties and property losses. More than 290 people were killed, and the direct economic losses reached CNY 53.2 billion.

This paper mainly focuses on the numerical simulation of rainstorm waterlogging under different return periods in 9 districts and counties of Jinshui District, Huiji District, Zhongyuan District, Guancheng Hui District, Erqi District, Shangjie District, Xingyang City, Xinzheng City, and Zhongmu County to the east of Zhengzhou City. The schematic diagram is shown in Figure 1.

At the same time, University Road Street in Erqi District is selected as the research area of the community residential level. The schematic diagram of the research area is shown in Figure 2. The study area extends from Central Plains East Road in the north, Longhai Expressway in the south, Songshan South Road in the west, and University North Road in the east. It mainly includes 11 communities, such as Xianghe Community, Central Plains Community, Songshan Community, and Kangqiao Huacheng Community, with an area of 2.7 square kilometers. There are many types of buildings in the area, such as residential areas, parks, schools, hospitals, and enterprises. The types of buildings are diverse and very representative.

2.2. Data Source

The original data used in this study mainly include remote sensing image data, soil data, land use data [29] and building height. The time and source of the data are shown in

Table 1. Data sources.

ID	Data Name	Date	Sources
1	Soil data	2022	Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences
2	Land use data	2020	Global coverage products of GLC_FCS30-2020 fine classification system
3	Remote sensing image data	2023	U.S. Geological Survey USGS
4	Building height	2020	Institute of Geography and Resources, Chinese Academy of Sciences

.

The original data used in this study mainly include four types of data: remote sensing image data, soil data, land use data, and building height data. The specific source, time, and acquisition method of the data are as follows. The soil data are from the Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences, with a spatial resolution of 30 m, and the acquisition time was 2022. These soil data are used to analyze the effects of different soil types on waterlogging. We cleaned the data, removed outliers, and performed coordinate matching and projection transformations to ensure consistency with other data. Land use data, were obtained in 2020 using the GLC_FCS302020 fine classification system for global coverage products. The spatial resolution of the data is 30 m, and the overall accuracy of the main land cover types is 80.88% (±0.27%). Land use data are derived from the global fine classification system and provide detailed land use types within the study area. The data on the land use type in the study area were visually processed, and the study area was divided into 7 types of land use: impervious surface, forest land, grassland, cultivated land, water area, wetland, and bare land. Remote sensing image data from the United States Geological Survey (USGS), obtained in 2023, have a spatial resolution of 30 m. Building height and terrain information are extracted from remote sensing image data. We processed the image of University Road Street in the study area and corrected it according to Google Maps and Gaode Maps. The building height data were obtained from the Institute of Geographic Sciences and Natural Resources Research of the Chinese Academy of Sciences in 2020, with a spatial resolution of 20 m. The building height data were used to calculate the actual inundation depth of the building. In order to ensure the accuracy of the data, we carried out local corrections according to the latest remote sensing images. The data preprocessing steps include the following: First, all the original data collected are cleaned to remove outliers and missing values to ensure the integrity and accuracy of the data. Then, coordinate matching and projection transformation are carried out on the data from different sources to ensure the spatial consistency of all data. Next, the data of different units and dimensions are normalized for subsequent analysis and calculation. Finally, the spatial superposition and height correction of building height data are carried out to ensure that they match the waterlogging model.

During the data processing and model construction, we made the following reasonable assumptions: It is assumed that the rainfall scenarios of once in twenty years and once in a hundred years occur independently and are not affected by other factors. It is assumed that the height above the ground of the same type of buildings is relatively uniform within the study area, with little difference. It is assumed that within a certain range, there is a linear relationship between the inundation depth and the economic loss per unit area. Through the above explanations of the data sources, sampling methods, data preprocessing steps, and the rationality of the assumptions, it is possible to better understand and analyze the economic loss situation of different land use types within the study area under waterlogging conditions.

3. Methods

3.1. Study Framework

The method framework proposed in this paper is shown in Figure 3, which is mainly divided into two stages: (1) numerical simulation of rainfall waterlogging based on SCS-CN (Soil Conservation Service—Curve Number) and GIS (Geographic Information System) and (2) street community-level refined urban waterlogging risk assessment. The numerical simulation mainly includes two processes: rainfall simulation and runoff–submergence simulation. The risk assessment establishes three assessment models: hazard, exposure, and vulnerability.

3.2. Rainfall Simulation

3.2.1. Rainstorm Intensity Formula

In the study of waterlogging disasters caused by urban rainstorms, measured rainstorm data or designed rainstorm models are often used for simulation [30]. Design rainfall refers to the local rainfall that may occur in accordance with the design standards for the planning and design of water conservancy projects or flood control and disaster reduction. Due to the lack of long-term rainstorm data, this paper has selected the design rainstorm model to study the rainfall process of different rainfall return periods.

The rainstorm intensity formula, which can objectively reflect the characteristics and laws of urban rainfall, is the basis of design rainfall deduction and has important value for urban drainage design planning and urban waterlogging control [31]. According to the ‘Code for Design of Outdoor Drainage’, the definition of rainstorm intensity formula is shown in Equation (1):

$$q={\displaystyle \frac{167A_1\left(1+ClgP\right)}{{\left(t+b\right)}^n}}$$

(1)

In the formula, q is the rainstorm intensity [unit: L/(hm²)]; t is the rainfall duration [unit: min]; P is the return period [unit: a]; A₁ is the rainfall parameter [unit: mm]; C is the rainfall variation parameter; b is the rainfall duration correction parameter [unit: min]; and n is the rainstorm attenuation parameter.

By consulting the data, Zhengzhou Natural Resources and Planning Bureau revised the design rainstorm intensity formula and design rain pattern standard of Zhengzhou City on 1 February 2023 and gave the calculation formula of rainstorm intensity in different districts and counties of Zhengzhou City. Among them, Shangjie District, Xingyang District, Zhongyuan District, Huiji District, Jinshui District, Guancheng Hui District, Erqi District, Xinzheng District, and Zhongmu County are 9 districts and counties. The determination of parameters is as follows: C = 3.264; b = 24.8; n = 0.856; A₁ = 11.987. Therefore, the rainstorm intensity formula is as shown in Formula (2)

$$q={\displaystyle \frac{2001.829\left(1+3.264lgP\right)}{{\left(t+24.8\right)}^{0.856}}}$$

(2)

3.2.2. Chicago Rainfall Hydrograph Method

The “Technical Specification for Urban Waterlogging Prevention and Control” (GB51222-2017) stipulates that the design rain pattern refers to the process of reflecting the change of rainfall over time in typical rainfall events [32]. In the 1960s, Kefir, a scholar, found that there was a correlation between rainfall intensity, rainfall duration, and frequency in the study of urban pipe network system. According to this relationship, the Chicago rainfall hydrograph model was proposed, which is widely used abroad [33]. In 1998, Chinese scholar Cen Guoping used the fuzzy recognition method to study the Chicago rainfall process line. The analysis shows that the Chicago rainfall process line model is in line with the characteristics of China’s heavy rain, and can simulate the heavy rain process in different cities in China.

The Chicago rain pattern design formula is shown in Formulas (3) and (4):

$$i\left(t_b\right)={\displaystyle \frac{A\left(1+c\lg P\right)\left[\frac{\left(1-n\right)t_b}{r}+b\right]}{{\left[\frac{t_b}{r}+b\right]}^{n+1}}}$$

(3)

$$i\left(t_a\right)={\displaystyle \frac{A\left(1+c\lg P\right)\left[\frac{\left(1-n\right)t_a}{\left(1-r\right)}+b\right]}{{\left[\frac{t_b}{1-r}+b\right]}^{n+1}}}$$

(4)

In the formula: i is the instantaneous rainfall intensity (mm/min); t_b is the pre-peak time (min); t_a is the post-peak time (min); A, c, b, n are the parameters of rainstorm intensity formula; and r is the rain peak position coefficient.

In this study, the rainfall duration was set to 120 min. According to the notice issued by the Natural Resources and Planning Bureau of Zhengzhou City, the rain peak position coefficient r was 0.433.

3.3. Runoff Generation and Submerged Simulation

3.3.1. Runoff Model Based on SCS-CN

The SCS model is an empirical model constructed by the United States Department of Agriculture to complete the calibration parameters based on a large number of experimental data in the northern United States. When performing model calculations, it is necessary to query the list based on the number of runoff curves (CN) in the model [34]. The SCS-CN model has a simple structure, few parameters, and high precision and is suitable for rainfall runoff estimation in small and medium-sized basins or regions. The model only needs a few input parameters, such as rainfall, land use type, and soil water status, to effectively estimate rainfall runoff. Therefore, the model is widely used in the prediction of rainfall surface runoff-producing area processes at different time scales. The model shows good applicability and accuracy in different geographical regions and climatic conditions, especially in areas lacking detailed hydrological data. The SCS-CN model can reflect the influence of an underlying surface type and soil type on rainfall runoff, and it is a better method for runoff calculation in the catchment [35].

The basic rainfall–runoff relationship of the SCS-CN model is:

$${\displaystyle \frac{F}{S}}={\displaystyle \frac{Q}{P-I_a}}$$

(5)

In the formula, P is the total amount of rainfall (mm); Q is runoff (mm); I_a is the initial loss of rainfall (mm); F is the late rainfall loss (mm); S is the possible maximum retention amount (mm) of the region at that time; and P = Q + F + I_a.

Formula (5) is an empirical relationship derived from the analysis and summary of measured data, and there is no strict theoretical explanation. However, since it comes from the actual data and represents the law of nature itself, a large number of application results prove the rationality of Formula (5).

According to the principle of water balance, rainfall is equal to the sum of runoff and the initial loss value of rainfall and the later loss. If the rainfall does not reach the initial loss, runoff will not be generated.

$$Q={\displaystyle \frac{{\left(P-I_a\right)}^2}{P+S-I_a}},P>I_a$$

(6)

In the process of runoff generation, the initial loss I_a is not easy to obtain. The empirical relationship I_a = 0.2S is introduced, and the final runoff generation formula is:

$$Q={\displaystyle \frac{{\left(P-0.2S\right)}^2}{P+0.8S}},P>0.2S$$

(7)

In Equation (7), the variation range of the possible maximum retention S value in the region may be very large, and it is not convenient to take the value. In order to solve this problem, the model maker introduces a dimensionless parameter CN, called curve number, and stipulates the following relationship:

$$S={\displaystyle \frac{25400}{CN}}-254$$

(8)

The CN value is a comprehensive parameter reflecting the pre-rainfall characteristics of the region, which is closely related to the characteristics of the underlying surface, land use type, and soil type [36]. The value is generally an integer within one hundred, indicating the size of the infiltration capacity. The greater the CN value, the weaker the infiltration capacity, and the greater the S value. In the SCS-CN model, there are four groups of soil types with different runoff capacities, and the CN values corresponding to different land use types in each soil type are different.

A soil type is an important factor affecting runoff and waterlogging when carrying out flood risk assessment. Different types of soil have different water absorption rates and infiltration capacities, which, in turn, affect the amount of runoff and the depth of water accumulation after the rainfall. In this study, we used the SCS-CN model to consider the effect of soil type on runoff. According to different types of soil (e.g., clay, sand, loam, etc.), we set different curve values (CN values) to reflect the water absorption and infiltration characteristics of different soil types, respectively. Specifically, the soil types and corresponding CN values we used in the model are shown in the table below.

Because the model is based on the data set in the United States, some parameters are not very consistent when used in other parts of the world, and the direct application of fixed parameters will cause poor simulation accuracy. Therefore, based on the CN value provided by the U.S. Department of Agriculture and the research experience of other researchers, this paper calibrated the CN value in the study area and determined the corresponding relationship of CN value in the study area as shown in

Table 2. Comparison table of land use type and hydrological soil group.

ID	Land Use Types	Hydrological Soil Group
		A	B	C	D
1	Impervious surface	89	90	91	93
2	Waters	98	98	98	98
3	Forest land	30	55	70	77
4	Cultivated land	64	75	82	85
5	Grassland	39	61	74	80
6	Wetland	62	71	78	81
7	Bare ground	39	61	74	80

:

According to the soil classification standards of the Soil and Water Conservation Service of the United States Department of Agriculture, similar flow-producing soils are classified into hydrologic soil groups, and hydrologic soil types are divided into four categories—A, B, C, and D—according to the difference in permeability and water conductivity. The detailed relationship between soil hydrological types and soil types is shown in

Table 3. Comparison table of hydrological soil classification.

Hydrosoil Type	Soil Texture	Soil Type
A	Sandy soil	Meadow wind-sand soil
	Gravelly soil	Coarse soil
B	Silt loam	Tidal soil, detidal soil, yellow spongy soil, alluvial soil
	Sandy loam soil	Alkaline tidal soil
C	Clay loam	Brown soil and calcareous brown soil
D	Silty clay loam	Sandy clay

.

By incorporating soil types into the model, we are able to more accurately simulate flooding in different areas and assess the impact of different soil types on flooding risk. This addition helps improve the accuracy and usefulness of the model and provides a more scientific basis for urban flood control.

3.3.2. A GIS-Based Local Isovolumetric Method

The process of flood submergence is from the beginning of the flow spread to the process of balanced rise [37]. Considering that the study area in this paper is flat and there are no more depressions, passive inundation is selected for inundation calculation. The basic principle is that the total amount of runoff at a certain time step is equal to the total amount of inundation at that time step. The calculation formula is as follows (9):

$$W=\underset{A}{\iint }\left[h_{\mathrm{a}}\left(x,y\right)-h_g\left(x,y\right)d\delta \right]$$

(9)

In the formula, W represents the total amount of water in the flooded area; A represents the area of the flooded area; $h_{\mathrm{a}}\left(x,y\right)$ is the water surface elevation of the water area; $h_g(x,y)$ represents the ground elevation; and $d\delta$ denotes the unit of submerged area.

In fact, the water surface of rainstorm waterlogging is complex, and the water surface with larger flow velocity is more complex. Due to the slow speed and relative simplicity of urban rainstorm water accumulation, the formed water surface can be approximated as a plane, and the ground elevation can be expressed by terrain grid DEM data. Formula (9) can be simplified into Formula (10):

$$W={\sum }_{i=1}^N\left[h_{\mathrm{a}}-h_g\left(x,y\right)\right]\mathsf{\Delta }\delta$$

(10)

In the formula, $h_{\mathrm{a}}$ is the elevation of the water surface of the whole basin. By dividing the water area into multiple grid values, this method is used to solve the problem in each grid. Therefore, Equation (10) can be further simplified as:

$$W={\sum }_{i=1}^N\left[h_{\mathrm{a}}-h_g\left(i\right)\right]\mathsf{\Delta }\delta$$

(11)

In the formula, $\mathsf{\Delta }\delta$ is the area of the grid unit; N is the total number of grids in the submerged area, and $h_g\left(i\right)$ is the elevation value of the ith grid unit. The smaller the area of the grid unit is, the closer the result is to the calculation result of the above formula. At this time, the unknowns are N and $h_{\mathrm{a}}$, and the N value can be obtained by the relationship between $h_g$ and $h_{\mathrm{a}}$.

In the submerged simulation, the equal volume method is an important analysis method. It is based on the principle of volume conservation, that is, the total water volume in the submerged area remains unchanged during the simulation process. It is mainly used to simulate the scope and degree of waterlogging. The process of the equal volume method is shown in Figure 4.

The determination of the water elevation is mainly based on the idea of dichotomy. The operation steps to determine the water elevation are as follows [38]:

(1)

Determine the value range of water surface elevation (H_min, H_max): Hmin is not lower than the minimum value of ground elevation, and H_max is lower than the maximum value of ground elevation.

(2)

Set the initial value of elevation as the average value of the minimum and maximum values of the water accumulation elevation at the beginning of h—that is, the beginning of $h_{\mathrm{a}}$ = h₀ = (H_min + H_max)/2, substitute the initial value into the formula and sum up the grid of each small catchment area. If $h_{\mathrm{a}}-h_g\left(i\right)>0$, it means that when $h_{\mathrm{a}}$ = $h_0$, the grid has accumulated water and W increases $[h_{\mathrm{a}}-h_g\left(i\right)]\mathsf{\Delta }\delta$; and $h_{\mathrm{a}}-h_g\left(i\right)\le 0$, indicating that in the grid without water, the W value does not increase. After completing the calculation of all the grids in the water accumulation area, obtain the value of the total water volume W in the water accumulation submerged area in the case of $h_{\mathrm{a}}$ = $h_0$.

(3)

Compare the total water volume W with the total water volume V simulated during the runoff generation process: if V − W > 0, the assumed water surface elevation is low, and the water surface elevation needs to be reset. At this time, in H_min = $h_0$, calculate the initial value of h and substitute it into the formula to continue to calculate the total water volume W; if V − W < 0, it indicates that the set water surface elevation is high, and the water surface elevation needs to be reset. At this time, H_max = $h_0$. At this time, calculate the initial h and substitute it into the formula to continue to calculate the total water volume W value.

(4)

Repeat Step (3), gradually making the values of W and V approach infinity until their difference is within the allowable error range. The elevation at this point is the water surface elevation.

(5)

In each catchment area, calculate the water surface elevation according to this method. Finally, visualize the water surface elevation of several small catchment areas to obtain the water area and water depth of the whole study area.

In the study of floods in rivers and other basins, it is very common to use the equal volume method. However, the research scope of this paper is limited to the city. The structure of urban topography, geomorphology, and underlying surface is much more complex than that of topographic natural basins. Various buildings in the city will present certain obstacles to the flow of water.

Considering the particularity of the city, this paper proposes a local equal volume method based on GIS to simulate urban waterlogging. According to the topography of the study area, the whole study area is divided into 93 sub-catchment areas. The division of sub-catchment areas is shown in Figure 5. At this time, each sub-catchment area is considered as a whole study, and the equal volume method will be transformed into a local equal volume method. The inundation simulation of 93 sub-catchment areas, respectively, was carried out, and the inundation status of urban areas was summarized.

3.4. Construction of Depth Damage Curve

The depth damage curve, also known as the vulnerability curve, is usually expressed by the loss or loss rate of different water depths. It is mostly used to evaluate the inundation loss of houses or other construction institutions [39], indicating the relationship between disaster intensity and loss. The establishment of an urban waterlogging vulnerability curve mainly includes two steps [40]: one is to establish the value curve of different buildings according to the statistical data of indoor property value, building structure type, and building material repair cost of different buildings; the second is to collect the inundation depth of waterlogging disasters in the study area and combine the value curve with the inundation depth to construct the depth–loss curve.

The purpose of this paper is to evaluate the loss of the community property caused by waterlogging. Due to the different types of buildings on different land types, their property values are different, so it is important to carry out the evaluation work more carefully. According to the GDP and living standards of residents in Zhengzhou City, referring to other relevant literature [41,42], the deep disaster loss curve of the study area (Figure 6) is determined, which mainly includes nine types of residential land, two or three types of residential land, four types of residential land, service activity land, entertainment land, industrial land, science and education land, commercial land and other established areas. We further divided the study area into a nine-category land use grid, and the results are shown in Figure 7.

Analyzing Figure 6 yields the following conclusions:

(1)

Differences in economic losses for different residential site types

Residential land use Type 1 and residential land use Types 2 and 3 generate higher economic losses, even at shallower inundation depths. This is due to the availability of municipal utilities, better environment, and higher value of buildings in these areas. Residential sites in Type 4, though susceptible to inundation, are of lower value and, hence, incur relatively less economic loss.

(2)

Land for services, recreation, industry, science and education, commerce, and other uses

Losses of land for service activities and recreational land are significant at greater depths of inundation, and the value of facilities and equipment in these areas is high. The loss curves for industrial and science and education land uses show that the losses in these areas are more variable at different inundation depths, possibly due to the high value of their equipment and facilities and the additional losses associated with the disruption of their production and research activities.

Commercial sites incur significant losses at relatively shallow inundation depths, reflecting the high sensitivity of commercial activities to inundation. Other established district sites have relatively even loss profiles across the range of inundation depths, but their combined losses still need to be emphasized. The table in Figure 6 shows the differences in economic losses from inundation for different land use types, which will help subsequent studies develop targeted flood prevention and mitigation measures. Particularly in high-value residential and commercial areas, higher standards of flood protection and contingency planning are needed.

We further divide the study area into nine types of land use grids (Figure 7), and use the land use layer to intersect with the waterlogging simulation layer in ArcGIS to determine the submerged area and depth of each land use type as follows:

$$L_i=D_i$$

(12)

$L_i$ represents the depth of the ith grid, and $D_i$ represents the land type of the ith grid.

Combined with the depth loss curve shown in Figure 6, the initial waterlogging risk $R^{ini}$ of the study area under the design rainfall scenario is estimated by Formula (13).

$$R^{ini}={\sum }_1^k{\sum }_1^n\left(Y_i^k\ast A_i\right)$$

(13)

Among them, $Y_i^k$ represents the unit area loss value of land use type in the kth of the ith grid, which is obtained by substituting the previous depth loss curve fitting formula into the waterlogging depth of the grid; $A_i$ represents the area of the ith grid; K represents the number of land use type classifications; and n denotes the number of grids in the study area.

4. Result

4.1. Numerical Simulation Results

4.1.1. Rainfall Simulation

Based on the Zhengzhou rainstorm intensity formula and the Chicago rainfall process line method, the 120-min rainfall scenarios under four different return periods of a ten-year return period, a twenty-year return period, a fifty-year return period, and a one-hundred-year return period were simulated, using the rain peak coefficient of 0.433. In this study, the rain peak coefficient is chosen as 0.433 based on the notification issued by the Zhengzhou Municipal Natural Resources and Planning Bureau, combined with the climatic characteristics and rainfall distribution of Zhengzhou City. The rainfall peak coefficient of 0.433 can better reflect the distribution of rainfall intensity during a typical rainstorm in Zhengzhou City, which can help more accurately simulate the flooding situation and assess the risk of flooding.

The rainfall simulation process is presented in Figure 8, which mainly shows the changes in rainfall intensity and the total rainfall with a rainfall duration. Figure 8a is a ten-year rainfall simulation process diagram. The total rainfall of a 120 min of rainfall duration is 87 mm, and the rainfall intensity is 3.27 mm/min at the peak. Figure 8b is a 20-year rainfall simulation process diagram. The total rainfall of a 120 min of rainfall duration is 107 mm, and the rainfall intensity is 4.02 mm/min at the peak. Figure 8c shows the simulation process of the 50-year rainfall. The total rainfall in 120 min is 133.5 mm, and the rainfall intensity is 5.02 mm/min at the peak. Figure 8d shows the simulation process of the once-in-a-century rainfall. The total rainfall in 120 min is 153.6 mm, and the rainfall intensity is 5.78 mm/min at the peak.

According to the results of the rainfall simulation, the intensity and total amount of rainfall also change with the change in the rainfall return period. The longer the return period, the greater the rainfall intensity per minute, and the rainfall also increases substantially. Therefore, in urban drainage planning or the construction of sponge cities, the rainfall conditions under a longer return period should be considered to avoid the occurrence of heavy rain disasters.

4.1.2. Simulation of Urban Waterlogging Runoff Based on SCS-CN Model

Taking the simulated rainfall obtained by the rainfall simulation as input data, the SCS-CN runoff model was used to calculate the urban runoff yield under four rainfall scenarios. The main process is as follows:

(1)

Determine the hydrological soil grouping in the study area

The soil type distribution map was obtained by visualizing the soil data, as shown in Figure 9a. It can be seen from the soil type map that there are 13 soil types in the study area, most of which are fluvo-aquic soil and cinnamon soil. The soil texture corresponding to the soil type mainly includes sand, gravel soil, sandy loam, silt loam, and clay loam.

According to the soil classification standard of the United States Department of Agriculture (USDA), the soils with similar runoff capacity were classified to form the hydrological soil group, and the hydrological soil types were divided into A, B, C, and D according to the difference in permeability and water conductivity. Type A soil refers to sandy soil with a high infiltration rate. Type B soil is mainly loam; class C soil is sandy clay loam with high clay content. The clay type with the lowest infiltration rate is a D-type soil.

All the soil species included in the study area were merged one by one to obtain the soil classification results in line with the SCS model (Table 3). It can be seen from Figure 9b that the hydrological soil groups in the study area mainly include three types of ABC, among which type C is mainly distributed in the western region, type B is distributed in most areas, and type A is scattered in the southeastern region. The distribution of hydrological soil groups is one of the factors affecting the runoff yield capacity in the study area. Because the soil properties are relatively stable, the soil classification results will be superimposed with the land use data as invariant values for the rainfall runoff simulation.

(2)

Determine the land use type

The land use data in the study area were visualized, reorganized, and divided into seven land use types: impervious surface, forest land, grassland, cultivated land, water area, wetland, and bare land. The specific results are shown in Figure 10. It can be seen from the figure that the impervious surface of the study area occupies the main part and is mainly distributed in the central urban area. Grassland, woodland, and cultivated land are mainly distributed around the study area.

(3)

Determine the CN value

According to the relationship between the land use type and hydrological soil type in Table 2, the resulting map of CN value in the study area is obtained, as shown in Figure 11. It can be clearly seen from the figure that the CN value in most areas of the study area exceeds 70, which also indicates that the surface water infiltration capacity is poor, it is easier for the surface runoff to form, and the surface runoff is relatively high after rainfall. The main reason for this phenomenon is that the development of urban areas has increased the proportion of impervious surface.

(4)

Determine the average CN value

Because of the study area and the urban area, the urban waterlogging submergence method adopted is the local equal volume method, which divides the whole study area into 93 sub-catchment areas. Therefore, the average CN value of each sub-catchment area should be calculated when calculating rainfall runoff, and then, the runoff yield of each sub-catchment area should be obtained. The average CN values of 93 sub-catchments are shown in

Table 4. The average CN value of sub-catchment areas.

ID	Average CN Value	ID	Average CN Value	ID	Average CN Value	ID	Average CN Value
1	76.85	25	77.46	48	81.18	71	67.29
2	69.29	26	80.81	49	80.86	72	70.00
3	86.82	27	82.53	50	79.74	73	73.00
4	76.52	28	71.37	51	79.12	74	74.99
5	71.91	29	67.31	52	81.25	75	70.80
6	77.36	30	61.24	53	79.02	76	71.63
7	75.05	31	70.22	54	79.63	77	77.37
8	76.51	32	79.18	55	70.04	78	75.55
9	75.42	33	87.90	56	69.26	79	79.54
10	75.33	34	80.70	57	78.49	80	78.78
11	79.80	35	70.15	58	71.34	81	81.12
12	75.53	36	85.03	59	68.00	82	77.84
13	80.93	37	75.32	60	76.31	83	69.16
14	81.64	38	86.62	61	82.96	84	72.14
15	84.54	39	85.89	62	72.71	85	81.16
16	75.35	40	55.65	63	69.49	86	76.31
17	76.86	41	81.98	64	75.08	87	68.86
18	78.21	42	80.03	65	75.32	88	80.12
19	75.86	43	75.47	66	68.66	89	78.85
20	88.14	44	74.56	67	71.41	90	74.85
21	86.32	45	77.93	68	82.92	91	74.79
22	82.40	46	72.80	69	79.33	92	77.09
23	65.95	47	71.74	70	69.23	93	77.37
24	81.00

.

(5)

Determine the rainfall runoff

The rainfall runoff yield under different scenarios in 93 sub-catchment areas is calculated according to Formula (6).

4.2. Street Community-Level Refined Risk Assessment

According to the rainfall runoff obtained by the waterlogging runoff model based on the GIS waterlogging submergence model, the submergence range and submergence elevation under different rainfall conditions in each catchment area are obtained, and the data are visualized.

The simulation results are visualized to obtain the distribution map of the submerged area of the study area under different rainfall conditions (as shown in Figure 12). Among them, Figure 12a is a 10-year rainfall inundation range map, Figure 12b is a 20-year rainfall inundation range map, Figure 12c is a 50-year rainfall inundation range map, and Figure 12d is a 100-year rainfall inundation range map. It can be seen from the figure that the inundation caused by rainfall mainly occurs in the low-lying plain area, and the high-lying mountainous area in the southwest does not inundate. This is due to the fact that the southwest is mostly mountainous, with high terrain and steep slopes. The rainwater generated by the rainfall can quickly condense into streams and flow out and will not accumulate, while the flooded areas in the plain area are mostly distributed in low-lying areas. Due to the gentle terrain and low slope, the rainwater generated cannot flow out quickly and accumulates easily to form puddles. It can also be seen from the figure that most of the submerged areas are concentrated in the urban area, which is closely related to the high proportion of impervious surface in the urban area.

In order to more intuitively represent the impact of waterlogging, the depth of water accumulation was visualized. According to the design specification of GB51222-2017 “Technical Specification for Urban Waterlogging Prevention and Control” and referring to relevant literature, when the water depth is less than 0.15 m, it will not affect the passage of pedestrians and motor vehicles, and it can be considered that there is no waterlogging risk. When the depth of water accumulation is 0.15~0.3 m, it will affect pedestrian traffic, resulting in a slower speed of motor vehicles, generally will not cause casualties and property losses, and can be considered a low waterlogging risk. When the water depth is 0.3~0.5 m, it has a great impact on pedestrians and motor vehicles, which may cause casualties and property losses, and can be considered as a medium waterlogging risk. When the water depth is greater than 0.5 m, it will affect daily life, and the probability of casualties and property losses is high, which can be considered a high risk of waterlogging.

According to the above specifications, the results of the submerged range are further processed and divided into four categories according to the submerged depth, as shown in Figure 13. Among them, Figure 13a is the inundation depth map of the once-in-a-decade rainfall, Figure 13b is the inundation depth map of the 20-year rainfall, Figure 13c is the inundation depth map of the 50-year rainfall, and Figure 13d is the inundation depth map of the once-in-a-century rainfall.

According to the water submergence map of different recurrence periods, it can be seen that as the recurrence period increases, the submerged water depth gradually increases. The submergence depth is mostly low risk in the situation of once in ten years and once in twenty years, while the submergence depth is mostly medium risk and high risk in the situation of once in fifty years and once in one hundred years, which means that more casualties and property losses will occur.

4.2.1. Interpretation and Classification of Building Types

In this paper, the street image of University Road is processed and corrected according to Google Maps and Gaode Maps. The interpretation results of buildings in the study area are shown in Figure 14.

According to the attributes of the acquired points of interest and the field investigation, the types of buildings are determined and are mainly divided into five categories: science and education buildings, commercial buildings, entertainment buildings, service buildings, and residential buildings. For residential buildings, there are usually differences in geographical location, building quality, number of floors of buildings, and community service facilities, which cannot be evaluated uniformly according to one standard. Therefore, referring to the GBJ137-90 land use standard formulated by the Ministry of Housing and Urban-Rural Development, combined with the field survey results and the research results [43,44,45] on the classification of residential types, this study divides residential types into four categories, combined with the detailed classification criteria (see

Table 5. Residential classification description.

Category	Range	Remote Sensing Interpretation Instructions
First-class residence	Mostly low-rise residential buildings complete with municipal public facilities and a good environment	Villa
Second-class residence	Mainly the middle and high-rise residential buildings complete with municipal public facilities and a good environment	Commercial and residential buildings or residential areas with better environment
Third-class residence	Municipal public facilities are relatively complete and the environment is general; there is a mix of housing and industry	Multistory residential areas with a poor environment
Fourth-class residence	Urban villages, shanty towns, and other simple residential land-based	A bungalow with a poor environment

).

The distribution of different types of buildings obtained by dividing buildings into categories is shown in Figure 15. There are 11 first-class residential buildings, 293 s-class residential buildings, 155 third-class residential buildings, 61 fourth-class residential buildings, 101 science and education buildings, 30 commercial buildings, 5 entertainment buildings, and 10 service buildings, with a total of 666 buildings.

4.2.2. Risk Analysis of Disaster Events

Based on the numerical simulation results of Zhengzhou City, we selected two rainfall conditions of the 20-year return period and 100-year return period to conduct a community-level refined risk assessment, combined with the fine underlying surface information, a more detailed analysis of the risk of water accumulation on various types of residential buildings in the study area. Using GIS to resample the simulated maximum water depth and superimpose it with the extracted residential buildings, the risk distribution maps of buildings with the 20-year return period and 100-year return period are obtained, as shown in Figure 16 and Figure 17, respectively.

It can be seen from the map that when encountering the 20-year rainfall, waterlogging in the study area is mainly concentrated in the four types of residential areas in the northeast and the eastern middle area, while the area around the river and the entertainment building area does not have obvious waterlogging. Under the one-hundred-year rainfall, the simulation results of the water accumulation area are roughly similar to those under the twenty-year rainfall, but the range is slightly expanded, and the depth of the water accumulation with serious waterlogging is further increased.

Among them, the waterlogging around the river and the entertainment building is not harmful because the impervious surface of the land is small and the soil seepage capacity is strong. The four types of residential areas in the northeast are characterized by the aging of the drainage pipe network and the weak drainage capacity. The reason why the middle area in the eastern part of the study area is waterlogged is the low terrain. According to the above analysis, the risk of disaster events is not only related to rainfall intensity and elevation but also depends on land use, drainage network, and other factors.

4.2.3. Analysis of Building Exposure

As a disaster-bearing body, the exposure of a building depends on two aspects. One is whether the building is subject to waterlogging, and the other is whether the depth of waterlogging is higher than the height of the building from the ground. Only when these two points are met, the residential building is exposed to waterlogging disasters. Therefore, this section determines the ground height of different types of residential buildings based on field surveys and previous research results. The actual submerged depth layer of the house is obtained by subtracting the water depth from the ground elevation of the house, and the house with submerged water depth greater than zero is the house exposed to waterlogging.

(1)

Building height from the ground

In the process of urban waterlogging, only when the water depth exceeds the ground height of residential buildings, the flood can overflow the threshold or steps, resulting in the loss of indoor property. Therefore, it is very important to obtain the height of residential buildings from the ground more accurately for the risk assessment of waterlogging disasters. The design height of the house in the national architectural design standard is 15 cm, but the results of the field survey show that the height of the ground is related to the type of residential building. The high-end residential buildings such as commercial and residential buildings and villas in Zhengzhou generally have a high height from the ground. With the increase of the number of building layers, the height of the building from the ground also increases, and some even exceed 1.5 m, while the height of the low-end residential buildings is relatively low, and some residential buildings are lower than the national standard.

Based on the field survey results of Zhengzhou City, this paper comprehensively refers to the national architectural design standards and the research results of Yin and assigns different ground height values to different residential types. The ground height standard is shown in

Table 6. Different types of buildings per the ground height division standard.

Building Category	Building Height (m)	Ground Height (cm)
First-class residence	——	35
Second-class residence	≤15	15
	>15	35
Third-class residence	——	15
Fourth-class residence	——	5
Commercial buildings	——	20
Science and education buildings	——	30
service building	——	15
Entertainment building	——	10

, and the visual ground height map is shown in Figure 18.

(2)

Analysis of building exposure

Using GIS tools, the water depth values simulated by the waterlogging numerical model under the 20-year return period and the 100-year return period rainfall scenarios are assigned to the building off-ground height map layer, and the water depth of the building is subtracted from the off-ground height value. The actual submerged depth of the building can be obtained, and then, the spatial distribution characteristics, area, number, and the exposure degree of the buildings exposed to waterlogging disasters under different return periods of rainfall are obtained.

After processing, the sensitivity distribution map of buildings can be seen in Figure 19. Figure 19a is the exposure distribution of buildings with a return period of 20 years, and Figure 19b is the exposure distribution of buildings with a return period of 100 years. Under the two return period rainfall scenarios, the geographical location of flooded houses is roughly similar, most of which are located in the north and southwest of the study area.

According to the statistics of the experimental results, the number and proportion of various types of buildings exposed to waterlogging disasters are obtained.

Table 7. Exposure degree of different building types under the 20-year rainfall situation.

Building Type	Number of Buildings Exposed to Waterlogging Risk (Building)
	Level 1 Waterlogging D ≥ 70	Level 2 Waterlogging 50 ≤ D < 70	Level 3 Waterlogging 20 ≤ D < 50	Level 4 Waterlogging 10 ≤ D < 20	Mild Waterlogging D < 10	Total
First-class residence	0	0	0	0	0	0
Second-class residence	1	2	13	0	17	33
Third-class residence	2	5	9	0	19	35
Fourth-class residence	0	0	18	0	19	37
Science and education buildings	1	2	7	0	2	12
Service building	0	0	0	0	0	0
Entertainment building	0	0	0	0	0	0
Commercial buildings	0	0	1	0	1	2
Total	4	9	48	0	58	119

is the statistical table of the exposure degree of various types of buildings under the once-in-twenty-year rainfall scenario.

Table 8. Exposure degree of different building types under the 100-year rainfall situation.

Building Type	Number of Buildings Exposed to Waterlogging Risk (Building)
	Level 1 Waterlogging D ≥ 70	Level 2 Waterlogging 50 ≤ D < 70	Level 3 Waterlogging 20 ≤ D < 50	Level 4 Waterlogging 10 ≤ D < 20	Mild Waterlogging D < 10	Total
First-class residence	1	1	0	0	0	2
Second-class residence	23	9	14	25	0	71
Third-class residence	12	6	29	27	0	74
Fourth-class residence	5	14	15	13	0	47
Science and education buildings	5	4	4	13	0	26
Service building	0	0	0	0	0	0
Entertainment building	0	0	0	0	0	0
Commercial buildings	1	5	0	5	0	11
total	47	39	62	83	0	231

is the statistical table of the exposure degree of various types of buildings under the once-in-a-century rainfall scenario.

Table 9. Proportion of waterlogging in all kinds of buildings.

Building Type	Total	20-Year Rainfall	Proportion (%)	100-Year Rainfall	Proportion (%)
First-class residence	11	0	0	2	18.2
Second-class residence	293	33	11.3	71	24.2
Third-class residence	155	35	22.6	74	47.7
Fourth-class residence	64	37	57.8	47	73.4
Science and education buildings	101	12	11.9	26	25.7
Service building	10	0	0	0	0
Entertainment building	5	0	0	0	0
Commercial buildings	30	2	6.7	11	36.7

shows the proportion of the number and the total number of various types of buildings affected by waterlogging.

By analyzing Table 7, Table 8 and Table 9, the sensitivity distribution of buildings in different rainfall scenarios can be obtained. In the 20-year rainfall scenario, the number of buildings affected by waterlogging in the four types of houses is the largest (37), followed by three types of residential buildings (up to 35); one type of residential, service buildings, and entertainment buildings were not affected by waterlogging. In the one-hundred-year rainfall scenario, the three types of residential buildings affected by waterlogging are the most skilled (74 buildings); second, there are 71 s-class residential buildings, of which service buildings and entertainment buildings are not affected by waterlogging. The analysis of the sensitivity of the building according to the proportion clearly shows that the four types of residential buildings are most vulnerable to the impact of waterlogging disasters, so the sensitivity of the four types of residential buildings to waterlogging disasters is higher.

In the 20-year rainfall scenario, the proportion of the four types of houses affected by waterlogging is 57.8%, while in the 100-year rainfall scenario, the proportion of the four types of houses affected by waterlogging is as high as 73.4%. This is because the four types of houses are mostly shantytowns and simple houses, with poor surrounding environments and inadequate infrastructure construction. When a rainstorm comes, the accumulated water cannot be discharged in time, resulting in the current situation of vulnerability to disasters.

4.2.4. Economic Vulnerability Analysis

Disaster economic loss assessment can not only reflect the severity of disaster losses but also contribute to disaster prevention and mitigation and post-disaster recovery and reconstruction decision-making. With the increase in the frequency of hydrological phenomena and the continuous development of the social economy, the direct economic losses caused by waterlogging disasters in China are on the rise. In order to formulate disaster prevention and mitigation and adaptation strategies, and post-disaster compensation and recovery policies, it is important to comprehensively evaluate the economic losses due to waterlogging disasters.

According to the results of different land use types and water depth, the loss value is calculated according to the depth–disaster loss curve. The distribution of economic losses in the study area is shown in Figure 20. According to the results, the distribution of direct economic losses caused by a 20-year return period and a 100-year return period is relatively consistent. However, under the 100-year return period rainfall, the area with extremely high direct economic losses has increased significantly. Therefore, in the early warning of disasters, emergency rescue and disaster relief should be carried out in advance in areas with high losses.

5. Discussion

In this study, a numerical simulation method coupled with the SCS-CN hydrological model and GIS is used to provide a detailed assessment of the flooding risk for Zhengzhou City under different rainfall return periods and to visualize the inundation extent and the depth of waterlogging. Based on the numerical simulation results, we constructed a refined framework for urban flood risk assessment at the street community level with buildings as the basic assessment unit. The results of the study show the following:

(1)

Flood risk assessment results:

Under the conditions of ten-year, twenty-year, fifty-year, and one-hundred-year return period rainfall, there are significant differences in the inundation range and water depth across various districts and communities in Zhengzhou. Simulation results indicate that under the ten-year and twenty-year return period rainfall conditions, local waterlogging occurs in central urban areas (such as Jinshui District and Erqi District), but the overall impact is minor. In contrast, under the fifty-year and one-hundred-year return period rainfall conditions, low-lying areas (such as Guancheng Hui District and Zhongyuan District) face severe inundation risks, potentially causing more serious casualties. Specifically, the central urban area, due to its flat terrain, dense buildings, and high drainage system pressure, exhibits higher water depth under shorter return period rainfall. Meanwhile, the peripheral areas, with low-lying terrain, show a wider range of inundation under longer return period rainfall.

(2)

Analysis of waterlogging causes

The main causes of waterlogging in the study area include the following three points: extreme climatic events, topographic factors, and imperfect drainage systems. The terrain characteristics of the northern and eastern central areas of the study area lead to the accumulation of rainwater and the formation of waterlogging areas. In addition, some areas in the study area, such as the northern part, have old infrastructure, incomplete drainage networks, and poor drainage capacity. According to the experimental results, the high-risk areas in the study area are mainly concentrated in low-lying areas with imperfect drainage systems. Flood risk in these areas is not only affected by heavy rainfall but also by land use types and soil permeability. Through a detailed analysis of the causes of floods, we can draw the following conclusions: First, the increase in the proportion of impervious water in the central urban area leads to a significant increase in surface runoff after rainfall. Different types of land use and buildings have different impacts on the flood risk. For example, commercial and residential areas usually face a higher risk of flooding due to their high proportion of impervious surfaces, while green spaces and wetlands can mitigate the impact of flooding to a certain extent due to better permeability.

In order to further illustrate this point, we take University Road Street in the Erqi District of Zhengzhou City as an example for in-depth analysis. The area includes 11 communities such as Xianghe Community, Zhongyuan Community, Songshan Community, and Kangqiao Huacheng Community, covering an area of 2.7 square kilometers. There are various types of buildings in the area, including residential areas, parks, schools, hospitals, enterprises, etc., with strong representation. Through numerical simulation, it is found that the water depth of Daxue Road Street is large under extreme rainfall conditions, especially in low-lying areas and areas with insufficient drainage facilities. The waterlogging disaster of University Road may lead to the damage of residential areas and school buildings, the difficulty of personnel evacuation, and serious economic losses. In addition, waterlogging could affect traffic in the region and hinder rescue and recovery efforts.

(3)

Comprehensive risk assessment:

A comprehensive analysis of the hazardous event risk, building sensitivity, and economic vulnerability of the study area leads to the following conclusions:

(i)

It is obvious from the results of the study that in the event of a heavy rainfall disaster, the generation of waterlogging is not only related to the elevation of the ground but also to the surrounding environment and facilities. The center of the study area is crossed by a river, and the river, as well as the two sides of the river, are not susceptible to internal flooding as the water generated by heavy rainfall can flow out of the river despite the low topography; whereas, there is an obvious waterlogged area in the eastern part of the study area caused by the low topography; and there is an obvious waterlogged area near the Category 4 dwellings in the north, which is attributed to the old infrastructure of the area, incomplete drainage pipe network, and poor drainage capacity.

(ii)

For buildings, the height above ground also affects whether a building is at risk as water can flow indoors and cause indoor damage if the water height in the waterlogged area exceeds the lowest opening of the building. Among the buildings in the study area, the height above ground of Type 1 dwellings and Type 2 dwellings with a building height of more than 15 m reaches 35 cm, so this type of dwelling will not cause indoor damage even if water accumulates up to 35 cm at the location where it is situated; however, Type 4 dwellings have a height above ground of only 5 cm, and the accumulated water can easily enter the indoor area and cause more serious damage.

(iii)

The economic losses caused by urban flooding disasters are related to the depth of standing water and the value of the site type. Although the sensitivity of Type 4 dwellings is very low and they are vulnerable to flooding, most of the Type 4 dwellings are squatters or makeshift houses, which are of low value and, therefore, generate low economic losses. The areas with high and very high economic losses in the study area are mainly located in residential land use Categories 2 and 3 and scientific and educational land use, with the proportion of Category 2 and 3 dwellings affected by the 1 in 20 years rainfall scenario and the 1 in 100 years rainfall scenario being 15.2% and 32.4% respectively, and the proportion of scientific and educational land use affected by the 1 in 20 years rainfall scenario and the 1 in 100 years rainfall scenario being 11.9% and 25.7% respectively. Although the proportion of the affected area is not the highest, its housing value is high, and the economic loss caused by the disaster is also high.

By evaluating the flood risk levels of different areas, this paper finds that the high-risk areas are mainly concentrated in low-lying urban areas and urban areas with relatively weak drainage facilities. For example, University Road Street in Erqi District and Future Road Street in Jinshui District have a high-risk level due to their low terrain, high population density, and aging drainage systems. In contrast, newly developed areas such as the High-Tech Zone and the Economic Development Zone are at relatively low risk due to their newer drainage systems.

(4)

Implications and recommendations:

The research results provide important references for urban planning and flood management in Zhengzhou. It is recommended that urban planning fully consider flood risk assessment results, optimizing the layout of urban drainage systems according to the principles of “source reduction, process control, and end treatment” and strengthening the construction of flood control facilities in low-lying areas to enhance the city’s overall flood prevention capacity.

Based on the flood risk assessment, specific recommendations are made for areas with different risk levels. In the medium- to high-risk areas, improving infrastructure is crucial. First, the drainage system needs to be expanded and improved, increasing pipeline capacity to ensure rapid drainage during heavy rainfall. Additionally, more drainage pumping stations should be constructed, especially in water-prone low-lying areas, to effectively remove accumulated water. For flood control facilities, large rainwater storage tanks and flood dikes should be constructed to divert and prevent the spread of floodwaters. In terms of land use planning, increasing green spaces and open areas to utilize the infiltration capacity of vegetation and soil can reduce surface runoff. Simultaneously, high-density development in high-risk areas should be restricted to reduce the proportion of impermeable surfaces. To ensure effective early warning and emergency response, it is recommended to install intelligent monitoring systems to monitor rainfall and water level changes in real time and promptly issue warning information. Detailed emergency plans, including evacuation routes and temporary shelters, should be developed to ensure rapid response in case of flooding. Additionally, disaster prevention and mitigation knowledge training should be conducted to raise residents’ awareness and self-rescue capabilities, and community volunteer teams should be organized to assist in flood prevention and rescue efforts, further enhancing community participation and response capabilities.

In low-risk areas, infrastructure maintenance is particularly important. Regular inspection and maintenance of the drainage system are needed to ensure that drainage pipes and outlets are unblocked and promptly cleared of obstructions. At the same time, maintaining existing green spaces and vegetation is essential to ensure their effective role in infiltration and water retention during heavy rainfall. Reasonable land use strategies include protecting green spaces and wetlands, restoring the natural storage and infiltration functions of wetlands, and controlling development density to prevent the increase of impermeable surfaces due to overdevelopment. Simplified early warning systems should be established to promptly communicate weather changes, and basic emergency measures should be developed to ensure rapid response during rare extreme weather conditions. Regular flood control drills should be conducted to improve community residents’ emergency awareness and capabilities, and weather forecasts and warning information should be promptly communicated to enhance community participation and response capabilities.

6. Conclusions

In this paper, a numerical simulation model coupled with hydrology and GIS is constructed, and a refined risk assessment is carried out at the street community level. According to the rainstorm intensity formula of Zhengzhou City and the Chicago rain pattern method, the rainfall under four different recurrence periods was designed, and the SCS-CN and GIS-coupled urban waterlogging numerical simulation model was constructed. According to the rainfall results, the urban waterlogging numerical simulation was carried out in Zhengzhou City, and the inundation range and water depth were visualized. At the same time, according to the inundation of waterlogging, the risk assessment of Daxue Road Street is carried out from three aspects: building risk, exposure, and direct economic loss in the study area. The final risk assessment results can provide valuable waterlogging risk information for relevant departments and decision-makers and assist scientific disaster relief and early warning.

In summary, through numerical simulation and refined assessment, this study provides a scientific basis and practical guidance for flood risk management in Zhengzhou. The research results provide not only an important reference for flood control and disaster reduction in Zhengzhou but also a framework and method for other cities to carry out similar research. However, the SCS-CN model is the most suitable for rainfall runoff estimation in small and medium-sized basins or regions. For areas with large watersheds or complex terrain, the accuracy of the model may be affected. Second, the model assumes that the rainfall runoff process is uniform and stable, so the model may not accurately reflect the actual situation when the rainfall intensity and spatial distribution change greatly.

In addition, the SCS-CN model is mainly based on factors such as land use, soil type, and initial water conditions but fails to fully consider the impact of complex factors such as vegetation cover change, soil erosion, and groundwater flow. In future studies, we will consider a variety of factors, such as vegetation cover change, soil erosion, groundwater flow, etc., to conduct a more comprehensive assessment and simulation of flood risk.