Changes in Population Exposure to Rainstorm Waterlogging for Different Return Periods in the Xiong’an New Area, China

Abstract

In the context of global climate change and urban expansion, urban residents are encountering greater rainstorm waterlogging risk. Quantifying population exposure to rainstorms is an important component of rainstorm waterlogging risk assessments. This study utilized a two-dimensional hydrodynamic model to simulate the inundation water depth and inundation area resulting from rainstorms, with return periods of 5, 10, 50, and 100 years, in the Xiong’an New Area, and overlaid the gridded population data in 2017 and in 2035 under SSP2 to assess the change in population exposure. The results show that the average inundation depth and area increase were from 0.11 m and 207.9 km² to 0.18 m and 667.2 km² as the rainstorm return period increased from once in 5 years to once in 100 years. The greatest water depths in the main urban areas were mainly located in the low-lying areas along the Daqing River. The total population exposed to rainstorm waterlogging for the 5-, 10-, 50-, and 100-year return periods was 0.31, 0.37, 0.50, and 0.53 million, respectively, in 2017. However, this is projected to rise significantly by 2035 under SSP2, increasing 2–4-fold compared with that in 2017 for the four return periods. Specifically, the projected population exposure is expected to be 0.7, 1.0, 1.8, and 2.0 million, respectively. The longer the return period, the greater the increase in population exposure. The proportion of the population exposed at the 0.05–0.2 m water depth to the total population exposure decreases as the return periods increases, whereas the proportion changes in the opposite direction at the 0.2–0.6 m and >0.6 m depth intervals. Spatially, high-exposure areas are concentrated in densely populated main urban regions in the Xiong’an New Area. In the future, more attention should be paid to densely populated low-lying areas and extreme recurrence rainstorm events for urban flood-risk management to ensure population safety and sustainable urban development.

1. Introduction

The world is experiencing unprecedented and significant warming, and extreme climate events such as heavy precipitation are becoming more frequent and intense [1]. In the future, it is projected that extreme precipitation will continue to increase, potentially exacerbating flood risk [2,3]. Direct economic losses due to flood disasters in China increased significantly from 1984 to 2021 The annual average value of the losses was approximately CNY 114.29 billion, accounting for 0.94% of the GDP [4]. With the acceleration in urbanization, impervious surface areas continue to expand, yet the urban drainage systems have not been updated in parallel. Urban waterlogging has become a major natural disaster in densely populated areas, posing a substantial threat to human life, productivity, and socio-economic activities [5,6,7,8,9,10,11]. Between 2015 and 2019, approximately 209 cities in China experienced various degrees of rainstorm waterlogging annually [12,13]. For example, in July 2021, Zhengzhou, located in Henan Province in China, experienced an exceptionally heavy rainfall event, which affected 14.786 million people across 150 counties (cities and districts). Additionally, in late July 2023, North China and the Haihe River Basin experienced extreme rainfall, resulting in heavy casualties in Fangshan in Beijing, Baoding, and the Xiong’an New Area in Hebei Province, among others. The Xiong’an New Area is a state-level new area established in 2017 in China, which is located in central Hebei Province. As part of the non-capitalization of Beijing, the Xiong’an New Area will be built into a super-level modern city and become an important point on the world-class Jing–Jin–Ji (short for Beijing, Tianjin, and Hebei) Economic Belt. Historically, flood disasters have occurred frequently in the Xiong’an New Area. From 1510 to 2020, there have been flooding disasters for 225 out of the 511 years, with an average occurrence of approximately once every 2.3 years [14]. During extremely severe flooding years, approximately 80% of the area’s territory was submerged [15,16]. Looking ahead, it is expected that the Xiong’an New Area and its surrounding regions will face a significant increase in the frequency and intensity of heavy rainfall and flooding. From 2026 to 2045, extreme precipitation events in the Xiong’an New Area are projected to increase by 10% to 25% [17]. As the construction of the Xiong’an New Area progresses, a substantial influx of people is anticipated, with the population estimated to reach approximately 5 million by 2035 [18]. This will result in a significant rise in the urban population size and density. Therefore, to which degree the urban population will be affected by rainstorm waterlogging in the future is the key focus of urban planning and construction and urban flood-risk management.

The modeling of rainstorm waterlogging has progressed from rudimentary empirical and conceptual models to more advanced hydrodynamic models [19]. Hydrodynamic models organize discrete urban areas into grids and utilize Saint-Venant equations or shallow-water equations to describe surface runoff processes, so as to simulate the urban surface inundation processes, which are difficult to achieve in traditional hydrology, and obtain spatial and temporal distribution information such as the inundation depth, inundation range, and inundation duration during waterlogging. At the same time, they show a stronger advantage in considering the hydraulic characteristics of urban pipe networks, river channels, lakes, buildings, and other terrain [20]. Therefore, hydrodynamic models, such as the FloodArea model, have been widely used in urban waterlogging and inundation simulations. Researchers have used the FloodArea model to simulate the spatial distribution of flood inundation in the Hun River and Datong River basins, as well as the urban waterlogging of different recurrence periods in Nanchang, Hefei, and Ningbo, and all of them have shown a good simulation effect [21,22,23,24].

People in affected communities are the most critical subjects affected by urban waterlogging disasters, so population exposure analysis is an integral part of flood-risk management [25,26]. Indeed, some researchers have integrated the spatial distribution information derived from extreme precipitation, flood events, or inundation simulations with vulnerable populations to assess the population exposure to rainfall-induced flooding disasters. From a global perspective, approximately 255 to 290 million people were affected by large flood events from 2000 to 2018 [27]. The regions with the highest population exposure are located in Asia, with China (395 million) and India (390 million) accounting for over one third of the global exposed population [26,28]. At the national level, about 21.8 million people (6.87% of the population) in the United States faced a once-in-a-century flood risk in 2015, with most of the exposed population residing near bodies of water [29]. In China, the multi-year average population exposure to heavy-rainfall-induced flooding between 1984 and 2012 was 126 people per square kilometer, with a significant increase in exposure levels [30]. On a regional scale, some researchers used the FloodArea model to simulate the major flood process in the upper reaches of the Hun River in 2013, and estimated that approximately 83,000 people were affected [22]. Additionally, other researchers employed hydrodynamic models to simulate and assess the impact of extreme urban flooding during the “7.20” event in Zhengzhou, China, providing precise evaluations of population exposure [31]. Nevertheless, the current studies do not specifically focus on population exposure to rainfall-induced waterlogging on a city scale. Moreover, studies on population exposure often rely on historical demographic data, and fewer studies have considered dynamic socio-economic scenarios.

In pursuit of this objective, the Xiong’an New Area has been selected as the study area. The research draws upon various data sources, including observatory data on daily precipitation spanning from 1951 to 2020, a digital elevation model (DEM), land-use data, and population data from 2017 to 2035. The FloodArea model was employed to simulate the extent and depth of rainstorm inundation for different return periods (5, 10, 50, and 100 years) within the Xiong’an New Area. Then, historical and future population data were employed to analyze the dynamic changes in exposure. The future population was estimated using Shared Socioeconomic Pathways (SSPs). SSPs reflect the current global and regional development status, as well as potential future development changes. There are five emission scenarios, known as SSP1–5. The SSP2 scenario selected for this paper represents a world that follows a historical approach with moderate challenges encountered regarding both mitigation and adaptation [32]. Subsequently, the population grid data in 2017 and 2035 under SSP2 were overlaid with the rainstorm inundation area to estimate the population exposure. Finally, the dynamic changes in population exposure from 2017 to 2035 were analyzed. This study aimed to provide scientific support for mitigating the impact of floods and effectively responding to rainfall-induced waterlogging disasters in the Xiong’an New Area. Through a dynamic exploration of the spatial and temporal changes in population exposure, it sought to offer insights into the development of long-term, sustainable risk-management strategies in the future for the Xiong’an New Area.

2. Materials and Methods

2.1. Study Area

The Xiong’an New Area is situated in the eastern region of Baoding, Hebei Province, China. It covers an area of approximately 1700 square kilometers. It encompasses the Xiong, Rongcheng, and Anxin counties, along with their surrounding areas (Figure 1). It is positioned on the alluvial fan of the Daqing River system in Haihe River Basin. The terrain of the Xiong’an New Area is high in the northwest and slightly low in the southeast, with a natural longitudinal slope of about 1‰. The soil layer is deep, and the vegetation coverage is very low, while there are several ancient river channels within the area. The mountainous area accounts for 64.1% of the land area, comprising mainly forests and grasslands in the west of the basin. The plains comprise mainly farmland and cities [33]. Baiyangdian Lake, North China Plain’s largest freshwater lake, is located southeast of the Xiong’an New Area. It receives water from nine upstream river branches of the Daqing River System. The Xiong’an New Area has a warm, temperate, monsoonal continental climate with a dry spring, wet and rainy summer, cool and dry autumn, and cold and little snow in winter. The average annual temperature is 12.4 °C, the average annual precipitation is approximately 500 mm, and the maximum annual precipitation is 942 mm. The Xiong’an New Area has a history of frequent flooding. During flood years, the low-lying areas near rivers and lakes are prone to submersion, accounting for 20% to 30% of the total region. During severe flood years, except for the high-lying areas of Rongcheng County, approximately 80% of the area is inundated [15,16].

2.2. Data

The source and attributes of the research data used in this paper are shown in

Table 1. Basic research data.

Data Type	Attribute	Source
Observed precipitation	Daily, 1951–2020	National Meteorological Information Centre, China Meteorological Administration
Observed flood-prone sites	3 sites, 12 August 2020	Hebei Meteorological Disaster Prevention Centre
DEM	30 m	https://earthexplorer.usgs.gov/
Administrative boundaries	Shpfile, 2015	The National Geomatics Center of China
Land use	30 m, gird file, 2015	Resource and Environment Science Centre of the Chinese Academy of Sciences (http://www.resdc.cn/)
	30 m, gird file, 2035	The land-use plan of the Xiong’an New Area in 2035 (http://www.xiongan.gov.cn/)
Population	2017–2020	Baoding Economic and Statistical Yearbook and the China Statistical Yearbook
	2021–2035	Future population data under SSP2 were projected using the Population, Development, Environment (PDE) model

. Among them, daily precipitation data from 1951 to 2020 of three meteorological stations in Xiong, Anxin, and Rongcheng Counties within the Xiong’an New Area were obtained from the China Meteorological Administration’s National Meteorological Information Centre. Rigorous preprocessing was applied to the original precipitation data, involving thorough checks and handling of any missing values, outliers, or erroneous data, ensuring the analysis was based on high-quality data. The above data were mainly used for the calculation of rainstorms for different return periods. The measured depth of inundation caused by heavy rainfall on 20 August 2020 in the Xiong’an New Area was obtained from the Hebei Meteorological Disaster Prevention Centre, and was used to verify the model’s adaptability.

Population data from 2017 to 2020 were derived from the Baoding Economic and Statistical Yearbook and the China Statistical Yearbook. Future population data for the 2021–2035 period under SSP2 were projected by the Population, Development, Environment (PDE) model [18] and then gridded to a resolution of one kilometer using the land-use classification method [34].

The basic geographic data include the land use, DEM, and administrative boundaries of the Xiong’an New Area. The land-use data for 2015 were derived from the Resource and Environment Science Centre of the Chinese Academy of Sciences, and the land-use data for 2035 were digitized from the land-use plan of the Xiong’an New Area in 2035, with a spatial resolution of 30 m for both phases. DEM was obtained from the SRTM (Shuttle Radar Topography Mission), with an original spatial resolution of 30 m. The administrative boundary data were obtained from the National Geomatics Center of China. Due to the different sources of data, the data were subjected to coordinate and projection conversion, with the uniform projection of Albers and the coordinate system of China Geodetic Coordinate System 2000 (CGCS2000).

2.3. Methodology

The primary objective of this study was to analyze the dynamic changes in population exposure to rainstorm waterlogging in the Xiong’an New Area. The flow chart of this study is shown in Figure 2, below. Initially, this study utilized the generalized extreme value distribution (GEV) function to calculate the daily precipitation of different return periods and assess the fitting results using the Kolmogorov–Smirnov (K–S) test [35]. Subsequently, the FloodArea hydrodynamic model incorporated daily precipitation data, DEM, ground roughness, and runoff coefficient data to simulate the depth and extent of inundation caused by the 24 h rainstorm. Finally, the inundation results were overlaid with historical and future population grid data to conduct a comparative analysis of the scale and spatial distribution of the affected population in the Xiong’an New Area for different return periods of rainstorm-induced inundation.

2.3.1. FloodArea Model

The FloodArea model, a hydrodynamic model developed by Geomer in Germany that focuses on flood-risk analysis, has been widely used in recent years. Unlike traditional static flood simulations, the FloodArea model provides accurate information on flood paths, arrival times, and inundation depths by simulating the progress of flooding evolution, and the results are displayed and stored visually and clearly in a raster format [36]. The calculation of inundation areas is based on a hydrodynamic approach. All eight neighbors of a raster cell are considered. The discharge volume to the adjacent cells is calculated using the Manning–Stricker formula [37]:

$$V=k_{st}\cdot r_{hy}^{\frac{2}{3}}\cdot I^{\frac{1}{2}}$$

(1)

where $r_{hy}^{\frac{2}{3}}$ is the hydraulic radius (unit: m), $I^{\frac{1}{2}}$ is the slope (unit: m/m), and $k_{st }$ is the hydraulic roughness coefficient. The appropriate values of $k_{st }$ vary for different land surfaces, and these values are related to the Manning coefficient [24]. The inundation depth is taken from the difference between the water level and maximum terrain elevation:

$$Flowdepth=Waterlevel-elevation$$

(2)

Flowdepth is the inundation depth at a point on the ground, Waterlevel is the water level elevation at that point, and elevation is the topographic elevation at that point. The direction of water flow during the whole inundation process is determined by the slope of the terrain and is calculated as follows:

$$aspect=270-\frac{360}{2\pi }\cdot \alpha \tan 2\left[\frac{\partial z}{\partial y},\frac{\partial z}{\partial x}\right]$$

(3)

α is the slope of the terrain, $\frac{\partial z}{\partial y}$ is the rate of change in elevation in the north–south direction, and $\frac{\partial z}{\partial x}$ is the rate of change in elevation in the east–west direction.

The FloodArea model has three simulation models, including the water level, the hydrograph, and rainstorm. The rainstorm module was used in this article. The rainstorm module requires precipitation process data, ground roughness data, runoff coefficient data, and DEM for the study area, where the ground roughness data are assigned empirical values of 18, 25, 33, 40, and 50 for woodland, settlement, paddy field, dry field, and water, respectively [24]. Runoff coefficients are assigned by land-use type from the standard for the design of outdoor wastewater engineering of China (GB50014-2021) [38]. Precipitation was allocated to 24 h for different return periods based on rainstorm pattern. According to the research results of Yu et al. (2021) [39], the rainstorm pattern is consistent for different return periods. The precipitation process of the 24 h rainstorm in the Xiong’an New Area was set as a single-peak type, with its peak occurring in the latter two thirds of the rainfall period.

2.3.2. Calculation of Rainstorms for Different Return Periods

The extreme value theory has been widely applied to investigate the occurrence probability of extreme precipitation. At the beginning of the 20th century, Fisher and Tippet proposed three types of extreme value distributions: Gumbel, Fréchet, and Weibull [40]. Later, Jenkinson and Coles unified these three distributions into the generalized extreme value (GEV) with three parameters, which allows for greater flexibility [41,42]. The GEV method, widely utilized in hydrometeorology, has been extensively employed in China to analyze the probability distribution of precipitation extremes [43,44]. In this study, the annual maximum daily precipitation was screened, and then the GEV distribution function was used to calculate the rainfall for the 5-, 10-, 50- and 100-year return periods for Rongcheng, Anxin, and Xiongxian Counties in the Xiong’an New Area. The goodness of fit of the results was assessed using the Kolmogorov–Smirnov (K–S) test [35]. The K–S test is a statistical method that compares the cumulative frequency distribution of sample data with a specific theoretical distribution, enabling us to determine the degree to which the observed data fits the expected theoretical distribution.

The formulas and parameters of the GEV distribution function are as follows:

Parameters: κ, σ, μ (σ > 0)

Scope:

$$\begin{array}{cc}1+\kappa \frac{x-\mu }{\sigma }>0& \kappa \ne 0\hfill \\ -\infty <x<+\infty & \kappa =0\hfill \end{array}$$

(4)

Probability density functions:

$$f(x)=\left\{\begin{array}{ll}\frac{1}{\sigma }\exp \left(-{\left(1+\kappa z\right)}^{{}^{-\frac{1}{\kappa }}}\right){\left(1+\kappa z\right)}^{{}^{{}^{-1-\frac{1}{\kappa }}}}& \kappa \ne 0\\ \frac{1}{\sigma }\exp \left(-z-\exp \left(-z\right)\right)& \kappa =0\end{array}\right.$$

(5)

Cumulative probability density function:

$$F(x)=\left\{\begin{array}{ll}\exp \left(-{\left(1+\kappa z\right)}^{-\frac{1}{\kappa }}\right)& \kappa \ne 0\\ \exp \left(-\exp \left(-z\right)\right)& \kappa =0\end{array}\right.$$

(6)

$z=\frac{x-\mu }{\sigma }$, $x$ is a sequence of precipitation extremes, and κ, σ and μ are the shape, size, and position parameters, respectively.

The K–S statistic is calculated as follows:

$$\mathrm{Dn}=\max \left|Fn(x)-\left.F(x)\right|\right.$$

(7)

Given a sample xl…… xn of random variables has distribution function F, where Fn is the empirical distribution function of the sample, and Dn reflects the maximum difference between the empirical distribution function Fn(x) and the specified distribution function F(x), so a smaller Dn indicates a better fit and can be used as a basis for preference. This paper concluded the test by comparing Dn with the critical values that were calculated at different levels of significance. The K–S test statistics of 0.0422, 0.0371, 0.0491, and 0.0475 for Rongcheng, Xiong, and Anxin Counties, and the Xiong’an New Area, respectively, were smaller than the critical value (0.1623) at a significance level of 0.05, and the correlation between the theoretical cumulative probability distribution and the empirical cumulative probability distribution reached more than 0.97 (Figure 3). This indicated that the generalized extreme value distribution function was a good fit.

2.3.3. Population Exposure to Rainstorm Waterlogging

Exposure refers to the location where people, livelihoods, the environment, and infrastructure are likely to be adversely affected [45]. Population exposure to rainstorm waterlogging is defined as the population in inundation areas. It was calculated by overlaying inundation-depth data and population data, measuring the water depth and corresponding population under each grid point, and accumulating the total population under the same water-depth interval. Exposure intensity can be computed by multiplying the different inundation depths and populations for each grid.

3. Results

3.1. Analysis of the FloodArea Model Applicability

Because of the different climatic and geographical backgrounds, the applicability of the FloodArea model had to be verified before the simulation. This study conducted a comparison between the observed water depths and the model-simulated water depths from three flood-prone sites to validate the simulation applicability of the model. The actual measured data were provided by the Disaster Risk Management Bureau (DRMB), and the simulated water depths were generated by the FloodArea model simulating heavy rainfall events. A rainstorm on 12 August 2020, with daily precipitation of 58.8 mm, 90.2 mm, and 154.9 mm at the Rongcheng, Anxin, and Xiong County stations, respectively, caused the water depth in the northeast corner of Shengtang Community to reach 0.6 m, the water depth at the entrance of the Third Primary School to reach 0.5 m, and the water depth at the entrance of the Housing and Urban–Rural Development Bureau in Xiong County to reach 0.35 m.

According to the comparison of the observed and simulated inundation depths at flood-prone sites shown in

Table 2. Comparison of the observed and simulated inundation depths at flood-prone sites.

Flood-Prone Site	Coordinates	Observed Water Depth (m)	Simulated Water Depth (m)	Relative Deviation (m)
1. Northeast corner of the Shengtang District	39.01° N, 116.12° E	0.6	0.36	0.24
2. Front of the Third Primary School	39.00° N, 116.12° E	0.5	0.55	0.05
3. Front of the Housing and Urban–Rural Development Bureau	38.98° N, 116.11° E	0.35	0.34	0.01
Average value		0.48	0.41	0.1

, the relative deviation in the northeast corner of Shengtang Community was 0.24 m, 0.05 m at the entrance of the Third Primary School, and 0.01 m at the entrance of the Housing and Urban–Rural Development Bureau. The mean error between the observed and simulated inundation depths was 0.1 m, which reflected the actual inundation situation to some extent. The model had good simulation accuracy and reliability, and could be used for flood simulation in the Xiong’an New Area.

3.2. Rainstorms for Different Return Periods

Table 3. Rainstorms for different return periods in the Xiong’an New Area (mm).

Station	5 Years	10 Years	50 Years	100 Years
Rongcheng	88.48	106.38	150.36	171.06
Anxin	98.18	112.02	185.77	218.24
Xiong	95.34	115.00	164.03	187.41
Xiong’an New Area	84.86	102.75	149.26	172.33

presents the rainstorms in the Xiong’an New Area and three counties for different return periods. We calculated the rainfall with return periods of 5, 10, 50, and 100 years in the Xiong’an New Area; the rainfall was 84.86, 102.75, 149.26, and 172.33 mm, respectively. Under the same return levels, the rainfall in Anxin was the highest and in Rongcheng the lowest. With the increase in the return level, the disparity in rainstorms among the three counties increased. From the 5- to the 100-year return periods, the difference in rainstorms between Rongcheng and Xiong increased from approximately 10 mm to 47 mm. Taking the arithmetic mean of daily precipitation at the three stations as the daily rainfall of the Xiong’an New Area, the rainstorm days and annual maximum daily precipitation across the Xiong’an New Area were calculated. The results are shown in Figure 4. The Xiong’an New Area has been experiencing 85 days of rainstorms (daily precipitation ≥ 50 mm) since 1951. The highest number of rainstorm days was recorded in 1994, with 5 days. There has been no significant change in the annual maximum daily precipitation in recent years, but the highest annual maximum daily precipitation occurred on July 2, 2016, with 199.3 mm rainfall. From 1951 to 2020, the number of days for rainstorms with return periods of 5, 10, 50, and 100 years was 16, 7, 2, and 1, respectively.

3.3. Changes in Inundation Depth and Inundation Area

The FloodArea model simulated the inundation depth and inundation area for 24 h for 5-, 10-, 50-, and 100-year rainstorms in the Xiong’an New Area. The results are depicted in Figure 5 and

Table 4. Changes in the inundated areas of rainstorm waterlogging for different return periods.

Depth of Inundation/m	Inundated Areas/km2
	5 Years	10 Years	50 Years	100 Years
0.05–0.2	186.9	253.1	481.1	531.3
0.2–0.6	18.2	28.0	73.7	120.2
>0.6	2.8	4.3	11.4	15.7
Total inundated areas/km2	207.9	285.4	566.2	667.2

. It was observed that the impact resulting from water depths below 0.05 m was relatively insignificant. Consequently, the calculation of inundation areas was conducted based on three distinct depth categories: 0.05–0.2 m, 0.2–0.6 m, and greater than 0.6 m. Notably, Baiyangdian Lake was excluded from the calculation.

The extent of inundation increased for each water-depth interval as the return period level got longer (Figure 5 and Table 4). The total inundation area of the Xiong’an New Area caused by rainstorms with a 5-year return period was 207.9 km², accounting for 11.8% of the total region. The water depth was mainly concentrated between 0.05 and 0.2 m, with an average inundation depth of 0.11 m. The deepest water in the main urban areas of the three counties was 1.8 m. It was located west of Xiongzhou Town in Xiong County, where the Daqing River system enters the county. The overall inundated area increased by 77.5 km² for the 10-year return period rainstorm, with a growth rate of 37.3%. The 0.05–0.2 m water depth had the greatest increase in inundated areas, at 66.2 km². The average water depth changed little, with the deepest water depth at 1.9 m. During the 50-year return period rainstorm, the depth and extent of inundation grew rapidly. The total inundated area reached 566.2 km², representing 32.0% of the total area of the Xiong’an New Area. The total flooded area expanded by 280.6 km², with an overall increase of 98.3% compared to the 10-year return period rainstorm. The average depth of inundation increased to 0.16 m, though the greatest depth was 2.1 m. By the 100-year return period rainstorm, the growth rate of the total inundated area slowed to 17.8%. Nearly 40% of the region ended up submerged, with a maximum water depth of 2.6 m.

In general, the distribution of the greater inundation depths was basically consistent under the return periods of 5-, 10-, 50-, and 100-year rainstorms in the Xiong’an New Area, and mainly covered the urban and town areas of Rongcheng and Xiong Counties, as well as some village areas located in the west and south of Anxin County. The deepest water was mostly found in low-lying places along the Daqing River in Xiong County. The maximum inundation range was found at an inundation depth of 0.05–0.2 m for the four return-period rainstorms, accounting for 89.9%, 88.7%, 85.0%, and 79.6% of the total flooded areas, respectively. The largest expansion in the inundation area occurred in the 0.05–0.2 m range as the return period increased, but the growth rate in this interval was the smallest. The largest growth rate of the inundation area occurred at the water depths above 0.6 m between the 10- and 50-year return period rainstorms, and the growth rate approached 165%.

3.4. Population Exposure to Rainstorm Waterlogging

3.4.1. Population Changes

Figure 6 illustrates the population changes in the Xiong’an New Area from 2017 to 2035. The projected population of the Xiong’an New Area in 2035 is expected to reach 5.03 million. Compared with the establishment of the Xiong’an New Area in 2017, the population is expected to increase by 3.9 million, and the population growth rate by 342%. The projected results show that the population in 2035 will closely align with the planned population of 5 million noted in the planning report for the Xiong’an New Area.

The Xiong’an New Area Planning Outline states that the Xiong’an New Area will be developed first by selecting a specific area (the border area between Rongcheng and Anxin counties) as the start-up area, building five peripheral clusters around the start-up area that leave space for major national development strategies and sustainable urban development, and finally forming a general spatial pattern of “North City, Central Park, and South Lake”. Figure 7, which illustrates the spatial distribution of the population in 2017 and 2035 under the SSP2 scenario, indicates a notable change in population concentration over time. The pattern of population distribution in the Xiong’an New Area in 2035 is projected to be consistent with the planning characteristics, with the population mainly gathered in the start-up area and peripheral areas.

3.4.2. Population Exposure

The population exposure to rainstorm waterlogging for different return periods is presented in Figure 8. The total population exposure increased as the return period rose. In 2017, the total population affected by rainstorm waterlogging with return periods of 5, 10, 50, and 100 years was 0.31, 0.37, 0.50, and 0.53 million, accounting for 27.7%, 33.0%, 44.1%, and 46.5% of the total population of the Xiong’an New Area, respectively. The population exposure in 2035 is estimated to increase to 1.06, 1.39, 2.31, and 2.52 million under the four return periods, which corresponded to approximately 21%, 27.7%, 46%, and 51% of the total population, respectively. It was observed that the growth rate of population exposure increased with the length of the return periods, with growth rates of 236%, 271%, 361%, and 378% under the return periods of 5-, 10-, 50-, and 100-year rainstorms, respectively. A rapid increase in population exposure is the result of these significant changes in the total population from 2017 to 2035.

The variations in population exposure to different water depths in 2017 and 2035 showed similar characteristics. The majority of the population was exposed to water depths ranging from 0.05 to 0.2 m. Within this range, the proportion of the exposed population to the total population exposure was 93.2%, 89.7%, 74.8%, and 60.6% in 2017, and 94.9%, 91.4%, 84.1%, and 73.6% in 2035 for the 5-, 10-, 50-, and 100-year return period rainstorms, respectively. Conversely, the lowest population exposure was found at inundation depths above 0.6 m, and it accounted for 0.4%, 0.9%, 2.4%, and 3.6% of the total population exposure in 2017, and 0%, 0.05%, 1.3%, and 1.6% in 2035. The percentage of population exposure at the 0.05–0.2 m water-depth interval decreases as the return period increases, while the proportion changed at the water depths of 0.2–0.6 m and above 0.6 m in the opposite direction.

The spatial distribution patterns in population exposure exhibited a consistent trend across the four return periods, as depicted in Figure 9. In 2017, population exposure was low (<100 person-m) in most of the regions, with only a few high-population-exposure (>1000 person-m) areas occurring in urban and rural residential regions. The high-population-exposure areas increased by 16 km², with a rate of 533%, from the 5-year to 100-year return period rainstorms. From 2017 to 2035, the population exposure areas expanded significantly under the four return periods. The areas with high population exposure increased by 25 km², 44 km², 91 km², and 112 km² for the return periods of 5-, 10-, 50-, and 100-year rainstorms, respectively. The increase became greater as the return period lengthened. From the 5-year to the 100-year return period rainstorms, the high-population-exposure areas increased from 28 km² to 131 km², and the proportion of the total area increased from 1.6% to 7.4%. According to future urban planning and development, the residential population will become increasingly concentrated, and more people will migrate to the start-up and peripheral areas of the Xiong’an New Area, which will have a high population exposure in 2035.

4. Discussion

In this study, we utilized the FloodArea two-dimensional hydrodynamic model to simulate the inundation extent and water-depth changes for different return periods of rainstorms. Then, we combined demographic and future projection data to analyze the changing characteristics of population exposure to rainstorm waterlogging. The results indicate that the population exposure to urban waterlogging is expected to increase in the future. Under the conditions of rainstorm events with return periods of 5, 10, 50, and 100 years, the population exposure in 2035 is projected to be 2–4 times higher than in 2017. Our conclusions are consistent with previous research findings [46,47,48]; the population exposed to extreme precipitation and flood events is on the rise. The variation in the total global flood-exposed population can be primarily ascribed to the effects of population change. Factors such as urbanization, sea-level rises, and increased population activity are leading to a growing proportion of the world’s population residing in flood-prone areas [27,49,50]. High population density in urban areas is associated with higher flood risk [51,52], and more people will be affected by floods in the future, inevitably resulting in more severe losses [53,54].

This research provides a new perspective on assessing the population exposure to urban waterlogging. In the analysis of rainstorm waterlogging hazard factors, differently from the existing studies, which often rely directly on the spatial distribution of rainfall or the database of historical flood hazard [26,55], our study considered not only the inundation area, but also the inundation depth. In terms of the analysis of population exposure, our study considered the dynamic change in population from 2017 to 2035, which is different from most existing studies that only consider the static historical population [56,57,58]. However, there are still some aspects to be considered and improved for further studies.

(1) There are some uncertainties within the results of this paper. The uncertainties primarily originate from hydrodynamic modeling and population projection. In this study, only a single urban-flooding model was used to simulate the inundation process, and multiple models could be used in subsequent studies for comparative analyses to reduce model uncertainty. The resolution and accuracy of DEM have also been identified as comprising one of the major contributors to the sources of uncertainty in flood modelling [59,60,61]. Studies have shown that in urban areas, simulation errors are magnified when the DEM resolution exceeds the width and spacing of buildings [62]. Higher-resolution DEM data are more accurate in describing the subsurface and reproducing small-scale flow paths, meaning that flood-simulation results can be expected to have a lower level of error and uncertainty [63]. The DEM selected in this paper has a resolution of 30 m and may not represent particularly dense urban cities well; for example, the grid values may represent the average elevation of low-lying roads/areas and high-rise buildings within that grid. Therefore, in order to identify the effects of microscopic urban surface dynamics, such as fence walls or small topographic obstacles, a more detailed DEM should be utilized. Future research could either directly acquire fine DEM data where available, or explore the synergistic integration of two or more DEM sources to achieve centimeter-level accuracy in topographic data [64], aiming for more precise urban flood simulation results. The rapid development of the Xiong’an New Area also has had a significant impact on the results of the flood simulation. Urbanization has led to increased surface runoff and the construction of urban drainage systems, etc. For instance, due to the early stage of construction in the Xiong’an New Area, the drainage network is not yet fully established, making it difficult to obtain data for this study. Therefore, the influence of drainage has not been considered in this paper, which may also have a certain impact on the simulation results. Urbanization also leads to the rapid expansion of residential areas, industrial parks, and other infrastructure. Studies have shown that regions experiencing rapid urbanization are often prone to flooding [65]. Future urban waterlogging simulations should pay more attention to the impacts of urbanization. In terms of population projection, we exclusively relied on the assumptions of the SSP2. Although we factored in dynamic urban development to estimate forthcoming shifts in the total population, it is crucial to acknowledge that variables such as births, migration, and other determinants, including disease outbreaks, natural disasters, global competition, and technological advancements, will exert an impact on future population trends [66]. As a result, it is advisable for future research to incorporate a range of shared socioeconomic pathway scenarios in order to gauge population changes across different pathways and mitigate the inherent uncertainty in population projections.

(2) The changes in population exposure are not solely dependent on rainstorm waterlogging but are influenced by multiple factors, including climate change effects and population change effects, as well as their combined impacts [26]. This paper mainly considered population change effects. Climate change will increase the precipitation, upstream runoff, and flood risks in the Xiong’an New Area and its surrounding regions [33,67]. Due to the small size of the study area and the low resolution of global climate models, the effects of future precipitation changes on inundation were not considered in this study. Not only will the Xiong’an New Area be affected by local precipitation, due to its geographical location and topographic distribution, but the upstream flooding will also inundate the urban area. Hence, our research results likely underestimate the population exposure to rainstorm waterlogging. To better apply the research results, it is important to consider seeking high-resolution regional climate model data for future projections of local precipitation and simulating the joint impacts of the upstream inflow and local precipitation in the Xiong’an New Area. Meanwhile, calculating the contributions of different factors helps us to understand which factors dominate the changes in population exposure, thereby making disaster-prevention strategies more targeted and effective.

(3) Urban waterlogging risk can be attributed to not only the frequency and magnitude of natural hazards, but also the level of exposure and vulnerability of the disaster-bearing body. For the near future (2021–2040), changes in flood risk primarily depend on changes in exposure and vulnerability [68], which are driven by complex interactions among the climate system, land and hydrological systems, and socioeconomic systems [69,70]. However, this study only analyzed the hazards and population exposure. In the future, we can strengthen vulnerability and rainstorm waterlogging disaster-risk research to better serve urban flood prevention and mitigation.

These research findings can offer specific guidance for urban planning and infrastructure development. Urban planners can craft tailored strategies, including enhancing drainage systems, strengthening flood-resilience building standards, and integrating green infrastructure designs. These measures will ensure that urban infrastructure is better prepared to withstand waterlogging threats, thus reducing potential economic losses. Additionally, this analysis enables the customization of early warning systems and emergency plans. It ensures that individuals in affected areas receive crucial information early on and can take appropriate emergency actions to mitigate potential disaster impacts.

5. Conclusions

Based on the daily precipitation data between 1951 and 2020 from three meteorological stations in the Xiong’an New Area, rainstorms with return periods of 5, 10, 50, and 100 years were calculated by using a generalized extreme value distribution. The changes in inundation depth and extent caused by rainstorms were simulated by the FloodArea two-dimensional hydrodynamic model, and combined with the gridded population data of the Xiong’an New Area in 2017 and in 2035 under SSP2, and the population exposure was assessed. The main conclusions are as follows.

(1) The rainstorm with return periods of 5, 10, 50, and 100 years in the Xiong’an New Area was 84.86, 102.75, 149.26, and 172.33 mm, respectively. For the same return period, the rainfall was the highest in Anxin County, and the lowest in Rongcheng County. With an increase in the return period, the disparity in rainfall in different counties continued increasing.

(2) The depth and extent of inundation increased with the increasing rainstorm return period. The average depths of inundation with return periods of 5, 10, 50, and 100 years in the Xiong’an New Area were 0.11 m, 0.13 m, 0.16 m, and 0.18 m, and the total inundation areas were 207.9 km², 285.4 km², 566.2 km², and 667.2 km², respectively. The most obvious change in inundation area occurred when transitioning from the 10-year to the 50-year return period rainstorms, with a growth rate of 98.3%. The majority of the submerged areas had inundation depths of 0.05–0.20 m. The spatial distribution of inundation water depth showed a similar pattern under four return period rainstorms. The deepest water was mainly found in the main urban areas of the three counties along the low-lying areas of the Daqing River.

(3) The total population exposed to rainstorm waterlogging under 5-, 10-, 50-, and 100-year return periods was 0.31, 0.37, 0.50, and 0.53 million in 2017, and increased to 1.1, 1.4, 2.3, and 2.5 million in 2035 under SSP2, with growth rates of 236%, 271%, 361%, and 378%, respectively. The proportion of population exposure at different water depths varied. The greatest population exposure was found at the inundation-depth interval of 0.05–0.2 m. The proportion of population exposure at the 0.05–0.2 m water-depth interval decreased as the return period increased, while the proportion changed in the opposite direction at the 0.2–0.6 m and >0.6 m water-depth intervals. High-exposure areas were scattered among the urban and rural residential areas in 2017. Due to the concentration of people in the start-up and peripheral areas of the Xiong’an New Area, the areas with high population exposure will appear in densely populated built-up regions by 2035.

Conducting a comprehensive analysis of population exposure is essential for assessing the societal impacts of urban waterlogging risk. This entails identifying which communities, demographic groups, or specific areas are more susceptible to the threat of urban waterlogging. All in all, future urban flood-risk management should pay greater attention to densely populated low-lying areas and extremely frequent heavy rainfall events. An in-depth examination of population exposure related to urban waterlogging provides essential insights to enhance our understanding and management of, and support a reduction in, disaster risks. This not only contributes to improving urban resilience but also safeguards the lives, property, and sustainable development of the Xiong’an New Area.