**1. Introduction**

Urbanisation and changes in land use have had a considerable impact on the processes and elements of the water cycle [1,2], whereas in recent years, global climate change and extreme weather have occurred frequently, and urban flooding caused by heavy rainfall has gradually become a hot topic of concern for scholars [3]. The frequent occurrence of urban flooding disasters has brought substantial economic losses and casualties to society and seriously threatened urban public safety [4]. Therefore, it is essential to understand the risk areas of urban flooding and conduct flood risk assessments to prevent and control urban flooding and reduce the losses caused by such disasters [5].

A common method for conducting an urban flood risk assessment is the historical disaster statistics method [6], which focuses on using statistical methods to analyse the development pattern of flooding, predict possible future flooding hazards, and estimate possible losses due to flooding based on historical flooding information and rainfall data

**Citation:** Wei, H.; Zhang, L.; Liu, J. Hydrodynamic Modelling and Flood Risk Analysis of Urban Catchments under Multiple Scenarios: A Case Study of Dongfeng Canal District, Zhengzhou. *Int. J. Environ. Res. Public Health* **2022**, *19*, 14630. https:// doi.org/10.3390/ijerph192214630

Academic Editors: Luis Garrote and Alban Kuriqi

Received: 25 August 2022 Accepted: 3 November 2022 Published: 8 November 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

369

in the study area [7]. For example, Hans de Moel et al. analysed trends in flood risk in time and space and made projections for future land use and flood inundation risk in the Netherlands. Their findings show that over spatial spans, flood losses are greater in areas of high economic growth than in areas of low economic growth. However, high-economicgrowth areas are more resilient to flood risk than low-economic-growth areas [2]. Man Qi et al. analysed the effects of topography, rainfall, and impervious surfaces on urban flooding and their spatial patterns of variation for four recent storm events in Cincinnati, USA. They used the kriging interpolation of estimated rainfall depths to measure the impact of rainfall on urban flood hazards [8].

With the development of computer technologies, hydrological hydrodynamic simulation methods are increasingly being applied to urban flood risk assessment. Some of the more widely used numerical models in urban hydrological simulations are SWMM, MIKE, and InfoWorks ICM [9]. For example, Zhao et al. used the coupling of two models, SWMM and MIKE21, to simulate in detail the spatial distribution and water depth of the inundated area in the central part of Cangzhou City under different rainfall return periods and evaluate the economic losses from flooding in the risk area [10]. Sidek et al. used the InfoWorks ICM hydrological-hydraulic model of the Baisala basin as an example to model the response of the basin to rainfall based on the Probabilistic Distributed Moisture (PDM) model to generate flood hazard maps based on several average repetition intervals (ARI) and uniform rainfall depths and analyse the main influences affecting the flood depth and extent [11]. Tabari et al. used the InfoWorks ICM hydraulic model to quantify the impact of anthropogenic climate on urban rainfall flooding in Antwerp, Belgium, using a risk assessment framework and causal counterfactual probability theory [12]. Compared to the analysis of flood risk through historical hazard scenarios, hydraulic modelling provides a more accurate risk assessment method and allows for a more comprehensive analysis of flood risk conditions [13–17]. The existing models work well on a large regional scale. For urban areas, flood risk modelling needs to be more precise, for example, down to a particular road or square. However, the lack of monitoring of relevant data at the urban scale, such as check-well level data and data from drainage networks, has led to only a few academics working on flood risk models for small urban catchments [16,18–26], so this paper constructs an urban flooding model using the Dongfeng canal area in Zhengzhou City as an example to analyse the flood risk within this small urban catchment.

The objectives of this study are (1) to comprehensively consider the hydrological processes between river–urban drainage system–surface runoff and construct a 1D/2D coupled urban flood model based on InfoWorks ICM and an analysis of urban flood processes; (2) to use the storm intensity formulae to design different rainfall scenarios; analyse the inundation depth, duration of inundation, and inundation extent under different recurrence period design storms; and analyse the overflow distribution and drainage capacity of the pipe network at the nodes; and (3) to construct an urban flood risk assessment system to analyse the urban flood risk for the study area.

#### **2. Materials and Methods**

#### *2.1. Study Area*

Zhengzhou is the capital of Henan Province, located in the north-central part of Henan Province, where the middle and lower reaches of the Yellow River divide, between longitude 112◦42 –114◦14 E and latitude 34◦16 –34◦58 N. Zhengzhou is mostly a plain, except for the hills in the southwest, and the terrain is flat, with elevations generally less than 284 m, the lowest being only 79 m. There is a difference of 205 m between the highest and lowest points in the territory.

The Dongfeng Canal drainage area of Zhengzhou City was selected as the study area for this study. This study area is located in the northern part of the main urban area of Zhengzhou. The study area covers an area of approximately 80 km2.

The study area has a temperate continental monsoon climate with an average annual precipitation of 632.4 mm and an average of 78 days of precipitation per year [27]. The extreme annual maximum rainfall is 1339 mm and the extreme annual minimum rainfall is 380.6 mm, with rainfall concentrated between June and August each year and the heaviest rainfall occurring in August.

The main major river network in the study area is the Dongfeng Canal. The Dongfeng Canal is a man-made river that was dug in 1958. Originally used as a channel to divert water from the Yellow River for irrigation, it now fulfils important functions in flood control, ecology, and landscape. The Dongfeng Canal starts at the Yellow River embankment and joins the Sosu and Jalu rivers to the south. At the same time, the Dongfeng Canal also intersects with a number of tributaries in the city centre. The Dongfeng Canal is one of the most important north–south oriented rivers in the city and has the task of draining flood water. The total length of the stormwater pipe network laid in the study area is 295.62 km. Stormwater in the study area is mainly discharged into the Dongfeng Drainage Canal through the stormwater pipe network laid on arterial roads such as the North Third Ring Road, Garden Road, Zhongzhou Avenue, East Yellow River Road, and East Dongfeng Road. Figure 1 shows the location of the study area, the distribution of the stormwater pipe network, and the elevation schematic.

**Figure 1.** Location of the study area, stormwater network distribution, and elevation data.

#### *2.2. Data Collection and Manipulation*

The basic data required to build the hydrodynamic model during the study included the following: road-building data, stormwater pipe network and node distribution data, 5 m accuracy DEM data, and land use/land cover data.

Road construction data were obtained from OSM (https://www.openstreetmap.org (accessed on 6 April 2022)). Stormwater pipe network and node data with a 5 m accuracy provided by Zhengzhou Planning and Survey Design Institute were used for the construction of the 1D stormwater pipe network and 2D surface diffuse flow model underlying the study area.

Land use data were classified using ArcGIS PRO for the supervised classification of satellite image data. Image data were obtained from the Geospatial Data Cloud (www. gscloud.cn (accessed on 6 April 2022)), Landsat8 OLI-TIRS remote sensing imagery [28]. Based on the current land use situation in the study area and the needs of the study, the land use in the study area was classified into five categories: building land, green space, water bodies, bare soil, and roads.

The river network data in the study were river shapes determined using satellite imagery and DEM data were used to extract features from river cross-sections.

#### *2.3. Research Methodology*

2.3.1. InfoWorks ICM Hydrodynamic Modelling

#### (1) Basic theory

This study integrates the water exchange between the pipe network and the twodimensional surface and river channels and uses InfoWorks ICM software to construct a one-two-dimensional coupled urban flood model [29].

The model is mainly concerned with hydrohydraulic processes such as rainfall, surface runoff, and the drainage of the pipe network [30]. The Infoworks ICM model is used to simulate the diffusion and transport of water in pipes by completely solving the system of St.Venant equations with the control equations in Equations (1) and (2).

$$\frac{\partial A}{\partial t} + \frac{\partial \mathbb{Q}}{\partial \mathbf{x}} = 0 \tag{1}$$

$$\frac{\partial A}{\partial t} + \frac{\partial}{\partial x} \left( \frac{Q^2}{A} \right) + gA \left( \cos \theta \frac{\partial h}{\partial x} - S\_0 + \frac{Q|Q|}{K^2} \right) = 0 \tag{2}$$

where *Q* is the flow rate, m3/s; *A* is the pipe section area, m2; *t* is the time, s; *x* is the length of the pipe along the flow direction, m; *h* is the water depth, m; *g* is the acceleration of gravity, m/s; *θ* is the horizontal angle in degrees; *K* is the water transfer rate, determined by Manning's formula; *S*<sup>0</sup> is the slope of the pipe bottom.

The Infoworks ICM model generalises the river channel to a piped open channel when simulating the flood evolution of the river network and uses a one-dimensional hydrodynamic model for the simulation [31,32], with the basic control equations being

$$\frac{\partial A}{\partial t} + \frac{\partial \mathbb{Q}}{\partial \mathbf{x}} = q \tag{3}$$

$$\frac{\partial A}{\partial t} + \frac{\partial}{\partial x} \beta \left(\frac{Q^2}{A}\right) + gA \left(\frac{\partial y}{\partial x}\right) + gAS\_f - uq = 0 \tag{4}$$

where *Q* is the flow rate, m3/s; *A* is the cross-sectional area of the river crossing, m2; *t* is the time, s; *x* is the horizontal coordinate along the flow direction, m; *y* is the water level, m; *g* is the acceleration of gravity; *β* is the momentum correction factor in degrees; *K* is the water transfer rate, determined by Manning's formula; *Sf* is the frictional slope drop; *u* is the flow rate of the lateral incoming flow in the river direction; *q* is the lateral incoming flow rate of the river, m3/s.

It uses the two-dimensional finite volume method to solve the shallow water equations in the simulation of two-dimensional surface diffuse flow by using the TVD excitation technique and the Riemann solver to solve the model computationally. The two-dimensional surface model can effectively and accurately simulate the flow of water on complex urban surfaces and provide support for engineering planning and design [31,33]. The shallow water control equations used in the simulation are as follows:

$$\frac{\partial h}{\partial t} + \frac{\partial (hw)}{\partial x} + \frac{\partial (hv)}{\partial y} = q\_{1D} \tag{5}$$

$$\frac{\partial(hu)}{\partial t} + \frac{\partial}{\partial x}\left(hu^2 + \frac{gh^2}{2}\right) + \frac{\partial(huv)}{\partial y} = S\_{0,x} - S\_{f,x} + q\_{1D}u\_{1D} \tag{6}$$

$$\frac{\partial(hv)}{\partial t} + \frac{\partial}{\partial x}\left(hv^2 + \frac{gh^2}{2}\right) + \frac{\partial(huv)}{\partial y} = S\_{0,y} - S\_{f,y} + q\_{1D}v\_{1D} \tag{7}$$

where *h* is the water depth, m; *u* is the velocity component in the x-direction, m/s; *v* is the velocity component in the y-direction; *S*0,*<sup>x</sup>* is the bottom slope component in the x-direction; *S*0,*<sup>y</sup>* is the bottom slope component in the y-direction; *Sf* ,*<sup>x</sup>* is the friction component in the x-direction; *Sf* ,*<sup>y</sup>* is the friction component in the y-direction; *q*1*<sup>D</sup>* is the outflow rate per unit area, m3/s; u1D is the velocity component of *q*1*<sup>D</sup>* in the x-direction, m/s; *v*1*<sup>D</sup>* is the velocity component of *q*1*<sup>D</sup>* in the y-direction, m/s.

#### (2) One-dimensional stormwater pipe network data pre-processing

Before constructing a 1D drainage model of the study area, the raw data were preprocessed, for example, by checking the connections to the pipe network. For areas without a drainage network, the discharge of rainwater directly into the nearby mains network was considered. To reduce the calculated pressure, the stormwater pipe network was generalised. For example, pipes with several branches located in the same catchment area were combined into one drainage pipe based on their drainage capacity. After the simplification, the total length of the stormwater pipes was 295.62 km, with a total of 1837 inspection wells and 69 outlets.

#### (3) Catchment delineation

The overall topographical variation in the study area was not significant and the data from the stormwater network were relatively similar, so the Tyson polygon method was used for sub-catchment delineation. The Tyson polygon method was used to delineate the sub-catchments. All adjacent inspection shafts were joined into a triangle and the perpendicular bisectors of the sides of these triangles were made so that a number of perpendicular bisectors around each inspection shaft formed a polygon. In addition, some sub-basins were manually adjusted according to the layout of the pipe network and field surveys. A total of 1837 sub-catchments were finally delineated.

(4) Determination of model parameters

The flow-producing surfaces were divided into four categories according to land use: building sites, roads, green spaces, and bare soil. Green areas and bare soil are permeable surfaces. The fixed runoff coefficient method was used for impervious surfaces and the Horton method was used for permeable surfaces [32]. The SWMM nonlinear reservoir method was used to simulate the confluence of the study area based on topographic data. The parameters of different types of flow-producing surfaces were also set according to previous research results in similar areas [34–37]. The specific values of the parameters are given in Table 1.

**Table 1.** Model production sink parameters.


(5) Two-dimensional model setup

The gridding interval was set for the study area and the size of the triangular grid was adjusted to meet the study requirements. Considering that the accuracy of the DEM may not reflect the inundation of the road, the grid elevation of the area where the road is located was reduced by 15 cm in order to better simulate the actual road conditions in the study area. In order to be able to reduce the amount of computation as much as possible while meeting the conditions of simulation accuracy, in the modelling process, the calculation grid was encrypted for key areas such as roads, and as large a grid as possible was used for areas with a single land use type. The final triangular grid was divided into 97,112, with a minimum grid area of 20 m2 and a maximum grid area of 1000 m2.
