1 Introduction

Rapid urbanization, particularly land urbanization, has induced significant changes in land use types globally, resulting in a significant disruption of traditional hydrological circulation (Jacobson 2011; Trudeau and Richardson 2016; Pumo et al. 2017). Numerous countries worldwide encounter substantial and complex challenges related to flood control (Balling and Goodrich 2011). Several Chinese cities, particularly megacities such as Beijing, Zhengzhou, Wuhan, and Guangzhou, experience frequent and varying levels of damage from rainstorm waterlogging disasters (Lo et al. 2015; Yin et al. 2015; Lyu et al. 2019).

Traditionally, gray infrastructure plays an important role in mitigating urban flooding as a necessary stormwater drainage system (Bakhshipour et al. 2019). China’s stormwater drainage system is typically designed to handle rainfall events that occur once every 3 to 5 years, and certain locations suffer from issues such as siltation and deterioration (Wang, Liu, et al. 2021). Furthermore, during the processing of runoff, drainage pipelines may generate transient air-water flows that exacerbate flooding (Li et al. 2023). Currently, the concept of exclusively depending on gray infrastructure for rapid stormwater drainage is no longer sufficient to meet the requirements of urban flooding control. With the development of society, various solutions have developed that emphasize the utilization of urban green spaces, such as low-impact development (LID), sustainable drainage systems, and sponge city (Fletcher et al. 2015; Salvadore et al. 2015). These practices aim to maximize the incorporation of green spaces into urban planning using a combination of engineered measures such as green roofs, bioretention cells, dry creeks, permeable pavements, and artificial wetlands, intending to meet the low-impact requirements of the hydrological water cycle for urban development (Pappalardo et al. 2017; Liu et al. 2021). Koc et al. (2021) assessed the performance of seven different LID scenarios and found that the combination of green roofs and bioretention cells can effectively reduce the impact of urban flooding. A study conducted by Tansar et al. (2022) proved that the unified layout strategy of the LID can be effective in enhancing the capacity of urban drainage systems.

Recent studies have indicated that combining gray and green infrastructure has emerged as a promising method for managing stormwater in urban regions (Browder et al. 2019; Xu et al. 2019; Wang et al. 2023). This will boost each strategy’s advantages due to the synchronization effect (Tansar et al. 2023). Many studies have provided valuable insights into the coupled gray–green approach, such as life-cycle cost analysis and resilience assessment of measures (Mei et al. 2018; Islam et al. 2021; Wang, Zhang, et al. 2021), as well as applicability analysis considering future emission scenarios (Wang et al. 2018; Leng et al. 2021; Dougaheh et al. 2023). However, few studies have clarified the precise mechanisms through which gray and green infrastructures impact urban hydrology and their exact role in mitigating urban flooding.

For better analysis, numerical modeling has become an important technique to support urban stormwater management and urban drainage planning. The United States Environmental Protection Agency’s (EPA) Storm Water Management Model (SWMM) is currently the most popular model for drainage networks globally (Gironás et al. 2010; Peng et al. 2021; Ma et al. 2022). It is adopted to investigate urban stormwater processes, design urban drainage networks, and evaluate LID construction (Babaei et al. 2018). In addition, the model can be applied to meet various secondary development requirements (Hsu et al. 2000; Yang et al. 2020). In order to address the limitations of SWMM in calculating surface water flow dynamics, scholars have integrated it with the Geographic Information System (GIS), with the aim of determining the extent and depth of flooding near waterlogged nodes by considering topographic factors (Huang and Jin 2019; Hou and Du 2020). However, integrated SWMM-GIS models often miscalculate the level of flooding due to the inadequate consideration of bidirectional water exchange. Thus, studies have developed the bidirectional linking of one-dimensional (1D) SWMM pipe network models with two-dimensional (2D) surface water flow models (Chang et al. 2015; Bisht et al. 2016; Chen et al. 2018). This approach offers both situational plausibility and operational efficiency (Yu and Huang 2014) and is particularly favored by researchers using open-source modeling.

This study used a SWMM-based hydrological-hydrodynamic model that combines a unit hydrologic model, a pipeline system hydrodynamic model, and a surface hydrodynamic model. By selecting Xiaoguwei Island in Guangzhou City as the case site, we employed numerical simulation to investigate the effects of green infrastructure and gray pipe network modifications on runoff and inundation in the area. In particular, we analyzed the role of structural pipe network modifications in mitigating regional overflow and flooding. This study identified the specific contribution of gray and green infrastructure to urban flooding mitigation, and it provides insights and recommendations for urban construction and development, including a broader utilization of urban green spaces and the adoption of pipe network upgrading initiatives.

2 Materials and Methodology

The primary technique employed in this study involves simulating rainstorm waterlogging using a self-developed model called the integrated hydrology and hydrodynamics urban flood model (IHUM). Considering the local land use retrofit opportunities and recommendations from previous studies, three types of LID practices are chosen, namely, green roofs, sunken green spaces, and permeable pavements. These practices are implemented using a fixed-proportion method to determine the appropriate layout scale for the study area. The gray pipe network renovation relies on the current pipeline capacity analysis, which is used to implement structural problem rehabilitation.

2.1 Study Site

We selected Xiaoguwei Island (Fig. 1), Guangzhou City, China, as the study area. Xiaoguwei Island is an isolated island situated in the Pearl River, with a total area of approximately 18 km2. It is located around 17 km away from the center of Guangzhou City, below the Tropic of Cancer, and experiences a subtropical oceanic monsoon climate. Xiaoguwei Island is subject to consistently high temperatures and humidity throughout the year. The average annual precipitation is about 1923 mm, with the majority of rainfall occurring during the flood season (April to September), accounting for around 81% of the total annual precipitation.

Fig. 1
figure 1

Location of Xiaoguwei Island, Guangzhou City

Full size image

The Guangzhou municipal government aims to ensure that by the end of 2025, the flood prevention and control measures on Xiaoguwei Island will be capable of adequately managing a downpour once every 50 years without resulting in a disaster.Footnote 1 Currently, Xiaoguwei Island has successfully implemented rainfall and sewage diversion, and the development of the sponge city construction is gradually progressing. However, the flood prevention and control measures are yet to reach the intended levels. Therefore, this study focused on Xiaoguwei Island and developed a coupled stormwater management model that examines the rainfall and flood conditions and proposed effective measures to enhance flood prevention and control in the area.

2.2 Integrated Hydrology and Hydrodynamics Urban Flood Model (IHUM)

This study introduced the IHUM for analytical computations. The IHUM is a linked model that allows for 1D and 2D bidirectional flow exchanges between surface flow, sewer flow, and river water bodies, enabling the floodwater to flow freely between the surface and the pipeline, well reproducing the actual scenario. The 1D model utilizes the SWMM model to perform hydrological calculations for rainfall production and confluence, as well as pipeline hydrodynamic calculations. The 2D model is based on shallow water equations and is primarily used to simulate the flow of rainwater overflowing the inspection wells to the surface. These two models are connected through dynamic link libraries.

2.2.1 Storm Water Management Model (SWMM)

The SWMM model was first developed in 1971. After over five decades of development, the model has proven to be globally effective in simulating rainfall and runoff, designing drainage and flood control systems, planning underground pipelines, and implementing low-impact development strategies. In this study, we used the aforementioned multiple functions to conduct analysis on the study site. Prior to the inspection well, SWMM describes the runoff production process and the slope catchment process in the subcatchment using the hydrological method. Once the rainwater enters the inspection wells, it is then transported via the pipe network system and the hydrodynamic method is used to determine the flow rate. This process continues until a superimposed flood flow is generated at the exit.

The Horton model was selected as the infiltration model in SWMM to align with the flow production mechanism in urban areas. The calculation of slope catchment is achieved by generalizing the subcatchment area using a nonlinear reservoir. The runoff outflow from the nonlinear reservoir is described in Eqs. 1 and 2. Equation 1 is the water balance equation, while Eq. 2 is the runoff outflow from the subcatchment after incorporating Manning’s equation:

$$ \frac{\partial y}{{\partial t}}A = A\left( {I - f} \right) - Q $$
(1)
$$ Q = {{1.49W\left( {h - h_{p} } \right)^{5/3} S^{1/2} } \mathord{\left/ {\vphantom {{1.49W\left( {h - h_{p} } \right)^{5/3} S^{1/2} } n}} \right. \kern-0pt} n} $$
(2)

where y is the surface runoff depth (mm), A is the watershed area (km2), f is the infiltration rate (mm/s), I is the rainfall intensity (mm/s), Q is the outflow rate (m3/s), W is the characteristic width (m), n is the mean value of Manning’s coefficient, S is the average slope, h is the depth of the reservoir (m), and hp is the depth of the puddle storage (m).

The progression of fluid movement in the piping system is determined by solving the 1D Saint-Venant equations through the dynamic wave method. The topology created by the inspection wells and the pipes serves as the transport medium for the pipe flow. The governing equations for the dynamic wave are expressed in Eqs. 3 and 4, representing the continuity and momentum equations, respectively:

$$ \frac{\partial Q}{{\partial t}} + \frac{\partial A}{{\partial t}} = 0 $$
(3)
$$ {\text{g}}A\frac{\partial h}{{\partial x}} + \frac{{\partial \left( {Q^{2} /A} \right)}}{\partial x} + \frac{\partial Q}{{\partial t}} + {\text{g}}AS_{f} = 0 $$
(4)

where A is the water cross-section area (m2), h is the water depth (m), Sf is the friction ratio of the drop, and g is the acceleration of gravity (m/s2).

2.2.2 Two-Dimensional Model

Although the SWMM model is economical in terms of computation, it lacks the capability to simulate the spreading and receding of the surface water. As a consequence, it is unable to accurately represent the complete process of urban flooding. The IHUM uses a 2D approach based on the shallow water equations to solve partial differential equations using an implicit double time-step method. This results in the creation of a robust and efficient non-constant-flow 2D model, with the governing equations represented by Eq. 5:

$$ S = S_{b} + S_{f} = \left[ {\begin{array}{*{20}c} 0 \\ {{\text{g}}\left( {h + b} \right)S_{ox} - {\text{g}}hS_{fx} } \\ {{\text{g}}\left( {h + b} \right)S_{oy} - {\text{g}}hS_{fy} } \\ \end{array} } \right] $$
(5)

where b is the bottom elevation (m), u and v are the flow velocity in the x and y directions (m/s) respectively, Sox and Soy are the components of the bottom-slope source term, Sb, in the x and y directions respectively, and Sfx and Sfy are the components of the resistance source term in the x and y directions respectively.

2.2.3 Integrated Hydrology and Hydrodynamics Urban Flood Model (IHUM) Coupling

The vertical water flow interaction of the IHUM model, occurring between the surface and the subface pipeline, is achieved by employing the weir flow and orifice flow equations described in Eqs. 6 and 7, respectively:

$$ Q_{n \to s} = c_{0} A_{mh} \sqrt {2g\left( {H_{{1{\text{D}}}} - H_{{2{\text{D}}}} } \right)} $$
(6)
$$ Q_{s \to n} = \left\{ {\begin{array}{*{20}c} {c_{w} wh_{2D} \sqrt {2gh_{2D} } } & {{\text{if}}\;H_{1D} \le Z_{2D} < H_{2D} } \\ {c_{o} A_{mh} \sqrt {2g\left( {H_{2D} - H_{1D} } \right)} } & {{\text{if}}\;Z_{2D} \le H_{1D} < H_{2D} } \\ \end{array} } \right.\} $$
(7)

where \({{Q}}_{{{n}}\to{{s}}}\) is the overflow flow rate (m3/s), \({{c}}_{0}\) is the orifice flow coefficient, \({{A}}_{{mh}}\) is the area of the node for water storage (m2), g is the acceleration of gravity (m/s2), \({{H}}_{{1D}}\) is the node water level (m), \({{H}}_{{2D}}\) is the surface grid water level corresponding to the node (m), \({{Q}}_{{{s}}\to{{n}}}\) is the node flow rate (m3/s), \({{c}}_{{w}}\) is the weir flow coefficient, \({{w}}\) is the circumference of the node or the width of the rainwater outlet (m), \({{h}}_{{2D}}\) is the surface water depth (m), and \({{Z}}_{{2D}}\) is the ground surface elevation (m).

2.3 Simulation and Optimization

In order to meet the requirements of waterlogging control on Xiaoguwei Island, this study attempted to determine the appropriate type and scale of LID measures as well as a system strategy for upgrading and reconstructing the pipeline network for the study area.

2.3.1 Integrated Hydrology and Hydrodynamics Urban Flood Model (IHUM) Construction

The pipe network and topography information of the study area was acquired through surveys and visits to local hydrological units, ensuring the reliability of the data sources and quality. We did not have access to the measured runoff data for Xiaoguwei Island. As a result, the parameter transplantation method (Yang et al. 2023) was utilized alongside the integrated runoff coefficient method to determine the model parameters. Parameter transplantation includes two techniques—similar transplantation and nested transplantation. Similar transplantation involves transferring parameters from an area with available information to an area without information based on similarities in watershed characteristics such as precipitation, temperature, humidity, and other meteorological factors (Samuel et al. 2011). This technique considers surface data such as land use, slope, and other relevant factors. Nested transplantation needs to fulfill the condition of nesting small watersheds within large watersheds, assuming that they have similar meteorological conditions. This technique only requires the consideration of topographic characteristics, such as slope, and soil texture and land use. If these features are similar, the hydrographic parameters can be transferred.

In this study, the hydrological parameters of the SWMM model were applied to the entire study area using the nested transplantation method. After conducting fieldwork and GIS analysis, it has been determined that Xiaoguwei Island has a relatively uniform elevation, a consistent topographic slope, and stable soil conditions. Xiaoguwei Island, which is the main part of Guangzhou University Town, hosts 10 universities. The land use patterns within these universities are mostly similar, making the natural background conditions of this study area suitable for using the nested transplantation method. More specifically, the hydrological parameters of part of the South China University of Technology campus and its surrounding area (SCUT), which is located on the eastern side of Xiaoguwei Island and spans approximately 2 km2, were extended to cover the entire study area. The pipeline network in SCUT is self-contained, allowing it to be treated as a separate analysis area for model construction and the determination of model parameters. The resulting subcatchments delineated by the Thiessen polygon were organized according to the topography and water-catchment conditions in the region, using the road network and buildings as boundaries. Table 1 presents the parameters of the SCUT SWMM model after the parameter calibration. The parameters were incorporated into the SWMM model of Xiaoguwei Island (Fig. 2), which comprises 1718 pipelines, 1704 wells, and 717 subcatchments.

Table 1 Storm Water Management Model (SWMM) parameter settings
Full size table
Fig. 2
figure 2

Storm Water Management Model (SWMM) of Xiaoguwei Island

Full size image

The 2D model calculations in IHUM utilize a grid system and depend on predetermined boundary constraints. The study area was partitioned into 18,903 grids through the extraction and analysis of elevation and land use/land cover data. This partitioning was employed to simulate the entire process of overflow events and generate data on water depth and volume within each grid. The coupled model calculation has a step size of 0.5 second and is run for a total of 4 h.

2.3.2 Scenario Setting and Optimization

The storm intensity equation (Eq. 8) was applied to calculate the design storm process in Panyu District, Guangzhou City. The rainfall duration was set at 2 h, and the Chicago rain pattern (Keifer and Chu 1957) was used as a distribution of rainfall processes. Based on the rainfall pattern in the study area, the rain peak coefficient, r, was determined to be 0.415. Then, this allowed for the establishment of the designed storm processes with return periods of 1-year (1a), 2-year (2a), 5-year (5a), 10-year (10a), 20-year (20a), 50-year (50a), and 100-year (100a) in the study area (Fig. 3).

$$ q = 2458.657(1 + 0.476\;{\text{In}}\;{{P)/}}\left( {t + 8.873} \right)^{0.749} , $$
(8)

where q is the designed storm intensity (L s−1 ha−1), P is the designed return period (year), and t is the rainfall duration (min).

Fig. 3
figure 3

Designed rainstorm processes

Full size image

The practical limitations of LID practices prevent any one measure from covering the entire watershed based on the specific site type. The feasibility of implementing a green roof is contingent upon various factors, including the roof’s slope, load-bearing capability, and permeability. Permeable pavements should consider the pavement’s bearing capacity, as it is not suitable for handling the heavy load of a freeway, which may result in failure and destruction. To prevent secondary disasters, the implementation of sunken green spaces should be kept at a horizontal distance of at least 3 m from the building. In this study, we adopted these three LID measures as they are popular. The measures are implemented using the fixed proportion method for the entire basin subcatchment area, with a deployment ratio of 40%. The LID parameter design for the SWMM model (Table 2) is based on the values provided in the model user manual (Rossman 2015). This allows for the identification of seven deployment programs, including individual deployments, combinations of two, and combinations of three (Table 3).

Table 2 Low-impact development (LID) parameter design for the Storm Water Management Model (SWMM)
Full size table
Table 3 Low-impact development (LID) program design for the study site
Full size table

By evaluating the hydraulic properties of congested nodes, pipe segments, locations prone to overflow inundation, and other relevant factors, we can identify hydraulic problems in the study area. Examples of the typical causes of waterlogging in the context of pipe networks in gray infrastructure are as follows: inadequately built internal pipelines and lower elevation compared to surrounded pipes; inadequate water collection systems, such as a small amount of rain grates, siltation, or destruction; insufficient capacity of local pipelines to drain water, such as a large pipe connected to a small pipe and subsequently connected to a larger pipe or local damage to the pipeline; and insufficient capacity of the main pipeline to drain water. By utilizing the IHUM model results of the waterlogged area and water depth information, along with the SWMM model outputs of the maximum overflow node and pipe section longitudinal profile, the waterlogged points can be identified and the causes of waterlogging can be analyzed. This allows for the implementation of specific local gray infrastructure renovations, such as modifying the pipe diameter and rectifying errors in connecting pipes of different sizes.

3 Results

This section analyzes the effects of LIDs and gray pipe network modifications from multiple perspectives. Additionally, it offers a detailed discussion and interpretation of these results.

3.1 Assessment of the Effectiveness of Low-Impact Development (LID) Measures

Low-impact development measures impact the runoff process when rainwater drops onto subcatchments. In this study, the IHUM model was used to evaluate the impact of LID deployment scenarios on reducing runoff and node overflow volume.

Figure 4 illustrates the runoff reduction effects of the seven LID placement scenarios. As the frequency of the designed storm rises, the rainfall intensity and runoff volume also increase. Consequently, the effectiveness of different LID practices in reducing the overall runoff volume decreases. Among the measures, G7, which combines three different measures, achieves the highest decrease in runoff throughout each rainfall scenario. In particular, it is able to reduce runoff by up to 41.7% during a 1-year storm and still achieves a reduction of 25.89% under the 100-year designed storm. The initial three LID deployment scenarios demonstrate the efficacy of the LID practices. The results indicate that the G2 green roofs have a higher runoff treatment volume compared to the G3 sunken green spaces in the study region. In addition, the G1 permeable pavements have the lowest treatment volume. G4, G5, and G6 indicate the efficacy of the two-by-two combination of deployment scenarios. G6 exhibits a higher treatment volume compared to G4 and G5 across all rainfall intensity levels. This is because G6 combines green roofs and sunken green spaces, which aligns with the treatment volume findings observed in the first three programs. Overall, the most effective approach in reducing the overall runoff rate is determined as the retrofitting of green spaces, roads, and building roofs for each subcatchment within the entire watershed. Thus, the green retrofit option G7 is used for the study region.

Fig. 4
figure 4

Comparison of runoff reduction rates for seven low-impact development (LID) programs under different rainfall scenarios

Full size image

We integrated G7 into the IHUM model to calculate the reduction in the node overflow caused by the green retrofit compared to the current situation (Table 4). Figure 5 presents the overflow processes. The results indicate that there is a general drop in the node overflow reduction rate with increasing rainfall intensity. Over the first four return periods, the overflow is reduced by approximately 50% compared to the status quo. The occurrence of the maximum overflow may increase the risk of flooding in the study area. As the rainstorm return period increases, the moment gradually advances.

Table 4 Node overflow statistics for low-impact development (LID) program G7
Full size table
Fig. 5
figure 5

Hydrographs of inundation volumes for the low-impact development (LID) program G7

Full size image

3.2 Assessment of the Effectiveness of the Gray Pipe Network Renovation

The gray pipe network modification aims to address the issue of local overflow that occurs after runoff is flowing into the pipe network system, as well as the subsequent reintroduction of surface water back into the pipe network system. To evaluate the efficacy of localized pipe network alteration, we examined both qualitative and quantitative results.

Figure 6 compares the inundation results in two areas. In particular, the top row illustrates the levels of inundation at two sites under the current scenario, while the bottom row presents the effects of implementing gray retrofit measures in the areas across different recurrence periods. The flooding extent and floodwater depth in the two affected areas are greatly reduced following the implementation of the pipe network retrofit. In addition, the level of flood mitigation varies among different locations. We also determined the decrease in the highest level of overflow following the implementation of gray retrofitting (Table 5). The results reveal that renovating the localized pipe network system in flood-prone areas has reduced the maximum overflow volume by 26.19% to 15.24% for rainstorm return periods ranging from 1 year to 20 years. This implies that upgrading and enhancing the capacity of the pipe network system in a limited area can effectively reduce flooding. However, it also indicates that the small-scale gray pipe network modifications have limited impact on extreme storms. Therefore, we attempted to integrate gray and green strategies to assess the efficiency of the combined methods in managing stormwater for urban flood mitigation.

Fig. 6
figure 6

Comparison of partial flooding before and after the application of gray modifications

Full size image
Table 5 Maximum overflow reduction rates with gray retrofits under different return periods
Full size table

3.3 Assessment of the Effectiveness of Comprehensive Gray–Green Practice

We conducted a coupled gray–green facility practice and assessed the impact of the enhanced measures on rainfall and flood management in the study area. The outlet J61 was chosen to examine the flow of water through the pipeline during a rainfall return period of 1 year (Fig. 7).

Fig. 7
figure 7

Comparison of outflow processes for different retrofitting methods at a typical outlet (J61) under a 1-year return period rainfall

Full size image

Based on Fig. 7, the area covered by the curves reflects the overall volume of outflow. The total outflows are approximately the same for the status quo situation and the gray retrofit, and approximately the same for the green retrofit and the gray–green retrofit. Furthermore, the first group contains a larger quantity of water compared to the second group. The level of peak flow reflects the drainage capacity of the pipeline to a certain extent. The peak of the curve associated with the status quo situation is not high, but it remains so for an extended duration, suggesting that the pipeline system has been consistently operating at maximum capacity for a long time. The shape of the flow curve for the green retrofit and the status quo situation exhibits overall similarity, although the time at which the peak occurred is postponed. The gray renovation has significantly increased the drainage capacity of the pipe network, with a peak of 16.86 m3/s, demonstrating its ability to rapidly drain rainwater. The peak flow rate of the integrated gray–green retrofit is 11.11 m3/s, which is much lower compared to the gray pipe network renovation. Meanwhile, it is considerably higher than both the current state and the green retrofit. Implementing a gray–green retrofit can simultaneously decrease water volume and enhance the drainage capacity of the pipe network. This helps alleviate pressure on the drainage pipes and ensures a smoother flow of water.

Table 6 reports the statistics of the number of pipes that have experienced pipe overloading for more than 2.5 h. The results indicate that integrated gray–green retrofit is superior to gray retrofit, while gray retrofit is superior to green retrofit. Note that while the gray retrofit does not reduce the volume of water entering the pipe network system, it does lead to a substantial increase in the drainage capacity of the pipeline system (Fig. 7). This results in a higher maximum flow rate that can be discharged, a significant improvement in the overload condition of the pipes, and a more pronounced recovery of the pipe capacity.

Table 6 Statistics of the number of pipes with a surcharging duration ≥ 2.5 h under different rainfall return periods with various retrofitting methods
Full size table

Figure 8 compares the distribution of flooding under the gray–green retrofit scenario to the current condition over several return periods. The overflow water volume, flooding extent, and water ponding level all decreased following the implementation of the gray–green retrofit. In general, the gray–green retrofit demonstrates a higher ability to control stormwater flooding compared to the other measures.

Fig. 8
figure 8

Inundation for coupled gray–green retrofit versus current state under different rainfall scenarios

Full size image

The above results highlight three important aspects: (1) Green infrastructure primarily functions by improving the infiltration, retention, and storage of stormwater runoff, thus reducing overflow and surface water accumulation. (2) The effect of gray pipe network renovation is to improve the water transmission capacity of the pipe network, serving as a “fast drainage” system for rainwater. (3) Both gray and green retrofitting strategies can be implemented at current locations with a limited scale and intensity. They are more effective in dealing with small return periods rainfall and flooding events. By integrating the gray and green facilities, we can effectively boost their flood management performance.

4 Discussion

Parameter transplantation is a crucial technique used for hydrological analysis and parameter estimation in hydrological models. It is particularly useful for areas or watersheds where there is limited information available. Inter-nested basins and catchments are commonly seen as indicative of hydrological similarity (Gottschalk et al. 2011; Müller and Thompson 2015; Betterle and Botter 2021). Nevertheless, small catchments nested inside larger catchments may exhibit greater variability (Botter et al. 2013), characterized by heightened variations in terms of changes in topographic slope and elevation when compared to larger catchments. Based on comparisons using GIS and field exploration, we determined that the topography and geomorphology between the two areas (Xiaoguwei Island and the SCUT area inside the island) in this study are generally consistent, which allows for the appropriate transplantation of parameters.

The treatment volume of LID measures differs from the treatment capacity. Using the current study site as an example, the predominant features within the area are buildings and green spaces, with a small percentage allocated to roads (49.113% for buildings, 32.757% for green spaces, and 6.726% for roads). Therefore, the impact of retrofitting can be more effectively showcased by applying the same proportion to both building roofs and green spaces (G2, G3, and G6). Nevertheless, by examining the correlation between these reduction rates and the corresponding area percentages, it is possible to presume that, in the scenarios investigated in this study, the capacity to reduce runoff follows the order of permeable pavements, sunken green spaces, and green roofs. Permeable pavements have a remarkable ability to reduce runoff, which agrees with the findings of Brattebo and Booth (2003). Additionally, we employed a rainfall category with a medium rainfall peak (rain peak coefficient of 0.415) in this study. Qin et al. (2013) also indicated that permeable pavements are more effective than sunken green spaces and green roofs in this rainfall pattern.

Treatment systems that combine multiple LID solutions can effectively manage runoff (Eckart et al. 2017). Generally, prior to designing a LID practice framework, it is crucial to consider the specific characteristics of the local urban environment, including the layout of buildings and roadways, as well as the permeability of the soil (Wang et al. 2019; Liu et al. 2021), for effective compensation of the impact of natural vegetation loss caused by construction and to establish suitable pathways for water and air. The three LID measures used in this study were validated in the physical pilot study conducted by Mai et al. (2018), demonstrating the applicability of these measures for typical red soils found near the Tropic of Cancer, including Guangzhou City.

The results pertinently indicate that the integration of LID with gray infrastructure is often a more effective option in achieving stormwater control objectives, as supported by previous studies (Liu et al. 2023; Wang et al. 2023). LID and drainage network systems have distinct functions in stormwater treatment. Low-impact development practices aim to decrease the overall amount of water that enters the pipeline system, resulting in a notable alleviation in pipe pressure. Gray infrastructure retrofits target specific areas in the network where pipe capacity is limited and requires enhancement. Hence, the integration of LID and infrastructure renovation significantly contributes to the reduction in flooding.

Given the unique characteristics of the studied area, the approaches for implementing LID measures and upgrading gray infrastructure, as suggested in this study, may vary from those applicable to other locations (Neupane et al. 2021). However, the idea developed in this study about implementing LID and upgraded gray pipe networks for the control of runoff and prevention of overflow can be applied to any site. By engaging urban planners, engineers, and stakeholders, it is possible to successfully integrate the elements of urban water safety and security, such as LID and pipe network upgrading, into the community environment. Prior to implementation, it is essential to conduct a comprehensive evaluation and analysis of the suitability and long-term sustainability of these measures in the specific area. Ultimately, this will lead to a considerable enhancement in urban resilience and social well-being.

5 Conclusion

This work introduced the IHUM model as a tool for numerically simulating flooding on Xiaoguwei Island in Guangzhou, China. Two measures were considered for urban flood control—LID and pipe network renovation. In order to identify the most effective strategy for mitigating flood threats in the study area, various scenarios of LID deployment and the repair of structural defects in the pipe network were examined. The primary findings can be summarized as follows: (1) The effectiveness of several LID techniques in mitigating floods decreased as the overall runoff quantity increased. (2) Implementing specific upgrades to the local pipe network will significantly reduce the extent and depth of flooding. (3) The advantages of integrating gray and green strategies for managing runoff and flooding are apparent, as they not only decrease the volume of water entering the pipeline network but also possess a higher capability to convey stormwater without overflowing to the surface.

This study offered novel perspectives on the potential interaction between urban pipe network systems and LID measures. It highlighted the significance of addressing structural issues in pipe networks to facilitate area drainage. Additionally, it verified the mutual benefit between pipe network retrofitting and LID measures in the treatment of stormwater. It is advisable for future urban land use planning to include space for LID implementation; and after conducting proper topological and economic evaluations of the pipeline network, its upgrading could be prudently conducted.

Admittedly, there are several limitations that exist in this study. They arise from multiple factors that influence the design and optimization of LIDs, which include, but are not limited to, deployment size, location of the deployment point, soil conditions, cost effectiveness, and future climatic change. Similarly, the upgrading of pipeline networks needs to take into account practical factors, such as socioeconomic impacts. Furthermore, the gray–green retrofit scenario proposed in this study was derived from a region-specific analysis and is subject to constraints in terms of its applicability to other regions. Additionally, this study did not incorporate blue water bodies into the model for comprehensive consideration. In future studies, it will be necessary to integrate the subjective and objective factors of multiple LID deployments and gray infrastructure upgrading in order to develop a comprehensive retrofit framework, and some socioeconomic impacts of the integration can be explored.

The task of urban flood control and drainage is one that requires integrated planning and the collective efforts of various stakeholders. To establish a robust blue-gray–green infrastructure protection system, local authorities and communities need to reach a consensus on the installation, operation, and maintenance of the water safety and security system, to ensure a more resilient and sustainable urban living environment.