Impact of Spatial Configuration of Bioretention Cells on Catchment Hydrological Performance Under Extreme Rainfall Conditions with Different Stormwater Flow Paths

Abstract

Bioretention cells (BCs) are widely used to manage urban runoff due to their positive impact on runoff control. Current research primarily focuses on optimizing the internal structural design of bioretention cells, while studies on the interactions between their spatial configuration, topography, and land use types are limited. This study employs the Storm Water Management Model (SWMM) and uses extreme rainfall to analyze the influence of typical stormwater flow paths, determined by various land use types and topography, as well as the spatial configurations of bioretention cells on catchment hydrological performance. The results show the following: (1) Different stormwater flow paths significantly affect catchment hydrological performance, with series-type pathways performing best. (2) The spatial configuration of bioretention cells significantly influences catchment hydrological performance. Decentralized BCs under series-type pathways showed better performance for reducing total outflow and peak runoff, with reduction rates increasing by 7.1% and 8.8%, while centralized BCs better delayed peak times. (3) Stormwater flow paths affect BC efficiency in catchment hydrological performance. Decentralized BCs under a series-type stormwater flow path are recommended for priority use. This study provides a novel perspective for optimizing the spatial arrangement of BCs and urban stormwater management, thereby contributing to flood risk mitigation.

1. Introduction

In recent years, rapid urban development has led to a substantial conversion of permeable areas into impervious areas [1], resulting in a sharp increase in the imperviousness of urban catchments. The increasing imperviousness of urban catchments has limited stormwater infiltration and exacerbated surface runoff. At the same time, recent changes in rainfall patterns, characterized by more frequent and intense extreme precipitation events [2,3], have overwhelmed existing drainage systems, leading to severe urban flooding. In China, southern provinces such as Guangxi, Sichuan, Hunan, and Jiangxi [4], as well as cities like Shenzhen [5], Wuhan [6], and Nanning [7], have reported significant flood disasters. Urban flooding presents a substantial threat to sustainable urban development [8], prompting extensive research efforts focused on urban flood control [9,10,11].

Traditionally, increasing the scale and capacity of grey infrastructure, such as drainage pipes and storage tanks, has been considered an effective means of mitigating flood risks. However, in highly urbanized areas, the implementation of grey infrastructure has faced significant spatial constraints due to the scarcity of land resources [12]. Meanwhile, the substantial construction and maintenance costs of this solution pose significant challenges in terms of economic feasibility and sustainability [13]. Within the broader framework of grey–green infrastructure integration, green infrastructure, as a powerful complement to grey infrastructure, can synergize with grey infrastructure to achieve more efficient stormwater management outcomes, and has become an important component of urban flood mitigation [14,15]. Green infrastructure, as an innovative and sustainable stormwater management measure, has been widely applied in highly urbanized upstream areas in recent years. It simulates natural hydrological processes by employing measures such as vegetative swales, permeable pavements, and bioretention cells. These measures effectively control stormwater runoff, reduce peak runoff, and delay its occurrence [16], thereby alleviating peak drainage pressure on urban drainage systems and enhancing the city’s resilience and adaptability to floods.

As one of the most widely implemented and studied forms of green infrastructure, bioretention cells (BCs) consist of structural components such as a surface layer, a soil media layer, and a drainage layer. They have the capabilities of detention, infiltration, and discharge, enabling effective control of site stormwater runoff [17,18,19,20]. Autixier et al. found that under different rainfall scenarios, the hydrological control performance of bioretention cells in catchment areas varied, with total outflow reduction ranging from 13% to 62% and peak runoff reduction ranging from 7% to 56% [21]. Kumar and Singh found that the higher the percentage of compost in the planting soil media of bioretention cells, the greater the infiltration rate. As the compost proportion increased from 5% to 25%, the infiltration rate of bioretention cells increased by eightfold on bare surfaces and by tenfold on surfaces planted with Calendula officinalis [22]. Dudrick et al. found a positive correlation between the infiltration rate of bioretention cells and plant diversity [17]. Zhang et al. increased runoff reduction rates by 19.1%, 38.7%, and 5.0% through increasing the thickness of the surface layer, soil media layer, and storage layer [23]. Although numerous studies on bioretention cells have been conducted, most focus on optimizing key design parameters of their internal structure to enhance hydrological control performance or address climate change. Limited research has been conducted on how the spatial configuration of bioretention cells interacts with topography and land use types to influence site-scale hydrological control performance, including total outflow reduction, peak runoff reduction, and peak time delay.

In general, the topography and land use types of a site have a significant impact on the flow paths of stormwater runoff, which in turn can lead to considerable differences in the final stormwater runoff volume. Land use types directly influence the generation and accumulation of runoff by altering the infiltration, interception, and detention characteristics of rainwater. Specifically, impervious surfaces (such as concrete, asphalt, etc.) significantly reduce stormwater infiltration, causing stormwater to rapidly form surface runoff, thereby increasing runoff volume and velocity. In contrast, permeable surfaces (such as bioretention cells, lawns, etc.) promote the infiltration and detention of stormwater, slowing down runoff generation and reducing runoff volume [24,25]. Le Floch et al. found that as the impervious area increased, the peak runoff increased by 100%, and the total outflow increased by 20% [26]. These studies indicate that differences in land use types have a significant impact on stormwater runoff. As another key factor, topography influences the flow paths and velocity of stormwater by altering surface characteristics such as slope and direction, thereby affecting the reduction in runoff during its flow process. For instance, natural topography, such as steep slopes, accelerates the flow of rainwater, reducing its detention and infiltration time, thereby increasing runoff volume. In contrast, gentle terrain facilitates the even distribution and infiltration of rainwater, thereby reducing runoff generation. Additionally, artificial topography such as landscaped mounds, curbs, and retaining walls also significantly affects stormwater runoff paths and velocities. Facilities like curbs and retaining walls can alter the direction of stormwater runoff, potentially causing rainwater to accumulate in low-lying areas, thereby increasing the risk of localized flooding. Yao et al. found that the stormwater flow path of roof → road → stormwater inlet generated more runoff compared to the path of roof → lawn → stormwater inlet under the same rainfall conditions [27]. Mao et al. found that redirecting runoff from permeable areas to wetlands instead of directly to the outlet effectively reduced site runoff [28]. Given this, bioretention cells are key structures for regulating stormwater runoff. Their effectiveness depends not only on their own detention and infiltration capabilities but also on the stormwater flow paths influenced by the combined effects of different land use types and topography. Therefore, exploring the interactions between bioretention cells and diverse land use types and topography is crucial for improving the effectiveness of urban stormwater management.

To address this, the present study focuses on typical stormwater flow paths formed by common land use types in source catchments such as roofs, roads, and green spaces, under varying topography. This study aimed to do the following: (1) Systematically analyze the differential impacts of different typical stormwater flow paths on site hydrological control performance. (2) Clarify the effects of the spatial configuration of bioretention cells (including centralized and decentralized layouts) on site hydrologic control performance. (3) Propose optimized configuration recommendations for bioretention cells tailored to different flow paths based on hydrological control benefits. The findings of this study, derived from extensive rainfall simulations, offer a novel exploration of spatial configurations of bioretention cells under varying flow paths and provide practical guidance for their implementation.

2. Methods

2.1. Study Area

As one of the cities with frequent flooding in southern China, Nanning has developed a series of low-impact development (LID) projects to mitigate urban flooding disasters. This study focuses on an old residential area in Nanning as a case study. The area is characterized by outdated infrastructure and limited space available for renovation. As shown in Figure 1a, the study area is located in the Sponge City Construction Demonstration Zone in Qingxiu District. The area covers 7604.6 m², comprising 2209.8 m² of roof, 3485.6 m² of road, and 1909.2 m² of green land, with an elevation ranging from 76 to 84 m (Figure 1b). The terrain is relatively flat with a slight slope.

Stormwater in the catchment is collected by pipes laid along the site and discharged into the municipal drainage network. Through field surveys and engineering drawings, the landscape of the catchment was identified in detail. The main land cover types are roofs, roads, and green spaces (Figure 1c).

2.2. Model Building

2.2.1. Models and Model Settings

Because the stormwater flow paths are determined by the construction conditions, it is not feasible to investigate the potential paths through actual rainfall–runoff monitoring. Therefore, an analysis based on model simulation was conducted in this study. The Storm Water Management Mode 5.2 (SWMM 5.2) developed by the United States Environmental Protection Agency (EPA) was used for modeling [29]. SWMM is widely applied to simulate hydrological processes in urban settings and to design stormwater systems. In this study, the bioretention cells within the LID module of SWMM were utilized for stormwater management. In order to simulate the runoff generation in the study area, data on catchment characteristics and drainage systems such as permeability, topography, land use were collected from the engineering design schemes. Historical rainfall data and some hydrological monitoring data were collected, ranging from 2005 to 2017. The parameters of the bioretention cells used in engineering practices were also gathered to implement stormwater management.

The sub-catchments were delineated into detailed patches based on site elevation, slope orientation, landscape mounds, and the presence of curbs, allowing for a more detailed consideration of the impact of stormwater flow paths on runoff [27].

In this study, the rainfall–runoff processes of the SWMM are simulated using the modified Horton equation for infiltration and the Dynamic Wave method for runoff routing. The initial values of the model parameters were set based on the collected catchment characteristics. All simulations in this study were conducted under a single rainfall event, and thus, the effects of evaporation are not considered.

Regarding the input parameters of the model, physical parameters such as catchment area, percent of impervious area (%Imperv), and slope were directly determined from engineering design schemes. Manning’s roughness coefficients of the pervious and impervious areas were set based on values provided in the SWMM manual, while infiltration parameters were derived from soil permeability experiments. Other parameters such as depression storage for pervious and impervious areas in the catchment were obtained through calibration. Specific parameters are detailed in Supplementary Materials, Table S1.

Three rainfall events were randomly selected for model validation. Nash–Sutcliffe efficiency (NSE) was employed to assess the fit between the observed data and the simulated data, and the model was considered acceptable when the NSE value was above 0.5 [30]. Figure S1 presents the validation results of the model.

$$\mathrm{N}\mathrm{S}\mathrm{E}=1-{\displaystyle \frac{{\displaystyle {\sum }_{t=1}^T{\left(Q_o^t-Q_m^t\right)}^2}}{{\displaystyle {\sum }_{t=1}^T{\left(Q_o^t-\overline{Q_o}\right)}^2}}}$$

(1)

where $Q_{\mathrm{o}}^t$ is the observed value at time t; $Q_m^t$ is the simulated value at time t; and $\overline{Q_o}$ is the observed mean value.

2.2.2. Scenario of Stormwater Flow Paths

The generation and migration of runoff create different stormwater flow paths due to variations in land use types and topography. This difference can influence the hydrological response of the entire catchment. Three types of stormwater flow paths were defined to examine the impact of bioretention cells under different stormwater flow paths on stormwater management, as shown in Figure 2. Different flow paths mean that the runoff generated by the plot flows in different directions. The first type is the series flow path, where runoff flows sequentially through all traversable parcels before discharging into the drainage system; for example, runoff flows from roofs to roads to green spaces and finally to the outlet. The second type is the parallel flow path; for example, runoff from roofs, roads, and green spaces is directly discharged to outlets without flowing through other parcels. The third type is the hybrid flow path; for example, runoff from roofs and roads is directed to green spaces, where it is attenuated and then discharged through the outlet.

2.2.3. Bioretention Cell Design

Scenario with bioretention cells were established in the SWMM to study the hydrological effects of bioretention cells under different types of stormwater flow paths. Since the spatial distribution of LID affects its control efficiency [26,31,32], scenarios for both centralized and decentralized configurations are established for each stormwater flow path. Centralized treatment means that runoff from the catchment is treated by lager bioretention cells near the outfall, while decentralized treatment means that runoff flows into small bioretention cells immediately after generation to be treated before flowing downstream, as shown in Figure 3. For both decentralized and centralized scenarios, the total area of bioretention cells is equal, accounting for 10% of the catchment area. The recommended ratio of bioretention cell area to contributing catchment area is 5–10% [33,34]. Considering that this study was conducted under extreme rainfall conditions, bioretention cells were designed to cover 10% of the catchment area to prevent rapid saturation due to undersized facilities, thereby ensuring effective hydrological control. In the centralized scenario, three larger bioretention cells were strategically placed near the outlet of each sub-catchment area. In the decentralized scenario, a total of 11 bioretention cells were dispersed across the site to manage runoff from individual plots. The specific areas of the bioretention cells under each scenario are detailed in Table S2. The parameters for the bioretention cells were determined from the engineering design drawings, as shown in

Table 1. The parameters of bioretention cells.

Layer	Parameter	Determination Method	Value
Surface	Berm height (mm)	Design drawing	150
	Vegetation volume fraction	Parameter calibration	0.2
	Surface roughness	Parameter calibration	0.19
	Surface slope (%)	Measurement	0.2
Soil	Thickness (mm)	Design drawing	500
	Porosity	Design drawing	0.5
	Field capacity	Parameter calibration	0.2
	Wilting point	Measurement	0.1
	Conductivity (mm/h)	Measurement	18
	Conductivity slope	Measurement	8
	Suction head (mm)	Measurement	80
Storage	Thickness (mm)	Design drawing	300
	Void ratio (voids/solids)	Parameter calibration	0.75
	Seepage rate (mm/h)	Measurement	9.036
	Clogging factor	Not considered	0
Drain	Flow coefficient	Design drawing	2.36
	Flow exponent	Design drawing	0.5
	Offset height (mm)	Design drawing	60

.

2.3. Rainfall–Runoff Description

2.3.1. Rainfall Data

Many studies now use rainfall data based on specific rainfall models (such as the Chicago rainfall model) [11,27,35]. However, due to the significant impact of rainfall patterns on hydrology, using a specific rainfall model may affect the reliability of the results [36,37]. Therefore, in this study, hourly precipitation data from 2005 to 2017 recorded at the monitoring station in the demonstration area of the sponge city project in the subtropical city of Nanning were used. Rainfall events were classified by a rainfall interval of 6 h, and the distribution of rainfall was analyzed according to the amount of precipitation, as shown in

Table 2. The distribution of rainfall events from 2005 to 2017.

Rainfall (mm)	Rainfall Events (Times)	Percentage of Rainfall Events (%)	Total Rainfall (mm)	Percentage of Total Rainfall (%)
≤40.4	1586	94.4	9462.0	54.2
40.4~54.4	31	1.8	1461.9	8.4
54.4~66.5	19	1.1	1189.8	6.9
≥66.5	46	2.7	5297.6	30.5
Grand total	1682	100	17,381.3	100

. It was found that large rainfall events, which account for only 5.6% of the total rainfall events, contributed 45.8% of the total precipitation. The Chinese government requires Nanning to achieve an annual runoff control rate of 70% to 85% [38], with an 85% control rate equivalent to a design precipitation of 40.4 mm. The control rate of runoff below 40.4 mm is 54.2%. However, the effectiveness of runoff control for precipitation exceeding 40.4 mm is uncertain, and flooding is more likely to occur under such heavy rainfall. Therefore, a total of 96 rainfall events were identified from 2005 to 2017, corresponding to design precipitation amounts for annual runoff control rates of 85% to 90%, 90% to 95%, and ≥95%. These rainfall events were input into the model for rainfall–runoff process simulation. Some characteristics of rainfall events are presented in Table S3.

2.3.2. Runoff Descriptors

Total outflow (V) and peak runoff (P) are used as indicators in many studies [26,39]. They were employed in this study to represent the runoff of the catchment without bioretention cells. However, these indicators focus solely on differences in runoff volume and neglect temporal variations in runoff, which may lead to changes in hydrological benefits at larger catchment scales. Therefore, peak time (T) is also considered in this study.

In the scenarios with bioretention cells, the total outflow and peak runoff were compared to the scenarios without bioretention cells to obtain the total outflow reduction rate (R_V) and the peak runoff reduction rate (R_P).

$$R_{V\left(n\right)}={\displaystyle \frac{V_{0\left(n\right)}-V_{\left(n\right)}}{V_{0\left(n\right)}}}\times 100\%$$

(2)

$$R_{P\left(n\right)}={\displaystyle \frac{P_{0\left(n\right)}-P_{\left(n\right)}}{P_{0\left(n\right)}}}\times 100\%$$

(3)

where V_0(n) is the total outflow without bioretention cells for the nth rainfall event, m³; V_(n) is the total outflow with bioretention cells for the nth rainfall event, m³; P_0(n) is the peak runoff without bioretention cells for the nth rainfall event, L/s; and P_(n) is the peak runoff with bioretention cells for the nth rainfall event, L/s.

The peak time latency rate (ΔT) before and after the installation of bioretention cells was also selected as one of the indicators for hydrological control benefits.

$$\Delta T_{\left(n\right)}={\displaystyle \frac{T_{\left(n\right)}-T_{0\left(n\right)}}{T_{0\left(n\right)}}}\times 100\%$$

(4)

where ΔT_(n) is the ratio of changes in peak runoff timing under two scenarios for the nth rainfall event; T_0(n) is the time of peak runoff occurrence without bioretention cells for the nth rainfall event, h; and T_(n) is the time of peak runoff occurrence with bioretention cells for the nth rainfall event, h.

3. Results and Discussion

3.1. The Impact of Different Stormwater Flow Paths on Catchment Hydrology

This section evaluates the impact of stormwater flow paths on the hydrological control performance of the study area. To avoid theoretical errors caused by a single rainfall event, this study uses the statistical mean of multiple rainfall events to analyze the influence of different stormwater flow paths on catchment hydrological performance, thus mitigating the uncertainty of rainfall. Figure 4 shows the simulation results of the three stormwater flow paths for 96 rainfall events. The red dots represent the mean values, which are also displayed numerically. From Figure 4, it can be seen that the hydrological control performance of the catchment varies under different stormwater flow paths.

As shown in Figure 4, for the total outflow of the catchment, the series-type stormwater flow path achieved the best hydrological control performance. However, the total outflow did not decrease significantly, with reductions of only 1.46% and 6.92% compared to the hybrid-type and parallel-type stormwater flow paths. Under the series-type stormwater flow path, stormwater runoff passes through more plots and has the longest overland flow length, resulting in more infiltration losses during the flow process. Therefore, it exhibited the best hydrological control performance in the catchment. However, due to the simulation conducted under high precipitation conditions in this study, the plots quickly become saturated, which led to minimal differences in performance.

For the peak runoff at the catchment outlet (Figure 4), the series-type stormwater flow path achieved the best hydrological control performance, with a reduction of 32.71% and 56.20% compared to the hybrid-type and parallel-type stormwater flow paths. The different stormwater flow paths cause differences in the types and numbers of plots through which the runoff passes. Under the series-type stormwater flow path, stormwater runoff was more intercepted and detained during the confluence process, resulting in a more dispersed surface runoff flowing out from the outlet, leading to the lowest peak runoff at the outlet. Although the percentage reduction in total outflow for the series-type stormwater flow path was small, it still has a significant impact on the peak runoff at the outlet, contributing to its optimal peak runoff performance.

For the peak time at the outlet of the catchment (Figure 4), the peak time under the series-type stormwater flow path arrives more slowly, delayed by 2.58% and 7.28% compared to the hybrid-type and parallel-type stormwater flow paths. The longer flow length and the more complex plot types under the series-type stormwater flow path, along with the stronger interception and detention capacity of permeable plots, compensate for the rapid runoff generation and convergence of impervious plots. This leads to a later arrival of the peak time under the series-type stormwater flow path.

The results of the hydrological performance under different stormwater flow paths indicate that the series-type stormwater flow path is the optimal stormwater flow path. Although the percentage reduction in total outflow of the catchment is small, it can significantly reduce the risks of flooding and associated damages [9]. However, solely designing stormwater flow paths cannot fully mitigate the impacts of high imperviousness caused by urbanization and extreme rainfall induced by climate change. Green infrastructure, such as bioretention cells, must be integrated to fully address these challenges.

3.2. The Impact of Spatial Configuration of BCs on Catchment Hydrology

3.2.1. The Impact of Centralized BCs on Catchment Hydrology

This section introduces the impact of centralized bioretention cells under different stormwater flow paths on catchment hydrology. Figure 5 illustrates the benefits of hydrological control achieved by installing centralized bioretention cells covering 10% of the catchment area compared to scenarios without BCs. Significant hydrological control benefits were observed after installing centralized bioretention cells. Under the parallel-type stormwater flow path, the total outflow of the catchment decreased by 50.8%, the peak runoff was reduced by 36.4%, and the peak time was delayed by 44.0%. Under the hybrid-type stormwater flow path, the total outflow decreased by 53.6%, the peak runoff was reduced by 40.1%, and the peak time was delayed by 38.5%. Under the series-type stormwater flow path, the total outflow decreased by 55.1%, the peak runoff was reduced by 38.7%, and the peak time was delayed by 39.8%.

As shown in Figure 5, centralized bioretention cells under the series-type stormwater flow path performed best in reducing the total outflow of the catchment. Under the hybrid-type stormwater flow path, the centralized bioretention cells exhibited the best performance in reducing the peak runoff at the outlet, while under the parallel-type stormwater flow path, they achieved the best performance in delaying the peak time. High rainfall intensity produces a large volume of runoff in a short period, and the shorter and more numerous flow paths in the parallel type caused runoff to concentrate and reach the bioretention cells more quickly. However, the limited infiltration and retention capacity of bioretention cells restricts the rapid capture of such large runoff volumes [29,40], resulting in the poorest performance in controlling total outflow and peak runoff under the parallel-type stormwater flow path. The series-type stormwater flow path, with the longest flow path, was expected to have the best control performance for total outflow, peak runoff, and peak time. However, the best performance in controlling peak runoff was observed under the hybrid-type stormwater flow path, and the best performance in delaying peak time was observed under the parallel-type stormwater flow path. Combined with Figure 4, this outcome is attributed to the differences in catchment hydrology under different stormwater flow paths in the scenario without BCs. For example, in the scenario without BCs, the parallel-type stormwater flow path had the fastest peak time. Although installing centralized bioretention cells achieved a better delay rate for peak time, the peak time under the parallel-type stormwater flow path still arrived faster than those of other flow paths (

Table 3. Changes in catchment hydrology after placement of centralization BCs.

Runoff Descriptor	Flow Path	No Centralized BC Scenario	Centralized BC Scenario	Reduction/Latency Rate
Total outflow (m3)	Parallel type	549.9	270.6	50.8%
	Hybrid type	521.8	242.1	53.6%
	Series type	514.3	230.9	55.1%
Peak runoff (L/s)	Parallel type	46.89	29.82	36.4%
	Hybrid type	39.84	23.86	40.1%
	Series type	30.02	18.40	38.7%
Peak time (h)	Parallel type	14.4	20.7	44.0%
	Hybrid type	15.1	20.9	38.5%
	Series type	15.5	21.7	39.8%

). While the installation of centralized bioretention cells can partially compensate for the differences in catchment hydrology caused by stormwater flow paths, it cannot fully eliminate these effects.

3.2.2. The Impact of Decentralized BCs on Catchment Hydrology

This section introduces the impact of decentralized bioretention cells under different stormwater flow paths on the catchment hydrology. Figure 6 illustrates the hydrological control benefits achieved by installing decentralized bioretention cells, covering 10% of the catchment area, compared to scenarios without BCs. Similarly to the centralized bioretention cell scenario, installing decentralized bioretention cells resulted in significant hydrological control benefits in the catchment.

Under the parallel-type stormwater flow path, the total outflow of the catchment decreased by 51.6%, the peak runoff was reduced by 37.0%, and the peak time was delayed by 27.3%. Under the hybrid-type stormwater flow path, the total outflow decreased by 53.0%, the peak runoff was reduced by 47.9%, and the peak time was delayed by 15.4%. Under the series-type stormwater flow path, the total outflow decreased by 62.2%, the peak runoff was reduced by 47.5%, and the peak time was delayed by 31.7%.

As shown in Figure 6, decentralized bioretention cells under the series-type stormwater flow path exhibited the best performance in reducing the total outflow of the catchment and delaying the peak time, consistent with expectations. Under the hybrid-type stormwater flow path, decentralized bioretention cells exhibited the best performance in reducing the peak runoff at the outlet, which can be attributed to the higher peak runoff observed in the no decentralized BC scenario (

Table 4. Changes in catchment hydrology after placement of decentralization BCs.

Runoff Descriptor	Flow Path	No Decentralized BC Scenario	Decentralized BC Scenario	Reduction/Latency Rate
Total outflow (m3)	Parallel type	549.9	266.2	51.6%
	Hybrid type	521.8	245.2	53.0%
	Series type	514.3	194.4	62.2%
Peak runoff (L/s)	Parallel type	46.89	29.54	37.0%
	Hybrid type	39.84	20.76	47.9%
	Series type	30.02	15.76	47.5%
Peak time (h)	Parallel type	14.4	18.3	27.3%
	Hybrid type	15.1	17.4	15.4%
	Series type	15.5	20.4	31.7%

). However, decentralized bioretention cells performed poorly in delaying the peak time at the catchment outlet. This may be due to the relatively smaller scale of individual decentralized bioretention cells. Under high-intensity rainfall, the surface layer of decentralized bioretention cells is quickly filled by the large volume of runoff generated in a short time, especially in decentralized bioretention cells built downstream of impervious areas. A significant amount of runoff bypasses the bioretention cells and discharges quickly, resulting in poorer performance in delaying the peak time.

3.2.3. The Comparison of the Impact of Centralized and Decentralized BCs on Catchment Hydrology

This section compares the impact of centralized and decentralized bioretention cells under different stormwater flow paths on catchment hydrology. Figure 7 presents a comparison of the effects of centralized and decentralized bioretention cells on hydrological control performance in the catchment under three stormwater flow paths. The hydrological performance of the catchment is influenced by the spatial configurations of bioretention cells.

For the parallel-type stormwater flow path, decentralized bioretention cells performed better in controlling total outflow and peak runoff, but the differences were minor. Centralized bioretention cells, however, demonstrated a significantly better performance in delaying peak time, improving the peak time delay rate by 16.7%. For the hybrid-type stormwater flow path, centralized bioretention cells outperformed decentralized bioretention cells in controlling total outflow and delaying peak time. While the difference in total outflow control was minimal, centralized bioretention cells improved the peak time delay rate by 23.1% compared to decentralized bioretention cells. On the other hand, decentralized bioretention cells excelled in reducing peak runoff at the outlet, achieving a 7.8% higher peak runoff reduction rate than centralized bioretention cells. For the series stormwater flow path, decentralized bioretention cells showed clear advantages in reducing total outflow and peak runoff, achieving 7.1% and 8.8% higher reduction rates, compared to centralized bioretention cells. However, decentralized bioretention cells had an 8.1% lower peak time delay rate.

The limited surface storage capacity of bioretention cells restricts the capture of large runoff volumes. Decentralized bioretention cells benefit from distributed treatment of runoff, as those located downstream of permeable plots handle less runoff due to reduced runoff generation, thereby demonstrating higher hydrological performance. Yang et al. also found that the performance of LID is spatially heterogeneous, with the heterogeneity related to land cover distribution in a given area [40]. In the series-type stormwater flow path, decentralized bioretention cells located downstream fill more slowly, allowing them to treat more stormwater runoff, which results in a higher total outflow reduction rate compared to centralized bioretention cells. In both series and hybrid flow paths, runoff is intercepted and delayed by decentralized bioretention cells at each plot before reaching downstream plots. This leads to more dispersed runoff outflow, yielding better benefits compared to centralized bioretention cells.

Centralized bioretention cells consistently perform better in delaying peak time across all stormwater flow paths. Runoff generated from each plot is handled individually by decentralized bioretention cells. This causes the surface layers of bioretention cells located downstream of impervious areas to fill up quickly during rainfall events, preventing some runoff from entering the cells and leading to rapid flow to the outlet. In contrast, centralized bioretention cells, designed to accommodate runoff from both permeable and impervious areas, fill up later, resulting in superior performance in delaying peak time.

In the parallel-type stormwater flow path, since there is no significant difference between centralized and decentralized bioretention cells in controlling total outflow and peak runoff, centralized bioretention cells are recommended due to their superior peak time delay capability. In the series-type stormwater flow path, decentralized bioretention cells are preferable for their better performance in reducing total outflow and peak runoff, effectively lowering flood risks and associated damages. In the hybrid-type stormwater flow path, both types of bioretention cells exhibit distinct advantages in peak time delay and peak runoff reduction. The choice of spatial configurations should be based on specific needs and requirements.

4. Conclusions

Through modeling, the catchment hydrology under 96 large rainfall events was simulated to evaluate the impact of stormwater runoff flow paths and the spatial configurations of bioretention cells. The main conclusions are as follows:

(1) Different stormwater flow paths significantly affect the catchment hydrology, with the series type being the optimal flow path. In the case without BCs, the total outflow decreased by 1.46% and 6.92% and the peak runoff was reduced by 32.71% and 56.20% compared to the hybrid and parallel types. The peak time was delayed by 2.58% and 7.28%. However, under higher rainfall intensities, the reduction in total outflow and peak time was limited, and the combined effects of urbanization and extreme rainfall could not be solved solely by optimizing the stormwater flow path.

(2) The spatial configurations of bioretention cells impact the hydrological control benefits in the catchment. Both centralized and decentralized bioretention cells have their respective advantages. Decentralized bioretention cells achieved the best performance for reducing total outflow and peak runoff under the series-type stormwater flow path, achieving 7.1% and 8.8% higher reduction rates compared to centralized bioretention cells. Centralized bioretention cells, however, demonstrated superior performance in delaying peak time, with delay rates higher by 8.1% to 23.1% compared to decentralized bioretention cells.

(3) The stormwater runoff flow path affects the control effect of bioretention cells on the catchment hydrology, and different stormwater flow paths require different strategies. Decentralized bioretention cells under the series-type stormwater flow path showed better performance, and were recommended for priority use. If the stormwater flow path is fixed, there are other recommended cases. In the parallel-type stormwater flow path, centralized bioretention cells, due to their significant peak time delay ability, are more suitable for this scenario. In the hybrid-type stormwater flow path, both centralized and decentralized bioretention cells have distinct advantages in peak time delay and peak runoff reduction, and the appropriate distribution form should be chosen based on actual needs.

This study explores the interaction between the spatial configuration of bioretention cells, topography, and land use types, and examines the impact of bioretention cells spatial configuration on site hydrological control performance under different stormwater flow paths, providing a new perspective for reducing urban flood disasters.