Effect of sponge city construction on urban waterlogging reduction in semi-humid areas of China

With the continuous acceleration of urbanization in China, the over exploitation of the natural environment by human construction and development activities and the lag of the traditional rainwater and flood management system have brought about frequent hydrological problems to cities, such as floods, rainwater runoff pollution, and water resources shortage. Among them, urban waterlogging is one of the most frequent and high-risk disasters in China, which seriously restricts the sustainable development of cities in China (Yin et al. 2015; Sang & Yang 2017; Zhou et al. 2018). China's semi-humid area is about 1.44 million square kilometers, accounting for 15% of the country's land area (Kang & Ge 2019), including 77 cities in 11 provinces. Waterlogging is caused by the rainwater runoff generated by high-intensity rainfall exceeding the design flow of drainage pipes in low-lying areas (Wang et al. 2012a), and peak flow is an important indicator affecting urban waterlogging (Wu 2021). Therefore, the characteristics of heavy and concentrated summer rainfall in semi-humid areas lead to frequent waterlogging disasters. Since the concept of ‘sponge city’ was proposed in April 2012, construction work has been carried out nationwide, and new development has been achieved. In 2015, more than 130 cities had formulated sponge city construction plans (Dai et al. 2018), including 29 cities in semi-humid areas. In 2014, Qian'an City, Hebei Province, was established as the first batch of sponge city pilot cities, and there is still waterlogging in urban built-up areas. Green space can be an important carrier of sponge city construction (Li et al. 2016). By using the additional green space to build low impact development (LID) systems (Li 2019) and giving play to the infiltration, retention, and storage of rainwater, which can significantly reduce the peak flow of rainwater, thereby reducing the pressure of instantaneous drainage from the pipe network and preventing waterlogging at source (Zhang 2010).

In recent years, solving the problem of urban waterlogging through LID facilities has become a research hotspot. Nowogoński (2021) believed that LID technology is an essential step in rebuilding the natural water cycle in urbanized catchment areas. LID systems shift the traditional approach to stormwater runoff control by using decentralized, small-scale control measures to maintain the pre-development hydrology of the site (Baek et al. 2015). Moreover, sunken green space, bioretention pond, vegetation swale, permeable pavement, and green roofs were common LID measures in various studies (Wang et al. 2013; Su et al. 2014; Meng et al. 2018; Zhu et al. 2019). Xiong et al. (2018) took Guangming New District of Shenzhen as a research object to verify that sunken green space, permeable paving, and green roofs can effectively reduce surface runoff and peak flood flow. Yan et al. (2014) took the parking lot in Jinan as an example to compare the hydrological effects before and after adding bioretention pond, and simulated the runoff patterns under different indicators by changing the depth and slope of bioretention pond. Kim & Kim (2021) analyzed the construction project of the Naju-Noan waterfront area. They considered that multiple combinations of LID facilities are effective measures to control stormwater pollution. Jemberie & Melesse (2021) verified that combined LID facilities have a significant effect of up to 75% on urban flood mitigation. Li et al. (2017) used SWMM to simulate and analyze the hydrologic and water quality process of three scenarios before urbanization, after urbanization, and after LID measures. The results showed that with the increase in rainfall return period, the runoff reduced by LID gradually increases, and water pollution is effectively controlled. Wang et al. (2017) took the old urban area as the research area and obtained the influence of different LID facility layout ratios on runoff control based on SWMM software simulation. Zhang et al. (2018a) coupled the SWMM model and the optimization objective function to calculate the economic-benefit optimal solution of the proportion of different LID facilities in the city. Jiang et al. (2021) combined the model simulation to study the spatial optimal allocation scheme of coupling system of low impact development measures, green infrastructure and urban gray infrastructure, and multi-objective cost–benefit optimal curve, thus realizing the benign cycle and sustainable development of urban eco-hydrology.

Sponge city construction in China is still in its early stages. Most research on LID systems to reduce waterlogging focuses on a single dimension of the cell scale, with little consideration and comparison in terms of LID function types and spatial layout at the regional scale (Xiong et al. 2015; Xiang et al. 2017; Zhu et al. 2018). There are few research results on optimizing the LID facility area ratio layout. This paper extended the research to the entire catchment area, mixed various land types in the city, and explored the impact of the combination of functional types and spatial layout on the effect of urban waterlogging reduction.

This paper took the green space of Yanshan South Road and its surrounding areas, a historical waterlogging section of Qian'an city, as the research object. In urban built-up areas, the green space system is stable. The land conditions are relatively tight (Ge & Li 2016), while the attached green space occupies a larger area in urban green space construction and has a more flexible spatial layout, which can be used to build a LID system to treat urban rainwater runoff (Hu et al. 2015). A storm flood management model was established based on mastering detailed primary data, integrating spatial form thinking, and using peak flow reduction as the core indicator. SWMM software was used to study the function type and spatial layout of LID facilities in the additional green space to reduce the waterlogging effect and to find a land-saving scheme with excellent hydrological performance, which will provide reference and suggestions for the design of rainfall-type attached green space in semi-humid areas.

The research area is located in the Hedong New Town Area in Qian'an, the southern section of Yanshan Road. The road width is about 20 m, the horizontal width of the one-sided road green space is between 60 and 510 m, the length of north and south is about 3 km, the total area is about 2.94 km², and a catchment zoning area of 0.63 km² for undertaking external runoff. According to the zoning map of waterlogging risk in Qian'an, there are areas prone to waterlogging in the project plot and surrounding blocks.

According to the geotechnical research report, the topsoil is Loam with a thickness of 0.3–0.7 m; the second layer is Loamy Sand with a thickness of 0.6–4.6 m; the third layer is Arenosols with a thickness of 0.6–5.6 m. In general, the upper soil is relatively loose, the lower soil permeability is better, and infiltration can be used as the primary function of LID facilities.

According to the construction requirements of the sponge city, the storage volume is determined to ensure to meet the stormwater control objectives. We selected four common measures in the two LID facilities, including three types of infiltration LID facilities (sunken green space (SG), bioretention ponds (BP), and infiltration ponds (IP)) and one storage LID facility (reservoir (RE)). Under the premise of ensuring the stormwater runoff control objectives, the four LID facilities were arranged in centralization (CE) and decentralization (DE) space (‘centralization’ layout means that LID facilities are placed only in road-accessory green spaces to concentrate runoff, while ‘decentralization’ layout means that LID facilities are placed in all types of additional green spaces to disperse runoff), and eight single LID facility schemes were initially constructed (Table 1). Then, two of the above four LID facilities were selected and combined by the centralized and decentralized layout in a different order according to the ratio of 1:3, 1:1, and 3:1 of the total runoff reduction target, a total of 72 combined schemes were prepared. This way, the impact of LID facilities' active type and spatial layout on runoff regulation was investigated (Table 2). Taking the reduction of peak flow as the core indicator, the aim is to study the effect of reducing waterlogging of the active type and spatial layout of LID facilities in the attached green space to determine the active type and spatial layout of the LID system.

The runoff coefficient of the comprehensive rainwater volume was calculated using the standard H1 = 29.60 mm of internal runoff control rate in the study area, the standard H2 = 42.60 mm of external runoff control rate of the site the area of each catchment area, and the runoff coefficient of the comprehensive rainwater volume. From the volume method formula, the total amount of runoff inside and outside the site is 62,578.91 m³ (Table 3).

SWMM storm water management model is a rainfall-runoff model based on hydrodynamics developed by the United States Environmental Protection Agency (EPA). It is the most widely used auxiliary design software for storm water management at present. It can conduct a dynamic simulation of hydrology and water quality and generalize the catchment area based on site status. The basic units such as node, catchment area, canal, and outlet are simulated by adjusting attribute parameters. Most importantly, the LID module included in SWMM can further simulate the rainwater concourse situation after site development by adding LID facilities such as Permeable Pavement, Rain Garden, Bioretention Cell, Vegetative Swale, and Green Roof in the sub-catchment partition attribute editor (Mogenfelt 2017). Compared with other models, the SWMM model is more suitable for the urban water system and LID evaluation. It is widely used in LID-related research (Lee et al. 2018).

The study region was generalized according to the current conditions and the dispersion of the municipal network in the surrounding area. The entire region was divided into nine sub-catchment zones, three stormwater network sections, and one end drainage outfall. The LID facilities were set up as a discrete sub-catchment zone for each scenario. The LID facilities were expressed in terms of the property parameters of the sub-catchment zone or the defined LID facilities were set up within the sub-catchment zone and fully covered (Wang et al. 2012b).

SWMM parameters include hydrologic parameters and hydraulic parameters. Hydraulic parameters include the diameter, length, shape, and elevation of rainwater drainpipes, which can be set according to drainage engineering design drawings. Hydrological parameters include the area of sub-catchment, the width of overland flow path, average surface slope, percent of impervious, Manning'N for impervious area (N-imperV), depth of depression storage on impervious area (Dstore-imperV), and so on. In addition, relevant parameters of LID facilities are set in ‘LID Controls’ module, which contains eight types of LID facilities. Some parameters are purely empirical, or empirical parameters with certain physical significance (Peng et al. 2021).

SWMM model uses a variety of methods to simulate surface infiltration, and the Horton infiltration model is generally used for urban areas (Rui et al. 2015), in which the maximum infiltration rate was set to 19.71 mm/min. The minimum infiltration rate was set to 0.50 mm/min. The attenuation constant was 4 h⁻¹, and the Manning coefficients of permeable paving, impermeable paving, shared green space, and rainwater pipes were set to 0.4, 0.014, 0.6, and 0.013, respectively, according to the SWMM user manual. All other parameters are set according to the current situation and relevant specifications, and the specific parameters of LID facilities are listed in Tables 4,⁵⁶–7.

As shown in Table 8 and Figure 3, no matter what the LID scheme is, it can effectively reduce the peak flow in the study area during each rainfall return period, to control the surface runoff in the area and relieve the drainage pressure. In the 1-year and 2-year rainfall return periods, the peak flow of DE-SG and DE-RE was the smallest, which were 1.86 and 2.63, respectively, and the peak flow reduction rate was 90.31 and 89.02%, respectively. In the 3-year, 5-year, 10-year, and 20-year rainfall return periods, the peak flow of the DE-BP scenario was the smallest, which were 3.07, 3.3, and 3.56, respectively. The peak flow reduction rate was 89.07, 90.02, 90.88, and 91.49%, respectively.

In the 1-year rainfall return period, the reduction rate of each scheme to the peak flow in the region from large to small was as follows: DE-SG > DE-RE > DE-BP > DE-IP > CE-BP > CE-IP > CE-RE. In the 2-year rainfall return periods, the reduction rate of each scheme to the peak flow in the region from large to small was as follows: DE-RE > DE-SG > DE-BP > CE-BP > DE-IP > CE-IP > CE-RE. In the 3-year rainfall return periods, the reduction rate of each scheme to the peak flow in the region from large to small was as follows: DE-BP > DE-RE > DE-SG > CE-BP > DE-IP > CE-IP > CE-RE. In the 5-year rainfall return periods, the reduction rate of each scheme to the peak flow in the region from large to small was as follows: DE-BP > DE-RE > DE-SG > CE-BP > DE-IP > CE-RE > DE-IP. In the 10-year rainfall return periods, the reduction rate of each scheme to the peak flow in the region from large to small was as follows: DE-BP > DE-SG > DE-RE > CE-BP > CE-RE > CE-IP > DE-IP. In the 20-year rainfall return periods, the reduction rate of each scheme to the peak flow in the region from large to small was as follows: DE-BP > DE-SG > DE-RE > CE-BP > CE-RE > DE-IP > CE-IP.

Based on the above data, the decentralized layout scheme generally had a better reduction effect on peak flow than the centralized layout scheme. Among them, the sunken green space and reservoir had a better reduction effect on peak flow under a low rainfall return period, consistent with the prior work (Hou et al. 2007; Lin et al. 2016; Zhu et al. 2020). The Bioretention ponds had a better reduction effect on peak flow under a high rainfall return period, which is consistent with the previous studies (Qin 2014). With the increase in rainfall return period, the peak flow reduction rate decreased gradually for all schemes including the infiltration pond.

It can be seen from Figure 4 that, all of the different LID combination schemes can reduce the peak flow and relieve urban water logging problems. In the 1, 2, 3, 5, 10, and 20-year rainfall return periods, the scheme of DE-SG-BP (1:1), DE-IP-SG (1:1), DE-SG-IP (3:1), DE-SG-BP (3:1), DE-SG-IP (1:3), and DE-SG-BP (3:1) had the highest peak flow reduction rate, accounting for 90.26, 92.57, 93.27, 94.28, 95.08, and 95.46%, respectively.

In the 1-year rainfall return period, the top five schemes reduce the peak flow in the region from large to small was as follows: DE-SG-BP (1:1) > DE-IP-BP (3:1) > DE-IP-SG (1:1) > DE-IP-SG (3:1) > DE-SG-IP (3:1). In the 2-year rainfall return periods, the top five schemes reduce the peak flow in the region from large to small was as follows: DE-IP-SG (1:1) > DE-SG-IP (1:3) > DE-IP-BP (3:1) > DE-SG-BP (3:1) > DE-IP-SG (3:1). In the 3-year rainfall return periods, the top five schemes reduce the peak flow in the region from large to small was as follows: DE-IP-SG (3:1) > DE-SG-IP (3:1) > DE-IP-SG (1:1) > DE-SG-BP (3:1) > DE-SG-IP (1:3). In the 5-year rainfall return periods, the top five schemes reduce the peak flow in the region from large to small was as follows: DE-SG-BP (3:1) > DE-SG-IP (1:3) > DE-SG-BP (1:1) > DE-BP-SG (3:1) > DE-IP-SG (1:1). In the 10-year rainfall return periods, the top five schemes reduce the peak flow in the region from large to small was as follows: DE-SG-IP (1:3) > DE-SG-BP (3:1) > DE-BP-SG (3:1) > DE-SG-IP (1:1) > DE-SG-BP (1:1). In the 20-year rainfall return periods, the top five schemes reduce the peak flow in the region from large to small was as follows: DE-SG-BP (3:1) > DE-BP-SG (3:1) > DE-SG-IP (1:3) > DE-SG-BP (1:1) > DE-SG-IP (1:1).

Based on the above data, the decentralized layout scheme generally had a better reduction effect on peak flow than the centralized layout scheme. With the increase of rainfall return period, the reduction rate of peak flow decreased obviously for schemes with a large proportion of reservoirs, such as the scheme of DE-RE-IP (1:1), CE-RE-IP (1:1), DE-SG-RE (1:3), and DE-BP-RE (1:3). When the reservoir was connected with other infiltration LID facilities, the peak cutting effect remained stable, and it is consistent with the prior work (Zhang 2018).

According to the above simulation results, the LID system was further optimized in combination with the overall site design and the current situation of green space in urban built-up areas. The LID system with infiltration and retention facilities as a base and interspersed with purification and storage facilities were finally constructed (Mo & Yu 2012; Wang et al. 2015; Lin 2018b), then the scale and capacity of each facility were determined (Table 9).

In this study, we combined the regional meteorological characteristics, hydrology, and soil conditions of the green area of Yanshan South Road. It determined the catchment zoning and maximum storage capacity according to the regional sponge city construction target total annual runoff control rate, rainfall during the return period, surrounding catchment areas, and pipe networks to ensure that the rainfall control target was met. Then, the spatial layout type and functional facilities of the LID system were determined after the comparison and selection of multiple scenarios through SWMM. Finally, according to local conditions, suitable LID facilities were selected for the final construction of the LID system in different catchment areas.

For the single LID facility scenario, under the 1-year and 2-year rainfall return periods, the peak flow reduction rates of the DE-SG and DE-RE scenarios were the highest. Under the 3-year, 5-year, 10-year, and 20-year rainfall return periods, the peak flow reduction rates of the DE-BP scenario were the highest. For the combined LID facility schemes, under the 1-year, 2-year, 3-year, 5-year, 10-year, and 20-year rainfall return periods, the highest peak flow reduction rates were found for the DE-SG-BP (1:1), DE-IP-SG (1:1), DE-SG-IP (3:1), DE-SG-BP (3:1), DE-SG-IP (1:3), and DE-SG-BP (3:1) scenarios. In conclusion, the decentralized layout scheme generally has a better peak flow reduction effect than the centralized layout scheme, among which the sunken green space and reservoir had a better peak flow reduction effect under the low return period, and the bioretention pond had a better peak flow reduction effect under the high return period. With the increase in return period, the reduction rate of peak flow decreases.

There was a positive relationship between the surface area of the LID combination scenario and the peak flow abatement rate. Specifically, the larger the surface area, the greater the peak flow abatement rate is. This relationship is mainly due to the long runoff surface flow time and large infiltration interface, which helps to infiltrate the rainwater and increase the amount of infiltration, thus reducing the peak flow, it is consistent with the prior work (Yang et al. 2015; Lin 2018a). In addition, as the rainfall return period increased, the peak flow reduction for each scenario decreased, mainly because the storage function of the LID facilities for each scenario gradually reached its maximum and the effect on peak flow reduction began to approach its limit, consistent with the previous studies (He et al. 2013). Moreover, with the increase in return period, the peak flow reduction rate of schemes with a large proportion of reservoirs decreased obviously, such as CE-RE-IP (3:1) and CE-IP-RE (3:1). However, when the reservoir is connected with other permeable facilities, the peak cutting effect will remain stable.

According to the results of the multi-scenario simulation, the larger the surface area of LID, the more pronounced the effect on runoff regulation. In fact, in the actual sponge city design, the layout of the green space system tends to be stable due to the tight land conditions in the built-up area, so the use of decentralized infiltration type LID facilities combined with storage type LID facilities spatial layout was more efficient to reduce peak flow per unit area, with a relatively small surface area and a more significant peak flow reduction. The land can be saved while the hydrological performance was better, thus giving full play to the advantages of LID systems in rainfall regulation.

There are some limitations in the study, the vertical, soil, and pipe network information obtained in this study was from engineering drawings or survey reports, and the information was perfect and exhaustive. However, due to the investment and cycle time of engineering construction, this study cannot verify the model through actual measured flow, and further research is needed in the future.

This research was funded by the National Natural Science Foundation of China (grant number 31800606) and Social Science Foundation of Beijing in 2021 (grant number 21JCC094).

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.