ABSTRACT

Global warming is associated with rising precipitation intensities, challenging urban drainage systems, and policymakers worldwide. Densely populated, highly sealed cities face high pluvial flooding risks. Nature-based Solutions have been identified as a promising and multifunctional approach to mitigating pluvial flooding impact. This study investigates the flood mitigation potential of various Nature-based Solutions scenarios and a green-grey infrastructure hybrid solution in a neighbourhood in Bochum, Germany. Using an integrated 1D-2D drainage model in PCSWMM, different sub-hourly storm events were simulated for current and future periods.

The green-grey hybrid solution was the most effective in reducing flood area and depth. Among Nature-based Solutions, permeable pavement had the greatest impact, followed by rain gardens and tree pits. All Nature-based Solutions were able to prevent pluvial flooding in design storms with return intervals of 10 years. Runoff reduction rates exhibited relatively stable behavior throughout different precipitation intensities, suggesting that Nature-based Solutions’ potential to reduce runoff exceeds the standard design applications.

The results suggest Nature-based Solutions are effective against pluvial floods in Bochum. Extensive, holistic Nature-based Solutions implementation is crucial for adapting sewer systems and enhancing city-wide resilience. While individual interventions can protect vulnerable infrastructures, city-level resilience must be prioritized to effectively address urban pluvial flood challenges.

1. INTRODUCTION

In the context of climate change, cities, and metropolitan regions worldwide are facing an increasing number of extreme weather events (IPCC 2012). Due to rising precipitation intensities, the Intergovernmental Panel on Climate Change (IPCC) identifies a very high risk of climate change impacts on inland flooding (IPCC 2022). In urban areas, high precipitation intensities pose the hazard of pluvial flooding. Distinctive to fluvial flooding, pluvial flooding does not originate from existing watercourses, but from surface runoff of rainwater that can no longer be fed into the sewer system (Rosenzweig et al. 2018). Next to insufficient drainage capacity, the poor availability of green infrastructure can be blamed for the increasing problems of managing pluvial floods in urban areas (Costa et al. 2021; Shen et al. 2019). Prospective urban water systems will indeed shift towards resource-oriented, integrated sustainable, and decentralized solutions (e.g., Adem Esmail and Suleiman 2020; Oral et al. 2020). In this context, Nature-based Solutions (NbS) have often been presented as successful flood mitigation strategy by reconnecting spheres of the natural water cycle and restoring the soil hydrological functions many times (Costa et al. 2021; Huang et al. 2020; Rosenberger et al. 2021). NbS is described as an umbrella concept for multifunctional solutions that create green infrastructure and provide ecosystem services (Albert et al. 2021; UN Water 2018; World Bank 2021). A key advantage of NbS is their multifunctionality, operating at different scales and meeting resilience needs at different times (Orta-Ortiz and Geneletti 2022; UN Water 2018; World Bank 2021).

In stormwater management, the implication of NbS is gaining increasing attention and has been adopted in resolutions of the G7, G20, and the United Nations General Assembly (Oral et al. 2020). Much research is devoted to investigating the effects of implementing NbS in stormwater models and providing a tool for decision support in stormwater management (Garcia-Cuerva et al. 2018; Johnson and Geisendorf 2019; Palla and Gnecco 2022; Sun and Hall 2016; Wright et al. 2016). Differently than structural flood protection measures, NbS, increasing surface roughness and infiltration, mimic the effect of vegetation or natural soils to reduce flow depth and/or velocities through green roof rain garden, retention basins, and permeable pavements. Their effectiveness has been demonstrated in several studies, worldwide (e.g., Berndtsson 2010; Huang et al. 2020; Trinh and Chui 2013; Zhou et al. 2024; Zölch et al. 2017), and novel social-ecological system frameworks for their design, planning, and implementation show that moving towards equitable and sustainable NbS improve human well-being and ecosystems health (Zhou et al. 2024).

Despite the high potential and increasing interest in NbS, the pluvial flood mitigation potential of NbS has not yet been simulated for Bochum, Germany. Through a case study approach, this study aims to investigate the pluvial flood mitigation potential of different NbS implementation scenarios on a local scale for different storm intensities and durations. It is driven by three research questions:

1. Which infrastructures in the case study area are frequently exposed to pluvial flooding?
2. How can different NbS implementation scenarios effectively reduce the pluvial flood area and depth in the case study? 
3. What is their potential impact in terms of peak and total runoff reduction in the case study?

The Langendreer neighborhood in Bochum, Germany, was chosen as the study area, as recent flood events have been more severe due to higher precipitation levels (Baumeister 2018). Based on feasibility considerations, we focus on three illustrative NbS types: tree pits, rain gardens, and permeable pavement.

2. MATERIALS AND METHODS

2.1 Study area

As shown in Figure 1, Bochum is located in western Germany in North-Rhine-Westphalia, a large agglomeration facing urban sprawl, densification, and increasing population density (Eurostat 2016). Situated between 43 m and 196 m above sea level, Bochum is characterized by an ever moist oceanic warm temperate climate with hot summers (City of Bochum n.d.; Kottek et al. 2006). The annual mean temperature is 10.5°C, and the total annual precipitation is 982 mm (Climate-data.org 2022).

Figure 1 Location of the study area, Bochum Langendreer, and study area delineation used in PCSWMM.

The Langendreer neighborhood in the district of Bochum East was chosen as the study area because recent flood events have been more severe due to higher precipitation levels (Baumeister 2018). Bochum’s hydrology is divided into the Emscher basin in the north, and the Ruhr basin in the south. Both rivers drain westward into the Rhine. Our study area is in the Ruhr basin in the Ölbach sub-catchment (LANUK 2010).

Bochum has a combined sewerage system that covers most of the city and handles both rainwater and sanitary inflow. The study area selected for analysis includes parts of Hauptstraße and Auf dem Jäger, comprising conduits with a total length of 509 m and 15 junctions (Figure 2). Additionally, two sewer tributaries located to the south lie outside of the study area, but their sub-catchments are generalized to account for their inflows. In addition, generalized sub-catchments, and conduits in the west of the study area (Stiftstraße), are added to the simulations, as initial model runs have shown that the model ended too early to account for backwater processes. This involves the addition of nine nodes, 428 m of conduits, and 21.7 ha of sub catchment area to the model.

Figure 2 Sewer system in the study area in Bochum Langendreer.
NOTE: Purple conduits and junctions are used in the hydrological model. Sewer catchments of conduit tributaries are generalized. Green nodes and conduits are not integrated into the model.

2.2 Modeling

Modeling data and parameters

The hydrological modeling was carried out using Personal Computer Storm Water Management Model (PCSWMM version 7.5). The version used here was derived from EPA SWMM5 5.1.015. PCSWMM was chosen for this study because of its global recognition, user-friendly interface, and ease of implementation of NbS (Hou et al. 2020; James et al. 2010). Flood parameters such as maximum flood area and maximum flood depth are products of the integrated 1D-2D model application.

Geodata with coordinates, shape, invert elevation, area, maximum depth, length, inlet and outlet nodes, offset, material, slope, and impervious area of junctions, conduits, and sewer sheds were provided by the City of Bochum. Initial flow, initial depth, surcharge depth, and ponded area are assumed for all junctions. To represent the combined nature of the sewer systems, a dry weather inflow is set for each junction according to the size of the houses in the respective sewer shed.

The depth of depression storage for the impervious and pervious portions of the sub catchment is set to 1.9 mm for impervious surfaces, and 3.8 mm for all pervious surfaces (ASCE 1992; Rossman 2015). The percentage of the impervious area with no depression storage is set to the default value of 25. The subarea routing is set to ‘OUTLET’ with 100% routed. The exchange with the groundwater flow, as well as the seepage rate, are neglected due to the short time intervals.

The maximum infiltration rate of the urban soils is set to 150 mm/hr, based on the findings of Wolff (1996) and Rossman (2015). The minimum infiltration rate and the maximum infiltration volume are taken from the soil characteristics given in Geologischer Dienst NRW (2020). The decay constant is set at 4.14 hr-1 based on Pitt et al. (2000) and Akan (1998). A drying time of eight days and ten antecedent dry days are assumed. A constant evaporation value of 3.2 mm/day is chosen based on Van Osnabrugge et al. (2019).

Street trees play an essential role in the sustainable urban stormwater management and specify the position of tree pits (TP). Accordingly, street trees in the study area are digitized prior to the implementation of all NbS. To account for the hydrological function of trees, the maximum infiltration rate is raised by 2 mm/day for each tree in any of the sub-catchments. An overview of the simulation options is given in Table 1.

Table 1 Overview of the simulation parameters used in PCSWMM.

Parameter	Value
Flow units	CMS
Infiltration method	Horton
Flow routing method	Dynamic wave
Link offsets defined by	Depth
Allow ponding	Yes
Skip steady flow periods	No
Inertial dampening	Partial
Define supercritical flow by	Both
Force main equation	H-W
Variable time step	On
Adjustment factor (%)	75

Selection of precipitation

The selection of precipitation events is based on the 3-Point Approach (3PA), which quantitatively distinguishes three distinct rainfall domains (Figure 3). Originally developed by Fratini et al. (2012) this conceptual approach is well-known in integrated stormwater management and has been adopted and modified, inter alia, by Sørup et al. (2016) and Rosenzweig et al. (2018). It considers three distinct frequency domains, delineated by threshold “points” of management. The domains differ not only in terms of recurrence interval, rainfall intensity and potential damage but primarily reflect different decision thresholds (Sørup et al. 2016).

Figure 3 Classification of the simulated rainfall intensities in the three-point approach.
NOTE: The domains increase with increasing rainfall intensity and return period. Presumed damage (writing in light grey) also increases. This study distinguishes between three rainfall domains, whereby each domain is used concerning (a) current rain intensities, and (b) future rain intensities (Adapted from Rosenzweig et al. 2018; Sørup et al. 2016).

The design precipitation used in this model is taken from the KOSTRA-DWD-2010R dataset from the Climate Data Centre (CDC) of the German Weather Service (DWD) (CDC 2010). The precipitation depths provided have the reference period 1951 to 2010 (both years inclusive) and are spatially resolved to 8.15 km x 8.20 km (CDC 2018). Statistical precipitation heights in a high emission future projection assuming the representative concentration pathway (RCP) 8.5 are extrapolated from KOSTRA using determined percentage increases from Dahm et al. (2019). Given the significant uncertainties inherent in high temporal resolution future projections, only the 60-minute duration level is modeled for the reference period 2070–2100. Dahm et al. (2019) modeled future Intensity-Duration-Frequency (IDF) curves for the Netherlands using the Advanced Delta Change Method (ADCM). Simulations for RCP8.5 were provided by Global Climate Models (GCM) from the EC-Earth ensemble for the period 2071–2100 and supplemented with rainfall by the Royal Netherlands Meteorological Institute (Dahm et al. 2019). Rainfall volumes for the 2071–2100 period under RCP8.5 scenario were not determined for a one-year return interval by Dahm et al. (2019). In this study, a value is interpolated on the assumption of linearity in the distribution of proportional increases in rain intensity from 1981–2010 to 2071–2100, with precipitation volumes between return intervals of one to 100 years. All precipitation depths are divided into 10-time steps, ranging from one minute to six minutes, depending on the simulated rain event. Table 2 provides an overview of the used total rain volumes and maximum rain intensities of all rainfall domains, durations, and reference periods.

The distribution of the precipitation total over the duration of the rainfall event is derived from a Euler type II hyetograph, which is recommended for modeling sewage systems in Germany (Arbeitsblatt DWA-A118 2006) and Poland (Wartalska et al. 2020). The Euler type II model is characterized by 70% of the rain falling in the first third of the rainfall event (Wartalska et al. 2020). In this study 8%, 20%, 40%, 10%, 7%, 5%, 4%, 3%, 2%, and 1% of the total precipitation volume are distributed among the ten timesteps (Table 2).

Table 2 Overview of the total rainfall volumes and maximum rainfall intensities of the simulated design storms with variable duration levels, reference periods, and domains reflecting the return intervals.

Reference period	Duration (min)	Total rain volume (mm)			Max. rain intensity (mm)
		1st domain	2nd domain	3rd domain	1st domain	2nd domain	3rd domain
1951–2010	10	8.3	15	21.7	3.32	6.00	8.68
1951–2010	20	11.5	21.3	31	2.30	4.26	6.20
1951–2010	30	13.2	25.3	37.4	1.76	3.37	4.99
1951–2010	25	14.7	29.7	44.7	1.31	2.64	3.97
1951–2010	60	15.5	33.1	50.6	1.03	2.21	3.37
2071–2100	60	17.7	37.2	56.1	1.18	2.48	3.74

2.3 NbS implementation scenarios

Five scenarios are compared in this study to investigate the hydrological performance of the selected NbS in the study area (see Table 3). While no stormwater management measure is implemented in Scenario 1 (S1), enhanced tree pits, rain gardens, and permeable pavement are the NbS investigated in Scenarios 2 (S2), 3 (S3), and 4 (S4), respectively. The size per tree pit unit is 4 m², and per rain gardens 10 m². In Scenario 5 (S5) all NbS are combined. The total NbS area was calculated based on specific site conditions. Tree pits were implemented at the location of street trees, rain gardens in front yards, and permeable pavements on pedestrian walks and street parking lots. For example, in the study area only the trees on the sidewalk along Hauptstraße owned by the municipality are here classified as street trees; while trees with high proximity to Auf dem Jäger are in open-spaced front yards and are considered unsuitable for implementing extensive tree pits (Figure 4).

Table 3 Description of the nature-based solutions scenarios simulated in the model. NbS types: tree pit (TP), rain gardens (RG), and permeable pavement (PP).

Scenario	NbS		Description
S1	-	-	Business as usual scenario: No additional NbS are implemented.
S2	TP	116	All street tree pits are enhanced with Bioretention cells.
S3	RG	480	Rain gardens are installed in private front yards.
S4	PP	3012	100% of all pedestrian pavements and street parking lots are transformed into permeable pavements.
S5	TR + RG + PP	3608	All NbS measures from Scenarios 2, 3 and 4 are implemented.
S6	TR + RG + PP	3608	All NbS measures from Scenarios 2, 3 and 4 are implemented, and the grey infrastructure is adjusted by extending the pipe diameter.

Figure 4 Location of NbS in the study area.
NOTE: Tree pits (TP) were implemented at the location of street trees, rain gardens (RG) in front yards, and permeable pavements (PP) on pedestrian walks and street parking lots. Inconsistencies on the Auf dem Jager street result from inaccuracies in the base map.

NbS are implemented in PCSWMM using the Low Impact Development (LID) tool. The properties of each NbS element are displayed in Table 4. They are based on previous simulations by Chui et al. (2016), Bai et al. (2019), Rosenberger et al. (2021), and (Liang et al. 2020). The initial saturation of each NbS is set at 10%.

Table 4 LID properties used in the hydrological model.

		TP	RG	PP
	LID type	Bio-retention cell	Rain garden	Permeable pavement
Surface	Berm height [mm]	25	250	5
	Vegetation volume [fraction]	0.1	0.9	0
	Surface roughness (Manning's n)	0.1	0.1	0.12
	Surface slope [%]	0	0	0
Soil	Thickness [mm]	200	100	200
	Porosity [volume fraction]	0.5	0.65	0.5
	Field capacity [volume fraction]	0.2	0.45	0.45
	Wilting point [volume fraction]	0.1	0.1	0.1
	Conductivity [mm/hr]	250	72	72
	Conductivity slope [%]	10	10	10
	Suction head [mm]	87.5	38	45
Pavement	Thickness [mm]	-	-	100
	Void ratio [Voids/Solids]	-	-	0.13
	Impervious surface [fraction]	-	-	0.3
	Permeability [mm/hr]	-	-	180
	Clogging factor	-	-	0
	Regeneration intervals [days]	-	-	0
	Regeneration fraction	-	-	0
Storage	Thickness [mm]	1000	-	300
	Void ratio [Voids/Solids]	0.75	-	0.43
	Seepage rate [mm/hr]	750	30	30
	Clogging factor	0	-	0
Underdrain	Flow coefficient	0.5	-	0.5
	Flow exponent	0.5	-	0.5
	Offset height [mm]	150	-	150

NOTE: Parameters for tree pits (TP) are summarized from Chui et al. (2016). Permeable pavements (PP) properties were obtained from Bai et al. (2019), Chui et al. (2016), and Rosenberger et al. (2021).

2.4 Calibration, uncertainty, and limitations

To ensure representative results, 60-minute design storms with return intervals of three and 100 years were simulated and compared with modeling results generated by the drainage department in the civil engineering office of the City of Bochum with the same climatic conditions. The most considerable uncertainty in the amount of runoff indicated by the runoff quantity continuity error occurs with -0.267% in the 10-minute rain event of the third domain. S4 in the second domain (0.266%) is characterized by the highest flow routing continuity error. Since the modulus of every continuity error is < 1%, the uncertainties of the hydrological and hydraulic portions of the model are considered acceptable. This study focuses solely on the potential reduction of the flood and runoff parameters of simulated NbS under the same conditions and uncertainties. Therefore, the exact amount of runoff and discharge is less relevant here and a possibly diminished representation of the reality justifiable.

3. RESULTS

3.1 Frequently exposed infrastructures in the study area

The first scenario (S1) reveals that three main areas are most frequently exposed to pluvial flooding. These areas are identified as approximately Hauptstraße 294 to 296, the garage yard next to Auf dem Jäger 72, and the depression under the railroad bridge that crosses the road Auf dem Jäger, south of the study area. Although design storms within the domains of year-to-year operations 1a and 1b do not result in pluvial flooding, Auf dem Jäger experiences flooding under all precipitation durations within Domains 2a and 2b, except the current 60-minute design storm. In these domains, flooding on Hauptstraße only occurs in response to 20-minute, 45-minute, and future 60-minute storms. The most significant flooding event with a 10-year return interval, characterized by a maximum flood depth of 0.13 m and flooded area of 603 m², is observed during the simulation of a future 60-minute design storm within Domain 2b (see Figure 5).

Figure 5 Maximum flooded area (a), and depth (b), for the pluvial floods triggered in the simulated design ss in Scenario 1 as a function of return interval and duration.
NOTE: Only Domains 2a, 2b, 3a, and 3b cause pluvial floods. 60a refers to current design storm intensities and 60b to worst-case future design storms of a 60-minute precipitation event.

The future 60-minute storm exhibits the most intense flooding within the study area in the third domain (3b), resulting in a flooded area of 3,695 m² with a maximum flood depth of 0.33 meters. Following closely, the second most intense rain event occurs within 30 minutes, occurring consistently across all second and third domain design storms. The flood extents within Domains 3a and 3b are depicted in Figure 6. During these extreme precipitation events, large portions of both Hauptstraße and Auf dem Jäger experience flooding. Additionally, in some instances, the footpath along the railroad tracks becomes completely submerged (refer to Figure 6).

Figure 6 Maximum pluvial flood areas and maximum depth.

The simulated precipitation lasted 10 (D10), 20 (D20), 30 (D30), 45 (D45), and 60 (D60) minutes, where D60a refers to current design storm intensities and D60b to worst-case future design storms. Rain intensities in the domain of urban resilience and spatial planning (3a and 3b) are used in the simulation.

3.2 NbS potential to mitigate urban pluvial flooding

The pluvial flood mitigation potential of different NbS scenarios compared to the status quo (S1) is summarized in Figure 7 and Figure 8. The extent of the flood area reduction depends on the scenario, the rainfall intensity, and the rainfall volume. Notably, for the 10-minute design storm of Domain 2a, pluvial flooding can be entirely prevented with any scenario. Moreover, the flood area reduction rate across all scenarios and durations is generally higher in Domains 2a and 2b compared to Domains 3a and 3b, with some exceptions. In the domains of technical optimization 2a and 2b, there is a slight trend between rain duration and reduction rate, with lower potential flood area reduction observed in higher-duration storm events. However, this trend does not hold for Domains 3a and 3b, nor for high intensity 10-minute precipitation events. In most cases, the reduction rate increases with the area of the implemented NbS, with S5 being the most effective scenario, and S2 the least effective scenario for our case study. Of note, S4 can prevent all floods in Domains 2a and 2b, while S3 fails to entirely prevent pluvial flooding under any design storms on Hauptstraße. The maximum flood area reductions on Hauptstraße in Domain 3a are 42.2% in S5, 100% in S6, and 27.5% and 55% for S5 and S6, respectively in Domain 3b (Figure 7).

Figure 7 Heat map of the maximum flood area reduction (%) of Scenarios 2 to 6 compared to the business-as-usual Scenario 1.
NOTE: S2 implements tree pits (TP), S3 rain gardens (RG), and S4 permeable pavements (PP). S5 combines S2 to S4, and S6 is combined with S5 with a conduit diameter extension of the underground sewer system.

The simulated precipitation lasted 10 (D10), 20 (D20), 30 (D30), 45 (D45), and 60 (D60) minutes, where D60a refers to current design storm intensities and D60b to worst-case future design storms. Reduction rates depend on the rain intensity, measured at the return period T1 to T100. Greyed-out rain events have not triggered pluvial flooding. While S3 to S6 refer to flood reductions of the entire study area, the results of S2 only record flood area reduction on Hauptstraße, as tree pits were not implemented elsewhere.

A similar pattern is observed when comparing flood depth instead of flood area (see Figure 8). The pluvial flood reduction potential of a single NbS is lower compared to the combined Scenario 5. Across all considered scenarios, greater effectiveness is observed in the domains of technical optimization 2a and 2b (return period of 10 years) than in the Domains 3a and 3b (storm events of a 100-year return interval). Reduction rates generally tend to increase with longer durations, probably due to lower intensities, although exceptions are noted for the 10-minute current design storm and the 60-minute future design storm. However, in contrast to the flood area reduction, the potential reduction in maximum flood depth is relatively modest.

Figure 8 Heat map of the maximum flood depth reduction (%) for S2 to S6 compared to the business-as-usual scenario S1.
NOTE: S2 implements tree pits (TP), S3 rain gardens (RG), and S4 permeable pavements (PP). S5 combines S2 to S4, and S6 combines S5 with a conduit diameter extension of the underground sewer system.

The simulated precipitation lasted 10 (D10), 20 (D20), 30 (D30), 45 (D45), and 60 (D60) minutes, where D60a refers to current design storm intensities and D60b to worst-case future design storms. Reduction rates depend on the rain intensity, measured at the return period T1 to T100. Greyed-out rain events have not triggered pluvial flooding. While S3 to S6 refer to flood reductions of the entire study area, the results of S2 only record flood area reduction on Hauptstraße, as tree pits were not implemented elsewhere.

As observed for flood area reductions, our results reveal that Scenario 5 can reduce the maximum flood depth more effectively on Hauptstraße. This insight is particularly significant given the assumed vulnerability of the houses along Hauptstraße 294 to 296. Once again, the possible reduction in flood area outweighs the reduction in flood depth on Hauptstraße. The only exception is S5 under the 10-minute design storm in Domain 2a, which is by far the most intense rain event simulated. Maximum flood depth reductions achieved by S5, considering only floods on Hauptstraße, amount to 22.2% in Domain 3a, and 12.1% in Domain 3b. Under similar circumstances, S6 may prevent up to 100% of flooding in Domain 3a, and 39.4% in Domain 3b. Notably, the discrepancy between the potential flood area and flood depth reduction on Hauptstraße is smaller in S6 compared to S5.

3.3 Potential runoff reduction

To assess the potential of NbS in pluvial flood mitigation, the reduction rates of runoff from sub-catchments are analyzed for S2 to S5. Surface runoff is closely linked to the precipitation volume and intensity of a storm event. Table 5 presents the total runoff volumes and maximum peak runoff rates for the study area. Both metrics increase with higher precipitation volumes and intensities. The greatest runoff volume is observed in the storm event with the highest precipitation volume: the future 60-minute storm event in Domain 3b. Conversely, the highest peak runoff occurs in the storm event with the highest precipitation intensity: the 10-minute storm event in Domain 3a.

Table 5 Total runoff volumes and peak runoff in the simulated design storms in S1 as a function of return interval and duration.

Return interval (years)	Duration (min)	Total runoff volume (mm)	Max. peak runoff (cm/s)
1 year	10	348	0.6
	20	509	0.7
	30	592	0.6
	45	667	0.5
	60a	706	0.4
	60b	815	0.5
10 years	10	692	1.4
	20	1008	1.57
	30	1207	1.4
	45	1427	1.15
	60a	1597	0.98
	60b	1811	1.11
100 years	10	1094	2.48
	20	1610	2.52
	30	1961	2.21
	45	2371	1.84
	60a	2732	1.6
	60b	3229	1.82

NOTE: 60a refers to the current design storm intensities, and 60b to worst-case future design storms of a 60-minute precipitation event.

While NbS were unable to entirely prevent pluvial flooding during high-intensity storm events in Domains 3a and 3b, Figure 9 demonstrates that the reduction rates in runoff volume for storm events with a 100-year return interval only decrease slightly compared to storms with 10-year return intervals. Interestingly, the reduction in runoff volume through the implementation of rain gardens (RG) in S3 shows an unusual trend, reaching its maximum in the third domain for every storm duration. All other scenarios perform better in the medium-intensity precipitation domains with return intervals of 10 and 100 years.

Figure 9 Heat map of the total sub-catchment runoff volume reduction (%) of S2 to S6 compared to the status quo (S1).
NOTE: S2 implements tree pits (TP), S3 rain gardens (RG), and S4 permeable pavements (PP). S5 combines S2 to S4, and S6 is combined with S5 with a conduit diameter extension of the underground sewer system.

The simulated precipitation lasted 10 (D10), 20 (D20), 30 (D30), 45 (D45), and 60 (D60) minutes, where D60a refers to current design storm intensities, and D60b to worst-case future design storms. Reduction rates depend on the rain intensity, measured at the return period T1 to T100. Greyed-out rain events have not triggered pluvial flooding. While S3 to S6 refer to flood reductions of the entire study area, the results of S2 only record flood area reduction on Hauptstraße, as tree pits were not implemented elsewhere.

It must be noted that the reduction rates displayed in Figure 9 do not refer to the runoff volumes in Table 5, but only to those sub-catchments in which the corresponding NbS are installed. When considering the entire study area, tree pits (S2) can reduce the total runoff volume by 2.3%, rain gardens by 1.1%, permeable pavements by 26.9%, and a combination of all NbS by 28.5%. However, when computing the runoff volume reduction rates for each sub-catchment individually, the maximum runoff volume reductions can reach up to 100% for permeable pavements (S4) (see Figure 10). Tree pits can prevent up to 17.2% of sub-catchment runoff (S2, D60b, T100), rain gardens up to 25.6% (S3, D60b) and a combination of all measures by up to 73.4% across different rain intensities and durations (S5). Furthermore, Figure 10 indicates that not only rain gardens but also tree pits can reduce a higher fraction of the runoff volume with increasing rain intensities. The relatively stable runoff reduction values across different rain durations suggest that rain intensity has a greater impact on the effectiveness of NbS than rain volume.

Figure 10 Heat map of the maximum single sub-catchment runoff volume reduction (%) of S2 to S6 compared to the status quo (S1).
NOTE: S2 implements tree pits (TP), S3 rain gardens (RG), and S4 permeable pavements (PP). S5 is combined with S2 to S4, and S6 is combined with S5 with a conduit diameter extension of the underground sewer system.

The simulated precipitation lasted 10 (D10), 20 (D20), 30 (D30), 45 (D45), and 60 (D60) minutes, where D60a refers to current design storm intensities, and D60b to worst-case future design storms. Reduction rates depend on the rain intensity, measured at the return period T1 to T100. Greyed-out rain events have not triggered pluvial flooding. While S3 to S6 refer to flood reductions of the entire study area, the results of S2 only record flood area reduction on Hauptstraße, as tree pits were not implemented elsewhere.

While the peak runoff in the study areas is rarely affected by the implementation of NbS, they do succeed in reducing the peak runoff in the sub-catchments in which they are positioned. Rain gardens (S3) can only reduce peak runoff in exceptional cases, with peak runoff reductions reaching up to 100%. In most sub-catchments, the peak runoff remains unchanged in S3 compared to S1. This also applies to tree pits (S2) in many storm events, especially for high precipitation volumes. Only S4 and S5 appear to consistently produce stable peak runoff reductions. Permeable pavements (S4) exhibit a high variance in their potential to reduce peak runoff among the sub-catchments, and this variance, along with the average peak runoff reduction, decreases with increasing return interval. On the other hand, the combination of all selected NbS (S5) shows less variability in the reduction potential. Interestingly, the maximum peak runoff reduction potential among the rain intensity domains is observed in the second domain with return periods of 10 years.

4. DISCUSSION

4.1 Potential flood reduction

The maximum flood area reduction results suggest that permeable pavement is the most effective NbS. This finding is consistent with other studies conducted globally, which have demonstrated that permeable pavements offer the greatest peak flow and volume reduction when used in isolation, and are most effective when combined with multiple NbS (e.g. Mabrouk et al. 2023). Since S5 does not lead to significantly higher reduction rates, the need for additional tree pits and rain garden implementation is questionable. In fact, tree pits and rain gardens contribute to the maximum flood depth reduction only marginally for our study area. This can be partially justified by the fact that trees mainly contribute to the processes of interception and evapotranspiration (Orta-Ortiz and Geneletti 2022; Zölch et al. 2017). With no pluvial flooding in the first domains, it is hard to estimate whether the flood reduction by NbS is linearly or non-linearly influenced by rainfall volume and intensity. The notable performance difference between the second and third domains underlines the need to investigate the relationship between rainfall volume and maximum flood area and depth reduction potential of NbS. Identifying when NbS cannot lead to any reduction in the pluvial flood area or depth would be insightful. This study’s results suggest that this threshold for permeable pavement and rain gardens is only reached during infrequent extreme weather events. However, this applies only to the potential flood area reduction of rain gardens. The maximum flood depth remains unaffected by this measure in some design storms. One reason for this could be that the effectiveness of rain gardens is influenced by unknown factors other than the area of the NbS, the size of the sub-catchment, flow length, slope, rainfall intensity, and rainfall volume. According to Rosenberger et al. (2021), additional influences on the performance of rain gardens are the temperature and previous dry periods, which are consistent in all simulations in this study. Tree pits, on the other hand, seem to get to zero flood depth reduction sooner than permeable pavement, between return intervals of 10 and 100 years, at least under these model conditions. Since flood depth reduction rates are lower than flood area reductions, it must be pointed out that the “flood depth is often regarded as the main factor responsible for flood damages” (Costa et al. 2021).

The comparability of the different scenarios is limited, as tree pits only focus on pluvial flooding on Hauptstraße. Had only rain gardens been studied on Hauptstraße, a potential 100% reduction in flood depth and area would have been achieved. This suggests that the effectiveness of NbS depends on sub-catchment characteristics. The ability for rain gardens to reduce maximum flood depth only in certain storm events (i.e., design storm durations of 10 minutes in Domain 2a, and 60 minutes in Domains 2b and 3b) is intriguing. Even if no regularity can be established, these rain events have higher intensities than other design storms. This may be since, in this model, rain gardens were only implemented on previously pervious portions of the sub-catchment. It can be assumed that in most cases, the rainfall on pervious surfaces infiltrates just as well as on the rain gardens. The modeled vegetation cover acts as a resistance and slows down the input rate into the soil only at high rainfall intensities. Without the RG, there will likely be excess infiltration overland flow, which in turn contributes to the flood volume (Beven 2012). Soulsby et al. (2017) confirmed that vegetation cover and the associated interception rates reduce net precipitation and precipitation intensities at the surface for Scottish forests and heather moorland. Accordingly, vegetation heterogeneity, tree species composition, crown porosity, root structure and depth, leaf area index, canopy volume, and diameter-at-breast-high also impact the interception loss of net precipitation and intensity (Baker et al. 2021; Kermavnar and Vilhar 2017; Orta-Ortiz and Geneletti 2022). Among others, Zölch et al. (2017) confirm that trees contribute to flood mitigation mainly through interception and evapotranspiration. It is quite possible that the interceptive capacity of the NbS is underestimated in this study.

Besides vegetation cover, soil structure, and texture, presence of bare land or water bodies, slopes, and land cover determine NbS water retention capacity (Oral et al. 2020). Comparing current and future design storms 60a and 60b, the hybrid scenario’s (S6) potential to reduce maximum flood depths decreases with increasing rainfall intensity, whereas rain gardens and permeable pavements can concentrate the flood depth better during higher intensity rain events. For flood area reduction, this only applies to RG. Other scenarios lose potential flood area reductions with increasing rain intensity. However, conclusions must be drawn with caution, as flood depth and flood area depend on each other, and the topographical conditions of the sewer shed. Since some of the flood depths are very low and were only processed to the second decimal place, the reduction rates are poorly resolved, and nuanced differences were possibly lost. Neither tree pits nor rain gardens could reliably reduce the number of flooded nodes. In S2 and S3, this number sometimes exceeds that of S1. Only the widespread implementation of permeable pavement and the following scenarios can reliably reduce the number of flooded nodes. It must be noted that in the most intensive rain event, there is hardly any reduction in the number of flooded nodes. Although only a small number of junctions flood in the first domains, none of the scenarios can reduce this proportion. However, it must be emphasized that no node flooding in Domains 1a and 1b is sufficient to be perceived as a flood in the 2D mesh, so reductions in the number of flooded nodes in the second and third domains are more crucial. Although the hybrid scenario shows promise in Domains 2a and 2b of 45- and 60-minute precipitations, the shorter rainfall intervals offer only slightly improved reduction rates. This shows a limit to the expansion of grey infrastructure.

Larger pipes can transport high rainfall volumes quickly, but they can hardly cope with increased rainfall intensities. One reason for this could be that backwater processes are accelerated by the sudden occurrence of large amounts of precipitation in a short period. However, as measured by the flooded nodes, NbS also fails due to the high rainfall intensities. The fact that in some simulations, NbS scenarios produce a higher number of flooded NbS could be due to the underdrain integrated with permeable pavement and TP. However, the increased number of flooded nodes is often the case for rain gardens that do not have underdrains installed. Compared to other scenarios, comparably low soil thickness may lead to drainage disadvantages. Because the maximum number of flooded nodes is reached in the higher return intervals despite flood reduction, the flood duration is a better indicator than the flooded nodes.

The limitations of the hybrid scenario (S6) in high-intensity rainfall events are not noticeable here. On the contrary, due to the larger volume of the sewer pipes, flooded nodes are drained more quickly. In this light, the additional expansion of grey infrastructure is very efficient. However, planners must decide whether the duration of the pluvial flood or the depth of the flood should be the primary consideration. If the time of the flood is chosen, the development of grey infrastructure is the most successful methodology. As the results have shown, tree pits and rain gardens fail to reduce the duration of flooded nodes. The flooded node duration reduction rate of permeable pavement is low, yet stable across rainfall intensities and volumes. Since S6 reflects the hybrid version, the adaptation of the grey infrastructure alone cannot be invested in. Only the difference between S5 and S6 indicates the performance of the conduit diameter extension. The transformation of grey infrastructure alone should be assessed to examine all sides of green and grey infrastructure measures.

4.2 Potential runoff reduction

Unlike flooding, runoff depends strongly on rainfall volume, and only to a small extent on precipitation intensity, as most runoff generates from saturation excess. The potential reduction of total sub-catchment runoff volumes decreases non-linearly with increasing rainfall domain. The relatively stable runoff volume reductions of all NbS in Domains 1a, 1b, 2a, and 2b suggests that runoff can be reduced at far higher return intervals. However, the runoff reduction may not affect the pluvial flood generation strongly. Even in Domains 3a and 3b, the reduction potential is not significantly lower than in the earlier domains. The case of RG, whose runoff volume reduction potential increases with increasing storm event recurrence interval, is now receiving attention. This behavior also affects S5, in which all NbS were implemented, at least in Domain 2a, which has slightly higher reduction rates than Domain 1a for 10-, 20-, and 30-minute rain events.

As previously assumed for the maximum pluvial flood depth and area, it is crucial for the rain gardens that the previously pervious areas are converted. If no runoff occurs in these pervious areas, hardly any advantages of the rain gardens can be determined. Only through material changes like rapidly draining sandy soil and reduced infiltration excess overland flow, can runoff volumes be reduced by the rain gardens. With increasing rainfall intensity, infiltration excess overland flow in these areas increases in S1, whereas the input rate is reduced in S3 for the reasons mentioned above. Considering the maximum runoff volume reduction rates of single sub-catchments, the reduction potential of tree pits also increases with the increasing domain. As with rain gardens, tree pits only occupy pervious areas. The potentially higher runoff reduction volume can also be attributed to this reasoning. However, significantly less vegetation was simulated for the tree pits, and this behavior is only revealed when considering the maximum reduction rate of single sub-catchments.

Another reason may be that tree pits treat parts of the impervious area fraction, and in the higher domains this results in much more additional runoff that can be reduced than in earlier domains. It must be considered that the number of rain gardens and tree pits implemented is distributed very heterogeneously among the sub-catchments, which is reflected in the differences between the maximum runoff volume reduction of a single sub-catchment, and the total sub-catchment runoff reduction. Rain gardens’ contributions to runoff reduction are meagre in some sub-catchments. Of course, this is also true for PP, which shows high differences due to the relatively large number of sub-catchments treated. S2 and S5 show little difference between total and maximum runoff volume reduction, as the number of sub-catchments considered is minimal, unlike rain gardens and PP. Another likely reason is that the sub-catchments in which the respective NbS were implemented need to overlap sufficiently. The potential reduction rates of rain gardens relate to sub-catchments other than the reference area of the PP. Whether and to what extent tree pits and permeable pavement can reduce the maximum runoff in the sub-catchment selection of the rain gardens remains unanswered. The results discussed here must therefore be viewed critically in terms of comparability, as the selection of the sub-catchments considered does not match. Therefore, a reliable comparison of the scenarios in runoff volume reduction is impossible. However, since the runoff volume reductions of converted sub-catchments are similar to the results of the entire study area, the evaluation of the NbS in terms of runoff volume reduction is considered representative. Again, the differences are strongly influenced by the implementation area of the NbS. The differences between the various scenarios can most likely be classified according to the size of the NbS measure. Baker and colleagues suggest that larger sized tree pits would most definitely promote infiltration and lead to an increase in the runoff reduction potential of tree pits (Baker et al. 2021). Exceptions to this are rain gardens, which prevent less runoff volume than TP, although they are more extensive. This is because tree pits also treat parts of the runoff from neighboring impervious areas, whereas rain gardens only treat the runoff in pervious areas. Further studies should investigate different NbS under the same conditions, especially in the same sub-catchment, to make the comparison more robust. This study lacks the scientific accuracy of the comparison because the implementation of NbS is made dependent on the sub-catchment characteristics. The peak runoff of the entire study area can likely not be reduced by the NbS because it occurs in sub-catchments outside the measures. The differences between the scenarios when considering individual sub-catchment areas can also be traced back to the quantity and heterogeneity of the sewer sheds viewed as the distribution of the reduction potentials of the peak runoff is consistent with previous results. Peak runoff reductions of rain gardens are perceived as outliers because a high quantity of sub-catchments does not reduce peak runoff at all. S5 does not have the statistical possibility to identify outliers due to a much smaller sample number of sub-catchments.

For many rain durations and domains, the peak runoff reduction rate is a less sensitive variable than the runoff volume. Especially in the higher, more intense domains, the potential peak runoff reduction drops stronger than the comparison of runoff volume reduction results. This could be because the peak runoff is influenced more by the rain intensity than the runoff volume, which depends linearly on the rain volume. Other drivers of the peak runoff generation are surface slope and flow length, parameters which are not changed by the implementation of NbS in this study. Other than runoff volumes, peak runoff rates are meagre, just like the flood depth, and were only recorded up to the second decimal place so that possible minor changes in the reduction rate might have gotten lost. Even if tree pits and rain gardens can only reduce the peak runoff in exceptional cases, the maximum calculated reduction rates are higher than the maximum reduced runoff volumes per sub-catchment. As discussed earlier, vegetation cover of tree pits and rain gardens has a resistance capability that slows down the rain intensity. This proves effective in the maximum peak runoff reduction rates of S2 and S3. This theory is challenged by the fact that S4 almost always causes a higher peak runoff reduction despite the absence of vegetation cover. Again, it is not advisable to compare the scenarios because of the different reference sub-catchments.

4.3 Modeling uncertainties

There are deep uncertainties when modeling of any kind as it only depicts abstractions of reality. Uncertainties might originate from initial and boundary conditions, including model inputs, model structure uncertainties, model parameter estimates, and uncertainties that have been overlooked. These uncertainties may be aleatory, epistemic, or ontological (Beven 2012). A successful summary of rainfall-runoff modeling theory, including uncertainties, is given in Beven (2012). The degree of uncertainty of the model used here cannot be estimated, as neither calibration nor validation nor an error and uncertainty analysis were undertaken. SWMM models are limited in explicitness, accounting for only pervious or impervious land covers for the rainfall-runoff responses and fail to reflect complicated urban catchment with a variety of urban land uses (Qiu et al. 2019). Additionally, the evapotranspiration rates of this study are likely oversimplified (Feng and Burian 2016). The assumptions made strongly influenced the modeling results.

To ensure representative results, a comparison was made between the results obtained in this case study and those generated by the drainage department in the civil engineering office of the City of Bochum. The flow rates of two randomly chosen conduits were compared for 60-minute design storms with return intervals of three and 100 years. The comparison demonstrated that the drainage course of the model employed in this study aligns with the model results available to the planning unit of the City of Bochum. Nevertheless, the quantity of water discharged into the sewage system in this study’s model is slightly higher than in the authority's comparative model, particularly in the 100-year domain. This discrepancy can be attributed to the greater size of the comparative model, as well as simplifications made in this study. It is noteworthy that the comparative model does not take permeable surfaces into account. 

Changing parameters would lead to different runoff results. For example, the sub-catchment fraction with no depression storage is assumed here. A higher value would result in more runoff volume and possibly more pluvial flooding. This might be one of the reasons why predicted infiltration capacities of NbS are not usually met in practice (Kasprzyk et al. 2022). A sensitivity analysis has yet to be performed for the model in this case study to estimate the impact of parameter assumptions made. The LID tool itself is likely a source of uncertainty, depending on the size of implemented NbS and their parameters (Zhang et al. 2022). Once again, reference must be made to the uncertainties inherent in projections of future temperatures and rainfall intensities by the IPCC (2014), and Dahm et al. (2019). According to Rosenzweig et al. (2018) it is inadvisable to base pluvial flood mitigation strategies on a “predict-then-adapt” approach and rely on their successfulness in having robust predictions of the future (Rosenzweig et al. 2018). Effect-centered alternative methods exist that consider the “function of infrastructure systems and policies in response to a range of plausible extreme events, including those that may not have been predicted by a given climate change pathway” (Rosenzweig et al. 2018).

5. CONCLUSION

Due to rising temperatures, the pluvial flood hazard is expected to increase in central Europe. Societal challenges such as urbanization and population growth simultaneously cause a higher vulnerability of urban systems to floods. NbS are increasingly taking over the research of sustainable stormwater management alternatives. Models that have simulated various rainfall events have confirmed their flood regulation services for both separate and combined sewer systems worldwide. This study aimed to investigate the pluvial flood mitigation potential of tree pits, rain gardens, and permeable pavement on a neighborhood scale in Bochum Langendreer by simulating different sub-hourly storm events in three decision domains in an integrated 1D-2D drainage model in PCSWMM, as well as identifying their co-benefits indicated by recent literature.

Our analysis identified three hotspots exposed to pluvial flooding particularly often in this case study, including houses and car parks, a front yard, and a nearby tunnel under the railroad bridge. The effects of tree pits, rain gardens, and permeable pavement on the pluvial flood and runoff generation showed that individual implementation of tree pits, rain gardens, and permeable pavement could completely prevent pluvial flooding in small volumes rain events. In more extreme scenarios, permeable pavement was more effective than rain gardens and tree pits in reducing flood area and depth, with no significant differences between tree pits and rain gardens. Considering individual sub-catchments, the maximum runoff volume reductions were 100%, 25.6%, and 17.2% for permeable pavements, rain gardens, and tree pits, respectively. While tree pits exceeded the runoff reduction potential of rain gardens on average, maximum runoff reductions were higher in the rain gardens scenario. Peak runoff reductions were strongly variable from sub-catchment to sub-catchment.

A combination of all selected NbS in this study could further increase the flood area reduction of permeable pavements. Still, an additional reduction of the flood depth or the number and duration of flooded nodes than permeable pavement could not be found. Runoff volume reductions, however, were reliably higher when combining the NbS. Expanding the combined NbS scenario with widening the conduit diameter was the most successful measure in this case study regarding pluvial flood reduction. This applies to the flood area and depth reduction, and duration of flooded nodes. However, expanding the conduit diameter did not contribute to any runoff reduction.

Even though the effects have shown to be dependent upon the specific site factors of this study and might not be generalizable outside of this case study, a few insights into the urban pluvial flood mitigation potential of NbS can be conveyed. However, the following conclusions must be drawn with caution. All NbS have shown to be effective, especially permeable pavement and a combination of all NbS could reduce the hazard of pluvial floods throughout. The comparison between grey and green infrastructure revealed two different scales of impact. While NbS could not keep up with grey infrastructure adjustments throughout the study area, flooding could be mitigated just as successfully on individual sub-catchments. A particular focus must rely on the substantial advantages NbS have over cost-efficiency and feasibility. It must be considered in decision-making processes that NbS not only address the challenge of pluvial flooding but also create co-benefits and contribute to the well-being of residents. There is no perfect answer to pluvial flood concerns yet. It is probably a lengthy ongoing planning process due to the constant changes in the initial situation, both climatically and socio-economically. Due to the lack of alternatives, NbS must be considered the fastest and most effective means to implement in practice. Especially in the short term, NbS may build resilience for highly exposed, possibly vulnerable, infrastructures.

On a larger scale, this study suggests that an extensive implementation and combination of different NbS in a holistic approach, such as the Sponge City Concept, is necessary to successfully reduce pluvial flooding. Only then can NbS keep up with the drainage potential of an underground sewer system extension. The potential of above-ground green infrastructure measures has yet to be exhausted in many places. Multifunctional areas such as water parks and retention basins that change topographical conditions could further shift the exposure to flooding. A high mitigation value must be given to the vegetation cover of NbS, which can reduce the precipitation intensity due to its interception and evapotranspiration functions. As rain intensity challenges urban water systems more remarkably than the rain volume, this information should be used in urban stormwater management practices. To make full use of all capacities, not only impervious areas must be unsealed, but pervious areas must be upgraded and planted.

It must be expected that pluvial flooding will always be a problem in cities. Even though the exposure to pluvial flooding can be reduced with green or grey infrastructure adjustments, future metropolitan areas likely have higher population densities and, thus, a higher vulnerability. Climate change causes higher rain intensities that possibly exceed commonly used projections. This will challenge not only mitigation adaptation efforts, but the anticipation, evaluation and communications of probabilities and consequences associated with pluvial floods (IPCC 2012). To sustainably reduce the pluvial flood risk in urban areas, the focus must be on creating resilience. Infrastructure-related mitigation and adaptation measures, carried by the municipalities, and object-related measures, held by property owners, must be realized on a grand scale to reduce the vulnerability towards flood events. Public awareness must be created to make this possible. Public relations, communication, and consulting services may be able to do this as it is not the lack of information that leads to low private sector participation, but rather how this information is communicated and perceived (Wang et al. 2019). Weather station networks must be expanded to pick up locally limited heavy rain events that cause pluvial floodings, and warning systems must be adapted to operate on small temporal resolutions. The design storms that create the planning basis should consider larger return intervals, as is common practice in floodplain management (Rosenzweig et al. 2018). Official plans of action against pluvial flooding in Germany must be drawn to support local municipalities and enhance cross-sectional communication and collaboration. This communication is also necessary to account for the ecosystem services of NbS. Instead of respective authorities tackling one problem at a time, the multifunctional benefits of NbS should be explored and discussed by urban and regional planning, urban drainage planning, traffic planning, green space planning, disaster control, and the health department together. A multi-criteria decision process should evaluate all economic, social, and environmental impacts of NbS (European Commission 2021).

Different approaches and concepts to stormwater management need to be united in a holistic approach like the Sponge City Concept. It is challenging to transform urban areas to meet the demands of sustainable stormwater drainage systems. Future building processes must, therefore, completely change their paradigm and reinvent the image of metropolitan regions. Architecture and urban planning must be adapted to accommodate NbS more easily (Adem Esmail and Suleiman 2020; Oral et al. 2020). It must be expected that more than the existing obligations anchored in German building law and land use, or development plans will be required. Further research should be devoted to retrofitting and optimizing NbS in the case study of Bochum. Alternative return intervals shall be investigated to identify the saturation point of NbS, and stormwater management models should be improved to replicate the interception capacity of vegetation. A 3D modeling approach might be more suitable as decision-making support for flood risk management and should therefore also be implemented in Bochum.

ACKNOWLEDGMENTS

The authors would like to express their gratitude to the Civil Engineering Department of the City of Bochum for providing comprehensive geodata pertaining to the sewer system utilized in this study; in particular, to Herr. Ralf Engels from the Drainage and Waterways Department for his unwavering support throughout the implementation of the model. The authors would also like to express their gratitude to HydroPraxis, France, for providing an Educational Grant to utilize PCSWMM EU for this study. The authors would like to extend their sincerest thanks to Prof Riccardo Rigon (University of Trento, Department of Civil, Environmental and Mechanical Engineering) for his invaluable feedback on an earlier draft.

References

1. Adem Esmail, B., and L. Suleiman. 2020. “Analyzing Evidence of Sustainable Urban Water Management Systems: A Review through the Lenses of Sociotechnical Transitions.” Sustainability 12 (11): 4481. https://doi.org/10.3390/su12114481
2. Akan, O. 1998. Urban Stormwater Hydrology: A Guide to Engineering Calculations. Pennsylvania: Technomic Publishing Co.
3. Albert, C., M. Brillinger, P. Guerrero, S. Gottwald, J. Henze, S. Schmidt, E. Ott, and B. Schröter. 2021. “Planning nature-based solutions: Principles, steps, and insights.” Ambio 50 (8): 1446–1461. https://doi.org/10.1007/s13280-020-01365-1
4. Arbeitsblatt DWA-A118. 2006. Hydraulische Bemessung und Nachweis von Entwässerungssystemen. Hennef, Germany: DWA.
5. ASCE. 1992. Design & Construction of Urban Stormwater Management Systems. New York: ASCE and the Water Environment Federation.
6. Bai, Y., Y. Li, R. Zhang, N. Zhao, and X. Zeng. 2019. “Comprehensive performance evaluation system based on environmental and economic benefits for optimal allocation of LID facilities.” Water (Switzerland) 11 (2). https://doi.org/10.3390/w11020341
7. Baker, H.J., M.G. Hutchins, and J.D. Miller. 2021. “How robust is the evidence for beneficial hydrological effects of urban tree planting?” Hydrological Science Journal 66 (8): 1306–1320. https://doi.org/10.1080/02626667.2021.1922692
8. Baumeister, A. 2018. “GIS als Werkzeug für eine dynamische wassersensitive Klimaanpassung in urbanen Landschaften, dargestellt am Beispiel von Oberflächenabfluss und Sickerwasserbildung in Bochum." (Dissertation for the attainment of the doctoral degree). Ruhr-Universität Bochum.
9. Berndtsson, J.C. 2010. “Green roof performance towards management of runoff water quantity and quality: A review.” Ecological Engineering 36 (4): 351–360. https://doi.org/10.1016/j.ecoleng.2009.12.014
10. Beven, K. 2012. Rainfall-Runoff Modeling The Primer. Oxford, UK: John Wiley & Sons, Ltd.
11. CDC (Climate Data Center). 2010. "Raster der Wiederkehrintervalle für Starkregen (Bemessungsniederschläge) in Deutschland (KOSTRA-DWD) Version 2010R." Accessed 13.11.2022. https://www.dwd.de/DE/leistungen/kostra_dwd_rasterwerte/kostra_dwd_rasterwerte.html
12. CDC (Climate Data Center). 2018. "Datensatzbeschreibung- Raster der Wiederkehrintervalle für Starkregen (Bemessungsniederschläge) in Deutschland (KOSTRA-DWD) Version 2010R." Accessed 13.11.2022. https://opendata.dwd.de/climate_environment/CDC/grids_germany/return_periods/precipitation/KOSTRA/KOSTRA_DWD_2010R/gis/BESCHREIBUNG_gridsgermany_return_periods_precipitation_KOSTRA_KOSTRA_DWD_2010R_gis_de.pdf
13. Chui, T.F.M., X. Liu, and W. Zhan. 2016. “Assessing cost-effectiveness of specific LID practice designs in response to large storm events.” Journal of Hydrology 533, 353–364. https://doi.org/10.1016/j.jhydrol.2015.12.011
14. City of Bochum. n.d. “Bochumer Fakten/Infos.” https://www.bochum.de/Bochumer-Fakten-/-Infos
15. Climate-data. org (Ed.). 2022. "Bochum Climate (Germany)." Accessed 15.07.2022. https://en.climate-data.org/europe/germany/north-rhine-westphalia/bochum-2136/
16. Costa, S., R. Peters, R. Martins, L. Postmes, J.J. Keizer, and P. Roebeling. 2021. “Effectiveness of nature-based solutions on pluvial flood hazard mitigation: The case study of the City of Eindhoven (The Netherlands).” Resources 10 (3): 24. https://doi.org/10.3390/resources10030024
17. Dahm, R., A. Bhardwaj, F.S. Weiland, G. Corzo, and L.M. Bouwer. 2019. “A temperature-scaling approach for projecting changes in short duration rainfall extremes from GCM data.” Water (Switzerland) 11 (2): 1–15. https://doi.org/10.3390/w11020313
18. European Commission, Directorate-General for Research and Innovation. 2021. Evaluating The Impact Of Nature-Based Solutions: A Handbook for Practitioners. (A. Dumitru and L. Wendling, eds.). Brussels: Publications Office of the European Union. https://data.europa.eu/doi/10.2777/244577
19. Eurostat. 2016. “Urban Europe – Statistics on cities, towns and suburbs.” Luxembourg: Publications Office of the European Union.
20. Feng, Y., and S. Burian. 2016. “Improving Evapotranspiration Mechanisms in the U.S. Environmental Protection Agency’s Storm Water Management Model.” Journal of Hydrologic Engineering 21 (10). https://doi.org/10.1061/(ASCE)HE.1943-5584.0001419
21. Fratini, C.F., G.D. Geldof, J. Kluck, and P.S. Mikkelsen. 2012. “Three Points Approach (3PA) for urban flood risk management: A tool to support climate change adaptation through transdisciplinarity and multifunctionality.” Urban Water Journal 9 (5): 317–331. https://doi.org/10.1080/1573062X.2012.668913
22. Garcia-Cuerva, L., E.Z. Berglund, and L. Rivers. 2018. “An integrated approach to place Green Infrastructure strategies in marginalized communities and evaluate stormwater mitigation.” Journal of Hydrology 559: 648–660. https://doi.org/10.1016/j.jhydrol.2018.02.066
23. Geologischer Dienst NRW. 2020. “Bodenkarte von Nordrhein-Westfalen 1:50 000 (BK50).” Accessed November 8, 2022. https://www.gd.nrw.de/pr_kd_bodenkarte-50000.php
24. Hou, X., H. Guo, F. Wang, M. Li, X. Xue, X. Liu, and S. Zeng. 2020. “Is the sponge city construction sufficiently adaptable for the future stormwater management under climate change?” Journal of Hydrology 588: 125055. https://doi.org/10.1016/j.jhydrol.2020.125055
25. Huang, Y., Z. Tian, Q. Ke, J. Liu, M. Irannezhad, D. Fan, M. Hou, and L. Sun. 2020. “Nature-based solutions for urban pluvial flood risk management.” Wiley Interdisciplinary Review Water 7 (3): 1–17. https://doi.org/10.1002/WAT2.1421
26. IPCC. 2012. “Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. A Special Report of Working Groups I and II of the Intergovernmental Panel on Climate Change.” Cambridge, UK, and New York, NY, USA: Cambridge University Press.
27. IPCC. 2014. “Summary for Policy Makers.” Climate Change 2014: Impacts, Adapt. Vulnerability. Contrib. Work. Gr. II to Fifth Assess. Rep. Intergovernmental Panel on Climate Change: C.B. Field, V.R. Barros, D.J. Dokken, M.K.J., M.D. Mastrandrea, T.E. Bilir, M. Chatterjee, et al. eds., 1–32 pp. Cambridge, UK, and New York, USA: Cambridge University Press.
28. IPCC. 2022. “Summary for Policymakers.” Climate Change 2022 – Impacts, Adapt. Vulnerability. Contributions of Working Group II to Sixth Assessment Report of the Intergovermental Panel on Climate Change. C.B. Field, V.R. Barros, D.J. Dokken, K.J. Mach, M.D. Mastrandrea, T.E. Bilir, M. Chatterjee, et al. (eds.) 3–34 pp. Cambridge, UK, and New York, NY, USA: Cambridge University Press.
29. James, W., L.E. Rossman, W.R.C. James. 2010. User’s Guide to SWMM5, 13^th Edition. Guelph, Canada: CHI.
30. Johnson, D., and S. Geisendorf. 2019. “Are Neighborhood-level SUDS Worth it? An Assessment of the Economic Value of Sustainable Urban Drainage System Scenarios using Cost-Benefit Analyses.” Ecological Economics 158, 194–205. https://doi.org/10.1016/j.ecolecon.2018.12.024
31. Kasprzyk, M., W. Szpakowski, E. Poznańska, F.C. Boogaard, K. Bobkowska, and M. Gajewska. 2022. “Technical solutions and benefits of introducing rain gardens – Gdańsk case study.” Science of the Total Environment 835, 155487. https://doi.org/10.1016/j.scitotenv.2022.155487
32. Kermavnar, J., and U. Vilhar. 2017. “Canopy precipitation interception in urban forests in relation to stand structure.” Urban Ecosystems 20 (6): 1373–1387. https://doi.org/10.1007/s11252-017-0689-7
33. Kottek, M., J. Grieser, C. Beck, B. Rudolf, and F. Rubel. 2006. “World Map of the Köppen-Geiger climate classification updated.” Meteorologische Zeitschrift 15 (3): 259–263. https://doi.org/10.1127/0941-2948/2006/0130
34. LANUK (Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westphalen). "Nature, Environment and Climate in North Rhine-Westphalia." 2010. https://www.lanuk.nrw.de/
35. Liang, C., X. Zhang, J. Xia, J. Xu, and D. She. 2020. “The Effect of Sponge City Construction for Reducing Directly Connected Impervious Areas on Hydrological Responses at the Urban Catchment Scale.” Water (Switzerland) 12 (4): 1163. https://doi.org/10.3390/W12041163
36. Mabrouk, M., H. Han, C. Fan, K.I. Abdrabo, G. Shen, M. Saber, S.A. Kantoush, et al. 2023. “Assessing The Effectiveness of Nature-Based Solutions-Strengthened Urban Planning Mechanisms in Forming Flood-Resilient Cities.” Journal of Environmental Management 344: 118260. https://doi.org/10.1016/j.jenvman.2023.118260
37. Oral, H.V., P. Carvalho, M. Gajewska, N. Ursino, F. Masi, E.D. van Hullebusch, J.K. Kazak, et al. 2020. “A review of nature-based solutions for urban water management in European circular cities: A critical assessment based on case studies and literature.” Blue-Green Systems 2 (1): 112–136. https://doi.org/10.2166/bgs.2020.932
38. Orta-Ortiz, M.S., and D. Geneletti. 2022. “What variables matter when designing nature-based solutions for stormwater management? A review of impacts on ecosystem services.” Environment Impact Assessment Review 95: 106802. https://doi.org/10.1016/j.eiar.2022.106802
39. Palla, A., and I. Gnecco. 2022. “On the Effectiveness of Domestic Rainwater Harvesting Systems to Support Urban Flood Resilience.” Water Resources Management 36 (15): 5897–5914. https://doi.org/10.1007/s11269-022-03327-6
40. Pitt, R., J. Lantrip, and T.P. O’Connor. 2000. “Infiltration Through Disturbed Urban Soils.” Build Partnerships 1–10. Reston, VA: American Society of Civil Engineers.
41. Qiu, Y., A. Ichiba, I.D.S.R. Paz, F. Chen, P.-A. Versini, D. Schertzer, and I. Tchiguirinskaia. 2019. “Evaluation of Low Impact Development and Nature-Based Solutions for stormwater management: A fully distributed modeling approach.”1–27. https://doi.org/10.5194/hess-2019-347
42. Rosenberger, L., J. Leandro, S. Pauleit, and S. Erlwein. 2021. “Sustainable stormwater management under the impact of climate change and urban densification.” Journal of Hydrology 596: 126137. https://doi.org/10.1016/j.jhydrol.2021.126137
43. Rosenzweig, B.R., L. McPhillips, H. Chang, C. Cheng, C. Welty, M. Matsler, D. Iwaniec, and C.I. Davidson. 2018. “Pluvial flood risk and opportunities for resilience.” WIREs Water 5 (6): 1–18. https://doi.org/10.1002/wat2.1302
44. Rossman, L.A. 2015. Storm Water Management Model (SWMM) User ’s Manual Version 5.1. U.S. Environmental Protection Agency (EPA).
45. Shen, J., S. Du, Q. Huang, J. Yin, M. Zhang, J. Wen, and J. Gao. 2019. “Mapping the city-scale supply and demand of ecosystem flood regulation services—A case study in Shanghai.” Ecological Indicators 106, 105544. https://doi.org/10.1016/j.ecolind.2019.105544
46. Sørup, H.J.D., S.M. Lerer, K. Arnbjerg-Nielsen, P.S. Mikkelsen, and M. Rygaard. 2016. “Efficiency of stormwater control measures for combined sewer retrofitting under varying rain conditions: Quantifying the Three Points Approach (3PA).” Environmental Science and Policy 63, 19–26. https://doi.org/10.1016/j.envsci.2016.05.010
47. Soulsby, C., H. Braun, M. Sprenger, M. Weiler, and D. Tetzlaff. 2017. “Influence of forest and shrub canopies on precipitation partitioning and isotopic signatures.” Hydrological Processes 31 (24): 4282–4296. https://doi.org/10.1002/hyp.11351
48. Sun, N., and M. Hall. 2016. “Coupling human preferences with biophysical processes: Modeling the effect of citizen attitudes on potential urban stormwater runoff.” Urban Ecosystems 19 (4): 1433–1454. https://doi.org/10.1007/s11252-013-0304-5
49. Trinh, D.H., and T.F.M. Chui. 2013. “Assessing the hydrologic restoration of an urbanized area via an integrated distributed hydrological model.” Hydrology and Earth Systems Sciences 17 (12): 4789–4801. https://doi.org/10.5194/hess-17-4789-2013
50. UN Water. 2018. Nature-Based Solutions for Water. Paris: UNESCO.
51. Van Osnabrugge, B., R. Uijlenhoet, and A. Weerts. 2019. “Contribution of potential evaporation forecasts to 10-day streamflow forecast skill for the Rhine River.” Hydrology and Earth Systems Sciences 23 (3): 1453–1467. https://doi.org/10.5194/hess-23-1453-2019
52. Wang, M., D. Zhang, Y. Cheng, and S.K. Tan. 2019. “Assessing performance of porous pavements and bioretention cells for stormwater management in response to probable climatic changes.” Journal of Environmental Management 243, 157–167. https://doi.org/10.1016/j.jenvman.2019.05.012
53. Wartalska, K., B. Kaźmierczak, M. Nowakowska, and A. Kotowski. 2020. “Analysis of Hyetographs for Drainage System Modeling.” Water 12 (1): 149. https://doi.org/10.3390/w12010149
54. Wolff, R. 1996. “Das Infiltrationspotential von Stadtböden am Beispiel Hamburgs.” Urbaner Bodenschutz, Arbeitskr. Stadtboden der Dtsch. Bodenkundlichen Gesellschaft. 69–81. Essen.
55. World Bank. 2021. A Catalogue of Nature-Based Solutions for Urban Resilience. Washington, D.C.: World Bank Group.
56. Wright, T.J., Y. Liu, N.J. Carroll, L.M. Ahiablame, and B.A. Engel. 2016. “Retrofitting LID Practices into Existing Neighborhoods: Is It Worth It?” Environment Management 57 (4): 856–867. https://doi.org/10.1007/s00267-015-0651-5
57. Zhang, Z., W. Hu, W. Wang, J. Zhou, D. Liu, X. Qi, and X. Zhao. 2022. “The hydrological effect and uncertainty assessment by runoff indicators based on SWMM for various LID facilities.” Journal of Hydrology 613, 128418. https://doi.org/10.1016/j.jhydrol.2022.128418
58. Zhou, K., F. Kong, H. Yin, G. Destouni, M.E. Meadows, E. Andersson, L. Chen, et al. 2024. “Urban flood risk management needs nature-based solutions: A coupled social-ecological system perspective.” npj Urban Sustainability 4 (1): 25. https://doi.org/10.1038/s42949-024-00162-z
59. Zölch, T., L. Henze, P. Keilholz, and S. Pauleit. 2017. “Regulating urban surface runoff through nature-based solutions – An assessment at the micro-scale.” Environmental Research 157, 135–144. https://doi.org/10.1016/j.envres.2017.05.023