1 Introduction

Floods are frequent and destructive natural hazards that can result in the loss of lives, livestock, and economic assets on a global scale (Slater and Villarini 2016; Alfieri et al. 2017; Kreibich et al. 2017; Zhai et al. 2020). Studies have shown that while floods are a natural phenomenon, excess urbanization and climate change have an increasing impact on their recurrence and extent (Alfieri et al. 2017; Du et al. 2019). Furthermore, the Intergovernmental Panel on Climate Change (IPCC) reported that future projections under various climate change scenarios suggest an increase in river flooding and extreme runoff and flows (Alfieri et al. 2017; Kreibich et al. 2017; Zhai et al. 2020; Slater et al. 2021; Yılmaz et al. 2022, 2023b; IPCC 2022).

To further understand the impacts of climate change on floods, Alfieri et al. (2017) conducted a global assessment of the frequency and magnitude of river floods under global temperature increase scenarios of 1.5 °C, 2 °C, and 4 °C. The study found an increasing trend in annual precipitation maxima for the entire land surface. Moreover, it is found that flood events with occurrence intervals larger than the return period of present flood protections are projected to rise on all continents under all scenarios leading to a widespread increase in the flood hazard.

In a similar study, Döll et al. (2018) projected a significant increase in high flows in the event of a global temperature increase of 1.5 °C and 2.0 °C. Furthermore, the IPCC's report (2022) highlights a high likelihood that global warming of 2.0 °C above pre-industrial levels will lead to an expansion of flood-prone areas and substantial increases in runoff. The report provides a summarized global warming scenario, with a medium level of confidence in this projection.

In addition to the negative effects of climate change, excessive and incorrect urbanization can increase the effects of natural hazards, particularly floods. Urbanization typically involves a rapid reduction in pervious areas, leading to urban flooding. The process of urbanization typically involves a consistent increase in impervious surfaces (Ercolani et al. 2018). Urban scale sealing with impervious materials affects the volume and velocity of runoff and results in flooding during intense rainfall (Skougaard Kaspersen et al. 2015). Over half of the world's population currently resides in urban areas, and this proportion is expected to increase by approximately 60% by 2030 (Nations et al. 2018). According to the report by European Environment Agency, various scenarios suggest that by the 2080s, an estimated 250,000 to 400,000 additional people per year in Europe will be impacted by floods, with a majority of them residing in urban areas (EEA 2020). In addition, many cities are expanding in size at an accelerated rate (Pathirana et al. 2014), and land use changes resulting from urbanization can have significant impacts on hydrometeorology due to modifications of surface and boundary layer atmospheric properties (Ren et al. 2008).

In a prior investigation, the correlation between rainfall and runoff in urban settings was examined, highlighting the adverse impact of population density and impervious surfaces on infiltration capacity. This analysis indicated a notable increase in flood magnitudes across sixteen drainage basins in Michigan (Brater and Sherrill 1975). Subsequently, (Darama 1985, 2000) conducted further research on the influence of impervious surfaces and urbanization on infiltration capacity within the same geographic region. These preceding studies collectively affirmed that the expansion of impervious surfaces diminishes infiltration capacity, consequently elevating the occurrence of flooding events. As a more recent study, Du et al. (2019) conducted a study to investigate the impact of varying urbanization levels on flooding in three cities located in the same river basin in China. The study revealed that urbanization significantly increases the risk of flooding by causing an increase in peak flow and flood volume. (Wang et al. 2023) also studied the effect of urbanization on surface runoff by utilizing multiple scenario-based designs, correlation analysis, and a stepwise regression model. Findings indicate a significant increase in the conversion rate of rainfall to runoff when rainfall intensity reached 12.5 mm/12 h, highlighting an elevated waterlogging risk. Areas characterized by impervious surface proportions of 25–50% and 75–100% exhibited heightened sensitivity of surface runoff depth to increases in impervious surfaces, necessitating increased attention. It was determined that maintaining original high-density vegetation or augmenting vegetation density from 0–25 to 25–50% in urban green spaces is the most cost-effective approach. In addition, the arrangement and characteristics of the landscape, such as its fragmentation and shape, influence the quantity and depth of surface runoff. The excess impervious layer and its relationship with runoff have been the focus of interest for researchers recently (Prosdocimi et al. 2015; Du et al. 2019; Adetoro et al. 2022). In a different perspective, multi criteria decision making models such as Analytical Hierarchy Process (Demir et al. 2023; Demir and Dinçer 2023; Dinçer et al. 2023, 2024) have been used to assess the flood risks due to rapid urbanization and changing climate patterns (Nsangou et al. 2022; Waseem et al. 2023). In addition, studies on flood prevention determined the details of non-structural and structural measures (Ginige et al. 2022; Kreibich et al. 2015; WMO 2012). Furthermore, flood prevention methods integrated with ecology and environment have become prominent in recent years (Vojinovic et al. 2021). These methods, which are integrated with ecology and environment, are generally referred to as nature-based solutions (NBSs). The NBSs involve the increase of both pervious surfaces and the storage of runoff in basins, which are then released back into the natural environment. According to IPCC report, green roofs and permeable pavements are identified as NBSs for urban green space creation, reducing urban heat islands, mitigating the effects of global warming, and serving as an urban flood management strategy (IPCC 2022).

Pervious pavement has been the subject of numerous studies with respect to its impact on surface water quality and hydrologic characteristic (Thives et al. 2018; Alam et al. 2019). However, there are limited studies on its efficacy in reducing the extent of urban flood. Similarly, there have been many studies on the effects of rainfall intensity, substrate configuration, humidity, and the retention thickness and covered area of green roofs, as well as their aging, on reducing peak flow (Speak et al. 2013; Zhang et al. 2018, 2019; Liu et al. 2020). However, there are only a few studies on the effect of these parameters on reducing surface runoff. Despite these factors, the vast majority of studies revealed that green roofs are effective in reducing peak flow during storm events, with varying efficiency depending on the characteristics of the roofs and the study area (Sims et al. 2016; Zhang et al. 2018). The extent to which pervious pavement and green roofs can mitigate urban flooding has not been thoroughly studied.

The aim of this study is to investigate how excess urbanization, and the resulting reduction of pervious layers affect the extent of urban floods. To achieve this, a simulation of a flood event that occurred on September 22, 2022 in Erkilet District of Kayseri was conducted. To validate the hypothesis, which emerged from the field investigation suggesting that urbanization is the primary driver of flooding in the Erkilet District, the simulation was repeated on the topography of 2006. The results confirmed that urbanization is indeed one of the main causes of flooding in the area. Furthermore, the effectiveness of green roofs and permeable pavements as NBSs for increasing the permeability of urban areas and mitigating floods was analyzed. The study solidly demonstrates the impact of excessive urbanization, occurring in the 16-year period, on the occurrence of urban flood and to concretely put forth the degree to which the application of green roofs and permeable pavement, considered as NBSs, is effective in reducing the adverse effects of floods. The novelty of our study lies in its exploration of the impact of urbanization on flooding and the effectiveness of green roofs and permeable pavements in reducing surface runoff. While previous literature has extensively investigated these factors separately, our study uniquely assesses their combined effect on a flood event and its extent, a perspective that has not been explored before.

2 The study area

The study area is situated approximately 10 km north of Kayseri city center, a major metropolitan city in central Anatolia and in close proximity to Türkiye's capital, Ankara. Erkilet experiences a dry climate during the summer season, while mild precipitation is observed during the winter months. Erkilet's climate sees temperatures typically ranging between -8 °C and 29 °C throughout the year, occasionally dipping below -16 °C or rising above 34 °C. Rainfall days vary throughout the year, defined as days with at least 1 mm of liquid or liquid-equivalent precipitation in Erkilet. The rainy season extends for 8.2 months, from October 15th to June 20th, with a probability of any given day being rainy at over 13%. May registers the highest average rainfall, with an average of 6.6 days experiencing at least 1 mm of rainfall. In contrast, the dry season commences on June 20th and spans 3.8 months until October 15th. Precipitation types are categorized into solely rain, solely snow, or a mix of both, with their prevalence varying throughout the year in Erkilet. Solely rain predominates for 11 months, commencing from February 12th to January 3rd, with May having the highest number of solely rainy days, averaging 6.5 days. Snow is most common for 1.1 months, starting from January 3rd to February 7th, with January experiencing the highest number of solely snowy days, averaging 1.8 days. A mix of rain and snow is most frequent for 5.0 days, beginning from February 7th to February 12th, with February having the highest number of mixed precipitation days, averaging 1.4 days Erkilet exhibits some seasonal variations in monthly rainfall amounts. The rainy period of the year, with a minimum of 13 mm of moving 31-day rainfall amount, spans 9.4 months from September 24th to July 5th, with May being the rainiest month in Erkilet, boasting an average rainfall of 35 mm. The long-term average annual precipitation recorded from 1991 to 2020 by the Turkish State Meteorological Service (MGM) is 390.2 mm (MGM 2023). These climatic conditions and geographical location are detailed in Fig. 1, which provides information on watersheds, drainage lines, the computational domain, and control points used for examining flow depths in various scenarios.

Fig. 1
figure 1

The study area and computational domain

Full size image

3 Methodology

The aim of this study is to put forth the detrimental effect of urbanization on urban flood regarding historical flood event, and to evaluate the effectiveness of green roofs and pervious pavement in mitigating the negative impacts of the historical flood. To achieve this, relevant data was acquired to simulate the existing flood event that occurred on September 22, 2022, in Erkilet District of Kayseri province, Türkiye. The hourly precipitation over Erkilet on September 22 was obtained from the ERA-5 reanalysis product, a multi-dimensional raster with proven accuracy for global and nationwide extent (Hersbach et al. 2020; Bell et al. 2021). Furthermore, to investigate the impact of urbanization on floods, the study utilized satellite imagery to reconstruct the building and topographic characteristics of the study area as it existed in 2006.

Fieldwork was carried out to analyze soil samples from 10 locations, accurately determining the soil structure and permeability of the area. The geometric features of hydraulic structures such as bridges, culverts, and decks, as well as their coordinates, were also determined. The Metropolitan Municipality of Kayseri provided vector data of buildings, roads, and other features, as well as a digital elevation model (DEM) of the region. The surface roughness of the study area was determined by considering land use classes of CORINE 2018 and CORINE 2006. All the obtained and processed data were used to implement a hydraulic model for simulating the urban flood that occurred on September 22, 2022.

To calibrate the model, variables such as water depth and flood propagation velocity, which were not recorded by institutions, were determined through field studies. After simulating the urban flood event, the implementation of green roofs and pervious pavements as a mitigation measure was analyzed, and changes in flood propagation and hazard classes were determined. A detailed flow chart of the study is presented in Fig. 2.

Fig. 2
figure 2

Flowchart of the study

Full size image

3.1 Geotechnical characteristics and field study

The case study area has undergone rapid and intense urbanization over the past decade, leading to a considerable decrease in the ratio of free fields to the total area. This has resulted in a significant reduction in the area's water absorption capacity. Thus, it was crucial to determine the soil properties of the area, along with its potential water absorption capacity. To achieve this, a comprehensive study was conducted, which included both site surveys and laboratory experiments.

3.1.1 Soil sampling

To delineate suitable sampling sites within the designated area, an observational site survey was undertaken. This survey, supplemented by consultations with local experts, facilitated the identification of potential zones where soil characteristics are likely to exhibit variability. Accordingly, ten locations were selected, distributed across the entire area (see Fig. 3). From each of these locations, two disturbed samples were collected for sieve analysis, water content (Wn), and consistency limit experiments. In addition, at five of these ten locations, soil samples were taken using a steel thin-walled sampler to determine the in-situ relative density. The data from these samples were used to calculate the in-situ natural unit weight (γn) of the soil to determine the water absorption capacities of the samples.

Fig. 3
figure 3

Soil sample locations

Full size image

3.1.2 Laboratory studies

After being collected from the investigated area, the samples were sealed and transported to the Geotechnics Laboratory of Çankaya University's Civil Engineering Department. At each location, each test was repeated twice (once for each sample) to ensure accuracy. The results of these experiments were found to be reasonably similar, with a margin of error of ± 5%. Therefore, the average of the two tests was taken as the representative value for each sample location.

To obtain the particle size distribution curves, sieve analysis tests were conducted in accordance with ASTM D 422–63 (2007) on all samples. Particle size distribution curves of 4 representative samples were presented in Fig. 4 where it can be seen that the curves obtained from both samples at each presented location show similar characteristics. Accordingly, the average particle size distribution curves were obtained for each sample location and given in Fig. 5. All samples were classified as poorly graded sand “SP” according to ASTM D 2487 (2017) with a fines content lower than FC < 5%.

Fig. 4
figure 4

Separate and average particle size distribution curves for selected locations

Full size image
Fig. 5
figure 5

Average particle size distribution curves

Full size image

Consistency limit tests were conducted according to ASTM D 4318 (2010), and the tested samples were found to be non-plastic (NP). The natural water contents of the samples were obtained according to ASTM D 2216 (2019) and were found to vary within the range of 9.06–20.94% for the tested samples. The basic soil properties of the tested samples are summarized in Table 1.

Table 1 Soil classification and basic properties of tested samples
Full size table

The water absorption capacities of the soil samples were determined by utilizing the test results obtained. To accomplish this, the natural unit weight (γn), natural water content (Wn), and specific gravity (Gs) of the soil samples extracted by steel thin-walled samplers were initially ascertained following the guidelines provided by ASTM D 7263 (2021), ASTM D 2216 (2019), and ASTM D 854 (2014), respectively. Subsequently, the in-situ void ratio (e) and degree of saturation (\({{\text{S}}_{\text{r}}}_{\text{n}}\)) values were calculated utilizing the following equations.

$${\upgamma }_{\text{dry}}=\frac{{\upgamma }_{\text{n}}}{(1+{\text{w}}_{\text{n}})}$$
(1)
$${\upgamma }_{\text{dry}}=\frac{\text{Gs}*{\uprho }_{\text{w}}}{(1+\text{e})}$$
(2)
$$\text{e}*{{\text{S}}_{\text{r}}}_{\text{n}}={\text{w}}_{\text{n}}*\text{Gs}$$
(3)

The degree of saturation (\({{\text{S}}_{\text{r}}}_{\text{n}}\)) of the soil samples in their natural state was determined by utilizing equations Eq. (1–3). As all of the tested soil samples were identified as non-plastic poorly graded sands (SP), their maximum water absorption capacities are restricted to the amount of water that can be absorbed from their natural (partially saturated) state to a fully saturated state (Sr = 1). In addition, the depth of the partially saturated soil (DPSS) was measured at each sampling location by excavating the ground until a relatively wet soil layer was visible, and "DPSS" was defined as the distance from the ground surface to the uppermost level of this relatively wet soil layer. Finally, the total amount of water absorption per square meter of surface area (WAC) was determined by utilizing Eqs. (4) and (5), which are presented below.

$$\text{e}*{(\text{S}}_{\text{r}}=1)=\text{w}*\text{Gs}$$
(4)
$${\text{W}}_{\text{AC}}=\frac{\left(\text{w}-{\text{w}}_{\text{n}}\right)*{\upgamma }_{\text{dry}}}{9.81*100}$$
(5)

The related test results and the resulting water absorption capacities (WAC) of the tested samples are summarized in Table 2.

Table 2 Absorption capacities of the tested soil samples
Full size table

4 Summary of field and geotechnical studies

The site surveys revealed that the soil profile of the investigated area could be categorized into four primary regions, with "Erkilet Boulevard" serving as a natural boundary between them (Fig. 6). In "Region 1," the depth of the partially saturated soil was measured to be DPSS ≈ 40–60 cm, and the soil profile beneath this depth was predominantly composed of sandy soil layers. Most of the open areas in "Region 1" are currently being used as agricultural land. However, in "Region 2," the DPSS was measured to be DPSS ≈ 30 cm, and a significant difference from "Region 1" was that rocky layers were observed beneath the surficial soil layer. Additionally, these rocky layers were observed to come to the surface at a short distance, resulting in DPSS ≈ 0 cm, which was defined as the boundary between "Region 2—Region 3" (Fig. 7). As a result, the water absorption capacity of "Region 3" is practically zero, since there is no surficial soil. In "Region 3," there is a small area covered with a 10 cm thick surficial soil. Although the WAC of this region is limited due to its small area and DPSS, it is still evaluated separately as "Region 4" to account for all possible water absorption in the investigated area.

Fig. 6
figure 6

Regions of Erkilet according to WAC

Full size image
Fig. 7
figure 7

The border of Region 2 – Region 3

Full size image

4.1 Hydrologic characteristics

The re-analysis product of ERA-5 was used to obtain precipitation data for the rainfall event that occurred on September 22, 2022. The obtained precipitation data is visually represented in Fig. 8.

Fig. 8
figure 8

Satellite precipitation data of September 22, 2022 on Erkilet (ERA-5 reanalysis product)

Full size image

The obtained precipitation data serves as a gridded boundary condition for the hydraulic model, enabling the simulation of the flood event. However, to determine the surface runoff generated by the rainfall, it is necessary to have information regarding the permeability and curve number of the region. To achieve this, the study area was divided into polygons as shown in Fig. 9 based on considerations such as land use classes, geotechnical studies conducted in the field, and the presence of buildings, roads, and other features for both the simulation years of 2006 and 2022. In the figure buildings are represented by yellow dots. The figure distinctly illustrates the degree of urbanization between 2006 and 2022.

Fig. 9
figure 9

Hydrologic characteristics of the study area

Full size image

4.2 Hydraulic model

In this study, the excess precipitation and corresponding surface flow of the Erkilet District on 22nd September 2022 was simulated by implementing an unsteady 2D HEC-RAS model. The governing equations for solving the 2D unsteady flow throughout the computational domain are given below.

$$\frac{\partial \text{h}}{\partial \text{t}}+\frac{\partial (\text{hu})}{\partial \text{x}}+\frac{\partial (\text{hv})}{\partial \text{y}}=\text{q}$$
(6)
$$\frac{\partial \text{u}}{\partial \text{t}}+\text{u}\frac{\partial \text{u}}{\partial \text{x}}+\text{v}\frac{\partial \text{u}}{\partial \text{y}}-{f}_{c}\text{v}+\text{g}\frac{\partial {Z}_{s}}{\partial \text{x}}-\frac{1}{h}\frac{\partial }{\partial \text{x}}\left({\upsilon }_{t,xx}\text{h}\frac{\partial \text{u}}{\partial \text{x}}\right)-\frac{1}{h}\frac{\partial }{\partial \text{y}}\left({\upsilon }_{t,yy}\text{h}\frac{\partial \text{u}}{\partial \text{y}}\right)=\frac{{\tau }_{b,x}}{\rho R}+\frac{{\tau }_{s,x}}{\rho h}$$
(7)
$$\frac{\partial \text{v}}{\partial \text{t}}+\text{u}\frac{\partial \text{v}}{\partial \text{x}}+\text{v}\frac{\partial \text{v}}{\partial \text{y}}-{f}_{c}\text{u}+\text{g}\frac{\partial {Z}_{s}}{\partial \text{y}}-\frac{1}{h}\frac{\partial }{\partial \text{x}}\left({\upsilon }_{t,xx}\text{h}\frac{\partial \text{v}}{\partial \text{x}}\right)-\frac{1}{h}\frac{\partial }{\partial \text{y}}\left({\upsilon }_{t,yy}\text{h}\frac{\partial \text{v}}{\partial \text{y}}\right)=\frac{{\tau }_{b,y}}{\rho R}+\frac{{\tau }_{s,y}}{\rho h}$$
(8)

where \(\text{u}\) and \(\text{v}\) are the velocities in the \(x\) and \(y\) coordinates, \(\text{g}\) is the gravitational acceleration, \({Z}_{s}\) is the water surface elevation, \({\upsilon }_{t,xx}\) is the horizontal eddy viscosity, \({\tau }_{b,x}\) is the bottom shear stress, \(R\) is the hydraulic radius, \({\tau }_{s,x}\) is the wind stresses, \(\text{h}\) is the water depth, and \(f\) is the Coriolis parameter.

When the flow is primarily dominated by gravity and friction forces, a simplified form of Shallow Water Equations (SWEs), known as Diffusion Wave Approximation or Diffusive Wave Equations (DWEs), can be implemented as the governing equations. During the derivation of DWEs, the local and convective acceleration terms were excluded from the SWEs, considering the domination of gravity and friction forces. As a result, the momentum equations of DWEs can be expressed as follows:

$$\text{g}\frac{\partial {Z}_{s}}{\partial \text{x}}-\frac{1}{h}\frac{\partial }{\partial \text{x}}\left({\upsilon }_{t,xx}\text{h}\frac{\partial \text{u}}{\partial \text{x}}\right)-\frac{1}{h}\frac{\partial }{\partial \text{y}}\left({\upsilon }_{t,yy}\text{h}\frac{\partial \text{u}}{\partial \text{y}}\right)=\frac{{\tau }_{b,x}}{\rho R}$$
(9)
$$\text{g}\frac{\partial {Z}_{s}}{\partial \text{y}}-\frac{1}{h}\frac{\partial }{\partial \text{x}}\left({\upsilon }_{t,xx}\text{h}\frac{\partial \text{v}}{\partial \text{x}}\right)-\frac{1}{h}\frac{\partial }{\partial \text{y}}\left({\upsilon }_{t,yy}\text{h}\frac{\partial \text{v}}{\partial \text{y}}\right)=\frac{{\tau }_{b,y}}{\rho R}$$
(10)

In the analysis employed in the current study, it is notable that the flow dynamics are principally governed by gravitational and frictional forces (Darama et al. 2021; Yılmaz et al. 2023a). Consequently, DWEs were selected as the appropriate numerical tool.

The gridding process and boundary conditions of the hydraulic model are depicted in Fig. 10. The computational domain utilizes a maximum grid size of 10 m, while smaller grids of up to 1 m are employed to cover roads, rivers, and culverts. The regions in which refined grids are employed are also shown in Fig. 10.

Fig. 10
figure 10

Study area grids used in hydraulic model

Full size image

To allow excess water to exit the computational domain, a normal depth boundary condition is selected with a slope of 0.0008. The determination of these normal depth values is based on an examination of the topographic data of the study area.

4.2.1 Digital elevation model (DEM)

The DEM, which has a 5-m resolution and is presented in Fig. 11, was provided by the Greater Municipality of Kayseri in a TIN format and subsequently converted to raster format. In order to obtain the geometric properties of buildings, such as their total area and locations, a satellite view from 2006 and 2022 was assessed. The buildings were then integrated into the DEM prior to implementing the hydraulic model. The buildings 2006 and 2022 can be clearly seen in both DEMs shown in Fig. 11.

Fig. 11
figure 11

Digital elevation model of the study area for 2006 and 2022

Full size image

4.2.2 Surface roughness

Surface roughness is a critical variable in calibrating a hydraulic model. Therefore, the determination of roughness is of utmost importance. Typically, the calibration process involves examining field data from historical events or evaluating data from gauging stations. However, there are no gauging stations in the study area that could be used to determine roughness based on land cover. To address this limitation, the computational area was classified according to the land-use data of CORINE 2018 for simulating the hydraulic model of 2022 while land-use data of CORINE 2006 was used for simulating hydraulic model of 2006, as illustrated in Fig. 12. Notably, the figure vividly illustrates the discernible expansion of the urban area marked in red over time.

Fig. 12
figure 12

Land cover classification of the computational domain according to CORINE 2006 and CORINE 2018

Full size image

Table 3 presents the average values of surface roughness coefficients for the study area based on the land cover classification. These coefficients were determined using a study conducted by Papaioannou et al. (2018), which utilizes the CORINE 2018 land cover data to establish surface roughness coefficients for each specific land cover type.

Table 3 Variation of surface roughness with respect to land cover for the computational domain (Papaioannou et al. 2018)
Full size table

4.3 Scenarios

The primary objective of developing scenarios is to create a depiction of an excessively urbanized area during heavy precipitation, the conditions of the pre-urbanized area, and the impact of nature-based solutions (NBS) on reducing the adverse effects of surface runoff. To achieve this goal, five different scenarios were developed, and presented in Table 4. Scenario 1 (Sc-1) focuses on the urbanized era, specifically examining the flood event occurred on September 22, 2022. Scenario 2 (Sc-2) represents the pre-urbanized era of 2006, offering insights into conditions prior to significant urbanization. Scenarios 3, 4, and 5 (Sc-3, Sc-4, Sc-5) involve the implementation of green roofs, permeable pavement, and a combination of both, respectively, in Sc-1.

Table 4 Scenarios under varying physical condition
Full size table

5 Results and discussion

5.1 Permeability and curve number calculations

The constructed polygons shown in Fig. 9 were incorporated into the hydraulic model to define the permeability and curve number (CN) of each region. The permeability and CN values for each region were calculated, taking into account the field study and land use classes, and are presented in Tables 5 and 6. It is important to mention that the calculations excluded the area occupied by buildings and Erkilet Boulevard. However, the CN values for buildings and Erkilet Boulevard were assumed to be 98 consistent with the suggestions in (Feldman 2000). Green roofs and permeable pavements were obviously excluded from the calculations presented in Tables 5 and 6.

Table 5 CN calculations for 2022
Full size table
Table 6 CN calculation for 2006
Full size table

The CN and permeability values for green roofs and permeable pavement were established based on existing literature (Andersen et al. 1999; Collins et al. 2008; Schwartz 2010; Liu et al. 2020). Previous research indicates that the CN for green roofs is 93 and for permeable pavement is 75, while their respective permeability rates are 50% and 65%.

5.2 Effect of urbanization on urban flooding

To ascertain whether excess urbanization is the main driver of urban flooding over a 16-year period, a comparison is conducted between Sc-1 and Sc-2. Sc-1 represents the current state of excessive urbanization, while Sc-2 depicts the pre-urbanization conditions.

Although the grid size and boundary conditions remain consistent across scenarios, there is a discrepancy in computational time. This disparity can be attributed to the inclusion of CN and permeability calculations, which introduce additional computational complexities into the models.

Figure 13 shows the flood inundation for the topography of 2006 and 2022 on a land use map. The figure illustrates that the overall inundated area does not exhibit a significant difference in the case of excessive urbanization. However, the primary disparity arises within the urbanized region. The excess urbanization extent the inundation area, causing excess flood water out of the urbanized area (Fig. 13a). Furthermore, the excess urbanization also caused flood water to reach the regions that should not be inundated if the urbanization does not take place (Fig. 13ab). To further assess the variations between the two scenarios, Table 7 presents the water depths at the pre-defined control points. This allows for a detailed examination of the discrepancies between the scenarios in terms of water depth measurements.

Fig. 13
figure 13

Effect of urbanization on flood (Sc-1 vs Sc-2)

Full size image
Table 7 Flow depth at the control points
Full size table

The analysis of water depths at the control points reveals notable discrepancies in flow depths between the two different topographies representing the years 2006 and 2022, particularly in the urbanized area. The increase in flow depth ranges from approximately 9.8% up to 195%. These findings from the simulation provide concrete evidence that the impacts of the historical flood's precipitation event would have been less significant if it had occurred during the pre-urbanized era of 2006.

5.3 Effect of green roofs and permeable pavements on flood mitigation

In Sc-3, all the roofs within the computational domain are replaced by green roofs, assuming they have the same permeability and water retention capacity. This scenario focuses exclusively on the implementation of green roofs as a nature-based solution. In Sc-4, the assumption is made that Erkilet Boulevard is covered with permeable pavement. This scenario specifically investigates the effects of permeable pavement as a mitigation measure. Lastly, in Sc-5, the assumptions made in both Sc-3 and Sc-4 are simultaneously applied. This scenario combines the implementation of green roofs and permeable pavement to assess the combined effectiveness of these nature-based solutions.

Figure 14 provides insights into the effectiveness of different measures in reducing floodwater depths. The results indicate that green roofs are capable of mitigating flood risks by reducing water depths by up to 4%. Similarly, permeable pavement contributes to the reduction of total surface runoff depth up to 5.2%.

Fig. 14
figure 14

The decrease in water depths due to NBSs

Full size image

When both green roofs and permeable pavement are implemented together, the combined approach proves to be slightly more effective, reducing flow depths up to 6.2%.

It is important to note that while increasing the pervious layer through nature-based solutions (NBS) in urbanized areas can help reduce the adverse effects of floods, the extent of their effectiveness is limited. Therefore, structural measures remain crucial and should be complemented and supported by the implementation of NBSs.

6 Discussions

Numerous studies have indicated that urbanization plays a prominent role in the escalation of flood events (Yang et al. 2013; Shao et al. 2020). Supporting this notion, a report by the European Environment Agency (EEA) in 2020 highlighted a correlation between increased urbanization and elevated flood flows (EEA 2020). Furthermore, research examining flooding patterns in the Xiang River Basin by Du et al. (2019) demonstrated a direct relationship between urbanization levels and peak flood flows, revealing that the ratio of peak floods incrementally corresponds to urbanization percentages. A study conducted in Ayamama Watershed, Istanbul, Türkiye, by Nigussie and Altunkaynak (2019) similarly affirmed that unrestricted urbanization leads to an expansion of flood inundation areas. In alignment with these prior investigations, our study in the Erkilet area confirmed that excessive urbanization results in inundation areas and flow depths up to 195% higher as depicted in Fig. 13 and in Table 7.

On a distinct note, NBSs for flood prevention involve the strategic use of natural features and processes to manage and mitigate flooding (Slinger et al. 2023). The efficacy of NBSs in reducing flood risks has been extensively discussed in previous research. According to a comprehensive review conducted by Huang et al. (2020), NBSs exhibit significant effectiveness in mitigating urban flooding triggered by high-frequency precipitation events, offering additional economic, ecological, and social advantages. However, NBSs have limitations in addressing pluvial flooding resulting from intense, short-term precipitation events, while conventional gray infrastructure also faces constraints as a mitigation measure against extreme flooding. In congruence with existing literature, our study underscores that the effectiveness of NBSs in the case study area is constrained primarily due to the relatively small area allocated for permeable pavement and green roofs. Notably, the reduction in water depths achieved by NBSs in the best-case scenario is 6.2%, highlighting the need to identify flood risk management strategies by evaluating the performance of alternative combinations of NBSs with gray infrastructure in the context of flood prevention within urban areas.

The significance of this study lies in its capacity to demonstrate the ramifications of excessive urbanization on flood risks and to assess whether the combinations of NBSs can serve as a viable solution to counteract the increased flood risks associated with urbanization.

7 Conclusions

Urbanization has been observed to have a detrimental impact on urban flooding, leading to an increased frequency of peak floods and larger areas of inundation due to the reduction of pervious layers. This trend is expected to worsen in the face of population growth and global warming, which contribute to more frequent and severe hydrometeorological extremes. Therefore, the aim of this study is to prove the effect of urbanization on urban flooding by simulating a flood event that occurred in Erkilet, Kayseri on September 22, 2022, and comparing it with the pre-urbanized conditions of the same region. Furthermore, the effectiveness of NBSs in mitigating urban flooding is also explored.

The results of the study confirm that urbanization has indeed increased the risk of flooding by reducing the availability of pervious surfaces. However, the implementation of green roofs has shown promise in reducing urban flooding and associated flood hazards, with the extent of reduction depending on the configuration of the green roofs, which directly affects the curve number. Regardless of the specific curve number or roof configuration, there was a decrease in flood hazard ranging from 1 to 5% throughout the rainfall event. Similarly, the use of permeable pavements has also demonstrated its effectiveness in reducing urban flooding and associated flood hazards, depending on the configuration of the pavements and their impact on the curve number. The implementation of permeable pavements resulted in a reduction in peak flow discharges and flood hazards. Moreover, when both green roofs and permeable pavements were combined, there was a further reduction in peak discharges. However, it is important to note that nature-based solutions alone are not sufficient to completely prevent floods. They should be considered as supporting measures alongside gray infrastructure as literature suggests.