The magnitude and frequency of extreme rainfall due to climate change is increasing. Increasing rainfall causes serious hydrological problems in cities. Rainfall does not infiltrate the soil, but mostly flows through the sewer pipes into the stream. Most old urban watersheds have combined sewer pipes. When rainfall exceeds the capacity of the combined sewer pipes, sewage mixed with stormwater overflows the sewer pipes and flows directly into the stream. This is called Combined Sewer Overflows (CSOs). CSOs enter the stream with non-point source pollutants accumulated on the surface and pollute the stream. CSOs are one of the major water quality problems in older urban watersheds. This can be solved by replacing the combined sewer pipes with separated sewer pipes, but in reality it requires astronomical costs. As an alternative, the Low Impact Development (LID) technique has recently been introduced. In this study, we analyzed the effects of climate change on CSOs in urban watersheds and applied LID techniques to offset the effects. The LID facility was applied with the most commonly used Bio-Retention cells.

  • Applicability of bio-retention to offset the negative effects of climate change on urban streams are investigated using EPA SWMM.

  • Run-off, TP load, and CSOs in urban watershed show high potential for increases by climate change.

  • The bio-retention cell is useful for offsetting the negative effects of climate change.

Recent studies on climate change emphasize the severity of changes in future rainfall patterns (Christensen et al. 2007; Sagarika et al. 2016; Pathak et al. 2017). Increased rainfall causes serious hydrological problems in urban areas (Tamaddun et al. 2016). When looking at rainfall data from various regions, rainfall in most regions is increasing (Easterling et al. 2000). In some areas, rainfall intensity is likely to increase even though rainfall due to climate change has decreased (Solmon & Coauthors 2007).

When rainfall occurred in natural watersheds, rainfall infiltrated into the soil smoothly and did not generate much stormwater. Because of the good infiltration, most of the rainfall is stored in the soil, and evapotranspiration naturally occurs. However, urbanization has caused many problems (Hewitt & Rashed 1992; Perdikaki & Mason 1999; Maniquiz et al. 2012). Increasing impermeable area due to urbanization reduces rainfall infiltration and results in increased stormwater (Semadeni-Davies et al. 2008). In addition, the peak flow increases and the time of concentration decreases, resulting in much stormwater in a short time (Anderson 1970). In addition, the magnitude and frequency of rainfall due to climate change are very likely to cause greater problems (Kharin et al. 2007; Loo et al. 2015; Burt et al. 2016).

Most old urban watersheds have combined sewer pipes. Combined sewer pipes are a system for transporting rainwater and sewage together, and have the advantages of low cost and convenient construction. However, combined sewer pipes have a disadvantage in that a large amount of pollutants enter the stream when it rains a lot. In Korea, it is the principle to install separated sewer pipes in urban areas that have been recently developed. However, combined sewer pipes have been installed in most old towns. During rainfall in these areas, non-point source pollutants accumulated on the surface during the dry season are swept away with stormwater, and contaminated stormwater enters combined sewer pipes and merges with sewage. Increasing impermeable areas and increasing frequency and intensity of rainfall due to climate change often exceed the capacity of combined sewer pipes, and sewage and stormwater flowing through the combined sewer pipes overflows directly into the stream.

Various climate change adaptation measures can be applied to solve these problems, one of which is the introduction of Low Impact Development (LID) techniques (Pyke et al. 2011). LID facilities that can be applied to various urban watersheds include bio-retention cells, green roofs, infiltration trenches, and permeable pavements. The principles for reducing stormwater and water pollution are similar. The main purpose of LID is to maintain natural hydrological functions while developing cities (Coffman & France 2002; U.S. Department of Housing, and Urban Development 2003). The effects of LID techniques on stormwater reduction and water quality improvement in urban watersheds have already been validated in various literature (Hunt et al. 2006; Bedan & Clausen 2009; Hurley & Forman 2011).

In this study, we investigated the effect of reducing combined sewer overflows (CSOs) by applying Bio-Retention (BR) among LID facilities in Oncheon stream basin located in Busan, Korea. First, climate change scenarios were applied to the EPA SWMM constructed for the Oncheon stream basin to analyze the impact of climate change. Then, when BR was installed in the Oncheon stream basin, it was revealed how much the adverse effects of climate change were offset.

Study area

Korea is geographically located in the mid-latitudes of the northern hemisphere, with annual average precipitation varying by region but about 1,500 mm. Seasonally, between 60 and 70 percent of annual precipitation is concentrated in summer, and the spatial variability of precipitation is very high due to topographical and meteorological factors (Kang 2000). The study basin is the Oncheon stream basin, a city stream in Busan, located southeast of the Korean Peninsula (see Figure 1). The Oncheon stream basin consists of a typical old town area with combined sewer pipes. Therefore, it is appropriate to investigate the change of CSOs in urban areas due to climate change. Oncheon stream basin area is 56.28 km² and river length is 14.85 km (BETCE 2007).

Figure 1

Study area.

Climate change scenarios

Future climate change scenario data are mainly generated using Global climate models (GCMs) and Reginal climate models (RCMs). The low spatial resolution of GCMs makes it difficult to apply to the small area of the Korean peninsula, which is heavily influenced by the ocean. Therefore, it is desirable to use RCMs to investigate the impact of future climate change. In this study, we used the future data of the Busan site, which dynamically downscaled the MPI-ESM-LR data using the Weather Research and Forecasting model (WRF).

MPI-ESM-LR is a model developed by the Max Plank Institute in Germany and consists of representable models the interactions between the atmosphere (ECHAM6) and the ocean (MPIOM), the effects of the ground and vegetation (JSBACH), and the biogeochemical processes in the ocean (HAMOCC5). WRF is a mid-scale forecasting system for weather forecasting and research. WRF features a multidisciplinary core that includes a three-dimensional variant data assimilation system (3DVAR data assimilation), a parallel computer platform, and a software architecture that follows system expandability. CORDEX-East Asia uses WRF3.2 with spectral nudging for long-term climate simulations (von Storch et al. 2000). For more information on WRF, please visit web-site (www.wrf-model.org).

The spatial resolution of the generated future precipitation time series is 12.5-km and the temporal resolution is 3-hour. The present period of the simulated data is from 1981 to 2010, and the future period is from 2021 to 2050. RCP 8.5 climate change scenarios were applied. In this study, the data belonging to the grid corresponding to the Busan site of the Korean Peninsula were extracted from the spatial resolution 12.5-km grid data. This data was regarded as the future precipitation time series of the Busan site and a climate change impact assessment was conducted.

Construction of EPA SWMM

Subcatchment topographic data

In this study, most subcatchment input data were composed by referring to Busan City Master Plan for Sewage and Drainage and Busan City Urban Information System (UIS). The basin was divided into 43 subcatchments using the drainage system and contour maps of the Oncheon stream basin. The area of subcatchment was calculated by GIS. The slope of the subcatchment was calculated using the highest altitude of the subcatchment, the length of the sewer pipe, and its slope.

The impervious area of the terrain data has a big impact on stormwater. The concept of impervious area in SWMM applies the effective impervious area (EIA). EIA is the area directly connected to the body of water in the Total Impervious Area (TIA). TIA was calculated from land use of 1: 25000. The portion of TIA by land use is provided in the SWMM manual (U.S.EPA 2015). However, these figures are based mainly on data from the United States and are unlikely to match Korea. Therefore, in this study, the portion of EIA by land use proposed by NIER (2014) was applied (see Table 1).

Table 1

Percent of impervious area of EIA for urban land use in Korea (NIER 2014)

Urban land useAverage percent impervious area of EIARange in percent impervious area of EIA
Residential 0.44 0.32–0.60 
Commercial 0.62 0.44–0.92 
Industrial 0.79 0.59–0.93 
Transportation 0.95 0.85–1.00 
Education 0.47 0.30–0.77 
Urban land useAverage percent impervious area of EIARange in percent impervious area of EIA
Residential 0.44 0.32–0.60 
Commercial 0.62 0.44–0.92 
Industrial 0.79 0.59–0.93 
Transportation 0.95 0.85–1.00 
Education 0.47 0.30–0.77 

Infiltration process was applied to CN method. The necessary parameter, CN, was estimated using land use and soil map. The subcatchment width was estimated as a function of the area of the subcatchment. In the case of the subcatchment roughness coefficient, the pervious area was applied to 0.13 of the natural area, and the impervious area was applied to 0.011, which is equivalent to smooth asphalt. In the case of initial depression storage depth, the pervious area was 5.0 mm and the impervious area was 1.85 mm. Of these, the CN, subcatchment width, and initial surface storage depth are later modified during parameter tuning.

Pipe network data and CSOs simulation method

Using UIS, the stream consisted of 64 pipes and the sewer pipe networks consisted of 169 pipes. In order to simulate CSOs, the Flow Divider function built into SWMM is included in sewer pipe networks (Jang et al. 2007).

Pollutants input data

The average amount of daily sewage planned for each subcatchment was entered into that subcatchment. The water quality applied in this study is T-P. Sewage water quality was equally applied to all subcatchments at 6.4 mg/L, the concentration entering the sewage treatment plant.

Stormwater quality in the EPA SWMM is simulated through a build-up process that simulates the buildup of contaminants during the dry season and a wash-off process that simulates the removal of contaminants during rainfall. In this study, an exponential function was used as a build-up process and EMC was used as a wash-off process.

Finally, 30,000 tons per day are taken from the nearby Nakdong River and supplied to the Oncheon stream as maintenance water. The maintenance water of the Oncheon stream is input to the starting node of the stream, and it is set so that the maintenance water is not supplied in case of rainfall.

Calibration and verification

After the model is built, the parameters of the model must be tuned so that the model reproduces the observations well. In this study, we used a module that links SWMM and Matlab (Choi et al. 2018). The parameters were estimated based on the Pattern-Search optimization technique built into Matlab. The objective function for the optimization technique is Kling-Gupta-Efficiency (KGE) as follows (Gupta et al. 2009):
formula
(1)
where is the correlation coefficient between the observed and simulated data, is the ratio of the mean of the observed data to the mean of the simulated data, and is the ratio of the standard deviation of the observed data to the standard deviation of the simulated data. Figure 2 shows the schematic diagram of the Matlab-SWMM linkage module.
Figure 2

Schematic diagram for SWMM-Matlab linking module.

Figure 2

Schematic diagram for SWMM-Matlab linking module.

Close modal

River flow and precipitation data were provided by Pusan National University (http://pnuhydro.pusan.ac.kr/). For T-P concentration, data from SeByeong Bridge site observed at Clean Water Lab in Pukyong National University were used. Figure 3 shows the locations of precipitation and stream flow and water quality locations.

Figure 3

Location of SeByeong Bridge and Pusan National University.

Figure 3

Location of SeByeong Bridge and Pusan National University.

Close modal

Application of bio-retention cells

In SWMM, BR serves as a generalized facility for all LID facilities. That is, other LID facilities are considered to be modified facilities based on BR. BR is used all over the world and is one of the best performing LID facilities (Davis et al. 2009; Ahiablame et al. 2012). BR also provides other effects such as aesthetic effects and ecological restoration (Houdeshel et al. 2012; Demuzere et al. 2014; Liu et al. 2014). BRs are installed in various ways (different configurations or sizes) depending on the purpose of adoption (Yang & Chui 2018). On the other hand, the parameters constituting BR in SWMM consist of a total of 18 parameters including surface layer height, soil layer thickness and porosity, storage layer thickness and porosity, and drainage related parameters. The implementation of various BRs in SWMM can be reflected by adjusting the numerical values of these parameters. In this study, we tried to understand the general performance of BR by setting BR of standard configuration. Choi et al. (2018) investigated the parameters of the BR-related SWMM presented in a number of documents (U. S. EPA 1999; KECO 2009; Palhegyi 2010; DER 2002; CVC 2012; DOEE 2013; VWRRC 2013; U. S. EPA 2015), and from these investigations, the parameters of the standard configuration for SWMM were presented (see Table 2 and Figure 4).

Table 2

General parameters of standard bio-retention cell (Choi et al. 2018)

LID typeParameterValueUnit
Bio retention cell Area Contributing area 2,000 m² 
Percent of facility area 
Surface Berm height 300 mm 
Vegetation volume fraction – 
Surface roughness – 
Surface slope 
Soil Thickness 600 mm 
Porosity 0.45 – 
Field capacity 0.30 – 
Wilting point 0.15 – 
Conductivity 50 mm/hr 
Conductivity slope 46.9 – 
Suction head 61.3 mm 
Storage Thickness 300 mm 
Void ratio (voids/solids) 0.625 – 
Seepage rate mm/hr 
Clogging factor – 
Drain Coefficient 0.23094 – 
Exponent 0.5 – 
Offset height 300 mm 
LID typeParameterValueUnit
Bio retention cell Area Contributing area 2,000 m² 
Percent of facility area 
Surface Berm height 300 mm 
Vegetation volume fraction – 
Surface roughness – 
Surface slope 
Soil Thickness 600 mm 
Porosity 0.45 – 
Field capacity 0.30 – 
Wilting point 0.15 – 
Conductivity 50 mm/hr 
Conductivity slope 46.9 – 
Suction head 61.3 mm 
Storage Thickness 300 mm 
Void ratio (voids/solids) 0.625 – 
Seepage rate mm/hr 
Clogging factor – 
Drain Coefficient 0.23094 – 
Exponent 0.5 – 
Offset height 300 mm 
Figure 4

Capacity of bio-retention by each layer.

Figure 4

Capacity of bio-retention by each layer.

Close modal

Among the parameters presented in Table 2, the seepage rate, which means the ability to penetrate into existing soil, is the most conservative soil infiltration rate of Type B of the NRCS hydrological soil group. The reason is that in case of Type C and D which have poor drainage characteristics, it is difficult to expect the effect of the facility, so the principle was not to install the LID facility on such soil.

In this study, the BR with a facility area of 80 m² was defined as a standard BR (ie, 1 unit of BR), and the interception capacity of the standard BR was calculated. From the parameters in Table 2, we can calculate the interception capacity of the standard BR. The interception capacity of BR is calculated by adding storage capacity and infiltration capacity. In the case of storage capacity, the surface layer capacity + soil layer porosity + storage layer porosity can be calculated as follows:

  • (1)

    Capacity of surface layer = 300 mm × 80 m² = 24 m3

  • (2)

    Porosity of soil layer = 600 mm × 0.45 × 80 m² = 21.6 m3

  • (3)

    Porosity of storage layer = 300 mm × [0.625/(1 + 0.625)] × 80 m² = 9.2 m3

  • (4)

    Total storage capacity = 24 m3 + 21.6 m3 + 9.2 m3 = 54.8 m3

Infiltration capacity can be estimated by multiplying soil infiltration capacity by average rainfall duration. For infiltration into existing soils, 4.0 mm/hr applied to BR's standard specifications were applied. Since the average rainfall duration of Busan site was 6-hour and median 4-hour, the conservative 4-hour was applied. Therefore, the infiltration capacity was estimated to be 4 mm/hr × 4 hr × 80 m² = 1.28 m3. Finally, the interception capacity of BR was calculated as 56.8 m3, which is the sum of storage capacity 54.8 m3 and infiltration capacity 1.28 m3. The interception capacity corresponding to each layer is shown in Figure 4.

The cost of the standard BR was estimated using the method proposed in CWRA (2010). CRWA (2010) proposed the cost of interception capacity of LID facilities as follows:
formula
(2)
where, is the cost of LID facility by unit area ($), V is the LID capacity (ft3), is the unit cost estimate ($/ ft3), and F is the adjustment factor (see Tables 3 and 4).
Table 3

Proposed BMP cost estimates (CRWA 2010)

BMPCost ($/ft3)
Bioretention (includes rain garden) 15.46 
Dry pond or detention basin 6.80 
Infiltration trench 12.49 
Porous pavement – porous asphalt pavement 5.32 
BMPCost ($/ft3)
Bioretention (includes rain garden) 15.46 
Dry pond or detention basin 6.80 
Infiltration trench 12.49 
Porous pavement – porous asphalt pavement 5.32 
Table 4

Example of cost adjustment factors (CRWA 2010)

BMP typeCost adjustment factor
New BMP in underdeveloped area 
New BMP in partially developed area 1.5 
New BMP in developed area 
Difficult installation in highly urban settings 
BMP typeCost adjustment factor
New BMP in underdeveloped area 
New BMP in partially developed area 1.5 
New BMP in developed area 
Difficult installation in highly urban settings 

In Equation (2), V used 54.8 m², the estimated interception capacity, used 15.46 $/ft3, which corresponds to bio-retention in Table 3, and F applies 2.0, which is the newly installed BMP in the area developed in Table 4. Therefore, the cost of installing the calculated standard BR was estimated to be 59,836 $.

Scenarios

Projects implemented to mitigate the effects of climate change will involve manpower and costs. Therefore, it is important to understand how much of the impact of climate change can be offset by limited resources. This study benchmarked the case of New York City, USA. New York City, USA, has been implementing a long-term Green Infrastructure Plan since 2002. The project aims to capture 1 inch of stormwater by LID facilities that occurs in an area of 10% of urban land use patches. It is planned to install LID facilities in all subcatchments by 2030 (NYC-DEP 2017). This study examined the extent to which the adverse effects of future climate change were offset when the project was applied to the Oncheon stream basin.

The total area of the Oncheon stream basin is 56,280,000 m² and urban land use is 27,575,847 m². Urban land use area is composed of residential area, commercial area, industrial area, public facility area, transportation area and amusement facility area. The target stormwater capture is 70,043 m3. By dividing this value by 56 m3, the interception capacity of standard bio-retention cells, the number of standard BRs to be installed in the Oncheon stream basin can be estimated. A total of 1,251 units should be installed and the cost is about 74.8 million $. In terms of cost, installing BR in every subcatchment is practically impossible.

In order to apply BR effectively, it is necessary to determine the subcatchment that should be installed first. In this study, it is considered effective to preferentially install a subcatchment with a high proportion of urban land use area, and the subcatchments where the ratio of land use land area exceeds 70% are summarized (see Table 5).

Table 5

Subcatchments with high urban land use area rate

Subcatchment No.Area (m2)Urban land use area (%)The number of bio-retention cells
178,656 90.50 7.3 
11 213,561 77.27 7.5 
22 283,765 95.15 12.2 
27 669,198 84.20 25.6 
33 1,703,575 84.88 65.6 
34 532,625 91.81 22.2 
36 670,477 79.87 24.3 
37 2,244,853 80.93 82.4 
38 556,518 93.58 23.6 
39 634,490 91.15 26.2 
40 269,705 77.42 9.5 
41 922,286 83.21 34.8 
Subcatchment No.Area (m2)Urban land use area (%)The number of bio-retention cells
178,656 90.50 7.3 
11 213,561 77.27 7.5 
22 283,765 95.15 12.2 
27 669,198 84.20 25.6 
33 1,703,575 84.88 65.6 
34 532,625 91.81 22.2 
36 670,477 79.87 24.3 
37 2,244,853 80.93 82.4 
38 556,518 93.58 23.6 
39 634,490 91.15 26.2 
40 269,705 77.42 9.5 
41 922,286 83.21 34.8 

Most of the subcatchments, which have a high proportion of urban land use, are densely populated with residential and commercial areas, and are also located downstream. In this study, seven subcatchments located downstream were used to construct a BR installation scenario. Five scenarios for installing BR in seven subcatchments and one scenario for installing all subcatchments were constructed (see Table 6). Figure 5 shows the subcatchments where BR is installed in each scenario.

Table 6

Bio-retention installation scenario in Oncheon stream basin

ScenariosSubcatchment No.The number of bio-retention cells (unit)Cost (million $)
[A] 39, 40 35.7 2.50 
[B] 36, 38, 41 82.7 5.79 
[C] 36, 38, 39, 40, 41 118.4 8.29 
[D] 34, 37, 39, 40 140.1 9.53 
[E] 34, 36, 37, 38, 39, 40, 41 222.8 15.16 
[F] Full installation 1249 84.97 
ScenariosSubcatchment No.The number of bio-retention cells (unit)Cost (million $)
[A] 39, 40 35.7 2.50 
[B] 36, 38, 41 82.7 5.79 
[C] 36, 38, 39, 40, 41 118.4 8.29 
[D] 34, 37, 39, 40 140.1 9.53 
[E] 34, 36, 37, 38, 39, 40, 41 222.8 15.16 
[F] Full installation 1249 84.97 
Figure 5

Selected subcatchments for study scenarios.

Figure 5

Selected subcatchments for study scenarios.

Close modal

Calibration and verification results

Parameters were calibrated using flow rates and TP loads observed at the SeByeong Bridge site in the Oncheon stream basin. From 2014 to 2015, 22 rainfall events were extracted and used for flow calibration, and from 2016 to 2018, 14 rainfall events were extracted and used for verification. The parameters used for correction are the width of the subcatchment (), CN, the initial depression storage depth () of the permeable zone, and the initial depression storage depth () of the impervious zone. Table 7 shows the parameter initial estimates and the corrected parameter estimates.

Table 7

Stormwater depth calibration parameters

Parameter (coefficient) (coefficient) (mm) (mm)
Original value 1.0 5.0 1.875 
Optimized value 2.0 0.450 4.0 1.5 
Parameter (coefficient) (coefficient) (mm) (mm)
Original value 1.0 5.0 1.875 
Optimized value 2.0 0.450 4.0 1.5 
Width W inputs initial values for each subcatchment assuming the subcatchment as a square, and inputs initial values for each subcatchment using land use and soil map. If the initial W (or ) of one subcatchment is greater than the initial W (or ) of another subcatchment, it is undesirable to change this order of magnitude during parameter estimation. Therefore, a method of correcting by multiplying the initial input value by the coefficient was taken. In other words, W and were estimated using Equations (3) and (4), respectively.
formula
(3)
formula
(4)
where is the calibrated width of the subcatchment, is the initial width, is the calibration coefficient of the width, is the calibrated of the subcatchment, is the initial , and is the calibration coefficient of . In the case of CN, the reason for the configuration as shown in Equation (4) is that the calibration was not smooth due to insufficient stream flow. In other words, the large initial CN value is smally corrected, and the small initial value is largely corrected. The depression storage depths and were applied equally to all subcatchments.

Figure 6 shows the simulated stormwater depth (SIM run-off depth) and observed stormwater depth (OBS run-off-depth) for rainfall events. Figure 6(a) is the calibration result. Figure 6(b) is the verification result. As a result of calibration, the determinant coefficient () was 0.9989 and the model efficiency coefficient () was 0.9746. In verification, = 0.95471 and = 0.93639. Therefore, it can be seen that the simulated stormwater depth reproduces the observation data well.

Figure 6

(a) Flowrate calibration result, (b) flowrate verification result.

Figure 6

(a) Flowrate calibration result, (b) flowrate verification result.

Close modal

The data observed at Pukyong National University was used to reproduce the T-P loads. T-P loads observed for 15 rainfall events from August 2013 to August 2017 were used to calibrate the water quality parameters. The calibrated water quality parameters are the and of the build-up process and the of the wash-off process. Table 8 shows the initial and final corrected values of the parameters.

Table 8

T-P loading calibration parameter

Parameter (coefficient) (/day) (coefficient)
Original value 1.0 0.2 1.0 
Optimized value 4.7578 0.98135 1.25 
Parameter (coefficient) (/day) (coefficient)
Original value 1.0 0.2 1.0 
Optimized value 4.7578 0.98135 1.25 
Since parameter is the mass of pollutant per unit area, it is conceptually similar to land-based unit load in Korea. Therefore, uses 5 times land-based unit load as the initial value. Parameter used for land use provided by the Korea Institute of Environmental Research as the initial value. As in the stream flow calibration, and were also calibrated by applying the initial value to the coefficient.
formula
(5)
formula
(6)
where is the calibrated of each land-use patch, is the initial , is the calibration coefficient of , is the calibrated of each land-use patch, is the initial , and is the calibration coefficient of . The parameter was applied equally to all subcatchments.

Figure 7 shows the simulated T-P loads (SIM T-P Loading) and the observed T-P loads (OBS T-P Loading). As a result of the calibration, = 0.70283 and = 0.65922. Loads are generally not as easy to calibrate parameters as stream flows. This is because, firstly, the observation error occurs easily compared to the flow rate, and secondly, the load is calculated as the product of the flow rate and the concentration, and the error occurs twice.

Figure 7

Water quality calibration results.

Figure 7

Water quality calibration results.

Close modal

Analysis of the effects of climate change

There are biases between climate change scenario data and observational data (Boe et al. 2007). In general, when applying climate change scenarios, priority is given to correcting these biases. The input climate data needed to simulate the SWMM are precipitation and evapotranspiration, and both data were bias-corrected using the quantile mapping method. However, the results of SWMM simulations using present climate simulations and SWMM simulations using observational climate data may be different even if the present climate simulations with bias-correction are applied. By comparing these, the applicability of the climate change scenario data was examined. Observational data consists of 30 years of data from 1981 to 2010, corresponding to the Busan site, among the data provided by the Korea Meteorological Administration (KMA). Present climate simulations were obtained from the same period of MPI-ESM-LR/WRF/RCP 8.5 combinations.

Table 9 shows the water quality items (NPS pollutants, CSOs) and hydrologic components (precipitation, run-off depth, evaporation) simulated from observational data to present climate simulations. Precipitation, run-off depth, and evaporation showed very similar observation data to present climate simulations. In terms of water quality, NPS pollutants differed by about 850 kg per year and CSOs by about 2,000 kg per year. There seems to be a big difference, but the error is about 5% for NPS pollutants and about 7% for CSOs. Water quality can vary greatly due to the complex and varied factors involved. In addition, CSOs are not calculated directly in the subcatchment but rather by adding routing processes. This leads to greater uncertainty. In view of this, it is unlikely that the results from observational data and current climate simulations will vary significantly. Through this verification, the applicability of the relevant climate change scenario data can be judged as sufficiently verified.

Table 9

Comparison of observed with present (ref. Obs is a simulated result using the observed data, Pre is a simulated result using the present data)

Major categorySub categoryObsPre
Hydrological quantity Precipitation (mm/yr) 1519.06 1464.91 
Run-off depth (mm/yr) 1076.30 1027.13 
Evaporation (mm/yr) 119.74 109.90 
Water quality NPS pollutants loading (kg/yr) 14894.94 14052.26 
CSOs loading (kg/yr) 28109.05 26155.39 
Major categorySub categoryObsPre
Hydrological quantity Precipitation (mm/yr) 1519.06 1464.91 
Run-off depth (mm/yr) 1076.30 1027.13 
Evaporation (mm/yr) 119.74 109.90 
Water quality NPS pollutants loading (kg/yr) 14894.94 14052.26 
CSOs loading (kg/yr) 28109.05 26155.39 

Next, current and future data were compared to identify the effects of climate change. The hydrologic components and water quality items compared in Table 10 were configured in Table 9. Change rate (%) was added to identify changes in the future compared to the present. In the case of hydrologic components, all components increased, but the change in run-off depth was relatively higher than that in precipitation. This means that more runoff occurs and relatively less evaporation occurs than rainfall. For water quality items, NPS pollutants and CSOs were similarly increased by about 10%. In summary, these results suggest that future climate change may deteriorate hydrologic systems in urban watersheds and contribute to increased loads of water pollutants.

Table 10

Comparison of present with future (ref. Pre is a simulated result using the present data, Fut is a simulated result using the future data)

Major categorySub categoryPreFutChange (%)
Hydrological quantity Precipitation (mm/yr) 1464.91 1738.95 +18.7 
Run-off depth (mm/yr) 1027.13 1285.50 +25.15 
Water quality NPS pollutants Loading (kg/yr) 14052.26 15571.38 +10.81 
CSOs loading (kg/yr) 26155.39 28949.48 +10.68 
Major categorySub categoryPreFutChange (%)
Hydrological quantity Precipitation (mm/yr) 1464.91 1738.95 +18.7 
Run-off depth (mm/yr) 1027.13 1285.50 +25.15 
Water quality NPS pollutants Loading (kg/yr) 14052.26 15571.38 +10.81 
CSOs loading (kg/yr) 26155.39 28949.48 +10.68 

Off-set effect of BR for climate change

Table 10 uncovers the effects of climate change on the hydrologic components and water quality in the future. This section looks at how BR offsets the effects of climate change. The effect of BR is quantified as follows:
formula
(7)
where, is the result in the present climate, is the result in the future climate, and is the result when the BR is installed in the future climate. Thus, E means the extent to which BR offsets the effects of climate change. Table 11 shows the effects of BR on climate change offsetting stormwater depths. BRs, on average, have reduced the stormwater depth increased by climate change by 13.5%. However, scenario F did not show a good offsetting effect compared to other scenarios. This is because all the subcatchments that do not expect BR efficiency are included.
Table 11

Stormwater depth by scenario (mm/yr)

ScenariosABCDEF
 1350.80 1247.86 1278.34 1248.20 1248.08 1027.14 
 1620.08 1513.76 1545.24 1513.76 1513.76 1303.72 
 1583.32 1477.73 1509.00 1478.21 1478.04 1280.69 
(%) 13.65 13.55 13.58 13.39 13.44 8.33 
ScenariosABCDEF
 1350.80 1247.86 1278.34 1248.20 1248.08 1027.14 
 1620.08 1513.76 1545.24 1513.76 1513.76 1303.72 
 1583.32 1477.73 1509.00 1478.21 1478.04 1280.69 
(%) 13.65 13.55 13.58 13.39 13.44 8.33 

Tables 12 and 13 show the effect of BR's climate change offset on NPS pollutants and CSOs. The effect of BR on NPS pollutants was found to offset, on average, 40% of the adverse effects of climate change. This offset is more than three times the stormwater depth. BR's climate change offset effects on CSOs average 70%. It can be seen that the effect of BR on CSOs is very high. This can be understood by looking at the mechanism by which LID is applied in the EPA SWMM. NPS pollutants are directly reduced by BR in subcatchments. CSOs are generated by both stormwater and NPS pollutants from subcatchments. Thus, stormwater and NPS pollutants reduced by BR drive changes in CSOs. This overlapping effect makes BR's climate change offset effects on CSOs larger than those for stormwater depth and NPS pollutants.

Table 12

NPS pollutants loading by scenario (kg/yr)

ScenarioABCDEF
Pre 436.62 806.78 1243.40 1471.37 2278.15 14052.26 
Fut 488.55 895.85 1384.4 1634.63 2530.48 15571.38 
Fut_BR 469.26 856.7 1325.96 1564.87 2421.57 15061.58 
E (%) 37.15 43.95 41.45 42.73 43.16 33.55 
ScenarioABCDEF
Pre 436.62 806.78 1243.40 1471.37 2278.15 14052.26 
Fut 488.55 895.85 1384.4 1634.63 2530.48 15571.38 
Fut_BR 469.26 856.7 1325.96 1564.87 2421.57 15061.58 
E (%) 37.15 43.95 41.45 42.73 43.16 33.55 
Table 13

CSOs by scenario (kg/yr)

ScenarioABCDEF
Pre 675.89 1475.67 2151.56 2452.22 3927.89 26155.39 
Fut 737.95 1609.33 2347.28 2657.95 4627.28 28949.48 
Fut_BR 696.48 1507.67 2204.14 2510.3 4017.96 27919.47 
E (%) 66.82 76.06 73.13 71.77 87.12 36.86 
ScenarioABCDEF
Pre 675.89 1475.67 2151.56 2452.22 3927.89 26155.39 
Fut 737.95 1609.33 2347.28 2657.95 4627.28 28949.48 
Fut_BR 696.48 1507.67 2204.14 2510.3 4017.96 27919.47 
E (%) 66.82 76.06 73.13 71.77 87.12 36.86 

Oncheon stream catchment is a typical urban area and generates a lot of stormwater due to the large number of impervious areas caused by urbanization. Rainfall does not infiltrate into the soil layer, so stormwater flows into the stream with large amounts of pollutants without natural purification through soil and vegetation, and groundwater is depleted, reducing the base flow of the stream. Climate change is increasing the frequency and intensity of rainfall. This disturbance of hydrological systems continues to occur. Most of the Oncheon stream watershed consists of a combined sewer pipe network, which results in frequent CSOs. Therefore, this study analyzed the effects of climate change on the hydrologic systems and CSOs in urban watersheds, and investigated the effects of bio-retention cells among LID facilities to mitigate the adverse effects of climate change.

The results of applying various climate change scenario data to SWMM revealed that the increase in stormwater depth is greater than the increase in precipitation. Future NPS pollutants and CSOs are projected to grow by about 10% over present ones. In other words, water pollution loads in urban watersheds are likely to increase due to climate change.

To reduce the impact of climate change, bio-retention cells were applied among LID facilities. Bio-retention cells were assumed to be installed in subcatchments, referring to the Green Infrastructure Plan in New York City, USA. In the high impermeability subcatchments, the stormwater depth offsets the adverse effects of climate change by 13.5%. The offset effects on NPS pollutants and CSOs were 40 and 70%, respectively. In particular, bio-retention cells have been shown to be a useful means of countering the increase in CSOs caused by climate change.

However, there are various types of BRs (e.g., rain gardens), and its size, composition, and cost can change depending on the conditions of the installation site. The key purpose of this study is to derive and examine the average effect of the standard BR, with no specific BR method. Hence, in this study, the standard BR was constructed using general BR parameters obtained through literature research for standard BR, without a selection of certain kinds of vegetation, material, and so on. However, when designed with specific BRs suitable for certain sites, the effects may differ from the results of this study, so a sensitivity analysis to reflect various BR methods is one of the good topics of further research. In addition, many of the literature examined ignores some characteristics such as the vegetation volume fraction, the roughness of surface, and so on, thus this study also intactly reflected the values investigated without considering the characteristics (i.e., set to zero; see Table 2). However, note that these parameters may largely affect the effectiveness of the facility in certain cases.

This study is a pilot application of climate change scenarios. In other words, urban hydrological and water quality responses to one type of climate change scenario data were investigated through the SWMM. In the future, various climate change scenario data need to be applied and analyzed. In addition, since more than nine LID facilities can be simulated in the SWMM, it is possible to investigate the effects of various LID facilities. The application of various climate change scenarios and various facilities could contribute to the search for the optimal urban drainage design to adapt to climate change.

This work was supported by Korea Environment Industry & Technology Institute(KEITI) through Smart Water City Research Program, funded by Korea Ministry of Environment(MOE)(2019002950004).

All relevant data are included in the paper or its Supplementary Information.

Ahiablame
L. M.
,
Engel
B. A.
&
Chaubey
I.
2012
Effectiveness of low impact development practices: literature review and suggestions for future research
.
Water, Air, & Soil Pollution
223
,
4253
4273
.
Anderson
D. G.
1970
Effects of Urban Development on Floods in Northern Virginia (P. 22)
.
US Government Printing Office, Washington, DC
.
Bedan
E. S.
&
Clausen
J. C.
2009
Stormwater runoff quality and quantity from traditional and low impact development watersheds 1
.
JAWRA Journal of the American Water Resources Association
45
(
4
),
998
1008
.
BETEC (Busan Environmental Technology Center)
2007
A Study on the Optimal Management of Combined Sewage Overflows (CSOs) Considering the Characteristics of Precipitation in Busan
.
ME (Ministry of Environment) R&D Final Report, Busan, South Korea
.
Boé
J.
,
Terray
L.
,
Habets
F.
&
Martin
E.
2007
Statistical and dynamical downscaling of the Seine basin climate for hydro-meteorological studies
.
International Journal of Climatology
27
(
12
),
1643
1655
.
Burt
T.
,
Boardman
J.
,
Foster
I.
&
Howden
N.
2016
More rain, less soil: long-term changes in rainfall intensity with climate change
.
Earth Surface Processes and Landforms
41
(
4
),
563
566
.
Charles River Watershed Association (CRWA)
2010
Stormwater Management Plan for Spruce Pond Brook Subwatershed
.
Memorandum. EPA Region-I, Franklin, MA
.
Choi
J.
,
Lee
O.
,
Kim
Y.
&
Kim
S.
2018
Improvement of estimation method of load capture ratio for design and evaluation of bio-retention LID facility
.
Journal of Korean Society on Water Environment
34
(
6
),
569
578
.
Christensen
J. H.
,
Hewitson
B.
,
Busuioc
A.
,
Chen
A.
,
Gao
X.
,
Held
R.
,
Jones
R.
,
Kolli
R. K.
,
Kwon
W.
&
Laprise
R.
2007
Regional climate projections. In Climate change, 2007: the physical science basis
. In:
Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change Chapter 11
(S. Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis, K. G. Averyt, M. Tignor, H. L. Miller, eds.).
University Press
,
Cambridge
, pp.
847
940
.
Coffman
L. S.
&
France
R. L.
2002
Low-impact development: an alternative stormwater management technology
. In:
Handbook of Water Sensitive Planning and Design
(R. L. France, ed.). Lewis, Washington, DC, USA,
pp.
97
123
.
CVC (Credit Valley Conservation)
2012
Low Impact Development Stormwater Management Planning and Design Guide
,
Credit Valley Conservation
,
Canada
.
Davis
A. P.
,
Hunt
W. F.
,
Traver
R. G.
&
Clar
M.
2009
Bioretention technology: overview of current practice and future needs
.
Journal of Environmental Engineering
135
(
3
),
109
117
.
Demuzere
M.
,
Orru
K.
,
Heidrich
O.
,
Olazabal
E.
,
Geneletti
D.
,
Orru
H.
,
Bhave
A.
,
Mittal
N.
,
Feliu
E.
&
Faehnle
M.
2014
Mitigating and adapting to climate change: multifunctional and multi-scale assessment of Green urban infrastructure
.
Journal of Environmental Management
146
,
107
115
.
DER (Department of Environmental Resources)
2002
The Bioretention Manual
.
Prince George's County Department of Environmental Resources Programs and Planning Division
,
Maryland
,
USA
.
DOEE (Department of Energy and Environment)
2013
Stormwater Management Guidebook
.
Department of Energy and Environment
,
Washington, DC
,
USA
, pp.
99
128
.
Easterling
D. R.
,
Evans
J. L.
,
Groisman
P. Y.
,
Karl
T. R.
,
Kunkel
K. E.
&
Ambenje
P.
2000
Observed variability and trends in extreme climate events: a brief review
.
Bulletin of the American Meteorological Society
81
(
3
),
417
426
.
Gupta
H. V.
,
Kling
H.
,
Yilmaz
K. K.
&
Martinez
G. F.
2009
Decomposition of the mean squared error and NSE performance criteria: implications for improving hydrological modelling
.
Journal of Hydrology
377
(
1–2
),
80
91
.
Houdeshel
C. D.
,
Pomeroy
C. A.
&
Hultine
K. R.
2012
Bioretention design for xeric climates based on ecological principles
.
Journal of the American Water Resources Association
48
(
6
),
1178
1190
.
Hunt
W. F.
,
Jarrett
A. R.
,
Smith
J. T.
&
Sharkey
L. J.
2006
Evaluating bioretention hydrology and nutrient removal at three field sites in North Carolina
.
Journal of Irrigation and Drainage Engineering
132
(
6
),
600
608
.
Jang
J. H.
,
Kim
S. D.
,
Sung
K. J.
&
Shin
H. S.
2007
Eco-hydrologic assessment of maintenance water supply on Oncheon stream
.
Journal of Environmental Science International
16
(
8
),
973
983
.
Kang, M. S.
2000
The regionality of the variation of summer precipitation in Korea
.
Journal of The Korean Association of Regional Geographers
6
(
3
),
139
152
.
KECO (Korea Environmental Corporation)
2009
The Study Report of Zeroing the Feasibility for Non-Point Source Pollution in Urban Areas
.
Korea Environmental Corporation, Incheon, Republic of Korea [Korean Literature]
.
Kharin
V. V.
,
Zwiers
F. W.
,
Zhang
X.
&
Hegerl
G. C.
2007
Changes in temperature and precipitation extremes in the IPCC ensemble of global coupled model simulations
.
Journal of Climate
20
(
8
),
1419
1444
.
Maniquiz
M. C.
,
Lee
S.
,
Min
K. S.
,
Kim
J. H.
&
Kim
L. H.
2012
Diffuse pollutant unit loads of various transportation landuses
.
Desalination and Water Treatment
38
(
1–3
),
222
229
.
NIER (National Institute of Environmental Research)
2014
A Research on Control Targets and Strategies for Impervious Surface Management. R&D Final Report
.
NIER, Incheon, Korea.
NYC-DEP (New York City Department Environmental Protection)
2017
Green Infrastructure Annual Report
.
NYC-DEP, New York, NY.
Palhegyi
G. E.
2010
Modeling and sizing bioretention using flow duration control
.
Journal of Hydrologic Engineering
15
(
6
),
417
425
.
Pathak
P.
,
Kalra
A.
&
Ahmad
S.
2017
Temperature and precipitation changes in the Midwestern United States: implications for water management
.
International Journal of Water Resources Development
33
(
6
),
1003
1019
.
Perdikaki
K.
&
Mason
C. F.
1999
Impact of road run-off on receiving streams in eastern England
.
Water Research
33
(
7
),
1627
1633
.
Pyke
C.
,
Warren
M. P.
,
Johnson
T.
,
LaGro
J. J.
,
Scharfenberg
J.
,
Groth
P.
,
Freed
R.
,
Schroeer
W.
&
Main
E.
2011
Assessment of low impact development for managing stormwater with changing precipitation due to climate change
.
Landscape and Urban Planning
103
(
2
),
166
173
.
Sagarika
S.
,
Kalra
A.
&
Ahmad
S.
2016
Pacific Ocean SST and z500 climate variability and western US seasonal streamflow
.
International Journal of Climatology
36
(
3
),
1515
1533
.
Semadeni-Davies
A.
,
Hernebring
C.
,
Svensson
G.
&
Gustafsson
L. G.
2008
The impacts of climate change and urbanization on drainage in Helsingborg, Sweden: suburban stormwater
.
Journal of Hydrology
350
(
1–2
),
114
125
.
Solomon
S.
, ,
Coauthors
2007
IPCC, 2007: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of IPCC
.
US Department of Housing, and Urban Development
2003
The Practice of low Impact Development
.
US Department of Housing and Urban Development, Washington, DC
.
U.S.EPA (United States Environmental Protection Agency)
1999
Storm Water Technology Fact Sheet: Bioretention
.
United States Environmental Protection Agency
,
Washington, DC, USA
.
U.S.EPA (United States Environmental Protection Agency)
2015
Storm Water Management Model User's Manual Version 5.1
.
United States Environmental Protection Agency
,
Washington, DC, USA
.
Von Storch
H.
,
Langenberg
H.
&
Feser
F.
2000
A spectral nudging technique for dynamical downscaling purposes
.
Monthly Weather Review
128
(
10
),
3664
3673
.
VWRRC (Virginia Water Resources Research Center)
2013
Virginia DCR Stormwater Design
.
Virginia Water Resources Research Center
,
Blacksburg, VA, USA
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).