Abstract
Urbanization and climate change increase the frequency of urban floods worldwide which in turn poses a great challenge in the management of the urban stormwater drain system. This study investigated the impact of climate change and surface imperviousness in the urban stormwater runoff and the impact of washed off pollutants load in the conduit efficiency for two census wards Triplicane (ward 116) and Valasaravakkam (ward 152) in Chennai City. The climate change data for the study area wards were taken from the regional climate model (RCM) of Representative Concentration Pathway (RCP) 4.5. Storm Water Management Model (SWMM) with quality and quantity scenarios were developed and the results of the present year scenario were validated using a questionnaire survey. Intensity Duration Frequency (IDF) curves were generated to forecast the observed rainfall and climate model data for various return periods, i.e., 2, 5, 10, 50, and 100 years for future climatic scenarios (2030s and 2050s). The model results show that infiltration is only 3–7% of the total rainfall in the study areas and the maximum blockage and reduction in conduit capacity were estimated to be 65 and 40%, respectively. This paper presents the observations and suggestions for improving the conduit efficiency from an Integrated Water Management Approach (IWMA).
HIGHLIGHTS
Imperviousness of the study area was calculated based on LULC classification using Google Earth.
Bias correction was carried out for the climate data using observed rain gauge data.
Evaluation of the impact of pollutants load in the conduit carrying capacity using the SWMM.
Questionnaire survey outcomes were used to understand the past extreme events and also used in conjunction with the model outputs to derive suggestions.
Graphical Abstract
INTRODUCTION
Unplanned infrastructure development and paved surface in urban area increase imperviousness and it is one of the fundamental reasons for urban floods (Ahamed & Agarwal 2019). Land-use changes such as loss in natural water bodies, encroachment of rivers/streams, and other drainage channels, along with uncontrolled development of built-up areas have been identified as contributory factors for the increasing flood events in Chennai City (Gupta & Nair 2010). The built-up area in Chennai City has increased up to 173.83% in 2017 which has significantly increased the imperviousness (Mathan & Krishnaveni 2020). For a 44% increase in the imperviousness area, the total runoff volume increases by 72% in an urban catchment (Lee & Heaney 2003). It is identified that Land Use Land Cover (LULC) changes and an increase in imperviousness of an urban region increase the urban stormwater runoff by manyfold.
The urban regions in developing countries are more prone to the adverse effects of climate change due to their vulnerability, less adaptivity, and increase in population density (Pickett et al. 2001). Repercussions of climate change on the urban areas led to a reduction in the water quantity, quality, and increased flood events (Miller & Hutchins 2017). The number of extreme precipitation events is likely to be increased and there will be a decrease in the annual mean precipitation in the Asian region landmass, as per the Intergovernmental Panel on Climate Change (IPCC) Assessment Report 5 (IPCC 2013). This shows the importance of using climate models in urban studies. The daily observed rainfall data also can be used to study the impact of climate change in a major urban area (Shahid et al. 2016) and the trend of rainfall in Chennai City can be studied using Coordinated Regional Climate Downscaling Experiment (CORDEX) data for Representative Concentration Pathways (RCPs) 4.5 and 8.5 (Athira & Agilan 2020).
Increase in the imperviousness and climate change made the stormwater network-related study a complex process. The models used for these studies are the Hydrologic Modelling System (HEC-HMS), Soil and Water Assessment Tool (SWAT), Storm Water Management Model (SWMM), MIKE System Hydrologique European (MIKE SHE), and MIKE URBAN. Among these SWMM is one of the most widely used models for stormwater network studies (Janga et al. 2007; Ahamed & Agarwal 2019; Brendel et al. 2021). The SWMM can be used to study the impact of climate change on the quality of stormwater runoff and the flood-carrying capacity of the stormwater drain network in urban areas for both the present and future scenarios (Leboutillier et al. 2000; Andimuthu et al. 2019; Sonavane et al. 2020). The SWMM can also be used to estimate the pollutants load using EMC values for different land use and land cover data (Ryu et al. 2016; Tu & Smith 2018; Tuomela et al. 2019; Choi et al. 2021). The four water quality indicators, namely biological oxygen demand (BOD), total suspended solids (TSS), total phosphate (TP), and total nitrate (TN) were predominantly used in the evaluation of stormwater quality in urban regions in various studies (LeBoutillier et al. 2000; Li et al. 2015; Tu & Smith 2018; Tuomela et al. 2019; Choi et al. 2021). So, we have taken only these four quality indicators in this study.
Several studies (Agarwal & Kumar 2019; Hussain et al. 2022) were carried out to evaluate the impact of imperviousness and climate change on the urban stormwater runoff and storm drain network using the extreme precipitation event. It is noted that some of these studies were carried out without adjusting the climate data for the local climatic conditions. These approaches will not be adequate to understand the future impact on the stormwater drain network completely. In order to estimate the accurate impact of climate change on the stormwater runoff and pollutant load, bias correction needs to be carried out for the regional climate model (RCM) data to adjust for the local climatic conditions. To the best of our understanding, the combined impact of climate change, surface area imperviousness, and pollutants load need to be studied for the effective evaluation of storm drain network performance. This paper presents the findings based on this combined impact on the storm drain network performance with an integrated water management approach.
METHODS
Study area description
Chennai is an ancient city with rich history and cultural heritage. The town was named Madras by the British East India Company on 22 August 1639. Today, it is one of the most populous urban agglomerations and is located on the East Coast of India. Chennai mainly depends on the North East (NE) monsoon during which around 65% of the annual rainfall is received every year and it is prone to flood during this season. Chennai City is now divided into 15 zones, and each zone has more than 10 wards accounting for 200 wards in the city. Among these wards, ward 116 (Triplicane) and ward 152 (Valasaravakkam) from zones 9 and 11, respectively, were chosen as the study area and these wards are within the urban limits of Chennai City.
Data used for the study
The secondary data collected for this study from various sources are presented in Table 1. Precipitation data for the study area were collected from the Institute for Water Studies (IWS) and the gridded data were downloaded from the online database of the Indian Meteorological Department (IMD) Pune. The SWMM uses the topographical, LULC, soil type, and storm network data for structuring the model. The LULC data required for this study were taken from Google Earth for the study area and it was georeferenced and digitized using ARCGIS10.5. The elevation and slope values for the study area were taken from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Digital Elevation Model (DEM) data. The soil property data for the study area are extracted from the Harmonized World Soil Database (HWSD) of the Food and Agriculture Organization (FAO). The stormwater network map was downloaded from the Greater Chennai Corporation online database, and it was digitized using QGIS 3.16.
Data used for the study and their source
Data . | Time period/version . | Spatial resolution . | Sources . |
---|---|---|---|
Precipitation | 1990–2020 | 0.25° × 0.25° | IWS & IMD |
Climate data | 2020–2100 | 0.44° × 0.44° | ESGF-CORDEX |
DEM data | Version 3 | 30 m × 30 m | ASTER DEM |
LULC | Real time | 15 cm | Google Earth |
Soil property | Version 1.2 | 1 km | FAO |
Storm network | – | – | Chennai Corporation |
Data . | Time period/version . | Spatial resolution . | Sources . |
---|---|---|---|
Precipitation | 1990–2020 | 0.25° × 0.25° | IWS & IMD |
Climate data | 2020–2100 | 0.44° × 0.44° | ESGF-CORDEX |
DEM data | Version 3 | 30 m × 30 m | ASTER DEM |
LULC | Real time | 15 cm | Google Earth |
Soil property | Version 1.2 | 1 km | FAO |
Storm network | – | – | Chennai Corporation |
The global climate model (GCM) data are available for different RCP emission scenarios, namely RCP 2.6, RCP 4.5, RCP 6.0, and RCP 8.5. In which, the RCP 2.6 scenario is very stringent with carbon dioxide emissions because it assumes CO2 emission to decline from 2020 and becomes zero by 2100. RCP 4.5 is a medium stabilization scenario with a carbon dioxide emission equivalent concentration range of 580–720 ppm for the year 2100. RCP 6.0 is a stabilization scenario where total radiative force is stabilized before 2100 without overshoot employment of a range of technologies and strategies for reducing greenhouse gas emissions, whereas RCP 8.5 is a very high emission scenario with a carbon dioxide equivalent concentration of 1,000 ppm for the year 2100 (IPCC 2014). One can observe, various measures being taken around the globe to cut down carbon emissions by 2050, and the RCP 4.5 model seems to reflect these measures for the future scenario. So, in this study, we have only considered RCP 4.5 data for our modelling purposes.
In this study, climate data from the World Climate Research Programme (WCRP), CORDEX dataset for South Asia (WAS 44) from the Atmosphere Ocean coupled General Circulation Model (AOGCM) with a resolution of about 50 km were shortlisted for this study. The models were further filtered based on the criteria such as the availability of historical paths, different RCP values, and models with the horizon year 2100. Based on the review of earlier studies (Komaragiri & Kumar 2014; Shashikanth & Sukumar 2017; Bhadran et al. 2020; Vishwakarma et al. 2022) carried out in India for evaluating the impact of climate change on precipitation, four models, namely CNRM-CM5, ICHEC-EC-EARTH, MIROC-MIROC5, and MPI-M-MPI-ESM-LR, were chosen for this study and the details are presented in Table 2. All the necessary data were georeferenced, georectified, geocorrected, and the data resolution was homogenised using the data resampling method in ARCGIS 10.5 before preparing the model inputs and analysis.
Description of selected RCMs
RCM . | GCM name . | Institution . |
---|---|---|
CORDEX | CNRM-CM5 | National Centre for Meteorological Research and Centre Européen de Recherche et de Formation Avancée en Calcul Scientifique |
MIROC-MIROC5 | National Institute for Environmental Studies, and Japan Agency for Marine-Earth Science and Technology, MIROC), Japan | |
ICHEC-EC-EARTH | Consortium of European Research Institution and Researchers | |
MPI-M-MPI-ESM-LR | Max Planck Institute for Meteorology, low resolution (Germany) |
RCM . | GCM name . | Institution . |
---|---|---|
CORDEX | CNRM-CM5 | National Centre for Meteorological Research and Centre Européen de Recherche et de Formation Avancée en Calcul Scientifique |
MIROC-MIROC5 | National Institute for Environmental Studies, and Japan Agency for Marine-Earth Science and Technology, MIROC), Japan | |
ICHEC-EC-EARTH | Consortium of European Research Institution and Researchers | |
MPI-M-MPI-ESM-LR | Max Planck Institute for Meteorology, low resolution (Germany) |
Bias correction
GCM is the fundamental source of information for developing climate scenarios and the basis for the impact assessment due to climate change at all levels. For impact analysis, GCM output cannot be used directly because the model exhibits a systematic error (bias). It is necessary to do a bias correction for the precipitation data to use it for hydrological studies. The RCM is a sub-area model of the GCM, and its accuracy depends on the accuracy of the GCM for the corresponding sub-area.
In this study, RCM data were used for the climate scenarios. There are various methods available for bias correction of RCM simulated data. In general, the available methods are classified into parametric and non-parametric methods. The parametric methods estimate the extreme precipitation more accurate than the non-parametric methods (Teutschbein & Seibert 2012; Kumar et al. 2022). The various parametric methods used for bias correction are (1) Linear Scaling (LS); (2) Local Intensity (LOCI) scaling; (3) Power Transformation (PT); (4) Delta–Change (DC); and (5) Distributing Mapping (DM) method.
The LS method is one of the simplest bias correction methods, which is used in several studies (Schmidli et al. 2006; Teutschbein & Seibert 2012), but this method has a disadvantage in that it bias corrects the RCM simulated data only based on the monthly mean precipitation values. The LOCI scaling method improves the raw data by correcting the wet frequency and intensity based on the monthly wet day threshold. However, this method is only suitable if the study area has several drizzle days (Schmidli et al. 2006). The PT method corrects the variance using the standard deviation of the data (Teutschbein & Seibert 2012), but this method is ineffective in correcting the wet day data. The DC method is a widely used method to correct the precipitation of the RCM data (Bosshard et al. 2011; Teutschbein & Seibert 2012). However, this method does not allow for a change in the extreme precipitation value due to its limitation of using a simple transformation function and multiplicative function to correct the precipitation. The DM method corrects the RCM simulated data by adjusting the mean, standard deviation, and quantiles (Teutschbein & Seibert 2012). The DM method uses a quantile mapping technique to improve the fitting of RCM simulated data to observed data, regardless of extreme values. So based on our best understanding, this method provides a more accurate and micro-level estimate of the data required for the study area.
The CMhyd tool helps to correct the precipitation data with respect to the metrological gauges in the study area (Takele et al. 2021). The CMhyd tool is used for bias correction of the precipitation data using the DM method.
Significance test for the rainfall data
Various statistical techniques are available for testing the significance of the forecasted rainfall from climate models. These techniques can be broadly classified into parametric and non-parametric methods. The parametric methods are sensitive to the skewness and also assume a normal distribution of the variables, whereas the non-parametric method does not follow the normal distribution and is not affected by the skewness of the data. So, in this study, the widely used non-parametric method Mann–Kendall and Sen's slope method (Panda & Sahu 2019; Rathnayake 2019) was used to show the statistical significance of the projected precipitation from different climate models for the confidence intervals of 95 and 90%.
IDF curve generation

SWMM simulation
The SWMM is a dynamic rainfall runoff simulator used to estimate a single-event or a continuous simulation, estimate storm runoff quantity, and quality for urban areas (Rossman 2005). Pollutant loading and washoff from sub-catchment areas are determined using the land use assigned to the sub-catchment (Rossman 2010). In this study, the time-series rainfall data, the infiltration flow derived using the curve number (CN) method, and the flow routing developed using dynamic wave method were used to estimate the surface runoff. For pollutants load estimation, land use was assigned to the SWMM using the mixed-land use method. The power function method in SWMM was used to calculate the pollutants load accumulated in the sub-catchment and the washoff pollutants load from the sub-catchment was estimated using the Event Mean Concentration (EMC) method in SWMM.
Blockage estimation
Performance evaluation of the models
Parameters used in the SWMM
ASTER DEM data were used to find the elevation and slope for the study area. The sub-catchments for the study area were delineated using the Arc Hydro tool available in Arc GIS 10.5 by adjusting the distribution of conduit features. CN was calculated using the LULC, ASTER DEM, and Soil Property data with the help of the Arc Hydro tool. Manning's Roughness coefficient value was taken from the study carried out by Andimuthu et al. (2019). The imperviousness area were calculated using the vector analysis method in QGIS 3.16. For quality analysis, the pollutant EMC values for BOD, TSS, TN, and TP were taken from various studies carried out for urban areas (Tu & Smith 2018; Tuomela et al. 2019; Choi et al. 2021). The parameters required for the SWMM were calculated using various methods and software as explained above and it is presented in Table 3.
Parameter values used in the SWMM
Parameter . | Values for ward 116 . | Values for ward 152 . |
---|---|---|
Slope for conduit | 0.01–0.028 | 0.005–0.01 |
Imperviousness percentage | 70–89 | 67–84 |
Impervious manning | 0.024 | 0.024 |
Pervious manning | 0.018 | 0.018 |
Impervious depression storage | 0.05–0.10 | 0.05–0.1 |
Pervious depression storage | 0.10–0.20 | 0.10–0.20 |
Curve Number | 81–100 | 75–86 |
Manning's roughness for conduit | 0.015 | 0.015 |
TSS (EMC) | 28.8–273 | 28.8–273 |
TP (EMC) | 0.092–0.8 | 0.092–0.8 |
TN (EMC) | 1.048–2.42 | 1.048–2.42 |
BOD (EMC) | 200–1,000 | 200–1,000 |
Parameter . | Values for ward 116 . | Values for ward 152 . |
---|---|---|
Slope for conduit | 0.01–0.028 | 0.005–0.01 |
Imperviousness percentage | 70–89 | 67–84 |
Impervious manning | 0.024 | 0.024 |
Pervious manning | 0.018 | 0.018 |
Impervious depression storage | 0.05–0.10 | 0.05–0.1 |
Pervious depression storage | 0.10–0.20 | 0.10–0.20 |
Curve Number | 81–100 | 75–86 |
Manning's roughness for conduit | 0.015 | 0.015 |
TSS (EMC) | 28.8–273 | 28.8–273 |
TP (EMC) | 0.092–0.8 | 0.092–0.8 |
TN (EMC) | 1.048–2.42 | 1.048–2.42 |
BOD (EMC) | 200–1,000 | 200–1,000 |
Questionnaire survey
RESULTS AND DISCUSSION
Present scenario results
Estimation of imperviousness area
Impervious area for the study area wards
LULU characteristics . | Ward 116 . | Ward 152 . | ||
---|---|---|---|---|
Area (m2) . | Percentage (%) . | Area (m2) . | Percentage (%) . | |
Developed land use types | ||||
Residential | 6,05,541 | 71.38 | 11,40,286 | 66.61 |
Road | 1,21,108 | 17.85 | 2,28,057 | 16.85 |
Undeveloped land use types | ||||
Park/playground | 5,703 | 0.84 | 4,007 | 0.296 |
Open space | 41,417 | 6.10 | 1,40,512 | 10.38 |
Green space | 19,941 | 2.94 | 35,719 | 3.647 |
Water bodies | 6,074 | 0.001 | 32,663 | 2.41 |
Total area (A) | 6,78,676 | 100 | 13,53,187 | 100 |
Total pervious area (B) | 73,135 | 10.78 | 2,12,901 | 16.73 |
Total impervious area (A−B) | 6,05,541 | 89.22 | 11,40,286 | 83.27 |
LULU characteristics . | Ward 116 . | Ward 152 . | ||
---|---|---|---|---|
Area (m2) . | Percentage (%) . | Area (m2) . | Percentage (%) . | |
Developed land use types | ||||
Residential | 6,05,541 | 71.38 | 11,40,286 | 66.61 |
Road | 1,21,108 | 17.85 | 2,28,057 | 16.85 |
Undeveloped land use types | ||||
Park/playground | 5,703 | 0.84 | 4,007 | 0.296 |
Open space | 41,417 | 6.10 | 1,40,512 | 10.38 |
Green space | 19,941 | 2.94 | 35,719 | 3.647 |
Water bodies | 6,074 | 0.001 | 32,663 | 2.41 |
Total area (A) | 6,78,676 | 100 | 13,53,187 | 100 |
Total pervious area (B) | 73,135 | 10.78 | 2,12,901 | 16.73 |
Total impervious area (A−B) | 6,05,541 | 89.22 | 11,40,286 | 83.27 |
Runoff estimation using SWMM for observed rainfall data
Ward 116
Details of subcatchment, conduits, and junctions in the study area ward 116.
The single event model runs were carried out for both extreme rainfall conditions in 2015 and minimum rainfall conditions in 2012. The event-based model results show that conduits C3, C9, and C13 (storm drain network) flooded for more than a day, and the remaining conduits flooded for 1–5 h. Out of 13 nodes, 12 nodes flooded in the 2015 extreme rainfall model. In specific, nodes E2 and E3 flooded for more than 10 h, and other nodes flooded for 1–5 h. The conduits (storm drain network) C3 and C13 are in Dr Beasant Road and C9 in Bharathi Salai, respectively, shown in Figure 6. All the outfalls are connected to Buckingham Canal and the maximum and minimum outflow from ward 116 was observed in E1OUT3 and F1OUT4, respectively.
The percentage of infiltration in the minimum scenario model (2012) is similar to other modelled scenarios. In the minimum rainfall model, conduit C3 flooded for more than an hour, and the nodes E2 and E3 flooded for 1 h. All three model results show that infiltration is very minimum due to the high imperviousness of the study area and the results are presented in Table 5.
Runoff quantity results for observed rainfall data from the SWMM for the ward 116
Model results for the Ward 116 . | Continuous model runoff quantity (1998–2020) . | Event-based model runoff quantity – extreme scenario (2015) . | Event-based model runoff quantity – minimum scenario (2012) . |
---|---|---|---|
Average annual depth (mm) . | Depth (mm) . | Depth (mm) . | |
Precipitation | 1,358.39 | 1,168.49 | 222.29 |
Infiltration loss | 71.32 | 9.252 | 22.36 |
Surface runoff | 1,287.02 | 1,109.252 | 200.43 |
Model results for the Ward 116 . | Continuous model runoff quantity (1998–2020) . | Event-based model runoff quantity – extreme scenario (2015) . | Event-based model runoff quantity – minimum scenario (2012) . |
---|---|---|---|
Average annual depth (mm) . | Depth (mm) . | Depth (mm) . | |
Precipitation | 1,358.39 | 1,168.49 | 222.29 |
Infiltration loss | 71.32 | 9.252 | 22.36 |
Surface runoff | 1,287.02 | 1,109.252 | 200.43 |
Ward 152
Details of subcatchment, conduits, and junctions in the study area ward 152.
The SWMM results for the observed rainfall data show that the infiltration is only 3 and 7% for the wards 116 and 152, respectively, the highly impervious nature of the study area wards reduced the infiltration of stormwater runoff significantly. The runoff for both the study area wards was more than 90% which is mainly because of the low infiltration. From the questionnaire survey, we found that the study area wards also lack a proper rainwater harvesting system which leads to an increase in the stormwater runoff.
Validation of study
Due to the unavailability of a discharge gauge for the study area wards, the event-based extreme rainfall model 2015 was validated using real-time flood mapping data and a questionnaire survey. The survey results show that ward 116 was heavily flooded during the 2015 extreme rainfall event and the streets worst affected were near Dr Beasant Road and Bharathi Salai. The SWMM results show that the conduits C3, C9, and C13 flooded for more than a day and it is important to note that the conduits C3 and C13 are located in Dr Beasant Road and C9 in Bharathi Salai. The real-time flood map (Karmegam et al. 2021) for ward 116 during the 2015 extreme rainfall event shows a water depth between 0.98 and 1.89 m. The depth of water estimated by the SWMM was 1.109 m which is within the specified range from the real-time flood map. The performance evaluation based on the surface runoff depth was found to be statistically satisfactory with NSE, R2, and PBIAS values of 0.807, 0.87, and 11.52, respectively. Hence, the SWMM developed is validated and observed to have produced accurate results.
Estimation of reduction in conduit efficiency for present scenario
Ward 116
The pollutants load was estimated using the SWMM quality model and it is used to find the reduction in conduit efficiency due to blockage. For the present scenario, the observed rainfall data for 20 years (1998–2020) was used to estimate the pollutants load. The pollutants load was estimated from the SWMM in terms of TSS, TN, TP, and BOD. The average annual pollutants load and daily pollutants load estimated for the entire catchment area from the SWMM for ward 116 is presented in Table 6. The BOD and TSS values are higher than the TN and TP pollutants load because ward 116 is completely residential. The pollutants load such as BOD, TN, and TP only affects the quality of the drain storm, but TSS not only affects the quality of the drain storm and it also blocks the conduit due to large suspended solids in it.
SWMM results
Pollutant load parameters . | Annual pollutant load (kg) . | Daily pollutant load (kg) . |
---|---|---|
TSS | 88,573 | 242.66 |
TP | 391.1 | 1.07 |
TN | 725.8 | 1.99 |
BOD | 76,162 | 208.66 |
Pollutant load parameters . | Annual pollutant load (kg) . | Daily pollutant load (kg) . |
---|---|---|
TSS | 88,573 | 242.66 |
TP | 391.1 | 1.07 |
TN | 725.8 | 1.99 |
BOD | 76,162 | 208.66 |
Forecast scenario
Bias correction for RCMs
Bias correction adjustment of RCM data for the study area wards 116 and 152.
The performance evaluation of the GCM before and after bias correction is presented in Table 7. The results show that the RCM data fit well with the observed data after bias correction. The value of R2 and NSE has improved significantly with the bias corrected data. For example, the NSE value of MPI-M-MPI-ESM-LR has improved from 0.59 to 0.94 after bias correction. Table 7 also shows that the PBIAS values of the corrected data lies within the recommended range. Hence, the model performance is found to be statistically satisfactory.
Performance evaluation of RCM data before and after bias correction
Climatic model . | Raw data . | Bias-corrected data . | ||||
---|---|---|---|---|---|---|
NSE . | R2 . | PBIAS . | NSE . | R2 . | PBIAS . | |
CNRM-CM5 | 0.845 | 0.762 | −25.75 | 0.969 | 0.960 | 10.10 |
ICHEC-EC-EARTH | 0.754 | 0.665 | −32.74 | 0.959 | 0.97 | 14.54 |
MIROC-MIROC5 | 0.791 | 0.794 | −44.74 | 0.952 | 0.956 | 5.40 |
MPI-M-MPI-ESM-LR | 0.599 | 0.682 | −46.86 | 0.937 | 0.951 | 11.26 |
Climatic model . | Raw data . | Bias-corrected data . | ||||
---|---|---|---|---|---|---|
NSE . | R2 . | PBIAS . | NSE . | R2 . | PBIAS . | |
CNRM-CM5 | 0.845 | 0.762 | −25.75 | 0.969 | 0.960 | 10.10 |
ICHEC-EC-EARTH | 0.754 | 0.665 | −32.74 | 0.959 | 0.97 | 14.54 |
MIROC-MIROC5 | 0.791 | 0.794 | −44.74 | 0.952 | 0.956 | 5.40 |
MPI-M-MPI-ESM-LR | 0.599 | 0.682 | −46.86 | 0.937 | 0.951 | 11.26 |
Projected change in the precipitation
Significant test result details for baseline and RCMs
Models . | α = 0.05 . | α = 0.1 . | ||||
---|---|---|---|---|---|---|
Z-stat . | Slope (β) . | Statistical significance . | Z-stat . | Slope (β) . | Statistical significance . | |
Base Line (1998–2018) | 1.177 | 16.34 | No | 1.177 | 16.34 | No |
CNRM-CM5 | 1.896 | 7.74 | No | 1.896 | 7.92 | Yes |
ICHEC-EC-EARTH | 2.382 | 4.65 | Yes | 2.382 | 5.11 | Yes |
MIROC-MIROC5 | 1.754 | 7.29 | No | 1.754 | 7.28 | Yes |
MPI-M-MPI-ESM-LR | −0.304 | −0.74 | No | −0.304 | −0.21 | No |
Models . | α = 0.05 . | α = 0.1 . | ||||
---|---|---|---|---|---|---|
Z-stat . | Slope (β) . | Statistical significance . | Z-stat . | Slope (β) . | Statistical significance . | |
Base Line (1998–2018) | 1.177 | 16.34 | No | 1.177 | 16.34 | No |
CNRM-CM5 | 1.896 | 7.74 | No | 1.896 | 7.92 | Yes |
ICHEC-EC-EARTH | 2.382 | 4.65 | Yes | 2.382 | 5.11 | Yes |
MIROC-MIROC5 | 1.754 | 7.29 | No | 1.754 | 7.28 | Yes |
MPI-M-MPI-ESM-LR | −0.304 | −0.74 | No | −0.304 | −0.21 | No |
Design storm
Mean and standard deviation for observed rainfall data for various duration
Rainfall duration . | Mean . | Standard deviation . |
---|---|---|
1 h | 46.17 | 25.32 |
2 h | 17.58 | 31.90 |
4 h | 73.29 | 39.53 |
6 h | 83.90 | 45.25 |
12 h | 105.70 | 57.96 |
24 h | 133.18 | 73.02 |
Rainfall duration . | Mean . | Standard deviation . |
---|---|---|
1 h | 46.17 | 25.32 |
2 h | 17.58 | 31.90 |
4 h | 73.29 | 39.53 |
6 h | 83.90 | 45.25 |
12 h | 105.70 | 57.96 |
24 h | 133.18 | 73.02 |
Intensity of rainfall for 2-year and 100-year return period
Data . | Intensity of rainfall – 2-year return period (mm/h) . | Intensity of rainfall – 100-year return period (mm/h) . |
---|---|---|
Observed rainfall | 42.01 | 157.08 |
CNRM-CM5 | 84.57 | 331.38 |
ICHEC-EC-EARTH | 54.36 | 263.52 |
MIROC-MIROC5 | 84.45 | 294.23 |
MPI-M-MPI-ESM-LR | 73.19 | 316.31 |
Data . | Intensity of rainfall – 2-year return period (mm/h) . | Intensity of rainfall – 100-year return period (mm/h) . |
---|---|---|
Observed rainfall | 42.01 | 157.08 |
CNRM-CM5 | 84.57 | 331.38 |
ICHEC-EC-EARTH | 54.36 | 263.52 |
MIROC-MIROC5 | 84.45 | 294.23 |
MPI-M-MPI-ESM-LR | 73.19 | 316.31 |
Estimation of stormwater runoff and conduit efficiency for single event
In order to estimate the carrying capacity of the conduits during various flood scenarios, single-event SWMM runs were conducted for five different return periods (2, 5, 10, 50, and 100 years) for both observed and forecasted rainfall data for the study area wards 116 and 152. In addition, the effect of pollutants load on the conduit efficiency was also studied. The model results for various rainfall return periods are presented in Table 11. The simulation results show that most of the rainfall became runoff and the main reason for the flood are inadequate stormwater drain and impervious nature of the study areas.
SWMM quantity model results for wards 116 and 152
Model Results . | Ward 116 . | Ward 152 . | ||
---|---|---|---|---|
2-year return period . | 100-year return period . | 2-year return period . | 100-year return period . | |
Rainfall (mm) | 435.68 | 1,627.37 | 435.67 | 1,627.37 |
Runoff volume (mm) | 426.67 | 1,618.19 | 420.96 | 1,613.11 |
Infiltration loss (mm) | 8.99 | 9.20 | 14.71 | 14.42 |
Model Results . | Ward 116 . | Ward 152 . | ||
---|---|---|---|---|
2-year return period . | 100-year return period . | 2-year return period . | 100-year return period . | |
Rainfall (mm) | 435.68 | 1,627.37 | 435.67 | 1,627.37 |
Runoff volume (mm) | 426.67 | 1,618.19 | 420.96 | 1,613.11 |
Infiltration loss (mm) | 8.99 | 9.20 | 14.71 | 14.42 |
Number of conduits flooded under different return periods in ward 116.
Number of conduits flooded under different return periods in ward 152.
Estimation of stormwater runoff and conduit efficiency for continuous simulation
Estimation of reduction in the conduit efficiency using continuous simulation method was carried out for two different scenarios, i.e., 2030s (2021–2040) and 2050s (2041–2060) to replicate the real-time situation where the conduits already have accumulated pollutants load.
Ward 116
For both the 2030s and 2050s scenario, the rainfall predicted by the MIROC-MIROC5 data are higher than other GCMs and ICHEC-EC-EARTH has the lowest rainfall volume. The pollutants load estimated for ward 116 in the 2030s (2021–2040) and 2050s (2041–2060) scenarios are presented in Table 12. The trend shows that with an increase in rainfall, there is an increase in the washed off pollutants load from each sub-catchment. So, the total washed off pollutants load from the entire study area has also increased.
Pollutant load estimated from the model for 2030s and 2050s for the ward 116
Scenario . | TSS (kg) . | TP (kg) . | TN (kg) . | BOD (kg) . |
---|---|---|---|---|
Model results for 2030s | ||||
CNRM-CM5 | 1,183,909 | 5,231 | 9,702 | 937,675 |
ICHEC-EC-EARTH | 880,780 | 3,890 | 7,219 | 697,801 |
MIROC-MIROC5 | 2,176,706 | 9,615 | 17,835 | 1,723,731 |
MPI-M-MPI-ESM-LR | 1,263,996 | 5,584 | 10,357 | 1,001,065 |
Model results for 2050s | ||||
CNRM-CM5 | 1,552,323 | 6,858 | 12,719 | 1,229,247 |
ICHEC-EC-EARTH | 1,051,695 | 4,645 | 8,618 | 833,115 |
MIROC-MIROC5 | 2,204,306 | 9,737 | 18,061 | 1,745,636 |
MPI-M-MPI-ESM-LR | 1,189,415 | 5,254 | 9,747 | 942,038 |
Scenario . | TSS (kg) . | TP (kg) . | TN (kg) . | BOD (kg) . |
---|---|---|---|---|
Model results for 2030s | ||||
CNRM-CM5 | 1,183,909 | 5,231 | 9,702 | 937,675 |
ICHEC-EC-EARTH | 880,780 | 3,890 | 7,219 | 697,801 |
MIROC-MIROC5 | 2,176,706 | 9,615 | 17,835 | 1,723,731 |
MPI-M-MPI-ESM-LR | 1,263,996 | 5,584 | 10,357 | 1,001,065 |
Model results for 2050s | ||||
CNRM-CM5 | 1,552,323 | 6,858 | 12,719 | 1,229,247 |
ICHEC-EC-EARTH | 1,051,695 | 4,645 | 8,618 | 833,115 |
MIROC-MIROC5 | 2,204,306 | 9,737 | 18,061 | 1,745,636 |
MPI-M-MPI-ESM-LR | 1,189,415 | 5,254 | 9,747 | 942,038 |
Ward 152
The pollutants load estimated for ward 152 in the 2030s (2021–2040) and 2050s (2041–2060) scenario are presented in Table 13. For both the 2030s and 2050s scenario, the maximum washed off pollutants load was observed for the MIROC-MIROC5 model. Similar to ward 116 the washed off pollutants load is directly proportional to the volume of rainfall.
Pollutant load estimated from the model for 2030s and 2050s for the ward 152
Scenario . | TSS (kg) . | TP (kg) . | TN (kg) . | BOD (kg) . |
---|---|---|---|---|
Model results for 2030s | ||||
CNRM-CM5 | 3,093,002 | 22,463 | 25,256 | 2,339,386 |
ICHEC-EC-EARTH | 2,279,369 | 16,562 | 18,504 | 1,724,037 |
MIROC-MIROC5 | 5,687,947 | 41,311 | 46,371 | 4,302,806 |
MPI-M-MPI-ESM-LR | 3,304,837 | 24,000 | 26,989 | 2,499,693 |
Model results for 2050s | ||||
CNRM-CM5 | 4,074,599 | 29,586 | 33,296 | 4,074,599 |
ICHEC-EC-EARTH | 2,727,995 | 19,820 | 22,170 | 2,063,411 |
MIROC-MIROC5 | 5,753,183 | 41,786 | 46,894 | 4,351,951 |
MPI-M-MPI-ESM-LR | 3,105,002 | 22,550 | 25,350 | 2,348,399 |
Scenario . | TSS (kg) . | TP (kg) . | TN (kg) . | BOD (kg) . |
---|---|---|---|---|
Model results for 2030s | ||||
CNRM-CM5 | 3,093,002 | 22,463 | 25,256 | 2,339,386 |
ICHEC-EC-EARTH | 2,279,369 | 16,562 | 18,504 | 1,724,037 |
MIROC-MIROC5 | 5,687,947 | 41,311 | 46,371 | 4,302,806 |
MPI-M-MPI-ESM-LR | 3,304,837 | 24,000 | 26,989 | 2,499,693 |
Model results for 2050s | ||||
CNRM-CM5 | 4,074,599 | 29,586 | 33,296 | 4,074,599 |
ICHEC-EC-EARTH | 2,727,995 | 19,820 | 22,170 | 2,063,411 |
MIROC-MIROC5 | 5,753,183 | 41,786 | 46,894 | 4,351,951 |
MPI-M-MPI-ESM-LR | 3,105,002 | 22,550 | 25,350 | 2,348,399 |
The summary of reduction in the carrying capacity of the conduits for various scenarios are presented in Table 14.
Reduction in carrying capacity of the conduits for various scenarios
Scenario . | Ward 116 – reduction in carrying capacity of conduits . | Ward 152 – reduction in carrying capacity of conduits . | ||||||
---|---|---|---|---|---|---|---|---|
2030s . | 2050s . | 2030s . | 2050s . | |||||
Min value (%) . | Max value (%) . | Min value (%) . | Max value (%) . | Min value (%) . | Max value (%) . | Min value (%) . | Max value (%) . | |
CNRM-CM5 | 0.046 | 25.95 | 0.052 | 21.65 | 0.09 | 19.26 | 0.12 | 24.91 |
ICHEC-EC-EARTH | 0.024 | 17.30 | 0.039 | 18.96 | 0.07 | 16.79 | 0.08 | 18.67 |
MIROC-MIROC5 | 0.080 | 35.28 | 0.068 | 40.03 | 0.17 | 34.48 | 0.18 | 39.34 |
MPI-M-MPI-ESM-LR | 0.054 | 16.54 | 0.051 | 15.28 | 0.10 | 19.02 | 0.10 | 17.32 |
Scenario . | Ward 116 – reduction in carrying capacity of conduits . | Ward 152 – reduction in carrying capacity of conduits . | ||||||
---|---|---|---|---|---|---|---|---|
2030s . | 2050s . | 2030s . | 2050s . | |||||
Min value (%) . | Max value (%) . | Min value (%) . | Max value (%) . | Min value (%) . | Max value (%) . | Min value (%) . | Max value (%) . | |
CNRM-CM5 | 0.046 | 25.95 | 0.052 | 21.65 | 0.09 | 19.26 | 0.12 | 24.91 |
ICHEC-EC-EARTH | 0.024 | 17.30 | 0.039 | 18.96 | 0.07 | 16.79 | 0.08 | 18.67 |
MIROC-MIROC5 | 0.080 | 35.28 | 0.068 | 40.03 | 0.17 | 34.48 | 0.18 | 39.34 |
MPI-M-MPI-ESM-LR | 0.054 | 16.54 | 0.051 | 15.28 | 0.10 | 19.02 | 0.10 | 17.32 |
Among all the scenarios the MIROC-MIROC5 had the maximum volume of rainfall and maximum amount of washed off pollutants loads. The high volume of rainfall and pollutants load caused a domino effect, as a result, there was maximum blockage in the conduits and significant reduction in their carrying capacity. As expected, the maximum reduction in carrying capacity of the conduits occurred in the 2050s scenario for most of the climatic models. In some cases, the 2030s scenario had maximum reduction in carrying capacity. This is due to the trend in the rainfall volume forecasted by respective climate models. As discussed earlier, the rainfall volume is directly proportional to the runoff volume and pollutants load and indirectly proportional to the carrying capacity of the conduits. This underlines the importance of using the appropriate climatic model to study the impact of rainfall in the design and management of conduits.
CONCLUSIONS
The conclusions and suggestions from this study are as follows:
The highly impervious nature of the study area wards has significantly reduced the infiltration rate to about 3–7%. The runoff in all the scenarios was more than 90%, which is mainly because of the low infiltration rate. So, this study suggests that the installation of infiltration galleries and rain water harvesting systems to improve the infiltration rate.
Evaluation of conduits based on 100 years return period flood from climate models could provide more insight in managing flood events. Outcomes from such evaluation can be used to propose additional stormwater drain networks wherever necessary or improving the existing drainage network with appropriate slope and size.
In this study we have noticed, the single event method cannot be used to study the impact of pollutants load on the conduit efficiency, and it was observed the continuous simulation method is the appropriate way.
The highest blockage and maximum reduction in carrying capacity of conduits were about 65 and 40%, respectively. The outcomes from these kinds of studies can be used to devise proper stormwater drain network management plans.
This study also observes that for an effective design and management of an urban stormwater drain network, it is important to study the combined effect of rainfall conditions based on climate data, surface imperviousness, and pollutants load.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.