Simulation of flood hazard, prioritization of critical sub-catchments, and resilience study in an urban setting using PCSWMM: a case study

Floods in urban settings are identified as one of the key natural and humanmade disasters across the globe. Moreover, floods are considered as the most expensive disasters due to rapid urban growth and changing climate (Pathak et al., 2020). In addition, with the increase in the frequency and magnitude of extreme precipitation, the present stormwater drainage system is not capable to carry the excess stormwater and subsequently increases the flood risk (Li et al., 2015; Duan et al., 2016). These disruptions have direct and indirect effects on the socioeconomic (Chang et al., 2008) and environmental conditions (Dawson et al., 2008). Most significantly, in many countries, urban flooding causes more damage per annum in comparison with other types of natural disasters (Moradi et al., 2019). Historically, most of these flood-related problems in the urban settings are centered on the larger cities and along the major streams. Since the urban settings are filled up with a larger impervious surface area, the runoff caused is up to six times greater than that with the natural land cover (Feng et al., 2021). For these reasons, flood events with 100 year return period occur more frequently (Milly et al., 2002). According to the United Nations Office for Disaster Risk Reduction (UNISDR), the statistics for 1980–2008 around the globe reported that the rate of urban flood disaster events per annum is 100 with a loss of 6,700 lives (Wannous & Velasquez, 2017). Thereafter, in 2019, out of 396 natural disasters, 194 are related to flooding compared to an average of 149 during 2009–2018 (EM-DAT, 2019) with deaths averaging 5,000 people/year making the urban floods the worst type of disaster.

Similarly, India is also experiencing floods in urban settings at an alarming rate (Table 1). In Bangalore city, 134 areas are identified as prone to flooding due to its urban growth by 63% between 1973 and 2009 (Ranganathan, 2015). Researchers and urban planning authorities are confronting a great challenge to revamp the situation in urban settings because of the increased flooding trend in recent years (Bisht et al., 2016). Using remote sensing Kit et al. (2011) demonstrated that the storm peak occurs when the stormwater flow and informal settlements overlap along the flow path of stormwater in Hyderabad.

Table 1 depicts the recent record of urban flood events in India and it was noticed that Hyderabad is facing frequent flood events. Hyderabad is one of the rapidly developing cities in India and its urbanization is increasing at an average annual rate of 3.7% since 1999. In 2000 the city faced a major flood event causing destruction of around 35,693 houses with a value of approximately INR 135 lakh (€169,000) and affected 200,000 people. Moreover, the recent event in Hyderabad city during the end of monsoon season of 2020 (October) claimed the lives of more than 80 people, displaced about 40,000 families and resulted in INR 5,000 crores (€0.57 billion) economic loss.

Generally, Hyderabad experiences peak rainfall in the period July–September, followed by significant decline from October onwards. However, in the flood event in October 2020, according to the India Meteorological Department (IMD), the city witnessed torrential rainfall of 31 cm which is the highest since 1916 and the highest rainfall for October recorded since 1903. The city has grown by 16.5% in the last 20 years. Due to this rapid urbanization and reducing number of detention ponds, the natural and artificial streams are transformed into an artificial stormwater network (O'Driscoll et al., 2010). This situation could lead to flooding in urban settings in rainy seasons. More specifically, lack of proper drainage system to drain the stormwater, reduction in infiltration rates, development along the rivers blocking the natural drains, and encroachment of lakes and other natural water bodies are some of the major causes of increased flooding in Hyderabad.

The flood situation in Hyderabad was the motivation to identify the appropriate model to simulate the flood inundation. Hence, effective and updated real time flood monitoring of the flood inundations (Sahoo & Sreeja, 2016) and hazard distribution in urban settings is of paramount importance to assist urban planners (Feng et al., 2020).

Past studies examined the ability of different models such as the Environmental Protection Agency's Storm Water Management Model (EPA-SWMM5) (Rossman et al., 2003), Hydrologic Engineering Center's Hydrologic Modeling System (HEC-HMS) (Cook & Merwade, 2009), MIKE URBAN (Bisht et al., 2016), Hydrologic Engineering Center's River Analysis System (HEC-RAS) (Rangari et al., 2019), Personal Computer Storm Water Management Model (PCSWMM) (Nazari et al., 2016), and InfoSWMM (Dongquan et al., 2009) in order to generate flood risk maps. The flood calculation for Patna, Chennai and Hyderabad urban settings are estimated using EPA-SWMM through the existing drainage networks (Vemula et al., 2019). Many studies have simulated flood inundation using 1D HEC-HMS (Rangari et al., 2019) and 1D-SWMM (Prabhakar et al., 2014). The present study agreed with the use of 1D-2D model, which performs better among 1D, 2D, 1D-1D models to visualize the flood depth and extent (Werner, 2004), whereas PCSWMM covers both hydrological and hydraulic modeling. The rainfall-runoff model in PCSWMM results gives assistance to identify inundating drainage junctions and to prepare flood inundation and hazard maps for real time flood monitoring.

In urban settings, different approaches were used previously to prioritize the critical catchments for the implementation of flood mitigation measures. Qi et al. (2022) proposed a comprehensive analysis of source tracking framework for identifying the spatial prioritization in the management of urban floods. In another study, Babaei et al. (2018) prioritized critical sub-catchments using EPA-SWMM and Preference Ranking Organization Method for Enrichment Evaluation-II (PROMETHEE-II) for Urmia city in Iran. In the present study, we attempted to integrate the PCWMM-hydrological model with the multiple-criteria decision-making (MCDM) technique, i.e., compromise programming method (CPM). This CPM has been applied successfully with GIS in the spatial variability study (Rogowski & Engman, 1996) and is also a popular method among other MCDM techniques. CPM is also used in ranking of general circulation models (Srinivasa Raju et al., 2017; Anil et al., 2021) in hydrology and most notably this method is not well used in urban flood studies.

Dong et al. (2017) found that the dual pressures of urbanization and climate change are contributing to the decrease in resilience of the urban storage system. Lim et al. (2010) also suggested adaptation of effective urban water policies to manage the urban water storage infrastructure issues caused by rapid urbanization. Butler et al. (2014) defined resilience as ‘the degree to which the system minimizes level of service failure magnitude and duration over its design life when subject to exceptional conditions’. Renschler et al. (2010) presumed that assessing urban system resilience using remote sensing and GIS plays a major role and there is still scope to explore more. To reduce the risk of flooding in urban settings, the system should store water for dry seasons and give support to recharging the groundwater table. Mugume et al. (2015) used total flood volume and flood duration for the quantification of system resilience. Wang et al. (2021) assessed flood resilience of an urban drainage system based on a ‘do-nothing’ benchmark. Hyderabad city had a huge number of storage units that acted like storage reservoirs (Chigurupati, 2009). Presently, urbanization is encroaching on the storage units as observed from the land use/land cover (LULC) of 1988 and 2020. In the present study, we attempted to simulate the resilience of the urban water storage infrastructure against excess stormwater with PCSWMM.

The purpose of the present study is to demonstrate the importance of real time flood monitoring and hazard distribution in urban settings to assist urban planners using existing and globally accepted PCSWMM. The objectives of this study are: (1) to simulate the flood inundation using coupled 1D-2D PCSWMM; (2) to create flood inundation, risk and hazard maps; (3) to prioritize sub-catchments using PCSWMM and CPM; and (4) to assess flood resilience based on integrated GIS and PCSWMM. The results of the present study are validated with the observed water logging points (WLPs) and social media marker maps.

In the present study, the urban stormwater zones of zone-XII and zone-IV&V were selected and these zones are in the northern part and southern part of Hyderabad, respectively. The study areas of zone-XII, IV&V are highly urbanized and densely populated. The extracted study area boundary maps of Hyderabad urban setting with stormwater zones are shown in Figure 1.

Generally, in an urban flooding scenario the analysis of duration and frequency of the event is necessary (Kumar et al., 2022). The present study used 43 years of annual hourly maximum precipitation data from the period 1971–2013 to find the return period of the October 2020 event (55.55 mm hourly maximum). Generalized extreme value distribution is fitted to observed precipitation data. The return period of the October 2020 event is obtained as 6 years. Figure S1 in the Supplementary Material represents the monthly rainfall comparison for the year 2020 with past IMD rainfall data.

PCSWMM is a GIS-based decision support version of EPA-SWMM developed by Computational Hydraulics International and also capable of simulating single event-based or continuous rainfall-runoff processes. The PCSWMM is a physical based, discrete-time simulation model that employs the principles of mass conservation, energy conservation and momentum. In the present study, the coupled 1D and 2D model of PCSWMM is incorporated with the routing model of semi-distributed system. Here, 1D and 2D models are used for the study of hydraulics in the storm drainage network and the overland flow model, respectively. PCSWMM is a representation of storm drainage network and the overland flow model. It shows the various sub-catchments that receive and generate runoff (Huber & Dickinson, 1988). The overland flow is calculated by specifying each sub-catchment into a nonlinear reservoir model and a dynamic wave routing model is selected to simulate the hydraulic calculations (inflow, outflow) of conduits. The pressure flow conditions in PCSWMM solve the 1D shallow water equations or Saint-Venant equations and adopt continuity-momentum equations for conduits and junctions (Xu et al., 2018). Hence, theoretically it produces more accurate results.

PCSWMM is used to extract the drainage network and to delineate the sub-catchment boundary along with their morphological data from DEM. The PCSWMM is formulated using the following parameters: catchment surface area, % area of imperviousness of sub-catchments, sub-catchment slope, sub-catchment average width, Manning coefficients, depression storage depth of impervious and pervious area, and infiltration. These parameters are obtained based on the properties of the sub-catchment and the recommendations followed from reference tables of PCSWMM support. The parameter values and ranges are summarized in Table S2 in the Supplementary Material. The infiltration attributes required by the sub-catchment layer are SUCTHEAD (suction head), CONDUCT (conductivity) and INITDEFICT (initial deficit), which are the main required input parameters. The Hyderabad study area belongs to sandy loam soil texture class. The conduit parameters of inlet and outlet nodes, conduit length, Manning's roughness of conduit and cross-sectional geometry type of each conduit should be assigned accordingly from the existing drainage network dataset. The conduits layer helps visualize the flow of water from one point to another. The conduit elevations are also derived from DEM and the conduit cross-sections are collected from GHMC.

The study areas of zone-XII and zone-IV&V are divided into several sub-catchments using PCSWMM-Watershed Delineation Tool. The rim and invert elevation of junctions are measured by using DEM. There are 1,114 conduits and 1,115 junctions in zone-XII, and 569 junctions and 568 conduits in zone-IV&V. Figure S2 in the Supplementary Material shows the delineated sub-catchments of zone-XII and zone-IV&V. The storage unit details of storage curves are prepared using Google Earth Pro. The details of zone-XII and zone-IV&V are depicted in Table 2.

The sub-catchment attributes, i.e., area (km²), width (m), % imperviousness from land use map, and slope values, are summarized for both zones in Table S3 in the Supplementary Material.

1D-PCSWMM calculates the runoff and depth of flow in the conduits, which are prepared using the existing storm drainage network data. The storm drainage network shapefile has been prepared using ArcMap and adjusted accordingly using the base map of Bing satellite, which is inbuilt in PCSWMM. The conduit cross-sections, length, conduit roughness, sub-catchment outlet connecting to junction, infiltration and attributing event rainfall data are the main inputs to simulate 1D model. The percentage of impervious and pervious areas for each sub-catchment can be derived from LULC dataset. The runoff generated from each sub-catchment is computed by considering non-linear routing and dynamic wave routing methods during model run.

The flood modeling of overland flows in 1D models is difficult to study due to multiple flow paths and flow through obstructions such as settlements. Here PCSWMM accurately combines 1D with 2D to provide depth and velocities of floods for urban settings. The Manning's roughness value of 0.025 is constant for all the 2D cells because PCSWMM does not allow representation of variable roughness values. To develop the 2D model in PCSWMM the bounding layer and 2D nodes layer are required. In PCSWMM, the DEM layer is used in the background, which defines the surface elevations of the modeled area. Moreover, the 2D cells preparation is an important segment that calculates 2D surface overland flow model and it represents the behavior of stormwater flow, which is mainly influenced by average elevation, LULC type and settlements existence in each cell. The obstruction layer shapefile (settlements such as buildings) is also incorporated in the background while creating the node layer. PCSWMM creates 2D mesh with 2D nodes layer to define overland flow paths. The hexagonal mesh created with 2D nodes and obstruction layer representing buildings are shown in Figure S3 presented in the Supplementary Material.

The directional mesh type is created for storage units. The mesh areas (m²) of the 2D cells are allowed from 10 to 250 while the majority of the cell areas are around 100 m². In PCSWMM, the resolution of grid node points should preferably be less than 100,000 to generate mesh formation which changes according to the area of catchment.

The flood inundation map can be created using the transect creator tool (TLC) in PCSWMM with the help of DEM and conduits layer. The TLC tool creates transect lines with a spacing of 25 m and length 100–200 m. In addition, to set up flood inundation analysis, a flood inundation polygon or flood inundation grid layer must be prepared. After model simulation run, in order to visualize the results of flood extent of overland flow and maximum water surface elevation, the render 2D network tool is used in PCSWMM-2D modeling option.

Urban settings have varying degrees of flood hazard depending on the type of flood and its severity. This concept is a set of interrelated behaviors (i.e., flood extent, velocity of moving stormwater, depth and duration of inundation, etc.) that indicate the likelihood of flooding in different scenarios. A proper calculation of flood hazard is necessary to manage the risk of flood mitigation activities. Risk map for the junctions is prepared after 1D model simulation based on risk classification and is presented in Supplementary Material as Table S5.

The 1D-2D flood model simulation results are presented in terms of both inundation velocity and depth. After post processing of 1D-2D urban flood analysis, the thematically rendered 2D cells help to indicate the maximum flood extent, maximum velocity and maximum flood depths.

Earlier the flood hazard studies widely followed depth of flood to classify the hazard index (Sharif et al., 2016). The flood hazard quantification process is carried out using the product of the depth of inundation (D) and velocity of moving stormwater (V) as shown in Table 3. The urban flood hazard (UFH) index is categorized into four classes (UFH1–UFH4).

Recent studies on integration of hydrological models with decision-making techniques such as the analytic hierarchy process and PROMETHEE-II are used to compute the ranking of economic performance (Rossman, 2015; Babaei et al., 2018; Safari et al., 2021).

CPM is a distance-based decision-making approach to select the best alternates that uses a distance metric called L_p to identify the ideal solution (from minimum distance), which is computed on the basis of Equation (1).

In this study, the RI is calculated for the flood event 2020 scenario denoted as RI₀ and 1988 based storage level scenario as RI₁. RI ranges from 0 to 1. Zero indicates the lowest level of resilience and 1 is the highest level of resilience to the given flooding nodes. To find the RI parameters, the integrated GIS and PCSWMM is used. The storage units of storage curves for the year 1988 are prepared using Google Earth Pro module and those curves are incorporated in PCSWMM. In order to study the flood resilience, only zone-IV and V are considered. There is not much significant change of storage volume between 1988 and 2020 in zone-XII.

Here outcomes from the modeling results of the urban flood event of October 2020 for zone-XII and zone-IV&V of Hyderabad city through PCSWMM are presented. The PCSWMM results pertaining to runoff volume, surface runoff depth, flood risk hazard, inundation mapping, prioritizing critical sub-catchments and resilience study with integrated GIS and PCSWMM are presented in this section.

Initially, during the PCSWMM simulation run process, if the continuity errors are greater than 10%, then the model input parameters need to be verified (Rossman, 2015). Here, the continuity error obtained is very much less than 10% for both the study areas.

The results of runoff values for the 2020 event are mainly simulated at the outlet of zone-XII in sub-catchment S25 which drains to Hussain Sagar Lake and for zone-IV&V the outlets are in sub-catchments S30 and S31 towards Musi River. After 1D model simulation, the summary of runoff results mainly displays the peak flow and the average flow at outfalls along with runoff coefficients. The calculated peak flow of zone-XII is 260.071 m³/s and average flow is 16.212 m³/s. Similarly, the peak and average flow at the outlet of zone IV and V together are 234.521 m³/s and 8.482 m³/s respectively.

Zone-XII: The maximum runoff depths are reported in sub-catchments S25, S22, S18, S11, S17 and the minimum in sub-catchments S10, S8, S6, S9, S5. The peak runoff is highest for S2 and lowest for S8, at 208.22 m³/s and 18.46 m³/s, respectively. Sub-catchment S13 in Industrial Development Area-phase-III and S14 in Shapur Nagar have larger impervious areas of 89.93% and 88.13% respectively. In addition, the total runoff values are high for the sub-catchments S25, S22 and S11 which have imperviousness of 86.98%, 84.87% and 87.31%, respectively.

There is no gauge data available at the outlets to calibrate and validate the results. Hence, to check the accuracy of the PCSWMM results, we have prepared inundation maps from PCSWMM for both zones and those maps are overlaid with field WLPs which are recorded during the flood event. In addition, to check the efficiency of the inundation model, the WLP locations of the flood event are prepared with social media markers (SMMs) using social media sources such as YouTube, Twitter, and newspapers. In zone-XII, 60% of WLP locations (6 out of 10) are matched with the flood inundation map and 94.7% of WLP (36 out of 38) are matched in zone-IV&V when compared to flood inundation maps. Hence, the performed model results are satisfactorily in agreement with the field WLPs.

The risk maps and hazard maps of zone-XII and zone-IV&V are obtained after the post processing of results obtained in PCSWMM. Total number of junctions presented in zone-XII is 1,115 out of which 59.7% of nodes fall under low-risk type, 1.2% are in medium risk and 39% are in high-risk condition. For zone-IV&V, the total number of junctions is 569, out of which 47.6% of nodes fall under low-risk type, 9.8% are in medium risk and 42.5% are in high-risk condition, shown in Figure S4 presented in the Supplementary Material. To visualize the outcomes of flood depths and velocities, PCSWMM is built with 2D mesh. The flood depth maps and flood velocity maps (Figures S5, S6 and S7) are presented in the Supplementary Material.

In the 1D-2D model, the hexagonal type of mesh of grid size 50 m × 50 m is adopted. Here, while considering finer resolutions such as 30 m, 25 m, and 20 m, the model could not take those resolutions of grid node points because of the large area of catchment. The high-resolution mesh might provide better inundation results of water depth and velocity. However, the comparison of different mesh sizes did not indicate differences in HEC-RAS (2D) model performance for flood hazard generation in Yesil river, Kazakhstan (Ongdas et al., 2020).

Critical sub-catchments of zone-XII and zone-IV&V were prioritized using PCSWMM along with CPM. The PCSWMM showed that the zone-XII sub-catchments of S02, S04, S17, S01 and S11 are more critical based on the peak runoff. In case of zone-IV&V, the critical sub-catchments are S24, S25, S20, S29 and S21. Prioritization of critical sub-catchments based on CPM and PCSWMM runoff volume, runoff depths and mean runoff results are shown in Table 6. According to CPM, sub-catchments S24, S13, S20, S23 and S11 for zone-XII and S25, S26, S30, S31, S24 for zone IV&V are critical. The order of prioritization for critical sub-catchments using CPM for both catchments is presented in Supplementary Material Table S8 (lower value of L_p indicates higher priority for the sub-catchment).

To study the resilience of the system, integrated GIS with PCSWMM is applied. The storage curves of the storage units present in the year 1988 are prepared using Google Earth Pro module and those curves are incorporated in PCSWMM to check the system stability for the October 2020 flood event. The results demonstrated that compared to flood event of 2020 significant change occurred in zone-IV&V. In zone IV&V, 18.25% reduction in runoff depth, 18.5% reduction of runoff volumes and 12.26% reduction of peak runoff in the stormwater network system is observed. The summary of the changes in runoff characteristics based on the storage systems available in 1988 for zone-IV&V is shown in Tables S9 and S10 in the Supplementary Material. Moreover, significant reduction of flooding hours is noticed in Mantrala Cheruvu and Palle Cheruvu as shown in the Supplementary Material (Table S11).

The urban flood modeling and sustainability study over the Hyderabad urban setting is most significant due to the area's past record of flooding, loss of human lives and impacts on the economy. The quantification of flooding is imperative for the design and development of mitigation works and management studies. Therefore, it is very important to assess the flood modeling study using PCSWMM especially in Hyderabad city, which is frequently experiencing unprecedented precipitation events.

Generally, in urban flood modeling studies the uncertainty in the results arises due to the resolution of spatial and temporal rainfall data (Notaro et al., 2013). In previous studies, researchers used the hourly rainfall and single station data for flood modeling studies using HEC-HMS, HEC-RAS and EPA-SWMM in Hyderabad city (Rangari et al., 2020; Seenu et al., 2020; Swathi et al., 2020). The present study used fine temporal resolution of rainfall data from densely sited gauge stations to get better results.

In the present study, the modeling results from PCSWMM were in line with those of observed WLPs and SMMs helped to validate the model outcomes. The results indicated that the settlements of the zone-XII and zone-IV&V regions are flooded which is in line with previous studies (Rangari et al., 2021) and also demonstrates that the stormwater drainage system network had insufficient capacity for the flood event of October 2020.

The PCSWMM helped to represent the duration of flooding at the junctions and ponds, thereby allowing the designer to find solutions that minimize the impacts of floods using LIDs. Based on the outcomes of PCSWMM 1D-2D simulation, the depth of inundation and velocity of flow are considered for generating UFH maps. In zone-XII, GHMC recorded Allapur as a major WLP location with a depth of 1.52 m; however, PCSWMM simulated from the 1D-2D model with a depth of 1.54 m which is almost identical to the recorded value. The other regions such as Bharathnagar and Fathenagar are also simulated satisfactorily with the GHMC recorded data. The findings of flood hazard maps from PCSWMM can be useful to the urban planners, government and administrative authorities to take necessary action and can allow people to migrate to safe locations during extreme rainfall events.

There is a need to prioritize the critical urban sub-catchments to optimize LID strategies, for allocation of budget and implementing mitigation plans in urban settings. The sub-catchments of S20, S11 in zone-XII and S24, S25, S26 in zone-IV&V are the most critical, according to the results of PCSWMM and CPM.

The resilience of the urban water storage infrastructure against excess stormwater for the flood event of 2020 compared with the 1988 storage level is assessed using GIS and PCSWMM. The RI for the 2020 flood event and 1988 storage level-based model results (RI₀ and RI₁) is 0.826 and 0.875, respectively, in zone-IV&V. It is observed that the rejuvenation of the storage to 1988 levels improved the resilience by about 6%. This proves that the rejuvenation of storage units which are encroached upon due to urbanization is one of the best possible ways to mitigate floods in urban settings (Water Directors of the European Union, 2003).

Due to the dual pressure of rapid urbanization and climate change, the city faces flooding and subsequent damage in the near future. Therefore, to make flood-resilient cities, we must identify the flood management and mitigation strategies (Qi et al., 2022); adopt eco-friendly solutions and improve the drainage system in line with future population-projected scenarios (Rangari et al., 2021); implement flood control activities such as LIDs and best management practices; and design urban spaces to reduce flood risks. An urban water revolution is needed in cities of the future.

The present study is mainly focused on flood modeling, prioritization of critical watersheds based on flood hazard, and potential impacts of rejuvenation of storage units on extent of flooding using PCSWMM in urban catchments. The study is based on only a single event: the extreme rainfall that occurred in Hyderabad urban setting in October 2020. The results from the PCSWMM study produced important conclusions, which are discussed below.

The 1D model inundation simulations are in line with the GHMC-field WLP locations. Good agreement between the 1D-2D simulated depth results and observation can be seen in the comparison of maximum flood depths. In zone-XII, Allapur was recorded as a major WLP with 1.52 m depth by GHMC, whereas PCSWMM simulated from 1D-2D model with a depth of 1.54 m.

The flood hazard is high in sub-catchment S20, S19, S10 in zone-XII and S23, S27, S25 in zone-IV&V. Based on the inundation depth, Allapur in zone-XII and Al-Jubail Colony, Bhagya Nagar, Gowlipura Pochamma Basthi, Hashamabad, Kakatiya Colony, Malakpet road under bridge area, Sai Nagar, and Talab Chanchalam locations in zone-IV&V are found to be extreme-hazard areas.

According to PCSWMM and CPM-based prioritization, most of the critical sub-catchments are situated in the upper part of zone-IV&V and lower part of zone-XII.

The resilience study using integrated GIS and PCSWMM results showed that rejuvenation of water bodies to their original state for the year 1988 gives 18.5% flood runoff volume reduction, 12.26% reduction of peak runoff in the stormwater network system and significant reduction of flooding hours from 172.87 to 84.21 and 73.62 to 47.91 for Mantrala Cheruvu and Palle Cheruvu, respectively, in zone-IV&V. In addition, the RI improvement with reference to 1988 storage units is 6%. The RI may improve while considering more suitable distributed storages.

The parameters used in the PCSWMM are to be considered when performing further investigations of real time flood monitoring without calibration. In addition, if runoff data is available in the future, the study can be improved using performance measures. The CPM results can be improved by considering human and economic losses of each sub-catchment. In addition, the CPM results may become more accurate while considering the regional factors that are affecting the rainfall-runoff process which cannot be performed using PCSWMM. Yet, this first-time application of the MCDM technique to urban storm water management study in Hyderabad city appeared to be realistic and provided understanding of this case study. The new integration procedures with PCSWMM and findings from the present study could help as references for policy makers and flood mitigation planners, especially for the GHMC authorities in the post-disaster emergency response. Globally, there has been an increase in precipitation frequency and intensity, as reported in the Coupled Model Intercomparison Project Phase 6 (CMIP6) of the Intergovernmental Panel on Climate Change (O'Neill et al., 2016). The rainfall-induced hazards, such as urban floods, are expected to increase in the future. Therefore, a detailed investigation and analysis of the urban catchment's response to future changes in climate, land use, and urban drainage systems will be included in the future study scopes, using the calibrated and validated PCSWMM.

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

The authors declare there is no conflict.