An innovative tool for modeling the specific flood volume was presented that can be applied to assess the need for stormwater network modernization as well as for advanced flood risk assessment. Field measurements for a catchment area in Kielce, Poland, were used to apply the model and demonstrate its usefulness. This model extends the capability of recently developed statistical and machine learning hydrodynamic models developed from multiple runs of the US Environmental Protection Agency (EPA) Storm Water Management Model (SWMM). The extensions enable the inclusion of (1) the characteristics of the catchment and its stormwater network, calibrated model parameters expressing catchment retention, and the capacity of the sewer system; (2) extended sensitivity analysis; and (3) risk analysis. Sensitivity coefficients of calibrated model parameters include correction coefficients for percentage area, flow path, depth of storage, and impervious area; Manning roughness coefficients for impervious areas; and Manning roughness coefficients for sewer channels. Sensitivity coefficients were determined with respect to rainfall intensity and characteristics of the catchment and stormwater network. Extended sensitivity analysis enabled an evaluation of the variability in the specific flood volume and sensitivity coefficients within a catchment, in order to identify the most vulnerable areas threatened by flooding. Thus, the model can be used to identify areas particularly susceptible to stormwater network failure and the sections of the network where corrective action should be taken to reduce the probability of system failure. The simulator developed to determine the specific flood volume represents an alternative approach to the SWMM that, unlike current approaches, can be calibrated with limited topological data availability; therefore, the aforementioned simulator incurs a lower cost due to the lower number and lower specificity of data required.

A simulator to determine the specific flood volume is developed as an alternative to the SWMM model.

A sensitivity analysis extension considering rainfall and catchment topological data is employed.

The probability of failure of the stormwater system is used as a criterion to determine the necessity for corrective action under conditions of uncertainty.

Climate change and urbanization are the main factors increasing the pressure on the functioning of sewer networks, in particular the components responsible for stormwater management (Miller et al., 2014; Hettiarachchi et al., 2018; Lama et al., 2021a; Khan et al., 2022). This results in an increase in the frequency and volume of stormwater flooding, a deterioration in the living standards of the inhabitants, and pipe abrasion (Jiang et al., 2018; Zhou et al., 2019; Chang et al., 2020; Lense et al., 2023). Data from the literature (Siekmann et al., 2011) show that the basis for making decisions regarding the necessity for corrective action (replacement of a pipe, removal of sediments, construction of a reservoir, etc.) is the specific flood volume, which represents the volume of stormwater flooding on a unit of impervious surface. Limiting values for the specific flood volume have been determined by Siekmann and Pinnekamp (2011), based on simulations for urban catchments, as the basis for the maintenance of the sewage network and the criteria for making decisions regarding modernization or the necessity for corrective action.

In order to obtain the required hydraulic efficiencies, simulation models are typically used to plan corrective action (Kirshen et al., 2015). For this purpose, mechanistic models (MCMs) are used, such as the US Environmental Protection Agency (EPA) Storm Water Management Model (SWMM), which accounts for surface runoff, drainage of the sewage network, and stormwater flooding during system overload (Guo et al., 2021; Yang et al., 2022; Lama et al., 2021b). As is the case with other mechanistic models (e.g., MOUSE, PCSWMM, and MIKE URBAN), SWMM can incorporate the spatial characteristics and hydraulic conditions of a sewage network into calculations that predict and characterize stormwater flooding (Martins et al., 2018; Yang and Chui, 2021; Ma et al., 2022). However, such models are characterized by high specificity (one model can be used for one catchment), and they require the collection of detailed data and measurements (rainfall and runoff), which is labor-intensive and generates a high cost. Moreover, there are a strong interactions between the calibrated parameters (Wu et al., 2013; L. Chen et al., 2018; Sonavane et al., 2020; Shrestha et al., 2022; Ray et al., 2023), leading to uncertainty in simulation results (Ball, 2020; Kobarfard et al., 2022; Sun et al., 2022) and, thus, complicating the selection of specified corrective action (Kim et al., 2015; Babovic et al., 2018; Hung and Hobbs, 2019). To solve this problem, an important step in the development of the computational algorithm is the implementation of sensitivity analysis (Fraga et al., 2016; Cristiano et al., 2019; Razavi and Gupta, 2019). Simulations by Szeląg et al. (2021a) have shown the influence of uncertainty in calibrated SWMM parameters on the calculation of the specific flood volume and degree of flooding; this finding has also been confirmed by the simulations of Fraga et al. (2016) and Kelleher et al. (2017).

To overcome the limitations of MCMs, the implementation of statistical and/or machine learning (ML) methods is a prospective alternative (Rosenzweig et al., 2021; Lei et al., 2021; Bui et al., 2018; Shafizadeh-Moghadam et al., 2018; Chen et al., 2019; Yang and Chui, 2021; Mohammad et al., 2023). ML methods can estimate the specific stormwater flood volume for a catchment area with different topology. So far, however, no simulator model based on statistical and/or machine learning has been developed to simulate the specific stormwater flood volume while also considering the factors included in MCMs (Mignot et al., 2019; Guo et al., 2021; Rosenzweig et al., 2021). Nevertheless, some progress in the application of ML methods to the simulation of stormwater flooding has been made. Thorndahl (2009), based on simulation results of flooding from sewer utility holes, including the uncertainty in calibrated parameters, developed a model using the FORM (first-order reliability model) method. Jato-Espino et al. (2018) and Li and Willems (2020), conducting simulations with MCMs, presented (logistic regression) models for the identification of flooding from a single sewer utility hole based on rainfall frequency and catchment and stormwater network characteristics. Therefore, Szeląg et al. (2022a, b) proposed a model to calculate estimates of stormwater flooding in a catchment; however, due to the insufficient data used to construct the model, its application is limited. In the aforementioned models, interactions between land use, catchment and stormwater network characteristics, and system capacity were neglected. However, omitting these factors at the spatial planning stage reduces the applicability of the model.

Another important indicator of proper sewage network management is the assessment of the risk of system failure (exceedance of the maximum specific flood volume). Reliable risk assessment requires the integration of mechanistic models, a statistical approach, and simulation of rainfall data (Fu et al., 2011; Zhou et al., 2019; Venvik et al., 2021). Most of the methods (Ursino, 2014; Cea and Costabile, 2022; Taromideh et al., 2022) focus on determining the impact of changes in rainfall due to climate change on the efficiency of the sewage system and include the impact of parameters expressing terrain and sewer retention. Currently, there is no effective method of risk analysis that considers the uncertainty in the calibrated parameters used to simulate the specific flood volume for the different urban catchments.

The aim of this article was to develop an innovatory simulator, considering rainfall data and catchment characteristics and
topology, that could be combined with
risk assessment and sensitivity analyses to calculate the specific flood
volume. Recognition of the above factors enabled the application of the
proposed logistic regression model to the identification of stormwater flooding in
catchments with different characteristics, as an alternative approach to the
SWMM model. An important aspect of the proposed approach was the risk
assessment of system failure (a specific flood volume exceeding 13 m

The analyzed urban catchment is located in the southeastern part of Kielce in the Świętokrzyskie region,
central Poland (Fig. 1). Residential
districts, public buildings, and main and side streets are located in the study
area. The catchment area covers 63 ha and consists of 40 % impervious and
60 % permeable areas. The road density is 108 m ha

Study catchment area (Wałek, 2019).

The length of the main interceptor channel in the stormwater network is 1569 m, with an average slope of 0.71 %. The diameter of the main interceptor channel
expands from 600 to 1250 mm, while the diameters of side sewers vary between
300 and 1000 mm. The slopes of the sewers vary between 0.04 % and 3.90 %.
The analyzed stormwater system is separated from the municipal sewage.
Stormwater flows to the division chamber (DC); it then flows into a stormwater treatment plant (STP) after reaching a depth of
0.42 m. During heavy
rainfall, when the stormwater level in the DC exceeds the overflow level
(OV), it is discharged by the storm overflow into the S1 channel, which
transports the stormwater directly to the Silnica River (without treatment).
At a distance of 3.0 m from the inlet of the main interceptor channel to the DC, the MES1
flow meter is installed; this flow meter measures the flow rates during heavy
rainfall with a resolution of 1 min. Analysis of data from 2010 to 2020
showed that the measured flow rates varied between 1 and 9 dm

The analyzed catchment was divided into subcatchments (Szeląg et al., 2022a) that constituted the study areas for the identification of stormwater flooding. Due to the limited number of rainfall data, the obtained model for the simulation of stormwater overflow did not include all of the important factors, such as the dry-period duration between rainfall events and catchment retention, that impact flooding phenomenon; this meant that the model had a limited predictive capability. A detailed description of the subcatchments used for the construction of the flooding identification model and the justification of their selection were presented in Szeląg et al. (2022b). In reference to the approach proposed by Duncan et al. (2012), Jato-Espino et al. (2018), and Li and Willems (2020), the number of subcatchments used for the development of a logit model was increased to eight in the current analysis (Fig. 2). The subcatchments' boundaries and data on the spatial development and stormwater network (Table 1) were determined based on maps for design purposes, which were discussed in detail in Szeląg (2013).

Characteristics of subcatchments.

The characteristics listed in the table are as follows:

Data were verified using an independent analysis performed by Wałek (2019), who used the QGIS program to construct a spatial development model and stormwater network for Kielce. The location of closing cross-sections of subcatchments (J, K, L, M, M, O, P, R, and S) along the main interceptor channel were additionally supported by the simulation results of outflow hydrographs developed by Wałek (2019), with use of the Hydrologic Engineering Center – Hydrologic Modeling System (HEC-HMS) model, as well as by Szeląg et al. (2016, 2022b), with use of the SWMM.

The value of the specific flood volume was defined as the stormwater flooding per unit
paved area, which can be expressed using the following formula (Siekmann and Pinnekamp, 2011):

On this basis, they established a limiting

The concept of the proposed tool – a simulator integrated with risk assessment and a sensitivity analysis – to evaluate the operation of a sewage system is presented in Fig. 2. Applying the MCM of an urban catchment with separate subcatchments (varying land use and topology), a specific flood volume simulator was developed as an alternative approach to the SWMM. A logistic regression model simulator based on rainfall data, catchment and stormwater network characteristics, and SWMM parameters (width of runoff path, retention depth of impervious areas, the Manning roughness coefficient of impervious areas, the correction coefficient of impervious areas, and the Manning roughness coefficient of channels) was proposed. The resulting tool enables fast analysis of sewer network performance, even with limited data access, and can be applied to other catchments. The proposed methodology is based on the extension of algorithms given by Szeląg et al. (2021a, 2022a). In contrast to previous studies (Szeląg et al., 2022b), the current approach considers the retention of the catchment and the sewer network, and the performance criterion of the sewer network was the volume of flooding, not just the fact that it occurred. Integration of the simulator with an analytical relationship for sensitivity coefficient calculations for logistic regression allows fast evaluation of the impact of MCM parameters on flooding for arbitrary catchment characteristics and topological data.

Algorithm for developing an advanced tool to simulate the specific flood volume (situation maps in module 1a and 1b by Walek, 2019).

In order to provide more reliable simulation results, the proposed risk assessment considered the uncertainty in the SWMM parameters and enabled the optimization of the operation of the sewer network based on the maximum allowable values of the channel Manning roughness coefficients.

The proposed computation algorithm consists of eight modules. Modules 1, 2, 3, and 4
include identical steps to those in the work of Szeląg et al. (2021a, 2022a). In
the present study, the scope of the analyses was extended: in addition to
precipitation data and SWMM parameters (Szeląg et al., 2022a), the
characteristics of the catchment and the stormwater network of the separated
subcatchments were also included (module 1), which were used to determine
the computational model. On the basis of spatial data (module 1a and 1b), a
mechanistic model of the catchment was built (module 2), which allowed one to
perform an uncertainty analysis using the generalized likelihood uncertainty estimation (GLUE) method (module 3). On this
basis, simulations were performed in separated subcatchments for rainfall
events (module 1e) under uncertainty (module 4). Based on the simulation results, a
logistic regression model was developed (module 5) to calculate the local
sensitivity coefficients for calibrated SWMM parameters, with respect to
rainfall intensity and catchment characteristics (module 6). Modules 1, 2,
3, and 4 included analyses to determine a specific flood volume simulator that
could be applied to any catchment. Thus, future algorithm implementation for
the new catchment will ultimately only include modules 6, 7, and 8. Using
adopted rainfall data, the sensitivity coefficients of the SWMM model parameters
for subcatchments are computed, and maps showing sensitivity changes at the
catchment scale are drawn (module 6). While the model is applied to identify
stormwater flooding, the possible methods to improve stormwater network
operation are analyzed inside modules 7 and 8. Computations using the developed
algorithm consist of the following steps and substeps:

Input data are collected (catchment characteristics – module 1a; stormwater network characteristics – module 1b; rainfall–runoff episodes – module 1c), independent rainfall episodes are separated (module 1d), and the characteristics of subcatchments are divided and determined (module 1e).

A hydrodynamic model is developed (module 2) based on catchment characteristics (module 1a) and stormwater network characteristics (module 1b).

An uncertainty analysis is conducted with the GLUE method (Sect. 3.3.3) using a hydrodynamic model of a catchment based on rainfall–runoff episodes (module 1d).

Using independent rainfall events (module 1d), simulations with a hydrodynamic
model, including the uncertainty in the calibrated parameters, are conducted according to the following points
(4a, 4b, and 4c).

SWMM parameters (a posteriori distribution), shown in Table S1, are simulated using the results of uncertainty analysis.

Stormwater network operation during independent rainfall events is simulated (module 1d) including uncertainty (module 4a).

Specific flood volume in each sample of independent
rainfall events in subcatchments is computed, and the determined

The logistic regression simulator SWMM of the specific flood volume is determined as an alternative to MCMs based on the results of the computations undertaken in point 4c.

A sensitivity analysis is carried out according to the following points (6a and 6b).

Sensitivity coefficients (with respect to SWMM parameters) are computed for assumed rainfall data and catchment characteristics.

Sensitivity coefficients for subcatchments (J, K, L, M, N, O, P, R, and S) are computed.

The developed logistic regression model for the amelioration of
stormwater network operation is applied.

The impact of corrective variants on sensitivity coefficients in subcatchments is analyzed.

An analysis of failures occurrence is carried out.

The determination of independent rainfall events for the 2010–2021 period was
based upon criteria defined in the German Association for Water, Wastewater, and Waste (DWA) guidelines (DWA-A118E, 2006). The minimum time
period between independent rainfall events was set as 4.0 h. Computation
of stormwater flooding was performed for rainfall events with a minimum
depth of

Stormwater flood volume calculations were performed with the SWMM
using the “Flooding” function (Szeląg et al., 2021b). Based on the
results of

The model of the analyzed catchment covers 62 ha and is divided into 92
subcatchments with areas varying from 0.12 to 2.10 ha and impervious areas
ranging from 5 % to 95 %. The model comprises 82 nodes and 72 sections of
channels. At the “trial-and-error” stage of the calibration method, the
mean retention of the catchment was 4.60 mm. The Manning coefficient of
impervious areas was found to be 0.025 m

In the GLUE method, the identification of model parameters was considered to be
a probabilistic task due to the large number of parameters characterizing
processes occurring in urban catchments (e.g., runoff, infiltration, flow in
stormwater conduits, and flooding) (Szeląg et al., 2021a; Kiczko et al., 2018). The identification of model parameters in the
GLUE method depends on the transformation of an a priori distribution to an a posteriori distribution by means of a
likelihood function

Based on the results of the GLUE method (a posteriori distribution of the SWMM parameters, 5000 samples), the
computation of the stormwater network was performed separately for 175 independent
rainfall events and 9 subcatchments; 35 events were used to validate the
model. The specific flood volume values for subcatchments (J, K, L, M,
N, O, P, R, and S) were calculated, and zero–one variables were established to
develop the logistic regression model. For the computed specific flood
volume values (

A logistic regression model (LRM) is a tool used for classification. This
model has already been applied to model stormwater flooding (Szeląg
et al., 2020), identify stormwater flooding from sewer utility holes (Jato-Espino
et al., 2018), and determine the technical condition of sewage systems (Salman and
Salem, 2012). The logistic regression model is described by the following
equation:

rainfall characteristics (

SWMM parameters (

and catchment and stormwater network characteristics
(confidence level

According to data from the literature (Morio, 2011), despite simplifications, local
sensitivity analysis is widely applied at the calibration stage and while
analyzing the hydrodynamic catchment models. In our study, the sensitivity
coefficient was calculated from the following equation (Petersen et al., 2002):

The probability of a specific flood volume (

If the stormwater network ceases to function properly and the threshold
value of

The probability of a specific flood volume for rainfall with
durations in the range of

Simulation was carried out with a calibrated hydrodynamic model for rainfall data as in step a.

A comparison of the computation results obtained in steps a and b was undertaken; in the event of a good fit, i.e., proper identification of the specific flood volume, the results obtained from the LRM can be accepted. Three specific corrective variants have been defined, as presented in Table S2.

The probability of failure (Sun et al., 2022; Karamouz and Nazif, 2013) was
used to analyze the performance of the sewage network during a rainfall event.
In the calculations, a failure was defined as an episode (assumed rainfall
data, catchment characteristics, sewer network, and SWMM parameters described by the
a posteriori distribution – GLUE results discussed in Sect. 3.3.3) in which

Based on Eq. (5) for the assumed characteristics (rainfall, catchment, and
drainage network), the operating conditions of the stormwater network were
determined. Hence, an algorithm is given to calculate the performance
improvement of a sewer network in the context of failure probability
(

The a posteriori distribution of the calibrated SWMM model parameters was established (

The probability of a specific flood volume for

The Manning roughness coefficient for channels when

An empirical distribution describing the

The

The probability of a specific flood volume and the probability of
failure (

The empirical distribution (cumulative distribution function, CDF) for

Steps e to g are repeated;

Steps a to h are conducted for different catchment characteristics.

A graph of

Based on SWMM simulation results including the uncertainty in the calibrated parameters (Table S1), the likelihood functions were determined (Kiczko et al., 2018). For the observed events (30 May 2010 and 8 July 2011) used to identify the SWMM parameters, it was found that 96 % of the measurement points included the calculated confidence interval. For the validation sets, 90 % of the observation points fall within the bands for the 15 September 2010 event and 70 % fall within the bands for the 30 July 2010 event (Fig. S1). The results of the likelihood function calculations for the calibrated SWMM model parameters are given in Figs. S2 and S3 in the Supplement.

The results of variation in the specific flood volume for the separate
subcatchments are presented in Fig. 3. Based on the obtained curves,
it was stated that the uncertainty in the SWMM parameters influenced the
simulation results, which was confirmed by the great variability in the
1st and 99th percentile values for each subcatchment. The median
values enabled one to identify that the largest specific flood volume was for
subcatchment J (14.90 m

Variability in the specific flood volume for the subcatchments.

It was demonstrated that problems with the operation of the stormwater network
are present in each subcatchment, as the calculated values of the (75th and 99th)
percentiles are higher than 13 m

Comparison of LRM and SWMM simulation results for the number of
episodes in which the specific flood volume was greater than
13 m

The impact of rainfall duration (

An LRM was built based on the operational simulation of the stormwater
network. The model can be used to identify the specific flood volume and for
decision-making regarding corrective action on the stormwater system. The
relationship from Eq. (2) was described by the following linear
combination:

For the determined independent variables (Eqs. 7, 8), calculations were
performed with the LRM and SWMM models (for 35 rainfall events,

The maximum difference between the LRM and SWMM simulations (

For a rainfall depth of

These two parameters appeared to have the most significant impact on the
specific flood volume and, at the same time, they present a vastly different
impact on the dynamics of changes regarding

Probability of a specific flood volume in subcatchments for the

Due to the fact that an exceedance of
specific flood volume was observed in the analyzed stormwater network, possible improvements to the network
were considered in terms of correcting catchment imperviousness (Imp) and enhancing terrain retention and channel capacity. The results of

Simulation results for the sensitivity coefficients of other SWMM model
parameters (Table S1) and the probability of specific flood volumes are
presented in Figs. S9–S17 in the Supplement. A 10 % decrease in Imp in subcatchment
J has a negligible impact on the

Sensitivity coefficient (

The greatest reduction in flooding volume was obtained for variant III:

Based on the SWMM model parameters determined via the MCM method (Table S1), the probability of failure (

Figure 8b presents the impact of

A great impact of channel retention (

The highest failure probability (

Developing and calibrating mathematical models to simulate stormwater network operation under hydraulic overloads is one of the latest areas of research. In comparison to the models used so far (Li and Willems, 2020; Thorndahl, 2009), the LRM proposed in this study includes SWMM model parameters describing catchment retention and, at the same time, the characteristics of the catchment and stormwater network (Table 2).

Comparison of the model developed for the identification of the specific flood volume to literature data.

The following abbreviations are used in the table: M – method; R – rainfall; C – catchment; S –
sewer; P – calibration parameter; and I – interpretation model (based on
estimated factors, the impact of analyzed factors on stormwater flooding can
be determined). The models were divided into two groups: mechanistic
(•) and statistical (

Apart from the model developed in this study, the abovementioned factors are only included in MCMs that have a form of differential equations. Therefore, they require a large number of simulations in order to determine the impact of selected variables on the computation results of the specific flood volume. Free of such drawbacks are statistical models (Table S4) that take the form of empirical relationships. For models developed with neural networks, there is the need to perform additional analyses (Ke et al., 2020; Yang and Chui, 2021). Jato-Espino et al. (2018, 2019) and Li and Willems (2020) analyzed stormwater flooding from sewer utility holes based on catchment characteristics and stormwater network characteristics (Table 2). Szeląg et al. (2022b) confirmed their results and developed a model for the identification of stormwater flooding in a catchment, but they did not consider catchment retention. In this context, the approaches cited above were insufficient to analyze the impact of different types of surfaces (e.g., roof, road, and parking) on sewage flooding. Fu et al. (2011), Thorndahl (2009), and Szeląg et al. (2022a) analyzed the uncertainty in the identified parameters, which allowed them, for example, to correct for impervious area retention and the roughness coefficient without being able to correct for catchment imperviousness, which limited the use of the models in catchment management. The approach proposed in this study is a combination of these two solutions, thereby providing a tool which can be successfully implemented to manage other catchments.

The results of this study confirmed the major significance of and strong
interaction between catchment characteristics and SWMM model parameters.
This fact can be further compared to several publications (Li and Willems,
2020; Jato-Espino et al., 2019; Zhou et al., 2019) presenting comparisons
of flooding simulations in urban catchments. This analysis indicated that an
impervious area in a catchment (Imp and Impd) leads to an increase in
flooding; an inverse dependency was obtained by Jato-Espino et al. (2018)
when modeling flooding from sewer utility holes. Jato-Espino et al. (2018) found that an increase in channel volume above the
closing cross-section of a catchment (

The calculation results obtained in this study confirmed the relevant impact of
rainfall data, catchment characteristics, and stormwater network
characteristics on sensitivity coefficients (the relationships between SWMM
parameters and the specific flood volume). For rainfall data and catchment
characteristics (assumed to be constant), it was proved that the correction
coefficient of impervious area (

The sensitivity analysis development proposed in this study enabled its application to catchments with different characteristics, which is an improvement compared with previously applied, more specific, approaches (Cristiano et al., 2019; Fatone et al., 2021). Differences in the probability of occurrence/sensitivity coefficients indicate the influence of downstream catchments on the conditions in the catchment above. The variation in the sensitivity coefficients does not account for local conditions within the side channels. Due to the creation of successive subcatchments by combining them, the conditions of the sewer system in its area are averaged out, making the interpretation of the results difficult. Using the developed tool, catchment management may become difficult when there is a particularly hydraulically overloaded area within the catchment that impacts neighboring subcatchments.

As in the case of the sensitivity analysis, the extension of the sewer system failure assessment has been adapted in this study to enable its implementation in a random catchment (for a sewer system without pump stations). The calculations' outputs showed the influence of the catchment and sewage network characteristics on the failure probability. The introduction of the maximum allowable value of the Manning roughness coefficient for the sewer channel enabled one to model the improvement in the operating conditions of the sewage network under uncertainty. A similar approach was used in the study of Fu et al. (2011) by limiting the analysis to probabilistic rainfall characteristics (Del Giudice et al., 2013) and using an MCM to simulate the drainage system. Fu et al. (2011) modified the above approach by focusing on the impact of uncertainty in the calibrated parameters on flooding; however, it was not possible to analyze the effect of retention or channel capacity on system performance.

In this study, a novel simulator of logistic regression including an advanced
risk assessment extension was developed for modeling stormwater systems' operation
under uncertainty. The proposed model is an alternative approach to
MCMs, which can be used at the preliminary stage of analyses
related to spatial planning, urban development and expansion, etc. This is of
major significance because, at the preliminary stage, the data set for building
catchment models is limited, and urgent demand for a simulation algorithm to
assist decision-making is present. Assuming a Manning roughness coefficient
(

In the adopted hydrodynamic (LRM-based) model, the impact of rainfall data, catchment characteristics (impervious areas in the downstream and upstream regions), and stormwater network characteristics (the length of channel per unit of impervious area, the channel slope, and the volume) as well as the SWMM parameters (roughness coefficient for sewer channel, correction coefficient for percentage impervious area, and Manning roughness coefficients for impervious area) were included simultaneously. The obtained simulation results show the strong interaction between the above-listed parameters. This is extremely relevant in the context of model calibration that can be applied to analyze stormwater network operation and to support the decision-making process (management of stormwater in an urban catchment). As the proposed solution analyses the spatial distribution of sensitivity coefficients, it is possible to identify the most vulnerable areas inside a catchment that require specific attention while also identifying SWMM model parameters that could be considered when locating measuring facilities.

The model and codes used in this work are available from the corresponding author upon reasonable request.

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

The supplement related to this article is available online at:

BS: conceptualization; FF, BS, and AK: methodology; BS, AK, MS, and GW: formal analysis and investigation; BS, PK, AM, EW, GW, FF, and NC: writing – original draft preparation; PK, EW, AM, FF, and NC: writing – review and editing; BS, PK, AM, EW, and NC: supervision.

The contact author has declared that none of the authors has any competing interests.

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

This paper was edited by Nadia Ursino and reviewed by two anonymous referees.