Effectiveness of Hydraulic and Hydrologic Parameters in Assessing Storm System Flooding

Storm sewer systems face many challenges in urban areas, in particular those systems which are old and surpassing their design period.*is study has used data from an urbanized, subcatchment covering 360 ha of dead run watershed in the Alkadeer district, Karbala, Iraq. Physically based models, Autodesk Storm and Sanitary Analysis (ASSA) and multiple nonlinear regression (MNLR), were applied. Hydrology data from 1980 to 2013 were inputted and examined over three scenarios.*e results indicated that significant increase in peak flooding was produced by an increase in discharge values which may occur through a higher rainfall intensity.*emodel was examined and new equations were developed that may help us to better understand the hydraulic and hydrologic simulations that are identified as having the potential to better protect the environment against sudden rainfall intensities.*e ratio of area of subcatchment to cross-sectional area of a pipe (Ac/Ap), the ratio of slope of subcatchment to slope of a pipe (Sc/Sp), and the ratio of velocity in subcatchment to velocity in a pipe (Vc/Vp) were the most sensitive parameters informing the ratio of runoff discharge of a subcatchment pipe discharge (Qc/Qp). *is study suggests that a more effective management of the storm water system under critical circumstances could be achieved by limiting the above parameters and this increases the efficiency of storm water facilities.


Introduction
Flooding in urban storm sewer systems could cause loss of life and considerable damage [1,2] as well as posing a threat to property and the environment and public health.Many inner cities in Iraq have been exposed to flooding for a number of reasons including poor maintenance of storm sewer systems and outmoded and limited capacity of storm systems with reference to instances of intense rain which cause blockages in pipe, electrical issues, and numerous operational problems [3].
e daily use of storm sewer systems has rapidly resulted in outdated systems because of large variations in urban land use, rising growth in population, and changing rainfall patterns [4].
City flooding during heavy precipitation is considered a natural phenomenon [5,6].e FACADE pattern is used in an object-oriented design to integrate storm flow and overland flow as well as implement an extensible integrated system for flood simulation in urban areas [6,7].Urban flooding can be simulated by a two-dimensional flow model depending on ground elevation, land use, and type of soil; all these parameters have a large impact on the degree of surface runoff [5,8].
Efficient drainage of the rain network has a strong connection with the length of drain and time of concentration of subcatchment affecting velocity and peak runoff discharges.Urbanization also impacts hydrology, which is typically characterized by raising the values of peak flooding and reducing the time lag thus causing an increase in runoff volume [9,10].
Inundation and overland flow volume may increase up to ten times because of urbanization; flooding, which normally would have occurred once every 100 years, may be doubled by more than 30% increase in urbanization [11][12][13].
While it has been reported that regular flooding in municipal areas is probably the result of frequent high intensity and extreme rainfall events (Mailhot and Duchesne, 2009), the chance of inner city flooding may increase because of the increase in urban density, especially in low-lying areas [15].Although many hydraulic and hydrology models have been   Advances in Civil Engineering widely used to predict urban flooding, there has been limited focus on different parameters, particularly regarding rapid and frequent hydrological phenomena.erefore, an investigation involving simulation modelling and data analysis was conducted to explore the relationship between the above factors to produce the best hydraulic and hydrologic model.
In recent decades, there has been much research into urban flooding with a focus on obtaining equations linking certain parameters involving topography, slope of pipes, varying rainfall intensity, size of watershed and runoff discharges, and major flooding parameters for developed areas.Attention has also been paid to the effects of climate change, varying hydraulic data on storm sewer systems, and the increasing availability of knowledge about changes in hydrological systems and insights.Many urban drainage systems can reach full capacity during heavy rainfall [16].e issue of storm sewer modelling is currently one of the most dynamic areas in hydraulics and hydrology research.Many software packages have been used to simulate pipe above ground hydraulics and hydrology models in the runoff of rainfall and pollutants from urban areas, such as SWMM [17], XPSWAMM [18], ASSA [19], and MIKE MOUSE [20].One-dimensional sewer flow models (within the storm sewer-1D) and two-dimensional overland flow (within subcatchment-2D) models have been developed and used to examine urban flooding [21].
Even though there are a number of studies conducted on flooding in storm sewer systems during heavy rainfall, these studies are still very limited in their infancy.Some 2D models with high-resolution grids have examined the consequences of storm pipes [22].High-resolution digital terrain data and instances of flooding in inner city areas were used to simulate the multifaceted interaction between the storm flow and surcharge-induced inundations in order to  e analysis of storm sewer systems in urban areas is considered as the most important measure taken to protect the environment in the event of intense rainfall.e analysis of a storm sewer system, postdesign and deployment, is considered the best way to ensure reliable up to a certain frequency or level of service and the ability to predict potential of serious ooding.Although many physically based and statistical models have been used for these purposes, the concept of having a model to simulate ows in urban sewers and to predict risk of ooding in urban areas is still comparatively inaccessible in many countries.e present work aims to investigate a model that can provide a better understanding of the ability of hydraulics and hydrology to predict and reduce risk of ooding in urban areas.

Study Area.
Karbala is an ancient festival city, situated 100 km southwest of Baghdad, Iraq (lat.-N;32 °03′51″, long.-E;44 °01′29″), in a semiarid area [25].One of the Karbala city's districts has been chosen as the study area comprising 360 ha (3.6 km 2 ), as shown in Figure 1(a).e percentage of impervious area is 60-80%.e topography map of the Karbala city and the layout of the storm sewer system which ows under gravity are shown in Figure 1(b) [26,27].

Data
2.2.1.Hydraulic Data.e storm sewer system was built in 2006 [25], with the design based on a spatially uniform rainfall intensity of 13 mm/h, according to available data.
e parameters used as input data for the hydraulic and hydrology models are as follows: subcatchment discharge (Q c ), discharge of pipe (Q p ), area of the subcatchment (A c ),

6
Advances in Civil Engineering cross-sectional area of pipe (A p ), slope of subcatchment (S c ), slope of pipe (S p ), velocity in subcatchment (V c ), velocity in pipe (V p ), time of concentration in subcatchment (T cc ), time of concentration in pipe (T cp ), length of subcatchment (L c ), and length of pipe (L p ).Details of the storm sewer system were entered into the model: manholes, pipes, pump station, topography, and land use quality.ere are 51 subcatchments (Table 1) with 73 manholes (Table 2) and 72 pipes with various diameters (315-800 mm) as shown in Table 3 [26,27].

Hydraulic Data.
e climate in Karbala is described as cold in winter (below −41 °F) and hot in summer (above 131 °F) [26].Rainfall data recorded at a rain gauge station were used as input data for the hydrology model as shown in Figure 2. All of precipitation, evaporation, in ltration, surface runo , and nal surface storage have been provided by the Directorate of Karbala and Urban planning.

Autodesk Storm and Sanitary Analysis and
Multiple Nonlinear Regressions 3.1.Methods.As illustrated in Figure 3, three scenarios of rainfall intensities have been followed.Climate changes in Iraq and particularly in Karbala city are extreme and rapid; therefore, this study has been examined using three scenarios.e rst scenario used normal rainfall intensities which were recorded in Karbala city over the 33 last years [26,27].Moreover, the second scenario used increased rainfall intensities (multiple values), while the nal scenario used rainfall intensities three times the normal rainfall intensities.

Model Description.
e proposed methodology was rst applied to the urban storm sewer systems in di erent municipalities of Karbala city based on the hydrology simulation and statistical data analysis.In this study, models based on actual hydraulic and hydrologic data were developed using Autodesk Storm and Sanitary Analysis (ASSA) [29] and multiple nonlinear regressions (MNLR) in order to accurately examine all parameters involved in an urban ooding.ASSA is considered an advanced, powerful, and comprehensive model, which is used to analyse and design storm and sanitary sewer systems in urban areas [26].ASSA uses the rational equation to examine the relationships between Q c , A c , S c , V c , T cc , and L c subcatchment parameters Weather Forecast for Karbala [29], while the Manning equation was used to express the relationship between the storm network parameters Q p , A p , S p , V p , T cp , and L p [30].MNLR [31] is an important model for estimating the frequencies of di erent variables which have nonuniform distribution [32,33].
e general method of approximation in nonlinear regression uses advanced, multiple-linear regressions that have been a ected by logarithmic feature defects [34].Nonlinear methods for a choice of free parameters and nonnormally distribution for data allowed an intercepting, nonlinear derivation.
e standard nonlinear regression model is shown by the following equation:

Advances in Civil Engineering
normally assumed for standard nonlinear regression analysis [32,35,36].Goodness of t statistical analysis and the Nash-Sutcli e model e ciency (NSE) was used to examine the results for the calibration and validation of the model, as shown by ( 2) and (3), respectively, with the relative di erence error in Q c /Q p [37][38][39][40][41]: where where O i is the observed data, P i is the modelled data, and n is the number of data.In this study, two theories have been followed.e rst calculates all the parameters of the subcatchment (Q c , A c , S c , V c , T cc , and L c ) without an urban storm water system, while the second includes a storm sewer system and uses Q p , A p , S p , V p , T cp , and L p .erefore, the hypotheses examined, based on the ratios of entered parameters, include ratio of subcatchment discharge (Q c ) to discharge of pipe (Q p ); ratio of area of subcatchment (A c ) to cross-sectional area of pipe (A p ); ratio of slope of subcatchment (S c ) to slope of pipe (S p ); ratio of velocity in subcatchment (V c ) to velocity in pipe (V p ); ratio of time of concentration in subcatchment (T cc ) to time of concentration in pipe (T cp ); and ratio of length of subcatchment (L c ) to length of pipe (L p ) as shown in Figure 4.

Calibration and Validation of the Model.
For this study, the models were tested on three scenarios applied to the 121 most intense rainy season ood events in 2013 and rainfall data from 1980 to 2013.In the rst scenario, there were 29 ooding events with an event frequency less than 0.042, and these ranged between 0.126 and 6.587 cms.In the second scenario, there were 39 ooding events with an event frequency less than 0.057, and these ranged between 0.429 and 15.232 cms.In the third scenario, there were 59 ooding events with an event frequency greater than 0.077, and the range was between 0.267 and 30.76 cms as shown in Figure 5.

(6)
With the input of three values (minimum, average, and maximum) for each parameter (A c /A p , S c /S p , V c /V p , T cc /T cp , and L c /L p ) into the above (4)-( 6), the results of Q c /Q p (min.), Q c /Q p (ave.), and Q c /Q p (max.) are extracted as shown in Table 4.
e optimal case of Q c /Q p � 1 was based on the recommended runoff surface needed to prevent flooding.Storm sewers in urban areas are only partially designed, most concentrated downstream of streets and subareas, unlike sanitary sewers which are designed to be downstream of each house.e model examined the percentage completely filled by storm network.e results indicated that the average of Q c /Q p was 0.98 which supports the idea that the flood events are correctly defined.However, the maximum value of Q c /Q p was 7.03 meaning that the subcatchment runoff discharge was larger than the discharge in the pipes, leading to flooding in the storm sewer network as defined in this study.Dependant on the minimum, average, and maximum parameters entered into the equation model, the weight of A c /A p gave equations of 73.5% and therefore is considered the most effective parameter in the equation for Q c /Q p .e next dimensionless parameters in the model are the ratios V c /V p and S c /S p .e weight of these parameters in the model was 7.8%.e ratio of T cc /T cp was 7.2% influence, the lowest impact on Q c /Q p 3.7% with the ratio L c /L p .e computed water stage is overestimated if the dynamic interaction between surcharge and discharge of the manhole is neglected in the simulation [5].However, these equations could still be used to estimate the volume of flooding.
Using logarithmic regressions, the correlations between parameters are reported in Table 5 and shown in Figure 8.
e results showed that there are significant correlations found between dimensionless parameters listed in Table 5.As shown in Table 5, the ratios were 1, −0.992, 0.0128, 0.287, 0.136, and 0.182 for Q, A, S, V, T, and L, respectively.It was also found that Q c /Q p and S c /S p had low correlations at 0.0128.erefore, a regression analysis was conducted to estimate the effects of all the parameter ratios on storm water flooding.
e normality of the data was checked using the Kolmogorov-Smirnov test, and the empirical cumulative distribution function (CDF) for observation data is examined as shown in Figure 9.
is figure shows a lack of mapping of the CDF line to the line representing normality (blue), providing evidence for a nonlinear distribution.Based on the model findings and its scenarios, both Q c /Q p and V c /V p increase with increasing rainfall intensity as shown in Figure 10.
Regarding the calibration of the model, Figure 11 shows the observed Q c /Q p versus modelled Q c /Q p based on the parameters with a confidence interval of 95%.e data show a trend towards the observed Q c /Q p data, which could be due to the nonlinear relationships for Q c /Q p with the other parameters.
All parameters determined during the process of calibration were used to validate the model, and the percent of difference is 2 (significant if less than 10) between observed and predicted data for Q c /Q p , based on the ratios of entered parameters.e efficiency of NSE was 0.33 > 0, indicating that the model predictions were as accurate as the mean of observed data.Figure 12 presents images for the parameters A c /A p , S c /S p , and V c /V p predicting Q c /Q p , an important means to understand the relationships among the parameters, and particularly the dependence of Q c /Q p on the other parameters.One of the most significant current discussions  Advances in Civil Engineering regarding this model is the possibility of using the extracted model equations to predict ooding in any storm sewer and its subareas, under extremely intense rainfall.

Summary and Conclusions
One of the major challenges when designing a storm sewer system is identifying the likely impact of identi ed parameters at the planning stage in order to eliminate or mitigate frequency of ooding.Many storm sewer simulation models exist, but this study has attempted to nd a more accurate way to examine the relationships between runo discharge and storm pipe discharge, topography of subcatchment and network slope, velocity in subareas and velocity in storm pipe, size of urban areas and cross-sectional area of storm pipe. is numerical study was carried out to determine which factor was more important based on the equations developed in the model.Conclusions drawn from this model include the following: (1) e computed water stage (7.03Q c /Q p ) is overestimated when the dynamic interaction between surcharge and discharge in manholes is neglected in the simulation [6].ese equations (3)-( 5) can also be used to estimate the volume of ooding, a new approach in this area of research.
(2) is model can be used to predict ooding in urban areas and to estimate the percentage of storm networks needed in subcatchments to limit city ooding.

Advances in Civil Engineering
(3) e calculations suggest that the use of irregular classifications of rainfall data could increase the computational complexity essential to rebuild the model and significantly increase the time required for simulation runs.In compensation, the relationship between parameters entered into the model was logarithmic, and the distribution of data is nonlinear.( 4) is study recommends that the design and analysis procedure for a storm sewer system must include the mitigation of the risk flooding in urban areas.
ese results may simplify developments in the prediction of urban flooding and thus protect resources as well as reduce flood risk.Given that storm sewer systems have historically been designed to consider limited rainfall intensity, this model can allow for varying rainfall intensities in unusual situations.e model has also established a clearer relationship between relevant parameters.It may therefore be useful worldwide in a range of different scenarios.

Figure 1 :
Figure 1: (a) Elevation map of Karbala city.(b) Map of study area Alkadeer district in Karbala city centre, Iraq, generated by GIS based on available data adapted from [26, 27].

Figure 2 :
Figure2: Average monthly rainfall and temperature in the study area for the years 1980-2013 adapted from Weather Forecast for Karbala[28].

Figure 3 :
Figure 3: Rainfall intensities for the three scenarios of the model [28].

Figure 4 :
Figure 4: Model of the hypotheses model for the parameter ratios (3 scenarios).

Figure 5 :Figure 6 :
Figure 5: (a) Daily peak ooding based on three scenarios for available rainfall data (1980-2013).(b) Relationships between rainfall, runo , and ooding over time for three scenarios of the model.

Figure 7 :
Figure 7: Normal and detrended Q-Q plots for all the parameters entered into the model.

Figure 8 :Figure 9 :
Figure 8: Correlations for all the parameters related to Q c /Q p .

Figure 11 :
Figure 11: Model calibration of Q c /Q p based on the parameters with a confidence interval of 95%.

Figure 12 :
Figure 12: ree-dimensional surfaces of the model parameters.

Table 1 :
Data about the subcatchment area input to the model.

Table 2 :
Data about the storm sewer manhole input into the model.

Table 3 :
Data about the storm sewer pipes input into the model.

Table 4 :
e results of Q c /Q p by the model equations.

Table 5 :
Partial correlations for Q c /Q p , A c /A p , S c /S p , V c /V p , T cc /T cp , and L c /L p .