Natural Hazards Research xxx (xxxx) xxx

Contents lists available at ScienceDirect

Natural Hazards Research

journal homepage: www.keaipublishing.com/en/journals/natural-hazards-research

Urban stormwater management under various runoff conditions for Arba

Minch town of southern Ethiopia
Bereket Ayele Ayda a, Demelash Wondimagegnehu Goshime a, *, Mekuanent Muluneh Finsa b, d,
Abebe Temesgen Ayalew c
a
Faculty of Water Resources and Irrigation Engineering, Water Technology Institute, Arba Minch University, P. O. Box 21, Arba Minch, Ethiopia
b
Water Resources Research Centre, Water Technology Institute, Arba Minch University, P. O. Box 21, Arba Minch, Ethiopia
c
Faculty of Hydraulic and Water Resources Engineering, Water Technology Institute, Arba Minch University, P. O. Box 21, Arba Minch, Ethiopia
d
Institute of Hydrogeology, Engineering Geology and Applied Geophysics, Faculty of Science, Charles University, Czech Republic

A R T I C L E I N F O A B S T R A C T

Keywords: Ineffective urban stormwater management causes increased surface runoff, and impacts on infrastructure and
Urbanization human life. However, urban stormwater condition is not quantified in Arba Minch area, which cause significant
Stormwater damage on the infrastructures as residence and roads. Therefore, the main objective of this study is to evaluate the
SWMM
condition of urban stormwater under different runoff conditions using Stormwater Management Model (SWMM)
Runoff
in Arba Minch town, specifically Sekela sub city. A combination of field, satellite, observed and laboratory test-
Urban drainage
Arba Minch based data on drainage systems, infiltration and % imperviousness was collected. The catchment is divided into
47 sub-catchments depend on the building block, flow direction, and drainage networks. The sub-catchments with
the average surface runoff revealed 0.07 m3/s, 0.14 m3/s, 0.25 m3/s, 0.65 m3/s, and 0.75 m3/s over a 2-year, 5-
year, 10-year, 25-year, and 30-year return period, respectively. The maximum nodes flooded occur at junctions’
number (J7, J8, J24, and J25), as well as conduits (C14, C18, and C27), according to simulation findings of 2-year
return period design rainfall which align with the major cause area. For return periods of 2, 5, 10, 25, and 30
years, the peak runoff result was greatest in sub-catchments SC-1, SC-6, SC-8, SC-9, and SC-30. This depicts sub-
catchments in the area are flooded, hence resizing of canals and other drainage structures are necessary to
mitigate the stormwater. The SWMM simulated results for 2-year, 5-year, 10-year, 25-year, and 30-year return
periods showed peak runoff was greatest in sub-catchments No.8, 1, 6, 30, and 9 because of the slope of the sub-
catchment and their large area with high percent imperviousness that results in less infiltration into the soil. The
finding of this study shows a rapid increase in surface runoff, which mainly related to impermeable surface and
inadequate urban drainage systems leading to damage to infrastructures and properties. This study suggests, a
mitigation measures should be adopted to minimize the effect of stormwater using stormwater various inter-
vention approaches following multi program combined with various runoff condition which is very important for
road protection authorities, decision makers, and the scientific community to support future water related
researches.

1. Introduction aforementioned studies, urbanization (land use and land cover changes)
has a direct influence on runoff production. Ali et al. (2018) reported that
Flooding in cities is becoming a global problem as cities expand and replacement of pervious areas with impervious surfaces causing
the number of impermeable surfaces increases, resulting in enormous increased peak surface runoff and volumes, and less opportunity for
quantity of surface runoff (Feng et al., 2021; Zhang et al., 2020; Arnone infiltration. The replacement of land use alterations brought on by ur-
et al., 2018). Globally, the effects of urbanization and urban stormwater banization will result from a significant change in land use and land
management need special attention because they often have a visible cover, which replacement of highly pervious areas with impervious re-
impact on hydrologic and runoff systems (Gong et al., 2020; Q Wang sults in an increase in surface runoff volume and decrease infiltration rate
et al., 2020; Shen et al., 2019). We note that from all of the (Anim et al., 2019; Dennis et al., 2019; Shrestha et al., 2021).

* Corresponding author.
E-mail address: demelash.wondimagegnehu@amu.edu.et (D.W. Goshime).

https://doi.org/10.1016/j.nhres.2024.02.003
Received 27 November 2023; Received in revised form 23 January 2024; Accepted 25 February 2024
2666-5921/© 2024 National Institute of Natural Hazards, Ministry of Emergency Management of China. Publishing services provided by Elsevier B.V. on behalf of
KeAi Communications Co. Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article as: Ayda, B.A. et al., Urban stormwater management under various runoff conditions for Arba Minch town of southern Ethiopia,
Natural Hazards Research, https://doi.org/10.1016/j.nhres.2024.02.003
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

Fig. 1. Location of the study area.

Urbanization causes a shift in drainage patterns, faster overland flow, et al., 2023; Hou et al., 2019; Sun et al., 2013). It is reported that
flooding, and a rise in environmental issues like land degradation increased runoff, reduced runoff concentration-time, decreased infiltra-
(Ibrahim et al., 2023; Bekele, 2021). Authors reported that, it also causes tion will result in flooding and damage to drainage systems. Therefore,
an increase in impermeable surfaces, such as houses, paved roads, and structural or non-structural stormwater management is the primary
parking lots. Several studies reported that the consequence of urbaniza- concern, it has to blend in with the surroundings. In any location, there
tion can not only increase surface runoff but also disturb the planning and may be numerous options for managing stormwater and providing water
design of drainage systems (Willems, 2013; Davies et al., 2008). For quantity (Belachew, 2019).
example, long-term changes in hydrological regimes, especially surface In Ethiopia, where many urban watersheds receive the majority of the
flooding routes and characteristics, can lead to the transformation of city annual rainfall the intensity of the rain is typically high. Control of runoff
land use and land cover and drainage systems to avoid runoff distur- at the source, flood protection, and safe disposal of excess runoff through
bances and damage. Urbanization also decreases the quality of the sur- suitable drainage systems become critical in developing countries, where
face water as pollutants collected on the impervious surface are washed the watersheds of major city centers receive major amounts of annual
and mixed with the surface water. Ogden et al. (2011) reported that rainfall and rainfall intensity is generally high. Studies reported that
increased flood maximum magnitudes and increased runoff quantities having adequate drainage infrastructure is necessary, including flood
are the typical hydrologic responses to urbanization. protection, runoff control at the source, and safe runoff disposal through
The change caused by urbanization, which is related to land use, appropriate drainage systems (Ahmed et al., 2019; Jemberie et al., 2023;
mainly alters rainfall-runoff generation processes in multiple ways Bulti and Abebe, 2020; Dibaba 2018; Ishimatsu et al., 2017). Several
(Wang et al., 2020). The scholars revealed that in the process of urban studies studied for design of drainage system due to poor stormwater
development, population, impermeable surface, building density, and management (Ahmed et al., 2019; Jemberie et al., 2023; Nakhaei et al.,
green space are extremely distributed within urban areas. Land use land 2023). However. most of the studies do not apply integration study of
cover alterations have been considered as a serious concern in world stormwater with land use landcover and intensive return period to crit-
environmental change (Mekonnen et al., 2018; Yohannes et al., 2018). ically determine the causes for poor stormwater management. Hence, the
The expansion of urbanization has its impact to land use and land cover main motivation of this study is to address and fill this gap in aspiration
changes (Rufino et al., 2018). An area that has less infiltration exposed to with detail investigation.
peak runoff increased impervious surface percentage in built-up areas Proper urban storm water management plays a vital role in the
can raise peak discharges in an urban area, resulting in larger and more effective operation of urban drainage systems. Proper urban storm water
frequent runoff events. During high flood conditions urbanized areas are drainage is typically intended to collect and convey excess surface runoff
susceptible to flooding under high precipitation conditions (Alshammari successfully to prevent urban flooding (Arnone et al., 2018). However,

2
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

Fig. 2. Land use the land cover of the catchment.

most developing countries regularly find that the majority of the stormwater management measures as a sustainable solution to handle the
drainage systems do not operate at full planned capacity. The drainage drainage problems and the serious consequences on the environment
systems of planned capacity are decreases due to poor urban storm water including on the inhabitants, and existing infrastructure such as supply
management, such as improper maintenance, inappropriate design, lines, buildings, roads, and others, essentially during the wet seasons. The
siltation, and the absence of adequate integration between urbanization main significance of the study is to produce good understanding and
and storm water drainage network. filling the gaps for planner, decision makers, stakeholders and any con-
Naeimi and Safavi (2019) stated that stormwater management is a cerned person to the consequences of inadequate drainage systems
most significant consideration in the design of urban drainage systems. (stormwater management) on runoff production and solve urban
Stormwater management applications, when properly selected, designed, stormwater drainage systems problem that contributes to better service
and implemented, can be used to mitigate the adverse hydrologic and coverage in the area.
hydraulic effects caused by drainage facilities, thereby protecting
downstream areas from increased flooding, erosion, and water quality 2. Materials and methods
degradation (Tefera, 2020). Authors reported that downstream convey-
ance constraints, particularly in cases where the roadway drainage sys- 2.1. Study area
tem connects to existing drainage systems, may limit the peak discharge
to the capacity of the downstream system. The research area is located in Ethiopia's southern region, at 505 km
Previous studies on the impacts of urban stormwater management on distance from Addis Ababa. Geographically, it is located between
runoff production have been not assessed in Arba Minch town especially 6
10 14.1800 N of latitude and 37
330 50.800 E of longitude as indicated in
using best storm management model. Therefore, this study aims to apply Fig. 1. Arba Minch experience extreme seasonal variation with bimodal
Stormwater Management Model (SWMM) which subdivides the general climatic condition. The mean annual rainfall and temperature ranging
catchment into sub-catchments, predicting runoff from sub-catchments from 598 mm to 1245 mm and 15.8
C to 33.5
C, respectively.
based on their individual properties and combining their outflows The large portion of Arba Minch town land is existence occupied by
using a float routing scheme. In this model, surface runoff is determined the impervious surface which includes buildings, roads, or Asphalt,
by combining mass balance equations with Manning's equation. cobble stone roads. As well as in the town urbanization increasing
In Arba Minch town, this conveying stormwater through stormwater because the Arba Minch town is rapidly growing and there are more
drainage systems is not properly managed, which results in large chances of infrastructure development which will convert the major part
amounts of silt in most of the drainage networks. This creates poor of the town into an impervious surface. Overall, in the city the magnitude
integration between road and stormwater drainage systems, with over- of urban stormwater in the area is severe due to poor drainage system and
flowing on the road surface, and water stagnation during the rainy sea- other geophysical conditions which mainly cause the town to susceptible
son. These problems are the main causes of runoff production and to urban flooding and stormwater management.
overflow in drainage systems in Arba Minch town and still have not been
implemented on other types of stormwater management system prac-
tices. Hence, the main objective of this study is to model urban storm- 2.2. Database
water management and its contribution to runoff production. The
specific objectives include (i) determining the magnitude of surface In this study, ground-based and satellite climate data, land use, soil
runoff production for various recurrent intervals, (ii) assess the capacity and digital elevation map (DEM) spatial data were used. Daily climate
of existing stormwater drainage systems (iii) evaluating the spatial data were obtained from the Ethiopian Meteorological Institute (EMI),
variation in runoff production due to urbanization. mainly Arba Minch town station. For this study, 30 years of historical
The finding of this study will contribute to taking appropriate urban observation rainfall data (1990-2019) were used to calculate (forecast)
the peak drainage stormwater of the Arba Minch town, and hourly

3
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

Fig. 3. Conceptual methodological framework of the study.

rainfall data for calibration were obtained from the Arba Minch univer- construction records, and material property reports including building
sity automatic weathering station on a date that coincided with the flow drawings were obtained from the city administration. The drainage
depth measuring date at each outfall in the catchment. The historical network documentation for the town and direct survey measurements
rainfall data contain 2.9% missing record rainfall data. were used to determine the characteristics of the urban drainage network
The observed rainfall data gap was filled by using Climate Hazard (surface elevation, maximum depth of junctions, length, form, and slope
Infra-Red precipitation with stations (CHIRPS) satellite rainfall data of the storm drainage conduits).
(Belay et al., 2019; Roberts et al., 2019; Avalon and Al, 2023). CHIRPS is The land use and land cover map of the study area was derived from
a multi-source product database developed by the U.S. Geological Survey Google Earth Pro-satellite image by using on-screen (object-based) digi-
(USGS) and the University of California, Santa Barbara (UCSB) (Funk tization technique analysis. This technique is suitable at high spatial
et al., 2015). CHRIPS data available on a daily basis was resolution of less than 1m, to visually see the image such as building
https://chg-ftpout.geog.ucsb.edu website at a variety of time steps, roofs, roads, water bodies, vegetation areas, etc. Finally, the object-based
spatial domains, and formats from 1981 to the current time steps. The digitizing in ArcGIS can easily be performed to prepare the land use land
spatial resolution of CHIRPS data varies from 0:05
 0:05
degree and cover map. Then, these datasets were used as main input for SWMM
the data are provided in NetCDF, GeoTiff, and Esri BIL format. We per- model to determine the Manning coefficient, land cover factor, and
formed a data quality test to avoid uncertainty in the data before final use runoff coefficient. Accordingly, the main LULC map of the study area
in the modeling. covers digitized roof, road, bare land, and green area (Fig. 2).
The field survey data also conducted to identify the status of the
drainage that was properly functioning, the shape of drainage systems, 3. Materials and methods
and runoff flow depth data was measured in drainage systems at three (3)
outfalls with a staff gauge instrument for two events, which are then used For modeling urban stormwater purposes, the Stormwater Manage-
for model calibration (rainfall data of September 18, 2021), and valida- ment Model (SWMM) was calibrated and used to compute storm
tion (rainfall data of July 24, 2021). drainage system flows, conduit surcharged flows, and to simulate
The soil texture has been estimated in the soil mechanical laboratory flooding of the urban drainage system in the study area. The main ma-
by using a soil hydrometer test mainly to measure the infiltration rate. terials used in this study include Easy Fit software, Google earth Pro,
This infiltration rate is the input parameter in the SWMM model. ArcGIS, and SWMM model. Easy Fit software was used to select the
Other relevant secondary data such as design documents, appropriate probability distribution function. Whereas Google Earth Pro

4
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

was used to delineate the study area, divide sub-catchments, and along In this study, three goodness fit test namely Chi-squared, Ander-
with ArcGIS to estimate the imperviousness area. Infiltration rate data is son-Darling, and Kolmogorov-Smirnov statistical tests were used. The
measured with a double ring infiltrometer and measuring dimension, appropriate rainfall frequency distribution is found using the L-moment
invert elevation and surface elevation at each sub catchment (maximum ratio diagrams, which is based on approximations (Shao et al., 2021). A
and minimum elevation) with hand GPS for input parameters in the diagram is based on the coefficient of skewness (LCs) versus the coeffi-
SWMM model, and take a random soil sample to identify soil texture in cient of Kurtosis (LCk) is used to analyze the frequency of rainfall. For
the study area. The general procedure followed in this study was pre- instance, the generalized Pareto (GP) frequency distribution function
sented in Fig. 3. which is selected in this study were used to predict the extreme maximum
rainfall value, and the design rainfall for different return period is given
3.1. Intensity duration frequency curves (IDF) in (Eq. (2)).
a
Design rainfall intensity was derived for different return periods able XT ¼ ε þ ð1  TÞk (2)
k
to determine peak runoff from each sub-catchment and used as an input
parameter in the simulation of runoff from each sub-catchment using where the location (ε), scale (α), and shape (k) parameters were esti-
SWMM (Tesema et al., 2021; Bibi and Tekesa 2023). Due to the absence mated using Easy-Fit 5.5 professional software.
of a shorter duration of recorded rainfall data for periods less than 24 h, it
is difficult to develop an IDF curve for a given area for a rainfall for less
than 24 h. In this study, a total of 30 years with 24-h rainfall data sets is 3.3. Stormwater management model (SWMM)
prepared to develop the IDF curve. Ethiopian Road Authority (2013)
suggests to convert from 24-h duration rainfall into shorter duration For modeling urban stormwater, a Stormwater Management Model
rainfall using an empirical reduction formula shown in (Eq. (1)) for the (SWMM) was calibrated and able to compute stormwater flow, conduit
estimation of various durations 1-hr incremental for 24-hr rainfall values surcharged flows, and to simulate flooding condition of Arba Minch city.
from annual maximum values. The equation reads: For calibration, the Personal Computer Stormwater Management Model
(PCSWMM) was used.


tðb þ 24Þn SWMM is a comprehensive computer model for simulation of urban
Rt ¼ n *Rd;t (1)
24ðb þ tÞ runoff quantity and quality in the storm and combined sewer systems.
This model is capable of simulating the precipitation movement from the
Where n and b are constant value (b ¼ 0.3 and n ¼ 0.92), Rt is rainfall ground surface through channels and pipe networks (Trajkovi et al.,
depth for any duration (t), and Rd,t is the 24 h rainfall. 2020). A single event and a long continuous period of the event can be
simulated by using SWMM. It is similar to the most commonly used hy-
3.2. Rainfall frequency analysis drologic fashions in urban areas, flood modeling, and analysis, which
determines drainage size by taking into account pervious and impervious
Daily rainfall of the highest magnitude is used to select the most areas (Shafiquzzaman et al., 2020).
suitable one from several often-used probability distribution functions by The SWMM model has the advantages of simple operation, short
L-Moment ratio diagram and easy-fit- software goodness of fit test calculation time, and relatively reliable calculation results and it is
methods (Nakhaei et al., 2023; Giełczewski et al., 2022; Lei et al., 2018). appropriate for modeling areas served primarily by a drainage network

Fig. 4. Sub-catchments and existing drainage network of the catchment in SWMM.

5
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

Table 1 catchment, including average slope, perimeter, area, and width, was
Horton infiltration parameter input into SWMM using soil texture. produced by this division. The study area of the existing drainage
No Site location Soil fo fc K (1/ network is represented in SWMM as shown in Fig. 4. It consists of a rain
Sample Texture (mm/ (mm/ hr) gauge station, 47 sub-catchments that are connected by 43 nodes (40
Latitude Longitude
Class hr) hr) junctions and 3 outfalls), and 40 conduits.


1 6.042327 37.553318
Silty 231.52 60 3.8772
2 6.038597
37.559370
Silt 196.21 47.143 4.8147 3.5. Horton's equation parameter
Loam
3 6.042099
37.555281
Silty 144.49 9 3.37
Clay
In this study, infiltration rate measurements were made in the field of
Loam the study area at each selected soil sample point 1, soil sample point-2,
and soil sample point 3 depending on the soil texture shown in
Table 1. The Horton infiltration parameter was obtained from the Double
without a significant component of surface flow (Gong et al., 2020; Shen Ring Infiltrometer in the field of the digitalized area and the Horton
et al., 2019), but it has disadvantage such as rough division of infiltration parameter value varies with soil sample location.
sub-catchments and complex parameter adjustment (J Wang et al.,
2020). 3.6. Components of SWMM model
SWMM model is selected for this study due to its proven performance
on Ethiopian catchments (Takele, 2022). Furthermore, the calibration 3.6.1. Surface runoff
and validation of its uncertainty parameters need to rely on measured The continuity and Manning's equation are combined to achieve float
flood data and also it is suitable used to stormwater runoff simulation routing in the channel (Blum et al., 2020). According to the EPA-SWMM,
(Takele, 2022). In SWMM, surface runoff is determined by combining a sub-catchment area with a rectangular area that has a single outlet
mass balance equations with Manning's equation. The degree of data channel and a uniform slope (S), width (W), and the depth (d) of the net
needed and software limitations will determine the selection of a suitable excess ponds above the sub-catchment surface. Runoff outflow (q) can be
modeling package. caused by water that has accumulated above the depression storage
The main input data required for SWMM include Digital Elevation depth (ds). Storage of a depression is defined by its initial rainfall ab-
map (DEM), Land us, rainfall, width, area, percent of imperviousness, stractions including surface ponding, a plant's intercept, and surface
slope, and soil property, infiltration rate for the previous area, and wetting (Giudicianni et al., 2023). The difference between a single inflow
Manning's roughness coefficient and depression storage for both pervious or rainfall intake and outflow or discharge in terms of infiltration,
and impervious areas among them to mention. From this, all initial input evaporation, and surface runoff over the sub-catchment, as shown in (Eq
parameter will be provided for the model to run. All initial input pa- (5)), results from the conservation of mass. Then, inserting the specific
rameters were determined based on the SWMM user's manual and other parameter in (Eq (5)) gives (Eq (6)).
literature and estimated according to the properties of the studied area.
∂d
¼I O (5)
3.3.1. SWMM model processes ∂t
SWMM is a discrete-time simulation model with a physical founda-
tion which it applies the laws of conservation of mass, energy, and mo- ∂d
¼R e f q (6)
mentum (Takele 2022; Fandel et al., 2021; Charalambous et al., 2019). ∂t
The conservation of mass and momentum equations control the erratic
Where I (Input parameter); R (Rainfall rate in mm/s); O (output
movement of water through the drainage system's pipes and channels.
parameter); e (surface evaporation rate in mm/s), f (infiltration rate in
For flow via a single conduit, these equations, also referred to as the Saint
mm/s), and q (runoff rate in mm/s).
Venant equations (Eq. (3)), can be stated as follows (Baselt and Andreas,
The Manning equation can be used to define the runoff's volumetric
2022) and explicitly described in (Eq. (4))
flow rate (Q) as shown in (Eq. (7)), assuming it behaves as if it were
∂A ∂Q uniform flow within a rectangular channel with width W, height (h), and
þ ¼0 (3)
∂t ∂x slope (S). The equation reads:
   1:49 1=2 2=3
∂ Q ∂ Q2  A ∂H Q¼ *S *R *A (7)
þ þ gA þ gASf þ gAhL ¼ 0 (4) n
∂t ∂x ∂x
In this equation, n stands for the coefficient of surface roughness, S for the
where x is the distance along the conduit, t is the passage of time, A is the
apparent or average slope of the sub-catchment, A for the runoff flow
cross-sectional area, Q is the flow rate, H is the hydraulic head of water in
area throughout the sub-catchments’, and R for the hydraulic radius
the conduit (elevation head plus any potential pressure head), Sf the
associated with this area. A is a rectangular space with a width of W and a
friction slope (head loss per unit length), hL the local energy loss per unit
height (h) equal to d-ds, as shown in (Fig. 3). Since W will almost always
length of conduit, and g the gravitational acceleration. Note that for
be significantly larger than d, A ¼ W (d * ds) and R ¼ d * ds are the logical
known cross-sectional geometry, the area (A) is a known function of flow
conclusions. Hence, (Eq. (7)) is produced by substituting these terms to
depth (y) which in turn can be obtained from the head (H). Thus, the
produce (Eq. (8)).
dependent variables in the above equations are flow rate (Q) and head
(H), which are functions of distance (x) and time (t). 1:49 1=2
Q¼ *S *ðd  ds Þ5=3 (8)
n

3.4. Sub-catchment and drainage network system Eq. (8) is divided by the surface area of the sub-catchment A to reveal
(Eq. (9)) which is used to calculate the runoff flow rate per unit of surface
The boundary of the study area is digitized by different polygons area, q.
using Google Earth Pro and Arc GIS. These polygons were uploaded into =
1
1:49*W*S 2 5 =
SWMM, and then the sub-catchment and drainage networks were created q¼ *ðd  ds Þ 3 (9)
based on urban drainage systems, topography, building blocks, and A*n
drainage flow direction. The primary information for each sub- (Eq. (10)) is produced by substituting this in (Eq. (6)) into the mass

6
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

3.7.2. Percent of impervious area

In this study, the percentage of imperviousness was determined using
a map of land use and cover. ArcGIS 10.7 anticipated four categories of
land use based on the land use map: namely building roofs, green space,
bare ground, and roads (pavements). The quantity of impervious area
within the sub-catchment was added, and the percentage of impervi-
ousness was determined by dividing the total area of the sub-catchment
by that amount.

3.7.3. Sub-catchment width

The size of the sub-catchment is divided by the length of the overland
Fig. 5. Non-linear reservoir model of a sub-catchment (Giudicianni flow to determine the width of the sub-catchment, which is typically used
et al., 2023). by the model as a calibration parameter. Due to the difficulty in esti-
mating the hydraulic length of overland flow, the sub-catchment width
balance relation (Eq. (3)). Therefore, the values of the surface runoff flow parameter is calibrated to offer a good match between actual and simu-
rate (q) can be found in (Eq. (10)). lated values. In this study, the initial width parameter for the sub-
catchments is determined using (Eq. (11)). Note that the width param-
∂d
=
1
1:49*W*S 2 5 =
eter is susceptible to change throughout the calibration procedure. The
¼R  e  f  *ðd  ds Þ 3 (10)
∂t A*n width of the sub-catchment is estimated using:
Overall, the basic input parameters required for SWMM to simulate
ASC
the runoff quantity are rainfall and the catchment's physical character- W¼ (11)
L
istics. Fig. 5 shows the non-linear reservoir model for SWMM.
where W is the sub-catchment's width, ASC is a collection of sub-
3.6.2. Infiltration rate catchments, and L is the sub-catchment's longest hydraulic length
SWMM uses three models inside its system section such as Curve (overland flow).
Number models, Horton's models, and the Green-models for infiltration
computation (Singh, 2018). In this study, Horton's models were used to 3.7.4. Sub-catchment slope
determine the infiltration rate because of the Horton technique, which is The slope represents the typical slope along the overland flow channel
predicated on producing runoff when rainfall intensity reaches the to the entrance locations for the sub-catchment. In the current investi-
infiltration capacity, explains how infiltration capacity decreases as the gation, the sub-catchment slope which was calculated as the elevation
precipitation event develops. difference divided by the flow path length (hydraulic length) on the map
was thought to be equivalent to the flow path slope.
3.6.3. Flow routing
Flow routing is the process of moving each hydrograph downstream 3.7.5. Sub-catchment Manning's roughness
within a system of pipes, and networks. The three options for flow The Manning's roughness for impermeable and previous zones for
routing components in SWMM5 are dynamic wave kinematic wave and overland flow is taken from literature (Dennis et al., 2019; Singh, 2018).
Steady-flow routing. The dynamic wave routing method was employed in As the beginning value for various surfaces, Manning's roughness is given
this study to determine runoff because dynamic wave routing solves all of a single value for impermeable and pervious areas of 0.013 and 0.05,
the one-dimensional Saint-Venant flow equations, due to its accuracy in respectively.
the results. These formulas include a volume continuity equation at nodes
as well as the continuity and momentum equations. 3.7.6. Depression storage
When the initial abstraction losses in the form of depression storage
3.7. Determination of sub-catchment properties are satisfied in the SWMM model experiences overland flow sub-
catchment (Mani et al., 2019). Infiltration and evaporation losses have
3.7.1. Soil reduced the amount of storage in these depressions. Because this study
Horton's model was employed in the SWMM hydrologic module's uses an event-based simulation, the evaporation phenomenon is not
infiltration model, which determines the quantity of rainwater infiltrated taken into account. According to Tuttolomondo et al. (2020), the initial
into a pervious surface area's unsaturated upper soil zone. The rate of depression storage estimates for the pervious and impervious surfaces are
infiltration is determined by the drainage area's soil qualities, surface calculated. For ease of analysis, depression storage was assumed to be
slopes, and land use land cover. The Horton infiltration model was constant throughout all sub-catchments. For impermeable and permeable
selected as the preferred infiltration method in SWMM why because the surfaces, the initial depression storage values are set at 3 and 6,
values can be freely measured in the field (Tsai et al., 2017). respectively.

3.7.7. Conduit properties

Table 2 Conduit cross-sections and Manning's roughness are among the
Some main parameters used for sensitivity analysis. conduit properties. To assign conduit dimensions, conduit cross sections
were measured at various locations and for various conduits, while the
No Parameters Descriptions
initial Manning's roughness values were recorded. According to Tutto-
1 N-Imperv Manning's roughness coefficient for impervious area
lomondo et al. (2020) suggested Manning's roughness is initially set to
2 N-Per Manning's roughness coefficient for the previous area
3 Dstore-Imperv Depth of surface storage in impervious area
0.011.
4 Dstore-perv Depth of surface storage in the previous area
5 Zero-Imperv Impervious area without surface storage (%)
6 % Imperv Percent of imperviousness 3.8. SWMM parameter and sensitivity analysis
7 W Width of sub-catchment
8 % Slope % Slope The relevance of the various elements was assessed using a sensitivity
9 N- for conduit Manning's roughness coefficient for conduit
analysis, to aid model calibration by identifying which parameters

7
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

Table 3 3.9. Model calibration and validation

Calibration parameters and assigned uncertainties in PCSWMM.
No Calibration Parameter Uncertainty Value (%) To calibrate and validate the SWMM model, observed runoff depth
data from the outfall-3 drainage system were chosen, and runoff depth
1 Sub-catchment width(w) 100
2 Impervious percentage (Imperv. %) 50 was measured for 2-days of rainfall occurrences. Additionally, 2 days of
3 Depression storage for impervious area 50 rainfall data that occurred concurrently with the days of the runoff flow
4 Sub-catchment of slop 50 depth recorded were obtained from the automatic weather station at the
5 Manning's roughness for impervious area 25 meteorology and hydrology faculty of Arba Minch Water Technology
6 Manning's roughness for conduits 25
7 Zero Impervious area 25
Institute of Arba Minch University.
Manual calibration is a labor-intensive process and takes a long time.
Automatic calibration methods and parameter estimation have been
implemented to overcome this difficulty. In the PCSWMM software, the
change as parameter value change. According to several scholars such as
sensitivity-based Radio Tuning Calibration (SRTC) tool or a knowledge-
(Ranaee et al., 2022; Ballinas et al., 2020) state that sensitivity analysis is
based calibration system was used for model calibration processes. The
employed to determine the impact of various model parameters on the
choice of initial parameter values and stated uncertainties marks the
corresponding outputs and to separate influential parameters from
beginning of the model calibration method in PCSWMM (Fandel et al.,
non-influential parameters. It is also frequently applied before calibra-
2021; Hossain et al., 2019).). However, model performance and data
tion to minimize the number of parameters taken into account during
input file uncertainties are taken into account while selecting the cali-
calibration. The PCSWMM, a personal computer stormwater manage-
bration parameters. After the initial parameter values were assigned to all
ment model, was used to perform the automatic sensitivity analysis.
of the model parameters, the uncertainty of each parameter was assigned
Table 2 shows the model parameter selected for sensitivity analysis for
based on the data source. The key sources of uncertainty in SWMM's
calibration.
modeling were the sub-catchment width, slope, Manning's roughness for

Fig. 6. Sensitive parameter analysis in PCSWMM.

8
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

Fig. 7. Simulated runoff depths versus observed runoff depths.

the impervious area (N-Imperv), Manning's roughness for the conduit, (operational and procedural practices) are used to manage (treat polluted
percentage impervious area (%Imperv), and depression storage for the stormwater) and they are grouped under Best Management Practice
impervious area. According to Xiulan et al. (2019) for SRTC, the un- (BMP) (Bulti and Abebe, 2020; Dibaba 2018). Stormwater BMPs are of
certainties for the calibrated parameters were assigned and are shown in many forms including: Plan, design and construct stormwater systems to
Table 3. screen contaminants before reaching drinking water sources, manage
The model was run using the best parameter output after the auto- stormwater to control erosion and flooding, acquire and protect natural
calibration runs and then simulated and observed runoff depths were waterways so that they can be rehabilitated. Furthermore, build ponds,
compared using performance metrics such as Nash-Sutcliffe efficiency swales, or wetlands to work with pipes, channels, and other drainage
(NSE) and coefficient of determination (R2). Finally, validation will be structures (often referred to as hard structures), source control, storm-
done to compare the model's outputs to another dataset without altering water regulations to address stormwater needs, public outreach and
the parameter values. forecasting approaches are some of the important stormwater manage-
ments controlling practices.
3.10. Model performance evaluation
4. Result and discussions
In this study, the coefficient of determination (R2) and Nash-Sutcliffe
efficiency (NSE) were used for evaluation of the model performance. The 4.1. Sensitivity analysis
coefficient of determination (R2) reveals the degree of variance in the
observed data. R2 ranges from 0 to 1 and measures how much of the In this study, the proportion of impervious surfaces, Manning coef-
observed dispersion can be explained by the prediction. The equation ficient for conduit and impervious surfaces, and width of each sub-
reads: catchment were all used for calibration and validation among them to
mention from eleven SWMM model parameters were used for sensitivity
2 P 3
n analysis. Fig. 6 shows the sensitive model parameters over the study area.
ðOi  Oav ÞðPi  Pav Þ
6 7 The results indicate that among the model parameters considered,
R2 ¼ 6
4sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
i¼1
ffi7

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 5 (12)
P n P n Manning's roughness coefficient, percent of impervious area and storage
ðOi  Oav Þ2 ðPi  Pav Þ2 depth of impervious area, sub-catchment width, slope, and zero imper-
i¼1 i¼1
vious area are found to be the most sensitive while, and the others are
least sensitive parameters throughout the model simulation.
Where, O is the observed value, P is the predicted value (simulated), Oav
is the average of the observed value, and Pav is average of the predicted
value. 4.2. Model calibration and validation
The Nash-Sutcliffe efficiency (NSE) measures the relative magnitude
of the residual variance of the simulated flow compared to the observed The calibration was done for sensitive runoff parameters of SWMM
flow (Eq. (13)). NSE has a range of values between 1 and -∞. While NSE with observed runoff depth data. The model validation was also per-
values are between 0.6 and 1, the simulated values are better predictors formed without further adjustment of the calibration parameters. The
than the mean measured value, and it is generally viewed as acceptable runoff depths measured at outfall-3 on the day of September 18, 2021,
performance. were used for model calibration processes by comparing with simulated
P
n runoff depths. Then, runoff depths measured at outfall-3 on the day of
ðOi  Pi Þ2 July 24, 2021, were used for model validation. The SWMM simulation
NSE ¼ 1  P
i¼1
n (13) with the default value of parameters showed relatively weak matching
ðOi  Oav Þ2 between the simulated and observed runoff depth data hydrographs.
i¼1
Hence, parameter adjustment is necessary to better simulate the runoff
depth with the aid of calibration.
3.11. Stormwater management The parameters of sub-catchment width, slope, Manning's roughness
for conduit, percentage impervious area (%Imperv), depression storage
Stormwater management refers to managing of quality and quantity for the impervious area, Manning's roughness for the impervious area,
of water. Several structural or engineered control devices and techniques and N-Imperv were the main sources of uncertainty in the rainfall-runoff

9
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

Fig. 8. Runoff depths of observed versus simulated values.

Fig. 9. L-moments ratio diagram for various distribution functions.

modeling of SWMM. The most distinct parameters that significantly

Table 4
affect how well SWMM modeling works are the sub-catchment width,
Best of fit testing by using easy fit software for annual maximum series.
percentage impervious area, Manning coefficient for impervious, and
Manning's roughness for the conduit. Therefore, in this investigation, No Distribution Kolmogorov Anderson Chi-Squared
Smirnov Darling
calibration procedures were used with these sensitive parameters.
Fig. 7 shows comparison of simulated runoff depths against the Statistic Rank Statistic Rank Statistic Rank
measured runoff depth values at outfall-3 using the rainfall event on 1 Gen. Extreme 0.123 2 0.588 2 1.060 3
September 18, 2021 for calibration. The observed hydrographs show an value
over-stimulated peak runoff depth, but the simulated hydrographs have a 2 Gen. Pareto 0.112 1 0.386 1 0.132 1
3 Gumbel Max 0.158 7 0.978 7 2.440 6
slightly similar shape. There is good agreement between the observed
4 Log-Pearson 3 0.137 5 0.686 4 1.048 2
and predicted runoff depth with Nash-Sutcliffe efficiency (NSE) of 0.887 5 Log-Logistic 0.125 3 0.719 5 1.945 5
and a coefficient of determination (R2) of 0.890. 6 Log-Normal 0.141 6 0.728 6 1.121 4
The validation of the model also showed good agreement between the 7 Normal 0.133 4 0.637 3 2.877 7
observed and simulated runoff depth with the coefficient of determina-
tion (R2) of 0.88, and a Nash-Sutcliffe efficiency (NSE) of 0.84. The result
event-based time series data were used to calibrate and validate the
of validation for runoff depth is shown in (Fig. 8). Overall, the results
SWMM model. The results of this investigation are broadly consistent
indicate the SWMM prediction is good considering a single-event
with the coefficient of determination (R2) and Nash-Sutcliffe efficiency
hydrograph.
(NSE) values. The total of the peak event, the low flow, and volume is
The performance value of the SWMM model for runoff depth is
known as the Nash-Sutcliffe efficiency (NSE).
revealed a coefficient of determination (R2) of 0.88, and Nash-Sutcliffe
efficiency (NSE) of 0.84, which is above 0.6 for calibration, and coeffi-
cient of determination (R2) of 0.88, and Nash-Sutcliffe efficiency (NSE) of 4.3. Rainfall frequency analysis
0.887, which is again rated above 0.6 for validation. In this study, the
Fig. 9 shows the L-moment ratio diagram for various frequency

10
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

Table 5 was measured at several drainage system exits for various return dura-
Rainfall intensity related to each return period. tions. According to the findings, extending the return period enhanced
Duration(hr) Return period(year) the design rainfall intensity. As a result, there was a noticeable rise in the
magnitude of peak runoff. The average surface runoff from sub-
2 5 10 25 30
catchments was computed in this study for return periods of 2, 5, 10,
0.08 119.09 249.22 414.10 783.78 887.36 25, and 30 years. With an increase in the return period, average runoff
0.17 75.02 157.00 260.87 493.75 559.00
0.25 57.25 119.81 199.08 376.80 426.60
increased by 0.0651 m3/s, 0.1366 m3/s, 0.2545 m3/s, 0.6459 m3/s, and
0.50 36.07 75.48 125.41 237.37 268.74 0.7455 m3/s, respectively. There are three outfalls (OF) in this study:
1.00 22.72 47.55 79.00 149.53 169.29 OF1, OF2, and OF3. However, the outfall runoff flow was predicted for
2.00 14.31 29.95 49.77 94.20 106.65 various return times (2, 5, 10, 25, and 30 years). The overall outlet runoff
12.00 4.33 9.07 15.07 28.53 32.30
discharge increased as the return period increased. Consequently, an
24.00 2.73 5.71 9.50 17.97 20.35
increase in peak runoff was observed because of a large number of linked
conduits with a high amount of discharge flow and from more sub-
catchment high amounts of runoff discharge to OF3. The runoff
distribution functions which is used to identify the suitable rainfall fre-
discharge at the three outfalls is shown in Table 6.
quency distribution (Lie et al., 2018). The results show that from all of
In general, the peak runoff magnitude for return durations of 2, 5, 10,
the common distribution functions, the generated data sample charac-
25, and 30 years was 0.22, 0.46, 0.81, 1.9, and 2.18 m3/s, respectively, as
terized by the values of L-skewness, i.e., 0.1399, and L-Kurtosis, i.e.,
shown in (Fig. 11) and Table 7. These peak runoff magnitudes are crucial
0.0018, respectively, was determined to be most closely matched by the
for engineering design, such as the Arba Minch town urban drainage
generalized Pareto (GP) distribution. Additionally, it is discovered that
system buildings. The probability of the once-every-25-year runoff event
the GP distribution has the lowest absolute statistics of 0.035.
in the Arba Minch town was 20% and this result is comparable with the
Table 4 shows the results of best fit using first ranked through
findings of Shafiquzzaman et al. (2020).
calculation of statistical parameters using three (3) test methods in easy-
The peak runoff magnitude of 0.81 and 1.9 m3/s would occur within
fit software for annual maximum series. The rainfall intensity will be
the 10-to 25-year return period and had a probability of 60 and 80%,
estimated using various selected distribution functions. The L-moment
respectively which is a comparable result as (Deng 2020; Trajkovi et al.,
ratio diagram and easy fit software analysis identified the generalized
2020). The highest runoff magnitude was revealed for the 30-year return
Pareto (GP) distribution function from goodness of fit test to forecast the
period which shows the hydrograph of runoff magnitude at different
extreme maximum rainfall value using a frequency distribution function.
return periods are significant in their runoff risk management. We note
Table 5 presents the rainfall intensity for different return periods.
that, runoff causes more loss of life and distraction.
Then, using the data in the table, the IDF curve has been produced as
shown in (Fig. 10). Generalized Pareto distribution (GPD) were used a
frequency distribution function to predict the extreme maximum rainfall
value estimated using Eq. (2) (section 3.2 above) for 2, 5, 10, 25, and 30-
Table 6
year return periods. The location (ε), scale (α), and shape (k) parameters
Runoff discharge at outfall with different return periods.
of the generalized Pareto (GP) function distribution are found to be
Node Max. runoff (m3/s)
33.635, 36.493, and -0.64874 using Easy-Fit 5.5 professional software.
Using Table 5 data develop the IDF curve in (Fig. 10) for each return Return period (year)
period to capture important characteristics of point rainfall for any 2-yr 5-yr 10-yr 25-yr 30-yr
desired shorter durations using Arba Minch climate station.
OF1 0.687 1.351 1.552 2.049 2.177
OF2 0.669 1.189 1.379 1.997 2.155
4.4. Peak surface runoff for various return periods OF3 0.951 1.618 2.498 2.479 2.467
SUM 2.307 4.158 5.429 6.525 6.799

To assess current performance, the model was run with continuous

rainfall events with varying return periods. Peak surface flow discharge

Fig. 10. IDF curve of the study area.

11
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

Fig. 11. Runoff magnitude and return period in the study area.

SWMM to describe the runoff from each sub-catchment. The drainage

Table 7 network surcharge conduit for a 2-year return period is shown in
Relationships between the probability of occurrence and its return period. (Fig. 12).
T Q(m3/sec) Desc. order Rank(M) P ¼ M/N P (%) T ¼ 1/P We note that, the flooding at a node rises to water which overflows at
2 0.22 2.18 1 0.2 20 5 the node and produces surface flooding. In this study, the flooding
5 0.46 1.9 2 0.4 40 2.5 junctions are identified and presented in Table 8. The simulation results
10 0.81 0.81 3 0.6 60 1.67
25 1.9 0.46 4 0.8 80 1.25
30 2.18 0.22 5 1 100 1
N 5 Table 8
Summary of flooding at junctions for 2-year return period design storm.
Node Hours Maximum Rate Hour of Maximum Total Flood
4.5. Capacity analysis of existing drainage system Flooded (m3/s) Flooding Volume
10^6 ltr
The SWMM model's drainage network is developed for the study area. J7 2.14 0.183 2:00 0.828
The whole catchment area is divided into 47 sub-catchments by J8 0.79 0.034 2:00 0.093
considering the existing drainage network, slope, and building blocks. J24 0.91 0.208 2:00 0.608
Each sub-catchment in the area is calculated and provided as an input in J25 5.02 0.386 1:13 3.412

Fig. 12. Drainage network surcharge conduit for 2-year return period.

12
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

Fig. 13. Drainage network surcharge conduit for 5-year return period.

or a proper structural design for runoff transfer, as well as other justifi-

Table 9
cations and a study by (Shafiquzzaman et al., 2020; Song et al., 2020;
Summary of flooding at junctions for 5-year return period design.
D'Aniello et al., 2019) also justifies design of stormwater by considering
Node Hours Maximum Rate Hour of Maximum Total Flood the runoff at different junctions. The simulation results indicate that C13,
Flooded (m3/s) Flooding Volume 10^6 ltr
C14, C17, C18, C26, and C27 are read colors, indicating that they flooded
J4 0.79 0.089 1:59 0.239 due to insufficient design water depth shown in (Fig. 13).
J7 8.9 0.268 2:00 3.646
J8 2.98 0.378 2:00 1.945
J17 0.89 0.028 1:51 0.087
J21 0.84 0.107 2:00 0.306 4.6. Water depth and flow in the links
J23 0.93 0.231 2:00 0.735
J24 4.91 0.735 1:11 4.925 A flood can occur when the water surface at a node or connection
J25 16.98 0.387 1:07 11.447 exceeds the channel's maximum defined depth. One of the criteria that
J38 0.89 0.118 2:00 0.356
J39 0.86 0.093 2:00 0.27
should be considered when determining the capability of runoff
conveyance through conduits in an urban drainage system is capacity
(the ratio of depth to full depth). The channel slope and depth are
of 2-year return period design rainfall show that maximum flooding oc- responsible for the majority of the surcharged links and flooded nodes.
curs at J7, J8, J24, and J25. These four nodes are produced due to Maximum flooding has occurred at some conduits, C14, C18, and C27,
flooding. Of node J25 had the highest flood occurrence (0.386 m3/s), according to the simulation results of a 2-year return period design storm.
with flooding time up to 1:13. The four junctions were often predicted to Fig. 14 shows the water elevation profile plot for water elevation from
flood because high runoff from sub-catchments No.38, No.46, No.47, and Node J24 to Node J25 shows that the conduit's elevation increased from
No.40 was conveyed to these nodes directly (Fig. 11). 1228 m to 1228.8 m as the distance from sub-catchments to the outlet
Simulation results of 5-year return period design rainfall show that increased because the elevation of the sub-catchments is higher than the
maximum flooding occurs at J4, J7, J8, J17, J21, J23, J24, J25, J38, and conduit's mean water level. The flow in the channels is affected by the
J39. These ten nodes are produced by flooding. Those of node J24 had conduit slope, size, and roughness. Drescher et al. (2021) also justify this
the highest flood occurrence (0.735 m3/s) with flooding time up to 1:11. situation.
The flooded junctions were often predicted to flood because of high Fig. 15 depicts the flow of information within the link and nodes.
runoff from sub-catchments No. 8, No.9, No.39, and No.42, No.46, and Because of insufficient design water depth, both intersections 24 and 25
No.47 were conveyed to these nodes directly (Fig. 13 and Table 9). are flooded, according to the simulation results. As a result, the conduit
Fig. 13 depicts the hydraulic capacity of existing drainage systems, which (C27) is also congested at this time, and runoff has overflowed. Maximum
indicate several of the conduits were undersized. This demonstrates that flooding occurs at some conduits C2, C3, C4, C5, C10, C12, C13, C14,
the conduits are insufficient to transmit the stormwater generated by the C15, C16, C17, C18, C19, C21, C23, C24, C25, C26, C27, C34, and C36
contributing watersheds, resulting in flash floods and infrastructure that have attained maximum depth (full depth), according to simulation
damage mainly C14, C18, and C27 are inundated due to insufficient results of 5-year return period design rainfall.
design water depth. Overall, this study suggests that the dimensions and Figs. 15 and 16 shows the flow within the links and nodes at different
characteristics of the drainage systems in the studied area may lack junction's points. The simulation result indicates that junctions 17 and
sufficient hydrological data analysis, appropriate design implementation, 25, junctions 38 and 39, and junctions 23, 24, and 25 are flooded due to

13
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

Fig. 14. Water elevation profile: junction 24 to 25 for a 2-year return period of rainfall.

Fig. 15. Water elevation profile: junctions 17 to 25 for a 5-year return period of rainfall.

insufficient design water depth. Consequently, the conduits (C13, C14, 4.7. Spatial variation on runoff production due to urbanization
C18, C17, C26, and C27) are also busy, and the overflow of runoff has
happened. And from the descriptive statistics for the design storm sce- From the precipitation data entered, the runoff from the sub-
narios, the return period had a significant impact on node flooding. The catchment was computed, which varied from each other due to varying
respective number of node overflows for 2, 5, 10, 25, and 30 years are land features (urbanization) and other properties. As most of the land
found to be 4, 10, 17, 28, and 30, respectively. With a longer return in- area is covered with buildings and paved surfaces, the runoff coefficient
terval, the number of flooded nodes grew significantly. value is almost high. Natural drainage is in decreasing order due to the

14
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

Fig. 16. Water elevation profile: junctions 38 to39 for a 5-year return period of rainfall.

increase in concrete construction work. The impervious surface area is in To analyze the flows and simulate flooding, design rainfall intensity
increasing order. The higher the value of imperviousness of the surface, was generated for a 24-h storm with return periods of 2, 5, 10, 25, and 30
the runoff will be increased. The results of this study in agreement with years. The results indicate a significant risk of flooding, revealing that the
other similar studies by Dudley et al. (2020). We note that, this research current drainage system is inadequate for both current and anticipated
followed a comprehensive approach for detail design of drainage system future rainfall. SC8, SC9, and SC10 showed the highest surface runoff for
which will have a great potential for researchers, water and sewerage all return periods. Junction nodes such as J7, J8, J24, and J25 experi-
office, flood forecasting teams and road authorities as an input or a base enced the most flooding in the simulated for a 2-year return period design
for future studies related to stormwater water management including rainfall with a node J25 had the maximum flood incidence of 0.386 m3/s
spatio-temporal LULC and climate change. lasting up to 1:13, mainly due to excessive runoff from specific sub-
Simulated results for 2-year, 5-year, 10-year, 25-year, and 30-year catchments. Conduits C14, C18, and C27 reached full depth flow, and
return periods showed peak runoff was greatest in sub-catchments several nodes and conduits produced surface flooding for design rainfall
No.8, No.1, No.6, No.30, and No.9 because of the slope of the sub- periods of two years or more. The analysis revealed that conduit 20,
catchment and their large area with high percent imperviousness that connected to outfall 3, experienced the highest peak flow across all return
results in less infiltration into the soil. The estimation of peak runoff for periods. Despite having three outlets (outfalls) in the study area, OF1,
the sub-catchments is the most runoff producers with the main source of OF2, and OF3, runoff flow was anticipated for all return periods of years.
the runoff that becomes the drainage network flooded due to large areas The peak runoff and total outlet discharge increased with a longer
of high percent imperviousness. return period, with average surface runoff escalating from 0.0651 m3/s
to 0.7455 m3/s over return periods of 2, 5, 10, 25, and 30 years. The
5. Conclusion study concluded that the current storm drain system cannot effectively
handle rainfall beyond five years, highlighting the system's incapacity to
The present study modeled the one-dimensional drainage network for convey floodwater during extended return times. Inadequate drainage
the Arba Minch town area using the SWMM model for simulating the maintenance and unplanned development were identified as key con-
urban rainfall-runoff. The primary challenge faced by most developing tributors to the drainage system's insufficient capacity in the studied area.
countries in terms of infrastructure, such as roads, buildings, and land The situation is likely to worsen in the future due to climate change,
use, is the ineffective nature of urban drainage systems. This study which may lead to more frequent and intense precipitation. Hence, future
focused on evaluating the storm drainage system, conduit surcharged studies should include the impact of climate change and human activities
flows, and floods in the existing drainage system using the Stormwater in the area that causes blockage of conduits, junctions, and conduits'
Management Model (SWMM), calibrated through the Personal Computer overflow and surcharge, as well as under-designed conduits considering
Stormwater Management Model (PCSWMM). The field survey was also spatio-temporal variation of land use land cover.
conducted to identify the status of the drainage that was properly func-
tioning (partially blocked and fully blocked), the shape of drainage sys- CRediT authorship contribution statement
tems, and runoff flow depth data measured in drainage systems at outfall
(3) which are used for model calibration rainfall, and validation was Bereket Ayele Ayda: Writing – original draft, Validation, Software,
done. Infiltration rate data is measured with a double ring infiltrometer Methodology, Investigation, Data curation, Conceptualization. Deme-
and measuring dimension with a tap meter, invert elevation and surface lash Wondimagegnehu Goshime: Writing – review & editing, Super-
elevation at each sub-catchment with hand GPS for input parameters in vision, Resources, Methodology, Conceptualization. Mekuanent
the SWMM model, and take a random soil sample to identify soil texture Muluneh Finsa: Writing – review & editing, Supervision, Methodology,
in the study area. Investigation, Conceptualization. Abebe Temesgen Ayalew: Writing –

15
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

original draft, Visualization, Supervision. Dibaba, W.T., 2018. A review of sustainability of urban drainage system: traits and
consequences. Journal of Sedimentary Environments 3 (3), 131–137. https://
doi.org/10.12957/jse.2018.37825.
Declaration of competing interest Drescher, M., Sinasac, S., 2021. Correction to: social-psychological determinants of the
implementation of green infrastructure for residential stormwater management.
The authors declare that they have no known competing financial Environ. Manag. 67, 323. https://doi.org/10.1007/s00267-021-01443-4.
Dudley, R.W., Hirsch, R.M., Archfield, S.A., Blum, A.G., Renard, B., 2020. Low streamflow
interests or personal relationships that could have appeared to influence trends at human-impacted and reference basins in the United States. J. Hydrol. 580,
the work reported in this paper. 124254. https://doi.org/10.1016/j.jhydrol.2019.124254.
Ethiopian Road Authority (ERA), 2013. Pavement Design Manual Volume II Rigid
Pavements, pp. 1–68. https://research4cap.org/ral/ERA-Ethiopia-2013-RigidþPa
Acknowledgements vementþDesignþManual-ERA-v130322.pdf.
Fandel, C., Ferre, T., Chen, Z., 2021. A model ensemble generator to explore structural
The authors gratefully acknowledge the Ethiopian meteorological uncertainty in karst systems with unmapped conduits. Hydrogeol. J. 29, 229–248.
https://doi.org/10.1007/s10040-020-02227-6.
institute (EMI) for providing climate data that are used in this study. We Feng, B., Zhang, Y., Bourke, R., 2021. Urbanization impacts on flood risks based on urban
also appreciate the data providers of Climate Hazards Group (CHG) and growth data and coupled flood models. Nat. Hazards 106, 613–627. https://doi.org/
other institutes for making the satellite rainfall data available. 10.1007/s11069-020-04480-0.
Funk, C., Verdin, A., Michaelsen, J., Peterson, P., Pedreros, D., Husak, G., 2015. A global
satellite-assisted precipitation climatology. Earth Syst. Sci. Data 7, 275–287.
References Giełczewski, M., Piniewski, M., Doma nski, P.D., 2022. Mixed statistical and data mining
analysis of river flow and catchment properties at regional scale. Stoch. Environ. Res.
Ahmed, W., Hamilton, K., Toze, S., Cook, S., Page, D., 2019. A review on microbial Risk Assess. 36, 2861–2882. https://doi.org/10.1007/s00477-022-02169-3.
contaminants in stormwater runoff and outfalls: potential health risks and mitigation Giudicianni, C., Mohammed, N., Assaf, S.T., ECr, 2023. Comparison of nonlinear reservoir
strategies. Sci. Total Environ. 692, 1304–1321. https://doi.org/10.1016/ and UH algorithms for the hydrological modeling of a real urban catchment with
j.scitotenv.2019.07.055. EPASWMM. Hydrology 10 (1), 24. https://doi.org/10.3390/hydrology10010024.
Ali, E., John, S.G., Bruce, N.W., 2018. Estimating effective impervious area in urban Gong, P., Li, X.C., Wang, J., Bai, Y.Q., Chen, B., Hu, T.Y., Liu, X.P., Xu, B., Yang, J.,
watersheds using land cover, soil character and asymptotic curve number. Hydrol. Zhang, W., Zhou, Y.Y., 2020. Annual maps of global artificial impervious areas
Sci. J. 63 (4), 513–526. https://doi.org/10.1080/02626667.2018.1440562. (GAIA) between 1985 and 2018. Remote Sens. Environ. 236, 111510. https://
Alshammari, E., Rahman, A., Rainis, R., Seri, N., Fuzi, N., 2023. The impacts of land use doi.org/10.1016/j.rse.2019.111510.
changes in urban hydrology, runoff and flooding: a review. Curr. Urban Stud. 11, Hossain, S., Hewa, G.A., Wella-Hewage, S., 2019. A comparison of continuous and event-
120–141. https://doi.org/10.4236/cus.2023.111007. based rainfall-runoff (RR) modelling using EPA-SWMM. Water 11 (3), 611. https://
Anim, D.O., Fletcher, T.D., Pasternack, G.B., Vietz, G.J., Duncan, H.P., Burns, M.J., 2019. doi.org/10.3390/w11030611.
Can catchment-scale urban stormwater management measures benefit the stream Hou, J.M., Hao, H., Qi, W.C., Guo, K.H., Li, Z.B., Reinhard, H., 2019. Experimental
hydraulic environment? J. Environ. Manag. 233, 1–11. https://doi.org/10.1016/ investigation for impacts of rain storms and terrain slopes on low impact
j.jenvman.2018.12.023. development effect in an idealized urban catchment. J. Hydrol. https://doi.org/
Arnone, E., Pumo, D., Francipane, A., La Loggia, G., Noto, L.V., 2018. The role of urban 10.1016/j.jhydrol.2019.12176.
growth, climate change, and their interplay in altering runoff extremes. Hydrol. Ibrahim, O.A., Goshime, D.W., Gebrekristos, S.T., Absi, R., 2023. Panacea to the
Process. 32 (12), 1755–1770. https://doi.org/10.1002/hyp.13141. catastrophe through mapping in anticipation of mitigating the flood magnitude along
Avalon, C.C., Al, S.R., 2023. Assessing rainfall variability in Jamaica using CHIRPS: wabi shebele river basin of Somalia. Natural research hazards. https://doi.org/
techniques and measures for persistence, long and short-term trends. Geographie 3, 10.2139/ssrn.4538648 (in press), Journal Pre-proof.
375–397. https://doi.org/10.3390/geographies3020020. Ishimatsu, K., Ito, K., Mitani, Y., Tanaka, Y., Sugahara, T., Naka, Y., 2017. Use of rain
Ballinas, G., Hector, A., Víctor, H., Alcocer, Y., Javier, J., Canto, R., Roel, S., 2020. gardens for stormwater management in urban design and planning. Landsc. Ecol.
Sensitivity analysis of the rainfall–runoff modeling parameters in data-scarce urban. Eng. 13, 205–212. https://doi.org/10.1007/s11355-016-0309-3.
Catchment Hydrology 7 (4), 73. https://doi.org/10.3390/hydrology7040073. Jemberie, M.A., Melesse, A.M., Abate, B., 2023. Urban drainage:the challenges and failure
Baselt, I., Andreas, M., 2022. Determining the flow resistance of racks and the resulting assessment using AHP, Addis Ababa, Ethiopia. Water 15 (5), 957. https://doi.org/
flow dynamics in the channel by using the saint-venant equations. Water 14 (16), 10.3390/w15050957.
2469. https://doi.org/10.3390/w14162469. Lei, Y., Lars, S.H., Pengqi, D., Dingbao, W., Richard, M.V., 2018. The probability
Bekele, Z.T., 2021. Hydraulic performance assessment of storm water drainage systems of distribution of daily precipitation at the point and catchment scales in the United
dejen town using storm water management model (SWMM), Ethiopia. Global States. Hydrol. Earth Syst. Sci. 22, 6519–6531. https://doi.org/10.5194/hess-22-
scientific journals 9, 1358–1383. 6519-2018.
Belachew, M., 2019. Improving urban runoff drainage systems. In: Shire Indaselassie Mani, M., Bozorg, O., Loaiciga, H.A., 2019. A new framework for the optimal
Town. Addis Ababa. Addis Ababa University. management of urban runoff with low-impact development stormwater control
Belay, A.S., Fenta, A.A., Yenehun, A., Nigate, F., Tilahun, S.A., Moges, M.M., Dessie, M., measures considering service-performance reduction. J. Hydroinf. 21 (5), 727–744.
Adgo, E., Nyssen, J., Chen, M., Griensven, A.V., Walraevens, K., 2019. Evaluation and https://doi.org/10.2166/hydro.2019.126.
application of multi-source satellite rainfall product CHIRPS to assess spatio-temporal Mekonnen, D.F., Duan, Z., Rientjes, T., Disse, M., 2018. Analysis of combined and isolated
rainfall variability on data-sparse western margins of Ethiopian highlands. Rem. Sens. effects of land-use and land-cover changes and climate change on the upper Blue Nile
11 (22), 2688. https://doi.org/10.3390/rs11222688. River basin's stream flow. Hydrol. Earth Syst. Sci. 22, 6187–6207. https://doi.org/
Bibi, T.S., Tekesa, N.W., 2023. Impacts of climate change on IDF curves for urban 10.5194/hess-22-6187-2018.
stormwater management systems design: the case of Dodola Town, Ethiopia. Environ. Naeimi, G., Safavi, H.R., 2019. Integrated stormwater and groundwater management in
Monit. Assess. 195, 170. https://doi.org/10.1007/s10661-022-10781-7. urban areas, a case study. Int. J. Civ. Eng. 17, 1281–1294. https://doi.org/10.100
Blum, A.G., Ferraro, P.J., Archfield, S.A., Ryberg, K.R., 2020. Causal effect of impervious 7/s40999-018-0386-9.
cover on annual flood magnitude for the United States. Geophys. Res. Lett. 47 (5), Nakhaei, M., Mohebbi, T.A., Saadi, T., 2023. An evaluation of satellite precipitation
e2019GL086480. https://doi.org/10.1029/2019GL086480. downscaling models using machine learning algorithms in Hashtgerd Plain, Iran.
Bulti, D.T., Abebe, B.G., 2020. Analyzing the impacts of urbanization on runoff Model. Earth Syst. Environ. 9, 2829–2843. https://doi.org/10.1007/s40808-022-
characteristics in Adama city, Ethiopia. SN Appl. Sci. 2, 1151. https://doi.org/ 01678-y.
10.1007/s42452-020-2961-3. Ogden, F.L., Pradhan, N.R., Downer, C.W., Zahner, J.A., 2011. Relative importance of
Charalambous, K., Bruggeman, A., Eliades, M., Camera, C., Vassiliou, L., 2019. impervious area, drainage density, width function, and subsurface storm drainage on
Stormwater retention and reuse at the residential plot level green roof experiment flood runoff from an urbanized catchment. Water Resour. Res. 47 (12), W12503.
and water balance computations for long-term use in Cyprus. Water 11, 1055. https://doi.org/10.1029/2011WR010550.
https://doi.org/10.3390/w11051055. Ranaee, E., Inzoli, F., Riva, 2022. Sensitivity-based parameter calibration of single- and
Davies, A.S., Hernebring, C., Svensson, G., Gustafsson, L.-G., 2008. The impacts of climate dual-continuum coreflooding simulation models. Transport Porous Media 145,
change and urbanisation on drainage in helsingborg, Sweden: suburban stormwater. 549–587. https://doi.org/10.1007/s11242-022-01854-9.
J. Hydrol. 350 (1), 114–125. https://doi.org/10.1016/j.jhydrol.2007.11.006. Roberts, K., Dowell, A., Nie, J.B., 2019. Attempting rigour and replicability in thematic
D'Aniello, A., Cimorelli, L., Cozzolino, L., 2019. Correction to: the influence of soil analysis of qualitative research data; a case study of codebook development. BMC
stochastic heterogeneity and facility dimensions on stormwater infiltration facilities Med. Res. Methodol. 19, 66. https://doi.org/10.1186/s12874-019-0707-y, 2019.
performance. Water Resour. Manag. 33, 4667–4668. https://doi.org/10.1007/ Rufino, I.A.A., Alves, P., Ester, G.E., Santos, K., 2018. Dynamic scenarios and water
s11269-019-02394-6. management simulations: towards to an integrated spatial analysis in water urban
Deng, Y., 2020. Low-cost adsorbents for urban stormwater pollution control. Front. planning. In: La Loggia, G., Freni, G., Puleo, V., De Marchis, M. (Eds.), HIC 2018, 13th
Environ. Sci. Eng. 14 (5), 83. https://doi.org/10.1007/s11783-020-1262-9. International Conference on Hydro Informatics, 3, pp. 1796–1803. https://doi.org/
Dennis, M.F., Zahida, Y., Cyriel, A., Olivier, D., 2019. Relating imperviousness to building 10.29007/7zm8.
growth and developed area in order to model the impact of peri-urbanization on Shafiquzzaman, M., Haider, H., Ghazaw, Y.M., Alharbi, F.S., AlSaleem, S.,
runoff in a Mediterranean catchment (1964-2014). J. Land Use Sci. 14 (3), 210–224. Almoshaogeh, M., 2020. Evaluation of a low-cost ceramic filter for sustainable reuse
https://doi.org/10.1080/1747423X.2019.1681528. of urban stormwater in arid environments. Water 12, 460. https://doi.org/10.3390/
w12020460.

16
B.A. Ayda et al. Natural Hazards Research xxx (xxxx) xxx

Shao, Yuehong, Zhao, Jun, Xu, Jinchao, Fu, Aolin, Wu, Junmei, 2021. Revision of central Ethiopia. Environ. Monit. Assess. 193 (12), 1–14. https://doi.org/10.1007/
frequency estimates of extreme precipitation based on the annual maximum series in s10661-021-09574-1.
the jiangsu province in China. Water 13 (13), 1832. https://doi.org/10.3390/ Trajkovi, S., Milicevic, D., Milanovic, M., Gocic, M., 2020. Comparative study of different
w13131832. LID technologies for drainage and protection of atmospheric stormwater quality in
Shen, Y.W., Morsy, M.M., Huxley, C., Tahvildari, N., Goodall, J.L., 2019. Flood risk urban areas. Arabian J. Geosci. 13, 1101. https://doi.org/10.1007/s12517-020-
assessment and increased resilience for coastal urban watersheds under the combined 06093-0.
impact of storm tide and heavy rainfall. J. Hydrol. 579, 124159. https://doi.org/ Tuttolomondo, T., Virga, G., Licata, M., Leto, C., La-Bella, S., 2020. Constructed wetlands
10.1016/j.jhydrol.2019124159. as sustainable technology for the treatment and reuse of the first-flush stormwater in
Shrestha, S., Cui, S., Xu, L., Wang, L., Manandhar, B., Ding, S., 2021. Impact of land use agriculture – a case study in Sicily (Italy). Water 12, 2542. https://doi.org/10.3390/
change due to urbanisation on surface runoff using GIS-based SCS–CN method: a case w12092542.
study of xiamen city, China. Land 10 (8), 839. https://doi.org/10.3390/ Wang, Q., Xu, Y.P., Wang, Y.F., Zhang, Y.Q., Xiang, J., Xu, Y., Wang, J., 2020. Individual
land10080839. and combined impacts of future land-use and climate conditions on extreme
Singh, V.P., 2018. Hydrologic modeling: progress and future directions. Geosci. Lett. 5, hydrological events in a representative basin of the Yangtze River delta, China.
15. https://doi.org/10.1186/s40562-018-0113-z. J. Hydrol. 236 (15), 104805. https://doi.org/10.1016/j.atmosres.2019.104805.
Sun, Z., Li, X., Fu, W., Li, Y., Tang, D., 2013. Long-term effects of land use/land cover Wang, J., Hu, C., Ma, B., Mu, X., 2020. Rapid Urbanization Impact on the Hydrological
change on surface runoff in urban areas of Beijing, China. J. Appl. Remote Sens. 8 (1). Processes in Zhengzhou, pp. 1–21. China. Water.
https://doi.org/10.1117/1.JRS.8.084596. Willems, P., 2013. Revision of urban drainage design rules after assessment of climate
Song, Y.H., Lee, J.H., Lee, E.H., 2020. Developing a reliability index of low impact change impacts on precipitation extremes at Uccle, Belgium. J. Hydrol. 496,
development for urban areas. Water 12, 2961. https://doi.org/10.3390/w12112961. 166–177.
Tsai, LinY., Chi, F.C., Chi, H.F., Jen, Y.L., 2017. Using the HSPF and SWMM models in a Xiulan, W., Shun, H., Gui, Q., Dongxia, W., Aiguo, S., Chuan, S.K., Zhongyan, L., 2019.
high pervious watershed and estimating their parameter sensitivity. Water 9 (10), Uncertainty estimation of robot geometric parameters and end-effecter position
780. https://doi.org/10.3390/w9100780. based on new generation GPS. Math. Probl Eng. 2019, 7830489. https://doi.org/
Takele, S.B., 2022. Journal of hydrology: regional studies. Modeling urban stormwater 10.1155/2019/7830489, 11.
management in the town of Dodola based on landuse and climate change using Yohannes, A.W., Cotter, M., Kelboro, G., Dessalegn, W., 2018. Land use and land cover
SWMM 5.1 44, 101267. https://doi.org/10.1016/j.ejrh.2022.101267. changes and their effects on the landscape of Abaya-Chamo Basin, Southern Ethiopia.
Tefera, D, 2020. Assessment and Modeling of Urban Drainage System Performance (A Land 7 (2), 1–17. https://doi.org/10.3390/land7010002.
Case Study of Mojo Town). MSc thesis. Addis Ababa University, Addis Ababa, Zhang, W.T., Zhang, X.N., Liu, Y.Z., Tang, W.W., Xu, J., Fu, Z.M., 2020. Assessment of
Ethiopia. flood inundation by coupled 1D/2D hydrodynamic modeling: a case study in
Tesema, D., Birhanu, B., Abebe, G., 2021. Climate change – induced variations in future mountainous watersheds along the coast of southeast China. Water 12 (3), 822.
extreme precipitation intensity duration frequency in flood - prone city of Adama, https://doi.org/10.3390/w12030822.

17