Flood Modeling of Jemo River Catchement in Addis Ababa City, Ethiopia

Flood affects lives and livelihoods in parts of Ethiopia. The flood from the Jemo river is causing damages to river side houses, infrastructures and displacement of affected population that resulted overflow on the surface following heavy rains and inundated lowland areas in the Nifas Silk Lafto district of Addis Ababa city, Ethiopia. This research involves the integration of Hydrologic Engineering Center-Hydrologic Modeling System (HEC-HMS) and Hydrologic Engineering Center-River Analysis System (HEC-RAS) to develop a twodimensional (2D) river model for flood inundation determination and mapping. The model Nash-Sutcliffe efficiency (ENS) was found to be 0.751 during calibration and 0.79 during validation and the coefficient of determination (R) was found to be 0.786 during calibration and 0.801 during validation. The HEC -RAS model was calibrated by comparing the results of the water level in each selected cross section obtained by the model with the observed historic flood mark levels of the year 2010 and 2013. The peak estimated time series discharge of HEC HMS model result was used to simulate the unsteady state of flow to determine flood extent, water depth and velocity of the study river. Flood hazard vulnerable areas both left and right side of the Jemo river delineated for the return periods 2, 5, 10, 25, 50 and 100 years. The research showed community houses ranging from mud houses to regular story buildings can be affected during each return period maximum flooding event. Jemo river flow capacity also checked at different cross sections and less carrying capacity sections identified for recommended flood protection measures like dyke and retaining masonry walls construction to tackle the flooding impact and avoid possible erosion of the river banks.


1.INTRODUCTION
Flooding is the most common natural hazard that can happen any time in wide variety of locations due to high intensity rainfall events. Floods can be explained as excess flows exceeding the transporting capacity of River channel, lakes, ponds, reservoirs, damage system, dam and any other water bodies, whereby water inundates outside water body areas [1]. Flood affects lives and livelihoods in parts of Ethiopia. Especially during the rainy season (June-September), the major perennial rivers as well as their numerous tributaries forming the country's drainage systems carry their peak discharges. Floods are already having very large impacts on cities and smaller urban centers in many African nations for instance the floods in Mozambique in 2000 which included heavy floods in Maputo, the floods in Algiers in 2001(with around 900 people killed, and 45,000 affected); heavy rains in East Africa in 2002 that brought floods and mudslides forcing tens of thousands to leave their homes in Rwanda and the very serious floods in Port Harcourt and in Addis Ababa in 2006 [2]. Flash flood from the Jemo river is causing damages to houses, infrastructures and threat to loss of life for human beings every year in Nifas silk lafto sub city of Addis Ababa city. The need for delineating flood risk areas and determining its magnitude is quite important in order to take required protection measures. HEC-HMS and HEC-RAS Computer modeling software is used for determination of river characteristics as well as delineation of flood affected areas to indicate the extent of flooding arising from the river considering different return periods.

Description of the study area
The Jemo river catchment is located between 8 0 57.192' N Latitude and 38 0 40.923' and 38 0 43.728' Longitude, the river catchment delineated and covers an area of 12.67km 2 . The main river course is located centrally in the catchment generally creating a V-shaped terrain profile.  The Control Specifications section contains  information pertinent to the timing of the model such as  when a storm occurred and what type of time interval is to be used in the model, etc. Finally, the input data component stores parameters and boundary conditions for basin and meteorological models [3]. Watershed delineation done at selected river outlet and required physical parameters for the HEC HMS model collected.

HEC -RAS Model Development
The three steps to run the Hydrologic engineering center river analysis system (HEC RAS) model are pre-processing of data, modeling phase, and post processing of data [4].The first step in developing hydraulic model HEC -RAS is to establish which directory the researcher wishes to work in and enter a title for the new project. Then, the river cross section at each location will be opened and geometric data called by Geographical information system (GIS) format and edited. And all RAS layers like stream center line, bank lines; bank stations, flow path center lines, cross-section cut lines, and others were generated from DEM of the study area.
The discharge values for different return periods can be entered manually for unsteady flow. The roughness coefficients (Manning's coefficient) and boundary conditions were added to the model. The values selected for manning's coefficient were 0.05 and 0.17, for the stream channel and overflow banks, respectively referring Ethiopian roads authority (ERA) standard document and fixed during calibration. The model was run for mixed flow regime conditions and unsteady flow water surface profile computations. The iterative solution of the energy equation, using the standard step method, solved the unsteady flow, while Manning's equation and contraction/expansion coefficients determined head losses before applying the computation process the model set up for boundary condition. There are various methods of boundary condition used. The method used in this paper is the Normal depth at the downstream end of the reach. The model calculates the depth from the given elevation data and discharge. Finally, the plan must be established for each model simulation.

Terrain Preprocessing 2.4.1 HEC-GeoHMS
HEC GeoHMS allows to preprocess terrain in two approaches either step by step or batch mode. In this research step by step process was used in order to examine out puts and made necessary corrections.
HEC-GeoHMS is a set of ArcGIS tools specifically designed to process geospatial data and create input for the HEC-HMS. HEC-GeoHMS provides the connection for translating GIS spatial information in to model files for HEC-HMS. The GIS capability is for data formatting, processing and coordinate transformation. Currently, HEC-GeoHMS operates on DEM to derive sub-basin delineation and to prepare a number of hydrologic units. HEC-HMS supports these hydrologic inputs as starting point for hydrologic modeling.

…… (1) Equation-51
Where, RMSD is the root mean square deviation is simulated water depth at cross section i is observed flood mark depth at cross section i N is the total number of data (total number of cross sections).

HEC HMS & HEC RAS model evaluation
The performance of selected model was evaluated using statistical measures to determine the quality and reliability of predictions when compared to observed values. Coefficient of determination (R 2 ) and Nash-Sutcliffe simulation efficiency (ENS) were the goodness of fit measures used to evaluate model prediction. The R 2 value is an indicator of strength of relationship between the observed and simulated values. The Nash-Sutcliffe simulation efficiency indicates how well the plot of observed versus simulated value fits the 1:1 line. If the measured value is the same as all predictions, ENS is approximately1. If the ENS is between 0 and 1, it indicates deviations between measured and predicted values. If ENS is negative, predictions are very poor, and the average value of output is a better estimate than the model prediction. The R 2 and ENS values are explained in equations below.
Where: qsi is the simulated value and qoi is the measured value. The proposed model of this study was calibrated by adjusting sensitive parameters and validation was done using observed data of the existing gaging station.

Study area basin characteristics
ARC-GIS 10.4 modeling software is used to delineate catchment for different river system in the Big Akaki and Jemo river at selected outlet catchment and hydrological, physical parameters and spatial information of the catchments were obtained. Road Networks, houses and buildings as well as vegetation cover of the catchment is digitized and corresponding areas is calculated. Hydrological parameters of the study area extracted and curve number (CN), water shed area, soil types obtained and used to run the model. The study area soil, land use characteristics and curve number (CN) were extracted during terrain processing using GIS extensions (Arc Hydro and HEC-GeoHMS).

Best fit flood probability distribution
The test statistic for yearly maximum precipitation and discharge data of selected study area outlet was analyzed using EasyFit software and best fit probability distributions among Normal, Log Normal, General Extreme Value, Chi-Squared and Log Pearson 3 identified. The five probability distributions were compared using Kolmogorove Smirnov,Anderson Darling and Chi-Squared test statistic using EasyFit software. The rank with respective test statistic listed and accordingly, General Extreme Value distribution rank first using Kolmogorov Smirnov and Anderson Darling statistic and Normal distribution rank first using Chi-Squared statistic

HEC HMS hydrological model result 4.2.1 Calibration and Validation
Calibration is adjusting of model parameters based on checking results against observations to ensure the same response over time. This involves comparing the model results, generated with the use of historic meteorological data, to recorded stream flows. The calibration of HEC-HMS for this particular study area was carried out using nine years from 2000-2008 daily rainfall and daily stream flow data of nearby Akaki river. Optimization of the parameter values was carried out within the allowable ranges recommended by the US Army corps of Engineers Hydrologic Engineering Center [5]. Coefficient of determination(R 2 ) (0.68 to 0.88 for Calibration and 0.62 to 0.86 for Validation and Nash-Sutcliffe efficiency (ENS) 0.5 to 0.88 for Calibration and 0.61 to 0.85 for Validation). The Nash-Sutcliffe efficiency (ENS) was found to be 0.751 during calibration and 0.79 during validation. The stochastic nature of precipitation effect on the simulated hydrograph was handled by stochastic calibration. In the stochastic calibration observed and simulated discharge time series were arranged in descending order and the objective function, Σ(Qobs-Qsim) ² was minimized by observing the plot of observed and simulated discharge. The coefficient of determination (R 2 ) was found to be 0.786 (Figure 4) during calibration and 0.801 during validation ( Figure 5). The relationship of simulated and observed flow found to be good and coefficient of determination and Nash-Sutcliffe matches as per the standard indicated above.

HEC HMS model result
The calibrated and validated HEC HMS model was used to estimate peak flood magnitude of the Jemo river selected outlet area using rainfall depth calculated by the best fit distribution for respective return periods. The 24hour depth frequency curve values were used from the Ethiopian Roads Authority drainage manual [6]. Accordingly, the output of the peak flow result for each return period was found as in the below Table 1.

HEC RAS model result
The outcomes of the HEC-RAS model for different values of Manning's roughness coefficient (n) were compared with the observed historic water surface flood level marks collected from specific site which occurred during the year 2010 and 2013 as shown in the Table 2 and represented in Figure 7.  The root means square deviation (RMSD) was tested between observed and simulated water surface elevations as shown in the Table 3.

2D flow area, water depth and water surface elevation
The river two-dimensional (2D) flow area polygon was drawn using aerial imagery and existing terrain data. The 2D flow area mesh generation made and boundary conditions fixed at upstream and downstream locations. Time series flow hydrograph output of calibrated HEC HMS model for 2,5,10,25,50 and 100 years return period was used as an input for boundary conditions for channel and 2D flow area.
The unsteady state simulation model run and depth of flow, water surface elevation and velocity for different return periods of the study river area generated. The 100-year return period simulated water depth simulated high at river stations 902m and 595m from selected outlet and similarly simulated water depth for 50,25,10,5 and 2 years estimated with the HEC RAS model. The river analyzed at different cross sections and less carrying capacity of the river observed at low elevations where there are community settlements very close to the existing river.

Flood hazard water surface extents & velocity map
Water surface extents of the river for each return period was generated with GIS extension (HEC GeoRAS) tool. Areas flooded both left and right side of the river for respective return period delineated and shown as in the Figure 9. The 100-year return period result showed that majority of left side of the river where communities are settled in close proximity is vulnerable and can be affected by the flood. It is observed that the downstream culvert found around Anbessa Garage is overtopped by flash water during rainy season and cause problem to normal traffic movement. Storm water drainage which mark and will have great impact to erode the existing soil at concerned locations. The flood model simulated flow velocity of magnitude less than 2 m/s for major sections of the river and exclusively river stations at 183m,435m,595m,1753m and 3252m found to have flow velocity more than 2m/s and between 3-5m/s.

Flood hazard vulnerability
Flooding area vulnerable for different return period also identified with the model and identified as shown With respect to river sections. Communities settled very close to the river upstream of the Jemo 01 condominium houses and newly built private houses are susceptible by the recurrence flood that can happen any time to cause potential harms to the human lives and livelihoods.

Flood damage estimate
Using HEC RAS model flood inundated area and overlaying the layer on to the latest google satellite image and Addis Ababa master plan, specific river cross sections that were affected by the flood delineated and specific areas also calculated. In order to estimate cost of flood damage an average unit price of birr 12,500 per square meter were assumed and used to estimate the probable damage that can be caused by the model flood. The following table shows residential buildings at respective river cross section that was affected by flood of more than 2m water depth and estimated cost of flood damage.
The HEC RAS flood model showed that for the 2 year return period the flood can cover an area of 1,483 m 2 community properties of majority residential buildings and similarly an area of 1,654m 2 , 1,759m 2 ,1, The simulated depth of water was found in between 0 to 6m for the return periods and varied along the sections depending with the specific terrain. The flow velocity upstream of Jemo 01 condominium areas and upstream sections of the reach were higher than 2.0m/s and this can lead to erosion of the soil and collapse of the existing river banks to cause further flood vulnerability of communities residing closer to the reach. The hydrologic response of the river was for the river carrying capacity of each return period was checked and locations with less carrying capacity for probable maximum flow were at low elevations and river cross sections at 1142m and 435m, 2899m,902m,595m as discussed in the result section.
The study flood model showed that communities upstream of the reach and around Jemo 01 residential areas were identified within the flooding area and can be affected badly if no immediate measure is taken to tackle the probable flooding problem. Based on the flow simulation result of this research the decision makers should consider structural measures such as construction of dike and retaining walls at low elevations along the river which would prevent the flooding caused by over toping.
ACKNOWLEDGEMENT First and for most, I would like to thank my Almighty God for his care, limitless love and support during my study in Addis Ababa Science and Technology University and reach this point in my life. I would like to express my sincere gratitude to my Advisor, Dr. Brook Abate for his persistent support and guidance throughout this study work. I am grateful to Ethiopian Roads Authority for granting this sponsorship program. My appreciation also goes to Ministry of Water, Irrigation and Energy and National Metrological Agency for providing me necessary data to conduct thesis work. Finally, I would like to extend my deepest gratitude to my family, for there encouragement and moral support without your encouragement and care this would not have been possible.