Abstract:
Understanding the complex relationship between rainfall and runoff processes is judicious
for proper estimation of runoff generated at outlet point. The stream flow of Meki river
watershed is always fluctuating from season to season and this causes flooding problem on
surrounding agricultural land. Estimating the amount of flood is very important to take
appropriate action to mitigate its impacts. Hence, the aim of this study is to model stream
flow and forecast flood of Meki river watershed using Hydrologic Engineering Center
Hydrologic Modeling System (HEC-HMS). The data used for this thesis were soil, Land Use
Land Cover (LULC), Digital Elevation Model (DEM) 30 m x 30 m resolution, 31 years
(1987-2017) daily rainfall and 24 years (1987-2010) daily stream flow data. Daily rainfall
data was collected from Ethiopian Meteorological Agency whereas DEM, soil and daily
stream flow data were collected from Ethiopian Ministry of Water Resources, Irrigation and
Electricity and LULC data was collected from Ethiopian Mapping Agency. The missed
rainfall data was filled using station average and normal ratio method while the missed
stream flow data was filled using linear regression method. The HEC-HMS input parameters
were generated using Hydrologic Engineering Center Geospatial Hydrologic Modeling
System (HEC-GeoHMS). The Gage weight meteorological method was selected to assign the
areal rainfall computed by Isohyetal method to each sub-basin. The Soil Conservation
Service Curve Number, Soil Conservation Service Unit Hydrograph, monthly constant base
flow and Muskingum method were employed for loss computation, excess rainfall
computation, base flow modeling and flood routing respectively. The model was calibrated
and validated with stream flow data of 18 years (1987-2004) and 6 years (2005-2010)
respectively. Nash-Sutcliffe Error (NSE), Root Mean Square Error (RMSE), and Coefficient
of Determination (R
2
) were used to evaluate efficiency of the model, giving values of 0.832,
0.50, and 0.91 and 0.804, 0.40, and 0.89 during calibration and validation respectively. This
indicates very good performance rating of the model. Flood prediction was conducted using
24 hour rainfall depth of 2, 10, 25, 50 and 100 years return period and found to be 133.21,
178.1 239.7, 313.2 and 346.19 m
3
/s respectively. Hence, these predicted values will help
further researchers to prepare flood inundation map and take appropriate actions to control
flood for this study area.
