Flood Inundation Mapping And Hazard Assessment- A Google Earth Engine Approach: Case Of The Lower Awash River Basin, Ethiopia.

Abstract

In Ethiopia, flooding is a serious natural hazard that can result in fatalities, property damage, and interruption of livelihoods. Floods are becoming more frequent and powerful, putting sustainable development at risk and calling for efficient flood control measures. This thesis employed Google Earth Engine (GEE) and an integrated hydrodynamic model to map flood inundation and analyze hazards in the Lower Awash River Basin, Ethiopia. The project's objective was to simulate the Lower Awash River basin's water level with flood mapping, to make a hazard assessment for the sub-basin, to compare the Google Earth Engine capabilities with the integration of the hydrodynamic model. The stream flow data of ten gauging stations were analyzed for estimating flood quantiles on the gauged site by applying regional flood frequency analysis based on an L-moment-based index flood. From probability distribution functions; generalized Pareto, generalized extreme value, generalized logistic, and Pearson III were selected as best-fitted distributions for 2, 3, 4, and 1 gauging stations from the total stations respectively. The flood quantiles were estimated using the best-fitted distributions for the specified stations for the return periods of 5, 10,20, 50, 100, 200, 500. At-site quantiles are estimated and fed into HEC-RAS as profiles for developing inundation extent, depth, and flow velocity based on the regional quantiles and the index flood. HEC-RAS Simulation and inundation results were calibrated and validated. Calibration was undertaken for the water level of two stations, Mille and Logia. The performance of the model examined Adjusted n value for the main river channel was 0.042. NSE, PBIAS, and R2 give of values 0.714273 (Very Good), -7.200516629 (Very Good), and 0.8785 (Very Good) respectively for Mille River. NSE, PBIAS, and R2 give values 0.755333 (Very Good), 1.384083045 (Very Good), and 0.8576 (Very Good) respectively for Logia River. The historical flood extent from satellite images and the HEC-RAS simulation were compared for model validation, yielding a good agreement by the overlapping percentage of 88.72. Hazard level and affected land use were computed. A considerable increase in inundated areas was found with increased return periods. The bulk of LARB's floodplain is classified as having very low, low, medium, and high hazards. the results from the three analyses indicate the increasing risk and vulnerability of different areas to flooding events as the return period extends. Agriculture and range land consistently showed high vulnerability. Gewane and Dubti emerged as the most flood-prone locations across all three analyses and agriculture and rangeland were the most affected land uses from the 1D, 2D HEC-RAS, and GEE results. The distribution of hazard levels and affected land use varied across different regions, highlighting the complex interactions between flood hazards and land vulnerability. Therefore, it can be concluded that GEE is a good alternative for flood mapping, especially for data-scarce areas.