Geospatial Technology–AI (Machine Learning Algorithm) Application in Flood Inundation Frequency and Extent Analysis and Modeling in Becho Plain, Western Central Ethiopia
| dc.contributor.advisor | Tibebu Kasawmar | |
| dc.contributor.author | Jifara Dabessa | |
| dc.date.accessioned | 2025-09-12T21:09:59Z | |
| dc.date.available | 2025-09-12T21:09:59Z | |
| dc.date.issued | 2024-06 | |
| dc.description.abstract | Floods are one of the most devastating forces in nature. Flooding is a significant natural hazard affecting the Becho plain in western-central Ethiopia, causing extensive environmental, social, and economic damage. This study utilizes advanced geospatial technologies and artificial intelligence (AI) techniques, machine learning (ML) algorithms, to predict and analyze flood inundation frequency and extent in this region. The study integrates historical flood records from the Awash River (1995−2018), climate data (1990−2023), and SAR imagery (2015−2023) to examine the relationship between climatic variables, flood events, and the spatial extent of inundation. The analysis involves evaluating various flood frequency models such as Generalized Extreme Value (GEV), log-Pearson III, and Gumbel distributions, with the GEV distribution demonstrating the best fit for river flow data (K-S statistic of 0.17 and p-value of 0.39). Furthermore, machine learning models including Support Vector Regression (SVR), Random Forest (RF), and Linear Regression (LR) were applied to predict flood frequencies and extents. Linear Regression was identified as the most accurate model, achieving a Mean Absolute Error (MAE) of 17.71 and a Cross-Validated MAE of 20.63. The calculated flood frequency and river flow overall mean and rainy season, the trend analysis showed a general increasing trend in the mean flow, with an overall trend of 0.12 m³/s per year and a rainy season trend of 0.56 m³/s per year. Moreover, based on linear regression analysis of time series flood extent, the slope is positive that the area flooded is increasing per year. Furthermore, geospatial analysis revealed critical flood-prone areas and their temporal changes, providing valuable insights into flood dynamics. The results highlight that integrating geospatial and AI methodologies significantly enhances the prediction and modeling of flood-prone areas. This approach offers a robust framework for improving flood risk management and mitigation strategies, ultimately contributing to better preparedness and resilience in the Becho plain. | |
| dc.identifier.uri | https://etd.aau.edu.et/handle/123456789/7435 | |
| dc.language.iso | en_US | |
| dc.publisher | Addis Ababa University | |
| dc.subject | Climate Data | |
| dc.subject | Flood Inundation | |
| dc.subject | Frequency Analysis | |
| dc.subject | Machine Learning | |
| dc.subject | Remote Sensing | |
| dc.title | Geospatial Technology–AI (Machine Learning Algorithm) Application in Flood Inundation Frequency and Extent Analysis and Modeling in Becho Plain, Western Central Ethiopia | |
| dc.type | Thesis |