Flood susceptibility mapping using earth observation data and tree-based ensemble machine learning: case study of Wouri Estuary in Cameroon
- Paper number
IAC-23,B1,IP,31,x80184
- Author
Mr. Chukwuma Okolie, Space Generation Advisory Council (SGAC), Nigeria
- Coauthor
Mr. Stephane Lako Mbouendeu, Cameroon
- Coauthor
Mr. Abdulwaheed Tella, Space Generation Advisory Council (SGAC), China
- Coauthor
Mr. Charles-aimé Nzeussi Mbouendeu, International Space University (ISU), France
- Coauthor
Mr. Ikenna Arungwa, Federal University of Technology Owerri (FUTO), Nigeria
- Coauthor
Mr. Swarnajyoti Mukherjee, Apogeo Space Srl, Italy
- Coauthor
Mr. Krittanon Sirorattanakul, California Institute of Technology, United States
- Coauthor
Mr. Jubril Okeyode, Space Generation Advisory Council (SGAC), Nigeria
- Coauthor
Dr. Barthelemy Ndongo, Space Generation Advisory Council (SGAC), Cameroon
- Coauthor
Ms. Lisah Ligono, Kenya
- Coauthor
Ms. Chnomnso Onwubiko, Space Generation Advisory Council (SGAC), Nigeria
- Coauthor
Dr. Ngozi Johnson, Space Generation Advisory Council (SGAC), Nigeria
- Coauthor
Dr. Ugonna Nkwunonwo, United Kingdom
- Coauthor
Mr. Hassan Musa, AHMADU BELLO UNIVERSITY, Nigeria
- Coauthor
Mr. FRANCK ERIC TCHAMENI, Cameroon
- Coauthor
Mr. Ayila Adzandeh, Space Generation Advisory Council (SGAC), Nigeria
- Coauthor
Mr. Junior Iroume, Institute of Geological and Mining Research, Cameroon
- Coauthor
Mr. AbdulAzeez Onotu Aliyu, AHMADU BELLO UNIVERSITY, Nigeria
- Coauthor
Ms. Daniela Vargas-Sanabria, Universidad Estatal a Distancia (UNED), Costa Rica
- Coauthor
Dr. Desire Muhire, Space Generation Advisory Council (SGAC), Austria
- Coauthor
Mr. Ishaku Yakubu, Space Generation Advisory Council (SGAC), Nigeria
- Coauthor
Ms. Syeada Tasnim, Space Generation Advisory Council (SGAC), Bangladesh
- Coauthor
Mr. Abinash Silwal, Space Generation Advisory Council (SGAC), Nepal
- Coauthor
Dr. Carole Bonguen, Space Generation Advisory Council (SGAC), Cameroon
- Coauthor
Mr. Adedoyin Ajeyomi, Space Generation Advisory Council (SGAC), Nigeria
- Coauthor
Mr. Dan Yang Damakoa, Space Generation Advisory Council (SGAC), Cameroon
- Coauthor
Ms. Anshul Dixit, Space Generation Advisory Council (SGAC), India
- Year
2023
- Abstract
The REFRA-SOS (Realtime Flood Risk Assessment in developing countries using Social media, Optical and SAR satellite data) project aims to mitigate flooding disaster risk in Cameroon by using the latest technologies available. To date, our studies have mapped communities located in flood prone areas of Doula Estuary in Cameroon. The impact of flooding is debilitating on livelihoods and socio-economic activities. In the present study, we adopt tree-based ensemble machine learning algorithms integrated with earth observation data for flood susceptibility mapping in Douala. Tree-based ensembles present several advantages such as interpretability, less data preparation, tolerance to multicollinearity, versatility, and ability to handle non-linear and complex relationships. Eleven flood conditioning factors (elevation, slope, topographic wetness index, terrain ruggedness index, distance to water bodies, drainage density, annual rainfall distribution, land use/land cover, soil texture, normalised difference vegetation index and modified normalised difference water index) will be integrated for flood prediction using the random forest (RF), extreme gradient boosting (XGBoost), light boosting machine (LightGBM) and categorical boosting (CatBoost) algorithms. The overall accuracy of the flood susceptibility map will be assessed to determine its sensitivity and robustness, and the performance models will be compared in terms of training speed and prediction accuracy. The findings will have important implications for policy makers involved in flood management and disaster risk reduction in coastal cities, particularly in Cameroon. By promoting the use of satellite-based data and machine learning approaches, the study aims to improve disaster risk reduction strategies and promote sustainable development in coastal cities.
- Abstract document
- Manuscript document
(absent)