Developing a Global Model for Flood Susceptibility Mapping Using Machine Learning

2023 Fiscal Year Research-status Report

• Project/Area Number: 23K04328
• Research Institution: Kyoto University
• Principal Investigator: AHMED M.Saber 京都大学, 防災研究所, 特定准教授 (00818403)
• Co-Investigator(Kenkyū-buntansha): 角 哲也 京都大学, 防災研究所, 教授 (40311732) ; カントウシュ サメ・アハメド 京都大学, 防災研究所, 教授 (70750800)
• Project Period (FY): 2023-04-01 – 2026-03-31
• Keywords: global Model / Machine Learning / Flood susceptibility

Outline of Annual Research Achievements

The project goal is to develop a Global Flood Susceptibility Map (GFSM) by using the Machine Learning (ML) model. Up to date, we have examined machine learning algorism in 10 case studies. Then we used cross-validation among the case studies from Japan (Four cases), USA (three Cases), Indonesia (One case), Egypt (One case), Vietnam (One Case), Saudi Arabia (One Case). Now we are comparing the different developed ML functions to be selected to develop the global map.
Within the current year, we are going to develop the first draft of the global flood susceptibility map.

Current Status of Research Progress

1: Research has progressed more than it was originally planned.

Reason

We have already conducted the research plan for last year, by collecting datasets, and run the models on several case studies, and now we are working on developing the best function for flood susceptibility map.

The first draft of our planned global map is expected very soon. The accuracy of the model and results are acceptable.

Strategy for Future Research Activity

The plan as follows:
1. Applying the developed ML model to the collected case studies: Japan (four cases), USA (three cases), Indonesia (one case), Egypt (one case), Vietnam (one Case), Saudi Arabia (One Case).
2. Training and testing by the cross-validation method for all cases.
3. The best function will be used to develop the first draft map for the world.

Research Products

• [Journal Article] Enhancing flood risk assessment through integration of ensemble learning approaches and physical-based hydrological modeling (2023)
  • Author(s): Saber Mohamed、Boulmaiz Tayeb、Guermoui Mawloud、Abdrabo Karim I.、Kantoush Sameh A.、Sumi Tetsuya、Boutaghane Hamouda、Hori Tomoharu、Binh Doan Van、Nguyen Binh Quang、Bui Thao T. P.、Vo Ngoc Duong、Habib Emad、Mabrouk Emad
  • Journal Title: Geomatics, Natural Hazards and Risk Volume: 14 Pages: 1-38
  • Peer Reviewed
• [Presentation] Machine Learning Approaches and Hydrological Modeling for Flood Risk Assessment (2024)
  • Author(s): Mohamed Saber*, Sameh A. Kantoush, Tetsuya Sumi, Emad Mabrouk
  • Organizer: 2nd International Conference Water Resources Management & Sustainability: Solutions for Arid Regions, Dubai, UAE
  • Int'l Joint Research
• [Presentation] Machine Learning for Predicating Flood Inundation in Comparison with Hydrological Models (2023)
  • Author(s): Mohamed Saber, Tayeb Boulmaiz, Sameh A. Kantoush, Tetsuya Sumi, Mawloud Guermoui, Karim I. Abdrabo, Hamouda Boutaghane, Doan Van Binh, Binh Quang Nguyen, Thao T. P. Bui, Emad Habib, Emad Mabrouk
  • Organizer: the 40th IAHR World Congress, Vienna, Austria
  • Int'l Joint Research
• [Presentation] Machine Learning Approaches and Hydrological Modeling approach for Flood risk mapping (2023)
  • Author(s): Mohamed Saber, Sameh A. Kantoush, Tetsuya Sumi
  • Organizer: DPRI annual meeting, Kyoto university, Japan
• [Funded Workshop] 2nd International Conference Water Resources Management & Sustainability: Solutions for Arid Regions, Dubai, UAE (2024)
• [Funded Workshop] 40th IAHR World Congress Vienna-Austria (2023)
• [Funded Workshop] 7th ISFF2023, Algeria (2023)