[P38] Integrating machine learning in tsunami damage assessment for buildings and roads: insights from the 2011 Japan tsunami
| 所属 | University of LAquila |
|---|---|
| 筆頭著者・発表者 | Scorzini Anna Rita |
| 共著者 | Mario Di Bacco(University of Florence) Anawat Suppasri(IRIDeS, Tōhoku University) James H. Williams(University of Canterbury) Daisuke Sugawara(IRIDeS, Tōhoku University) |
キーワード
- Tsunami
- Roads
- Buildings
概要
Accurate damage assessment following a tsunami is essential for effective disaster risk management. Traditional methods relying on univariate fragility functions often fail to capture the complex mechanisms influencing tsunami damage. To overcome this limitation, this study explores the application of machine learning techniques to two building and road damage datasets collected after the 2011 Great East Japan tsunami. The original datasets contain detailed damage data and additional variables for over 250,000 buildings and 4,000 km of inundated roads. In this study, the datasets are enhanced with new explanatory variables, such as hydraulic characteristics of the event, structural and geometric properties of the exposed assets, and surrounding environmental factors. A novel buffer-based approach is applied to capture the interactions between structures, including shielding effects that might reduce damage to nearby buildings, as well as the influence of collapsed structures that could generate debris. The developed models integrate these factors to improve tsunami damage predictions. The analysis reveals that while inundation depth remains a critical predictor, other variables also play significant roles, with coastal topography and interactions between neighboring structures being crucial for buildings, and wave approach angle, road orientation, and potential overflow from inland watercourses being key for roads.
