Geomorphological characterization of fluvial flood hazard across the island of Oahu, Hawaii (USA)

Assessing floods hazard is critical to preventing damages and acting fast when a flood event occurs. Flood hazard assessment can be costly, time-consuming, and complex due to running complicated hydrodynamic models of rivers and floodplain systems. Besides, several model runs are normally necessary because different scenarios over large areas must be analyzed for a full characterization of flood hazards. To solve this problem and make the analysis easier with less cost and time, we can use machine-learning techniques. Machine-learning and data-driven methods can be the best alternative way to conventional methods because of their simplicity, faster processing, handling large amounts of data, and more importantly, learning from the past datasets and models to improve themselves. In this Thesis, data-driven DEM-based algorithms developed with the SaferPlaces project (funding: European Commission through EIT Climate-KIC) have been applied to analyze the fluvial and pluvial floods across Oahu island, USA.