Machine Learning for Flood Prediction: The Contribution of SAR and Multispectral Indexes

Flood forecasting is crucial for mitigating the impact of extreme hydrological events, especially in urban environments where sudden flooding can cause severe damage to infrastructure and human lives. This thesis explores a machine learning approach to predicting river flood events based on historical hydrological and meteorological data and leveragin information provided by two types of satellite images: SAR images and Multispectral images. The study focuses on a binary classification problem and aims to investigate whether SAR and multispectral indices can provide valuable information when combined with meteorological and topographical features in order to forecast a flood event.