Towards scalable design rainfall in Europe

Design rainfall is an important input in simulations of pluvial flooding. It is often based on so-called Intensity-Duration-Frequency (IDF) curves which represent statistics on past extreme rainfall events. Local rainfall observations, ideally at the sub-hourly temporal resolution, are necessary for constructing IDF curves. As these observations are scarce, representative design rainfall events can be difficult to come by, especially in ungauged areas.

By training a machine learning model on local IDF curves and geographical and climatological covariates, this project aims to develop a scalable method for generating IDF curves for present and future climates in Europe. The first step is to create a target dataset of local IDF curves using a generic method across countries.