Design flood estimation for global river networks based on machine learning models

Hydraullic modeling 22.11.2021

In this research led by Dr Gang Zhao, researchers propose a machine-learning-based approach to estimate design floods globally.

Research led by Dr Gang Zhao at the University of Bristol proposes a new three staged approach to estimating design floods at large scale. 

Work involved building a new RFFA by comparing a total 11 793 global gauging stations to develop a machine learning technique for predicting gauge flows.

Results observe considerable improvement in extreme flow estimation compared to Fathom’s previous RFFA methods, with low bias and prediction accuracy within typical measurement errors of gauge-based discharge. Zhoa’s discharge model is a hugely important development, that will now underpin all of Fathom’s future fluvial models.