Name: Advancing Flood Forecasting with Machine Learning at ECMWF
Start: 2026-01-30T11:00:00+00:00
End: 2026-01-30T12:00:00+00:00
Location: Worsley Building, Room 9.58a

Advancing Flood Forecasting with Machine Learning at ECMWF

A presentation on AIFL, a deterministic LSTM-based model under development for global daily streamflow forecasting.
About this Event

A hybrid event held by the SciML Community at Leeds Institute for Data Analytics

Speaker: Maria Luisa Taccari (University of Leeds Alumnus) from the European Centre for Medium-Range Weather Forecasts (ECMWF)

Abstract:

Flood forecasting at large scales is critical as extreme events become more frequent. ECMWF provides the computational backbone for operational systems such as GloFAS and EFAS, delivering global and European streamflow predictions. Recent advances in machine learning offer new opportunities to complement these models and improve predictive skill across diverse basins.

I present AIFL, a deterministic LSTM-based model under development for global daily streamflow forecasting. AIFL is designed to integrate into operational workflows, capturing key hydrological dynamics while providing robust predictions for both routine and extreme events. I will outline its architecture, training strategy, and preliminary evaluation results.

Bio:

Maria Luisa is a researcher in scientific machine learning and hydrology. She works at the European Centre for Medium-Range Weather Forecasts (ECMWF) within the Destination Earth initiative, developing machine-learning workflows for hydrological forecasting with a focus on river discharge, and is currently on secondment as a Visiting Scientist at the Joint Research Centre (JRC) of the European Commission.

She completed her PhD at the University of Leeds (2024) with the thesis “Deep Learning for Groundwater Prediction” (supervised by Professor Peter Jimack, Dr He Wang and Dr Xiaohui Chen). Prior to her PhD, she worked as a Consultant and Researcher at Deltares in the field of flood protection.