Research

Water extremes, like too much water (heavy rainfall, floods, or storm surges) or too little water (droughts), are becoming more severe and frequent due to human-driven impacts on Earth. Their prediction remains challenging given the Earth’s complexity and is typically done probabilistically.

At HEWSL, we aim to advance predictions of water extremes to enhance water security and support resilient and adaptive communities. 

1. Understanding water extremes.

We investigate the causes, changes, and interplay of water extremes. We aim to uncover how water extremes are generated, how they evolve, and how they interact and manifest. These are key to inform predictive modelling.

To address this, we use diverse methods, including attribution approaches, causal inference, process-based impact models (e.g., hydrological models), statistical analyses, and machine learning.

2. Next-generation prediction of water extremes under uncertainty. 

We develop probabilistic prediction approaches that reflect the Earth system with increasing accuracy and realism: capturing nonstationarity, interdependencies, multiple generating mechanisms, compound extremes (occurring simultaneously or sequentially), and spatial patterns.

Our work combines statistical and probabilistic modelling, hydrological models, machine learning, post-processing methods (bias correction/statistical downscaling), hydrological frequency analysis, data assimilation, and uncertainty quantification, among others.

3. Informing decision-making.

We build flexible, coherent, and reproducible prediction frameworks to inform decisions related to water extremes under uncertainty. These tools support adaptive water management, infrastructure design and planning, and policymaking.

Our work includes Our work includes impact-based prediction, catastrophe modeling, co-development of workflows and open-source tools, stakeholder engagement, and fostering communities of practice.