GWS: Research Areas

Research

We build probabilistic methods for hydroclimatic variability and extremes, and connect them to the decisions that infrastructure, insurance, and governance actually face.

Hydro-Climatic Extremes and Flood Risk

Designing infrastructure, pricing insurance, and protecting cities all depend on one question: how rare and how severe will extreme floods, droughts, and windstorms be? We provide rigorous, probabilistic answers.

We develop statistical and probabilistic models for extreme precipitation, streamflow, drought, windstorms and other hydro-climatic variables — with particular expertise in tail behaviour, return level estimation, and intensity-duration-frequency (IDF) analysis. Our methods are designed to handle the compound nature of hydroclimatic extremes: events where precipitation, wind, and antecedent conditions interact to produce consequences far more severe than any single variable would suggest. We work from urban catchments to continental scales, from present-day conditions to 100-year planning horizons under climate change.

Who it serves

Infrastructure designers and civil engineers · Insurance and reinsurance sector · Municipal and national flood risk authorities · Catastrophe model developers

Representative methods

Extreme value theory (GEV, GP) · Copulas and multivariate distributions · IDF curve construction and uncertainty quantification · Nonstationary frequency analysis

Climate Change and Water-Related Risk

Climate change is making historical records unreliable guides to future risk. Planning infrastructure for 50 or 100 years requires methods that account for shifting patterns and deep uncertainty about what lies ahead.

We model how statistics of precipitation, streamflow and other hydro-climatic variables evolve under climate change, using large climate ensembles, frequency analysis, and scenario-based risk assessment. We translate climate projections into actionable risk estimates for engineers, planners, and policy makers — replacing static design events with probabilistic futures.

Who it serves

Infrastructure asset owners and operators · National climate adaptation programmes · Reinsurance and catastrophe modelling (portfolio stress testing) · DFG / ERC climate-linked funding streams

Representative methods

Single model initial-condition large ensembles (CRCM5-LE) · Multi-model ensembles (EURO-CORDEX) · Downscaling of climate model projections · Scenario-based return level estimation · Compound event analysis

AI-Data-Driven and Operational Hydrology

The scale of modern water risk exceeds what traditional modelling approaches can handle alone. We combine our probabilistic foundations with machine learning to build tools that are physically grounded and operationally scalable.

We integrate our statistical foundations with machine learning surrogates, big-data hydroclimatic infrastructure, and operational forecasting systems. AI enters at multiple levels — not only in model development, but in workflow design, automated quality control, and the synthesis of heterogeneous datasets at scale. Applications range from global precipitation simulation to urban-scale ensemble flood forecasting. The reinsurance sector, which prices risk using copula-based dependence structures and stochastic wind and precipitation fields, is a natural partner for the tools we are building in this area.

Who it serves

Environmental monitoring agencies · Water utilities with operational forecasting needs · Insurance and reinsurance sector (wind and precipitation portfolio modelling) · Technology partners in water and climate risk · Large-scale international research consortia

Representative methods

CoSMoS (open-source stochastic simulation) · Machine-learning surrogates for ensemble-scale modelling · Wasserstein generative adversarial networks · Hybrid statistical-physical models · Space-time random field simulation

Uncertainty, Risk, and Decision Support

Getting the uncertainty wrong in a flood risk assessment is as dangerous as getting the prediction wrong. We connect probabilistic science directly to the decisions that infrastructure owners, planners, and policy makers actually face.

We develop frameworks for quantifying, propagating, and communicating uncertainty in hydrological systems — from model-parameter uncertainty through to risk metrics used in engineering design, insurance pricing, and governance.

Who it serves

Government water agencies and regulators · Water utilities and infrastructure investors · International development banks · Municipal planning authorities

Representative methods

Robust decision-making (RDM) and scenario discovery · Uncertainty propagation through model chains · Return level estimation under deep uncertainty · Governance frameworks for water conflict and adaptation

TU Hamburg

Hamburg University of Technology

Global Water Security

Harburger Schloßstraße 22A

21079 Hamburg

E-mail : igws(at)tuhh.de

Phone : +49 40 30601 3207

Research

Hydro-Climatic Extremes and Flood Risk

Designing infrastructure, pricing insurance, and protecting cities all depend on one question: how rare and how severe will extreme floods, droughts, and windstorms be? We provide rigorous, probabilistic answers.

Climate Change and Water-Related Risk

Climate change is making historical records unreliable guides to future risk. Planning infrastructure for 50 or 100 years requires methods that account for shifting patterns and deep uncertainty about what lies ahead.

AI-Data-Driven and Operational Hydrology

The scale of modern water risk exceeds what traditional modelling approaches can handle alone. We combine our probabilistic foundations with machine learning to build tools that are physically grounded and operationally scalable.

Uncertainty, Risk, and Decision Support

Getting the uncertainty wrong in a flood risk assessment is as dangerous as getting the prediction wrong. We connect probabilistic science directly to the decisions that infrastructure owners, planners, and policy makers actually face.

TU Hamburg

Links