Work Package 4 is focused on modelling near-term extreme weather events and assessment of potential hazards. The first task is the creation of a baseline estimate of extreme events. This will be expanded to include geographical characteristics, regional climate statistics and year-specific features to ensure accuracy for the Demonstrator locations. This will be done using extreme value theory (EVT) methods. As a part of this exporation, numerical weather prediction systems will be assessed to see if these systems can be used as an input for the EVT (to include both long-term changes and seasonal variability).
To ensure the data is relevant for stakeholders, co-production will be used to develop a toolkit and hazard indices tailored for climate adaptation planning. Workshops will be held with Demonstrator stakeholders to translate climate variable changes into the implementation of indices, storylines or other climate information needed by stakeholders for climate adaptation planning.
Some of the methods used include:
- Classic Extreme Value Theory Methods are traditional approaches to modeling extremes. They often involve generalized extreme value (GEV) methods for block maxima and peak-over-threshold (PoT) methods for exceedances. These methods assume year-to-year stationarity and model extremes at fixed locations and therefore work well within the scope of the Demonstrator Cities.
- Bayesian model averaging (BMA) approaches are employed to incorporate model uncertainty into the training process.
- Time of Emergence (TOE) estimation is crucial for stakeholders to understand when a hazard will emerge in a specific location. This is determined relative to global warming levels (GWL), considering uncertainty quantification associated with GWL thresholds.