Prediction methods have been developed through a cycle of designing, implementing, applying and evaluating prediction systems (see figure). Interactions between workpackages, and with the user community that informs all activity in the project, were key to the plan. This required close research collaborations to ensure that new developments were relevant, that methods were applied appropriately, and that results were interpreted correctly. Whilst important progress was made within workpackages, it is the workpackage interactions that have likely produced the most significant advances.
These closely related measures of extremes had been chosen because they were critical inputs to assessments of water resource and sustainability of living in large cities, and are a useful benchmark for decadal / near term prediction system and uncertainties, as they were better understood than some more localized variables, e.g. daily extreme rainfall causing flooding
The workpackages focussing specifically on prediction methods were:
Lead: Lenny Smith (LSE)
Contributing: Mat Collins (MOHC)
Role: worked directly with users and with other WPs to develop new approaches to the design of ensemble prediction systems that focus on information content and utility.
Lead: Chris Ferro (Exeter) Contributing: David Stephenson (Exeter), Simon Tett (Edinburgh)
Role: developed new methods for evaluating climate and impacts predictions, and supported the use of these methods to evaluate the predictions produced by other WPs.
Lead: Myles Allen (Oxford)
Contributing: Mat Collins (Hadley Centre)
Role: worked with other WPs in implementing uncertainty analyses and prediction systems and to conduct novel uncertainty analysis climate variables for AR5.