1. PRUDENCE STARDEX MICE http://www.cru.uea.ac.uk/projects/mps/

2. STARDEX STAtisical and Regional dynamical Downscaling of EXtremes for European regions Clare Goodess Climatic Research Unit, UK http://www.cru.uea.ac.uk/projects/stardex/

3. The STARDEX team University of East Anglia, UK King's College London, UK Fundación para la Investigación del Clima, Spain University of Bern, Switzerland Centre National de la Recherche Scientifique, France Servizio Meteorologico Regional, ARPA-Emilia Romagna, Italy Atmospheric dynamics group, University of Bologna, Italy Danish Meteorological Institute, Denmark Eidgenössische Technische Hochschule, Switzerland Fachhochschule Stuttgart - Hochschule für Technik, Germany Institut für Wasserbau, Germany University of Thessaloniki, Greece

4. The STARDEX objectives To rigorously & systematically inter-compare & evaluate statistical & dynamical downscaling methods for the reconstruction of observed extremes & the construction of scenarios of extremes for selected European regions. To identify the more robust downscaling techniques & to apply them to provide reliable & plausible future scenarios of temperature & precipitation-based extremes for selected European regions.

5. Statistical downscaling Relationships between larger-scale climate variables & local surface climate variables, derived from observed data, are applied to climate model output……... based on the two assumptions that: larger-scale variables are more reliably simulated relationships remain valid in a changed climate

6. Spatially coherent changes in extremes have occurred over the last 40 years...

7. 1958-2000 trend JJA heat wave duration Scale is days per year. Red is increasing Malcolm Haylock, UEA Property damage: US$ 13 bn Fatalities: 27,000 (14,800 in France) Western Europe August 2003

8. 1958-2000 trend in JJA heavy rain events Scale is days per year. Blue is increasing Central and Eastern Europe August 2002 Fatalities: > 100 Economic losses: > US$18 bn Insured losses: > US$3 bn Malcolm Haylock, UEA

9. In part, these changes can be explained by changes in circulation & other predictors e.g., Heavy winter rainfall and links with North Atlantic Oscillation/SLP Haylock & Goodess, IJC, 2004

10. The STARDEX SDS methods multiple linear regression canonical correlation analysis artificial neural networks multivariate autoregressive model conditional re-sampling & other analogue- based methods methods based on a potential precipitation circulation index & critical circulation patterns conditional weather generator local & dynamical scaling

11. Handling many combinations of different methods (20+), regions (7), indices (13) & seasons (4) was difficult!

12. Key messages from the STARDEX verification work Skill varies station- station, season- season, index- index, method- method SE England: Haylock et al., IJC WinterSpring SummerAutumn

13. A2 scenario changes for the W Alps: Schmidli et al., GRL In majority of cases no consistently superior model, so a major recommendation is to use a range of the better SDS methods – just as the recommendation is to use multiple GCMs/RCMs Winter

14. A2 scenario changes for the W Alps: Schmidli et al., GRL Uncertainties are larger/skill lower – so use scenarios of summer rainfall with care Winter Summer

15. Projected changes in flow duration Cochem gauge, Mosel (USTUTT-IWS) Black: present Green: B2, 2080s Red: A2, 2080s % change in greatest 5-day winter rainfall A2, 2080s % change in greatest 5-day winter rainfall, 1958-2001, 611 stations

16. We have produced recommendations and guidelines for those wanting to undertake SDS – and to help users identify the most suitable methods For example: Robustness criteria Applications criteria Performance criteria

17. http://www.cru.uea.ac.uk/projects/stardex We have produced outputs in a wide range of formats: reports, papers and information sheets

18. So STARDEX provides a sound scientific starting point for ENSEMBLES….. http://www.cru.uea.ac.uk/projects/stardex/