1. Does the Danube exist? Versions of reality given by climate models and climatological datasets Valerio Lucarini, PhD Technical Director HYDROCARE Conference Athens, 18-19/01/2007 Project part-financed by the EU

2. Introduction   Territorial planning and management require the development of deep knowledge concerning some key hydro meteorological and hydrological processes: – –Water is central to human and environmental welfare; – –About 70% of all natural disasters in the world are caused by hydro- meteorological events   HYDROCARE (INTERREG IIIB – CADSES): Time: 2006- 2007, Budget: 2.5 M€; Partnership 11 institutions from 6 countries (Italy, Germany, Greece, Poland, Romania, and Slovakia). Lead Partner, CINFAI, Italy – –Mission: Analysis of the hydrological cycle of the CADSES area by adopting an integrated and multidisciplinary approach. – –Web-site: http://www.hydrocare-cadses.net   The assessment of the reliability of the current RCMs for the climatology of the water balance (mean value & variability), of the basin of the Danube river is crucial, because of its relevance at social, economical and environmental level. This the reason for its centrality in the project HYDROCARE.

3. Back to the Danube. Basics   Data sources: – –ERA-40 reanalysis data – –NCEP/NCAR reanalysis data – –Regional Climate Models Control data – Prudence project – –Global Runoff Data Center – GRDC – –Met Office, Hadley Center, UK (driving data)   Daily values of: – –Precipitation (P) – –Evaporation (E) – –Runoff (R) – –Observed discharge data (GRDC)   Area of interest: – –Danube: length river 2850 Km, Area basin 807 000 km 2 – –Period of 30 years: 01.01.1961 – 31.12.1990 – –Calculation of integral values (over the area, using GIS tools) of: P, E, R, Precipitation – Evaporation (hydrological balance), (P - E) Mass conservation: Courtesy of CIA

4. Regional CM (PRUDENCE 5 FP) CodeModelDriving dataInstituteCountryDatalat x lonVL CLM GKSS CLMHadAM3H A2 GKSS Research Centre GeesthachtGermanyDaily0.50° x 0.50°20 HIRHAM METNO HIRHAMHadAM3H A2 Norwegian Meteorological InstituteNorwayDaily0.46° x 0.46°19 CHRM ETH CHRMHadAM3H A2 ETH - Swiss Federal Institute of Technology SwitzerlandDaily0.50° x 0.50°20 PROMES UCM PROMESHadAM3H A2 UCM - Universidad Complutense de Madrid SpainDaily0.50° x 0.50°26 RACMO KNMI RACMOHadAM3H A2 KNMI - The Royal Netherlands Meteorological Institute, University of Reading Netherlands, UK Daily0.44° x 0.44°31 REMO HadAM3H A2 MPI - Max-Planck-Institute for Meteorology GermanyDaily0.50° x 0.50°19 SHMI25RCAO – high resolution HadAM3H A2 SMHI – Swedish Meteorological and Hydrological Institute SwedenDaily0.22° x 0.22°59 SHMI50RCAOHadAM3H A2 SMHI – Swedish Meteorological and Hydrological Institute SwedenDaily0.44° x 0.44°24 DMI12HIRHAM – extra high res. HadAM3H A2 DMI - Danish Meteorological InstituteDenmarkMonthly0.15° x 0.15°19 DMI25HIRHAM – high resolution HadAM3H A2 DMI - Danish Meteorological InstituteDenmarkDaily0.22° x 0.22°19 DMI50HIRHAMHadAM3H A2 DMI - Danish Meteorological InstituteDenmarkDaily0.44° x 0.44°19 ICTPICTP – RegCM HadAM3H A2 ICTP The Abdus Salam Intl. Centre for Theoretical Physics ItalyDaily0.44° x 0.44°23

5. Other data (Verification) CodeDatasetInstituteCountryAvailable data lat x lonLevels ERA40ERA-40, T159 resolution – Reamalyses ECMWF –European Center for Medium-Range Weather Forecast UK4XDaily2.5° x 2.5°60 NCEP- NCAR NCEP-NCAR - Reanalyses National Center for Environmental Prediction – National Center for Atmospheric Research USA4XDaily1.905° x 1.875° 28 HadAM3HadAM3H model– A2 scenario (forced by observed SST and sea ice) Hadley Centre for Climate Change - Met OfficeUKDaily1.25ºx1.875º19 Obs. Disc. Danube discharge at Ceatal Izmail station Global Runoff Data CenterGermanyMonthly Est. Obser. Danube basin runoff reconstructed as in Hagemann et al. (2004) Global Runoff Data CenterGermanyMonthly

6. Data Gridding Voronoi Polygon

7. Statistics of the Yearly time series  Balance (Precipitation – Evaporation)  Precipitation  Evaporation  Runoff

8. Mean vs. Variability ERA-40 NCEP High Med Low μ σ

9. P vs. E μ(E) μ(P) ERA-40 NCEP Med Low High

10. Correlation with Driving AGCM (1) Precip Evap MODELSC(P,P)C(E,E) CLM_GKSS_germany0,910,66 HIRHAM_METNO_norway0,900,53 CHRM_ETH_swiss0,870,70 PROMES_UCM_spain0,87-0,16 RACMO_KNMI_netherland0,930,71 REMO_germany0,880,72 SMHI_25_sweden0,840,75 SMHI_50_sweden0,890,79 DMI_12_denmark0,850,73 DMI_25_denmark0,870,76 DMI_50_denmark0,800,75 ICTP_italy0,840,66

11. P-E Feedback P vs E DRIVING DATA 0,90 CLM_GKSS_germany0,81 HIRHAM_METNO_norway0,58 CHRM_ETH_swiss0,84 PROMES_UCM_spain0,04 RACMO_KNMI_netherland0,69 REMO_germany0,82 SMHI_25_sweden0,87 SMHI_50_sweden0,89 DMI_12_denmark0,91 DMI_25_denmark0,90 DMI_50_denmark0,92 ICTP_italy0,80 NCEP/NCAR0.72 ERA40-0.41

12. Correlation with Driving AGCM (2) P-E=B MODELSC(B,B) CLM_GKSS_germany0.92 HIRHAM_METNO_norway0.92 CHRM_ETH_swiss0.89 PROMES_UCM_spain0.90 RACMO_KNMI_netherland0.93 REMO_germany0.86 SMHI_25_sweden0.85 SMHI_50_sweden0.89 DMI_12_denmark0.86 DMI_25_denmark0.85 DMI_50_denmark0.78 ICTP_italy0.85

13. Runoff vs Balance ERA-40 NCEP High Med Low μ(R) μ(B) Not good!

14. Seasonal Cycle  Balance (Precipitation – Evaporation)  Precipitation  Evaporation  Runoff

15. PRECIPITATION Max Min 100%

16. EVAPORATION Max Min 100% Min Negative balance

17. BALANCE Min 100% Max Negative balance

18. RUNOFF Min Max Amplitude Phase

19. Geographical limits to water transport?   The Mediterranean Sea play a relevant role in the hydrology of the Danubian region both for the mean state and the extreme events.   The largest impact in terms of precipitation of the Mediterranean water vapor is in the regions downwind of the Sea, thus including Central-Eastern Europe.   The Danube depends almost entirely on precipitated water of Mediterranean origin. Similarly, a very strong Mediterranean influence exists for Elbe, Oder, and Vistula, since they or their main tributaries originate from mountains (Carpatians, Sudety, Erzebirge) which catalyze the precipitation of Mediterranean water

20.   Most of the major floodings occurred in central-eastern Europe are due to a typical Mediterranean meteorological pattern, the Genoa cyclone.

21. Conclusions   NCEP and ECMWF Reanalyses are largely inadequate for representing the hydrology of the Danube basin;   RCMs feature large discrepancies for the climatology of water balance: most underestimate the discharge of the Danube; they act as differently parameterized downscaling of the driving GCM;   Only few models (METNO, SHMI, KNMI) provide estimates which are consistent with the observed discharge values of the Danube at its Delta;   Most RCMs have a large and anticipated mean seasonal cycle (small damping); problems in representation of snow depletion: KNMI model agrees remarkably well with observed data;   The agreement between mean integrated P-E and runoff is not perfect;   The considered approach relies on the mass conservation principle at the air-land interface and bypasses the details of soil modelling and will be used for analyzing climate change scenarios.   Analysis of meteorological processes and of transport of water vapor of Mediterrabean origin is crucial – –Meteorological Hydrological Cycle, not Geographical Hydrological Cycle

22. Sligthly tragically..  While the RCMs actually act as strongly constrained downscaling models, at the same time, once outputs are upscaled via spatial integration procedure on a finite - not too large, not too small domain, as discussed earlier - domain, information may be, and actually in most cases is, degraded.