1. Multilevel drought reanalysis over France with Safran-Isba-Modcou hydrometeorological suite Jean-Philippe Vidal 1, Éric Martin 1, Laurent Franchistéguy 2, Jean-Michel Soubeyroux 2, Martine Baillon 2 and Michèle Blanchard 2 1 CNRM/GAME, Météo-France/CNRS 2 Météo-France, Direction de la Climatologie EGU General Assembly 2009, 20-24 April 2009

2. Contents  ClimSec project  Tools and methods –Safran-Isba-Modcou –Drought indices  Overview of drought patterns –Spatial extension –Propagation through the hydrological cycle  From indices to events –Local scale –National scale  Conclusions and further work

3. ClimSec project  2-year project: “ Impact of climate change on drought and soil moisture in France ”  Motivated by damages to buildings following 2003 drought due to soil shrinking and swelling; Aims at finding whether such events will be more frequent in the future  Funded by Fondation MAIF  Partners: –Météo-France, Climatology Department –CNRM/GAME –CERFACS –UMR Sisyphe  Focus on the reanalysis of drought events for characterizing present climate

4. Safran-Isba-Modcou (SIM) Safran atmospheric analysis system Optimal Interpolation of ERA-40 + ground observations Gridded rainfall, snowfall, temperature, humidity, wind speed, visible, infrared radiation (8km, hourly) Isba land surface scheme Computation of water and energy budgets with soil and vegetation databases (ECOCLIMAP) Evapotranspiration, soil moisture, snow cover, drainage, runoff, etc. (8km, 5 min) Drainage and runoff Modcou hydrogeological model Flow routing, multilayer aquifers Flow (3-hourly) and water table levels (1-8km) – w : water content – w wilt : wilting point – w fc : field capacity 50-year SIM hydrometeorological reanalysis (1 August 1958 to 31 July 2008) Vidal et al. (submitted) A 50-year high-resolution atmospheric reanalysis over France with Safran system. International Journal of Climatology

5. Drought indices  Drought : abnormal water deficit in one (or more) part of the land surface hydrological cycle  Choice of relevant drought indices depends on the socio-economic activity domain  No universal drought index  SIM: consistent computation of variables within the hydrological cycle VariableDrought type SafranPrecipitationMeteorological IsbaSoil Wetness IndexAgricultural ModcouRiver flowHydrological

6. Drought indices  Approach –Use of a monthly variable summed/averaged over n months –Kernel density estimates for each calendar month and grid cell –Quantile-quantile projection onto normal distribution  Advantages –Computation with reference to 50-year local climate –Correspondance index value / non-exceedance probability –Spatial consistency –Different time scales: 1 to 24 months  Indices –Standardized Precipitation Index (SPI) → 8km grid –Standardized Soil Wetness Index (SSWI) → 8km grid –Standardized Flow Index (SFI) → >900 hydrometric stations

7. Overview of dry periods Area affected by drought (% of France area), 3-month indices, threshold 5% Autumn 197819761989-19902003 SPI3 SSWI3 SFI3

8. SSWI3 Example of drought propagation  Year 1976, 3-month indices, traditional SPI drought classes Long startHeartRecovery Soil memory “Extremely wet” flow conditions propagating downstream Loire and Allier rivers Heavy convective rainfall on rather dry soil  High runoff Extreme drought Severe drought Moderate drought Mild drought Mildly wet Moderately wet Severely wet Extremely wet Soil and vegetation resistance SPI3

9. From indices to events – Local scale Grid cell located near Toulouse (south-west), SSWI3 Duration Severity Magnitude Threshold P = 5% Starting date Peak date End date

10. Some local characteristics  50-year analysis of SSWI1, threshold P = 10% Urban areas Sandy soils with crops Sandy soils with forests Autumn starting droughts Winter starting droughts Spring droughts

11. From indices to events – National scale  Droughts are developping in both space and time  Event identification –Event envelop: index threshold (ex: P=20%) –Spatial contiguity: cluster identification at each time step –Temporal continuity: minimum common area between two time steps  Spatio-temporal characteristics 1.[Duration, severity, magnitude] combined with area

12. Mean characteristics Mean duration (months) SSWI3 P=20% Mean area (% of France) Mean severity

13. From indices to events – National scale  Droughts are developping in both space and time  Spatio-temporal event identification –Event envelop: index threshold (ex: P=20%) –Spatial contiguity: cluster identification at each time step –Temporal continuity: minimum common area between two time steps  Spatio-temporal characteristics 1.[Duration, severity, magnitude] combined with area 2.Identification of most severe events through Severity–Area–Time Scale (SAT) curves inspired from depth-area- duration analysis of storm precipitation (WMO, 1969; Andreadis et al., 2005)

14. SAT curves of most severe events Severity SPI3 SPI6 SPI12 Area (%) SSWI3SSWI6 SSWI12 P=20%

15.  Safran-Isba-Modcou hydrometeorological suite –Consistent approach for monitoring drought propagation through the hydrological cycle  Standardized drought indices –Powerful analysis tools for characterizing events in space and time  Drought reanalysis –Identification and description of past drought events at both local and national scale in France Conclusions

16. Further work  Evolution of drought characteristics under climate change  Dealing with uncertainty 1.Emissions scenarios A1B A2 2.Range of climate models Transient projections from different versions of ARPEGE Time slices from IPCC AR4 models 3.Two downscaling methods Weather types (Boé et al., 2006) Quantile-quantile (Déqué, 2007)

17. Thank you for your attention Contact: jean-philippe.vidal@meteo.fr 1988-1990 SSWI3 drought (envelop P=20%)