1. COSMO-SREPS Priority Project C. Marsigli ARPA-SIM - HydroMeteorological Service of Emilia-Romagna, Bologna, Italy

2. limited-area model error: schemes, parameters, surface forcings COSMO-SREPS methodology  i.c. and b.c. perturbations -> INM multi-model multi- boundary (SREPS)  model perturbations -> parameter perturbation (fixed value throughout the integration to test the impact of the different parameter perturbations) LAM ensemble driving model error analysis error at the global model scale larger scale errors smaller scale errors

3. System applications  COSMO Priority Project -> Short-Range Ensemble  1D-Var DA of satellite data to provide retrievals to be nudged within 7km COSMO-I7: test of the use of COSMO-SREPS to estimate a flow-dependent B matrix  COSMO-SREPS provides boundary conditions for the ensemble system at 2.8 km under development at DWD (COSMO-DE-EPS)

4. Involved COSMO people  Marsigli, Montani, Paccagnella (ARPA-SIM): 1.5 FTE  Flora Gofa, Petroula Louka (HNMS): verification, 0.5 FTE Collaboration:  Susanne Theis, Christoph Gebhardt (DWD): EELMK  Uli Schaettler (DWD): COSMO model code

5. Task 0. Preliminary actions for the system set-up: completed. Task 1. Definition of the methodologies to perturb model trajectories: completed Task 2. Definition of the methodologies to perturb initial conditions: suspended Task 3. System set-up: completed (set-up for the JJASON 2007 period during the MAP D-PHASE DOP) Task 4. Test on case studies: on-going (21 runs in autumn 2006 done, analysis on-going) Task 5. Test on a regular basis: on-going (running on the JJASON 2007 period during the MAP D-PHASE DOP) Task 6. Verification: on-going

6. System set-up 16 COSMO runs 10 km hor. res. 40 vertical levels COSMO at 25 km on IFS IFS – ECMWF global by INM Spain COSMO at 25 km on GME GME – DWD global COSMO at 25 km on UM UM – UKMO global COSMO at 25 km on NCEP AVN – NCEP global P1: control (ope) P2: conv. scheme (KF) P3: tur_len=1000 P4: pat_len=10000

7. intra-group distance 14 days (00UTC runs) 2mT whole domain

8. Daily precipitation ROC area Northern Italy + Switzerland observations Average over 0.5 x 0.5 deg boxes - +42h fc range

9. Daily precipitation ROC area Northern Italy + Switzerland observations Average over 0.5 x 0.5 deg boxes - +66h fc range

10. Relationship between error and spread 12UTC runs Northern Italy observations Nearest grid point

11. ensemble mean 2m Temperature nearest grid point

12. Ensemble member verification Precipitation – 0-0.1 mm / 6h IFS GME UKMO NCEP

13. Ensemble member verification Precipitation – 0-0.1 mm / 6h Tiedtke, pat_len500 Kain-Fritsch, pat_len500 Tiedtke, tur_len1000, pat_len500 Tiedtke, pat_len10000

14. Deterministic scores – ave 0.5 x 0.5 IT+SV 1mm/24h5mm/24h10mm/24h 12 UTC

15. Deterministic scores – ave 0.5 x 0.5 IT+SV 1mm/24h5mm/24h10mm/24h 12 UTC

16. Conclusions  ic and bc perturbations contribute to the spread for an amount approximately double with respect to the parameter perturbations, statistically over the whole domain and the whole period in terms of 2mT. Locally (in space and time) the situation can be quite different  the selected parameters produce a detectable spread among members with the same father  rms error is grater than spread for all the events but the correlation is good  different fathers contribute differently to the skill, for different parameters this is less clear

17. On-going activities  run of COSMO-SREPS with model perturbations for the DOP MAP D-PHASE (6 months, June-November 2007) :  statistical analysis of the system  comparison with the other available mesoscale ensemble systems  verification - system analysis (HNMS)  provide b.c. for DWD ensemble tests  test a flow-dependent B-matrix for 1D-Var of satellite data  test of different parameters (16 runs using only the IFS initial and boundary conditions)

18. Test of parameter perturbations (same father) 16 LM runs at 10 km P1: control (ope) P2: conv. scheme (KF) P3: parameter 1 P4: parameter 2 P5: … IFS – ECMWF global

19. Fine