COSMOSREPS Priority Project C Marsigli ARPASIM Bologna Italy

COSMO-SREPS Priority Project C. Marsigli ARPA-SIM, Bologna, Italy

COSMO-SREPS v aim: to develop a limited-area ensemble system for the short-range considering also small scale uncertainty boundary conditions initial conditions LAM ensemble model (LM): schemes, parameters, surface forcings v i. c. and b. c. perturbations -> INM multi-model v LM perturbations

people involved in the PP Involved COSMO people: • Marsigli, Montani, Paccagnella (ARPA-SIM) + collaborations: • S. Theis (DWD) – discussion on the methodologies and sharing of perturbation techniques • U. Schaettler (DWD) – new version of LM developed to move from the code to a namelist a number of tuneable parameters • WG 2 and WG 3 people has provided a parameter list Non-COSMO: • INM (Spain): J. Garcia-Moya, J. Simarro, …

Task Suggested scientists (optional) % FTE per year start milestones date 0: Preliminary people external to the project 0 September 2005 - LM code to LM by December 2005 - INM initial and boundary conditions at ECMWF by June 2006 1: Definition of external collaboration of S. Theis 30% January 2006 - model parameters to be perturbed by December 2006 actions for the system set-up the methodologies to perturb model trajectories 2: Definition of the 40% October 2006 - ETKF implemented by the end of 2007 5% May 2006 50% June 2006 - data transfer by INM for the first tests - 1 -month test set-up by October 2006 - set-up of e-suite in 2007 - analysis of the impact of the different perturbations methodologies to perturb initial condition 3: System set-up 4: Test on case external collaborations of INM and U. Schaettler studies June 2008 June 2007 December 2006 5: Test on a 40% January 2007 December 2007 6: Verification 20% January 2007 October 2008 regular basis

ic and bc: multi-model approach v a MUlti-Model MUlti-Boundaries ensemble is operational at INM (Spain): 4 different limited-area models (HIRLAM, HRM, MM 5, UM) are nested on 5 different global models (IFS, GME, AVN, UM, LM) => 20 members ensemble (25 km horizontal resolution) v. LM has been added in May 2006 IFS LM 25 km GME LM 25 km AVN LM 25 km UM LM 25 km Courtesy of INM

IFS – ECMWF global LM at 25 km on IFS COSMO-SREPS (Short-Range Ensemble Prediction System) short range (up to 3 days) by INM Spain GME – DWD global LM at 25 km on GME P 1: conv. scheme (KF) P 2: turb. parameter 1 P 3: time filter UM – UKMO global LM at 25 km on UM AVN – NCEP global LM at 25 km on NCEP P 4: turb. parameter 2 16 LM runs at 10 km

Parameter list

13 15 UTC

47 N 44 N 6 W 9 W

13 18 UTC

47 N 45 N 7 W 10 W

14 00 UTC

46. 5 N 45 N 10. 5 W 13 W

14 06 UTC

COSMO-LEPS PROB. MAPS fc. range 0 -6 6 -12 12 -18 h 100 > 10 mm/6 h 90 COSMO-SREPS PROB. MAPS fc. range 12 -18 18 -24 24 -30 h 60 > 10 mm/6 h 30 10 13 May 12 -18 UTC

COSMO-SREPS fc. range 24 -30 ecmwf gme avn ukmo 14 May 0 -6 UTC

ensemble spread 12/05 13/05 12 UTC 00 UTC 12 UTC

member behaviour (“father”)

intra-group distance

Task 0. Preliminary actions for the system set-up Formal agreement between COSMO and INM: • use of LM by INM: support has been given to INM -> done • provision of the forecasts to COSMO: SREPS output available From May 2006 3 -hourly initial and boundary conditions for COSMO -SREPS are provided daily at ECMWF on ECFS: 4 LM runs at 25 km on an Euro-Atlantic area, nested on 4 different global models (IFS, GME, AVN, UKMO) interpolated with int 2 lm at 10 km over the COSMO-LEPS area

Task 1. Definition of the methodologies to perturb model trajectories • Schemes (physics and numerics): Some schemes have been selected to be perturbed: Tiedtke, Tiedtke with cape closure, Kain-Fritsch, 2 -time-level, Runge-Kutta • Parameters: • a first selection of the parameters has been done (analysis of few case studies) • the new tuneable parameters will be tested (provided by external collaborations of WG 3 people + Susanne Theis) • Surface forcings: LM surface fields -> interest of Luca Bonaventura to test a technique for the perturbation of surface parameters -> to be developed Task 2. Definition of the methodologies to perturb initial conditions: it will start in October with a visit to Bologna of Neill Bowler from UKMO.

Task 3. System set-up: • Data transfer between INM and ECMWF: done • Adaptation of source code to give the possibility to use tuneable or random parameters: the new model version model will include a new namelist file in which many physical parameters can be set (Uli Schaettler, WG 6) -> done • Extracting LM frames and nesting of LM on LM frames (WG 6): to be discussed • Discussion on computer resources: during 2006 SP SPITLAEF is being used. A new Special Project submitted to ECMWF for the developing phase for 2007 (SPITFEAR). Collection of COSMO units to be discussed. • Establish the link SIR-COSMO SREPS: this meeting • Possibility to have a COSMO originating centre in the GRIB coding: to be discussed (possible link with TIGGE)

Task 4. Test on case studies Test of SREPS prototype on real-time case studies and comparison with COSMO-LEPS and “model perturbed COSMO-LEPS”: • run of COSMO-SREPS for 1 case study to test that the chain from INM to COSMO-SREPS is working -> done • run of COSMO-SREPS with model perturbations only for 1 month of Autumn 2006: • statistical analysis of the system • provide b. c. for EELMK tests • comparison with COSMO-LEPS and PEPS

Plans for 2006 -2007 Task 5. Test on a regular basis • Task 5. 1 selection of the methodology to generate the perturbations : • parameters: based on the analysis of the results of the 1 -month test period • development of the i. c. perturbations technique • Task 5. 2 selection of the testing period: • possible link with MAP D-PHASE (summer-autumn 2007) • availability of observations (-> WG 5) • Task 5. 3 investigation on computer resources: • COSMO units available? • Task 5. 4 INM data availability

Plans for 2006 -2008 Task 6. Verification • Task 6. 1 selection of the verification methodologies (with WG 5) • Task 6. 2 selection of the reference data-set: COSMO observations, VERA analyses (MAP D-PHASE), … • Task 6. 3 computation of the score: comparison with COSMO-LEPS (and/or perturbed COSMO-LEPS? ) and PEPS + EELMK and SIR links