1. ALADIN & LACE recent and ongoing Development on LAMEPS Yong Wang ZAMG, Austria With contribution from Bellus, Hagel, Horanyi, Ivatek-Sahdan, Kann, Kertesz, Mladek, Radu, Tascu, Wittmann, Wimmer etc.

2. Overview of LAMEPS research activities in ALADIN countries LAMEPS operations in ALADIN LAMEPS R & D in ALADIN International collaboration Summary and future plan

3. Overview of LAMEPS research in ALADIN/LACE Austria: ETKF/ET, Breeding, Blending, downscaling, post-processing, LBC and physics perturbation Blegium: downscaling Croatia: downscaling Czech: downscaling, ALADIN SV Hungary: downscaling, ALADIN SV Portugal: downscaling Romania: multi-model EPS, downscaling, LBC and physics uncertainties

4. LAMEPS operations in ALADIN/LACE ALADIN-LAEF (Limited Area EnsembleForecasting) Since Mar. 2007 in pre-operation  ALADIN CY31  Hydrostatic, spectral, η in vertical  18km horizontal, 37 levels  SLSI, DFI, 600s time step  Bougeault deep convection, Lopez microphysics, ISBA, RRTM radiation scheme....  Forecast integration up to 54h

5. ALADIN-LAEF  LAEF ensemble size: 22 members  16 x perturbed members: ECMWF singular vector perturbation  1 x downscaling of ECMWF-EPS control  1 x downscaling of ECMWF deterministic  4 x ALADIN-AUSTRIA runs (00, 06,12,18) 22 Members  2 runs per day (00 und 12 UTC)  Products available: ca. 2.5h after ECMWF-EPS Operational procedure: ECMWF EPS Data from MARS  901  927  001  ftp to ZAMG  FA to GRIB1 conversion  products and visualization  ftp to LACE webpage

6. ALADIN-LAEF  Every 3 hours interval, from 6 to 54 hours forecast  0.15 degree Lat/Lon grid Mean/spread/probability/post stamp: T2m, 10m Wind, Precipitation, Mslp, EPS meteogram T2m, 10m Wind, Precipitation, Cloudiness 2m RH, wind gust, CAPE, convective cloudniss Mean/spread/spaghetti/probability/post stamp chart: 250, 500, 700, 850, 925hPa, T, Wind, Geopotential, RH (lower level), 1000-850hPa, 1000-500hPa, 850-700hPa thickness Mean/spread/spaghetti/probability: CAPE, Lifted Index, cloud water?  LACE domain for products

7. ALADIN-LAEF

8. LAMEPS R&D in ALADIN - ALADIN native SV - Blending SV and Breeding, ETKF/ET - LBC and physics perturbation - Downscaling - Verification

9. ALADIN SV (by Hagel, Horanyi and Mlaedk) 1.ALADIN SV computation has been implemented at ECMWF and Meteo-France. 2.Technical issues (CPU and memory) have been analysed. 3.Sensitivity tests with respect to horizontal resolution (22km and 44km for GLAMEPS domain) and to optimization time (12h and 24h) have been carried out, in which the optimization area is the LACE one. 4.There are also efforts on study of technical feasibility of semi- Lagrangian SV computation.

10. ALADIN native SV 44km, optimisation: 12h and LACE domain

11. ALADIN native SV 44km, optimisation: 24h and LACE domain

12. Blending SV and Breeding ALADIN-LAEF: Blending global SV with ALADIN LAEF Breeding To combine the large-scale uncertainty from global SV-EPS with the small-scale uncertainty generated by Breeding/ET in LAEF. It is expected that 1). reducing the inconsistency between global and limited area EPS. 2). combining the future uncertainty generated by SV and the uncertainty in the past generated by Breeding. Method: spectral analysis and digital filter.

13. Blending

14. Breeding, ETKF and ET 1.Similar results with Breeding ETKF and ET have been shown in the 10 members LAEF experiments. 2.New experiment with 18 members EPS, and longer test period (more than 2 sommer months) is going on. 3.Combination of Breeding, ETKF and ET with the multi-physics schemes of ALADIN has been designed. 4.The research results (if there is an improvement!) will be introduced into ALADIN LAEF.

15. LBC and physics uncertainties 1.The research on LAEF Breeding/ET coupling with ARPEGE SV and NCEP EPS Breeding/ET, and on the use of multi- physics scheme have shown encouraging results. Those will be introduced into the operational ALADIN-LAEF 2.Surface perturbation in LAEF is in consideration. 3.Work with different coupling techniques for dealing with the uncertainty on LBC is going on (Radu). 4.The research results (if there is an improvement!) will be introduced into ALADIN LAEF.

16. Downscaling Downscaling of ARPEGE EPS (by Mladek) 1.Downscaling of PEARP for the cases with which there are big problem in precipitation forecast. Finding the more added value of ALADIN downscaling of PERAP for those difficult situation 2.Results show that there is no real more added value with ALADIN downscaling of PEARP 3.The best result is the combination of PEARP and ALADIN downscaling. Downscaling of ECMWF EPS Publication on ECMWF Tech. Note by Ivatek-Sahdan, Brankovic and Matjacic. Efforts have been done also in HU, PO and BE.

17. Downscaling

18. Verification (by Mladek, Kann and Hagel) A common verification package has been developed. Namelist controlled

19. Common Verification Package Aim: Simple, stand-alone, easy use and portable, universal

20. LAEF Verification: upper air

21. LAEF Verification: precipitation

22. LAEF Verification: T2m

23. International Collaboration  MAP D-Phase  GLAMEPS  B08RDP  THORPEX/TIGGE-LAM

24. WMO/WWRP FDP: MAP D-PHASE Provided by Felix Ament, MeteoSwiss

25. MAP D-PHASE Provided by Felix Ament, MeteoSwiss

26. WMO/WWRP: B08RDP B08RDP: Beijing 2008 Olympics Mesoscale Ensemble Prediction Research and Demonstration Project Partners: NCAR, NCEP, MSC, JMA, CMA, ZAMG/Meteo-France acting as a single partner.

27. WMO/WWRP: B08RDP B08RDP is a five years research project 2005-2009, each participant will provide their LAMEPS products in real time to Beijing Olympics 08 for demonstration. LAMEPS tier 1: 15km, 30 levels, 10 members, 6-36h forecast, domain 3500x3000km LAMEPS tier 2: 2-4km, 30-60 levels, 1-2 members, 0-36h forecast, domian 1320x1100km ALADIN-LAEF is used for regional EPS action with resolution 15km, 37 levels: August 2007 for Test and in August 2008 for Olympic game AROME will be used for meso-scale EPS with very high resolution 2-4km.

28. WMO/WWRP: B08RDP

29. Verification Bias scores RR6h ETS RR6h

30. Bias correction and combination: T2m RMS error (2007.8.14-2007.8.24) left – before BC right –after BC RMS errors of all EPS after calibration have been decreased obviously. Diurnal changes have been lowered after calibration. Combined-Ensemble RMS error after calibration is the smallest of all, before calibration, it is the smallest in most cases. After calibration spread is more closer to RMS errors than before.

31.  The LAMEPS application in ALADIN LAEF has been put into pre-operation since Mar. 2007.  Optimation of LAEF is needed, in particular, some near surface parameters due to the coupling between ALADIN and ECMWF model  Many LAMEPS R&D activities in ALADIN, like ALADIN-SV, Blending, ETKF/ET, multi-physics, downscaling, multi-model EPS, verification, etc.  ALADIN LAMEPS colleagues are also active on different international collaborations. Summary

32. Future plan Improvment and optimization of ALADIN LAEF Continuing the research on SV, Blending, ETKF/ET, downscaling, clustering, Post-processing, surface perturbation, verification, etc. Demonstration and exchange the research result with international LAMEPS community (GLAMEPS, B08RDP, MAP D-PHASE, TIGGE-LAM).

33. Acknowledgment Thanks very much to all the ALADIN Colleagues, who has contributed the ALADIN LAMEPS work, in particular, Bazile, Bellus, Bouttier, Fischer, Hagel, Horanyi, Ivatek- Sahdan, Kann, Kertesz, Mladek, Nicolau, Radu, Tascu, Wittmann, Wimmer, etc. Thanks also the COSMO colleagues for help with EPS clustering, e.g. Paccagnella, Montani, Marsigli and so on. Thanks also the colleagues in CMA, NCEP, NRL, MSC, who has helped the EPS work at ZAMG.