Report about the meeting of the COSMO Steering Committee (STC) Moscow, 6 and 7 September 2010 Detlev Majewski, Head of STC Deutscher Wetterdienst, Offenbach, Germany Moscow, September 2010 COSMO, before and … Moscow, September 2010 after Moscow, September 2010 COSMO STC Meeting, 6 and 7 Sept. 2010 Participants: D. Majewski (Germany) T. Charantonis (Greece) M. Ferri (Italy) M. Ziemianski (Poland) R. Dumitrache (for G. Stancalie, Romania) D. Kiktev (Russia) P. Steiner (Switzerland) M. Arpagaus (SPM) For VERSUS (after lunch break) : A. Raspanti, U. Damrath, V. Stauch Moscow, September 2010 COSMO Science Plan 2010 - 2014 • Presentation by M. Arpagaus Moscow, September 2010 COSMO Software, additional software Current status; see COSMO contract COSMO model including nudging data assimilation Interpolation program int2lm Proposal by P. Steiner Additional programs: fieldextra, VERSUS Requirements for COSMO-Software: Portable, documented (internal + external), supported Procedures for the proposal of new COSMO-Software Moscow, September 2010 COSMO Licence Issues, Part I NCMS (United Arab Emirates) signed licence agreement in July 2010 Four more potential licence takers in the future: Brazil (INMET and DHN), Oman, Catalunya Support of licence takers (see Licence Contract): Installation of the COSMO software on customer‘s computer and second level support (by DWD only) Support of COSMO software for at least four years (all COSMO partners) How to use the licence fees? Further development of COSMO, e.g. mutual short-term visits in the framework of PPs, invitation of experts; external software contracts; participation at a conference with an oral presentation about COSMO related scientific results Moscow, September 2010 COSMO Licence Issues, Part II DWD asks for 50% of the licence fees for providing support to the customers and (usually) the initial/lateral boundary data based on analyses/forecasts of the global model GME. How to apply for support funded from the licence fees? 1-page proposal to the COSMO SPM and Head of STC detailing aim of activity, persons involved and costs forseen. SPM and STC will decide on a propsal after consultation with the WG-Leader(s). Moscow, September 2010 Current Users of HRM Moscow, September 2010 COSMO model for developing countries Goal: Provide a medium-term perspective for current HRM users in developing countries. Proposal: Provide an old cycle (question: How old?) of the COSMO model without support to users in developing countries; see http://data.worldbank.org/about/country-classifications/country-andlending-groups; here: Low-income economies ($995 or less) and Lowermiddle-income economies ($996 to $3,945) Decision of STC needed until end of 2010 because HRM has to be phased out until 2012 (GRIB2, ICON for lateral boundary values, grid spacing of 7 km or less) Head of STC will prepare letter to be send to COSMO Directors. From developing countries viewpoint: COSMO model or WRF (NOAA/NCAR) as replacement of HRM Moscow, September 2010 Status of PPs; New PP: POMPA; Status of VERSUS STC discussed the status of all PPs and PTs STC agreed to extend the following PPs: COLOBOC (by one year) CONSENS (by one year) STC agreed to create the new PP POMPA “Performance on massively parallel architectures” STC discussed the status and future development of PP VERSUS Moscow, September 2010 COSMO and important sport events Sochi 2014 (Winter Olympic) Brazil 2014 (Soccer World Cup) Brazil 2016 (Summer Olympics) Can the COSMO Partners offer some show pieces (FDPs, RDPs) to increase the visibility of COSMO? Head of STC will write a letter to the Directors of the COSMO Partners. Moscow, September 2010 FTE 2009/2010/2011 What to do if a COSMO member cannot provide 2 FTEs for two or even three years? Head of STC will ask the President of DWD to write a letter of concern to the Director of the COSMO Partner involved to ask for proposals of how to alleviate the situation. Missing FTEs should not be due to a lack of focus at the service because of running other regional NWP models! Moscow, September 2010 Communication issues How to improve communication within STC and with Head of STC? Jour fixe SPM and head of STC once a month (phone call) Once a year Head of STC and SPM will write a short (1-2 pages) paper about COSMO progress for Directors Moscow, September 2010 Next COSMO-GMs COSMO-GM 2011 from 12 – 16 September in Rome (Italy) COSMO-GM 2012 from 10 – 14 September in Lugano (Switzerland) Moscow, September 2010 Report of BRAC-Meeting 17 – 20 May 2010 Supetar, Brac Island, Croatia D. Majewski, DWD, Head of STC BRAC-HR BRainstorming on Advanced Concepts on High Resolution Modelling Workshop NWP at scales of a few hundreds of meters to a few kilometers Moscow, September 2010 Background • General Assembly of ALADIN partners 2009 in Istanbul • Problems seen in HARMONIE/AROME integrations at high resolutions detected by HIRLAM: - Fireworks (precipitation, cold pools, wind gusts) - Stability of integration • Concerns about scalability of HARMONIE on future massively parallel computer systems • Strategic planning process for the next 5 – 10 years • Two-stage process: Brainstorming meeting in Brac in May 2010 Scientific vision workshop in early 2011 Moscow, September 2010 Known deficiencies by Sander Tijm • • • • Daily cycle of convection Outflow (how to verify?) Low level clouds Strength of convection Moscow, September 2010 Daily cycle of convection by Sander Tijm Moscow, September 2010 Outflow: Temperature Cold pools as seen in the 2mtemperature Moscow, September 2010 Strength convection by Sander Tijm Moscow, September 2010 Impact of horizontal diffusion by Sander Tijm Moscow, September 2010 Future HPC challenges • New computer architectures like multi-/many-core CPUs and GPUs (Graphics Processing Unit like NVIDIA) • Massively parallel computer systems with > 100.000 cores. • Mixed OpenMP / MPI parallelization. • Portability of numerical model to different platforms. • Scalability of numerical model on > 10.000 cores. • Major challenges: Load balancing (e.g. physical processes) and efficient I/O on > 10.000 cores. • Stability of operational runs using 100.000 cores (with a potential small mean time between interrupts!). Moscow, September 2010 Scalability of COSMO-EU on 512, 1024 and 2048 cores of an IBM Blue Gene (PowerPC 450, 850 MHz) 24-h forecast; 665 x 657 gridpoints, 40 layers (852, 426, 213 gridpoints/core) Moscow, September 2010 Workshop in Brac, Croatia Organisers J. F. Geleyn (ALADIN), J. Onvlee (HIRLAM), D. Klaric (RC LACE), P. Termonia (Belgium), C. Fisher (France) Participants About 23 particpants from ALADIN, AROME, HIRLAM, LACE groups and D. Majewski (COSMO) T. Davies (UKMO) E. Kallen (ECMWF) Moscow, September 2010 Organisation of BRAC-HR Monday: Plenary with several key note lectures Tuesday and Wednesday: Working group meetings • WG1: Dynamical cores‘ development • WG2: Upper air physics • WG3: Physics – dynamics interface • WG4: Validation and verification • WG5: Nesting, LBC, coupling • WG6: Surface modelling Thursday: Reports of WGs and draft of standpoints Moscow, September 2010 Outcome of BRAC-HR • „Official“ 12+6 page report by the organizing committee (probably mainly written by J. F. Geleyn) dated 20/08/2010 • This report distinguishes between: - areas of broad consensus (6 pages) - areas of divergence of opions (5 ½ pages) - attempt of a synthesis for future steps (1 page) Moscow, September 2010 Areas of broad consensus • At high resolution, stochastic point of view more and more important • Increased sophistication for process description primarily for microphysics of precipitation and cloud-determination scheme(s) • Some 3D-effects of radiation like slope and shadow of clouds • Improvements of physics-dynamics interface needed • For verification: Radar and satellite data • Onset of (airmass) convection is one of the most fundamental problems to be solved Moscow, September 2010 Areas with divergence of opinions • Detailed priorities for the necessary evolutions of the diagnostic – validation – verification toolkit • Evolution or substitution of the current AROME spectral SI/SL dycore; two extreme solutions: Spectral SI/SL vs. gridpoint RK (like WRF, COSMO) • Relative priority to be asigned to the dynamical core issues (Météo France participants (F. Bouttier and C. Lac; see also paper by P. Benard and C. Lac) argue that the current dycore is the cause of the problems; J. F. Geleyn and many other participants (but not the COSMO representative) blame missing lateral mixing. Moscow, September 2010 Summary and outlook • Deep internal controversy between Météo France, ALADIN and HIRLAM about the cause of the problems of the current AROME and how to tackle them. • To shed some light on the problem: - Run Weisman/Klemp (1982) test with AROME and COSMO-Model. - Compare operational forecasts of AROME and COSMO for common areas (Southwestern Germany, Netherlands, Belgium). Moscow, September 2010 Orography (m) AROME (2.5 km) COSMO (2.8 km) Moscow, September 2010 Temperature (°C) at 2m and wind at 10m 12 July 2010 00 UTC +12h AROME (2.5 km) Moscow, September 2010 COSMO (2.8 km) Wind gusts (km/h) at 10m 12 July 2010 00 UTC +12h AROME (2.5 km) Moscow, September 2010 COSMO (2.8 km) Temperature (°C) at 2m and wind at 10m 12 July 2010 00 UTC +14h AROME (2.5 km) Moscow, September 2010 COSMO (2.8 km)