LAM activities in Romania Raluca RADU National Meteorological Administration, Bucharest, Romania ALADIN model ALADIN model ( Alexandru S., Banciu D., Caian M., Ibanescu I., Radu R., Soci C., Stefanescu S.) Operational suite Aladin Model is integrated on a SUN E4500 platform (8-CPU 400GHz, 64 bytes architecture) twice per day over two different domains up to 48h forecast range. statistical adaptation (MOS, Kalman filter) verification: local and common ALADIN verification input data for the further applications: wave models (WAM, VAGROM) and for diffusion and transport of pollutants models (MEDIA). Application Wave models (Stefanescu S.) Aladin integrated over the Black Sea domain at 24 km resolution for 60 h, twice per day. 2m temperature and specific humidity, 10m wind speed and precipitation, evaporation and heat fluxes used to couple the basin scales and coastal circulation models During July 2005, a pre-operational forecasting experiment in ARENA project framework. The results of the POM model (Princeton Ocean wave Model) were integrated for the Romanian coastal zone Figure : The current velocity (a), sea temperature (b) and salinity (c) 24h forecast of the POM circulation model for the Romanian coastal zone valid on UTC High resolution Regional Model and Lokal Modell (Pescaru V. I., Velea L., Dumitrache R.) Operational suite HRM is integrated on SUN BLADE 1000 workstation at 14 km resolution. The non-hydrostatic Lokal Modell, implemented firstly on SGI ALTIX workstation and then running on Linux Cluster: multi processor optimization pre-operational running procedures, simulations at 7 km and 2.8 km proved its functionality. running twice per day up to 78 hour forecast for a domain covering Romania (16 –32 E, 42 – 51 N). Figure: Total precipitation HRM (top left), LM (bottom left) at 14 km and wind at 10m at 2.8 km (right) Tests with Local Modell and HRM at 14 km shows that LM improves the forecast, as for instance by reducing the precipitation area and amount closer to observations (Figure). The Lokal Modell was also implemented and tested for the 2.8 km horizontal resolution, using to that aim more representative domains over the Romanian territory. Analysis of the obtained results proved the LM’s ability to detect fine scale phenomena. Application CONVEX experiment (11-12 th of May 2005) (Barbu A., Pescaru V. I., Banciu D., Velea L., Dumitrache R., Radu R.) 11 – 12th of May 2005 international exercise CONVEX-3, simulation of a nuclear accident at the power plant in Cernavoda (Romania) Coordination by IAEA, NEA/OECD (Nuclear Energy Agency/Organization for Economic Co-operation and Development), OCHA (United Nation Office for the Co-ordination of Humanitarian Affairs), WHO (World Health Organization) and WMO (World Meteorological Organization). Figure : a) 48 h trajectory with the initial point (925 hPa), b) pollutant concentration for 15, 18, 21 and 24 hour forecast ranges by MEDIA model, c) pollutant concentration and wind 10m by INPUFF model NMA provided: meteorological forecast by ALADIN and HRM models, pollutant concentration forecast obtained by the coupled systems ALADIN– MEDIA models and HRM-INPUFF models 48 hours trajectories (issued from the trajectory model by using ALADIN forecasted wind). maps containing the distribution of the pollutant cloud and trajectories have been communicated to the decision factors and stored on the NMA web page outputs in accordance with those provided from Meteo- France and MetOffice. Research Spectral coupling (Radu R.) Spectral coupling (Radu R.) The impact of spectral coupling method for treating LBC on forecasted fields in ALADIN was studied comparing with the operational coupling. A daily data-basis over almost an year was realized. The validation of the spectral coupling method for ALADIN continued with aspects concerning its behaviour at finer resolution in local phenomena as the tornado case of Movilita The results correlated with observations reveals the capacity of different coupling methods to simulate the both phases of the frontogenesis. The operational coupling was able to detect better the first phase of frontogenesis developed at the nonstationarity of surface and PBL forcing, meanwhile only the spectral coupling catches the second phase (regeneration) which was born through the interaction mechanism between the large and the small scales. Figure: HRV Satellite image on (left) and vorticity spectra for ALADIN-NH (3.5km) using operational grid-points coupling (conturing line) and spectral coupling (dashed line) before (left) and after (right) the frontogenesis produces. Characteristics : regular grid, Lambert projection Δx=10 km (144x144 points) from October 2005 Δx=24 km (120x90 points) hybrid vertical coordinate 41 vertical levels dynamical adaptation mode ARPEGE supplies initial and boundary conditions 2TL Semilagrangian scheme with time step of 450s for 10km, 900s for 24 km time integration on 48h : 40 min for 10 km, 22 min for 24km post processing (every 3 s), standard output (grib format routed towards the visualization systems in Bucharest and to the Regional Meteorological Centers), graphical products (meteograms, pseudo – satellite images, hight of specific isotherms, stability indexes, etc); availablity on the ALADIN-romanian web site operational implementation of Aladin Diagpack, especially used for nowcasting (CAPE, MOCON, different instability indexes) Figure: a) Integration domains and orography: x=10 km (top) and b) x =24 km (bottom) Implementation of ALADIN model, Lokal Modell and MM5 model on Linux cluster Linux Cluster platform : 14 nodes x 2 processors Intel XEON, 2,66Mhz per node, 32 bytes - processors 28, RAM 30 GB, computing power ~ 155 Gflops (peak) - MPI INTERCONECT HP Gigabit Ethernet, Full Duplex (~ 2GB) - Administrative INTERCONET HP FAST ETHERNET Full Duplex (~200 MB) - OS: GNU/LINUX RH MPI MPICH Compilers: INTEL FORTRAN 8.2 and C, PORTLAND Group and C