1 Weather forecast with WRF-EMSAfrica EPIKH Workshop, CNRST-Rabat, 16 June 2011 Weather forecast with WRF-EMS Weather and Climate Simulation using WRF Model on Moroccan Computing Grid Prof. Lahcen BAHI (Ecole Mohammadia d'Ingénieurs, Rabat, Morocco) Mr. Mohamed EL KHARRIM (Direction Régionale de la Météorologie au Nord, Rabat, Morocco)

2 WRF EMS features Complete, full-physics, numerical weather prediction (NWP) package that incorporates dynamical cores from both the National Center for Atmospheric Research (NCAR) Advanced Research WRF (ARW) and the National Center for Environmental Predictions' (NCEP) non-hydrostatic mesoscale model (NMM).

3 WRF EMS features All the capability of the NCEP and NCAR WRF packages are retained within the WRF EMS; complete and relatively easy to use numerical weather prediction (NWP) modeling package

4 WRF EMS features The installation, configuration, and execution of the cores have been simplified to encourage its use throughout the operational, private, and University forecasting and research communities.

5 WRF EMS features Every element of an operational NWP system has been integrated into the WRF EMS, including the acquisition and processing of initialization data, model execution, output data processing, and file migration and archiving.

6 WRF EMS features Tools for the display of forecast and simulation data are provided. Real-time forecasting operations are enhanced through the use of an automated process that integrates various fail-over options and the synchronous post processing and distribution of forecast files.

7 WRF EMS features The EMS supports the automated processing of forecast files concurrent with a model run. The user can view forecast fields while the model is running. Should a run fail, the WRF EMS will also send to the user(s) alerting of a problem.

8 WRF EMS features The system includes pre-compiled binaries optimized for 32- and 64-bit Linux systems running shared or distributed memory Linux environments. The MPICH2 executables are also included for running on local clusters across multiple workstations.

9 WRF EMS features The post processor supports a wide variety of display software including AWIPS, BUFSKIT, NCL, GrADS, GEMPAK, NAWIPS, and netCDF.

10 The need of Grid (Time consuming runs) High resolution Large domainLong range forecast (10 days) Runs with Data Assimilation

11 The need of Grid NWP guidance to NWS and River Forecast Centers (RFCs) at temporal and spatial scales (high resolution) not available from operational data sources Making a Global model Making 4 runs per day Re-modeling local forecast problems and historically significant weather events

12 WRF Modeling System

13 Sensitivity of a Simulation to Model Configuration (Case Study)Domain2: North of Morocco Grid resolution: 5 km Forecast length: 24h :00 to :00 Domain1: Morocco Grid resolution: 16 km Forecast length: 24h :00 to :00

14 Case Running Pre-processingems_prep --dset gfsptile --date cycle length 24 Running ems_run --option Post-processing ems_post –grads Visualizing

15 Results Model run time Domain2: North of Morocco (100x72)Grid resolution: 5 km Forecast length: 24h 3 h 9 min Domain1: Morocco (100x101) Grid resolution: 18 km Forecast length: 24h 54 min Platform: Intel CPU: Pentium D

16 Results Model run time Results obtained using the Eumedgrid infrastructure

17 WRF ARW Core Benchmark Model configuration for WRF EMS Benchmark simulation: Grid Dimensions : 5005 (77x65) Horizontal grid points x 45 vertical levels Forecast Length : 24 hours with 3 hour output frequency Grid Spacing : 15km Time Step : 90 seconds Cumulus Scheme : Kain-Fritsch

18 WRF ARW Core Benchmark ResultsCompiler Platform CPU Kernel Memory Time 2.1 PGI INTEL 1x 0.93 GhZ PIII _FC5 0.5 Gb 235 min 2x 0.80 GhZ PIII legacysmp 154 min 1x 2.00 GhZ P4 fc6 1.0 Gb 116 min 2x 1.70 GhZ Xeon ELsmp 64 min 2x 3.20 GhZ Xeon ELsmp 2.0 Gb 32 min 2x 2.80 GhZ Xeon fc5smp 33 min ELsmp 31 min 2x 3.20 GhZ Xeon 64-bit EL 4.0 Gb 29 min ELsmp 2x 3.06 GhZ Xeon ELsmp 8.0 Gb 21 min 1x 1.00 GhZ PIII fc6 236 min 243 min 170 min 1x 1.50 GhZ P4 1.5 Gb 160 min _FC4smp 156 min 118 min 1x 2.60 GhZ P4 legacy 87 min 80 min 1x 1.70 GhZ Xeon 71 min 40 min 42 min 2x 2.70 GhZ Xeon _FC4smp 41 min 1x 2.80 GhZ Xeon _FC4 57 min ELsmp AMD 2x 1.80 GhZ Opteron 244 2x 2.20 GhZ Opteron 248 _FC3smp 23 min 2x 2.00 GhZ Opteron 270 (Dual Core) _FC4smp 15 min 26 min

19 Thank You