GFS Model activities in India (NCMRWF) Saji Mohandas V S Prasad G R Iyengar R G Ashrit Surya Kanti Dutta M Das Gupta E N Rajagopal National Centre for.

GFS Model activities in India (NCMRWF) Saji Mohandas V S Prasad G R Iyengar R G Ashrit Surya Kanti Dutta M Das Gupta E N Rajagopal National Centre for Medium Range Weather Forecasting Global Spectral models at NCMRWF ModelYearForecast Range R40L days T80L days T170L days T170L days T254L days T382L days T574L days Recent developments in Global Forecast System Implementation of the T382L64 GFS from May 2010 (latest versions of upgraded model and GSI) Assimilation of additional data in T382L64 GFS The Advanced Microwave Scanning Radiometer - Earth Observing System (AMSR-E) winds Rainfall rates (TRMM, SSMI) NOAA19 radiances Atmospheric Infrared Sounder (AIRS) radiances GPSRO (COSMIC) Evaluation of the T382L64 GFS for Monsoon-2010 Implementation of the T574L64 GFS from mid-November, 2010 (July 2010 Version) Data Pre-processing In T80L18 system ECMWF decoders were used in data pre-processing and then data was packed them in prepqm format. From T254L64 onwards a complete NCEP data pre-processing system was implemented. Presently this data pre-processing system even linked UKMO OBSTORE data processing (for conventional observations only). New Observations Apart from the observations that are used in the earlier system the following new observations are being assimilated. 1. Precipitation rates from SSM/I & TRMM 2. GPSRO occultation 3. AIRS and AMSRE radiances 4. MODIS winds Atmospheric profiles from radio occultation data using GPS satellites GPSRO 1bamua, 1bamub, 1bmhs,1bhirs3, 1bhirs4,MHSNESDIS/POES ATOVS sounding and radiance data IASI,AIRS,AMSR-E brightness temperaturesNASA/AQUA and EUMETSAT/METOP/IASI brightness temperature data NASA/TRMM (Tropical Rainfall Measuring Mission) and SSM/I precip. rates Spssmi, SPTRMM NESDIS/POES, METOP-2 and AURA orbital ozone data sbuv ERS-2 wind, ASCAT winds from METOP-A satellite, Winsat winds from Coriolis satellite Scatwind (presently not going at NCMRWF) AMV from Meteosat-7, Meteosat-9, GOES-11, GOES-13, MTSAT-1R, MODIS (TERRA and AQUA), satwind AIREP, AMDAR, TAMDAR, ACARSAircar/aircft TEMP (land and marine), PILOT (land and marine), Dropsonde, Wind profiler ADPUPA Land surface, Mobile, Ship, Buoy (SYNOPs)ADPSFC Types of Observations Assimilated in GFS DAF system Conventional datasets Satellite datasets GPSRO RADIANCE Data Coverage AMV AWS Observations Data Coverage New GFS model-Highlights Higher resolution and modified science options Implementation on IBM Power 6 Directory structure suitable for easy portability Improved Initial conditions (GSI) and more non-conventional observations New post processors and more diagnostics Improved performance Directory Structure $HOME gfs nwprodnwdatanwworknwplot GFS Overview Dynamics : Spectral, Hybrid sigma-p, Reduced Gaussian grid Time integration : Leapfrog/Semi-implicit Time filter : Asselin Horizontal diffusion : 8th order wavenumber dependent Orography : Mean orography Surface fluxes : Monin-obhukov Similarity Turbulent fluxes : Non-local closure SW Radiation : RRTM LW Radiation : RRTM Deep Convection : SAS Shallow convection : Mass-flux based Grid-scale condensation : Zhao Microphysics Land Surface Processes : 4-layer NOAH LSM Cloud generation : Xu and Randal Rainfall evaporation : Kessler Air-sea interaction : Roughness length by Charnock Gravity Wave Drag : Based on Alpert Sea-Ice model : Based on Winton Major Changes Resolution and model parameters -T382L64/T574L64 Vertical coordinate system (Hybrid sigma-p) Upgraded ESMF library (esmf v 3.1.0rp2) Restructured GFS code (namelist options, DIAG3D and GOC3D o/p) Modified dynamics/physics (thru namelist) Post processing options (postgp/nceppost) 40 deg. North90 deg. East Saha et al., 2010 Resolutions T382L64 (~35Km) May x574 grid Timestep:180 sec CPU: 3min/24hr (IBM-P6 24 node x 8 proc) Physics: minor changes from T254L64 T574L64 (~23Km) November x880 grid Timestep:120 sec CPU: 15min/24hr (IBM-P6 16 node x 32 proc) Completely modified physics package Modified Physics Radiation and clouds Gravity wave drag Planetary boundary Layer Shallow convection Deep convection Tracer transport scheme Post Processing Options gfs_post_txxx_6hrly.sh/JGFS_nceppost.sh POSTGP (old program) - Fast computation, Smoothed fields NCEPPOST (Unified post processor) - Consists of CHGRES and NCEPPOST steps - CHGRES is computationally expensive thread level application - NCEPPOST is MPI code - More accurate algorithm for some diagnostics - Much more number of output fields Output Files Global model outputs: SF, BF and FLX DIAG3D and GOCART (Optional) Post processed outputs: PGB (Pressure level 3D atmospheric fields) Analysis and Forecast Verification (GEOP) SH CORR COEFF RSME (TEMP) SH CORR COEFF RSME RMSE (VECTOR WIND500) SH RMSE (VECTOR WIND500) NH Using NCEP VSDB verification Verification Against Mean Analysis Anomaly Correlation WIND500 Case Studies 1.Monsoon Depression (30 Aug 2010) 2.Tropical Cyclone JAL (06 Nov 2010) 3.Western Disturbance (8 Feb 2011) 4.Easterly Wave (2 Feb 2011) 5.False alarm (28 March 2011) MD (30 Aug 2010) T382L64 MD (30 Aug 2010) T254L64 MD (30 Aug 2010) T382L64 T254L64 MD (30 Aug 2010) T382L64 MD (30 Aug 2010) T254L64 MD (30 Aug 2010) T382L64 T254L64 TC (06 Nov 2010) T382L64 TC (06 Nov 2010) T254L64 TC (06 Nov 2010) T382L64 T254L64 TC (06 Nov 2010) T382L64 TC (06 Nov 2010) T254L64 TC (06 Nov 2010) T382L64 T254L64 WD (08 Feb 2011) T574L64 WD (08 Feb 2011) T382L64 WD (08 Feb 2011) T254L64 T574L64 T382L64 T254L64 WD (08 Feb 2011) T574L64 WD (08 Feb 2011) T382L64 WD (08 Feb 2011) T254L64 T574L64 T382L64 T254L64 abc d ef EW (02 Feb 2011) T574L64 EW (02 Feb 2011) T382L64 EW (02 Feb 2011) T254L64 T574L64 T382L64 T254L64 EW (02 Feb 2011) T574L64 EW (02 Feb 2011) T382L64 EW (02 Feb 2011) T254L64 T574L64 T382L64 T254L64 FA (28 Mar 2011) T574L64 FA (28 Mar 2011) T382L64 FA (28 Mar 2011) T254L64 T574L64 T254L64 T382L64 FA (28 Mar 2011) T574L64 FA (28 Mar 2011) T382L64 FA (28 Mar 2011) T254L64 T574L64 T254L64 T382L64 T254L64 50 km T382L64 35 km Evaluation of the Global Forecast Systems for Monsoon 2010 Model Systematic Errors 850 hPa WINDS(M/S) SYSERR T254 T382 700 hPa WINDS(M/S) SYSERR T254 T382 500 hPa WINDS(M/S) SYSERR T254 T382 200 hPa WINDS(M/S) SYSERR T254 T382 850 hPa TEMP(K) SYSERR T254 T382 200hPa TEMP(K) SYSERR T254 T382 850 hPa SP HUM (G/KG) SYSERR T254 T382 ANALYSIS 850 hPa ZONAL WIND (M/S) (60E-70E) T254 T382 ANALYSIS 850 hPa ZONAL WIND (M/S) (75E-80E) RMSE Day 1Day 3Day 5 T382 T hPa Zonal wind RMSE 200 hPa Zonal wind RMSE Day 1Day 3Day 5 T382 T254 850 hPa Meridonal wind RMSE Day 1Day 3Day 5 T382 T254 200 hPa Meridonal wind RMSE Day 1Day 3Day 5 T382 T254 850 hPa Temperature RMSE Day 1Day 3Day 5 T382 T254 200 hPa Temperature RMSE Day 1Day 3Day 5 T382 T254 850 hPa RH RMSE Day 1 Day 3Day 5 T382 T254 Mean RMSE u(m/s) PL T254 T382 UKMO Time series of RMSE of day-5 forecasts of Zonal Wind Mean RMSE v(m/s) PL T254 T382 UKMO Time series of RMSE of day-5 forecasts of Meridional Wind Mean RMSE T(K) PL T254 T382 UKMO Time series of RMSE of day-5 forecasts of Temperature For a quantitative rainfall forecast verification, the IMD's 0.5 daily rainfall analysis (Rajeevan and Bhate 2008, Rajeevan et al 2005) is used. Mean Error Equitable threat score Evaluation of rainfall forecasts Day 1Day 3Day 5 T382 T254 Mean Error The difference between the forecast mean and Observed rainfall Equitable Threat Score for forecast of rainy days during JJAS 2010 Day 1Day 3Day 5 T254 T382 Equitable Threat Score All India Rainfall (JJAS2010) T254 T382 All India Rainfall (Monsoon-2010) Seasonal, Monthly and Weekly* T254T382 *Weekly rainfall is accumulated from forecast with single IC once in every week Verification of NCMRWF operational global model Day 3 FCST 850hPa Wind against RS/RW over Indian Region (Jan Dec 2010) Summary T382L64 and T574L64 systems were implemented successfully on IBM-P6 The performance is better than T254L64 system T382L64 is predicting anomalously strong circulation features in some cases which are absent in T574L64 system