Huang et al: MTG-IRS OSSEMMT, June MTG-IRS OSSE on regional scales Xiang-Yu Huang, Hongli Wang, Yongsheng Chen and Xin Zhang National Center for Atmospheric Research, Boulder, Colorado, U.S.A. Stephen A. Tjemkes, Rolf Stuhlmann EUMETSAT, Darmstadt, Germany

Huang et al: MTG-IRS OSSEMMT, June Contents The OSSE configuration The nature run Simulated (conventional and MTG) observations Calibration experiments Data assimilation and forecast results 4D-Var results Summary

Huang et al: MTG-IRS OSSEMMT, June The OSSE Configuration Nature run model: MM5 Assimilation model: WRF Data assimilation scheme: WRF 3D-Var (4D-Var) Selected cases: –International H2O Project (IHOP, 13 May - 25 June 2002) –Southern Great Plains of US

Huang et al: MTG-IRS OSSEMMT, June OSSE setup

Huang et al: MTG-IRS OSSEMMT, June IHOP Case Three convection cases are selected from 11 June 2002 to 16 June 2002: 11 June: Dryline and Storm 12 June: Dryline and Storm 15 June: Severe MCS Map illustrating the operational instrumentation within the IHOP_2002 domain (Weckwerth et al. 2004)

Huang et al: MTG-IRS OSSEMMT, June Nature run configuration Nature model: MM5 Grid points: 505X505X35 Horizontal resolution: 4Km Time step: 20s Physics parameterizations: –Reisner 2 microphysics –No cumulus parameterization –MRF boundary layer Initial and Lateral boundary condition: –6-hourly ETA model 40-km analyses (phase I: GFS analyses) ~ 220 minutes with 256 CPUs model domain

Huang et al: MTG-IRS OSSEMMT, June observed 6-h rainfall simulated 6-h rainfall The observation is on Grid. The observation is on Polar Stereographic Projection Grid. The simulated rainfall is on Grid. The simulated rainfall is on Lambert Projection Grid. The color scales are different UTC 12 Jun Case A: 11 June Case

Huang et al: MTG-IRS OSSEMMT, June observed 6h-rainfall simulated 6h-rainfall 0600 UTC 13 Jun Case B: 12 June Case

Huang et al: MTG-IRS OSSEMMT, June Case C: 15 June Case observed 6h-rainfall simulated 6h-rainfall 0000 UTC 16 Jun

Huang et al: MTG-IRS OSSEMMT, June Simulated Dataset WRF-Var is employed to produce simulated conventional observations (NCEP ADP Upper Air sounding, Surface Observation, AIRCAR/AIRCFT and Satellite wind data ) –Simulated observations use the actual locations and times –Use realistic observation errors

Huang et al: MTG-IRS OSSEMMT, June Simulated Dataset NCEP ADP Upper Air sounding NCEP ADP Surface Observation 501 Surface33 Sounding Example of Simulated Data distribution within the time window: 2300 UTC 11 June to 0100 UTC 12 June 2002

Huang et al: MTG-IRS OSSEMMT, June Simulated Dataset NCEP ADP AIRCAR/CRAFT NCEP ADP Satellite wind 748 ACAR/CFT Example of Simulated Data distribution within the time window: 2300 UTC 11 June to 0100 UTC 12 June SATOB

Huang et al: MTG-IRS OSSEMMT, June MTG-IRS retrieved profiles MTG-IRS RP Example of Simulated Data distribution within the time window: 2300 UTC 11 June to 0100 UTC 12 June 2002

Huang et al: MTG-IRS OSSEMMT, June T q Error statistics for MTG-IRS

Huang et al: MTG-IRS OSSEMMT, June Calibration runs Generate pseudo radiosonde observations from the nature run. Exp 1. No-obs. Exp 2. Pseudo-obs. Data assimilation experiment using the pseudo observations. Exp 3. Real-obs. Data assimilation experiment using real (radiosonde) observations.

Huang et al: MTG-IRS OSSEMMT, June SOP: Simulated Observation Profiles; ROP: Real Observation Profiles Difference in T (K), 12 km results, averaged over 1800 UTC 11 to 1200 UTC 15 June 2002 At analysis time At 12 h FCST

Huang et al: MTG-IRS OSSEMMT, June Difference in q (g/kg), 12 km results, averaged over 1800 UTC 11 to 1200 UTC 15 June 2002 At analysis timeAt 12 h FCST SOP: Simulated Observation Profiles; ROP: Real Observation Profiles

Huang et al: MTG-IRS OSSEMMT, June Difference in u (m/s), 12 km results, averaged over 1800 UTC 11 to 1200 UTC 15 June 2002 At analysis timeAt 12 h FCST

Huang et al: MTG-IRS OSSEMMT, June Experiments design (3D-Var) Forecast model: WRF V3.0 Grid points: 169X169X35 Horizontal resolution: 12 Km Time step: 60s Physics parameterizations: WSM6 microphysics Grell cumulus paramerization MRF boundary layer Cases: Z to Z Data: MTG-IRS retrieved profiles + SOP Verification against truth

Huang et al: MTG-IRS OSSEMMT, June Lists of Experiments

Huang et al: MTG-IRS OSSEMMT, June RMSE: 6hourly cycling results Analysis Time12 h FCST

Huang et al: MTG-IRS OSSEMMT, June RMSE: 6hourly & 1hourly cycling results Analysis Time12 h FCST

Huang et al: MTG-IRS OSSEMMT, June Averaged ETS and BIAS at 18 h FCST ETSBIAS

Huang et al: MTG-IRS OSSEMMT, June Forecast model: WRF Data assimilation system: WRF 4D-Var Grid points: 169X169X23 Horizontal resolution: 12Km Time step: 60s Assimilation window: 3 hours Physics parameterizations: – WSM6 microphysics – New Grell cumulus paramerization – MRF boundary layer Case A: Z to Z Case B: Z to Z Background: GFS analysis and forecasts Data: Retrieved profiles + SOP or “Truth” Verification against truth Experiments design (4D-Var) D1

Huang et al: MTG-IRS OSSEMMT, June RMS error of Case A 12 h FCST

Huang et al: MTG-IRS OSSEMMT, June Case A at 18H FCST ETS BIAS

Huang et al: MTG-IRS OSSEMMT, June RMS error of Case B 12 h FCST

Huang et al: MTG-IRS OSSEMMT, June Case B at 18H FCST ETS BIAS

Huang et al: MTG-IRS OSSEMMT, June Summary Three storms are well reproduced in the 5 day nature run. The calibration experiment shows that the real and simulated observations have the similar impacts on the analyses increments and forecasts differences. The quality of the retrievals has been improved significantly. The forecast skill of wind, temperature and moisture forecasts is improved when MTG-IRS T and/or q profiles are assimilated. Rainfall forecast skill is improved also. In two short periods of experiments, assimilating MTG-IRS data using 4D-Var gives slightly better forecast skills over 3D- Var.

Huang et al: MTG-IRS OSSEMMT, June Current and planned work Continue the 4D-Var experiments Other nature runs Try other cases –Europe –Tropics –Over oceans –… Assimilate wind data