GSI-based EnKF-Var hybrid data assimilation system:
implementation and test for hurricane prediction Xuguang Wang, Xu
Lu, Yongzuo Li, Ting Lei University of Oklahoma, Norman, OK In
collaboration with Mingjing Tong, Vijay Tallapragada, Dave Parrish,
Daryl Kleist NCEP/EMC, College Park, MD Jeff Whitaker, Henry
Winterbottom NOAA/ESRL, Boulder, CO 1 Slide 2 2 control forecast
GSI-ACV Wang 2010, MWR control analysis data assimilation First
guess forecast control forecast Ensemble covariance EnKF Whitaker
et al. 2008, MWR EnKF analysis k member 1 forecast member 2
forecast member k forecast EnKF analysis 2 EnKF analysis 1 member 1
forecast member 2 forecast member k forecast member 1 analysis
member 2 analysis member k analysis Re-center EnKF analysis
ensemble to control analysis GSI-based Hybrid EnKF-Var DA system
Wang, Parrish, Kleist, Whitaker 2013, MWR Slide 3 3 GSI hybrid for
GFS: GSI 3DVar vs. 3DEnsVar Hybrid vs. EnKF Wang, Parrish, Kleist
and Whitaker, MWR, 2013 3DEnsVar Hybrid was better than 3DVar due
to use of flow-dependent ensemble covariance 3DEnsVar was better
than EnKF due to the use of tangent linear normal mode balance
constraint Slide 4 4 GSI-4DEnsVar: Naturally extended from and
unified with GSI- based 3DEnsVar hybrid formula (Wang and Lei,
2014, MWR, in press). Add time dimension in 4DEnsVar GSI hybrid for
GFS: 3DEnsVar vs. 4DEnsVar Slide 5 5 Wang, X. and T. Lei, 2014:
GSI-based four dimensional ensemble-variational (4DEnsVar) data
assimilation: formulation and single resolution experiments with
real data for NCEP Global Forecast System. Mon. Wea. Rev., in
press. GSI hybrid for GFS: 3DEnsVar vs. 4DEnsVar Results from
Single Reso. Experiments 4DEnsVar improved general global forecasts
4DEnsVar improved the balance of the analysis Performance of
4DEnsVar degraded if less frequent ensemble perturbations used
4DEnsVar approximates nonlinear propagation better with more
frequent ensemble perturbations TLNMC improved global forecasts
Slide 6 6 GSI hybrid for GFS: 3DEnsVar vs. 4DEnsVar 16 named storms
in Atlantic and Pacific basins during 2010 Slide 7 7 Approximation
to nonlinear propagation 3h increment propagated by model
integration 4DEnsVar (hrly pert.) 4DEnsVar (2hrly pert.) 3DEnsVar
-3h 0 3h * time Hurricane Daniel 2010 Slide 8 3DEnsVar outperforms
GSI3DVar. 4DEnsVar is more accurate than 3DEnsVar after the 1-day
forecast lead time. Negative impact if using less number of time
levels of ensemble perturbations. Negative impact of TLNMC on TC
track forecasts. 8 Verification of hurricane track forecasts Slide
9 9 Development and research of GSI based Var/EnKF/hybrid for
regional modeling system GSI-based Var/EnKF/3D- 4DHybrid GFS
Hurricane- WRF (HWRF) WRF ARW WRF-NMMB Poster: Johnson et al.
Development and Research of GSI based Var/EnKF/hybrid Data
Assimilation for Convective Scale Weather Forecast over CONUS.
Slide 10 10 GSI hybrid for HWRF Hurricane Sandy, Oct. 2012
Complicated evolution Tremendous size 147 direct deaths across
Atlantic Basin US damage $50 billion New York State before and
after nhc.noaa.gov Slide 11 11 Sandy 2012 Model: HWRF Observations:
radial velocity from Tail Doppler Radar (TDR) onboard NOAA P3
aircraft Initial and LBC ensemble: GFS global hybrid DA system
Ensemble size: 40 Experiment Design Slide 12 12 Model: HWRF
Observations: radial velocity from Tail Doppler Radar (TDR) onboard
NOAA P3 aircraft Initial and LBC ensemble: GFS global hybrid DA
system Ensemble size: 40 Experiment Design Oper. HWRF Slide 13 13
TDR data distribution (mission 1) P3 Mission 1 Slide 14 14
Verification against SFMR wind speed Last Leg Slide 15 15
Comparison with HRD radar wind analysis Slide 16 16 Comparison with
HRD radar wind analysis SN Slide 17 17 Track forecast (RMSE for 7
missions) Slide 18 18 Experiments for 2012-2013 seasons Case#
Correlation between HRD radar wind analysis and analyses from
various DA methods Slide 19 19 ISSAC 2012 (mission 7) Slide 20
Verification against SFMR and flight level data Slide 21 21
Experiments for 2012-2013 season Track MSLP Slide 22 22 Two-way
Dual Resolution Hybrid for HWRF 3km movable nest ingests 9km HWRF
EnKF ensemble Two-way coupling Tests with IRENE 2011 assimilating
airborne radar data 9km 3km Slide 23 Two-way Dual resolution hybrid
Slide 24 Summary and ongoing work 24 GFS a.GSI-based 4DEnsVar for
GFS improved global forecast and TC forecast. b.The analysis
generated by 4DEnsVar was more balanced than 3DEnsVar. c.the
performance of 4DEnsVar was in general degraded when less frequent
ensemble perturbations were used. d.The tangent linear normal mode
constraint had positive impact for global forecast but negative
impact for TC track forecasts. e.Preliminary tests showed positive
impact of the temporal localization on the performance of 4DEnsVar.
HWRF a.The GSI-based hybrid EnKF-Var data assimilation system was
expanded to HWRF. b.Various diagnostics and verifications suggested
this unified GSI hybrid DA system provided more skillful TC
analysis and forecasts than GSI 3DVar and than HWRF GSI hybrid
ingesting GFS ensemble. c.Airborne radar data improved TC structure
analysis and forecast, TC track and intensity forecasts. Impact of
the data depends on DA methods. d.Dual-resolution (3km-9km) two way
hybrid for HWRF showed promising results. e.Developing/enhancing
4DEnsVar hybrid and assimilation of other airborne data and other
data from NCEP operational data stream for HWRF. Slide 25 Outer
Domain assimilate operational conventional surface and mesonet
observations, RAOB, wind profiler, ACARS, and satellite derived
winds every 3 hours to define synoptic/mesoscale environment 12 km
25 Johnson, Wang et al. 2014 Development and Research of GSI-based
Var/EnKF/hybrid DA for Convective Scale Weather Forecasts over
CONUS Inner Domain assimilate velocity and reflectivity from NEXRAD
radar network every 5 min during last 3hr cycle Poster: Johnson,
Wang, Lei, Carley, Wicker, Yussouf, Karstens 4 km Slide 26
Precipitation forecast skill averaged over 10 complex, convectively
active cases GSI-EnKF forecasts are more skillful than GSI-3DVar
forecasts for all thresholds and lead times. Benefits of radar data
are more pronounced assimilated by GSI-EnKF than GSI-3DVar. 26
Slide 27 May 8 th 2003 OKC Tornadic Supercell Ref and vorticity at
1 km 27 W and Vort. at 4 km Lei, Wang et al. 2014 1hr forecast from
22Z GSI hybrid Slide 28 28 DA cycling configuration (mission 1)
0000Z28 Cold Start 1800Z25 Spin-up Forecast 0200Z26 Deterministic
Forecast DA Cycle 2200Z25 OBS GSI3DVar Spin-up Forecast 0000Z28
Deterministic Forecast OBS Hybrid 1800Z25 Ensemble Spin-up Forecast
0000Z28 0200Z26 Deterministic Forecast DA Cycle 2200Z25 OBS HWRF
EnKF Ensemble Perturbation Slide 29 29 DA cycling configuration
(mission 1) Spin-up Forecast 0000Z28 Deterministic Forecast OBS
Ensemble Perturbation 0200Z26 2200Z25 GFS ENS Hybrid-GFSENS Slide
30 30 GSI-3DEnsVar: Extended control variable (ECV) method
implemented within GSI variational minimization (Wang 2010, MWR):
Extra term associated with extended control variable Extra
increment associated with ensemble (4D)EnKF: ensemble square root
filter interfaced with GSI observation operator (Whitaker et al.
2008) GSI-based Hybrid EnKF-Var DA system
