Assimilation of TDR radial velocity using the unified GSI-based hybrid ensemble-variational data assimilation system for HWRF

GSI-based EnKF-Var hybrid data assimilation system: implementation and test for hurricane prediction

Xuguang Wang, Xu Lu, Yongzuo Li, Ting LeiUniversity of Oklahoma, Norman, OK

In collaboration with

Mingjing Tong , Vijay Tallapragada, Dave Parrish, Daryl KleistNCEP/EMC, College Park, MD

Jeff Whitaker, Henry WinterbottomNOAA/ESRL, Boulder, CO

1

2control forecast

GSI-ACV Wang 2010, MWR control analysis

data assimilationFirst guess forecastcontrol forecastEnsemble covarianceEnKF Whitaker et al. 2008, MWR EnKF analysis k

member 1 forecastmember 2 forecastmember k forecast EnKF analysis 2

EnKF analysis 1

member 1 forecastmember 2 forecastmember k forecastmember 1 analysismember 2 analysismember k analysisRe-center EnKF analysis ensemble to control analysisGSI-based Hybrid EnKF-Var DA system

Wang, Parrish, Kleist, Whitaker 2013, MWR3GSI 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

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 4DEnsVarGSI hybrid for GFS: 3DEnsVar vs. 4DEnsVar

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

6

GSI hybrid for GFS: 3DEnsVar vs. 4DEnsVar

16 named storms in Atlantic and Pacific basins during 20107

Approximation to nonlinear propagation

3h increment propagated by model integration 4DEnsVar (hrly pert.)4DEnsVar (2hrly pert.)3DEnsVar-3h 0 3h*timeHurricane Daniel 20103DEnsVar 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.

8Verification of hurricane track forecasts

9Development and research of GSI based Var/EnKF/hybrid for regional modeling system

GSI-based Var/EnKF/3D-4DHybridGFSHurricane-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.10

GSI hybrid for HWRFHurricane Sandy, Oct. 2012

Complicated evolution

Tremendous size

147 direct deaths across Atlantic Basin

US damage $50 billion

New York State before and afternhc.noaa.gov11

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

1112 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 1213

TDR data distribution (mission 1)

P3 Mission 11314

Verification against SFMR wind speed

Last Leg

15

Comparison with HRD radar wind analysis

16

Comparison with HRD radar wind analysis

SN

17

Track forecast (RMSE for 7 missions)

1718

Experiments for 2012-2013 seasonsCase#Correlation between HRD radar wind analysis and analyses from various DA methods

19

ISSAC 2012 (mission 7)

Verification against SFMR and flight level data

21

Experiments for 2012-2013 seasonTrack MSLP

2122

Two-way Dual Resolution Hybrid for HWRF3km movable nest ingests 9km HWRF EnKF ensemble Two-way couplingTests with IRENE 2011 assimilating airborne radar data

9km3km

Two-way Dual resolution hybrid

Summary and ongoing work24GFSGSI-based 4DEnsVar for GFS improved global forecast and TC forecast.The analysis generated by 4DEnsVar was more balanced than 3DEnsVar.the performance of 4DEnsVar was in general degraded when less frequent ensemble perturbations were used.The tangent linear normal mode constraint had positive impact for global forecast but negative impact for TC track forecasts.Preliminary tests showed positive impact of the temporal localization on the performance of 4DEnsVar.

HWRFThe GSI-based hybrid EnKF-Var data assimilation system was expanded to HWRF. 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.Airborne radar data improved TC structure analysis and forecast, TC track and intensity forecasts. Impact of the data depends on DA methods.Dual-resolution (3km-9km) two way hybrid for HWRF showed promising results.Developing/enhancing 4DEnsVar hybrid and assimilation of other airborne data and other data from NCEP operational data stream for HWRF.

Outer Domainassimilate operational conventional surface and mesonet observations, RAOB, wind profiler, ACARS, and satellite derived winds every 3 hours to define synoptic/mesoscale environment

12 km

25Johnson, Wang et al. 2014 Development and Research of GSI-based Var/EnKF/hybrid DA for Convective Scale Weather Forecasts over CONUS

Inner Domainassimilate velocity and reflectivity from NEXRAD radar network every 5 min during last 3hr cycle

Poster: Johnson, Wang, Lei, Carley, Wicker, Yussouf, Karstens4 km25Precipitation forecast skill averaged over 10 complex, convectively active casesGSI-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

May 8th 2003 OKC Tornadic Supercell

Ref and vorticity at 1 km27W and Vort. at 4 kmLei, Wang et al. 2014 1hr forecast from 22ZGSI hybridGSI hybrid28

DA cycling configuration (mission 1) 0000Z28Cold Start1800Z25Spin-up Forecast0200Z26Deterministic ForecastDA Cycle2200Z25OBSGSI3DVarSpin-up Forecast0000Z28Deterministic ForecastOBSHybrid1800Z25EnsembleSpin-up Forecast0000Z280200Z26Deterministic ForecastDA Cycle2200Z25OBSHWRF EnKFEnsemble Perturbation2829

DA cycling configuration (mission 1) Spin-up Forecast0000Z28Deterministic ForecastOBSEnsemble Perturbation0200Z262200Z25GFS ENSHybrid-GFSENS2930

30 GSI-3DEnsVar: Extended control variable (ECV) method implemented within GSI variational minimization (Wang 2010, MWR):Extra term associated with extended control variableExtra 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

EMBED Equation.3 _1442473026.unknown

_1442483431.unknown

stat 3DVAR static covariance; observation error covariance; ensemble size;

correlation matrix for ensemble covariance localization; kth ensemble perturbation;

3DVAR increment; total (hybrid) increment; innovation vector;

linearized observation operator; weighting coefficient for static covariance;

weighting coefficient for ensemble covariance; extended control variable._1242899166.unknown

_1242899307.unknown

_1242899363.unknown

_1242899712.unknown

_1242997707.unknown

_1242899327.unknown

_1242899242.unknown

_1242899268.unknown

_1242899196.unknown

_1242899098.unknown

_1242899125.unknown

_1242899058.unknown

Chart10.4255250.2456520.4555180.5101590.07700680.4038330.5568810.3503180.5975640.4736330.5845120.540420.7108220.5146580.7083730.7592910.1859690.6299550.6631630.3611390.5510210.7299410.6902760.7383120.7455550.6352710.6945370.7806380.4829360.6966810.6155910.5385450.3578160.8018560.4027670.7250490.5809930.4696060.6473850.7225960.6225180.7183020.7864680.5172260.7373580.824110.642460.6629020.480970.5152720.5575450.5510490.4485030.4536480.604310.5603090.6775660.7972880.3678990.4604010.5712560.1441340.5194650.7172290.3168130.628151

HybridGSI3DVARHybrid-GFSENS

hourly00.619441100.730321700.516009500.6725208-10.404978-10.5251511-10.4818875-10.3472888-20.6140794-20.7606046-20.536677-20.661033300.4255250.2456520.40839600.7806380.4829360.78279100.480970.5152720.4523200.604310.5603090.602113-10.5101590.07700680.427535-10.6155910.5385450.710225-10.5510490.4485030.621034-10.7972880.3678990.767187-20.5568810.3503180.564301-20.8018560.4027670.815457-20.5712560.1441340.556215-30.4736330.5845120.525795-30.5809930.4696060.62046-30.7172290.3168130.718618-40.7108220.5146580.674838-40.7225960.6225180.751963-50.7592910.1859690.790874-50.7864680.5172260.816433-60.6631630.3611390.695966-60.824110.642460.826903-70.7299410.6902760.767035-80.7455550.6352710.6719740.4736330.5845120.53533HybridGSI3DVAREnKFHybrid-GFSENSHybrid-6hourly10.4255250.2456520.4083960.4555180.587089EnKF20.5101590.07700680.4275350.4038330.3889940.40839630.5568810.3503180.5643010.5975640.6260340.43017740.4736330.5845120.5257950.540420.5339640.56430150.7108220.5146580.6748380.7083730.8024150.52579560.7592910.1859690.7908740.6299550.8100370.67483870.6631630.3611390.6959660.5510210.8075810.79087480.7299410.6902760.7670350.7383120.7850410.69596690.7455550.6352710.6719740.6945370.786330.767035100.7806380.4829360.7827910.6966810.839480.671974110.6155910.5385450.7102250.3578160.6699120.782791120.8018560.4027670.8154570.7250490.7736220.710225130.5809930.4696060.620460.6473850.6493110.815457140.7225960.6225180.7519630.7183020.7677610.62046150.7864680.5172260.8164330.7373580.7775960.751963160.824110.642460.8269030.6629020.7804610.816433170.480970.5152720.452320.5575450.7156720.826903180.5510490.4485030.6210340.4536480.618960.48279190.604310.5603090.6021130.6775660.2931570.765114200.7972880.3678990.7671870.4604010.712948210.5712560.1441340.5562150.5194650.6795580.6723051111220.7172290.3168130.7186180.6281510.7577370.587089 0.677694 0.7081250.388994 0.373247 0.3039080.626034 0.487795 0.6348030.533964 0.698106 0.5339640.802415 0.7569 0.7929050.810037 0.759163 0.8341250.807581 0.743512 0.7779360.785041 0.815162 0.828770.78633 0.762792 0.7468290.83948 0.83948 0.839480.669912 0.604599 0.7469660.773622 0.808948 0.8055270.649311 0.458239 0.697780.767761 0.725003 0.7576110.777596 0.719738 0.8246960.780461 0.735277 0.7922980.715672 0.653415 0.5753320.375058 0.749608 0.781796

hourly

HybridGSI3DVAREnKFHybrid-GFSENSHybrid-6hourly

gfs

HybridGSI3DVAREnKFHybrid-GFSENSHybrid-6hourly

once

HybridGSI3DVARHybrid-GFSENS

Sheet1

HybridGSI3DVARHybrid-GFSENS

00.571985900.668330800.430523500.42802100.7863100.391733-10.389299-10.417498-10.469314-20.593225-20.672389-30.36529-30.614611-40.676784-40.678448-50.650206-50.775066-60.534455-60.733994-70.761252-80.74934100.591059200.64935610)0.505596500.5713958-10.4113893-10.5810024-10.4818875-10.3472888-20.6356317-20.6979844-20.608717-29.97E+3600.4555180.2387490.52206800.6966810.5446860.78547200.5575450.5152720.4523200.6775660.560309-10.4038330.101740.511778-10.3578160.5385450.345439-10.4536480.4485030.765114-10.4604010.367899-20.5975640.3503180.564301-20.7250490.7319760.792533-20.5194650.144134-30.540420.5845120.474447-30.6473850.4696060.611534-30.6281510.316813-40.7083730.5146580.674838-40.7183020.6225180.72104-50.6299550.1859690.790874-50.7373580.5172260.80297-60.5510210.3554260.722076-60.6629020.642460.826903-70.7383120.6902760.767035-80.6945370.6808560.6932680.4555180.365290.786310.4038330.6767840.4174980.5975640.6502060.6723890.540420.5344550.6146110.7083730.7612520.7183020.6299550.7493410.7750660.5510210.786310.7339940.7383120.4174980.3917330.6945370.6723890.4536480.6146110.4280210.6784480.3892990.7750660.5932250.7339940.535330.3917330.6767840.4693140.6502060.534455-0.65129494440.7612520.749341

00.751163300.54536500.6808317-10.6987548-10.7015115-10.6749301-20.7806225-20.678564-20.684596100.83948 0.83948 0.8394800.715672 0.653415 0.57533200.587089 0.677694 0.708125-10.669912 0.604599 0.746966-10.375058 0.749608 0.781796-10.388994 0.373247 0.303908-20.773622 0.808948 0.805527-20.626034 0.487795 0.634803-30.649311 0.458239 0.69778-30.533964 0.698106 0.533964-40.767761 0.725003 0.757611-40.802415 0.7569 0.792905-50.777596 0.719738 0.824696-50.810037 0.759163 0.834125-60.780461 0.735277 0.792298-60.807581 0.743512 0.777936-70.785041 0.815162 0.82877-80.78633 0.762792 0.74682900.751163300.66731600.680831700.61085-10.6987548-10.691566-10.6749301-10.2705578-20.7806225-20.6523845-20.6845961-20.560944700.839480.7706670.84991900.7156720.6534150.57533200.5870890.6776940.69141900.2931570.3215120.292802-10.6699120.6045990.746966-10.618960.7297170.729437-10.3889940.3732470.303908-10.712948-0.08712190.611472-20.7736220.8089480.805527-20.6260340.4877950.634803-20.6795580.1842210.627119-30.6493110.4582390.69778-30.5339640.6981060.533964-30.7577370.663620.712386-40.7677610.7250030.757611-40.8024150.75690.792905-50.7775960.7197380.824696-50.8100370.7591630.834125-60.7804610.7352770.792298-60.8075810.7435120.777936-70.7850410.8151620.82877-80.786330.7627920.7468290.839480.7706670.8499190.6699120.6045990.7469660.7736220.8089480.8055270.6493110.4582390.697780.7677610.7250030.7576110.7775960.7197380.8246960.7804610.7352770.7922980.7156720.6534150.5753320.618960.7297170.7294370.5870890.6776940.7081250.3889940.3732470.3039080.6260340.4877950.6348030.5339640.6981060.5339640.8024150.75690.7929050.8100370.7591630.8341250.8075810.7435120.7779360.7850410.8151620.828770.786330.7627920.7468290.2931570.3215120.2928020.712948-0.08712190.6114720.6795580.1842210.6271190.7577370.663620.712386

_1258548698.unknown

_1258647986.unknown