1/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
Assimilation of Satellite Soil Moisture Assimilation of Satellite Soil Moisture Data Products in NCEP GFS Data Products in NCEP GFS
W. Zheng1,2, X. Zhan3, J. Liu2,3, J. Meng1,2, J. Dong1,2, H. Wei1,2, & M. Ek1
1NOAA/NCEP/EMC, 5830 University Research Ct, College Park, MD
2IMSG, Kensington, MD3NOAA/NESDIS/STAR, 5830 University Research Ct, College Park, MD
2/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
ObjectiveObjective GFS and LIS-EnKF CouplingGFS and LIS-EnKF Coupling Embed EnKF in GFSEmbed EnKF in GFS 11stst Test for AMSR-E SM Test for AMSR-E SM Testing with SMOS SM Testing with SMOS SM Next StepNext Step
OUTLINEOUTLINE
3/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
““Online” soil moisture data Online” soil moisture data assimilation for GFSassimilation for GFS
Examine how SM data impact Examine how SM data impact GFS forecastsGFS forecasts
OBJECTIVESOBJECTIVES
4/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
Air Quality
WRF NMM/ARWWorkstation WRF
WRF: ARW, NMMETA, RSM GFS, Canadian Global Model
Satellites99.9%
Regional NAMWRF NMM
North American Ensemble Forecast System
Hurricane GFDLHWRF
GlobalForecastSystem
Dispersion
ARL/HYSPLIT
Forecast
Severe Weather
Rapid Updatefor Aviation
ClimateCFS
1.7B Obs/Day
Short-RangeEnsemble Forecast
MOM3
Noah Land Surface Model
Coupled
Global DataAssimilation
OceansRTOFS/HYCOM
WaveWatch III
NAM/CMAQ
NCEP Global Forecast SystemNCEP Global Forecast System
From Louis Uccellini (2009)
5/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
Topography,Soils
Land Cover, Vegetation Properties
Meteorological Forecasts,
Analyses, and/or Observations
Snow Soil MoistureTemperature
Land Surface Models
Data Assimilation Modules
Soil Moisture &
Temperature
EvaporationSensible Heat
Flux
Runoff
SnowpackProperties
Inputs OutputsPhysics Applications
Weather/Climate
Water Resources
HomelandSecurity
Military Ops
Natural Hazards
NASA Land Information SystemNASA Land Information System
From Christa Peters-Lidard (2007)
6/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
Ensemble Kalman Filter (EnKF)Ensemble Kalman Filter (EnKF)
yk
Nonlinearly propagates ensemble of model trajectories. Can account for wide range of model errors (incl. non-additive).Approx.: Ensemble size.
Linearized update.
xki state vector (eg soil moisture)
Pk state error covariance
Rk observation error covariance
Propagation tk-1 to tk:
xki+ = f(xk-1
i-) + wki
w = model error
Update at tk:
xki+ = xk
i- + Kk(yki - xk
i- ) for each ensemble member i=1…N
Kk = Pk (Pk + Rk)-1 with Pk computed from ensemble spread
From Rolf Reichle (2008)
7/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
EnKF for Noah LSM in GFSEnKF for Noah LSM in GFS
Nonlinearly propagates ensemble of model trajectories. Can account for wide range of model errors (incl. non-additive).Approx.: Ensemble size.
Linearized update.
xki state vector (eg soil moisture)
Pk state error covariance
Rk observation error covariance
Propagation tk-1 to tk:
xki+ = f(xk-1
i-) + wki
w = model error
For Noah LSM 4 layer SM: xj
i+ = xji- + ( i - xj
i- )* Pj1 / (P11 + R)
No matrix inversion. Scalars only
8/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
GFS and LIS-EnKF CouplingGFS and LIS-EnKF Coupling
LISLIS
GFSGFS
CouplerCoupler
NoahNoah
NoahNoah
EnKFEnKF
GFS & LIS CouplingGFS & LIS Coupling
Pros: Flexibility for more LSMs, 2D, 3D EnKF, Multivariable EnKF, etc.
Cons: Coding of the coupling system may require more time
9/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
Embed Simplified EnKF in GFSEmbed Simplified EnKF in GFS
GFSGFSNoahNoah
EnKFEnKF
EnKF Embedded in GFSEnKF Embedded in GFS
Pros: GFS can demonstrate SM impact on forecasts GFS may take advantage of satellite SM obs ASAP
Cons: Hardwiring limits more flexibility for assimilating other observational data
10/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
Preliminary Test with AMSR-E SMPreliminary Test with AMSR-E SM
Data: NESDIS AMSR-E daily soil moisture SM observation rate set to be 3% vol/vol Date: 2007 July 1-7
EnKF: Simplified for Noah LSM. Perturb SM state only GFS_CTL: GFS run without any EnKF SM data assimilation
GFS_EnKF: GFS run with the simplified EnKF
11/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
CONUS 24hr Total Rainfall day 5
forecast
12/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
Method: A Simple Ensemble Kalman Filter (EnKF) embedded in latest version of GFS latest version
Assimilation time period: 00Z May 1 – June 18, 2012. (GFS/GSI)
Experiments: CTL: Control run EnKF: Sensitivity run Perturbations:
Precipitation, 4 layer soil moisture states
Testing with SMOS SMTesting with SMOS SM
13/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
GFS_CTL
EnKF-CTL GFS_EnKF
SMOS
Comparison of soil moisture 18Z, 1-17 June 2010
14/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
SMOS GFS_CTL
GFS_EnKF EnKF-CTL
Comparison of soil moisture 18Z, 1-17 June 2010
15/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
GFS Top Layer SM Validation GFS Top Layer SM Validation With USDA-SCAN Measurements With USDA-SCAN Measurements
1-17 of June, 20121-17 of June, 2012
East CONUS (28 sites) West CONUS (25 sites) Whole CONUS
RMSE Bias Corr-Coef RMSE Bias Corr-
Coef RMSE Bias Corr-Coef
CTLCTL 0.149 0.015 0.458 0.122 0.049 0.488 0.136 0.031 0.472
EnKFEnKF 0.139 0.001 0.596 0.117 0.046 0.559 0.129 0.023 0.579
16/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
GFS Top Layer SM Validation GFS Top Layer SM Validation With USDA-SCAN Measurements With USDA-SCAN Measurements
1-17 of June, 20121-17 of June, 2012
17/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
GFS Top Layer SM Validation GFS Top Layer SM Validation With USDA-SCAN Measurements With USDA-SCAN Measurements
1-17 of June, 20121-17 of June, 2012
18/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
Surface skin Temperature 2 m temperature
Comparison of Tsfc, T2m 18Z, 1-17 June 2010
SMOS soil moisture assimilation generally decreased GFS surface temperature forecasts
19/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
Sensible Heat Flux Latent Heat Flux
Comparison of SHF and LHF 18Z, 1-17 June 2010
SMOS soil moisture assimilation increased GFS latent heat flux and decreased sensible heat flux estimates
20/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
ObsCTL: 12-36h
CTL: 36-60h
EnKF: 12-36h
EnKF: 36-60h Obs
Precipitation forecast 24h Accum (mm) Ending at 12Z 4 June 2012
SMOS soil moisture assimilation have observable impact on rainfall forecasts of GFS
21/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
EnKF: 60-84h
EnKF: 84-108h
CTL: 60-84h Obs
Obs
Precipitation forecast 24h Accum (mm) Ending at 12Z 4 June 2012
CTL: 84-108h SMOS soil moisture assimilation have observable
impact on rainfall forecasts of GFS
22/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
Assimilating SMOS in NCEP GFSImproved GFS deeper layer soil moisture
estimates comparing with in situ measurements reduced GFS temperature forecast biases
positively;increased latent heat and decreased sensible
heat fluxes for most CONUS regions;had significant impact on precipitation forecasts.
Results SummaryResults Summary
23/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
Implement semi-coupling of GFS and LIS;Implement semi-coupling of GFS and LIS;
Optimize model perturbation;Optimize model perturbation;
More testing with AMSR-E, SMOS, ASCAT and More testing with AMSR-E, SMOS, ASCAT and AMSR2 soil moisture data;AMSR2 soil moisture data;
More validation with weather observations. More validation with weather observations.
NEXT STEPNEXT STEP
24/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
GFS and LIS “Semi-Coupling”GFS and LIS “Semi-Coupling”
LISLIS
GFSGFS NoahNoah
NoahNoah
EnKFEnKF
ForcingForcing
StatesStates
GFSGFS NoahNoah
LISLISNoahNoah
EnKFEnKF
ForcingForcing
StatesStates
25/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
Implement semi-coupling of GFS and LIS;Implement semi-coupling of GFS and LIS;
Optimize model perturbation;Optimize model perturbation;
More testing with AMSR-E, SMOS, ASCAT and More testing with AMSR-E, SMOS, ASCAT and AMSR2 soil moisture data;AMSR2 soil moisture data;
More validation with weather observations. More validation with weather observations.
NEXT STEPNEXT STEP
26/2610th Joint Center for Satellite Data Assimilation Workshop, Oct. 10-12, 2012
Thanks …Thanks …