Post on 29-Aug-2018
transcript
Soil moisture and temperature assimilation into the
GEOS-5 land surface model
Clara Draper, Rolf Reichle, Gabrielle de Lannoy, and Qing Liu
Global Modeling and Assimilation Office, NASA Goddard Space Flight Center,and University Space Research Association
October 11, 2012
Draper NASA GMAO
Soil moisture/temperature assimilation 1 / 20
Outline
1. Assimilation of passive and active microwave C/X-bandnear-surface soil moisture retrievals
◮ Improve model profile soil moisture
2. Calibration of microwave radiative transfer model◮ Enable direct assimilation of L-band brightness temperature
observations, to improve model profile soil moisture and surfacesoil temperature
3. Assimilation of GOES skin temperature retrievals◮ Improve surface turbulent fluxes◮ Enhance assimilation of surface-sensitive radiances in GEOS-5
ADAS
Draper NASA GMAO
Soil moisture/temperature assimilation 2 / 20
1. Assimilation of passive and active microwave C/X-band near-surfacesoil moisture
More details: Draper et al (2012), GRL
Draper NASA GMAO
Soil moisture/temperature assimilation 3 / 20
Outline
◮ Compare assimilation ofnear-surface soil moisture frompassive (AMSR-E, LPRM, X-band)and active (ASCAT) microwavesensors into the Catchment model(GEOS-5 LSM) forced withMERRA atmospheric fields
◮ Assimilate with an EnKF from Jan.2007 - May 2010
◮ Remove model-observation bias byCDF-matching the observations
◮ Evaluate against SCAN/SNOTEL& Murrumbidgee Soil MoistureMonitoring Network in situobservations
Schaefer et al (2007), Young et al (2008), Friedl et al (2002)
Draper NASA GMAO
Soil moisture/temperature assimilation 4 / 20
Assimilation skill by land cover classSkill: anomaly correlation with in situ observations
Mix.(26) Grass.(32) Crop.(27) All (85)
0.4
0.5
0.6
R (
surf
ace)
Mean R with 95% confidence intervals: surface
Mix.(26) Grass.(32) Crop.(27) All (85)
0.4
0.5
0.6
R (
root
−zon
e)
Mean R with 95% confidence intervals: root−zone
Mix.(26) Grass.(32) Crop.(27) All (85)OPEN DA ASCAT DA AMSR−E DA BOTH
◮ Mean root-zone R over all sites:OPEN 0.45, DA ASCAT 0.55, DA AMSR-E 0.54, DA BOTH 0.56
Draper NASA GMAO
Soil moisture/temperature assimilation 5 / 20
Contribution of observation skill to assimilation skill
0.2 0.4 0.6 0.8
0.2
0.4
0.6
0.8
Ope
n−lo
op s
kill
(R)
(sur
face
)
Observation skill (R) (ASCAT or AMSR−E)
Assimilation skill (R)for surface soil moisture
0.2 0.4 0.6 0.8
0.2
0.4
0.6
0.8
Ope
n−lo
op s
kill
(R)
(roo
t−zo
ne)
Observation skill (R) (ASCAT or AMSR−E)
Assimilation skill (R)for root−zone soil moisture
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.2 0.4 0.6 0.8
0.2
0.4
0.6
0.8
Ope
n−lo
op s
kill
(R)
(sur
face
)
Observation skill (R) (ASCAT or AMSR−E)
Assimilation skill improvement overopen−loop for surface soil moisture
0.2 0.4 0.6 0.8
0.2
0.4
0.6
0.8
Ope
n−lo
op s
kill
(R)
(roo
t−zo
ne)
Observation skill (R) (ASCAT or AMSR−E)
Assimilation skill improvement overopen−loop for root−zone soil moisture
−0.08
0
0.08
0.16
0.24
0.32
0.4
◮ Based on assimilation ofASCAT or AMSR-E
◮ Confirms results fromsynthetic experiments ofReichle et al (2008)
◮ If (obs skill − open-loopskill) > −0.2, assimilationimproved the model skill
Draper NASA GMAO
Soil moisture/temperature assimilation 6 / 20
Soil moisture assimilation summary
◮ Significant improvements to mean root-zone and near-surface soilmoisture model skill from assimilation of ASCAT and/or AMSR-Enear-surface soil moisture retrievals
◮ At individual sites observation skill must be substantially worsethan model skill for assimilation to degrade the model soilmoisture skill
◮ Recommend assimilation of both passive (AMSR-E, AMSR2) andactive (ASCAT) near-surface soil moisture
Draper NASA GMAO
Soil moisture/temperature assimilation 7 / 20
2. Calibration of microwave radiative transfer model
More details: De Lannoy et al (submitted), JHM
Draper NASA GMAO
Soil moisture/temperature assimilation 8 / 20
Radiative transfer model calibration
◮ Calibrate radiative transfer modelparameters to reduce large biasesbetween Catch./RTM and observedL-band brightness temperatures(TB)
◮ Use L-band TB from ESA’s SMOSmission (launched 2009) inpreparation for NASA’s SMAPmission (scheduled 2014)
◮ Optimization of objective functionmeasuring difference in long-termmean and standard deviation, anddistance from prior
◮ Calibrate over 2010, validate over2011
mean (Catch./RTM minus SMOS) TB , 2011
−50 −40 −30 −20 −10 0 10 20 30 40 50 [K]
Draper NASA GMAO
Soil moisture/temperature assimilation 9 / 20
Reduction in bias from calibrationMean (Catch./RTM - SMOS) Tb, 2011 (H-pol, 42.5o, asc.)
SMAP
L-MEB with ECMWF-SMOS roughness
L-MEB
Calibrated
−50 −40 −30 −20 −10 0 10 20 30 40 50 [K]
Best results: calibrate roughness, scattering albedo, and veg. optical depth
Draper NASA GMAO
Soil moisture/temperature assimilation 10 / 20
Remaining biases
Catch./RTM minus SMOS TB , ascending H-pol, all angles
−10 −5 0 5 10 [K]
Draper NASA GMAO
Soil moisture/temperature assimilation 11 / 20
Summary
◮ Calibration has greatly reduced the (very large) model-SMOSbiases, allowing direct assimilation of L-band radiances (includingSMAP)
◮ Remaining biases, due to both SMOS instrument calibration andCatch./RTM biases, are being addressed within assimilation
Draper NASA GMAO
Soil moisture/temperature assimilation 12 / 20
3. Assimilation of GOES skin temperature retrievals
Draper NASA GMAO
Soil moisture/temperature assimilation 13 / 20
Outline
◮ EnKF assimilation of GOES-E/W skin temperature (Tskin) overNorth America, for JJA 2012
◮ Assign model-observation bias to the observations using adynamic observation bias correction scheme
◮ Bias estimates based on model-observation difference over previous5-10 days
◮ Evaluate impact by comparison to twice-daily MODIS Tskin
observations
Draper NASA GMAO
Soil moisture/temperature assimilation 14 / 20
GOES Tskin data
◮ Global high resolution Tskin product, provided by NASA LangleyResearch Center
◮ Early results suggest comparable accuracy to MODIS◮ Currently available 3-hourly (cloud-free) at 0.25◦ resolution
Tskin observations per day (JJA 2012)
Scarino et al (submitted)
Draper NASA GMAO
Soil moisture/temperature assimilation 15 / 20
Daytime results (18:00 UTC)RMSD between model/GOES and MODIS Tskin, after removing 3-month bias
ubRMSD OPENLOOP (mean: 3.7 K)
ubRMSD GOES bias corrected to model (mean: 3.6 K)
ubRMSD GOES (mean: 2.6 K)
ubRMSD OPENLOOP - ASSIM. (mean: 0.15 K, 67% +ve)
Draper NASA GMAO
Soil moisture/temperature assimilation 16 / 20
Nighttime results (06:00 UTC)RMSD between model/GOES and MODIS Tskin, after removing 3-month bias
ubRMSD OPENLOOP (mean: 2.2 K)
ubRMSD GOES bias corrected to model (mean: 1.9 K)
ubRMSD GOES (mean: 1.3 K)
ubRMSD OPENLOOP - ASSIM. (mean: 0.13 K, 80% +ve)
Draper NASA GMAO
Soil moisture/temperature assimilation 17 / 20
Tskin assimilation summary
◮ GOES offers Tskin observations with high spatial resolution andtemporal frequency
◮ Offline assimilation of GOES Tskin brings model closer to MODISTskin
◮ Next: assimilate GOES Tskin data into GEOS-5 atmosphericDAS/model, test impact on assimilation of atmosphericobservations
Draper NASA GMAO
Soil moisture/temperature assimilation 18 / 20
Implementing the land data assimilation in GEOS-5
03z 06z 09z 12z 15z 18z 21z
ATMOS
ANALYSISIAU
CORRECTOR
PREDICTOR
IAU
ATMOS FORCING
09z 12z 15z
b
b
b
b
LDAS
b
b
b
b
L-IAU
LAND SURFACE UPDATES
Draper NASA GMAO
Soil moisture/temperature assimilation 19 / 20
THANK YOU FOR LISTENING.
◮ Further details: clara.draper@nasa.gov
◮ MORE DETAILS⊲ De Lannoy, G., Reichle, R., Pauwels, V. (submitted), Global Calibration of the GEOS-5 L-band MicrowaveRadiative Transfer Model over Land Using SMOS Observations, J. Hydromet.
⊲ Draper, C., R. Reichle, G. De Lannoy, and Q. Liu (2012), Assimilation of passive and active soil moistureretrievals, Geophys. Res. Lett., 39, L04401.
◮ REFERENCES⊲ de Jeu, R., and M. Owe (2003), Further validation of a new methodology for surface moisture and vegetationoptical depth retrieval, Int. J. Remote Sens., 24, 4559–4578.⊲ Dorigo, W., K. Scipal, R. Parinussa, Y. Liu, W. Wagner, R. de Jeu, and V. Naeimi (2010), Errorcharacterisation of global active and passive microwave soil moisture datasets, Hydrol. Earth Syst. Sc., 14,2605–2616.⊲ Friedl and coauthors (2002), Global land cover mapping from MODIS: algorithms and early results, Remote
Sens. Environ., 83, 287–302.⊲ Reichle, R., W. Crow, R. Koster, H. Sharif, and S. Mahanama (2008), Contribution of soil moisture retrievalsto land data assimilation products, Geophys. Res. Lett., 35, L01404.⊲ Scarino, B., Minnis, P., Palikonda, R.,Reichle, R., Morstad, D., Yost, C., Shan, B., and Liu, Q. (submitted),Retrieving surface skin temperature for NWP applications from global geostationary satellite data, Rem. Sens..⊲ Schaefer, G., M. Cosh, and T. Jackson (2007), The USDA Natural Resources Conservation Service Soil ClimateAnalysis Network (SCAN), J. Atmos. Oceanic Technol., 24, 2073–2077.⊲ Wagner, W., G. Lemoine, and H. Rott (1999), A method for estimating soil moisture from ERS scatterometerand soil data, Remote Sens. Environ., 70, 191–207.⊲ Young, R., J. Walker, N. Yeoh, A. Smith, K. Ellett, O. Merlin, and A. Western (2008), Soil Moisture andMeteorological Observations From the Murrumbidgee Catchment, Department of Civil and EnvironmentalEngineering, The University of Melbourne.
Draper NASA GMAO
Soil moisture/temperature assimilation 20 / 20
Remotely sensed near-surface soil moisture data
◮ AMSR-E: LPRM X-band (38 km resolution, depth < 1cm)
◮ ASCAT: C-band (25 km resolution, ∼ 1cm depth)
◮ Both scaled into Catchment climatology using CDF-matching
0.2 0.4 0.6 0.8
Mix. Cover Grassland Cropland
◮ ASCAT skill significantly lower fortopographic complexity > 10%(crosses): data discarded
◮ Otherwise skill of ASCAT andAMSR-E is broadly similar(skill is anomaly correlation with insitu observations)
de Jeu and Owe (2003), Wagner et al (1999)
Draper NASA GMAO
Soil moisture/temperature assimilation 21 / 20
Remaining biasesMean (Catch./RTM - SMOS) Tb (all angles)
Ascending H-pol
Descending H-pol
Ascending V-pol
Descending V-pol
−10 −5 0 5 10 [K]
Draper NASA GMAO
Soil moisture/temperature assimilation 22 / 20
Dynamic observation bias correction
Dynamically correct the observations to remove the model-observationbiasx−(t) = M(x+(t − 1))x+(t) = x−(t) + K [Hx−(t) − (yo(t)) + Hbo−(t))]
bo− = bo+(t − 1)bo+(t) = bo−(t) + λ[(Hx+(t) − yo(t)) − Hbo−(t)]λ = (1 − e−∆t/τ )
◮ ∆t is time since last observation
◮ τ is time scale of bias memory (5 days)
◮ Separate bias model for each time of day
Draper NASA GMAO
Soil moisture/temperature assimilation 23 / 20
TSURF in Catchment model◮ TSURF is blackbody radiative temperature, controlled by balance
of surface fluxes
∆Z(m)
0.1
0.2
0.4
0.8
1.5
10.0
Diffusive
heat flux
RN
LH + SH
TSURF = w(TCAN , TSOILSURF )
Surface specific heat capacity:
200 J/K
(70,000 J/K for broad-leaf)
dWdt
= RN − LH − SH − G
G
Draper NASA GMAO
Soil moisture/temperature assimilation 24 / 20