An Overview and Prospects for the Future Michele Rienecker Global Modeling and Assimilation Office NASAGSFC With many contributions borrowed from the Ocean Data Assimilation Community ID: 410711
Download Presentation The PPT/PDF document "Ocean Reanalyses-" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Ocean Reanalyses- An Overview and Prospects for the Future
Michele RieneckerGlobal Modeling and Assimilation OfficeNASA/GSFC
With many contributions borrowed from the Ocean Data Assimilation CommunityespeciallyCo-authors of Rienecker et al. OceanObs’09Stammer et al. OceanObs’09Smith et al., Mercator Ocean Newsletter, January 2010Masina et al., Mercator Ocean Newsletter, January 2010Heimbach and Ponte, CLIVAR Decadal Workshop, 2010Yan Xue, this workshopGuillaume Vernieres & Robin KovachMagdalena Balmaseda
US CLIVAR IESA Workshop
1Slide2
2Early ocean reanalysis drivers: seasonal forecasts (~1980 onwards) and decadal prediction (EU ENACT and follow-on ENSEMBLES project; CMIP5: 1960 onwards)GODAE and CLIVAR
: major advances through international cooperation, connections with the observationalists and with users – especially evident in the EUA variety of estimation methods:
Optimal interpolation (OI), asymptotic Kalman filters, 3DVar Ensemble methods – state-dependent error estimates Smoothers: 4Dvar; Kalman smoothersMany different models Surface fluxes from various sources: atmospheric reanalyses or RT NWP analysesUS efforts: CSFR (NCEP), GFDL, GMAO, SODA (UMD&TAMU), ECCO (JPL&MIT) The quality of the analyses depends on: the quality of the model the quality of the forcing the quality of the data and how we use them i.e., the error statistics used in the estimation machinery US CLIVAR IESA WorkshopSlide3
3 Uneven observational coverage in space and time Deep ocean and ice covered regions are poorly observed. OS in marginal seas is declining OS in coastal areas needs attention
However ….. Argo has had a dramatic impact on ocean analyses
The Observing System – one of the greatest challenges for ocean reanalysesUS CLIVAR IESA WorkshopGlobalNumber of temperature profiles per monthSlide4
US CLIVAR IESA Workshop4In situ profiles available for assimilation in January 2010Argo: 8853 XBT: 938 TAO: 1907
PIRATA: 265 Slide5
5Ocean Analysis Systems for Climate: global, 1/4° or coarser horizontal grid spacing comparisons of many metrics:
http://www.clivar.org/organization/gsop/gsop.php Many applications: monitoring climate - diagnostics of Earth’s climate variability, including unobserved quantities
initialization of climate forecasts observing system studies SODAECMWF
GECCO
SODA
ECMWF
ECMWF
GECCO
GECCO
SSH trends 1962-2001
*
Thermosteric (upper)
Halosteric (lower)
*
Note these analyses did not use corrected XBT data
From Stammer et al.
OceanObs’09 CWP
Atlantic MOC at 26N
⎯
ECMWF analysis
⎯
ECMWF – no data
assim
●
Bryden
et al. (2005)
●
Cunningham et al. (2007)
US CLIVAR IESA Workshop
From
Balmaseda
et al., GRL, 2007Slide6
6
Heimbach et al. OceanObs’09 CWP
Impact of SST, SSH, TSSSH, given SST, TS, BImpact of the observing system on MOC (Sv) in the ECCO systemRMS differences for 2006SST, given SSH, TS, BTS, given SSH, SST, B
ECMWF S3 (1-7 mo forecast)
Balmaseda & Anderson (2009)
US CLIVAR IESA Workshop
Impact of the observing system on
Seasonal SST Forecast Skill
Heimbach
& Ponte:
(1) Climatology only
poor constraint
- compare to
pre-1990’s era!
(2) Large impact
of altimetry
& Argo
(3) Complementary
nature of
Argo, altimetry
(4) Modest impact of
SSTSlide7
7 Argo: Biggest impact in less well-observed regions (Indian, S. Atlantic, S. Pacific, Southern Ocean)
Argo salinity also improves estimated temperature Smith and Haines (2009) T impact on S can differ from S data impact on S Balmaseda et al. (2007
)⇒ an issue for pre-Argo era ECMWF S3Impact of Argo on av. salinity in upper 300mBalmaseda et al., GRL
2007
US CLIVAR IESA Workshop
Impact of Argo T & S on analysis salinity (0-300m)
Impact of Argo T only on analysis salinity (0-300m)
ArgoSlide8
8From Stammer et al. OceanObs’09 CWP
North Atlantic heat contentUpper 700m
* Note these analyses did not use corrected XBT dataUS CLIVAR IESA Workshop(Obs)
From
Masina
et al., 2010
16
0
12
8
4
-4
-8
-12
2005
1960
1970
1980
1990
Global heat content
Upper 700m
T300
T300 Anomaly
70S – 70N
30N – 70N
From Yan
Xue’s
presentation
& Guillaume
Verniere’s
poster
Updated diagnostics - Upper ocean
Average T in upper 300m
Comparison across systems – uncertainty or discriminator?
A single system – uncertainty?Slide9
9From Stammer et al. CWP
Freshwater content
Upper 700mUS CLIVAR IESA WorkshopFrom Masina et al., 2010Slide10
US CLIVAR IESA Workshop10
GMAO ODAS-3
From Smith et al., 2010, Mercator Ocean NewsletterUsing Water masses as a validation metric: pdf of salinity mis-fits in S(T)Slide11
US CLIVAR IESA Workshop11Detection & Attribution of Atlantic salinity changes, P. Stott, R. Sutton, D. Smith, GRL, 2008Slide12
US CLIVAR IESA Workshop12Issues for ocean climate reanalyses: Error statistics – both model and observations (representation errors)
Reliable estimates of uncertainty in the analyses Understanding the analysis differences, esp. in modern “well-observed” era:
controlled experiments – same data, QC, forcing, …. Model and forcing biases – esp. important in the early periods Treatment of salinity in the pre-Argo era Freshwater fluxes – imbalance impacts SSH variability and trends as well as salinity SSH biases in the assimilation of altimetry Corrections to data bases (e.g., XBT, Argo) – need for continual updates to re-analyses & careful documentationSlide13
13Expected improvements in ocean state estimation in the next 10 years: Reduced model biases Increased model resolution (both horizontal and vertical)
Improved parameterizations Improvements in NWP analyses and re-analyses ⇒ improved forcing products Improved error covariance modeling – the basis of the estimation procedure
Improved RT and DM QC for input data streams Easy access to data bases with appropriate metadata ⇒ all data will be available to be assimilatedUS CLIVAR IESA WorkshopProspects for the Future:Slide14
14Emerging generation: “coupled” analyses or Integrated Earth System
Analyses State estimates consistent across components
Current efforts target improved initialization of climate forecastsSome current examples: 3DVar (NCEP) EnKF (GFDL, GMAO) MvOI (GMAO) Future: atmosphere-ocean-land-sea-ice-chemistry-biologyGFDL CDA - EnKF Impact on Forecast Niño-3 SST anomalies RMS as a fn of initial time and forecast lead
Courtesy Tony Rosati
Initial month
3Dvar
CDA
-
EnKF
Forecast lead (months)
1.0
1.8
0.0
1.0
1.8
0.0
J
F M A M J J A S O N D
J
F M A M J J A S O N D
5
11
1
9
7
3
5
11
1
9
7
3
US CLIVAR IESA Workshop
T300
T300 Anomaly
70S – 70N
30N – 70N
From Yan
Xue’s
presentation
& Guillaume
Verniere’s
poster
Updated diagnostics - Upper ocean
Average T in upper 300mSlide15
15GMAO: GEOS-5 “Coupled” Ocean Assimilation using Atmospheric Replay
Model predicted change
Replay Correction from Existing Analysis
Total “observed change”
09Z
12Z
15Z
18Z
00Z
21Z
03Z
06Z
09Z
Analysis
from
MERRA, ECMWF JRA-25, etc
Atmospheric Replay Background
Any existing atmospheric analysis
Assimilated atmos analysis
New ocean analysis
Coupled model forecast
“Initialized”
States for
Coupled forecast
03Z
06Z
Next coupled
analysis cycle
Coupled
analysis cycle
US CLIVAR IESA Workshop
NCEP: 1
st
guess for both ocean and atmosphere are from AOGCM integration
GFDL: assimilate NCEP reanalysis; “
All coupled components adjusted by observed data through instantaneously‐exchanged
fluxes”
Weakly coupled atmosphere-ocean assimilationSlide16
US CLIVAR IESA Workshop16Several examples of coupled ocean-sea-ice assimilation using co-variances estimated from EnKFLisæter
et al. Ocean Dynamics, 2003 Lisæter et al. JGR, 2007 –
CryoSat OSSECaya et al. JAOTech, 2010Adjoint investigation of sea-ice export sensitivities:Heimbach et al., Ocean Modelling, 2010Lisæter et al. JGR, 2007
Coupled Ocean-Sea-Ice Analyses
No Assimilation
Assimilation of synthetic
CryoSat
sea-ice concentration
Ice thickness
Ice thickness
salinity
salinitySlide17
μMmg m-3
NCEP Forcing2009
2009MERRA Forcing20092009Courtesy, Watson GreggUpper-ocean vertical fluxes are critically important to biological activity. Issues for assimilation: Assimilation perturbations to ocean physics Surface forcingOcean Biology
US CLIVAR IESA Workshop
17Slide18
Assimilated Global Annual Median Chlorophyllfor MODIS-Aqua mg m-3
Global Annual Median Chlorophyll for MODIS-Aqua
Courtesy, Watson GreggReduces sampling biasesfrom satellite in the annualdata18US CLIVAR IESA WorkshopSlide19
US CLIVAR IESA Workshop19Summary/Comments A lot of progress in ocean data assimilation!
For some metrics, current syntheses diverge in the modern “well-observed” era – somewhat surprising!
We need to understand these differences: controlled experiments – same data, QC, forcing, …. Missing reliable estimates of uncertainty in the analyses We need to track assimilation increment contributions to budgets – a measure of quality and of progress. There is a push for “coupled” atmosphere-ocean assimilation where observations from one medium constrain the state in the other. We need to make sure that we don’t ignore the issues still to be addressed in the uncoupled system: representation error treatment of biases ….Slide20
US CLIVAR IESA Workshop20Thank you for your attention!