GSOP Synthesis Evaluation2

1
GSOP Synthesis Evaluation-2

• Progress since last meeting
• Update on previous inter-comparison and related
  issues
• Discussion of discrepancies in synthesis results
• Synthesis Application Global Sea Level Trends
• Status of coupled approaches (incl.
  initialization)
• Uncertainties in surface fluxes
• Bulk formula and drag coefficients
• Data errors/use of different datasets
• Discussion on error covariances
• Ocean reanlayses that span the 20th Century
• Review of data requirements for assimilation
• Review of climate index computations and GSOP Web
  page
• Discussion of synthesis capabilities for serving
  CLIVAR needs

2
GSOP Synthesis Evaluation-2
3
GSOP Synthesis Evaluation-2
4
GSOP Synthesis Evaluation-2

• Requirements of observing system, influence
  future developments at OceanObs09
• Indispensable measurements calibration needs
  for synthesis
• Altimetry (Jason 2) (calibration with tide
  gauges)
• ARGO ? deep sampling (deep hydrography for
  calibration)
• Scatterometer
• Mass/GRACE
• SST/SSS

5
GSOP Synthesis Evaluation-2

• Statement is needed addressing what are the
  CLIVAR data requirements to observations
  community, where we are, where do we go next.
• What is best practice eg for use of XBTs. Major
  issues are virtually unsampled deep ocean and
  calibration (esp salinity where problems even
  greater than SST due to changes in technology..).
  
• Statement which data not good enough to
  distinguish between the models and to distinguish
  trends. What is going on in subsurface Southern
  Ocean?
• Ed ARGO altimetry not good enoughwhat else is
  needed? An assessment of what we have on our
  hands and how this is changingnecessary
  considering IPCC process and attention this
  brings.

6
GSOP Synthesis Evaluation-2

• We need to request that ERA-INTERIM be made
  available!
• We need seasonal better run-off data sets. GFDL
  also has a run-off dataset with error bars care
  needed in how to apply this onto the ocean and
  can share experience
• A new group to define the error in surface flux
  products.
• Need to make a recommendation that correction
  needs to be ultimately applied by data centres,
  at least that there needs to be consistency and
  transparency in what correction is being applied.
• Link this dataset from GSOP webpage, with
  associated documentation, error estimates etc
• Put the result of Carls survey of GSOP webpage
• ACTION entrain Levitus, NODC for combining
  separate efforts to correct eg for XBT biases

7
Max strength of Atlantic MOC at 25N
from the ECMWF meeting in 2006, plot generated by
Armin Koehl
8
1994-2001 time-mean Atlantic MOC for contributing
groups this time
New contributions this time K-7, MERCATOR,
ECMWF (26N time series only)
9
MOC strength at 900 m (near the depth of MAX MOC
strength)
25N
10
How much does it matter?
Mean 1952
NCEP
wind stress
with NCEP wind
with NCEP wind
11
Climate Forecasts

• An ultimate goal of ocean data assimilation is to
  improve SI and decadal climate forecasts.

12
Further Advances in (ENSO) Prediction

• Model Improvements - reducing systematic errors
• Better Constraining Initial Conditions
• Particularly important in ocean because the
  memory of ENSO resides there.
• The importance of ocean subsurface data in making
  ENSO predictions has been demonstrated in a
  number of studies.
• Is the best ocean analysis/reanalysis the best
  initialization?

13
The Initial Condition Practices

• (Best) State Estimate
• Data Assimilation in the Separate Component
  Models
• Leads to initialization shocks.
• Coupled Model Climate ? Observed Climate
• Anomaly Initialization
• Improve model
• Coupled Modes of Coupled Model ? Observed
  Coupled Modes
• Initializing the Coupled Modes
• Identify ensemble perturbations
• Way out Do the Coupled Assimilation Problem

14
Intermediate StepGFDL Ensemble Kalman Filter
coupled data assimilation system

• Multi-variate analysis scheme
• Maintenance of the T-S relationship in Oceanic
  Data Assimilation (ODA)
• Maintenance of geostrophic balance in Atmospheric
  Data Assimilation (ADA)
• Analysis of oceanic states using 20th century
  ocean observational network
• ENSO variability
• Large Impact of CDAs initialization on ENSO
  forecasts!

15
Coupled Model Initialization of K7
(T. Awaij)
Improvement of predictability Good
initialization by 4D-VAR 4D-VAR approach
allows us to make initial conditions as close as
possible to the observed climate attractors for a
target phenomenon
16
Four-Dimensional Variational Coupled
Data Assimilation by K7

• Oceanic initial condition
• Bulk parameters controlling Air-sea fluxes of
• For Smaller-scale Parameterization
• we made pre-optimization using the Green
  function approach.

Momentum Sensible heat Latent heat
17
For IOD case,
Observed time series
DMI index
4DVAR CDA improved DMI index
Full assimilation
Assimilation only for parameter fitting
Assimilation only for initial value
Correction by both oceanic initial conditions and
parameters plays an important role
T. Awaji, 2007
1997
1998
18
Comparison between Adjusted bulk parameter values
and experimental results
Comparison with Charnocks experiment
Standard deviation
Adjusted values lie in reasonable range
adjusted mean
Charnocks experimental law
U 10 m/s
19
Initialization of a Global Climate Model with
Oceanic Reanalysis

• Recent studies about climate model initialization
  with oceanic data
• Pierce et al., 2004, Clim. Change (T, S
  anomalies)
• Collins et al., 2007, Phil. Trans. R. Soc.
  (summary)
• Smith et al., 2007, Science (T, S anomalies)
• Troccoli and Palmer, 2007, Phil. Trans. R. Soc.
  (T, S absolute)
• Keenlyside et al., 2007, submitted (SST
  anomalies)
• Pohlmann et al., 2008, to be submitted.

20
Reducing Coupling Shock

• Initialization schemes all suffer from the
  inconsistencies between the interaction of the
  model and initial conditions.(eg. the model winds
  along the eq. do not support the assimilation
  thermocline slope)
• To mitigate coupling shock coupled model
  initialization schemes have been developed using
  only anomalies

21
Pohlmann et al. Method

• Hindcast experiments
• The predictability is estimated with a large
  number of initial conditions to tribute to the
  uncertainties in the initial state.
• The hindcast results are compared to the
  assimilation experiment to evaluate the
  prediction skill in the system.
• T, S anomalies nudged - in all layers except the
  surface layer
• - time constant of 10 days
• - equatorward of 80ºN and 80ºS
• - except where sea ice was present 1952-2001
• Basic question
• Are the hindcast experiments closer to the
  assimilation experiment than the control run is?

22
Results
Anomaly correlation coefficient (ACC) for surface
air temperatures
Cotrol Assimilation
Hindcast year 1 Assimilation
Hindcast year 2-4 Assimilation
Hindcast Year 5-10 Assimilation
23
Conclusions

• The method of initializing a coupled model with
  oceanic reanalysis is applicable.
• The anomaly coupling scheme avoids drift in the
  hindcast experiments.
• Global and North Atlantic mean surface
  temperature and North Atlantic meridional
  overturning circulation show predictive skill
  over the pentadal to decadal time scales due to
  the oceanic initialization.
• The hindcast experiments are closer to the
  assimilation experiment than the control
  experiment without the initialization.

For further progress in SI and DEC we must
develop full coupled assimilation capabilities.