1. Joint Proposal to WGOMD for a community ocean model experiment

2. Background
Seasonal Forecasts is an ideal test bed to asses the quality of the GCMS used in future climate projections:
Need ocean initial conditions to carry out experiments
Not all the groups have ocean data assimilation systems in place.
WGOMD is about to discuss experimental setup to asses ocean model performance in the representation of the inter-annual variability. Possibility of X-Panel interaction and feedback.
The proposed experiment could answer some outstanding questions about the key elements needed to improve the representation of the Pacific basin and ENSO.

3. Questions from the Pacific Panel
There are known model deficiencies in the representation of the Pacific climate:
Upwelling off the South American Coast
Cold tongue penetrating too far west
Too weak/too strong east-west slope of the thermocline
Equatorial heat content (IT/meridional transport, vertical mixing) …
Is it possible to say which are the dominant source of error?
Forcing fields?
Model parameterizations/configuration?
Model resolution?
Parameterization of air-sea interaction?

4. Proposal
Conduct a long ocean model simulation (20-40 years) with a variety of models, using inter-annually varying atmospheric reanalysis (ERA40-ENACT, NCEP-CORE, JMR- 25…). This long integration can be used as:
a CNTL experiment to answer some of the above questions (see later)
To provide initial conditions for concerted seasonal forecast experiments.
Conduct additional integrations with each model (optional), to test sensitivities:
Different forcing fields
Different resolution
Different parameterization, bulk formulae, ….
Need to agree the specifications for the CNTL integration.

5. Sensitivities: Compute and Compare
Sensitivity to Forcing fluxes:
Sensitivity to Resolution:
Sensitivity to Model:
Sensitivity to Bulk Formula:
Sensitivity to P-parameterization:

6. Sources of Uncertainty
All
ERA40
ERA40-No data
ERA40-Assim

7. Assimilation and Uncertainty T300
Long period :
Large part of uncertainty comes from forcing fluxes
EQPAC EQATL EQIND TRPAC
TRATL NPAC NATL GLOBAL
ALL
SAME FORCING (ERA40)
OCEAN MODEL
ASSIMILATION SCHEME

8. Sensitivities: preliminary results from ENACT/GSOP
MEAN STATE:
Equatorial Pacific heat content is model dependent
Slope of Pacific Eq. thermocline controlled by winds
ERA40 produces weak thermocline slope in most models
Too diffuse thermocline in Indian Ocean?
Analysis minus Observations
Western Pacific
Equatorial Indian
DATA ASSIM
NO DATA ASSIM
Mean Assimation Temperature Increment

9. Strawman for the CNTL integration
From spin up, launch the integrations, using daily atmospheric forcing fluxes plus some restoring terms:
3D relaxation to climatology of T and S (~10 year time scale)
Strong relaxation to time-varying values of SST (~2-3 days t.s)
Relaxation to climatological Surface Salinity (1 year t.s)
If free surface, volume preserving constrains.
The restoring terms are used to diagnose errors.
Atmospheric forcing (daily values):
Momentum flux: ERA wind stress
Heat flux:
ERA solar, latent and sensible heat fluxes. PLUS
Strong relaxation to daily values of SST (~2 days time scale) (Reynolds from 1982)
Fresh Water flux:
ERA PME + ENACT correction
Weak Relaxation to climatological SSS (~12 month time scale)
If free surface, some volume preserving constrains.

10. CNTL integration at ECWMF
Decent Inter-annual ENSO variability
Correlation with altimeter data ( )

11. CNTL integration at ECWMF
Some MOC variability

12. CNTL integration at ECWMF
As initial conditions, decent forecast skill.
S3 Nodata S3 Assim