the i nverse r egional o cean m odeling s ystem
DESCRIPTION
The I nverse R egional O cean M odeling S ystem Development and Application to Variational Data Assimilation of Coastal Mesoscale Eddies. Di Lorenzo, E. Georgia Institute of Technology Moore, A. UC Santa Cruz Arango, H. Rutgers University Cornuelle, B and A.J. Miller - PowerPoint PPT PresentationTRANSCRIPT
The Inverse Regional Ocean Modeling System
Development and Application to Variational Data Assimilation of Coastal Mesoscale Eddies.
Di Lorenzo, E.Georgia Institute of Technology
Moore, A.UC Santa Cruz
Arango, H.Rutgers University
Cornuelle, B and A.J. MillerScripps Institution of Oceanography
Chua B. and A. BennettOregon State University
Australia
Asia
USA
Canada
Pacific Model Grid SSHa
(Feb. 1998)
(source: modeling team Rutgers, UCLA, GaTech, Scripps)
Regional Ocean Modeling System (ROMS)
Inverse Regional Ocean Modeling System (IROMS)
Chua and Bennett (2001)
Inverse Ocean Modeling System (IOMs)
Moore et al. (2003)
NL-ROMS, TL-ROMS, REP-ROMS, AD-ROMS
To implement a representer-based generalized inverse method to solve weak constraint data assimilation problems
a representer-based 4D-variational data assimilation system for high-resolution basin-wide and coastal oceanic flows
Di Lorenzo et al. (2006)
OCEAN INIT IALIZE
FINALIZE
RUN
S4DVAR_OCEAN
IS4DVAR_OCEAN
W4DVAR_OCEAN
ENSEMBLE_OCEAN
NL_OCEAN
TL_OCEAN
AD_OCEAN
PROPAGATOR
KERNELNLM, TLM, RPM, ADM
physicsbiogeochemicalsedimentsea ice
Optimal pertubations
ADM eigenmodes
TLM eigenmodes
Forcing singular vectors
Stochastic optimals
Pseudospectra
ADSEN_OCEAN
SANITY CHECK S
PERT_OCEAN
PICARD_OCEAN
GRAD_OCEAN
TLCHECK _OCEAN
RP_OCEAN
ESMF
AIR_OCEAN
MASTER
ean M ode
earch C o m
Non Linear Model
Tangent Linear Model
Representer Model
Adjoint Model
Sensitivity Analysis
Data Assimilation
1) Incremental 4DVAR Strong Constrain
2) Indirect Representer Weak and Strong Constrain
Ensemble Ocean Prediction
Stability Analysis Modules
ROMS Block Diagram NEW Developments
Arango et al. 2003Moore et al. 2003Di Lorenzo et al. 2006
Best Model Estimate (consistent with observations)
Initial Guess
ASSIMILATION Goal
STRONG Constraint WEAK Constraint (A) (B)
…we want to find the corrections e
Best Model Estimate (consistent with observations)
Initial Guess
ASSIMILATION Goal
Cost Function
4DVAR inversion
Hessian Matrix
Model x Model
4DVAR inversion
IROMS representer-based inversion
Hessian Matrix
Stabilized Representer Matrixµ TºR GPG
Representer Matrix
Model x Model
Obs x Obs
Representer Coefficients
WEAK CONSTRAINT
TRUE Mesoscale Structure
SSH[m]
SST[C]
ASSIMILATION Setup
Sampling:(from CalCOFI program)5 day cruise 80 km stations spacing
Observations:T,S CTD cast 0-500mCurrents 0-150mSSH
Model Configuration:Open boundary cond.nested in CCS grid
20 km horiz. Resolution20 vertical layersForcing NCEP fluxesClimatology initial cond.
SSH [m]
1st GUESS day=5
TRUE day=5
SSH [m]
WEAK day=5
STRONG day=5
TRUE day=5
ASSIMILATION Results
1st GUESS day=5
WEAK day=5
STRONG day=5
ASSIMILATION Results
ERRORor
RESIDUALS
SSH [m]
1st GUESS day=5
WEAK day=5
ASSIMILATION Results ERRORor
RESIDUALSSea Surface Temperature [C]
1st GUESS day=5
WEAK day=0
STRONG day=0
TRUE day=0
Reconstructed Initial Conditions
1st GUESS day=0
Normalized Observation-Model Misfit
Assimilated data:TS 0-500m Free surface Currents 0-150m
TS
VU
before assimilation
observation number
Normalized Observation-Model Misfit
Assimilated data:TS 0-500m Free surface Currents 0-150m
TS
VU
after assimilation
Error Variance ReductionSTRONG Case = 92%WEAK Case = 98%
observation number
SKILL = 1 – (SST RMS error)
daysassimilation window
Initial Guess
Climatology
Persistence
WEAK
STRONG
forecast
Subsurface Temperature
Free Surface Height
Salinity
Velocity
Persistence
Initial Guess
•Choosing climatology as the 1st guess leads to dynamically unbalanced fields, a strong initial shock, which degrades the quality of assimilated solution.
•Assimilating the data greatly improves the model trajectory for 10 days after the assimilation window when compared to the 1st guess.
•We should be able to exploit the long persistence timescale associated with the slow moving California Current eddies.
•A 5 day assimilation window may be too short to extract the time dependent dynamical information required to improve the model trajectory.
•Different definition of skill may be more appropriate to isolate the ability of the model to correct and predict the spatial structure of the eddies.
•Explore and characterize the dynamical sensitivities of the flow field, and the predictability timescales of the California Current.
THOUGHTS on the SCB test
PROGRESS• Developed and tested assimilation capability of ROMS for a realistic nested model setup (the California Current eddies ROMS can be used with IOM framework IROMS
WEAK day=5
STRONG day=5 ERROR
orRESIDUALS
Velocity (V)ASSIMILATION Results
1st GUESS day=5
AHV=4550AHT=4550
AHV=0AHT=0
AHV=4550AHT=1000
AHV=4550AHT=0
TANGENT LINEAR INSTABILITY SST[C]
TANGENT LINEAR INSTABILITY
TLMAHV=4550AHT=4550
TLMAHV=4550AHT=1000
TLMAHV=0AHT=0
Non Linear Model Initial
Guess
PROGRESS• Developed and tested assimilation capability of ROMS for a realistic nested model setup (the California Current eddies ROMS can be used with IOM framework IROMS • Tangent Linear Dynamics are very unstable in realistic settings. Need to find the “optimal” combination of increased viscosity/diffusivity and reduced physics to recover stability.
• Background and Model Error COVARIANCE functions are Gaussian and implemented through the use of the diffusion operator. We are implementing spatially dependent decorrelation length scales and additional dynamical constraint (e.g. geostrophy)
PENDING TECHNICAL ASPECTS
…end
Vzeta
temp
salt
WEAK day=0
WEAK day=5
WEAK day=5
STRONG day=5
CLIMA day=5
TRUE day=5
Velocity FULL
Weak Constraint Results
WEAK - dimensionalzeta =
81.303
temp =
88.112
salt =
97.177
u =
96.591
v =
95.9
TOTAL non-dimnensional=98%
Strong - dimensionalzeta =
84.374
var_red =
92.131
var_red =
94.522
var_red =
86.654
var_red =
88.683
TOTAL = 92%