Towards an Operational Satellitebased AnalysisPrediction System using ROMS

Towards an Operational Satellite-based Analysis/Prediction System using ROMS: An Example for the East Australia Current Hernan G. Arango IMCS, Rutgers John L. Wilkin IMCS, Rutgers Javier Zavala-Garay IMCS, Rutgers

Outline • EAC application • Brief summary of IS 4 DVAR • Updated version on EAC-IS 4 DVAR analysis/prediction system • Background error covariance modeling • Some preliminary results • Future work

EAC

East Australia Current Application Configuration -24 -28 -32 -36 -40 -44 -48 145 150 155 160 165 Resolution 0. 25 x 0. 25 degrees Grid 64 x 80 x 30 DX 18. 7 - 29. 2 km DY 23. 6 - 30. 4 km DT (1080, 21. 6) sec Bathymetry 16 - 4895 m Decorrelation Scale 100 km, 100 -70 m Nouter, Ninner 3, 10 OBC NCOM (years 2001 and 2002) Forcing NOGAPS, daily

IS 4 DVAR

IS 4 DVAR • Given a first guess (a forward trajectory)

IS 4 DVAR • Given a first guess (a forward trajectory)… • And given the available data…

IS 4 DVAR • Given a first guess (a forward trajectory)… • And given the available data… • What are the changes (or increment) to the IC so that the forward model fits the observations?

The best fit becomes the reanalysis assimilation window

The final state becomes the IC for the forecast window assimilation window forecast

The final state becomes the IC for the forecast window assimilation window forecast verification

EAC-IS 4 DVAR

4 DVar Observations XBTs We have moved from SSH 7 -Day Averaged AVISO To SSH 4 -day avg. from CSIRO (compatible with SXBT=SCTD) SST Daily CSIRO Days since January 1 st 2001, 00: 00

EAC Incremental 4 DVar: Surface Versus Sub-surface Observations

EAC Incremental 4 DVar: Surface Versus Sub-surface Observations First Guess SSH/SST

EAC Incremental 4 DVar: Surface Versus Sub-surface Observations SSH/SST First Guess SSH/SST

EAC Incremental 4 DVar: Surface Versus Sub-surface Observations SSH/SST First Guess SSH/SST ROMS IS 4 DVAR: XBT Only SSH/SST

Comparison between ROMS fit and observed temperature from all XBTs. Note: actual XBT-data was not assimilated, it is used here to evaluate the quality of the reanalysis. SSH+SST

Comparison between ROMS fit and observed temperature from all XBTs. Note: actual XBT-data was not assimilated, it is used here to evaluate the quality of the reanalysis. SSH+SST Assimilation of SST

Comparison between ROMS fit and observed temperature from all XBTs. Note: actual XBT-data was not assimilated, it is used here to evaluate the quality of the reanalysis. SSH+SST Assimilation of SST Erroneous projection of SSH information

Example of synthetic XBT

Comparison between ROMS fit and observed temperature from all XBTs. Note: actual XBT-data was not assimilated, it is used here to evaluate the quality of the reanalysis. SSH+SST+Syn. XBT

Comparison between ROMS prediction and observed temperature from all XBTs.

some remarks • Good ocean state predictions for up to 2 weeks in advance • Need to correctly project the altimeter data • Proxies for subsurface information can be obtained based • • on surface information, but need lots of subsurface data to construct a robust empirical relationship The fact that an empirical (linear) relationship exist suggest that there could be a simple dynamical relationships linking the surface with the subsurface variability The idea is actually not new (Weaver et al 2006: “multivariate balance operator”)

What we have learned • Need a good unbiased background or first guess 1. From a “good” model forced by unbiased boundary 2. • • 1. 2. and surface forcing From assimilation of climatological seasonal cycle of T and S (e. g. Levitus, CARS). Adequate modeling the background error covariance B is a crucial. So far, to model B we need to specify Spatial decorrelation scales Standard deviations of the increments.

Assimilation of T-S climatology (CARS)

Background error covariance modeling

How can the information content of one variable be transferred to other variables? • Two possibilities: 1) Using the adjoint model 2) Modeling of the background covariance matrix

Balance operator for mesoscale variability • Forward problem (scalar SSH given a 1. 2. 3. 4. • vector d. T) Given d. T (e. g. , XBT) Find d. S Find density anomalies Find SHH. Inverse problem (given scalar SSH guess the vectors d. T and d. S).

NEED LOTS OF OBSERVATIONS

NEED LOTS OF OBSERVATIONS NEED CLIMATOLOGY

Future work • Test IS 4 DVAR in US east coast • Implement/evaluate the balance formulation and compare with statistics derived from observations • Ensemble prediction