1. www.cmar.csiro.au/bluelink/ Building Bluelink David Griffin, Peter Oke, Andreas Schiller et al. March 2007 CSIRO Marine and Atmospheric Research

2. Bluelink: a partnership between the Bureau of Meteorology, CSIRO and the Royal Australian Navy Introduction

3. Bluelink: a partnership between the Bureau of Meteorology, CSIRO and the Royal Australian Navy Talk Outline Ocean Forecasting Australia Model, OFAM Bluelink Ocean Data Assimilation System, BODAS Bluelink ReANalysis, BRAN Bluelink High-Resolution Regional Analysis HRRA Introduction

4. What users want: (a week in advance?)

5. Where they want it:

6. Ocean Forecasting Australia Model, OFAM … every 10 th grid point shown  Global configuration of MOM4  Eddy-resolving around Australia  10 m vertical resolution to 200 m, then coarser  Surface fluxes from ECMWF (for reanalyses) Minimum resolution: ~100km ~10km resolution

7. Bluelink Ocean Data Assimilation System, BODAS  Ensemble OI … sequential assimilation technique

8. Bluelink Ocean Data Assimilation System, BODAS  Ensemble OI … sequential assimilation technique  Assimilates observations of SLA, SST, in situ T and S

9. Bluelink Ocean Data Assimilation System, BODAS  Ensemble OI … sequential assimilation technique  Assimilates observations of SLA, SST, in situ T and S  To constrain the model to match reality, then make a forecast

10. Bluelink Ocean Data Assimilation System, BODAS  Multivariate assimilation system:  sea level obs correct h,T,S,U,V Single point assimilation …

11. Bluelink Ocean Data Assimilation System, BODAS  The spatial structure of the covariances are determined by the statistics of the free-running model. Influence of sealevel obs at x

12. Ensemble OI : vertical projection of surface observations - similar to multiple linear regression Cross-section of temperature increments Plan view of sea-level increments

13. Cross-section of temperature bkgnd (grey) & analysis (black- colour) Plan view of sea-level increments -> need both SST and SLA.

14. BRAN1.0  BRAN1.5  BRAN2.0 BRAN1.0BRAN1.5BRAN2.0 10/1992-12/20041/2003-6/200610/1992-12/2006 Assimilates along-track SLA, T(Z), S(z) Assimilates along-track SLA, T(z), S(z), AMSRE - SST Assimilates along-track SLA, T(z), S(z), AMSRE – SST or Rey 1/4 o OISST no rivers Seasonal climatological river fluxes SSS restoring (30 days); SST restoring (30 days) no SSS or SST restoring SSS restoring (30 days in deep water only); no SST restoring ECMWF surface heat, freshwater and momentum fluxes ECMWF surface heat and freshwater fluxes; and momentum fluxes from 10 m winds 3 day assimilation cycle7 day assimilation cycle with 1 day nudging using 1 day relaxation 7 day assimilation cycle with 1 day nudging using 0.25 day relaxation A few bugsNo known bugs (yet)Fingers crossed

15. BRAN1.0  BRAN1.5  BRAN2.0 BRAN1.0BRAN1.5BRAN2.0 Warm biasNo temperature bias Noticeably discontinuous in time (jumpy, shocks etc) Acceptably continuous (can track features easily) SST errors ~ 2-3 degreesSST errors ~ 0.6-0.8 degrees SLA errors ~ 15 cmSLA errors ~ 8 cm

16. Free-running model:

17. HRRA - Gridded altimetry and SST, statistically projected to depth:

18. BRAN1.0:

19. BRAN1.5 smoother, more realistic, no warm bias

20. BRAN1.5 cf HRRA – 2005

21. Conclusion BRAN1.0  plenty of lessons learnt BRAN1.5 realistically reproduces the 3-d time-varying mesoscale ocean circulation around Australia Major threat to real-time equivalent of BRAN: Less data available for assimilation

22. Thank you

23. An application: dispersal of the larvae of Southern Rock Lobster

24. Bluelink ReANalysis, BRAN BRAN1.5:  1/2003 – 6/2006  Forced with ECMWF forecast fluxes  Assimilates observations once per week  Assimilates SLA from Jason, Envisat and GFO (T/P with-held)  Assimilates AMSRE SST  Assimilates T and S from Argo and ENACT database

25. BRAN1.5 vs TAO ADCP zonal currents 165E170W 147E140W110W

26. BRAN1.5 vs CLS 1/3 o GSLA ANALYSIS 0-DAY FORECAST 7-DAY FORECAST

27. Comparisons with with-held T/P altimetry (top) and AMSRE (bottom) Comparisons between BRAN1.5 and with-held T/P altimetry:  RMS error of 8-10 cm  anomaly correlations of 0.6 Comparisons between BRAN1.5 and AMSRE (every 7 th day is assimilated):  RMS error of 0.7 o  anomaly correlations of 0.7

28. Observing System Experiments Experiment design  With-hold each component of the observing system  6-month integration (1 st half of 2003)  compare to with-held observations  treat BRAN1.5, with all observations assimilated, as the “truth”

29. Observing System Experiments Assimilation of Argo and SST reduces the forecast error of SLA by ~50% compared to the assimilation of altimetry Assimilation of Altimetry and Argo only reduces the forecast error of SST by a small amount 2003

30. Observing System Experiments For the 2003 - GOOS:  each component of the GOOS has a unique and important contribution to the forecast skill of upper ocean temperature  each component has comparable impact on the forecast skill of the upper ocean temperature Metric  Depth average (0-1000 m) of the RMS “error” in potential temperature