1. Consistency & Fidelity of Indonesian Throughflow (ITF) Transport Estimated by Ocean Data Assimilation (ODA) Products Tong Lee NASA Jet propulsion Laboratory, California Institute of Technology

2. ODA products used for the intercomparison SystemModel and resolution Assimilation method Data assimilatedPeriod 1CERFACS, France OPA8.1-ORCA2 2°x(0.5°- 2°), 31 levels 3D-VAR SST, T & S profiles from EU’s EN3 Project (XBT, CDT, Argo, TOGA-TAO) 1960-2006 2 ECCO-GODAE – v3 (MIT-AER), USA MITOGCM, 1°x1°, 23 levels Adjoint Altimetry; scatterometry; tide gauges; gravity; SST, SSS; T & S profiles from XBT, CTD, Argo, TAO & other buoys, elephant seals (SeaOS); Florida Current; RAPID moorings 1992-2007 3ECCO-JPL, USA MITOGCM, 1°x(0.3°-1°), 46 levels Kalman filter and RTS smoother Altimetry, T profiles from XBT/CDT, Argo, TAO and other buoys 1993-2009 4ECCO-SIO MITOGCM, 1°x1°, 23 levels Adjoint Altimetry; scatterometry; tide gauges; geoid; SST, SSS; T & S profiles from XBT, CTD, Argo, TAO & other buoys 1993-2001 5ECCO2 MITOGCM, 18 km (cubed-sphere grid), 50 levels Green’s functions Altimetry; T & S profiles from XBT, CTD, Argo, in-situ sea ice concentration 1992-2007 6 ECMWF ORAS3, EU HOPE, 1°x(0.3°-1°), 29 levels 3-D OI with online bias correction Altimeter (sea level anomalies and global trends), SST, T & S from XBT, CTD, Argo, TAO 1959-2009 7G-ECCO, Germany MITOGCM, 1°x1°, 23 levels AdjointAltimetry; scatterometry; tide gauges; geoid; SST, SSS; T & S profiles from XBT, CTD, Argo, TAO & other buoys 1952-2001 Many thanks to the following groups for contributing ITF transport estimates

3. ODA products used for intercomparison (cont’d) 8GFDL, USA MOM3, 0.5°x(1/3°- 0.5°), 31 levels 3D-VAR? SST, T profiles from XBT, CTD, ARGO, TAO & S profiles from CTD, Argo 1960-2005 9INGV, Italy MOM, 0.5°x(1/3°- 0.5°), 31 levels OI T profiles from XBT, CTD, ARGO, TAO & S profiles from CTD, Argo 1957-2007 10K-7, Japan MOM3, 1°x1°, 36 levels Adjoint Altimetry, SST, T from XBT, CTD, Argo, TAO 1987-2004 11 MERCATOR-2, France OPA8.1-ORCA2, 2°x(0.5°-2°), 31 levels SEEK filter Altimetry, SST, T & S profiles from EU’s EN3 Project (XBT, CDT, Argo, TOGA-TAO) 1993-2008 12 MERCATOR-3, France OPA8.1-ORCA2, 2°x(0.5°-2°), 31 levels 3D-VAR Altimetry, SST, T & S profiles from EU’s EN3 Project (XBT, CDT, Argo, TOGA-TAO) 1960-2005 13MOVE-G, Japan MRI.COM, 1°x(0.3°- 1°), 50 levels 3D-VAR Altimetry, SST, T & S from XBT, CTD, Argo, TAO 1992-2009 14SODA, USAPOP, 0.25°x0.4°, 40 levels OIAltimetry, Satellite and in-situ SST, T & S profiles from MBT, XBT, CTD, Argo and other float data, TAO and other buoys 1958-2007

4. Estimate from INSTANT (2004-06): 15 Sv (± 25%) 1993-2001 mean (a) & variability (b) of ITF volume transport Negative means from Pacific to Indian Ocean. 1 Sv = 10 6 m 3 /s Averaged spread of anomalies among different products = 1.7 Sv “Signal-to-noise ratio” > 1

5. Seasonal (a) & non-seasonal (b) anomalies of ITF transport ECCO2 INSTANT Color curves represent other ODA products that have lower resolutions ( 0.4° to 2°). The better agreement between ECCO2 & INSTANT is attributed to the higher resolution of ECCO2 model on a C- grid, which allows a better representation of flows through narrow channels (esp. deep signals from the IO, e.g., semi-annual waves) ECCO2 has a dominant semi-annual signal (like INSTANT). Other ODA products have dominant annual cycle

6. Impact of resolution on the deep ITF signals: ECCO-JPL (1°x0.3°) & ECCO2 (18 km) similar in upper layer but not at depth

7. Non-seasonal ITF transport anomaly referenced to the common 1993-2001 seasonal cycle Similar interannual variations: stronger (weaker) ITF during La Nina (El Nino)

8. Five-year low-pass ITF transport anomaly (color curves) and their ensemble average (black curve) Decadal signals of 10-15 year periods. A consistent strengthening during 1992-2000: consistent with observed wind and SSH. No substantial weakening associated with the 1976 “climate shift”: Wainwright et al. (2008) reported a 2.5-Sv reduction based on an analysis of IX1 XBT data (Fremantle – Sunda Strait).

9. Ensemble mean ITF transport has 10-15 yr “cycle”; anti- correlated with SOI, not-so-good correlation with PDO index

10. Stronger trade cause SSH pile-up in the west near equator Off-equatorial curl enhance SSH rise in the west Decadal changes of ITF since the 1990s are consistent with observed wind & SSH ERS scatterometers T/P altimeter Lee & McPhaden (2008)

11. Given interannual-decadal changes, how well does a 3-year average (like INSTANT) represent longer-term mean? 3-year low-pass ITF transport anomalies & ensemble average Centered time of INSTANT estimate

12. Summary of ITF transport comparison among ODA products Time mean: 13.6±3.2 Sv (‘93-’01), consistent with INSTANT estimate of 15 Sv (‘04-’06). Seasonal variations: ECCO2 (18 km resolution) has a dominant semi-annual signal like INSTANT.  All other ODA products (0.25°-2°) show a dominant annual signal.  Resolution is crucial to resolving deep signals (e.g., the semi-annual signal from the Indian Ocean). Interannual & decadal variations: Stronger (weaker) ITF during La Nina (El Nino). Decadal signal of 10-15 year period, strongest during 1992-2000 associated with strengthening Pacific trade wind & higher western Pacific SSH.  No substantial weakening associated with the 1976 “climate shift”. A 3-year average (i.e., period of the INSTANT Program) would vary by a few Sv due to interannual & decadal variations. Need sustained measurements of the ITF.!